Major Trends in Theoretical and Applied Linguistics Volume 1 Versita Discipline: Language, Literature

Home , English numerals, Pari (letter), Relativizer

Edited by: Nikolaos Lavidas Thomaï Alexiou Areti-Maria Sougari

Managing Editor: Anna Borowska

Language Editor: Edgar Joycey Major Trends in Theoretical and Applied Linguistics: Selected Papers from the 20th International Symposium on Theoretical and Applied Linguistics (April 1-3, 2011) / Edited by: Nikolaos Lavidas, Thomaï Alexiou & Areti-Maria Sougari.

Published by Versita, Versita Ltd, 78 York Street, London W1H 1DP, Great Britain.

This work is licensed under the Creative Commons Attribution-NonCommercial- NoDerivs 3.0 license, which means that the text may be used for non-commercial purposes, provided credit is given to the author.

ISBN (paperback): 978-83-7656-074-8

ISBN (hardcover): 978-83-7656-075-5

ISBN (for electronic copy): 978-83-7656-076-2

Managing Editor: Anna Borowska

Language Editor: Edgar Joycey www.versita.com

Cover illustration: © Istockphoto.com/skvoor Major Trends in Theoretical and Applied Linguistics contains 80 papers on Theoretical and Applied Linguistics by prominent and young researchers, representing a large variety of topics, dealing with virtually all domains and frameworks of modern Linguistics. These papers were originally presented at the 20th International Symposium on Theoretical and Applied Linguistics at the Aristotle University of Thessaloniki in April 2011. The first volume is organized into two parts: Invited Papers; Selected Conference Papers: Phonology - Phonetics, Syntax - Morphology - Semantics. Contents

Foreword...... 9 Reviewers...... 15

Part I Invited Papers

Joseph Emonds Q: Natural Language’s only Functional Head...... 19

Geoffrey Leech Growth and Decline: How Grammar has Been Changing in Recent English...... 47

Sophia Marmaridou Towards a Constructional Account of Indefinite Uses of Proper Names in Modern Greek...... 67

Amalia Moser Aspect and Aktionsart: A Study on the Nature of Grammatical Categories...... 99

Nicos C. Sifakis Asking the Right Questions in “New School” EFL Curriculum Design...... 121

Part II Selected Conference Papers

Section 1: Phonology - Phonetics

Evia Kainada and Angelos Lengeris The Acquisition of English Intonation by Native Greek Speakers...... 141 Katerina Nicolaidis and Mary Baltazani The Greek Rhotic in /rC/ Sequences: An Acoustic and Electropalatographic Study...... 157

Nina Topintzi and Mary Baltazani Where the Glide Meets the Palatals...... 177

Eleni Tsiartsioni The Production of English Aspirated Stops in Foreign Language Acquisition...... 197

Section 2: Syntax - Morphology - Semantics

Rusudan Asatiani The Proto-Kartvelian and Proto-Indo-European Common Typological Feature: An Active Alignment (?)...... 215

Ifigeneia Athanasiadou and Martha Lampropoulou A Conceptual Metonymy Account of Count and Non-Count Nouns: A Study of Modern Greek Nouns from the Domains of Eating and Drinking...... 233

Maria Chondrogianni Basic Illocutions of the Modern Greek Subjunctive...... 249

Aikaterini Delikonstantinidou Pride Concepts...... 273

Konstantinos Kakarikos Case Attraction in Free Relative Clauses of Ancient Greek: A Study of the Syntax–Morphology Interface...... 289

Φρύνη Κακογιάννη Ντοά Τα διαθεσιακά επιρρήματα: Επιμέρους υποκατηγορία προτασιακών επιρρημάτων της Νέας Ελληνικής ...... 305

Haritini Kallergi The Role of Repetition in the Rise of Concessivity...... 319

Γεωργία Κατσούδα Τα ρήματα σε -άμαι της Νεοελληνικής Κοινής...... 335 Axiotis Kechagias The Syntax-IS Interface: On the Functional Discrepancies between Clitic Left Dislocation and ‘Bare Left Dislocation’ in Modern Greek...... 353

Dimitra Lazaridou-Chatzigoga Distributivity and Genericity in Greek: The Case of kathe with the Definite Article...... 369

Elisabeth Mela-Athanasopoulou Adjectival Participles Bearing on Unaccusativity Identification. Evidence from Modern Greek...... 385

Masaki Ohno Stranded Quantifiers, Reconstruction and QR...... 393

Isabel Oltra-Massuet and Isabel Pérez-Jiménez On Scalar Predicative PPs in Spanish ...... 401

Theofanis Papoutsis Measuring the Productivity of Noun-Deriving Suffixes across Languages: Greek -tita vs. English -ness ...... 421

Antonio R. Revuelta Puigdollers Result Clauses in Modern Greek and Spanish: A Contrastive Study...... 439

Juan Romeu A vs. en in Spanish Locatives...... 459

Anna Roussou, Christos Vlachos and Dimitris Papazachariou In Situ, Ex Situ and (Non) Echo Questions...... 475

Marietta Sionti, Leonardo Claudino, Yiannis Aloimonos, Carolyn P. Rose and Stella Markantonatou Semantic Clusters Combined with Kinematics: The Case of English and Modern Greek Motion Verbs...... 495

Ekaterina Tarpomanova Dativus Ethicus in the Balkan Languages...... 511 Major Trends in Theoretical and Applied Linguistics

8 Foreword

The present edited book is based on papers presented at the 20th International Symposium on Theoretical and Applied Linguistics that took place in Thessaloniki, April 1-3, 2011. From the time we started organizing the Symposium, setting the date and location of the event, until the moment we had the book ready in hands we felt we were involved in a ‘marathon’…of Linguistics. All subfields, all schools, and all frameworks were present at the Symposium. We were honored to receive many high quality submissions (abstracts and articles). More than 150 scholars from 20 countries presented their work and exchanged thoughts, concerns and insights. This book consists of 80 papers organized in three volumes, two parts and seven sections. Part I includes papers presented by plenary speakers, Part II includes selected papers presented at the Symposium. All of the articles focus on contrastive linguistic questions that mainly concern Greek and English by aiming, in parallel, to show a large part of the linguistic research that has been accomplished in Greece. Papers written in Greek continue the long tradition of earlier volumes that were based upon and derived from conferences on Theoretical and Applied Linguistics organized by the Department of Theoretical and Applied Linguistics, School of English, Aristotle University of Thessaloniki. They present the significant work conducted in Greece and strengthen the contrastive linguistic studies that are available. The invited papers (Part I of volume 1) present the major dimensions of the linguistic study of today. Hence, the main questions that concern modern research, but that can also be considered classical questions, are represented in the first part of the book. These questions pertain to the status ofthe functional (modifier) categories as being independent of lexical categories and as stemming from the natural language ability to count and/or quantify (Joseph Emonds); the significance of comparable (sub)corpora to observe changes in progress in written English and to compare the rates of change from one genre to another and across regional varieties (Geoffrey Leech); the issue of proper nouns in Modern Greek, which as heads of nominal constructions unify with

Foreword 9 Major Trends in Theoretical and Applied Linguistics

determiners and adjectives in the same way that common nouns do, and, on the other hand, activate semantic frames associated with particular categories of entities (Sophia Marmaridou); the analysis of aspect and its relationship with the disputed category of Aktionsart, as well as its relationship to tense (Amalia Moser); and the ‘deep fundamentals’ of EFL curriculum design, i.e., assumptions that tend to be taken for granted by policy makers (Nicos Sifakis). The first section of the second part examines some crucial issues of Phonology and Phonetics –most of the articles connect theoretical analyses to L2 acquisition. These issues concern acoustic and electropalatographic data to analyze rhotic production in /rC/ sequences in Greek (Nicolaidis & Baltazani); the relationship of the glide /J/ to the high front vowel and the palatal consonants (Topintzi & Baltazani); the production of polar questions and their pitch range by Greek L2 learners of English (Kainada & Lengeris); and the effect of teaching on the production of L2 English aspirated stops among Greek EFL learners (Tsiartsioni). The second section is the largest of the theoretical linguistic sections that demonstrate, among other aspects, the significant research in the area of syntax/morphology and semantics, as well as in their interfaces, in Greece. The topics of Section 2 include the core problems of the syntactic analysis: echo and non-echo questions in Greek, realized either ex-situ or in-situ, and their syntactic differences, depending on the position where the wh-phrase is realized (Roussou, Vlachos, & Papazachariou); the complex area of split-ergativity, telicity, and syntactic reconstruction (Asatiani); the behavior of result subordinators in Modern Greek and Spanish (Revuelta Puigdollers); the differences between clitic left dislocation and ‘bare left dislocation’ with respect to Information Structure interface rules (Kechagias); the position of the stranded object-related quantifiers (Ohno); and the ethic dative case of the personal pronouns in the languages of the Balkan Sprachbund (Tarpomanova). Section 2 also addresses central questions for Semantics, including attitudinal phrasal adverbs, i.e., adverbs that express the speaker’s attitude to what s/he is saying (Kakoyianni Doa); countability preferences as motivated by the use of conceptual metonymies (Athanasiadou & Lampropoulou); conceptual metaphors that originate from the emotion concept of Pride (Delikonstantinidou); and the relation between linguistic notions and sensorimotor data (Sionti, Claudino, Aloimonos, Rose, & Markantonatou). Section 2 also extends to the results of research in the area of Morphology, such as the deponent verbs in Modern Greek, the problems of their categorization and lemmatization in grammar books and dictionaries and the changes in their morphology (Katsouda); the adjectival properties of past participles and the mismatches between predictions of past participle-to-adjective conversion and other unaccusativity diagnostics (Mela-Athanasopoulou); and the productivity of the Greek noun-forming suffix -tita, in comparison to its English counterpart -ness (Papoutsis). Section 2 further analyzes the relationship between the different levels of linguistic analysis; that is, the significant area of Interfaces

1 0 Foreword Lavidas, Alexiou & Sougari

as it appears in the following phenomena: the basic illocutions associated with the Subjunctive in Modern Greek (Chondrogianni); asymmetries in syntax and morphology interface as illustrated by case attraction in the free relative clauses (Kakarikos); types of lexical repetition that are involved in structures of asyndetic clause linkage with a concessive meaning (Kallergi); the interaction between determiners and quantifiers in the Greek Noun Phrase (Lazaridou- Chatzigoga); the locus of scalar meaning in Prepositional Phrases and how it is compositionally determined (Oltra-Massuet & Pérez-Jiménez); and the internal syntactico-semantic structure of the prepositions a and en in Spanish locative constructions (Romeu). The third section (Volume 2) attempts to portray the link of theoretical linguistic analysis to discourse and society. This relation can be shown in various aspects, including the ways in which sexual orientation is articulated with language (Makri- Tsilipakou), and the relationship between grammatical gender and cognition and the type of correlation between the two (Pavlidou & Alvanoudi). Section 3 evaluates the treatment of ‘negative’ terms in Modern Greek dictionaries and the problems that relate to usage labels (Efthymiou, Gavriilidou & Papadopoulou); and the dictionary users’ abilities and performance during receptive dictionary use (Gavriilidou). It further assesses the Greek-Cypriots’ view of the appearance of dialect and foreign features in local advertising (Papapavlou & Satraki) and persuasive language in political and media discourse (Tsaroucha). This section also addresses verbal semantics as a useful resource for reconstructing not only the semantics of whole sentences, but also for inferring relations between larger pieces of text (Tantos); the spoken classroom discourse and the correlation between meaning and interaction (Christodoulidou); the practices employed in a local newspaper during the production of news (Ntanopoulos); and the ways in which the lyrics of popular rock music are associated with change as it is imprinted in language (Sophiadi). The fourth section addresses important issues that inform the process of L1 and L2 language acquisition. More particularly, research findings allow conclusions to be drawn in relation to the effects of the story presentation method (booklet vs. card) on the linguistic variables of reference, loose linking (e.g. ‘and then…’), coordination, subordination and narrative length in children from 3;6 to 6;6 years old (Tsimpli, Papakonstantinou & Nicolopoulou); the acquisition of clitic placement in Cypriot Greek on the basis of spontaneous speech and elicited production data (Neokleous & Parodi); how motion verbs in L1 Greek interact with aspect in the expression of manner of motion (Kotroni); the production of aspectual forms by children acquiring Greek as a native language (Kaltsa); the acquisition of wh-interrogatives by Greek children (Asproudi); and syntactic errors and strategies found in children’s speech in an endeavor to outline linguistic development in early speech (Papadopoulou & Pavlou). Cross-linguistic studies enhance our understanding of the processes involved in

Foreword 1 1 Major Trends in Theoretical and Applied Linguistics

acquiring particular languages; with this view in mind, the study of cross-linguistic influences in the speech production of trilingual children in the Greek context focuses on interactional strategies and transfer lapses (Anastassiou & Andreou). Based on theoretical and research findings, a study was conducted with a view to assessing the proficiency level of learners of Greek when forming compounds in order to determine whether they can be considered bilinguals or second- language learners of Greek (Kalligiannaki & Tzakosta). The role of interlanguage receives attention in a corpus study of postverbal subjects in English as a second language within the generative grammar framework (Agathopoulou). Moreover, the role of interlanguage is examined to determine: the degree of interlanguage interference found in Greek speakers who also speak Italian and Spanish (Karatsiolis & Kambakis-Vougiouklis), the acquisition of morphological agreement in L2 Standard Modern and Cypriot Greek (Karpava & Grohmann), the acquisition of relativization by advanced Greek-speaking learners of French in oral narrative (Monville-Burston & Kounouni), and the L2 acquisition of English pronominal subjects by Greek learners (Prentza). The fifth section deals with critical issues in language disorders. More specifically, it includes studies of aspects of: the Relativized Minimality approach in Greek aphasia and the asymmetries between production and comprehension across question types (Nerantzini, Papadopoulou & Varlokosta); the inhibitory mechanism that serves to limit the generation and maintenance of irrelevant information in Greek-speaking individuals diagnosed with Broca’s aphasia (Peristeri & Tsimpli); bilingual aphasia and the effect of modality-specific noun– verb dissociations in L1 and L2 (Kambanaros); and narrative abilities in children with SLI (Kambanaros, Grohmann, Theodorou & Michaelides). The sixth section (Volume 3) deals with language learning and how teaching practices ought to accommodate the way learners learn languages. In addition, assessment procedures receive special attention. In particular, this section seeks ways of improving traditional Modern Greek vocabulary learning by suggesting an etymo-cognitive approach (García Aragón); of illuminating the lexical features of textbooks used in six curriculum subjects of Irish post-primary education (Kostopoulou); of analyzing mediation tasks included in the national exams for language proficiency (KPG) (Stathopoulou); of examing reading formulas and tools and their credibility in the examination process (Tzimokas & Mattheoudakis); and of investigating the extent of identification and comprehension success achieved by Greek EFL high school students when encountering phrasal idioms during reading English as a foreign language (Katsarou). Language learning in schools (and universities) is examined with regard to students’ attitudes towards the assessment process, which professed to develop learner autonomy (Everhard); the teacher observation procedures undertaken and suggestions for improving such practices (Kotsiomiti); strategies and strategy use by elementary school learners of English in a Greek context (Vrettou); and old and new primary

1 2 Foreword Lavidas, Alexiou & Sougari

English course books in terms of breadth, frequency and recycling of vocabulary items (Zapounidis). The development of the learners’ writing skill receives attention in three papers: process-writing is seen as a possible means of aiding the development of this particular skill (Anastasiadou), peer assessment is put forward as an alternative means of assessing students’ writing (Meletiadou & Tsagari), and the employment of cognitive and metacognitive strategies is traced in bilingual pupils’ creative writing (Geladari & Mastrothanasis). In this section, some other issues that receive attention are as follows: the varying motivations to learn different languages in an intercultural classroom setting (Iliopoulou & Sougari); the presentation of the new proficiency levels and the new format of the Certificate of Attainment in Greek (Karakyrgiou, Panagiotidou & Antonopoulou); the impact of the Common European Framework of Reference on teaching languages and assessment in the Greek educational context (Kokkinidou, Markou, Rousoulioti & Antonopoulou); the production of requests in the interlanguage of foreign language (FL) learners of Greek (Bella); the L2 acquisition and teaching of Greek complementizers (Nouchoutidou), the suitability of the topics utilized in students’ textbooks for learning Greek as a foreign language (Giannakou). The following research questions are also discussed: Linguistics textbooks in Greek, the practices used in the last 30 years, and the need for new textbooks (Xydopoulos, Tsangalidis & Prountzou); the big question ‘To CLIL or not to CLIL’ with respect to the effects of content and language integrated learning (CLIL) in the 3rd Experimental Primary School in Evosmos, Thessaloniki (Mattheoudakis, Alexiou & Laskaridou); the learners’ attitudes and strongly positive perceptions of CLIL in the tertiary section (Soulioti); the use of teacher illocutionary acts in raising pragmatic awareness (Michail); the effects of working on multi-word expressions, especially in the development of lexical competence (Piasecka); proximal predictors of L2 willingness to communicate in Polish adolescents (Piechurska-Kuciel); and distance education through the eyes of the students in contact sessions (Tsagari). The selection of papers and the reviewing process have been made possible with the help of a number of academics who offered their time and expertise. We would like to heartily thank our invited speakers, Joseph Emonds, Geoffrey Leech, Sophia Marmaridou, Amalia Moser, Nicos Sifakis, for accepting our invitation and sending us their papers. We must also thank Vicky Papachristou and Eirini Kelmali, members of the organizing committee, for their hard work and contribution to the success of the Symposium. Along with the main session, three workshops were organized, offering thematically unified sessions in both theoretical and applied areas of research. More specifically, the three workshops were as follows: Proto-Indo-European Syntax and its Development (organized by Leonid Kulikov and Nikolaos Lavidas), Challenges in Early Foreign Language Learning and Teaching (organized by Areti-Maria Sougari and Thomaï Alexiou), Translation Process Research and Translator Training: Trends and Perspectives (organized

Foreword 1 3 Major Trends in Theoretical and Applied Linguistics

by Kyriaki Kourouni). We are grateful to all the members of the Department of Theoretical and Applied Linguistics and the School of English for their assistance and advice, and especially to Ianthi Maria Tsimpli, Angeliki Athanasiadou, Edgar- Matthew Joycey and Leonid Kulikov for their exceedingly useful comments on the final manuscript of the book, and to our computer expert, Anastasios Paschalis, for his support in organizing the technical aspects of the Symposium. A special word of thanks goes to the sponsors of the Symposium (The Research Committee of Aristotle University of Thessaloniki, European Centre for Modern Languages, Greek Applied Linguistics Association) for their financial support. We would also like to express our gratitude to our students, who acted as volunteers and who all contributed to the successful organization of the Symposium. They made the event not only possible but memorable. This book would not have become a reality and a publication available to all the ‘linguistics’ world, without the support of Versita, and, mainly, of Anna Borowska and Agata Morka. We enjoyed an excellent three-day event, a ceaseless reading of articles, and a linguistic “journey” from semantic conceptions... to teaching... and from syntactic phi-features... to learning strategies. The 20th International Symposium on Theoretical and Applied Linguistics was both scientifically and socially enjoyable and satisfying. We hope the selected papers will have the same success as the event.

Nikolaos Lavidas Thomaï Alexiou Areti-Maria Sougari

April 2013

1 4 Foreword Reviewers

Eleni Agathopoulou (Aristotle University of Thessaloniki) Argyris Archakis (University of Patras) Evangelia Asproudi (Aristotle University of Thessaloniki) Angeliki Athanasiadou (Aristotle University of Thessaloniki) Mary Baltazani (University of Ioannina) Spyridoula Bella (National and Kapodistrian University of Athens) Paul Bouniol (National and Kapodistrian University of Athens) Despina Chila-Markopoulou (National and Kapodistrian University of Athens) Maria Dimitrakopoulou (Center for the Greek Language) Carol Everhart (Aristotle University of Thessaloniki) Marc Fryd (Université de Poitiers) Zoe Gavriilidou (Democritus University of Thrace) Michalis Georgiafentis (National and Kapodistrian University of Athens) Eleni Griva (University of Western Macedonia) Konstantina Iliopoulou (Secondary Sector) Edgar-Matthew Joycey (Aristotle University of Thessaloniki) Maria Kaltsa (Aristotle University of Thessaloniki) Eliza Kitis (Aristotle University of Thessaloniki) Kyriaki Kourouni (Aristotle University of Thessaloniki) Leonid Kulikov (University of Ghent) Geoffrey Leech (Lancaster University) Marianthi Makri-Tsilipakou (Aristotle University of Thessaloniki) Sofia Marmaridou (National and Kapodistrian University of Athens) Maria Mastropavlou (University of Ioannina) Marina Mattheoudakis (Aristotle University of Thessaloniki) Elsa Mela-Athanasopoulou (Aristotle University of Thessaloniki) Michalis Milapides (Aristotle University of Thessaloniki) Amalia Moser (National and Kapodistrian University of Athens) Katerina Nicolaidis (Aristotle University of Thessaloniki)

Reviewers 1 5 Major Trends in Theoretical and Applied Linguistics

Kiki Nikiforidou (National and Kapodistrian University of Athens) Despina Papadopoulou (Aristotle University of Thessaloniki) Eleni Peristeri (Aristotle University of Thessaloniki) Periklis Politis (Aristotle University of Thessaloniki) Angeliki Psaltou-Joycey (Aristotle University of Thessaloniki) Antonio R. Revuelta Puigdollers (Universidad Autónoma de Madrid) Anthi Revythiadou (Aristotle University of Thessaloniki) Nicos Sifakis (Hellenic Open University) Maria Sifianou (National and Kapodistrian University of Athens) Vassilios Spyropoulos (National and Kapodistrian University of Athens) Melita Stavrou (Aristotle University of Thessaloniki) Dina Tsagari (University of Cyprus) Anastasios Tsangalidis (Aristotle University of Thessaloniki) Ianthi Maria Tsimpli (Aristotle University of Thessaloniki) Angeliki Tsokoglou (National and Kapodistrian University of Athens) Marina Tzakosta (University of Crete) Ioannis Veloudis (Aristotle University of Thessaloniki) Giorgos Xydopoulos (University of Patras) Keti Zouganeli (EFL Teacher-Teacher Trainer)

Organizing Committee of the 20th International Symposium on Theoretical and Applied Linguistics

Nikolaos Lavidas Areti-Maria Sougari Thomaï Alexiou Vicky Papachristou Eirini Kelmali

1 6 Reviewers Part I Invited Papers

1 7 Major Trends in Theoretical and Applied Linguistics

1 8 Q: Natural Language’s Only Functional Head1

Joseph Emonds

Palacky University, Olomouc [email protected]

Abstract

Most current versions of Chomskyan syntax take for granted that maximal or extended projections of the fundamental lexical categories N, A, V and P contain elaborate systems of functional heads and projections (sometimes referred to as “the functional sequence”) which are also significantly different for each of these four categories. This study argues that this approach more than “takes to extremes” this proliferation of syntactic categories; I argue here that it is fundamentally misguided. All functional modifier categories truly independent of lexical categories stem from the natural language ability to count and/or quantify. Among its other advantages, this hypothesis reveals for the first time the close affinity of subject phrases and measure phrases, and moreover provides a simple account of differences between English and Japanese regarding two ways of counting and agreement vs. non-agreement of predicates with subjects.

1. Which Closed Class Modifiers are “Functional Category Heads”?

Strangely, one of the prototypical and widely accepted functional heads, the category of Definite morphemes (“D”), fails or is neutral with respect to essentially every empirical test for being a head (Emonds 2012, summarized here in section 2). Rather, the functional head of nominal phrases is a universal

1 The first parts of this study revise Emonds (2007), from a locally distributed Japanese journal. An expanded version appears in Kawashima, Philippe and Sowley (2008).

Invited Papers 1 9 Major Trends in Theoretical and Applied Linguistics

quantifying node Q for counting, which includes cardinal numerals and plurality. Section 3 proposes that this system is formally based on the different ways Q can be “valued” as ±PLUR. Section 4 explores contrasting systems of QP projections for nominals, proposing a parameter distinguishing English from Japanese. Sections 5 and 6 then extend this “QP hypothesis” for functional heads to the categories AP and PP, showing that degree words and intensifiers are instances of Q and that measure phrases are located in SPEC(Q). Finally the QP hypothesis also covers VPs (section 7), since number agreement with a subject reduces to a default valuing of a Q that would otherwise be unvalued. Section 8 concludes that that recent work on functional categories has simply missed generalizations expressable in terms of Q and QP, and has thus seriously overstated the number of syntactic primitives. It is widely accepted that four central lexical categories of language (N, V, A, P) serve as “heads” (notated X or X0) that project to phrases XP, and that only these categories can be “open,” i.e. contain many hundreds of members and accept coining by adult native speakers. Throughout this study, XP is equivalently written as X’. When I refer to X0 and XP together as a class, I write Xj, e.g. both types of nominal projections taken together are Nj. Referring to heads and phases of the same type in this way is called the “bar notation.” Moreover in a given language, lexical heads tend to systematically precede or follow their phrasal sisters or “complement” YPs. This property is often uniform in a language across different choices of lexical heads. English for example is “head-initial” and Japanese is “head-final.” In these terms, it is well known that several small closed classes of non- phrasal modifiers can be added to these head X. For N we can call them “n”, for V we can call them “v”, etc. In head-initial English, the x (=n, v, a, p) are typically free morphemes.

(1) a. [NP two nbunches of nother [N boys] [YP from the city ]

b. [VP vhas vbeen vgetting [V cut] [YP from a tree ] ]

c. [AP { areal/ apretty / ahow amuch more } [A important] [YP to you ] ]

d. [PP pdown pover [P into] [YP that forest ] ]

In head-final Japanese, corresponding modifying x are often bound suffixes. Straightforward examples in Japanese of such grammatical n, so-called “classifiers,” and grammatical v (causative, passive, and politeness verbal suffixes) are given in Emonds (2007). Now since n and N are not simply two names for the same thing, what differentiates x from X? One clear difference is whether a category has hundreds or thousands of members, i.e. is “open,” or has at most perhaps twenty members that adult speakers cannot add to, i.e. is “closed”:

2 0 Invited Papers Joseph Emonds

(2) Dictionary Insertion. In a single maximal XP, lexical insertion from open classes X of the Dictionary is limited to the most internal X0 position in XP.

That is, in a head-initial XP, [XP X1 - X2 - … - Xk -…(YP)…], an open class of lexical

N can appear only in the Nk position. The other Ni must be closed class modifiers n. Research often calls the “small” modifiers (n, v, a, and p) in (1) “functional categories,” but what is their actual formal status in a system of syntactic primitives? Van Riemsdijk (1998) convincingly argues for the following hypothesis about their categorial nature.2

(3) Categorial Identity Thesis. Each n ε N, each v ε V, each a ε A, each p ε P.

Some brief examples of arguments for (3), based on the constructions in (1), are as follows. Further arguments for the CIT appear in Emonds (2001).

Each n is an N. Bunch and other in (1a) have regular N plurals, and bunch accepts adjectival and numeric modifiers. Quantity n such as bunch, couple, etc. can also function as independent nouns, as can certain Japanese numeric classifiers: dai ‘box’, nen ‘year’. Each v is a V. English auxiliary verbs as in (1b) all inflect like verbs. Similarly, Japanese verbal suffixes are themselves verbs, since they are regularly followed by verbal inflections such as the present tense -(r)u: tabe-ru ‘eats’, tabe-sase-ru ‘makes eat’, tabe-rare-ru ‘is eaten’, tabe-mas-u ‘eats’ in polite speech. Each a is a A. Real and pretty modify A in (1c), yet are clearly adjectives in their own right. Similarly, several contexts reserved for A also accept bare how: How does he seem? How did they treat him? Each p is a P. Down, over, etc. can be modifiers of P: down in the street, over toward town. They can also be independent prepositions: right down the street, two miles over the hill. In other combinations like from behind the barn, both from and behind exhibit properties of the category P.

Under van Riemsdijk’s CIT, English head-initial structures are thus as in (4).

(4) Functional category structures in head-initial systems:

[XP X1-X2-…-Xk-…(YP)…]

2 This study takes no position on whether each functional xi in (1) projects to a separate phrasal category xP. Although most studies of functional categories assume that they do, there are empirical arguments for flat structures without such xP, such as Kubo (1996) and Emonds (2001).

Invited Papers 2 1 Major Trends in Theoretical and Applied Linguistics

Then (2) requires that open class Xi must be next to the YP sisters that they

select, and not be separated from them by other Xi. Though the CIT is appealingly simple, it cannot be the whole story on functional categories. For example certain modifiers of adjectives in English (too, as, quite, rather, somewhat) actually share no properties with adjectives. It’s similarly unlikely that demonstratives are “nouns” (e.g. Japanese kono, sono, ano or Spanish este, ese, aquel). Nor do lower numerals such as 5-19 typically exhibit properties of other grammatical N, cross-linguistically. These kinds of discrepancies suggest that we must somehow extend or modify the CIT. I claim nonetheless that the categories conforming to the CIT need only be supplemented with a single additional quantification head Q. For clarity, I notate 3 Q as QX in contexts ___XP across values of X.

(5) The Q-extended CIT. Across languages, a single functional category head Q

can extend all four types of XP to XPQ.

The Q-extended CIT implies that the familiar node DP is to be written as NPQ or [N, Q]’,

and that IP = VPQ = [V, Q]’. By the same token, APs and PPs containing degree words

and expressions (or any other closed class modifiers) are to be written as APQ and j PPQ. The subscript notation on phrases means that both Q and X jointly project or “percolate” to a containing extended XP. The subscript Q on a bar notation category Xi thus indicates a feature that can be referred to in stating syntactic principles. An important property distinguishes “plain XP” from those that project to

XPQ. A plain XP can always project to a higher XP by means of an adjunction,

e.g. of an adverbial PP, though it need not. But an XPQ that contains a phrasal quantification cannot further project. It is thus a “closed projection” in the sense of Fukui and Speas (1986). We will see below that languages differ as to if and which projections must be closed in this sense.

While QN is not limited to numerals (see note 8), it almost certainly includes in any language some numerals for counting items with reference, i.e. nouns. English Q is used for all counting, while some Slavic languages (Veselovská

2001) use it only for high counting, i.e. QN > 4. The potential of Q as a counting device can be expressed as (6).

(6) Universal Counting. The unique functional head Q is the category for numerals and can combine with both types of nominal projections Nj.

3 The lexical category subscripts on Q are just shorthand for the category of their

sister phrase. Thus, the quantifiers P Q and QA differ formally in the same way as verbs subcategorized differently, such as V, ___PP and V, ___AP. These subscripts do not affect the syntactic identity of the category Q that they appear on.

2 2 Invited Papers Joseph Emonds

In the standard use of English numerals to count, Xj = N’. English can also combine 0 QN with N to create “counting compounds”: The phrases in bold in (7) are usually wrongly thought to be a type of measure phrase, i.e. an N’.

(7) a. a crispy [N [N [Q fifty ] N[ dollar(*s) ] ] [N bill ] ]

those great [N [N [Q ten ] [N day(*s ] ] [N bus passes ] ] b. *a fifty dollar(s) crispy bill *those ten day(s) great bus passes

But, as can be seen from their singular form (7a) and their ordering after pre- 0 4 nominal adjectives (7b), they are clearly compound nouns of the form [N QN + N] . Finally, I venture to claim that this basic category Q for counting and quantification is absent in animal communication. Its introduction was thus a fundamental mutation leading to human symbolic communication. Plausibly, the initial possibility of Merging with Q involved the largest, most concrete open class, the nouns N or more generally nominal projections Nj, as stated in (6). Merge of Q and Nj in essence created existential quantification, a necessary precondition for counting known in set theory as the Axiom of Choice; counting itself then required in addition only some mental version of a successor function (Peano 1889). While counting itself may have had little survival value, a mutated early human controlling existential quantification could also assert existence in the absence of stimuli, the essential characteristic of human language known as Displacement (Hockett 1960), whose survival value seems unquestionable.

The formal extension of QX and SPEC(QX) to other categories, features, and their meanings, as in (5), was a further development after this first leap.

2. The Content and Feature Values of QN Inside Noun Phrases5

I first argue for the Q-extended CIT (5) by establishing its validity fornoun phrases. That is, I will show that extended projections of N can contain a single

quantifying functional head above N. Other than QN, grammatical modifiers closer to N are themselves of category N, as the CIT (3) predicts. Moreover, I contest a widely assumed position—but one never actually argued for—that

4 These English [N QN + N ] never appear as isolated head Ns of NP: *I like a crispy fifty dollar in my pocket; *An ample vacation requires a good ten day. The English setting of the “Q-Parameter” in Section 4 predicts this, because it requires that head nouns further

combine with QN in NPQ, yielding e.g. An ample vacation requires a good ten days. 5 This section summarrizes material presented in Emonds (2007).

Invited Papers 2 3 Major Trends in Theoretical and Applied Linguistics

noun phrases contain additional heads higher than Q such as demonstratives, definiteness, or other quantifier nodes.

2.1. Quantification of Nouns

A comprehensive generative description of a closed class modifier system for English noun phrases is laid out in Jackendoff (1977: Ch. 5). According to him, nouns can be pre-modified by two main independent categories whose most

characteristic elements don’t seem like Ns. Here I re-name them D and QN ; they

then appear in sequences D – QN - N.

(8) Closed class modifiers for English N D = the, demonstratives, WH-pronouns, universal quantifiers (each, every, all, both), some, any, no, which, what. Possessive NPs also compete for the unique D position in this system, co-occurring only with all and both.

QN = { numerals, many, few, much, little, several, a(n).

According to Jackendoff, a noun in an NP can be modified by only one D and one

QN. There are a few idiomatic or otherwise atypical uses of these words that don’t conform to this statement, not further treated here: every which way, his every step, what the hell, a few steps, etc. In the other direction, as Jackendoff shows, D and Q quantifiers with their usual logical meanings typically don’t combine in a single NP: *all few, *any many, *each several, *every many, *some much, *no a(n), *every a(n), etc. I propose to strengthen the categorial dichotomy in (8) by two general

principles for interpreting these categories: (i) The logical role of all QN items is existential quantification(this seems straightforward), and (ii) D houses what are arguably universal quantifiers. These correlations with meaning are interesting consequences of the division in (8), but are not necessary preconditions for the validity of such structure. The second correlation, that D is uniformly a universal quantifier position in LF, in fact depends on several non-obvious but intriguing and quite plausible auxiliary hypotheses. a. N. Chomsky (class lectures, early 1980s) proposed that a definite article is simply universal quantification over sets defined within a single universe of discourse. Their close relatives, the demonstratives, should be analyzable in similar terms. b. Chomsky also proposed that any is a universal quantifier with a special property of always taking wide scope. 6

6 We might treat no as a universal quantifier with a wide scope property similar toany : “We own no cars” = “For all x, x a car, ~(we own x).”

2 4 Invited Papers Joseph Emonds

c. Finally, which is often taken as a WH-counterpart to a definite article; like definites it is “Discourse-linked.” Space prevents developing fuller arguments for these hypotheses, but together they strongly suggest the accuracy of the dichotomy in (9).

(9) In LF, (i) QN is existential quantification, and (ii) D is universal quantification.

The only English D that seems to violate (9ii) is the existential some. So as to maintain these attractive LF generalizations, I propose that the D some

“alternatively realizes” the existential quantifier category QN; cf. (21) below. This means that some spells out as an uninterpreted D in PF, while its unpronounced 7 sister [Q Ø] is interpreted, as minimally marked existential quantification. Then, as predicted, no precise LF differences distinguish pairs such as three X/ some three X; few X/ some few X.

The general structure of NPQ for English I thus hypothesize is then (10).

(10) NPQ

D QN NP

{ those/ all/ which/ {three/ few} N1 Nk YP every/ any/ some } bunches of friends from school

2.2. QN as the Unique Functional Category Head of NPQ

The question immediately posed by (10) is whether D or Q or both are functional

category heads of NPQ . For Q, there can be little doubt: Giusti (1991), Ritter (1991), Veselovská (2001) and Cardinaletti and Giusti (2006) have established that a quantifying and counting head Q, sometimes termed NUM, is indeed

a functional head Fn above N within noun phrases; Jackendoff’s term for this category is SPEC(N’’). 8 In support of this, we see that Q exhibits many expected head properties listed in (11), which D conspicuously lacks. Contrary to a widespread assumption in

7 For numerous other examples of alternative realization, e.g., agreements, case-marking, Romance clitics, etc. see Emonds (2000: Ch. 4). 8 The English article a(n) and quantifiers many, few, much, little and several are in complementary distribution with cardinals and hence should be in the same categorial position (Jackendoff 1977: Ch. 5). Ritter’s (1991) label is NUM, but Q is preferable because Q has uses besides simple counting. Incidentally, since these quantifiers can be further

modified by AQ , they must be As in the QN position.

Invited Papers 2 5 Major Trends in Theoretical and Applied Linguistics

generative studies dating from Abney (1987), the very same tests that establish that Q is the head of a functional projection show conclusively that D is not a head. For more detailed paradigms and arguments for the following contrasts, especially (11d-f), see Emonds (2012).

(11) a. Q has a role in how NPs are selected, but D does not. b. Q has a role as a head that assigns case to NPs, but D does not. c. Q can serve as a right hand head of Nj’ in Japanese, but D cannot. d. As complement phrases of a head Q, NPs sometimes move. If D were a head, Q + NP would be a phrase D’ and hence should sometimes move, but it never does. e. NP sisters of Q can in certain cases be coordinated, but there is no such coordination of putative sisters of D. f. NP sisters of Q can sometimes undergo ellipsis, but there are no corresponding paradigms for putative complement sisters of D, i.e. no ellipted sequences Q+NP.

I now briefly exemplify each of the five points (11a-e). Selection (11a). Q plays a role in selection of extended noun phrases; for example verbs like disperse and gather require underlying object NPs with plural or collective count noun heads, thus involving a feature of Q. Similarly, Abney (1987: 86-88) observes that various Navajo verbs select for singular, dual or plural NPs, even though “Navajo does not actually mark any of these distinctions (object class or number) in its determiner.” Since he does not consider Q as a possible head of extended NPs, he declares the Navajo pattern “a curiosity.” In contrast, “D does not appear to be selected by a matrix head” (Abney 1987: 85). For example, no verbs select only definite phrases. Case Assignment (11b). Like other functional heads (in particular I), Q can sometimes assign case, as well as block case-assignment to its sister NP by a more distant head. In a number of languages, Qs such as existential quantifiers or high numerals assign morphological genitive case to their sister NP. D has no such role in assigning characteristic case within NPs. Veselovská (2001) amply illustrates these properties and the contrasting syntax of D and Q in Czech. Head Placement (11c). in purely head-final Japanese, numerals with classifier suffixes can appear in head position of extended NPs, exactly as expected if they

are functional heads QN with a preceding NP complement.

(12) [PP Teburu-no ue-ni ] [QP [NP ookina hon ] [Q yon-satsu ] ga ] aru. Table-of top-at big book four-CLAS-NOM be ‘There are four big books on the table.’

2 6 Invited Papers Joseph Emonds

(13) [QP [NP [YP Sono daigaku no ] [N gakusei ] ] [Q san- nin ] ga ] tsui-ta. that university-GEN student three- CLAS-NOM arrive-PAST ‘Three students of that university arrived.’9

In contrast, the Japanese demonstratives D kono/ sono/ ano ‘this/ that/ that’ and its WH N-modifier dono ‘which’, have no head-like behavior. Unlike uniformly final Japanese heads, these Ds must be pre-nominal, and can be ordered freely among other adjectival and possessive complements and modifiers. Movement (11d). In general, bare lexical projections such as VP exhibit less robust phrasal behavior than full extended projections (IP/ CP). Similarly, the NP sisters of Q have some phrasal properties (11d-f) though fewer than do extended

NPQ. For example, some constructions can exhibit movement of NP sisters of Q (14), though such movements are not so productive.

(14) a. [NP Flowers for Easteri ] we don’t have many of ti.

b. Not much ti was eaten [NP of leftover turkey ]i.

Coordination (11e). Examples (15) contain coordinated NP sisters of QN.

(15) a. We didn’t buy [QP many [NP books on culture] or [NP guides for tourists]].

b. [QP Two [NP students of music] and [NP friends of my sister]] live with me.

The contrasts in (11) thus all favor QN over D as a functional head above the

N head in nominal phrases. It appears that the place of D in NPQ is rather in its “Specifier,” a position almost universally accepted in bar notation studies of phrasal projections.

(16) Specifier Position. A functional head QX licenses a SPEC(QX) on its left, independently of a language’s word order.

There is thus extensive support for the prototypical structure (10) for extended NPs in English, Czech and probably many head-initial languages, where D and

9 This construction contrasts with a second way of counting in Japanese, whereby a numeral compound appears as a modifier inside an NP, set off from a final head N by the subordinating genitive marker no. Thus, the following example is an alternative to (12).

(i) [PP Teburu-no ue-ni ] [NP ookina [XP [Q yon-satsu ] no] [N hon ] ga ] aru. Japanese numeral compounds can appear with nouns in two further positions (Oga 2002); they can float off the NP rightward and also leftward (Okuda 2006). In these configurations Okuda shows they are exterior to NP, even if adjacent to NP.

Invited Papers 2 7 Major Trends in Theoretical and Applied Linguistics

QN are defined as in (8). Jackendoff’s “quantifier/ definiteness slot” corresponds

to SPEC(QN), while his “second quantifier slot” corresponds to the head position

QN. I thus conclude that QN is an independently justified unique functional head above N, analogous to I above V. The arguments for the structure (10) are more than extensive; they are overwhelming. Here are five further considerations which favor Q as a single functional head outside NP in extended nominal projections.

• In Abney’s (1987) original cross-linguistic arguments for a functional head above N, number agreement plays a central role. Since Q is the locus of ±PLUR, on this score alone Q is a more satisfying candidate than D for the head of extended NPs. • Taking Q as the head of extended NPs strengthens the parallels in Abney (1987 Ch. 4) between modifiers of As and Ns. He argues that degree words DEG (more, as, too etc.) are functional heads of APs, whose SPECs can be measure phrase NPs. Now since degree words indicate quantity, they are more semantically parallel with Q than D.

• Japanese now conforms to UG in having a (final) functional head NQ above N, though this extended projection is optional in Japanese. • English no longer has unexplained complementary distribution between possessive phrases and the functional head of extended NPs; rather we

observe them together: [SPEC(QP) John’s ] [Q three ] houses, [SPEC(QP) today´s ] [Q many ] lectures, etc. • Cross-linguistically it is no longer surprising that demonstratives and definite articles are often declined and/or ordered left-to-right like AP modifiers of N; in languages where this happens, that is precisely what they are.

A final advantage of structure (10) is that it makes plausible the following conjecture that relates syntax and reference:

(17) The locus of independent reference. All and only phrasal projections of

nouns (NP and NPQ) have independent reference.

The existential quantifier in a (non-generic) NPQ makes its “actual reference” different from the “virtual reference” of the plain NP it contains, as argued in

Milner (1978: chapter 1). In [ manyi/ fivei/ plentyi/ dozensi of [ young boys]j .],

the reference of the contained plain NP and the containing extended NPQ are

not the same. In contrast, a universal quantifier in an NPQ never changes the reference of the plain NP: (both) those toys, (any) three toys, (all) my child’s toys, etc.

2 8 Invited Papers Joseph Emonds

3. ±PLURAL as the Principal Feature Value of QN

Let’s review now the general structure of English NPs in terms of van Riemsdijk’s CIT (3), my extension of it (5) and the position of Specifiers (16). Since the main function

of QN is for recursive counting, I take its most basic interpretation in LF to be +PLURAL.

(18) [NP,Q SPEC(QN) (=D) [ QN, ±PLURAL ] [NP … N1 … N2 … Nk …(YP)…] ]

In this structure:

(19) (i) Nk is the open class lexical head;

(ii) any preceding Ni are closed class n such as couple, bunch and other;

(iii) QN is the unique and obligatory functional head of the extended NPQ;

(iv) (only) the exterior NPQ cannot further project (it is closed); and

(v) the functional head QN of this larger NP precedes its sister NP by the head-initial parameter of English, but follows D by principle (16).

As noted earlier in (8), English possessive nominals are in complementary distribution with the definite article and demonstratives, as well as with most D quantifiers {some, any, no, each, every, which, what}.10 I treat all these items as SPEC in schema (18), even though among them only possessives are overtly phrasal. This grouping corresponds to the “first SPEC(Nj) position” in Jackendoff’s nominal structures, which also accounted for this same complementary

distribution. I notate this frequently phrasal position as SPEC(QN). In the theory here, the SPEC position can occur only in the presence of Q (across categories). If Q is not present, no initial SPEC, phrasal or non-phrasal, is available either.

A salient English paradigm that confirms the obligatory nature of NQ (19iii) is that count nouns cannot appear “bare,” i.e. with no realization of either Q or D.

(20) *Soon book will be cheap. *Large house was for sale.

I propose to explain this by applying to (18) the idea of Chomsky (2001) that grammatical features are “unvalued” at the outset of a syntactic derivation, and then must receive interpretable values during a syntactic derivation. From this

perspective, we can reconceptualize +PLURAL in (18) as the LF values of QN, and

thereby actually eliminate PLURAL as a separate feature. That is, [QN, ±PLURAL]

10 This complementarity does not hold in many languages whose Ns project to NQ., including Czech. This study does not analyze this discrepancy.

Invited Papers 2 9 Major Trends in Theoretical and Applied Linguistics

is to be replaced by ±QN, i.e. QN receives a ± value from any lexical numeral or quantifier inserted under it, as follows. When a lexical N is a count noun, lexical

singular QN such as a(n) and one provide the value –QN, while all other lexical QN

(two, many, several, etc.) becomes +QN. A third possibility is that no morpheme

is inserted directly under QN. Then, if nothing else happens, this QN remains unvalued and the derivation is ill-formed (“crashes”) at LF:

However, another means of valuing a covert English QN with count nouns is by “Alternative Realization,” a widely applicable syntactic device for closed class items whose uses and restrictions are outlined in Emonds (2000: Ch. 4).

(21) Alternative Realization (AR). A syntactic feature F canonically associated in UG with category B can be alternatively realized in a closed class morpheme under B*, provided that projections of B and B* are sisters.11

In these terms the traditionally written +PLURAL is simply the positively valued

canonical feature QN. If a head N of QN’s sister NP contains a plural suffix, it has the

form [N N - +QN]. AR then applies with B = QN in canonical position and B* = N. That 0. is, QN is valued and because the plural morpheme alternatively realizes it under N

QN can remain covert in this configuration because AR operates in tandem with an “Invisible Category Principle,” which licenses empty categories (Emonds 2000: Ch. 4).

(22) Invisible Category Principle (ICP). If all marked canonical features F on B are alternatively realized by AR, then B may be empty.

Thus, if QN has no other marked features, i.e. is neither an existential quantifier

nor a numeral, the plural suffix on N is enough to permit QN to be empty: Soon books will be cheap; Large houses were for sale.

There is moreover a second way that AR and the ICP can value a covert QN. A

SPEC morpheme generally agrees in number with its Q, so that an overt SPEC(QN)

also alternatively realizes ±QN (= ±PLURAL). Since these SPECs are sisters of QN,

they can also license an empty ±QN in its base position: This [Q Ø] book was cheap; 12 Each [Q Ø] large house was for sale.

11 Throughout, one possible projection of a node is simply the node itself.

12 English mass nouns do not require an overt NQ or D. We might account for this by simply identifying the descriptive label “mass noun” with an alternative realization of –QN as a lexical feature on mass nouns. This move would involve extending AR to marked subsets of open class items. I leave for future research whether one can do this in a formally restricted way. French mass nouns behave more as this study’s framework expects, in that

they must appear with an overt singular –QN, namely a singular partitive article du/ de la.

3 0 Invited Papers Joseph Emonds

4. The Q Parameter: Obligatory Q-extended Projections in English

This previous section’s account of excluded English bare count nouns (20) is

based on assuming that all English noun phrases must project to NPQ, with a

functional head QN that must be valued as (±PLURAL) in LF. Since NPs include those with mass noun heads, gerunds and complex event nominals headed

by –ing (Grimshaw 1990), these heads must be specified as –QN. This forced

projection of NP to NPQ is a language-specific property, formulated here in a way similar to an earlier proposal of Fukui and Speas (1986):

(23) Q-Parameter. Maximal NP (=N1) in English must be closed by a Merge with

a QN head. NPs in Japanese need not be closed by merging with QN.

The Japanese setting of this Parameter is motivated by the fact that all its

open class nouns can be in bare NPs, i.e. its plain NPs need not project to NPQ. A further condition, which remains a stipulation here, applies to phrases in SPEC:

(24) SPEC Categories. Phrasal categories in SPEC(Q) positions must be nominal, i.e. Nj.

Since the category QN can receive its LF feature values from either the lexical item

it houses or (by AR) from the head of its sister phrase NP, material in the SPEC(QN)

position need not interact with QN. Consequently, as many studies remark, a “genitive”

NP in SPEC(QN) can stand in any pragmatic or argument relation to the head of NP. In

particular, if the definition of a subject of a phrase X’ picks out the lowest NPQ (= “DP”)

which c-commands X’, then a possessive noun phrase in the SPEC(QN) position can even be the subject/ external argument of any lexical head X0 of NP. Now according to the Q-Parameter (23), Japanese NPs can and most often

do lack a QN sister to NP; its NPs need not be “closed.” As a result, such NPs

have no SPEC(QN ) position. At the same time, since Japanese NPs are “open” and head-initial, a head NP can merge (repeatedly) with adjoined non-head NPs on its left, which can then satisfy the definition of subject/ external argument or a possessor for an N head. In fact, as is well known, several NP+no, not contained in each other, can modify a single Japanese N.

(25) Japanese NP with multiple adjoined internal subjects/ possessors:

a. [NP Daijobu – no [NP Taro - no [NP Kobe - no [NP shimbun ] ] ] ] Saturday’s Taro’s Kobe’s newspaper ‘Taro’s Kobe newspaper of Saturday’

b. [NP NPposs – no [NP NPposs - no [NP NPposs - no [NP …(YP)… - Nk - … - N1 ] ] ] ]

Invited Papers 3 1 Major Trends in Theoretical and Applied Linguistics

Notice that these multiple possessors are quite unlike the recursive possessive

NPs in English. In Japanese, each NPposs directly modifies the highest head N, whereas in English, a first possessive N must modify the next (as inJohn’s father’s newspaper’s headlines) rather than the highest head N. Since these exterior NPs are not in any structural relation with a functional

head QN (in this respect there is no difference from English), any of them can either serve as a subject or take on any thematic or pragmatically sanctioned role relative to the lexical N head of NP.

5. Q in the Context ___AP

5.1. Degree Words and Measure Phrases

Bresnan (1973) and Jackendoff (1977: Ch. 6) isolate a class of largely mutually exclusive adjectival modifiers, often called degree words (DEG). I propose that 13 this class instantiates Q in the context ___AP and so should be notated QA.

(26) QA = very, so, quite, rather, somewhat, this, that, more, most, less, least, as, too, how.

Since multiple members of QA generally cannot co-occur, as seen in (27), it

appears that the underlined QA must select APs lacking Q. That is, just like QN, a

single QA functions to close AP projections.

(27) a. These chairs are how old?/ so old. *These chairs are how so/ so how old? b. We want a less/ quite bright room. *We want a less quite/ quite less bright room. c. Is she rather/ that clever? *Is she rather that/ that rather clever? d. We consider John very/ too arrogant. *We consider John very too / too very arrogant.

Since adjectives are “properties” rather than “things,” a QA as in (26) can’t measure quantity with integers, but only in terms of stronger, weaker, equal or

13 Another candidate for QA is enough, which in Germanic languages surfaces after A.

3 2 Invited Papers Joseph Emonds

deictic degrees. But the counting potential of QA emerges clearly with more, less,

as, that and too. These QA license measure phrase NPs in the context ___QA-AP (Neeleman, van de Koot and Doetjes 2004).

(28) [AP [NP three times/ a bit ] [Q more/ less ] [AP [A clever] [YP in math] [ZP than you] ] ]

[AP [NP two days/ a good deal ] [Q too ] [AP [A short ] ] ]

[AP [NP three times ] [Q as/ that ] [AP [A far/ long/ old/ expensive ] ] ]

Adjective phrases apparently conform perfectly to the earlier statements (16) and (24):

(16) Specifier Position. A functional head QX licenses a SPEC(QX) on its left, independently of a language’s word order.

(24) SPEC Categories. Phrasal categories in SPEC(Q) positions must be nominal, i.e. Nj.

In the light of a second use of NPs in SPEC(Q) as measure phrases, a possible explanation of (24) may follow from a relation between quantities expressed in

Q and their “measure” in an NPQ in SPEC. That is, SPEC’s fundamental role is to

further specify number and/or quantity, which is a characteristic meaning of NP with a Q head.

A tree for an English quantified (measure) AP is thus as in (29). As with QN, the

structure is flat, as there is no motivation for grouping AQ with AP; both AP and QA

project as features to a closed extended projection APQ. And as with NPQ, I claim that no further functional head is needed for APs, again in conformity with the Q-extended CIT (5).

(29) APQ = [ A, Q ]’

SPEC(QA)=NPQ AP

QA A YP

three times more/ as/ too/ that clever in math

The structure (29) replicates the structure inside English NPs; compare (29)

with (10). Here, however, the sole LF role of the (again optional) NP in SPEC(QA)

is to associate certain QA with some discrete, counted measure, which inherent

features of QA in the context ___AP can’t provide. The difference between the two subtypes of Q categories is that the measure for discrete nouns is inherent

in QN’s own content, i.e. the numerals, existential quantifiers, and +PLURAL. In

contrast, a discrete “measure” for QA is external to it, in SPEC(QA).

Invited Papers 3 3 Major Trends in Theoretical and Applied Linguistics

Returning briefly to NPs, there is in fact a little noticed complementary distribution between subject phrases and measure phrases, which testifies to

their identical single structural position SPEC(QN).

(30) a. My mother didn’t like preparing for my father’s (one) vacation. My mother didn’t like preparing for several days more vacation. *My mother didn’t like preparing for my father’s several days more vacation. b. This grant provides two hundred dollars more salary every month. This grant provides that assistant’s salary every month. *This grant provides that assistant’s two hundred dollars more salary every month.

It is only because QN needs no external specification that SPEC(QN) is free to house NPs with any pragmatic relation to the head N, the notoriously varied

semantics of “possessive” NPs. The NPs in SPEC(QA) have no such freedom; they can serve only as “measure phrases.” Previous analyses have failed to identify measure phrases inside APs with possessive NPs inside NPs, even though in English both types must be unique, and both must be NPs; see again (30). Thus, the grammatical source of the much studied possessives is in “less frequent” measure phrases, which are in turn nothing but an extension of the primitive linguistic ability to count.14

5.2. Measure Phrases Without Degree Words

A small closed class of English adjectives (long, high, tall, deep, wide, old, early,

late, square) allow measure NPs in SPEC(QA) even in the absence of an overt QA.

(31) These chairs are ten years [Q Ø ] { old/ *obsolete/*creaky }.

The path seemed many miles [Q Ø ] { long/ *lengthy/ *rocky}.

His hedge got three feet [Q Ø ] { wide/ *broad/ *overgrown } last year.

These NP, naturally enough, cannot occur with any overt Qs that disallow measure phrases.

14 In general, less frequent grammatical variants of a construction reveal more than more frequent variants. Along such lines, less frequent dependent clauses better indicate underlying word order than main clauses; negated sentences reveal more about deep grammar than positive clauses, etc.

3 4 Invited Papers Joseph Emonds

(32) *These chairs are ten years [Q very ] old.

*The path seemed many miles [Q so ] long.

*His hedge got two meters [Q somewhat ] wide.

Since these adjectives form a closed class, I postulate a syntactic feature Fm

common to those QA (more, as, too, etc.) that permit measure phrases in SPEC(QA); the As in (31) then alternatively realize this feature. As a result, their (English)

lexical entries and the ICP (22) together allow their QA to be empty. Essentially, Fm = “compatible with discrete measures.”15

In summary, NPs in a SPEC(QA) position quantify properties expressed in APs

as greater or less, or as excessive or not. Only certain overt QA permit these phrases, even though they are also permitted by a few head adjectives in English

that license an empty QA.

6. Q in the Context ___PP

Consider PPs of space and time, whose P express these notions. Since one can’t “count” a spatial or temporal span without discrete units of measure, English

“intensifiers” of P such as the overt QP right are incompatible with any measure 16 phrase in SPEC(QP).

(33) John put his books [PP(Q) (*six inches) [Q right ] [PP behind the door ] ].

The doorbell rang [PP(Q) (*a few seconds) [Q right ] [PP after six ] ].

It was [PP(Q) (a few seconds)[Q right ] [PP after noon ] ] that they arrived.

Jim kicked the ball [PP(Q) (*30 meters) [Q clear ] [PP across the field ] ].

You’ll find some restaurants PP(Q)[ (*a few blocks) [Q straight ] [PP down this road ] ].

Just as in the contexts __AP, a preceding NP provides Q with a discretely measured value for many P: before, after, above, below, behind, inside, away, back, etc. Yet again because these P still form a closed class, they are susceptible to AR (21). Like the English As that license measure phrases (31), these P apparently m also alternatively realize the syntactic F of a QP that licenses such phrases, as in

(34). Then as a result of the ICP (22), QP is empty.

15 A language-particular treatment of (32) seems appropriate, as their exact French translations are ungrammatical: *Ces chaises sont dix ans vieilles; *Le sentier semblait plusieurs kilomètres long. 16 Jackendoff (1977: Ch. 6) notes that measure phrases in these PPs don’t occur with right.

Invited Papers 3 5 Major Trends in Theoretical and Applied Linguistics

m (34) John put his books [PP [NP, +PL three feet ] [Q, F Ø ] [PP behind the door ] ]. m Her ball landed [PP [NP, +PL a few paces ] [Q, F Ø ] [PP (away) from mine ] ]. m It was [PP [NP, -PL an hour ] [Q, F Ø ] [PP after midnight ] ] that they arrived.

Most English adjectives disallow a combination of an empty QA and a measure phrase, as seen in (31); so also many Ps are incompatible with the configuration in (34).

(35) John put his books [PP (*a few inches) at the door ].

Her ball landed [PP (*two steps) ] with mine.

It was [PP (*an hour) until the party] that they were singing.

Thus, the use of SPEC for measure phrases in PPs parallels that in APs. What

differentiates the two is that no overt English QP take a measure phrase, unlike

QA such as more, as and too.

7. QV in English Clauses: Where “Subject Agreement” Comes from

7.1. The Parallel Structure of English Noun Phrases and Clauses

Suppose by parsimony that English clause structure (36) parallels that of NP as in (18).

(18) [NP,Q SPEC(QN) (=D) [ QN, ±PLURAL ] [NP … N1 … N2 … Nk …(YP)…] ]

(36) [VP,Q SPEC(QV) [QV, ±PLURAL ] [VP … V1 … V2 … Vk …(YP)…] ]

The bolded VPQ, SPEC(QV) and QV correspond respectively to what Chomsky (1986) calls IP, SPEC(IP) and I. So let’s partly rewrite (36) with more familiar symbols as (37), though if the parallel in (18) and (36) is “real,” these special symbols should play no formal role.

(37) Clause structure: [IP SPEC(IP) [I QV, +PL ] [VP … V1 … V2 … Vk … (YP)…] ]

As with the NP structure, there are no empirical reasons for grouping together

QV (= I) + VP as a constituent I’. The only justification ever given for such an I’ is based on parenthetical adverbials after a subject NP:

3 6 Invited Papers Joseph Emonds

(38) Mary, within a month, should enroll for school. This process, I’ve learned, is a new way to make ice cream. Smoking upstairs, to my knowledge, doesn’t bother Bill much.

Is there any alternative to an I’ constituent for the post-parenthetical sequences in (38)? In fact, it appears that subjects in SPEC(IP) come to precede these parentheticals by raising leftward around them, apparently to a focus position, as in (39).

(39) Maryi, within a month, [IP ti should enroll for school].

This processi, I’ve learned, [IP ti is a new way to make ice cream].

Smoking upstairsi, to my knowlsege, [IP ti doesn’t bother Bill much].

Moreover, we know independently that expletive subjects cannot move into focus position (i.e. serve as new information), as seen in (40a). So if expletive subjects replace the full NP subjects in (39), the results are equally ungrammatical (40b). If follows that the pre-parenthetical NPs are in a focus position, outside of IP, and so cannot be used to argue for the existence of an I’ separate from IP.

(40) a. *Therei Bill believed ti to be no reason for a meeting.

*Iti Sue didn’t think ti bothered Bill much to smoke upstairs.

b. *Therei, I’ve learned, ti is a new way to make ice cream.

*Iti, to my knowledge, ti doesn’t bother Bill much to smoke upstairs.

Since an analysis with I’ is unable to account for examples like (40b), we are free to retain the structure in (37) in which I and VP do not form a constituent.

Let’s now see how the QPV structure for clauses relates to a description of Japanese. I introduced in Section 4 a Q-Parameter (23), according to which Japanese NPs need not be “closed” by a Merge with Q. If we extend (23) to Japanese and English clauses, it then follows that Japanese VPs can be “bare,” i.e. not project to an IP. That is, the structure (37) is not obligatory for Japanese clauses.

(41) Generalized Q-Parameter. Maximal NP and VP in English must be closed by a Merge with a Q head. Japanese NPs and VPs need not be closed by a Merge with Q.

This formulation is in fact formally equivalent to the central parameter distinguishing English and Japanese made explicit in the title of Kuroda (1992): “Whether We Agree or Not: A Comparative Syntax of English and Japanese.” However, though he discusses many insightful ramifications of his hypothesis, he does not extend his parameter to the structure of nominal phases, as under the QP Hypothesis; he treats only differences in clausal structure.

Invited Papers 3 7 Major Trends in Theoretical and Applied Linguistics

I do however have reservations about the way Kuroda uses “optional agreement” to analyze Japanese case alternations. These differences go well beyond the scope of this paper. In short, my view is rather that, since Japanese does not need to project its VPs to IPs, it is more economical not to, and so perhaps it never does; in this case there simply are no IP structures in Japanese. Its finite clauses are then actually traditionally termed VPs with subjects in (possibly multiple) adjoined positions, as in Fukui and Speas (1986).17

7.2. Valuing and Interpreting QV (=I) in Syntax

The inherent features of QV (=I) in (37) are those of tense and modals. This yields an English clausal structure as in (42). This tree is the familiar structure of finite clauses, but replaces terms such as I, INFL and Tense with the general

and (I propose) only functional category Q modifying VP. Vi represents possible

grammatical verbs v such as be, have, get, go and come, while Vk is the open class lexical head.

(42) VPQ ( = IP )

NPQ, ±PL QV [±PAST, ±MODAL] VP

Vi … Vk … (YP)

All English IPs that are finite have the obligatorily overt structural subject NP shown in (42); their head I either is a Modal or agrees in number with this 18 subject. This specification for number suggests thatV Q (= I) is in fact a sort

of “default quantification” over V, in that it provides VQ with ±PLURAL values in 19 case lexical members of this category, e.g. Modals, are absent. Just as with QN (modifying count nouns) that are unvalued by a lexical numeral or quantifier,

QV can receive its value by Alternative Realization, whose definition I repeat for convenience.

17 Since Japanese subjects are adjoined to VP rather than located in SPEC(QV), they can sometimes be PPs, with the Ps de ‘at’ or kara ‘from’, an analysis argued for on independent grounds in Inoue (1998). 18 More accurately, subjects of a finite verb must be overt or a trace of a subject fronted to a clause’s left periphery. 19 Gerunds and participles lack both Modals and agreement because they are not IPs to start with: participles have A heads (Emonds 2000: chs. 5 and 7), and gerunds have N heads (section 4.7); nor do “bare VPs” in causative constructions project to separate IPs (ch. 6). For reasons of space, this study cannot analyze the lack of agreement on infinitives.

3 8 Invited Papers Joseph Emonds

Here F is Q, B is the QN head of a subject phrase and B* = QV. Formally, QV must

be valued in a well-formed derivation that leads to an interpretable LF. Like QN, this value, which includes but is not limited to ±PLURAL, can be provided by

a lexical item in QV, i.e. a Modal. But when QV dominates no such item, it can alternatively realize ±PLURAL located on one of Q’s sisters in (42), either VP or

SPEC(QV). Since VP has no Q feature, the only possible source for valuing QV is

the ±PLRUAL of a subject NPQ. In more familiar terms, the category I must agree in number with an NP in SPEC(IP). This analysis derives from Chomsky’s (2001) conception of using syntactic derivations to value features, and thus implies that number agreement plays a role at LF. This conception overturns a long-standing assumption that English subject-verb agreement is “meaningless,” i.e. adds nothing to the simple specification of NPs as singular or plural. It also departs from my own previous working assumption, namely that alternatively realized features contribute to LF only by licensing features in their canonical positions.20 In addition to these matters, a reader might hesitate to relate “plural verbs” so closely to the rather more concrete counting system of numerals. Nonetheless, though syntactic categories invariably have a concrete cognitive basis, they are often used to express concepts not included in these original bases. For example, the category N is certainly based on naming material objects. Yet open class items such as flaw, vacuum, ubiquity, ether, immortality, existence don’t refer to observable or even material entities. They are “things” only circularly, in that they are grammatically Ns. Similarly, though P’s basic function is to locate in space and time, “marked’ P” like of, without, despite, most uses of for etc. don’t do this. It is typical of natural language to formally extend use of a syntactic category beyond its cognitive basis. In this sense, the category I

(= QV) simply extends counting and quantification into verbal domains. What then can be the semantics of verbs being “quantified” as ±PLUR? Traditional grammar remarks only that a plural verb doesn’t mean a plurality of successive events. That is, any predication, in English at least, is true if its verb holds of a subject at a given time, namely that of the verb’s Tense. However,

20 There are other constructions where alternatively realized features can make independent, if secondary, contributions to LF. In work in progress, I argue that LF representatiions of certain complex Tenses such such as English perfects must use together two values of Tense in one clause, one in its canonical (V) and one in its alternatively realized (I) position.

Invited Papers 3 9 Major Trends in Theoretical and Applied Linguistics

+PLUR on V does imply a plurality of simultaneous events/ states: those with

different subjects. That is, the unmarked interpretation of the agreement on VQ is a counting of simultaneous events (or states).21

(43) The boys were eating ice cream. (several “eatings”) The boys resemble their father. (several “resemblances”)

In this section’s analysis, number agreement with the subject NPQ (the AR of

the latter’s number feature) is a default means for valuing QV; what is obligatory

is not agreement itself but the valuing of QV as ±PLUR. This leaves open the

possibility that both [QV, +PLUR] and [QV, -PLUR] might be specified independently of the value of PLUR on the subject phrase. Such marked constructions indeed exist, and support the analysis here over a more traditional variant in which number agreement is simply obligatory. (i) In British English, when a subject is a collective noun (government, army,

team), QV can be independently specified as +PLUR, which seems to mean that the members of the group act severally but in concert.

(44) a. This government is/ *are known for its austerity program. (Normal agreement) b. The government are planning reforms. (The government is a group acting together)

(ii) QV can have an independent singular form, which then imposes on the NP in SPEC an interpretation as a single event, regardless of the latter’s inherent number.

(45) a. Normal agreements: Too many boys make a bad party. Sienna’s neighborhood flags waving in the wind were a colorful sight. Being late and not being apologetic are not considered polite.

b. When QV is inherently marked –PLUR, the subject NP is taken as a single event: Too many boys makes a bad party. Sienna’s neighborhood flags waving in the wind was a colorful sight. Being late and not being apologetic is not considered polite.

21 With “symmetric predicates” (we married; the boys met), plural “simultaneous events” are indispensable, and so might be pragmatically viewed as one event. But even here, two people marrying each obtain new legal status, so two legal events must have transpired.

4 0 Invited Papers Joseph Emonds

Of course, many and perhaps most plural NPs are simply incompatible with “single event” interpretations, as seen in (46).

(46) Three severe storms were/ *was due to global warming. Student answers on this test have/ *has provided entertainment.

Again, mechanical agreement itself is not obligatory. What is obligatory prior

to LF is only the valuing of QV as ±PLUR, which in some contexts such as (44b) and (45b) can occur independently and not as a default. As noted above, these extensions of a feature ±PLUR, beyond its original

cognitive basis with N, are typical in formal syntax. Even though QV is not used for quantifying temporal duration of an event or state, nor for counting their

repetitions, a QV expressed as agreement nonetheless does actually seem to count simultaneous events.

Recall finally that in the closed projections PPQ and APQ, the interpretation of Q can be further specified by a preceding measure phrase NP. We can extend

this condition to VPQ:

(47) Valuing Q. Whenever QX lacks inherently specified numeric features, i.e.,

when Q ≠ QN, it can receive a default LF interpretation by being specified

for quantity by an NP in SPEC(QX).

That is, a subject NP of an agreeing verb acts structurally as a measure phrase

that supplies a quantitative interpretation of QV, and thus satisfies a general requirement in Chomsky (2001) that syntactic derivations must value features.22 This study’s approach to functional categories has thus predicted number agreement of finite verbs with subjects for any languages which have the English setting for the Generalized Q-Parameter (41), whose NPs and VPs must Merge with Q. (41) moreover reveals why subject agreement is so central in syntax; it

signals that a closed VPQ rather than an open VP is structurally present.

22 This conclusion sheds some light on a puzzling asymmetry in Jackendoff (1977: Ch. 6). English measure phrases optionally precede all open class heads except Vs. In order to quantify an activity of a V over time, one must use a post-head adverbial phrase: a. *She may several hours talk about it. She may talk about it (for) several hours. b. *We two miles followed that car. We followed that car (for) two miles.

Now the Q-extended CIT (5) in fact is compatible with an English I (= QV) specifying some measure. But this measure apparently counts only simultaneous events or states specified

by the predication NP+VP, i.e. SPEC(QV)+VP. Consequently, there is no way for QV or SPEC(QV) to indicate any other kind of measure for V, either over time or space.

Invited Papers 4 1 Major Trends in Theoretical and Applied Linguistics

8. How Many Categories are there in Syntax?

This paper has widened the use of QP, via the Q-extended CIT (5), to English APs, PPs and VPs. In particular, measure phrase NPs have turned out to be counterparts in APs and PPs to subject NPs in IP and possessive NPs inside larger NPs (sections 5-7). Though the inclusion of VP projections under (5) in section 7 is far from obvious, it allows the Q-extended CIT to subsume an ingenious idea of Kuroda (1992), whereby the functional head I above VP is crucially identified with subject-verb agreement in English and an absence of agreement in Japanese. That is, agreement’s crucial component is the ±PLUR number on I, i.e. the syntax-

assigned values ±QN. In this perspective, English IPs should be considered to be

VPQ, which Japanese then lacks, as argued in both Fukui and Speas (1986) and Kuroda (1992). To express this difference, I have generalized a Q Parameter (23) for noun phrases to verb phrases as in (41):

(41) Generalized Q-Parameter. Maximal NP and VP in English must be closed by a Merge with a Q head. Japanese NPs and VPs need not be closed by a Merge with Q.

The basis of the Q-extended CIT (5) is that in English a QV must be valued in LF, and as a default quantified, even though a predicate V or VP cannot be. Unlike j j in other projections X , neither V itself nor a measure phrase in SPEC(QV) can

separately provide a value to QV. The only way an unvalued feature QV can satisfy the requirement that all LF features be valued is via a constituent whose Q is

already valued, i.e. by agreement with the ±Q on an NPQ in SPEC(QV). These NPQ of course structurally correspond to the familiar subject NPs in SPEC(IP). An

agreeing I thus turns out to be nothing other than a QV formally receiving its

value from a QN in subject position. All functional categories that are not lexical categories in disguise (i.e., functional categories of “small x” for the x = n, v, a, p which obey van Riemdijk’s CIT) thus reduce to a single functional head Q. And in light of the following additional considerations, there is no need for a significantly larger category inventory in syntax than that just reviewed.

(i) What are usually called D or DET are single words dominated by SPEC(QN). (ii) The only productive category of adverbs are heads that are of category A. (iii) C (= COMP) reduces to P (Emonds 1985: Ch. 7).

This reduced set of head categories, namely {N, V, A, P, Q}, recalls the categorical parsimony of generative semantics, whose advocates wished to reduce the set of syntactic categories to a small group of basic categories of

4 2 Invited Papers Joseph Emonds

logic. In my view, they rightly claimed that syntax needs only a reduced set of categories, comparable to those in some kind of “natural language logic,” i.e. what is called today LF. However, generative semantics prematurely substituted categories found in modern symbolic logic for those of empirically justified Logical Forms for natural language. Consequently, this approach emphasized items expressing predicates (V), reference (N) and quantification (Q). But since place and time are extraneous in symbolic logic, it wrongly ignore the critical LF roles of PP structures. Since symbolic logic was nothing but Bertrand Russell’s simplified, intuited version of LF, it is a circular exercise to hypothesize a natural language LF dependent on symbolic logic. Rather, natural language logic and its categories must be newly discovered on the basis of syntactic research, using the method (Chomsky 1957) of contrasting acceptabilities for similar syntactic sequences. We then find that natural languages distinguish (do not conflate) four kinds of categories N, A, V and P, which both take arguments (a property of symbolic logic predicates) and at the same time can all be constants and variables in larger propositions. These are supplemented by a single category Q which is first and foremost used to count Ns, and second to existentially quantify them, and then to measure properties (A) and locations and times (P). Finally, the role of Q in V projections, as a source and carrier of agreement, becomes almost totally formal.23

23 From an evolutionary perspective, this parsimonious scenario greatly improves on systems that proliferate functional categories.

Invited Papers 4 3 Major Trends in Theoretical and Applied Linguistics

References

Abney, S. 1987. The English Noun Phrase in its Sentential Aspect. MIT doctoral dissertation.

Bresnan, J. 1973. The Syntax of the Comparative Clause in English. Linguistic Inquiry 28, 119-164.

Cardinaletti, A. and G. Guisti. 2006. The Syntax of Quantified Phrases and Quantitative Clitics. In M. Everaert and H. van Riemsdijk (eds.), The Blackwell Companion to Syntax, vol. V. London: Blackwell, 23-93.

Chomsky, N. 1957. Syntactic Structures. The Hague: Mouton.

Chomsky, N. 1986. Barriers. Cambridge, MA: MIT Press.

Chomsky, N. 2001. Derivation by Phase. In Michael Kenstowicz (ed.), Ken Hale: a Life in Language Cambridge, MA: MIT Press, 1-60.

Corver, N. and H. van Riemsdijk. 2001. Semi-Lexical Categories: The Function of Content Words and the Content of Function Words. Berlin: Mouton de Gruyter.

Emonds, J. 1985. A Unified Theory of Syntactic Categories. Dordrecht: Foris.

Emonds, J. 2000. Lexicon and Grammar: the English Syntacticon. Berlin: Mouton de Gruyter.

Emonds, J. 2001. The Flat Structure Economy of Semi-Lexical Heads. In N. Corver and H. van Riemsdijk (eds.), Semi-Lexical Categories: The Function of Content Words and the Content of Function Words. Berlin: Mouton de Gruyter.

4 4 Invited Papers Joseph Emonds

Emonds, J. 2007. Q: the only Functional Head above N and A. Theoretical and Applied Linguistics at Kobe Shoin 10.

Emonds, J. 2008. Q: the one and only Functional Head. In R. Kawashima, G. Philippe and T. Sowley (eds.), Phantom Sentences: Essays in linguistics and literature presented to Ann Banfield. Peter Lang: Berne, 193-226.

Emonds, J. 2012. Which is the Extended Projection above N? In L. Brugè, A. Cardinaletti, G. Giusti, N. Munaro and C. Poletto (eds.), Functional Heads. Oxford: Oxford University Press.

Fukui, N. and M. Speas. 1986. Specifiers and Projection. MIT Working Papers in Linguistics 8, 128-172.

Giusti, G. 1991. The Categorial Status of Quantified Nominals. Linguistische Berichte 136, 438-454.

Grimshaw, J. 1990. Argument Structure. Cambridge, MA: MIT Press.

Hockett, C. 1960. The Origin of Speech. Scientific American203, 88-111.

Inoue, K. 1998. Sentences without Nominative Subjects in Japanese. Researching and Verifying an Advanced Theory of Human Language 2-A, Kanda University of International Studies report, 1-30.

Jackendoff, R. 1977. X-bar Syntax. Cambridge, MA: MIT Press.

Kubo, M. 1996. Some Considerations on Noun Clauses and Numeral Classifiers: A Study of (Pseudo)partitives in Japanese and English. Keio Studies in Theoretical Linguistics 1, 89-124.

Kuroda, S.-Y. 1992. Whether We Agree or Not: A Comparative Syntax of English and Japanese. Japanese Syntax and Semantics: Collected Papers. Dordrecht: Kluwer Academic Publishers.

Lieber, R. 1980. On the Organization of the Lexicon. MIT doctoral dissertation.

Milner, J.-C. 1978. De la syntaxe à l’interprétation. Le Seuil, Paris.

Neeleman, A., H. van de Koot and J. Doetjes. 2004. Degree Expressions. Linguistic Review 21.

Invited Papers 4 5 Major Trends in Theoretical and Applied Linguistics

Oga, K. 2002. The Syntax of Japanese Nominal Projections and some Cross- Linguistic Implications. University of Durham doctoral dissertation.

Okuda, M. 2006. Japanese Classifier Agreement and the DP–hypothesis. Kobe- Shoin University MA dissertation.

Peano, G. 1889. Arithmetices principia, nova methodo exposita. Torino: Bocca.

Ritter, E. 1991. Two functional categories in noun phrases: evidence from Modern Hebrew. In S. Rothstein (ed.), Perspectives on Phrase Structure: Heads and Licensing. San Diego: Academic Press, 37-62.

Van Riemsdijk, H. 1998. Categorial Feature Magnetism, the Endocentricity and Distribution of Projections. Journal of Comparative Germanic Linguistics 2, 1-48.

Veselovská, L. 2001. Agreement Patterns of Czech Group Nouns and Quantifiers. In N. Corver and H. van Riemsdijk (eds.), Semi-Lexical Categories: The Function of Content Words and the Content of Function Words. Berlin: Mouton de Gruyter.

4 6 Invited Papers Growth and Decline: How Grammar Has Been Changing in Recent English

Geoffrey Leech

Lancaster University [email protected]

Abstract

Grammatical change, as opposed to (for example) phonetic or lexical change, is generally felt to be a slow process, imperceptible to the native speaker. There is some truth in this judgement, and yet changes in English grammatical usage during the recent past have been considerable. The evidence for this comes from comparable corpora or subcorpora systematically sampled from the language at regular intervals of time. It has now become possible to • observe changes in progress in written English • compare rates of change from one genre to another • compare rates of change across the two regional varieties: American English (AmE) and British English (BrE)

1. The Brown Family of Corpora

The corpora I shall refer to, in demonstrating this, are known collectively as the Brown family of corpora. The name derives from the Brown Corpus, the earliest electronic corpus of the English language, collected at Brown University 1962-4, and since then widely distributed among researchers, teachers and students. The Brown Corpus consists of just over a million words, and is subdivided into 500 written (printed) text samples of approximately 2000 words each. Alongside the Brown Corpus, the Lancaster-Oslo-Bergen corpus was created as a British counterpart, to be as precisely comparable to the Brown Corpus as possible. It consists, like Brown, of 500 samples of written English, constructed according to the same corpus design, with subdivision into 15 genres, within which individual texts and groups of texts are matched. Like those of the American corpus, the text samples of the British corpus were selected by stratified randomized sampling from texts first published in 1961. It will be noticed that these corpora, like others subsequently created, were named after the universities at which they were compiled. Also, for convenience,

Invited Papers 4 7 Major Trends in Theoretical and Applied Linguistics

corpora created at more than one university have been informally identified by acronyms, such as “LOB corpus” for the “Lancaster-Oslo-Bergen corpus”. Considering the immense growth in computing power and text storage capacity since the 1960s and 1970s, the Brown and LOB corpora are nowadays regarded as somewhat small and as well as being limited to written English. Nevertheless, they are still valuable for research because of their precise comparability, and also because of their wide availability: researchers all around the world can obtain copies of these corpora, and can annotate them and analyse them for their own research purposes.

Table 1. The genres of the Brown family of corpora, and the number of text samples each contains.

Number Identifying Genres Subcorpora of text letter samples A Press: reportage 44

B Press: editorial Press 27

C Press: reviews 17

D Religion 17

E Skills and hobbies 36 General F Popular lore 48 Prose G Belles letters, biography, memoirs etc. 75

H Miscellaneous [mostly government documents] 30

J Learned [academic writing] Learned 80

K General fiction 29

L Mystery and detective fiction 24

M Science fiction 6 Fiction N Adventure and western fiction 29

P Romance and love story 29

R Humour 9

Total 500

Because the genre sections are relatively small, it has been found that more reliable comparisons can be made by grouping the genres into four subcorpora (Press, General Prose, Learned and Fiction), containing a set of similar genres, as shown in the third column of Table 1. Table 2 summarizes the contents of each corpus of the Brown family, referring to the subcorpora which are used in subsequent data analyses in this paper.

4 8 Invited Papers Geoffrey Leech

Table 2. The composition (in terms of major subcorpora) of the Brown Corpus and other members of the Brown family of corpora.

The Brown Family of Corpora Each 1-million-word corpus has 500 text samples allotted to 15 genres grouped into four subcorpora Press: Genres A, B and C c. 178,000 words

General Prose: Genres D, E, F, G and H c. 413,000 words

Learned/Academic: Genre J c.159,000 words

Fiction: Genres K, L, M, N, P and R c. 257,000 words

Each whole corpus is just over 1,000,000 words

Since the Brown and LOB Corpora were compiled, the Brown family has grown by the addition of the Freiburg-Brown Corpus and the Freiburg-Lancaster-Oslo- Bergen Corpus (Frown and FLOB for short), which again precisely match Brown and LOB, but with data sampled from publications in 1992 (American English) and 1991 (British English) respectively. In effect, these four corpora provide a means of tracking changes in usage over the period roughly of one generation, between the 1960s and the 1990s. Further, for the purpose of investigating recent change in standard written English, further comparable corpora have since been added, as shown in Table 3:

Table 3. The evidence of change: comparable corpora of American English (AmE) and British English (BrE) at roughly 30-year intervals.1

1901±3 1931±3 1961 1991, 1992 2006±2

British BLOB-1901 BLOB-1931 LOB FLOB BE06 English Corpus Corpus Corpus Corpus

American (no corpus B-Brown Brown Frown AmE06 English yet) Corpus Corpus Corpus

1 In the short names of the corpora, the initial “B” stands for “before”, and the initial “F” stands for “Freiburg”. Although it is clearly impossible to give due acknowledgement to all those who participated in the development of these corpora, the following list names those who took a leading role in the compilation of the corpora bracketed after them: Nelson Francis and Henry Kučera (Brown); Geoffrey Leech and Stig Johansson (LOB); Christian Mair and Marianne Hundt (Frown, FLOB); Nick Smith, Paul Rayson and Geoffrey Leech (BLOB-1931, BLOB-1901); B-Brown (Marianne Hundt), Paul Baker (BE06), Paul Baker, Amanda Potts and Matteo Di Cristoforo (AmE06).

Invited Papers 4 9 Major Trends in Theoretical and Applied Linguistics

On this diagram, all-white cells represent comparable corpora that are complete, and match in detail the original Brown Corpus, except for their date and regional provenance. The two shaded corpora on the left are at present incomplete. The BLOB-1901 corpus has been used in some of my analyses – but, because only a comparable but incomplete 1/3 of the corpus was available, the frequency data derived from that corpus had to be scaled up to be equivalent to the figures for the other corpora. The two lightly-shaded corpora on the right were collected from matching textual materials on the World Wide Web, and although in every respect they match the Brown family design, their web origin might conceivably have an influence on the style of language used, and hence of their comparability with the other corpora. (The reduced width of the rightmost column of Table 2 is a reminder that the most recent period represented by the Brown family, the period 1991-2006±2, is approximately only half as long as the 30-year periods separating the earlier corpora.) Obviously the results obtained from the shaded corpora have to be treated with some caution. The dates “1931±3” etc. indicate that for the more recently-collected corpora it was considered impractical to compile the corpus from data published in a single year (as had originally been done for the Brown, LOB, Frown and FLOB corpora). Instead, the three years preceding and following the target year were accepted as a window of time from which the corpus material could be collected. Thus “1931±3” for the BL0B-1931 and B-Brown corpora indicates that the window within which the text samples could be collected was 1928-1934 . It was found that the task of collecting exactly matching texts from a particular year is very difficult and time-consuming, particularly for the earlier periods of the Brown family, and it was decided that the change of grammatical usage within a period as short as seven years was likely to be fairly negligible.

2. How Frequency Data Reflect Change

To illustrate how data from the Brown family can trace changes in the language during the twentieth century, I take the straightforward example of the preposition upon, which has declined progressively and markedly in frequency between the 1901 and 2006 corpora in British English, so that today it is less than one-sixth as frequent as it was at the beginning of the twentieth century. Figure 1 shows the overall trend, measured in the number of occurrences of the item concerned (here upon) per million words. For the time being, we will accept this per-million-word measure (abbreviated to ‘pmw’) as the most convenient, though not necessarily the best, method of calculation. Figure 2 looks in more detail at the period 1931-1991, dividing the frequencies of the whole corpora into frequencies pmw for the four subcorpora Press, General Prose, Learned and Fiction.

5 0 Invited Papers

Geoffrey Leech

1400 1200 1000 800 600 400 200 0 |1901| |1931| |1961| |1991| |2006|

Figure 1. Declining frequency of the preposition upon in British written standard English 1901-2006

1,200

1,000

800 Press Gen Prose 600 Learned Fiction Overall 400

200

0 1931 1961 1991

Figure 2. Declining frequency of the preposition upon in British written standard English 1931- 1991, showing the four subcorpora Press, General Prose, Learned and Fiction

The main extra information we obtain from Figure 2 is that (a) the decline of upon is general across the four subcorpora, but (b) that it is steeper, especially in the period 1931-1961, in certain subcorpora than in others. Thus the steepest decline is found in the Press subcorpus, which is not surprising, given that journalistic writing is often in the vanguard of change. On the other hand, the Learned variety, representing the conservative style of academic writing, shows

Invited Papers 5 1 Major Trends in Theoretical and Applied Linguistics

the least change, so that by 1991 this subcorpus has ‘lagged behind’ and retains the highest frequency of upon.2 We may compare two methods of tracking change: the normalized frequency (pmw) method already illustrated, and the proportional (variable-and-variants) method which has been considered linguistically preferable, although it is often more problematic (see Smith and Leech, forthcoming). To illustrate the two methods, we show another instance of change, this time of marked increase. Figures 3 and 4 show the same phenomenon: the increasing use of negative contractions (-n’t instead of not) and contractions of auxiliary verb or copula. Figure 3 simply shows the increase use of -n’t in terms of instances per million words (pmw):

8,000

7,000

6,000

5,000 Press Gen Prose 4,000 Learned Fiction Overall 3,000

2,000

1,000

0 1931 1946 1961 1976 1991 2006

Figure 3. Normalized frequency method: Increasing frequency of negative contractions in occurrences pmw, 1931-2006.

As in Figure 2, the chart shows a general increase across the four subcorpora, except that the Learned variety (again, more conservative) shows a scarcely perceptible rise from zero, even declining slightly but non-significantly towards zero in the last 15-year period. It is also not surprising that the highest frequency and the steepest rise are in the Fictional subcorpus, which in general shows the closest similarity to spoken English, of which contractions are most characteristic.3

2 Hundt and Mair (1999) have described the contrasting slower and faster tendencies of ‘agile’ and ‘uptight’ genres (e.g. Press and Learned subcorpora respectively) in tracking linguistic change in corpora. 3 On increasing use of contractions in newspaper texts, see Axelsson (1998).

5 2 Invited Papers

Geoffrey Leech

Figure 4 shows the increase in contractions employing the proportional method of calculation. This means that frequency of the item concerned is measured against what would be the frequency of that item, if all negatives after an auxiliary were contractions. The frequency of negative contractions is here registered as a percentage of potential contractions. To show this, we sum occurrences not only of didn’t,Geoffrey isn’t, hadn’tLeech, etc. but also of did not, is not, had not, etc.

90%

80%

70%

60%

Press 50% Gen Prose Learned 40% Fiction Overall

30%

20%

10%

0% 1931 1946 1961 1976 1991 2006

Figure 4. Proportional method: Increasing frequency of negative contractions as a percentage of potential contractions, 1931-2006.

It will be seen that the two methods of tracking change show very similar results. However, in the rest of the paper, I will find it convenient, and sometimes necessary, to employ the normalized frequency method (pmw), since the alternative proportional method is often problematic, in that one has the problem of (a) determining, and (b) calculating, the number of potential, as well as actual occurrences. This is relatively straightforward in the case of negative contractions, in that one has to search the corpus for V + not as well as V+n’t, where V is a contractable verb. Even here, however, difficulties arise: for example, am not and may not are not able to contract to amn’t and mayn’t in contemporary English, although they belong to the class of verbs (copula and modals) which are normally contractable. Without going into detail into the reasons why the alternation ‑n’t ~ not does not apply to all cases, it will be clear that ‘contractability’ is not an easy property to define, let alone to quantify, even though it might be thought one of the simplest grammatical variables to handle. In other cases, it is virtually impossible to define the variable. For example, the modal auxiliary must has been declining rapidly in recent English, but to define the class of expressions of ‘obligation/necessity’ which are variants of the same

Invited Papers 5 3

Major Trends in Theoretical and Applied Linguistics

variable as must, in the sense that they could replace must without significant change of meaning, is extremely difficult. In different contexts it could be have to, need to, should, be obliged to, be necessary that, etc, but these have different syntactic privileges of occurrence and vary somewhat in meaning. The number of competing variants is virtually open-ended.

3. Possible Causes of Frequency Change

The question as to why such frequency changes take place is another issue which it is difficult to resolve. Are the observed changes due to internal, linguistic reasons, or to external social reasons? Are they brought about by conscious human influence, or by unconscious means? Although these questions are impossible to answer definitively, we can put forward conjectural explanations for change by noting related tendencies of growth or decline taking place across a range of different grammatical phenomena. The chief processes we postulate are: a. Colloquialization (or informalization) – adapting written language towards spoken norms (e.g. the increasing use of contractions, including negative contractions as already discussed) b. Grammaticalization – the process of converting lexical into grammatical material (e.g. historically, the growth in the use of the progressive and of semi-modals such as have to) c. Americanization – following the American lead – adopting American linguistic habits or conventions (e.g. the revival of the mandative subjunctive in British English) d. Densification – packing more meaning into less space – as found particularly in press headlines e. Prescriptivism – language change under the influence of overtly expressed attitudes either in favour of or against particular usages (e.g. the decrease of the use of the passive in recent decades).

Of these, a, c, d seem to be processes driven by social factors outside language, whereas grammaticalization (b) is chiefly internal to language (although spurred on by external influences). Prescriptivism (e) is the only case where conscious human motivation is involved. We will now consider each of these briefly in relation to the changing phenomena for which they offer an explanation.

3.1. Colloquialization

The trend whereby written language becomes more colloquial or speech-like has already been illustrated in the increasing use of negative contractions. Other

5 4 Invited Papers Geoffrey Leech

changes which appear to show the influence of colloquialization are increasing use of verb contractions, of semi-modals (see 3.2 below) of the present progressive (again see 3.2), of not-negation rather than no-negation (e.g. She didn’t say anything rather than She said nothing). The shift of preference away from wh-relativization towards that- and zero relativization can also be partially attributed to colloquialization. Indeed, colloquialization, although not a general trend within the language, may be considered a ‘natural’ process, in the sense that innovations in the language frequently occur first in the spoken medium and then to spread to the written medium. As the instance of relative clauses shows, the growth of one form (the that relativizer) may be associated with the decline of another, competing with it (the wh-relativizer). This negative side of colloquialization can arguably be a partial explanation for the declining use of the passive (see 3. 5 below).

3.2. Grammaticalization

The process of grammaticalization has been extensively investigated as one of the major factors accounting for language change, often over a long time-scale (see Hopper and Traugott 2003). Thus two textbook examples of grammaticalization come from the modal and aspectual systems of the English verb: the English modal auxiliaries (a rather well-defined class of verbs morphologically, syntactically and semantically) arose in a process of grammaticalization which reached its crucial period in late Middle English. More recently, a new ‘wave’ of grammaticalization has brought to prominence so-called semi-modals (or emerging modals – see Mair 1997, Krug 2000) such as be going to, have to, and want to. Grammaticalization is associated with increase in frequency, which can continue long after the main grammaticalizing process in other respects has concluded. These three semi-modals, with the addition of need to, have indeed been increasing steeply during the twentieth century, and one argument is that these have been encroaching on the more ‘traditional’ modals such as must and will (see Leech 2003, Smith 2003). The second case of grammaticalization we consider is the progressive aspect, which again appears to have arisen through grammaticalization in late ME and early ModE. Since then the progressive has gradually established itself in the language, and continues to increase in frequency, as is shown especially by the present progressive, which has increased c. 30% in both AmE and BrE over recent decades (Leech et al. 2009). It seems likely that the processes of grammaticalization and colloquialization act in synergy or in tandem – for example, the growing use of semi-modals is far more evident in the spoken language – where especially in American

Invited Papers 5 5 Major Trends in Theoretical and Applied Linguistics

conversation have to, got to and need to are all nowadays more common than must, the modal auxiliary in the same semantic field of obligation/necessity. Grammaticalization involves phonetic processes (e.g. vowel reduction, ellipsis and assimilation, as in the reduced forms of semi-modals informally spelt gonna, gotta, and wanna) associated with the spoken language, and it is therefore not unreasonable to suppose that changes through grammaticalization arise primarily in the spoken language and gradually affect the written language through colloquialization.

3.3. Americanization

The influence of American English on other regional varieties, such as British English, is easily accepted on the level of lexis – where in this author’s lifetime American usages such as movie, guys and I guess have been steadily taking over from corresponding British usages film, chaps/blokes, and I reckon/ suppose. To explain the largely one-way traffic from AmE to BrE, we hardly need to mention that the USA is easily the most influential English-speaking nation economically, culturally, and politically. Perhaps more relevant is the fact that the USA has by far the largest native English-speaking population – several times larger than the English-speaking population of the UK, the next most populous nation of the English-speaking world. Americanization is therefore a familiar process to speakers of BrE, as of other varieties of English around the world, even if the word itself is rarely used. The question to ask here is: can Americanization be applied not only to lexical change but to grammatical change? Change in grammar, more than change in lexis, takes place at a level below conscious notice. But the study of the comparable corpora of the Brown family yields the observation, in comparing these corpora, that the AmE and BrE often show the same process of change, but that in AmE the change is often ahead that in BrE. For example, the decline of the passive voice and the decline of the modal auxiliaries in the AmE corpora are both about a generation ahead of the BrE corpora. (Figures 5 and 6 – see also Leech et al. (2009).) For some, this is fairly convincing evidence that AmE, by showing a greater and earlier loss of frequency, is directly influencing BrE. However, others may regard it simply as evidence that AmE and BrE are on the same track, and that AmE happens to be somewhat more advanced, with no assumption of direct transatlantic influence. The evidence of Americanization is more conclusive if an AmE usage that was virtually non-existent in BrE at time time-1 becomes established in British usage at time time-2, or if at a later time time-3 the original British usage becomes rare and is supplanted by the American usage. Something like the former can

5 6 Invited Papers Geoffrey Leech

16,000

14,000

12,000

10,000

BrE 8,000 AmE

6,000

4,000

2,000

0 1931 1961 1991 2006

Figure 5. Frequency of the passive voice in the Brown family of corpora 1931-2006, showing a decline in the passive voice in BrE, ‘led’ by a decline in AmE.

16,000 14,000 12,000

10,000

8,000 BrE

AmE 6,000

4,000

2,000 0 1901 1931 1961 1991 2006

Figure 6. Frequency of modal auxiliaries in the Brown family corpora, showing a decline in frequency in BrE in 1961-2006 ‘led’ by a decline in AmE.

be observed with the competition between (a) the bare infinitive and (b) the to- infinitive following the verb help:

(a) help solve the problem (b) help to solve the problem

In the 1931 corpus, construction (a) occurred only a handful of times in the British corpus. By 1961 it was getting established, and by 1991 it had increased dramatically, imitating a similar rise in the American 1961 and 1992 corpora. Construction (b), on the other hand, declined slightly from 1961 to 1991 in the British, as well as in the American corpora. The trend, in both cases, was that the BrE corpora were ‘copying’ similar developments in the AmE corpora. This trend

Invited Papers 5 7

Major Trends in Theoretical and Applied Linguistics

can also be observed in the dramatic reversal, over 1931-1991 in BrE, of the preference for the construction (a) with have as a main verb behaving like an auxiliary over the ‘American’ construction with do-support (b):4 (a) Have you enough time? (b) Do you have enough time? (a) He hadn’t a chance. (b) He didn’t have a chance.

(On this topic, see Varela Perez 2007 and Leech et al. 2009: 255-6.)

3.4. Densification

One highly significant trend in both the American and British corpora ofthe Brown family is an increase in frequency of nouns (see Mair et al. 2002). Since nouns, both as types and as tokens, are the most numerous words in each corpus, and also tend to carry the most information, the most telling outcome of this increased ‘nouniness’ is an increase in density of information per n words. Densification, as this process can be called, is illustrated by a growth in the frequency of content words (especially nouns and adjectives) over function words (such as prepositions and determiners). In part, this can be explained by a growth in frequency of two particular syntactic features of the noun phrase: noun+noun sequences and s-genitives. The following is an extreme example of how frequency of nouns affects density of information in certain varieties of present-day English writing: the nouns are identified by underlining:

Five syntheses were completed this year and are expected to be published in early 2007: Commercial Motor Vehicle Driver Training Curricula and Delivery Methods and Their Effectiveness; Commercial Motor Vehicle Carrier Safety Management Certification; The Role of Safety Culture in Preventing Commercial Motor Vehicle Crashes; The Impact of Behavior-Based Safety Techniques on Commercial Motor Vehicle Drivers ; and Health and Wellness Programs for Commercial Vehicle Drivers. [AmE H06]

The italicized part of the third line shows how juxtaposed nouns can sometimes build up into complex sequences: here there is a sequence of five or six nouns in

4 The ratio in favour of (a) was 60:17 in the British 1931 corpus, and this had reversed to 21:126 in the 1991 corpus. The AmE corpora Brown and Frown showed heavy preference for (b), increasing by 1992: from 17:131, to 17:175. See Leech et al. (2009: 255).

5 8 Invited Papers Geoffrey Leech

a single noun phrase. In the Brown family, the increase of noun+noun sequences is steeper in AmE, but in BrE too, as Figure 7 shows, the increase is remarkable, and is consistent across all four subcopora.

40,000

35,000

30,000

25,000 Press Gen Prose 20,000 Learned Fiction 15,000 Overall

10,000

5,000

0 1931 1961 1991

Figure 7. Increasing frequency of noun+noun sequencies in BrE (1931-1991)

Noun+noun sequences are often more compressed alternatives to constructions using prepositional phrases: for example, land use can be replaced by (the) use of land, an expression which clearly employs more words and occupies more space. The construction of noun+noun sequences, although by no means a new development in the history of English, appears to have been growing considerably since at least the eighteenth century (Rosenbach 2006), and to have undergone a marked increase since the beginning of the twentieth century, with the growth of the popular press, particularly in the USA. Figure 7 confirms the popularity of this type of construction in journalism, as it shows Press as the subcorpus with the highest frequency of noun+noun sequences, and also shows a continuing marked rise (cf. Biber 2003). Clearly noun+noun sequencing is particularly characteristic of certain genres – not only journalistic, but also technological and bureaucratic. However, it is noticeable from Figure 7 that this feature of noun phrases has increased even in fiction writing, where technical compression of information is less expected. The increase in the s-genitive has also been remarkable over the twentieth century (see Hinrichs and Szmrecsanyi 2007), as Figure 8 shows for BrE: The increase of s-genitives, like that of noun+noun sequences, is steepest in the Press subcorpus, which also registers the highest frequency of this construction. Again, as in the case of noun+noun sequences, the effect of the s-genitive is often to compress information into the pre-modifying part of

Invited Papers 5 9

Major Trends in Theoretical and Applied Linguistics

10,000

9,000

8,000

7,000

6,000 Press Learned 5,000 Fiction 4,000 Overall

3,000

2,000

1,000

0 1931 1961 1991

Figure 8. Increase in the use of the s-genitive in BrE 1931-1991

the noun phrase, rather than to spread it out through the use of prepositions, especially of. Compare the length of the following examples:

the poll’s results ←→ the results of the poll the nation’s history ←→ the history of the nation

3.5. Prescriptivism

Prescriptivism is not so much a process as a state of mind: an attitude which favours certain usages and rejects others, often without good reason. Such attitudes can be widely and influentially advocated (e.g. by teachers, textbook writers, publishers, language mavens and grammar checkers), so that they can cause change in language behaviour, especially in written English, through the changing attitudes, often accompanied by guilt, they bring about in others. Two examples where prescriptivism appears to be part of the explanation of change are (a) the decline of frequency of the passive (see Figure 5) and (b) the marked change of preference from which to that as a restrictive relativizer, as in (1) and (2):

(1) Tarrant repeated at length the exchanges which had taken place in Scotland Yard. [LOB L02] (2) Tarrant repeated at length the exchanges that had taken place in Scotland Yard.

The declining frequency of the passive may be due to a combination of processes: colloquialization, Americanization, and prescriptivism. Since the

6 0 Invited Papers Geoffrey Leech

passive is far less frequent in conversation than in academic writing (where, as we have seen, conservative standards of writing are most strongly maintained), passive decline can be attributed, at least in part, to colloquialization – the spread of spoken habits of usage to writing. Also, as we have seen, AmE has been leading the way in this decline – so Americanization is also in play. However, it is also well known that powerful prescriptive forces have been advocating the avoidance of the passive during the twentieth century – a trend more recently reinforced by the effect of automatic grammar checking facilities incorporated in word-processing software. The evidence that the decline of the passive has been increasing suggests that such disapproval by ‘language authorities’ is having its effect. It also appears that this anti-passive propaganda is more powerful in the USA than in the UK (see Seoane and Williams 2006) – hence the association of prescriptivism in this case with Americanization. The second example of prescriptivism also emanates from sources in the USA, where the ‘sacred That Rule’ (as it has been called by Arnold Zwicky5) put forward forcefully by ‘language authorities’ decrees that which be avoided as a relativizer in restrictive relative clauses, and that that be used instead (i.e. in the examples above, (2) is prescribed instead of (1)). Here again, colloquialization and Americanization can be linked together as contributing processes, but the most influential factor in causing a switch from wh-relativizers to that (and, to a lesser extent, to zero relativization) appears to be prescriptive pressure. Figures 9 and 10 show the increase of that-relativization and the decline of

3500 3000 2500 2000 BrE 1500 AmE 1000 500 0 <1931> <1961> <1991> <2006>

Figure 9. Frequency pmw of that-relative clauses in the Brown family of corpora, with comparison between BrE 1931-1991 and AmE 1961-1992.

5 See the discussion of this issue in Zwicky’s ‘Language Log’ (posted 22 May 2005 at http://itre.cis.upenn.edu/~myl/languagelog/archives/002291.html#more.)

Invited Papers 6 1

Major Trends in Theoretical and Applied Linguistics

8000 7000 6000 5000 BrE wh- 4000 AmE wh- 3000 BrE which 2000 AmE which 1000 0 <1961> <1991>

Figure 10. Frequency pmw of wh-relative clauses in the Brown family of corpora 1961-1991/2, with which-relatives shown separately (with dashed lines) [NB: This includes both restrictive and non- restrictive relative clauses.]

wh- relativization in both AmE and BrE. In AmE, the decline of wh- relativizers is largely due to avoidance of which rather than of who, whom or whose. In BrE, although the use of which declines, who, whom and whose also account for part of the decrease. It appears that BrE, up to 1991, was little affected by the ‘sacred That Rule’, and that the moderate decline of wh- pronouns may have been due more to the negative effect of colloquialization.

4. Conclusion

The foregoing account of change in English grammar highlights some of the most significant findings that have emerged from the study of the Brown family of corpora over the last decade. What I have presented is no more than an outline of ‘work in progress’ which continues as more corpora are added to the family, and as the analysis is extended to new periods. Moreover, the Brown family, being restricted to printed and published resources, needs to be supplemented by other diachronic corpora, especially the Diachronic Corpus of Present-day Spoken English (DCPSE), the Corpus of Contemporary American English (COCA) and the Corpus of Historical American English (COHA). This is not to mention the many corpora that have been developed or are under development representing other regional varieties of English around the world. Although details of statistical significance have not been included in this account, other published sources (especially Leech et al. 2009) confirm that almost all of the trends illustrated are highly significant according to log likelihood tests.

6 2 Invited Papers Geoffrey Leech

The research reported here has been undertaken since 1998 with the support of grant-giving bodies (the Arts and Humanities Research Board, the British Academy and the Leverhulme Trust), and with the help of many co-researchers, especially Nicholas Smith and Paul Rayson, to whom I am grateful for continuing discussion and collaboration.

Invited Papers 6 3 Major Trends in Theoretical and Applied Linguistics

References

Axelsson, M.W. 1998. Contraction in British Newspapers in the Late 20th Century. Uppsala: Acta Universitatis Upsaliensis.

Biber, D. 2003. Compressed noun-phrase structures in newspaper discourse: the competing demands of popularization vs. economy. In J. Aitchison and D.M. Lewis (eds.), New media language. London: Routledge, 169-181.

Facchinetti, R., M. Krug and F. Palmer (eds.). 2003. Modality in Contemporary English. Berlin: Mouton de Gruyter.

Hinrichs, L. and B. Szmrecsanyi. 2007. Recent changes in the function and frequency of Standard English genitive constructions: A multivariate analysis of tagged corpora. English Language and Linguistics 11(3), 335-378.

Hopper, P.J. and E.C. Traugott. 2003 [1993]. Grammaticalization. 2nd edn. Cambridge: Cambridge University Press.

Hundt, M. and C. Mair. 1999. “Agile” and “uptight” genres: The corpus-based approach to language change in progress. International Journal of Corpus Linguistics 4, 221-242.

Krug, M. 2000. Emerging English Modals: A Corpus-based Study of Grammaticalization. Berlin and New York: Mouton de Gruyter.

Leech, G. 2003. Modality on the move: The English modal auxiliaries 1961-1992. In R. Facchinetti, M. Krug and F. Palmer (eds.), Modality in Contemporary English. Berlin: Mouton de Gruyter, 223-240.

Leech, G., M. Hundt, C. Mair and N. Smith. 2009. Change in Contemporary English: A Grammatical Study. Cambridge: CUP.

6 4 Invited Papers Geoffrey Leech

Mair, C. 1997. The spread of the going-to-future in written English: a corpus- based investigation into language change in progress. In R. Hickey and S. Puppel (eds.), Language History and Linguistic Modelling. A Festschrift for Jacek Fisiak. Berlin: Mouton de Gruyter, 1537-1543.

Mair, C., M. Hundt, G.N. Leech and N. Smith. 2002. Short term diachronic shifts in part-of-speech frequencies: A comparison of the tagged LOB and F-LOB corpora. International Journal of Corpus Linguistics 7(2), 245-64.

Rosenbach, A. 2006. On the track of noun+noun constructions in Modern English. In C. Houswitschka, G. Knappe and A. Müller (eds.), Anglistentag 2005 Bamberg. Proceedings. Trier: Wissenschaftlicher Verlag, 543-557.

Seoane, E. and C. Williams. 2006. Changing the rules: A comparison of recent trends in English in academic scientific discourse and prescriptive legal discourse. In M. Dossena and I. Taavitsainen (eds.), Diachronic Perspectives on Domain-specific English.Bern: Peter Lang, 255-276.

Smith, N. 2003. Changes in the modals and semi-modals of strong obligation and epistemic necessity in recent British English. In R. Facchinetti, M. Krug and F. Palmer (eds.), Modality in Contemporary English. Berlin: Mouton de Gruyter, 241- 266.

Varela Pérez, J.R. 2007. Negation of main verb have: Evidence of a change in progress in spoken and written British English. Neuphilologische Mitteilungen 108, 223-246.

Invited Papers 6 5 Major Trends in Theoretical and Applied Linguistics

6 6 Towards a Constructional Account of Indefinite Uses of Proper Names in Modern Greek1

Sophia Marmaridou

University of Athens [email protected]

Abstract

This paper addresses the issue of Modern Greek proper names, within the framework of Construction Grammar. More specifically, proper noun headed nominal constructions will be the focus of attention. Unlike earlier accounts that view common and proper nouns as grammatically, semantically and functionally distinct from each other, the present investigation provides evidence for constructional properties these categories have in common. It will be shown that in Modern Greek a proper noun as head of a nominal construction unifies with determiners and adjectives in the same way that common nouns do. Moreover, like common nouns, proper noun heads activate semantic frames associated with particular categories of entities (e.g. persons, cities, cars, etc.). Additionally, proper noun frames are contextually rich. More specifically, the use of a proper noun activates a frame which is so rich in information specificity that in this frame the name identifies a unique referent. This referent can provide metonymic access to various aspects of the frame when the proper noun is the head of a nominal construction. Significantly, metonymic uses of proper nouns in indefinite constructions, varying in degree of conventionality, are often associated with a subjective viewpoint and rate high in expressivity. Apparently, the findings of this study argue for a holistic, constructional approach to proper names and a frame-based analysis of their use. The present corpus-based analysis adopts a Construction Grammar

1 I am greatly indebted to the Dept. of English Language and Literature of the Univ. of Thessaloniki and the organizing committee of ISTAL20 for inviting me to give a plenary talk at the conference. This paper has benefited from several participants’ comments, questions, and suggestions and especially from a long discussion with Joseph Emonds. All shortcomings are, of course, my own.

Invited Papers 6 7 Major Trends in Theoretical and Applied Linguistics

framework and notation after Fried & Östman (2004) to highlight inheritance relations and unification principles among the relevant constructions. Moreover, the investigation makes reference to mental space theory and conceptual blends as in Fauconnier (1997) and the concept of frame-metonymy as in Sweetser and Fauconnier (1996). Therefore, the present work argues for a usage-based perspective on language and a principled exploration of its cognitive, cultural, and constructional motivations.

1. Introduction

The present study aims to show that a Construction Grammar framework provides useful insights to the analysis of proper names, which has been an issue of much controversy in both philosophical and linguistic discussions for more than a century. It is further argued that, being a holistic, usage-based theoretical model, Construction Grammar is compatible with other cognitive, usage-based approaches to linguistic analysis, including mental space theory. Finally, the theoretical stance taken in this paper entails a methodological commitment to a corpus-based approach to the data and a consideration of all semantic, syntactic, pragmatic, and discoursal features of proper noun headed constructions. Following Huddleston (1988: 96), the distinction between proper nouns and proper names is maintained; a proper noun is a grammatical noun subclass, while a proper name results from naming practices and can be a proper noun, such as Thessaloniki or John, but need not be, as names like The United Nations indicate. As will become evident subsequently, the nominal constructions to be analysed in this paper are headed by proper nouns which constitute personal proper names in Modern Greek. The debate about the origin, meaning and function of proper names dates at least as far back as the end of the 19th and the beginning of the 20th century within the framework of the philosophy of language (see, for example, Mill (1956[1843], Frege (1892), and Russell (1905), among others). More specifically, while Mill (1956[1843]) argued that proper names only have denotation, or reference, and do not connote (i.e. mean) anything at all, Frege (1892) claimed that apart form that which the name designates (i.e. denotation), there is also that which contains the way in which the denotation of a name is given. In this respect, like common nouns, proper names denote in virtue of the propositional content of the descriptions associated with them (e.g. Aristotle was a philosopher). Subsequent discussions by Russell (1905), Strawson (1950), and Searle (1969) essentially adopt Frege’s views, while Kripke (1972), Putnam (1975), and Donnellan (1975) not only consider proper names as rigid designators devoid of any kind of meaning, but also extend this account to the category of natural kinds, such as cats, cows, lions, etc.

6 8 Invited Papers Sophia Marmaridou

Their basic argument is that the extension of natural kinds is not determined by concepts that represent basic characteristics of these kinds, but instead are like proper names whose extension is determined by ostension and/or something like a causal chain that associates a term with its referent or the object to which it applies. More recently, this issue has been addressed from a variety of linguistic viewpoints, as in Zwicky (1974), Hawkins (1978), Allerton (1987), Carroll (1985), Marmaridou (1984), (1989), and (1991), Langacker (1991), Van Langendonck (1999), Lehrer (1992), Vandelanotte & Willemse (2002), and Dancygier (2009) and (2011), among others. The interest in proper names mainly follows from two issues. The first relates to the obvious fact that even though proper names are primarily used to uniquely identify referents in the universe of discourse, they are also used to indicate that this is not in fact the case, as in the examples below:

(1) John Carpenter arrived in a limo. (2) I knew a John Carpenter at school.

The co-occurrence of the name with the indefinite article in (2) suggests that there is possibly more than one referent known by that name. In fact, several uses of proper names relax the uniqueness requirement, as shown in almost all recent work cited above. The second issue concerns the grammatical form of proper names. For example, even though proper names in English prototypically occur in the bare form of nominals (i.e. without determiners or other type of modification), they are frequently part of constructions that are typical of common nouns in the language, as example (2) indicates above. Moreover, grammatical similarity in proper name uses does not guarantee similarity in meaning, as (3) and (4) below suggest:

(3) A furious Thatcher entered the Cabinet meeting earlier this afternoon. (4) A beautiful Standhal was hanging on their living-room wall.

Clearly, there is only one Thatcher in the current universe of discourse and the name uniquely identifies this referent on a particular occasion. Similarly, there is only one painting hanging on the wall, but the name is used metonymically to refer to it as one of several paintings. In this sense, the indefinite article cannot be dissociated from the meaning of singularity it also encodes. Moreover, the adjective modifying the proper name in (3) indicates a temporary quality of the referent, while in (4) the work of art is attributed a permanent quality by the modifying adjective. The variability in the meaning and use of proper names is also attested in the Modern Greek language, in which proper names are typically preceded by

Invited Papers 6 9 Major Trends in Theoretical and Applied Linguistics

the definite article o/i/to (‘the’) in their referential uses, but also frequently co- occur with the indefinite determinerenas/mia/ena (‘a/an’), for the three genders respectively. Their bare form occurs only in non-referential uses, as is the case with common nouns in this language. The following examples illustrate these points:

(5) o mitsotakis ektimise tis psifoθirikes ikanotites the mitsotakis appreciate-3rd.sg.pst the vote-soliciting capabilities tu neu politiku the-gen young-gen politician-gen. ‘Mitsotakis appreciated the vote-soliciting abilities of the young politician.’ (HNC 449816)2

(6) enas papavlasopulos se pire. Pjos ine; A papavlasopulos you call-3rd.sg.pst Who be-3sg.pr? ‘A Papavlasopoulos called you. Who is he?’ (oral example attested by the author)

(7) ala tus perimene enas δiaforetikos, apetitikos karamanlis but them wait-for 3sg.pst a different, demanding Karamanlis ‘But a different, demanding Karamanlis was waiting for them.’ (HNC 658997)

(8) periexi ke arketa plasta erγa anamesa sta opia enas maleas, contain-3rd.sg.pr and several fake works among in-the which a aleas enas parθenis ke poli piθanon ke enas tsaruxis a parθenis and very probably and a tsaruxis ‘It also contains several fake paintings, among them a Maleas, a Parthenis and, most likely, a Tsarouxis.’ (GkWaC 7695)

(9) an ke iuδas δen ejina pote if and Judas not become-1st.sg.pst never ‘even though I never became a Judas’ (GkWaC 1279)

The subject nominal in (5) illustrates the prototypical, referential use of proper names: reference is made to a specific individual, known to both the author of the article it appears in and the reader. In (6) the referent identified by

2 The Modern Greek corpora consulted include the HNC (Hellenic National Corpus, 47m. words) and the GkWaC (Greek Web as Corpus, 150m. words) operated by Sketchengine.

7 0 Invited Papers Sophia Marmaridou

the indefinite nominal is unknown to the speaker and, possibly, to the addressee. The only thing known about this person is his name, i.e. Papavlasopoulos, which morphologically identifies the referent as a male human being. The indefinite nominal in (7) has a different function from that of (6). Reference is made to a specific entity known to both speaker and addressee, but the indefinite determiner and the modifying adjectives present the referent on a specific occasion, i.e. when, unlike his typical self, he was more demanding. Apparently, the nominal signals a temporary characterization of the referent similar to that in (3). The indefinite noun phrases in (8) illustrate the well-entrenched metonymy whereby the name of a painter stands for one of his or her works to which reference is made. Finally, (9) exemplifies a non-referential use of a culturally salient name, aiming to evoke the sense of a traitor. In view of the above, the aim of this paper is to address the issue of the use of Modern Greek personal names, and surnames in particular, within the framework of Construction Grammar. More specifically, proper noun headed nominal constructions will be the focus of attention. Unlike earlier accounts that view common and proper nouns as grammatically, semantically, and functionally distinct from each other, the present investigation provides evidence for constructional properties these categories have in common. It will be shown that a proper noun as head of a nominal construction unifies with determiners and adjectives in the same way that common nouns do. Moreover, like common nouns, proper nouns activate semantic frames associated with particular categories of entities, e.g. human beings, animals, locations, etc. Additionally, proper noun frames are contextually rich. More specifically, the use of a proper noun activates a frame which is so rich in information specificity that, in this frame, the name identifies a unique referent. This referent can provide metonymic access to various aspects of the frame when the proper noun is the head of an indefinite nominal construction, as in examples (7) and (8). Significantly, metonymic uses of proper nouns in indefinite nominal constructions, varying in degree of conventionality, are often associated with a subjective viewpoint and rate high in expressivity. Therefore, the findings of this study argue for a constructional analysis of proper names and a frame- based analysis of their use. At a methodological level, the present corpus-based analysis adopts a Construction Grammar framework and notation after Fried & Östman (2004) to highlight inheritance relations and unification principles among the relevant constructions. Moreover, the investigation makes reference to mental space theory as in Fauconnier (1997) and the concept of frame-metonymy as presented in Sweetser and Fauconnier (1996). Overall, the present work draws on a usage- based perspective on language and a principled exploration of its cognitive, cultural, and constructional motivations. To the extent that generalizations pertaining to the English language and used in linguistic theorizing by several

Invited Papers 7 1 Major Trends in Theoretical and Applied Linguistics

scholars have been brought to bear on Modern Greek data, this study may be said to also adopt a cross-linguistic perspective. In the next section some principles of Construction Grammar will be set out with a view to highlighting those aspects of the theory that are particularly relevant for the analysis of the data. In the third section the challenges posed by the use of proper names will be set off against the background of more recent research that will inform the present study. In the next section referential and non- referential uses of proper names in Modern Greek will be shown to depend on the occurrence or non-occurrence of determiners respectively, following a general pattern of the grammar of nominal constructions in this language. In the fifth section indefinite nominals headed by proper nouns will be examined and the relevant constructions will be highlighted and fully described. In the last section, and on the basis of the analysis proposed, it will be argued that Construction Grammar, as a usage based approach to the study of linguistic units, is well suited to an account of proper names, whose use not only combines aspects of form and meaning, but also crucially involves pragmatic and discoursal attribute values.

2. The Construction Grammar Framework

Before proceeding any further it is perhaps important to set out some basic principles of Construction Grammar that bear on the proposed analysis, and the notation to be used in it.

2.1. Historical Background

Construction Grammar as a distinct model of linguistic analysis has been around for at least twenty years now. It started with work by Fillmore and Kay in Berkeley around the eighties and initially led to a number of important publications in Language in the eighties and nineties. It subsequently extended in the direction of information structure by Lambrecht (1994), argument structure by Goldberg (1995), and formal semantics by Kay (1997). The integration of Construction Grammar and Fillmore’s Frame Semantics – the latter developed much earlier – led to the well-known application of FrameNet, as reported in Johnson et al. (2001). The long established view that language change is pragmatically motivated and affects form and meaning at the same time (see Hopper and Traugott (1993); Traugott and Dasher (2002)), made Construction Grammar particularly suitable for investigating instances of language change and was therefore enhanced in works such as Leino & Östman (2005), Hoffmann & Trousdale (2011) and Gisborne (2011). It soon became clear that this model could be extended to also account for discourse regularities. Interest in that area is rapidly increasing, as reflected

7 2 Invited Papers Sophia Marmaridou

in Fried & Östman (2003), Nikiforidou (2010) and (2011), and Antonopoulou & Nikiforidou (2011), among others. Expectedly, grammatical constructions shed new light on the analysis of interaction as work by Fried & Östman (2005), Fox & Thomson (2007), Fried (2009), and Terkourafi (2010) reveals. Finally, given the theoretical commitment of Construction Grammar to account for linguistic units as constructs combining form and function, this framework was soon adopted by researchers such as Waara (2004), Broccias (2008), and Valenzuela Manzanares & Rojo López (2008) in the area of second language teaching and learning.

2.2. General Principles

Construction Grammar is a theory of language that, like many others, aims to represent the relationship between structure, meaning, and use. However, unlike other theories of language, it sees function and form as inseparable from each other, thereby introducing the notion of grammatical construction as the basic unit of analysis and representation, and as comprising information on all levels of linguistic analysis, from the level of phoneme to the level of text. In this framework language is seen as the inventory of its constructions; knowing a language is knowing its constructions. Moreover Construction Grammar has empirical goals in that, while aiming at maximally relevant generalizations and aspiring to capture cross-linguistic generalizations, it views the existence of language universals as an empirical matter, rather than an article of faith. More specifically Construction Grammar entertains the hypothesis that speakers rely on relatively complex meaning-form patterns – grammatical constructions – for building linguistic expressions. Constructions are organized into networks of overlapping patterns related through shared properties. Therefore, linguistic expressions reflect the effects of interaction between constructions and linguistic material (e.g. words) that occur in them. The primary motivation for Construction Grammar is the insight that the juxtaposition of two or more forms seldom results in a simple concatenation of the meanings those forms might have in isolation (Fried & Östman 2004: 12). This motivation justifies a number of methodological principles. First of all, a uniform representation of all grammatical knowledge in the speaker’s mind is sought. Second, both words and clauses have constructional properties and are represented as constructions. Third, linguistic units, as particular associations between form and meaning, are represented by multiple inheritance taxonomic networks (Croft & Cruse 2004: 262-265). Therefore, no priority (theoretical or methodological) is given in this model to particular language forms, e.g., active, declarative, transitive clauses, and no a priori decision is required about what should count as “basic” or the “core” in language. In this context, an account of what are usually considered “peripheral” parts of language is both

Invited Papers 7 3 Major Trends in Theoretical and Applied Linguistics

methodologically desirable and fully justifiable in that such parts constitute the majority of structures used in everyday discourse, while they are not explicitly taught. It follows that the relation between “productive rules” and “idioms” is seen as a cline from relatively productive to relatively frozen. Consequently, linguistic units such as “Thank you” or “See you” are as central to English as “John loves Mary”, or “The book, which Susan read, is on the table”. Support for the view that an adequate description of linguistic structure needs to make reference to complex constructions rather than to a generative capacity of a rule system comes from cognitive correlates of language processing. It has been argued that we process language in larger blocks, in gestalts (Chafe, 1994; Pawley, 1987; Pawley & Syder, 1983). Memory and language storage function in terms of larger formulas rather than in terms of words or phrases. We talk in terms of prosodic units rather than in terms of one word or one linguistic phrase at a time. As Bolinger (1951; 1976) has long argued, we make use of “prefabs” in language, including the prosodic patterning of language. Levels of an intonation contour are not particularly salient for speakers; rather, they serve as cues in the context of intonation contours as wholes. The above is also consistent with what we know about cognition and language acquisition. Recent advances in research on child language show that our cognitive processes manipulate linguistic structures of varying sizes and complexity (Peters, 1983; Praine & Brooks, 1995; Tomasello 1992, 2000; Goldberg 2006).

2.3. The Grammatical Construction

In Construction Grammar grammatical constructions are symbolic signs that represent the basic building blocks of linguistic analysis (e.g. morpheme, word, phrase, text, etc.). All grammatical constructions can be placed on a lexicon – syntax continuum. A construction is an abstract, conventional, representational entity providing a blueprint for licensing well-formed linguistic expressions (Fried & Östman 2004: 18). The actually occurring linguistic expressions such as phrases, sentences, etc. are constructs. It follows that we use constructs to communicate – not constructions. Constructions can vary in degree of complexity. They may represent simple configurations, such as [Det + N], which any phrase-structure grammar could account for. However, the Construction Grammar framework is particularly suited to represent much larger and complicated patterns, containing syntactic, semantic, pragmatic, etc. layers of information. For example, it should be able to account for instances of proper name use, such as those represented in (7) and (8) above, crucially making reference to the semantics of indefinite nominals and encyclopaedic information activated by the proper noun used. It is therefore a multi-dimensional framework. In this model it is important to realize that linguistic expressions typically “reflect the effect of interaction between

7 4 Invited Papers Sophia Marmaridou

constructional patterns and the words that fit in them: words, whether heads or dependents, contribute specific semantic properties to any larger construction they occur in, but a construction may also modify some of those properties, as well as add features of its own” (Fried & Östman 2004: 22). Given the complexity of linguistic patterns that Construction Grammar addresses, it becomes important to formalize relationships holding between linguistic elements in the attempt to explicate and understand them. Therefore appropriate notation is used as a convenient way of providing a construction with its phonological, morphological, syntactic, semantic, pragmatic, discourse, and prosodic characteristics. Three major notation devices are used in Construction Grammar. The first concerns boxes-within-boxes diagrams representing constituent structure and capturing dominance relations and the linear order of constituents. External characteristics of a construction as a whole are represented in outer boxes. Internal constituents of a construction are represented in inner boxes. The distinction between external and internal domains is crucial, given that a construction as a whole is typically not a sum of its parts. The possible discrepancies between the external and internal properties may have to do either with a mismatch in form, or with a non-compositional meaning of the construction.3 The second device concerns feature structures providing grammatical information in terms of attributes and values. For example, in the nominal construction in (5) above, the morphological attributes [gender], [case], and [number] take as their values the features [masculine], [nominative], and [singular], respectively. The third device is co-indexation for keeping track of unification relations. For example, in the same example, for the determiner to unify with the proper noun, they have to be co-indexed for the same gender, case, and number. An important aspect of Construction Grammar is that when two or more grammatical patterns show partial overlap, such that one pattern is a more constrained version of another, it is assumed that inheritance relations are at work. In this sense, inheritance relations between constructions keep track of properties along which these constructions resemble each other. For example, if the proper noun construction inherits the common noun construction, then it contains all the specifications that hold for the common noun construction, in addition to features idiosyncratic to the proper noun construction. The Construction Grammar notation and its theoretical underpinnings will become clearer in the proposed analysis of proper names.

3 There is no general agreement on the requirement for formalization among linguists subscribing to Construction Grammar. For example, boxes-within-boxes diagrams were introduced by Fillmore in the early stages of the particular framework and later adopted by Fried and Östman, among others, but not by Goldberg (see references at the end of this paper).

Invited Papers 7 5 Major Trends in Theoretical and Applied Linguistics

3. Philosophical and Linguistic Perspectives on Proper Names

This section aims to present a number of issues that have arisen in philosophical and linguistic approaches to proper names, with a view to highlighting some aspects of their use that justify further investigation. In linguistic treatments, in particular, the scope varies between morpho-syntactic accounts and semantico-pragmatic ones. It is perhaps not a coincidence that revived interest in this category has been noted in the last decade, when advances in Cognitive Linguistics and Construction Grammar have enabled a holistic approach to the investigation of linguistic units, whereby all information resides on one and the same level of representation and actual constructs are licensed, rather than generated, by particular abstract constructions. A perennial issue concerning the analysis of proper names is whether they have meaning or not, and if so, how this meaning relates to their referential function. Within the philosophical tradition of truth-conditional semantics, where the distinction between sense and reference is the cornerstone of semantic analysis, the ‘meaning’ of proper names is a matter of controversy. Frege (1892), Russell (1905), and Searle (1969), among others, claim that the meaning of a proper name consists of a number of propositions that truly describe its referent. This descriptivist view is juxtaposed to the communication chain view, in which the reference of a term is not given by a description, but rather by a chain of communication that starts with the initial dubbing of a name for an individual and continues with the successive use of this term by people in order to refer back to this individual (see e.g., Mill (1956[1843]); Kripke (1972)). It is not the aim of this paper to go into the arguments that have developed over this issue in any detail. What is of interest, though, is that, despite their differences, what both philosophical approaches seem to have in common is the assumption that proper names identify unique referents in the world, or universe of discourse. It is this assumption that has been successfully handed down to linguistic approaches to the analysis of proper names, and has motivated the interest in the grammatical status and function of members of this category as distinct from common nouns. In a discussion of issues of reference and definiteness, Marmaridou (1984) focuses on the Modern Greek definite article o/i/to and argues that this determiner is primarily a marker of reference. One of the arguments offered is its obligatory co-occurrence with proper names, which are definite par excellence in their referential function, and its absence from non-referential uses. Within a pragmatic, relevance-theoretic framework, Marmaridou (1989) is an attempt to approach the analysis of proper names from the point of view of communication and cognition. It is specifically argued that a proper name

7 6 Invited Papers Sophia Marmaridou

“itself activates the formation by the hearer of an appropriate context in which the name … can only identify one referent … the name is not interpreted against a pre-existing context, but rather itself selects one among several contexts in which it is appropriately paired with a specific referent” (ibid: 363). In this analysis, what determines the selection of a particular context is the search for optimal relevance in communication, in the sense of Sperber & Wilson (1986). Apparently, though, the selection of the appropriate context is also constrained by linguistic and cultural convention. For example, Allerton (1987) distinguishes seven broad categories that get names and Lehrer (1992) provides evidence for a system of culturally specified naming choices. More specifically, it is demonstrated that native speakers have intuitions about how to name members of a particular lexical field and that naming, though creative, is highly constrained. (ibid: 140). Similarly, Marmaridou (1991), adopting a socio-cultural perspective, examines the use of names in literary works and identifies meaning-creating and meaning-created proper names (e.g. Godot in Becket’s famous play and Sophia in Fielding’s Tom Jones as a wise and virtuous girl, respectively), thereby arguing for a cultural constraint on the use of proper names. A grammatical perspective on proper names has usefully brought forward the distinction between proper nouns and proper names. Following Huddleston (1988), Lehrer (1992: 124-125) notes that a proper noun is a grammatical noun subclass, while a proper name is the institutionalized name of some entity. A proper name may have the form of a proper noun, i.e. London, Jack, etc., but it need not do so. Thus, the Open University is a proper name, but the older university is not. It is also observed, though, that the difference between common and proper nouns is anything but clear-cut, given the variability of name forms to include common word classes and the fact that proper nouns are often used as non-referring expressions. By contrast, Van Langendonck (1999) argues for a clear distinction between proper and common nouns on the basis of the non-descriptive and ad hoc assignment of the former. In answer to the apparent fact that linguistic and cultural convention attribute minimal semantic (e.g. categorical) content to proper names, he claims that, semantically, proper names do not have asserted lexical meaning, but do display presuppositional meaning, including categorical (basic-level), associative (introduced via the name-bearer or via the name form), and grammatical meanings (ibid: 95). More importantly from the point of view of the present paper, Van Langendonck (ibid: 101-102) introduces a further distinction between a dictionary lemma and a word in a particular semantic-syntactic function. According to this view, a proprial lemma, i.e. a word that prototypically functions as a proper name (e.g. Mary, John, the Thames), is to be distinguished from its semantic-syntactic functions. This distinction motivates his analysis of the proper names in (10) and (11)

Invited Papers 7 7 Major Trends in Theoretical and Applied Linguistics

as employed in their primary identifying function, whereas in (12) and (13) as having a secondary, describing function:

(10) John attended a meeting today. (11) Napoleon was defeated at Waterloo. (12) You are talking about a different John. (13) He is becoming a second Napoleon.

Clearly, the proper names in (12) and (13) are acting as common nouns, i.e. they have the semantic-syntactic function typically assigned to common nouns. Van Langendonck claims that this secondary function is linked to the primary one by operations like metonymy, as in (12), and metaphor, as in (13). This analysis is not without its problems. The proper names in both (12) and (13) are heads of an indefinite nominal construction, but they are functionally distinct in that the one in (12) performs a referential function, while the one in (13) performs a predicate function. It is not clear whether the author is actually claiming that referential functions of proper noun headed indefinite nominals are metonymic, while non- referential ones metaphorical. Moreover, this analysis does not explain why in (12) one is speaking of at least two different persons called John, while in (14) below reference is made to one and the same individual called Putin, while both proper noun headed indefinite nominals perform the same referential function in the respective examples.

(14) It appears that we re now dealing with a different Putin… (WebNews, reported in Dancygier (2011: 223)

Apparently, there is more to the analysis of proper names than semantic- syntactic considerations of function, which needs to be specified at a more global level so that various parameters of use, including semantic, pragmatic, and cultural, can be taken into account. It will be proposed that such parameters can be specified at constructional level. Adopting the view that proper names are rigid designators, which is not without its problems as already noted, Vandelanotte and Willemse (2002) follow Van Langendonck’s distinction between proprial lemmas as dictionary entries and proper names as a semantic-syntactic class, with a view to classifying instances of the latter in terms of modification, including articles, numerals, demonstratives, and adjectives. Essentially, the authors identify grammatical constructions, in which the name either denotes a referent other than the one it typically associates with, thereby functioning as a common noun with a categorizing or type-specifying meaning, as in (15) and (16) below, or the name refers uniquely, as in (17) and (18), but some additional information is provided concerning the referent.

7 8 Invited Papers Sophia Marmaridou

(15) That Picasso is not on exhibition at the moment. (16) This country will be fine until the next Margaret Thatcher comes along. (17) I can’t live in a sad and misty Paris. (18) It was in a sad and misty Paris that the general was carried to his grave.

Notably, the definite article and demonstratives modify the name only when the proprial lemma functions as a common noun and therefore extends its ‘lexical’ meaning metonymically or metaphorically, as in (15) and (16). The indefinite article is said to modify the proprial lemma both when functioning as a common noun, as in (17), and when functioning to identify a unique referent, as in (18). However, the boundary between the two functions in the latter two examples is not clear. According to the authors, in both of these examples the modified name indicates a temporary property of the same referent, but (17) involves a restrictively modified proprial lemma used as a common noun (e.g. a sad and misty Paris as opposed to a bright and shining one), whereas (18) involves a non-restrictively modified proprial lemma used as a proper name (e.g. Paris, which, as you know, is sad and misty). However, this distinction is not borne out by the data, possibly because the context in which (17) and (18) occur is not provided. If there is such a distinction, as the authors assume, and context is important in identifying it, then it is possible to argue for the contextual, or discoursal, motivation of the respective functions. In the absence of any further contextual parameters, the identity of the referent is the same, but in (17) the referent is apparently modified for properties that characterize it time and again, whereas in (18) the modification seems to apply to a particular span of time, i.e. a day. In this understanding, the name in both examples identifies a unique referent whose description by the modifying adjectives applies on different occasions. It should also be noted that the constructional pattern instantiated by both examples seems to reflect the speaker’s attitude towards the referent identified. This is not unrelated to the type of adjectives used to modifythe specific referent and the use of the indefinite article. As it turns out, differences in the understanding of a modified proper name may relate to various attributes of the grammatical construction of which it is a part, such as contextual, discoursal, and attitudinal, expressing speaker involvement. It transpires from the above discussion that the distinction between proprial lemmas and proper names, like that between proper nouns and proper names, is a useful one in so far as it contributes to the realization that the analysis of proper names cannot be dissociated from the nominal constructions in which they occur. Moreover, this distinction brings forward the need to specify constructional attributes of proper names at various levels of linguistic analysis, including semantic, syntactic and pragmatic, thereby justifying a Construction Grammar framework for this task. In what follows, proper names in Modern Greek will be examined in more detail.

Invited Papers 7 9 Major Trends in Theoretical and Applied Linguistics

4. Referential and Non-Referential Uses of Proper Names in Modern Greek

The examination of proper names is related to the issue of the Modern Greek nominal construction in an obvious sense. Unlike English, proper nouns in Greek typically unify with the definite articleo /i/to in their referential function, thereby following the general pattern of noun headed constructions. Proper nouns unifying with indefinite determiners, such asenas /kapjios, are also systematically attested as referring expressions, cancelling the uniqueness presupposition, as it were. Therefore, the unification of proper nouns with determiners follows from a general pattern of Modern Greek nominals. It has long been argued (Marmaridou 1984) that in Modern Greek, determiners characterize the referential function of noun phrases, whereas non-referential predicate functions are typically characterized by absence of determiners. Grammatical descriptions of this language also reflect this point. In Construction Grammar terms, the unification of a noun head with a determiner specifies

an instance of a type, Ti. When a noun head does not unify with a determiner, the nominal specifies a type T (following Langacker 1991: 55). A special class of verbs, commonly termed copulative and/or auxiliaries, including ime (be), jinome (become), exo (have), and kano (do), typically select for complements specifying a type, i.e. they unify directly with noun heads, unless the noun head is modified by a relative clause (Marmaridou 1984; Mackridge 1985; Joseph and Philippaki 1987; Holton et al. 1997).4 In this latter case, the noun head unifies with a determiner. The two types of predicate nominals are exemplified in (19) and (20) below:

(19) ine pseftis be-3rd.sg/pl.pr liar-sg.masc ‘He is a liar.’ (HNC 605965) (20) o iroas ine enas fititis pu vioni mia katastasi stasimotitas the hero be-3rd.sg.pr a student who experience-3rd.sg.pr a situation stagnation-gen ‘The hero is a student who is experiencing a state of stagnation.’ (GkWaC 68282)

In (19), the predicate noun head is a phrasal constituent and directly unifies with the copula. Semantically, this predicate nominal specifies the type of entity

4 Manolessou & Horrocks (2007: 228) note that predicative nominals lack the article throughout the history of Greek.

8 0 Invited Papers Sophia Marmaridou

the subject of the sentence is, i.e. pseftis (a liar). By contrast, in (20), the noun head, modified by a relative clause, unifies with a determiner to form a phrasal constituent. In this case the predicate nominal specifies an instance of the type fititis (student), i.e. one that is experiencing a particular state. For the purposes of the present discussion it is assumed that the two types of predicate nominals instantiate two different constructions, i.e. the Type Nominal Construction and the Instance Nominal Construction, respectively. Proper nouns follow this pattern, as examples (5)-(9) indicate above, the first four illustrating the referential function and the last one the predicative function. Non-referential functions are also attested extra-sententially as vocatives and across verbs of naming, as the following examples indicate:5

(21) Katina, ise monaδiki. Katina, be-2nd.sg.pr unique ‘Katina, you are unique!’ (HNC 2421300) (22) lejete pavliδis leanδros called-3rd.sg.pr.pass Pavlidis Leandros ‘He is called Pavlidis Leandros.’ (GkWaC 689)

Focusing on definiteness as a complement taking category within the framework of formal syntax, Anderson (2004: 441) observes that proper names as arguments are complements of the definite article in Modern Greek, whereas in vocatives and across naming contexts, as in (21) and (22) above, definiteness is not assumed; rather identity is assigned and therefore the definite article is absent. He further concurs that, when used as arguments, or referential terms, they constitute the primary means of establishing the identity of arguments, i.e. of ensuring that both speaker and addressee know the identity of a particular argument (ibid: 439). There are at least two problems with this analysis. The first concerns the relation between definiteness and reference. If definiteness is defined as establishing identity, as in (5) above, it is not clear how it differs from assigning identity, as in (21) and (22), both of which, according to Anderson, do not express definiteness. Apparently, in both (5) and (21) speaker and addressee know the identity of the referent bearing the name. In (22) the identity of the referent is pragmatically available to speaker and addressee, possibly from previous discourse, but now it is simply paired with a name. If definiteness applies to referents known to both speaker and addressee, then

5 Even though vocatives are not used to refer to entities whose existence is presupposed, they do confer addressee status to such entities, thereby establishing them as referents and enabling future reference to them. This point will not be further addressed in the present discussion.

Invited Papers 8 1 Major Trends in Theoretical and Applied Linguistics

it characterizes the use of names in both (5) and (21) and no distinction can be drawn between these examples on the basis of this property. The difference between (5) on the one hand and (21) and (22) on the other seems to lie in the referential use of the name in (5) and the non-referential uses in (21) and (22), rather than in definiteness itself. Secondly, Anderson (2004) offers no account of the co-occurrence of proper names with indefinite determiners. Working within a formal syntactic model of analysis, in which definiteness, but not indefiniteness, is a complement taking phrasal head, Anderson (ibid.) overlooks the fact that not only as complements of the definite article, but also when preceded by indefinite determiners do proper names have a referential function, as example (6) illustrates above. As also argued elsewhere for Modern Greek (see Marmaridou 2012) and following Langacker (1991), in such cases, as in all instances of indefinite reference, the indefinite nominal indicates that the speaker, and possibly the addressee, does not have mental contact with the referent the name identifies.6 In what follows, an attempt will be made to provide an account of some indefinite uses of personal proper names in Modern Greek following the Construction Grammar framework. It will be argued that the uses illustrated in (6)-(8) above inherit the Indefinite Nominal Construction in the language, but (7), (8), and (23) below are further constrained semantically and pragmatically. More specifically, uses like those illustrated in (7) and (8) make reference to frame metonymy (see Sweetser and Fauconnier (1996)) as part of the semantics of the construction, while the semantics and pragmatics of the use illustrated in (23) depend on a conceptual blend of relevant mental spaces (after Fauconnier (1997):

(23) oste na anaδiθi kapjia stiγmi enas Mozart i enas Picasso so that to emerge-3rd.sg.prs.pass some moment a Mozart or a Picasso ‘so that a Mozart or a Picasso emerge at some point’

On the basis of the proposed analysis it will be argued that, even though proper nouns differ from common nouns in that the understanding of the former depends on much richer frames than that of the latter, both categories function as heads in the same nominal constructions in the language, which are further elaborated into more specific constructions accounting for their various uses.

6 For the view that the definite article in Modern Greek is a semantically neutral marker of reference and compatible with pragmatically established definiteness, whereas indefinite determiners are semantically marked for indefiniteness see Marmaridou (1984), and (2012) for a modified view concerning the indefinite article only. For a brief overview of definiteness in formal syntax, see Marmaridou (2011). For the view that enas/mia/ena is a marker of indefiniteness rather than an article in Modern Greek, see Chila-Markopoulou (2000).

8 2 Invited Papers Sophia Marmaridou

5. Indefinite Proper Noun Constructions in Modern Greek

For the purposes of the analysis, it will be assumed that the Indefinite Nominal Construction in Modern Greek, formalized in Figure 1 below, inherits the Instance Nominal Construction, as specified in section 4 above. As already mentioned, according to Construction Grammar notation, inner boxes represent the constructional specifications of the constituent parts, whereas the outer box represents the whole construction, which need not be a sum of the constituent parts, but may be characterized by additional features specific to the construction as a whole.

The Indefinite Nominal Construction

frame indef ' S and/or H in cat n Ti insufficient gender # 1 [ ] proper↓4[ ] ↓3 [...] syn sem prag mental contact morph case # 2 [ ] max + cnfg count with Ti ' num sg lex - num sg bounded +

role det role head cat n head cat det proper ↑4 [ ] syn max - syn lex + level max - lex +

frame indef frame ↑3 [...] cnfg count cnfg [ ] sem sem num sg num sg bounded + bounded [ ]

gender # 1 [ ] gender # 1 [ ] case # 2 [ ] case # 2 [ ] morph morph num sg num sg

lxm enas/mia/ena lxm [...]

Figure 1. The indefinite nominal construction in Modern Greek

One of the claims made in Figure 1 at the syntactic level is that both common and proper nouns may be nominal heads, hence the unspecified value of the syntactic head for the attribute [proper]. Moreover, each of the constituent constructions, i.e. the nominal head and the determiner, has lexical, but not

Invited Papers 8 3 Major Trends in Theoretical and Applied Linguistics

phrasal status, hence they are both characterized as [lex +] [max -], i.e. they are lexical but not phrasal constituents. At the semantic level, the semantic frame of the nominal head may take a number of values depending on the lexical choice of the head. As already mentioned, a common noun has meaning in terms of the semantic frame it activates and denotes a category, all members of which are characterized by the same frame.7 A proper noun has meaning in terms of a category-specific semantic frame (e.g. human being, city, animal, etc.), which is encyclopaedically enriched to the extent that in that frame only one member of the category can be identified. This seems to be the basis of the uniqueness requirement for the use of proper names. For example, the use of the name Mozart activates a frame of classical music in which a human being is uniquely identified as a famous composer. Apparently, proper nouns are associated with much richer frames than common nouns, in that they activate both category membership, or, in Langacker’s (1991) terms, type-specification, and rich encyclopaedic information, e.g. classical music, composers, etc., which need not be activated by respective lexical frames, e.g. the words ‘music’, ‘composer’, etc. As Dancygier (2011: 211) observes, if this information is not available to the addressee, he or she will not be able to pair the name with a referent, even though they will recognize the particular form as a proper name. Unlike Dancygier (ibid.), though, I further claim that a proper noun head of a construction activates an expectation of uniqueness, even when frame information is not available. In other words, it is part of cultural knowledge that, if a word form is recognized as a proper noun, it is expected that it should identify a unique referent in some frame or other. The use of a proper noun, referential or otherwise, may vary depending on the construction of which it is part, but the frame is activated in all of them. For example, in Modern Greek a frame is activated by a proper noun in both (either definite or indefinite) referential and non-referential uses, as shown in all examples above. A further claim made in Figure 1 is that, even though the nominal head itself may be semantically unspecified for number, count status, and mass/unit attribute values, when it unifies with an indefinite determiner, it selects for the attribute values of the determiner. In the examples above, the indefinite determinerenas /

7 Lexical analysis in terms of semantic frames originates in Fillmore (for an overview see Fillmore (1985)), who claimed that our knowledge of any linguistic form is available to us in connection with some personally meaningful setting or situation. These settings or situations comprise scenes which are associated with certain linguistic frames, so that scenes and frames activate each other in the language user’s memory and are hence mutually retrievable. It follows that the process of communication involves the activation within speakers and across speakers of linguistic frames and cognitive scenes. The implication of this position is that in lexical analysis a word cannot be separated from its context if an adequate account of its meaning is to be given.

8 4 Invited Papers Sophia Marmaridou

mia/ena selects for singularity, countability, and boundedness values, which also characterize the proper noun heads in the respective utterances. To the extent that these attribute values have morphological correlates, the latter are also specified for both the nominal head and the determiner in the respective constituent constructions. It follows from all the above and the outer box in Figure 1 that the Indefinite Nominal Construction is syntactically a phrasal, [max +], non-lexical, [lex -], constituent, headed by a common or proper noun, which semantically identifies

an instance of a type [Ti], a singular, countable and bounded entity, with which, at a pragmatic level, the speaker and/or the hearer are in insufficient mental contact (following Langacker 1991).8 On the basis of this analysis, I would like to claim that the use of the proper name in (6) above directly instantiates the Indefinite Nominal Construction in that it constitutes a phrasal syntactic category, which semantically activates a frame in which an instance of the type ‘human being’ is identified as bearing the culturally available family name Papavlasopoulos, morphologically marked as a male, singular, bounded entity. The question following the name expression (‘who is he?’) in example (6) confirms that the speaker, and possibly the addressee, has no mental contact with this person. As already mentioned, the use of the name in (7) does not identify a person who is unknown to the speaker or the addressee. Therefore, indefiniteness cannot apply to the referent. I would like to suggest that the semantics of the indefinite construction is further characterized by frame metonymy, as discussed in Sweetser & Fauconnier (1996) in relation to examples such as The ham sandwich wants his check. It has been argued that the use of the ham sandwich for the customer who had a ham sandwich is not simply a case of part-for-whole metonymy, whereby the ham sandwich is a part for what the whole customer activity involves, but rather both the meal and the customer are parts of a restaurant frame, including customers, meals, checks, tables, waiters, etc. The authors argue that, because this frame is available to speakers of the language, is it possible to metonymically use the meal for the customer, or even the table for the customer, as in Table 3 wants his check, for example. The frame of politics activated by Karamanlis in example (7), in which the name identifies a unique referent, includes specific information about this referent, namely that this person is generally a mild, easy-going, almost indifferent politician. This view is probably established by reports of his behaviour on

8 Strictly speaking, the semantic attribute value [Ti] is specified in the Instance Nominal Construction which, as already mentioned, is inherited by the whole Indefinite Nominal Construction. This feature is incorporated in Figure 1 for ease of reference.

Invited Papers 8 5 Major Trends in Theoretical and Applied Linguistics

several occasions requiring political decisions, activities, collaborations, etc. What (7) implies is that Karamanlis’ behaviour on a particular occasion was different from the typical one, so that now he appeared demanding.9 Apparently, the proper name does not only activate one among several contexts in which it is appropriately paired with a specific referent, as Marmaridou (1989:363) and Dancygier (2009: 166) have argued; rather, given that the frame in which Karamanlis is paired with a specific referent also includes the referent’s views, behaviour, decisions, etc., the proper name metonymically stands for this referent’s behaviour on a particular occasion. In short, the name activates a frame in which a referent is selected to provide access to his behaviour on a particular occasion. Note that this indefinite nominal construction is used to express the speaker’s assessment of the situation described. According to the author’s view, on a particular occasion Karamanlis behaved contrary to expectations. I would like to claim that this evaluation of the situation indicates a degree of subjectivity, i.e. speaker/author involvement as a pragmatic attribute value of this construction. Example (8) can be similarly analysed. Each name activates a corresponding frame in which a specific referent is identified as a famous painter. The frame also includes associated information about the outcome of a painter’s work, style, market price, originality, etc. In this frame, the name of the painter metonymically stands for each one of his/her works. The conventionality of this metonymy to refer to paintings is worth noting: one does not expect to find similar metonymic references to musicians’ or poets’ works (in fact, the corpora consulted confirm this view), even though such instances cannot be excluded. It is not inconceivable that one could say something like “exi ena afθendiko Skalkota sti siloji tu” (‘s/he has an authentic Skalkotas in his/her collection’) to refer to an original score by the famous Greek musician. Moreover note that the conventionality of the metonymy in (8) prompts for the understanding that reference is made to each famous painter’s painting, even though it is clearly stated in the rest of the utterance that each of these paintings is a forgery. The example in (23) differs from those in (6)-(8) in that, while each proper name activates a frame in which the name identifies a unique referent, Mozart and Picasso, each referent intended in (23) is other than the one typically associated with this name. The utterance, of which (23) is a part, constitutes an argument for the recognition of gifted children’s talent and their guidance by specialists in a field, so that musicians like Mozart, or painters like Picasso, can

9 The meaning of δiaforetikos (different) and alos (other) in collocations with prosopo (face) to refer to aspects of one’s character and behaviour is also discussed in Marmaridou (2011).

8 6 Invited Papers Sophia Marmaridou

eventually emerge. Clearly, (23) is an evaluative statement making reference to artistic quality associated with Mozart in the domain of music and Picasso in that of painting. It could be argued that the name in these cases metonymically stands for artistic quality in the respective domains. In fact, this is the line taken by Barcelona (2004) in relation to predicate uses of paragon names, as in That young man is a real Shakespeare, i.e. a great playwright. However, in (23) the names do not simply stand for qualities. Rather the use of each name in this example makes reference to a specific, yet unknown and hence indefinite, child that is gifted by Mozart’s or Picasso’s talent. In this case, the frame evoked by each proper name is applied to another referent selectively, in the sense that not everything we know about Mozart or Picasso is mapped on each of the children (Dancygier 2011: 215), but only the culturally entrenched and contextually motivated feature of artistic excellence. To explain this use, reference will be made to mental spaces and conceptual blends, after Fauconnier (1997). Both of these theoretical notions relate to meaning construction as resulting from high-level complex mental operations that apply within and across domains when we think, act, or communicate. The domains include background cognitive models (e.g. film-making, politics, art, music, etc.) as well as locally introduced mental spaces. Local spaces may be introduced by various linguistic expressions, including proper names, which set up elements in the discourse and can also associate such elements with properties like “having the name Mozart”, “being an excellent music composer”, etc. A connection of elements between two mental spaces, say (R) and (R΄), enables them to function as inputs to a third space that “blends” elements from the other two. Conceptual blending is a cognitive process operating on two input mental spaces to yield a third space, the blend. The blend inherits partial structure form the input spaces and, importantly, has emergent structure of its own (ibid: 149). As already mentioned, the individuals to whom reference is made in (23), the gifted children, are not those typically identified by the respective names. Following Marmaridou (2012), I put forth the view that this is an instance of a conceptual blend, in the sense of Fauconnier (1997), motivating the semantics of the nominal construction and represented in Figure 2 below. In one of the input spaces, (R), the two names, Mozart and Picasso, activate frames of classical music and modern art, respectively, in which two unique individuals are identified, the one associated with the property of being an excellent music composer and the other with that of being an excellent modern painter. In the second input space (R΄) there is a set of indefinite, unspecified individuals. A connector links Mozart and Picasso in the first input space with indefinite, unspecified individuals in the second input space. In the blend there is a set of indefinite, unspecified individuals, one of whom is associated with the property of being an excellent music composer and the other with that of being an excellent painter. A pragmatic constraint of this construction is that the proper

Invited Papers 8 7 Major Trends in Theoretical and Applied Linguistics

R R΄ • child 1 Mozart = • • child 2 • excellent music • child 3 composer . • Picasso = . • excellent painter • child n

•child 1 = •excellent music composer •child 2 = Blend •excellent painter •child n = ...

Figure 2. The blending semantics of the Indefinite Proper Name Construction instantiated in (23)

name(s) used must be culturally entrenched. Moreover, the utterance in (23) is used to argue for the benefits of early recognition of talent and in that sense the use of the names is associated with argumentative discourse. Importantly, each indefinite noun phrase expresses the speaker’s subjective evaluation of an unspecified individual to whom reference is actually made. It follows that the noun phrases in (23) instantiate a construction, i.e. the Indefinite Proper Noun Construction, as formalized in Figure 3, which inherits the Indefinite Nominal Construction, and is further specified for “blending” semantics, the cultural pragmatics of the proper name used, argumentative discourse type, and speaker involvement in expressing a subjective view. One of the claims formalized in Figure 3 is that it syntactically concerns the category of proper nouns. Moreover, in the proposed construction, the referent in

space (R), i.e. the instance of the type named, Ti (e.g. Mozart, Picasso), is culturally significant, as the unification index #1 indicates in the semantic and pragmatic specifications of the construction.10 Furthermore, unification index #2 shows

that the property associated with the Ti named (i.e. that of being an excellent composer or an excellent painter) is projected into the blend. More importantly for the understanding of the blend, the claim is also made that, by associating

the indefinite Ti in space (R΄) with the particular property of excellence, the

10 Strictly speaking, the cultural significance of the name used follows from the encyclopaedically rich frame activated by the proper noun.

8 8 Invited Papers Sophia Marmaridou

The Indefinite Proper Noun Construction inherit Indefinite Nominal Construction

R # 1 T named salient indef i sem associated_property # 2 [...] blend Ti property # 2 [...]

R΄ # 3 T indef [ i ]

cultural knowledge # 1

cat n argumentative prag d-frame text type proper + persuasive syn max + lex -

subjective_assessment # 3 involvement expressivity

Figure 3. The indefinite proper noun construction in Modern Greek

speaker is making a subjective assessment of this indefinite Ti, i.e. the speaker expresses his/her judgement that these children will become famous artists (if appropriately discovered and trained), hence the unification index #3. This judgement, instantiating speaker involvement, is used to argue for the need to discover such talented children early on in their lives; hence this construction is pragmatically motivated by the argumentative type of the discourse in which it occurs. Overall, Figure 3 shows that, apart from all the features that the proposed proper noun construction shares with the Indefinite Nominal Construction in the language, it is further characterized by additional semantic and pragmatic features comprising a conceptual blend operating on the basis of cultural information, as well as by discourse type and speaker involvement.

6. Conclusions

In this study an attempt has been made to analyse some indefinite uses of proper names in Modern Greek adopting a Construction Grammar framework and notation. On the basis of corpus examples it was argued that proper nouns, like

Invited Papers 8 9 Major Trends in Theoretical and Applied Linguistics

common nouns in the language, unify with definite and indefinite determiners to establish referents in discourse. Their semantic difference lies in that, while a common noun activates a semantic frame that applies to all members of the type identified, a proper noun activates a frame in which a referent is uniquely identified on every occasion of use. Therefore, the uniqueness criterion of proper nouns is satisfied in the activation of such frames. For example, in (5) above, the proper noun activates the frame of politics in which a unique referent is identified. Note that, if more than one referent is identified by that name, for example when there are several members of the same family currently involved in the domain of politics, the actual referent intended must be pragmatically/ contextually specified, as is the case with definite reference in general. As already mentioned, particular attention has been paid to indefinite uses of proper names, partly because their frequent occurrence challenges the long held view that proper names have definite reference by definition, but also because indefinite uses do not evoke the same understanding on each occasion. More specifically, the indefinite proper name sometimes identifies an unknown referent, as in (6) above, or the same referent on different occasions, as in (7), or a referent other than the one typically identified by the proper name, as in (8), or (23). Given such variability in the use of indefinite proper names, the present investigation proposes a unifying treatment of such uses within the framework of Construction Grammar. The analysis has shown that indefinite, proper noun headed constructions on the one hand inherit all those attribute values that characterize the Indefinite Nominal Construction in Modern Greek, as in (6) above, and, on the other, some of them are characterized by frame metonymy, as in (7) and (8), or are further specified semantically by conceptual blending and pragmatically by cultural knowledge, speaker involvement, and discourse type, as in (23). Moreover, it was shown that some indefinite proper noun headed constructions, such as the ones exemplified in (7) and (23), express subjective involvement and rate high in expressivity. Notably, these constructions, as special instances of the Indefinite Nominal Construction, license less conventional constructs in the language than those represented in (6) and (8). This fact relates to the investigation of the Indefinite Nominal Construction in Modern Greek by Marmaridou (2012), who has argued that the indefinite determiner enas/mia/ena, semantically specified for indefiniteness and singularity, also contributes to speaker involvement and expressivity in a family of less conventional indefinite constructions in the language, including the Indefinite Proper Noun Construction exemplified in (23). On the basis of studies that view subjectivity, expressivity, and synchronic variability as prompts for language change (e.g. Hoffmann and Trousdale (2011)), it has also been argued that the specification of this indefinite determiner for these properties is conducive to its change into a semantically neutral article in Modern Greek. In view of the above, I would like to propose that the proper noun headed

9 0 Invited Papers Sophia Marmaridou

construction licensing constructs like (7) above also belongs to this family of indefinite constructions that contributes to the synchronic variability of enas/ mia/ena. Evidently, this claim further supports the view put forth in Marmaridou (2012) concerning the grammatical specification of the indefinite determiner in the language. At least two implications follow from the proposed analysis. The first concerns the usefulness of a corpus-based approach to the data. Evidently, the identification of proper name uses has been facilitated by systematic corpus search, which has also revealed interesting frequencies of proper noun constructions. An informal count of seven names of politicians and artists indicates that indefinite uses constitute 17% of all their occurrences. A corpus-based approach has also enabled the identification of the contextual, discourse driven parameters regularly characterizing proper noun headed constructions, which would have otherwise gone unnoticed. The second implication relates to the appropriateness of Construction Grammar for an account of a linguistic category such as proper names. More specifically, the investigation of indefinite uses of proper names has necessitated the specification of a number or parameters at various levels of constructional analysis, which jointly characterize these uses. The Construction Grammar framework has been shown to provide the means for such integration since, by definition, a construction incorporates information at all levels. Moreover, given that it is a usage-based theoretical model, it can also incorporate insights from other usage-based approaches to linguistic analysis, such as frame semantics, extended to apply to frame metonymy, and mental space theory. It is certainly not a coincidence that both Construction Grammar and the theory of Conceptual Blends aim to account for the understanding of linguistic expressions whose meaning goes well beyond the meaning of the constituent parts, the latter represented as constituent constructions in the one model and input spaces in the other. The need to account for the systematicity of complex constructional patterns that do not simply constitute the sum of their parts can no longer be overlooked by linguistic theory.

Invited Papers 9 1 Major Trends in Theoretical and Applied Linguistics

References

Allerton, D.J. 1987. The linguistic and sociolinguistic status of proper names. What are they, and who do they belong to? Journal of Pragmatics 11: 61-92.

Anderson, J.M. 2004. On the grammatical status of names. Language 80/3: 435- 474.

Antonopoulou, E. and K. Nikiforidou. 2011. Construction grammar and conventional discourse: A construction-based approach to discoursal incongruity. Journal of Pragmatics 43: 2594-2609.

Barcelona, A. 2004. Metonymy behind grammar: The motivation of the seemingly “irregular” grammatical behaviour of English paragon names. In G. Radden and K.-U. Panther (eds.), Studies in Linguistic Motivation. Berlin and New York: Mouton de Gruyter, 357-374.

Bolinger, D.L. 1951. Intonation: Levels versus configurations. Word 7/3(1): 199- 110.

Braine, M.D.S. and P.J. Brooks. 1995. Verb argument structure and the problem of avoiding an overgeneral grammar. In M. Tomasello and W.E. Merriman (eds.), Beyond names for things: Young children’s acquisition of verbs. Hillsdale, NJ: Lawrence Erlbaum, 353-376.

Broccias, C. 2008. Cognitive linguistic theories of grammar and grammar teaching. In S. De Knop & T. De Rycker (eds.), Cognitive Approaches to Pedagogical Grammar. Berlin and New York: Mouton de Gruyter, 67-90.

Carroll, J.M. 1983. Toward a functional theory of names and naming. Linguistics 21: 341-371.

9 2 Invited Papers Sophia Marmaridou

Chafe, W. 1994. Discourse, consciousness, and time: The flow and displacement of conscious experience in speaking and writing. Chicago IL: University of Chicago Press.

Chila-Markopoulou, D. 2000. The indefinite article in Greek. A diachronic approach. Glossologia 11/12: 111-130.

Croft, W. and D.A. Cruse. 2004. Cognitive Linguistics. Cambridge: Cambridge University Press.

Dancygier, B. 2009. Genitives and proper names in constructional blends. In V. Evans and S. Pourcel (eds.), New Directions in Cognitive Linguistics. Amsterdam and Philadelphia: John Benjamins, 161-184.

Dancygier, B. 2011. Modification and constructional blends in the use of proper names. Constructions and Frames 3/2: 208-235.

Donnellan, K.S. 1975. Proper names and identifying descriptions. In D. Davidson and G. Harman (eds.), The Logic of Grammar. California: Dickenson Publishing Company Inc., 356-379.

Fauconnier, G. 1997. Mappings in Thought and Language. Cambridge: Cambridge University Press.

Fillmore, C. 1985. Frames and the semantics of understanding. Quaderni di Semantica 6: 222-254.

Fox, B.A. and S.A. Thomson. 2007. Relative clauses in English conversation: relativizers, frequency and the notion of construction. Studies in Linguistics 31/2: 293-326.

Frege, G. 1960[1892]. On sense and reference. In P.T. Geach and M. Black (eds.), Translation from the Philosophical Writings of Gottlob Frege. Oxford and New York: Oxford University Press, 40-65.

Fried, M. 2009. Construction Grammar as a tool for diachronic analysis. Constructions and Frames 1/2: 262-291.

Fried, M. and J.-O. Östman. 2003. The explicit and the implicit in the Suasion Frame. In E. Hajičová, A. Kotěšovoková, and J. Mírovský (eds.), Proceedings of CIL17. CD-ROM. Prague: Matfyzpress.

Invited Papers 9 3 Major Trends in Theoretical and Applied Linguistics

Fried, M. and J.-O. Östman. 2004. Construction Grammar. A thumbnail sketch. In M. Fried and J.-O. Östman (eds.), Construction Grammar in a Cross-Language Perspective. Amsterdam and Philadelphia: John Benjamins, 11-86.

Fried, M. and J.-O. Östman. 2005. Construction Grammar and spoken language. The case of pragmatic particles. Journal of Pragmatics 37: 1752-1778.

Gisborne, N. 2011. Constructions, Word Grammar and grammaticalization. Cognitive Linguistics 22/1: 155-182.

Goldberg, A. 1995. Constructions. A Construction Grammar Approach to Argument Structure. Chicago: Chicago University Press.

Goldberg, A. 2006. Constructions at Work. The Nature of Generalization in Language. Oxford: Oxford University Press.

Hawkins, J.A. 1978. Definiteness and Indefiniteness. A Study in Reference and Grammaticality Prediction. London: Croom Helm.

Hoffmann, T. and G. Trousdale. 2011. Variation, change and constructions in English. Cognitive Linguistics 22/1: 1-23.

Holton, D., P. Mackridge and I. Philippaki-Warburton.1999. Γραμματική της Ελληνικής [Grammar of Greek]. Athens: Patakis Publications.

Hopper, P.J. and E.C. Traugott. 1993. Grammaticalization. Cambridge: Cambridge University Press.

Huddleston, R. 1988. English Grammar: An Outline. Cambridge: Cambridge University Press.

Johnson, C.R., C.J. Fillmore, E.J. Wood, J. Ruppenhofer, M. Urban, M. Petruck, and C.F. Baker. 2001. The FrameNet Project: Tools for Lexicon Building (Version 0.7, January 2001). Berkeley, CA: International Computer Science Institute.

Joseph, B. and I. Philippaki-Warburton. 1987. Modern Greek Grammar. London: Croom Helm.

Kay, P. 1997. Words and the Grammar of Context. Stanford, CA: CSLI.

Kripke, S.A. 1972. Naming and necessity. In G. Harman and D. Davidson (eds.), Semantics of Natural Language. Dordrecht: Reidel, 253-355.

9 4 Invited Papers Sophia Marmaridou

Lambrecht, K. 1994. Information Structure and Sentence Form. Cambridge, MA: Cambridge University Press.

Langacker, R.W. 1991. Foundations of Cognitive Grammar. Volume 2: Descriptive Application. Stanford: Stanford University Press.

Lehrer, A. 1992. Names and naming: Why we need fields and frames. In A. Lehrer and E.F. Kittay (eds.), Frames, Fields, and Contrasts. Hillsdale, N.J.: Lawrence Erlbaum, 123-142.

Leino, P. and J.-O. Östman. 2005. Language change, variability, and functional load. In J. Leino (ed.), Constructional Reorganization. Amsterdam and Philadelphia: John Benjamins, 37-54.

Mackridge, P. 1985. The Modern Greek Language. Oxford: Clarendon Press.

Manolessou, I. and G. Horrocks. 2007. The development of the definite article in Greek. Studies in Greek Linguistics – In memoriam A.-F. Christidis, 224-236.

Marmaridou, S.A.S. 1984. The study of reference, attribution and genericness in the context of English and their grammaticalization in M. Greek noun phrases. Unpublished PhD dissertation, University of Cambridge, UK.

Marmaridou, S.A.S. 1989. Proper names in communication. Journal of Linguistics 25/2: 355-372.

Marmaridou, S.A.S. 1991. What’s so Proper about Names? Athens: Parousia Journal Monograph Series 15.

Marmaridou, S. 2011. The relevance of embodiment in lexical and collocational meaning: the case of prosopo in Greek. In Z. Maalej and N. Yu (eds), Embodiment via Outer Body Parts: Studies from Various Languages and Cultures. Amsterdam and Philadelphia: John Benjamins, 256-289.

Marmaridou, S. 2012. The constructional motivation of indefinite generics in Modern Greek. Constructions and Frames 4/2.

Mill, J.S. 1956[1843]. A System of Logic. New York: Longman.

Nikiforidou, K. 2010. Discoursal categories and grammatical description: A constructional integration. Paper presented in plenary at the 6th International Conference on Construction Grammar, Prague, 3-5 September, 2010.

Invited Papers 9 5 Major Trends in Theoretical and Applied Linguistics

Nikiforidou, K. 2011. Grammar and Discourse: A Constructional Approach to discourse-based Conventionality. Athens: Parousia Journal Monograph Series, 81.

Pawley, A. 1987. Encoding events in Kalam and English: different logics for reporting experience. In R.S. Tomlin (ed.), Coherence and Grounding in Discourse. Amsterdam and Philadelphia: John Benjamins, 329-360.

Pawley, A. and F.H. Syder. 1983. Two puzzles for linguistic theory: native-like selection and native-like fluency. In J.C. Richards and R.W. Schmidt (eds.), Language and Communication. London and New York: Longman, 191-227.

Peters, A.M. 1983. The Units of Language Acquisition. Cambridge: Cambridge University Press.

Putnam, H. 1975. Mind, Language and Reality. Cambridge: Cambridge University Press.

Russell, B. 1905. On denoting. Mind 14: 479-493.

Searle, J. 1969. Speech Acts. Cambridge: Cambridge University Press.

Sperber, D. and D. Wilson. 1986. Relevance: Communication and Cognition. Oxford: Basil Blackwell.

Strawson, P. F. 1950. On referring. Mind 59: 320-344.

Sweetser, E. and G. Fauconnier 1996. Cognitive links and domains: Basic aspects of mental space theory. In G. Fauconnier and E. Sweetser (eds.), Spaces, Worlds and Grammar. Chicago and London: The University of Chicago Press, 1-28.

Terkourafi, M. 2010. Don’t go V-ing in Cypriot Greek: Semantic, pragmatic, and prosodic aspects of a prohibitive construction. Constructions and Frames 2/2: 208-241.

Tomasello, M. 1992. First verbs: A Case Study of Early Grammatical Development. Cambridge, MA: Cambridge University Press.

Tomasello, M. 2000. Do young children have adult syntactic competence? Cognition 74: 209-253.

Traugott, E.C. and R.D. Dasher. 2002. Regularity in Semantic Change. Cambridge: Cambridge University Press.

9 6 Invited Papers Sophia Marmaridou

Valenzuela Manzanares, J. and A.M. Rojo López. 2008. What can language learners tell us about constructions? In S. De Knop and T. De Rycker (eds.), Cognitive Approaches to Pedagogical Grammar. Berlin and New York: Mouton de Gruyter, 197-230.

Van Langendonck, W. 1999. Neurolinguistic and syntactic evidence for basic level meaning in proper names. Functions of Language 6/1: 95-138.

Vandelanotte, L. and P. Willemse. 2002. Restrictive and non-restrictive modification of proprial lemmas. Word 53/1: 9-36.

Waara, R. 2004. Construal, convention, and constructions in L2 speech. In M. Achard and S. Niemeier (eds.), Cognitive Linguistics, Second Language Acquisition, and Foreign Language Teaching. Berlin and New York: Mouton de Gruyter, 51-75.

Zwicky, A.M. 1974. Hey, whatsyourname!, Papers from the Tenth Regional Meeting of the Chicago Linguistic Society. Chicago: Chicago University Press, 787-801.

Invited Papers 9 7 Major Trends in Theoretical and Applied Linguistics

9 8 Aspect and Aktionsart: A Study on the Nature of Grammatical Categories*

Amalia Moser

University of Athens [email protected]

Abstract

The topic of this paper seems perhaps over-ambitious, touching as it does on several questions which have become serious theoretical issues over the last decades and some of them over several centuries. The controversy on the nature of grammatical categories has marked Philosophy and Linguistics throughout the twentieth century; the more modest controversy on aspect and its relationship to the disputed category Aktionsart, as well as its relationship to tense, has been dealt with in a vast amount of literature. This short paper cannot provide a definitive answer; it has the more modest aim of showing a) that Greek aspect and Aktionsart form a continuum b) that the two categories remain nevertheless distinct and thus hopefully contribute to the broader discussion on gradience, in particular intersective gradience, as defined by Aarts (2007).

1. The Discreteness of Categories

Aristotle, with his clarity of thought and passion for precision, shaped the way in which the concept of category was almost uniquely understood until the 20th century: a category is defined by the necessary and sufficient conditions which determine, unequivocally, whether each entity belongs or not to the category; tertium non datur. Classical Logic never digressed from this principle, even though there were some dissenting voices even in Antiquity.1

* My warmest thanks to the organising committee for inviting me to present this paper, to the linguists in the audience for their stimulating questions and comments, to the anonymous reviewers and to all the other colleagues and students who have provided me with thought- provoking material, especially the objections which led to the fuller development of the theoretical arguments put forward in this paper. 1 Eubulides of Miletus, from the school of Euclides of Megara, with his sorites paradox, is the best-known example (cf. e.g. Aarts et al. 2004: 2).

Invited Papers 9 9 Major Trends in Theoretical and Applied Linguistics

These voices were few and far apart because at fi rst sight (that is, for the fi rst twenty-four centuries or so) this concept seemed to function perfectly well for natural classes, especially for the taxonomy of plants and animals. The biological hierarchy kingdom > class > order > genus > species > variety (as systematized by Linnaeus in the 18th century and with a few later amendments shown in Figure 1) took excellent care of the natural world. To take an example, the phylum of vertebrates was subdivided into classes, with the possibility of intermediate superclasses (Figure 2):

vertebrates (phylum)

[oviparous] [viviparous] [superclasses]

fishes birds mammals reptiles amphibians (classes)

Figure 1. Biological Classifi cation2

vertebrates (phylum)

[oviparous] [viviparous] [superclasses]

fishes birds mammals reptiles amphibians (classes) Figure 2. Traditional classes of vertebrates

In the 19th century, however, new data started coming in. The discovery of Australia, for instance, brought along the discovery of animals such as the platypus, which is at the same time a mammal and an egg-laying animal, a fact also mirrored in its appearance, as it is both furry and beaked.

2 Peter Halasz, based on Carl R. Woese, Otto Kandler, Mark L. Wheelis (1990) Towards a natural system of organisms: proposal for the domains Archaea, Bacteria, and Eucarya. Proceedings of the National Academy of Sciences of the United States of America 87: 4576-4579 ( the article in which the level of ‘domain’ is proposed). The representation was downloaded from Wikipedia (http://en. wikipedia.org/wiki/File:Biological_classifi cation_L_Pengo_vfl ip.svg ).

1 0 0 Invited Papers

Amalia Moser

Several other fi ndings, such as that vertebrates are closely related to non- vertebrate but craniated species, as well as non-craniated animals, led to the postulation of several intermediate categories, on which, incidentally, there is no absolute consensus among biologists: a superphylum of deuterostomia and a phylum of chordates, which includes vertebrates among other subphylums (tunicata, cephalochordata). The platypus and the echidna had to be accommodated into a sub-class of their own, thus seriously upsetting the symmetry of the model:

vertebrates [sub-phylum]

[fishes] [tetrapods] [superclasses]

agnatha amphibians reptiles [classes] osteichtyes mammals birds chondrichthyes ?non-monotremes monotremes [ sub-classes ]

Figure 3. Categorization of vertebrates

Ironically, therefore, it was progress in the natural sciences, the major supporters of classical categorization, that created the most serious rifts in the conception of categories. It showed that, rather than being discrete and clearly defi ned, even natural categories could be vague. The vigorous philosophical discussions that this development generated led to the theory of prototypes; Tsohatzidis (1990: 1) provides one of the most succinct and lucid descriptions in the literature:

[…] elements are assigned to a category not because they exemplify properties that are absolutely required of each one of its members, but because they exhibit to a greater or lesser extent […] certain types of similarity with a particular category member that has been (naturally or culturally) established as the best example (or prototype) of its kind.

Later developments led to defi nitions allowing for the possibility of the prototype being an abstraction based on (the attributes of) a number of focal members of a category; in the words of Jaszczolt (2002: 34) “the core prototype can be regarded as a real example or an abstract set of characteristic features”. Currently in Linguistics both approaches fi nd equally staunch supporters, with the two main theories of language, generative and cognitive, respectively adopting discreteness and vagueness or fuzziness.

Invited Papers 1 0 1 Major Trends in Theoretical and Applied Linguistics

The position adopted here is halfway between the two. It is maintained that the perception of category membership, and by extension of categories, is prototypical; both the relevant philosophical arguments and the psychological experiments (even the earliest ones, e.g. Rosch 1973, 1978) seem conclusive. Nevertheless, the extreme position of many advocates of gradience that there is no possibility of defining even prototypical members in a strict manner is not adhered to – instead, it is proposed that it is not only possible but desirable to define prototypical members in as rigid a way as possible; this would be compatible with the ease with which they are recognized. Paradoxically, perhaps, both of these convictions, which seem contradictory, stem from the same beliefs about the properties of the human mind. As Saussure and classical Structuralism have shown, categories are not imposed by reality, but are the result of our own decisions, within our attempt to put some order to the world around us. Not having a clear idea of at least the prototypical members of these categories would defeat the purpose of postulating them. From the 1980s onwards there has been a growing tendency to attribute prototypicality to grammatical categories. The seminal article by Hopper & Thompson (1984) on “The discourse basis for lexical categories in Universal Grammar” was particularly influential;3 it was followed by Taylor’s 1989 book on Linguistic Categorization, while most major cognitive works attribute fuzziness and prototypicality to basic grammatical notions, such as those of word, phoneme, verb, noun, adjective etc., often using for them the term ‘radial categories’. A belief in prototypicality leads naturally to a belief in the existence of continua. The fuzziness of categories, providing as it does for different degrees of membership as well as for membership of multiple categories, allows chains of entities linking different categories. Again, however, the view adopted here is that, if continua are to make any sense, their end points should be rigidly defined, even in cases where there are no clear instances of them in actual fact. This conviction is best expressed in Aarts (2007: 201):

If it is claimed that there is gradience between two categories α and β, then it must first be established that α and β actually exist as word classes. In other words, the well-motivated setting up of discrete categories of form classes is logically prior to claiming that gradience obtains between them. Paradoxically, adherents of gradience models must face the fact that their analytical point of departure should always be rigid discreteness.

3 In Moser (1988), e.g., the category perfect as well as auxiliary verb and periphrasis are analysed on the basis of prototypicality.

1 0 2 Invited Papers Amalia Moser

The term ‘gradience’, introduced by Bolinger (1961), has been gaining ground in the last few years, affording a terminological expression of the logical connection between prototypes and continua. Aarts distinguishes two types of gradience, subsective and intersective. The former is “an intra-categorial phenomenon which allows members of a class to display the properties of that class to varying degrees” (Aarts 2007: 79). Intra-categorial gradience leads to prototype effects.4 Intersective gradience, by contrast, is “an inter-categorial phenomenon which comes about when two form classes ‘converge’ on each other” (ibid.). Continua belong to this second type of gradience.

2. Time in Language

Tense, aspect and Aktionsart all have to do with time. Time is a very important concept for humans, not only in recent years, when it has become a valuable commodity, to be measured, allotted with precision to different tasks and priced accordingly, but throughout human history. It is important, however, to stress that it is a CONCEPT. There is no concrete physical entity to which it corresponds, such as there is for the concept of space, for instance. It is a construct of the human mind.5 There are two sources of our understanding of time: The most obvious one is human life and its inevitable progression from birth to death. This gives us our basic linear concept of time: we usually perceive it

4 The term ‘prototype effects’ is used here to imply that subsective gradience is not necessarily equated with prototypicality. In fact, Aarts (2007: 87-90) insists on distinguishing the two, on the grounds that gradience is predominantly a grammatical phenomenon, while prototypicality concerns conceptual entities in general, has an almost exclusively semantic basis and does not necessarily apply to linguistic entities; for one thing, the prototypicality criteria for the latter would be completely different to the concrete criteria applied to physical objects (shape, colour etc.). My own view is that prototype effects work in exactly the same way in both cases, hence it is entirely justified to speak of prototypicality in grammatical categories; much is made of the different type of criteria in the literature, but this difference seems immaterial, especially if the case of abstract entities is taken into account. Moreover, even though grammatical categories are the product of linguistic theorizing, it is probably fair to say that the process of categorization must be the same as in the case of physical entities. The point, nevertheless, need not be argued any further, since subsective gradience is not the object of this paper. 5 According to modern Physics, and specifically Albert Einstein’s and Stephen Hawking’s theories, time as a physical entity apparently has not much to do with time as we conceive of it, being one of the four dimensions and, if I understand correctly (mainly on the basis of Jaszczolt 2007), static. Movement, however, is a crucial feature of what we understand as time. This is verified by the existence in most languages of expressions such as “the flow of time”.

Invited Papers 1 0 3 Major Trends in Theoretical and Applied Linguistics

as a line with the present at its center extending infinitely towards the past and the future. The linear conception of time is usually represented as shown in Figure 4 (following the convention of left-to-right reading): ← past future → present

Figure 4. The Time Axis

A less obvious but more primitive source for our concept of time is the regular repetition of certain phenomena (the cycles of day and night, the succession of seasons etc.) We become aware of these long before we become aware of life and death, and they give us a sense of time moving in circles, repeated at regular intervals. This cyclical perception is reinforced by our sense of rhythm, which has its source in our earliest experiences: hearing our mother’s heartbeat and later feeling our own. All languages express time in their vocabulary and very many through their grammar, most commonly the verb. Grammatical expression usually involves the linear concept of time. There is some debate as to whether there might not be languages, in particular Australian ones, which reflect exclusively the cyclical conception of time (cf. e.g. Comrie 1985: 4-5); this is rather doubtful, since the cycles also proceed in a linear sequence. Most languages display both concepts, since both are equally essential in our experience of the world, even if linearity prevails in grammatical systems.

2.1. Tense

Tense is the grammatical category expressing the location of events on the axis of time. Events are located along the line shown in Figure 4, usually in respect to the present, which of course is the present of the speaker. This means two things: a) that we perceive time in more or less locational terms; in other words we have imagined time as space, and have thus created a metaphor based on space b) that we perceive it deictically, in egocentric terms. Nevertheless, the term absolute tense is used when the point of reference is the present (i.e. the ‘now’ of the speaker); the term relative is reserved for time defined in respect to another event. The termabsolute-relative tense is used when this second event is placed in respect to the speaker’s present. The somewhat controversial analysis of the Modern Greek tense system in Moser (1994, 2009) concludes that morphologically Modern Greek expresses

1 0 4 Invited Papers Amalia Moser

past, present and future, as well as anteriority in all three time spheres, as shown in Table 1 below.6

Table 1. Tense in the Greek Verb

aspect [+ PERFECTIVE] [-PERFECTIVE] tense

past Aorist (έτρεξα) Present (τρέχω)

present ― Imperfect (έτρεχα) tense Absolute future Perfective future (θα τρέξω) Imperfective future (θα τρέχω)

past Pluperfect (είχα τρέξει) ―

present Present Perfect (έχω τρέξει) ― Tense Relative Relative (anterior) future Future Perfect (θα έχω τρέξει) ―

The relationship of tense and aspect has been a subject of controversy; an association of the perfective with the past tense has often been pointed out (notably in Bybee, Perkins & Pagliuca 1994). The most obvious candidate for contributing to a gradient is the perfect, given the ongoing debate about whether it is a tense or an aspect. The relationship of tense and aspect in Greek has been explored in interesting ways in several recent works, notably by Xydopoulos & Tsangalidis (2007). The table above shows that the association of the perfective with the past is not an issue for Greek, while the (admittedly controversial) analysis of the perfect adopted here places it firmly in the category of tense. Under this analysis, then, the only indication of an interaction of the two categories lies in the absence of a perfective form for the (absolute) present7 and of an imperfective form for any of the anterior tenses.8 None of these issues will be addressed here; the table, however, can be used as an overview of the entire tense and aspect system.

6 This table leaves out the very interesting non-temporal uses of the tenses, which are outside the scope of this paper. Several studies of individual tenses are available; for a comprehensive analysis see Bella 2005. 7 In several analyses (e.g. Holton et al. 1997) the form τρέξω is dealt with as a perfective present, even though it is not used as an independent form. 8 While the form of the (main) verb used for the formation of the perfect is unquestionably perfective, there is no consensus about the perfect’s aspectuality. It has been analysed as perfective (the view adopted here), as a third aspect (both these analyses are given as alternatives in Clairis & Babiniotis 2005) and even as imperfective, because of the imperfectivity of the auxiliary (Tsangalidis 1999).

Invited Papers 1 0 5 Major Trends in Theoretical and Applied Linguistics

2.2. Internal Temporal Constituency

Aspect and Aktionsart are both perceived as being associated with the internal temporal constituency of situations.

2.2.1. Aktionsart

Regardless of their position on the time axis, situations occupy a point or a space on it, i.e. a chunk of time. Hence, they can be:

• durative (whether long- or short-lasting) or instantaneous cf. a situation like searching vs. finding • continuous or intermittent cf. the continuous ringing of a bell vs. a bell that rings every five seconds • homogeneous or consisting of clearly discernible phases cf. walking on a treadmill for exercise vs. walking from the station to a hotel • including an end point or open-ended cf. baking a batch of cookies vs. being a baker by profession

These distinctions were pointed out long ago, starting in fact with Aristotle. During the last century there have been several attempts to categorize them systematically, mostly amongst philosophers. The most widespread categorization is the one by Vendler (1957) into states, activities, accomplishments and achievements, based on the criteria of telicity (i.e. the inclusion of an end point) and divisibility into phases. The latter is particularly important for English, because of the nature of its aspectual system, but it is applicable cross- linguistically.

Table 2. Vendler’s categorization

[-PHASES] [+PHASES]

STATES ACTIVITIES [-TELIC] sleep run love paint be alive search ACHIEVEMENTS ACCOMPLISHMENTS [+TELIC] die run a mile find paint a portrait recognize read a book

The categorization by Mourelatos (Figure 5), while hierarchical, ends up with the same four categories (see the corresponding terms in brackets).

1 0 6 Invited Papers Amalia Moser

situations

states occurrences

processes events (activities)

developments punctual occurrences (accomplishments) (achievements)

Figure 5. Mourelatos’ categorization

Sasse (1991) includes a fifth category (inchoative statives) consisting of verbs like love, know etc., which double as statives and inchoatives

(1) I love Thessaloniki (2) I loved Thessaloniki instantly

Although such verbs have a rather special position in Greek and provide interesting information, especially for the history of aspect, in terms of internal temporal constituency they are covered by the other categorizations mentioned, since their two meanings/uses fall under states and achievements respectively. It should be mentioned at this point that there are serious objections in the literature not so much to the specific categorizations as to the category of Aktionsart itself as distinct from aspect, which will be discussed briefly after the description of the latter.

2.2.2. Aspect

The existence of a grammatical category “aspect” is not in dispute, even though it only entered Western European grammars in the 1840s via Slavic Grammar. Comrie’s (1976: 3) classic definition is more or less universally accepted:

Aspects are different ways of viewing the internal temporal constituency of a situation.

This is further elaborated in his definition of the subdivisions of the category, perfective and imperfective:

“Another way of explaining the difference between perfective and imperfective meaning is to say that the perfective looks at the situation from outside, without necessarily distinguishing any of the

Invited Papers 1 0 7 Major Trends in Theoretical and Applied Linguistics

internal structure of the situation, whereas the imperfective looks at the situation from inside, and as such is crucially concerned with the internal structure of the situation”. (Comrie 1976: 3-4)

It becomes immediately evident that aspect has in fact nothing to do with the actual internal temporal constituency of the situation. It depends on the vantage point (to anachronistically use a cognitive term, since this is essentially a cognitive definition) adopted by the speaker. The independence of aspect from the real temporal constituency of the situation becomes even more obvious when it is further stressed, with plenty of examples, that the perfective does not express punctuality, brevity or boundedness, nor the imperfective duration or non-boundedness (Comrie 1976: 16-40); these may arise as implicatures but as such are cancellable by the context. The important thing about aspect is that speakers have a CHOICE: they can select their vantage point, within or without the event. The result is that the two aspects are interchangeable: the same event can be described in two different ways, in grammatically acceptable sentences. Comrie’s own examples from Greek and French (1976:17) are shown in (3) and (4):

(3) ebasíleue déka étē – il régnait trente ans (4) ebasíleuse déka étē – il régna trente ans

The choice depends on a number of factors, including the implicatures arising from the adoption of different vantage points: the imperfective is more likely to be used when the speaker wishes to stress the duration or the tediousness of an event and the perfective when the speaker emphasizes its completion; none of this, however, is compulsory. Both of the following sentences would be acceptable with either aspect, although interestingly a perfective (5) would be rather awkward, even though it refers to a completed event (the reason being that the context leaves no doubt that the emphasis is on the length of the event):

(5) έπλενα όλη τη μέρα ρούχα και τελείωσα αργά τη νύχτα

‘I washed[imperfective] clothes all day and I only finished late at night’ (6) έπλυνα ένα σωρό ρούχα όλη την ημέρα και έχουν μείνει άλλα τόσα

‘I washed[perfective] a heap of clothes all day and there are as many left’

It was this interchangeability that led to the distinction being defined on the basis of subjectivity: while Aktionsart as an inherent characteristic is objective, aspect is subjective in the sense that it involves choice. The notion of subjectivity will be discussed in section 4 of this paper, after a brief look at the objections to the postulation of two grammatical categories.

1 0 8 Invited Papers Amalia Moser

3. Objections to the Aspect – Aktionsart Distinction

The aspect – Aktionsart continuum proposed here presupposes the existence of both these categories; this existence, however, has been seriously challenged. The objections come from opposite directions: Cognitive Linguistics and Formal Semantics. Langacker (e.g. 1987/1991) argues against the distinctions on the basis of the fact that speakers always have a choice of vantage point and hence can construe any situation as durative or not, the ‘objective’ temporal constituency being irrelevant. It is proposed here that this is exactly the argument that lends support to the claim of a viable distinction. There seems to be no doubt that there are objective differences in situations such as the one mentioned above between searching and finding and, more crucially, that they are obvious to speakers; the fact that they can be construed in ways that allow them to be expressed by the same aspect is precisely what renders the category so interesting, especially if it can be shown that objective temporal constituency interferes with aspect in linguistically significant ways. Verkuyl’s (1993, 1999, 2007) objections concern exactly this point: he does not deny the possibility of drawing the distinction, but claims that Aktionsart is a purely ontological category, with no bearing on language and therefore on Linguistics. However, there is evidence of systematic interaction between Aktionsart and aspect in language. There are few, if any, languages where aspect functions entirely independently of Aktionsart – even Greek, which is one of the freest in its use of aspect, is subject to a few Aktionsart restrictions. These concern the two more ‘extreme’ verb classes, achievements and states: Achievements, as a result of their instantaneous nature, have some difficulty combining with the imperfective, since, in order for a situation to be seen as developing, some duration is necessary. Their imperfective forms are normally habitual, with one exception: when they are used as a background or backdrop to another event:

(7) Τη στιγμή που έβρισκα τα κλειδιά μου χτύπησε το τηλέφωνο.

‘The moment I found[-perfective] my keys, the phone rang”.

They can also be used metonymically to refer to the process leading up to them, as in:

(8) Ο πατέρας τους πέθαινε κι αυτοί πήγαν διακοπές

‘Their father died[-perfective] (was dying) and they went away on holiday’.

Invited Papers 1 0 9 Major Trends in Theoretical and Applied Linguistics

On the other hand, states, which are open-ended and atelic, react in two ways to a combination with the perfective: it either presents them in a genuinely perfective manner, i.e. as a whole (9), or it singles out the point of entry into the state, as in (10):

(9) Κοιμήθηκαν τρεις ώρες

‘they slept[+perfective] for three hours’ (10) Κοιμήθηκαν στις τρεις

‘they slept[+perfective] at three’

The Slavic languages are much more constrained by Aktionsart and English is entirely bound by it: there are serious lexical restrictions on the use of the progressive, dealt with in section 4 of this paper. Verkuyl (e.g. 1994: 15) deals with this by postulating two parameters, [±SQA] and [±ADD TO]. The former refers to the type of arguments in the sentence; the latter refers to the presence or absence of movement, both literal and metaphorical, from Source to Goal, i.e. to the distinction between events with and without phases, or situations involving change and stative situations (Verkuyl 1999: 23-24) and thus essentially to Aktionsart. In the next sections it will be proposed that the varying degrees to which Aktionsart interferes with aspect, thus limiting the speaker’s freedom of choice, is what causes the gradience between the two categories.

4. Aspect and Subjectivity

4.1. The Concept of Subjectivity

The concept of subjectivity is the object of considerable controversy. It is one of the most popular subjects these days, mainly within the cognitive paradigm, and also in its historical dimension, i.e. as subjectivisation or subjectification, which features heavily in the literature on grammaticalization.9 The main concepts are those of Langacker (1990, 2006) and Traugott (1995, 2010), which differ considerably, despite their presentation as similar in most of the relevant literature.

9 See e.g. Langacker (1990), Nuyts (2001), De Smet, H. & J.-C. Verstraete (2006) and Stein & Wright (1995), Athanasiadou, Canakis & Cornillie (2006), Davidse, Vandelanotte & Cuyckens (2010) respectively.

1 1 0 Invited Papers Amalia Moser

Langacker’s concept is not related to the semantic content of the linguistic expression but to the viewing arrangement and the presence or absence of the speaker; maximal subjectivity is associated with the speaker functioning as a tacit conceptualizing presence. For Traugott subjectivity involves the expression of self and the representation of a speaker’s perspective or point of view in discourse. Nuyts’ (2001) concept is distinct from both, even though it is usually included in the Traugottian line of thought, to which it does bear greater affinity. He sees subjectivity as a subcategory of evidentiality, defining it not in the traditional terms of the quality of evidence, but rather in terms of whether the evidence is only available to the speaker or more widely known. In this paper subjectivity is seen as a choice of expressing the same, truth-conditionally determined, meaning through different linguistic forms, specifically through different aspectual forms of the verb; this is a very broad conception of subjectivity, to the point perhaps of being thought of as vacuous, but nevertheless quite rigidly defined and at the same time conforming to the common meaning of subjectivity.10

4.2. Problems With Choice and Subjectivity

Even in this elementary sense of availability of choice, subjectivity as the basis for defining aspect and Aktionsart presents difficulties, which have been pointed out in the literature.

4.2.1. Adverbials and Complements

Almost every book or article on Greek aspect mentions the imposition of one of the aspects (usually the imperfective) by adverbials: συχνά ‘often’, κάθε μέρα ‘every day’, συνεχώς ‘continually’ belong to this group and thus leave no room for choice. While this is true in a sense –these adverbials co-occur exclusively with the imperfective– it has to be taken into account that adverbials are themselves the result of choice on the part of the speaker. Outside the classroom there is no situation where a speaker is presented with an adverbial around which he or she has to build a sentence. The adverbials themselves, therefore, do not impose limitations to the speaker’s choice and to subjectivity as understood here.

10 For fuller argumentation see Moser (forthcoming).

Invited Papers 1 1 1 Major Trends in Theoretical and Applied Linguistics

Some matrix verbs, on the contrary, do impose restrictions on the choice of aspect in the complement. These fall mainly under three categories: • ‘aspectual’ verbs: αρχίζω ‘start’, συνεχίζω ‘continue’, σταματάω ‘cease’ etc. ⇒ imperfective (exception: verbs meaning ‘be about to’ ⇒ perfective) • verbs of learning/possession of skills: ξέρω ‘know’, μαθαίνω ‘learn’ ⇒ imperfective • verbs of perception: βλέπω ‘see’, ακούω ‘hear’ with na-complements ⇒ imperfective

It is easy to find ad hoc explanations: aspectual verbs explicitly refer to a point or time chunk of a process, while verbs of learning etc. refer to a permanent attribute; they still constitute an example of Aktionsart interfering with the choice of aspect. The case of verbs of perception is different: here the imperfective complement denotes the subject’s direct sensory knowledge of the situation and therefore has to present the situation as ongoing.11

4.2.2. Subjectivity, Habitual and Progressive

Objections of a more theoretical nature to subjectivity as the defining property of aspect have also appeared in the literature. Interestingly, almost all of them concern the subdivisions of the imperfective aspect, which are represented in Figure 6: aspect

perfective imperfective

habitual continuous

non-progressive progressive

Figure 6. Aspect and its subdivisions according to Comrie 1976: 26

11 These constructions, then, seem to function as evidentials, suggesting interesting paths of research, not least on subjectivity in Nuyts’ (2001) terms, with possible support for Nikiforidou’s (2004) analysis associating the imperfective with subjectivity; cf. the detailed analyses by Delveroudi et al. (1993) and Veloudis (2001). It should be pointed out that these finitena -complements of verbs of perception, rather than being descendants of infinitival complements, have replaced an Ancient Greek evidential participial construction.

1 1 2 Invited Papers

Amalia Moser

Oddly enough, this categorization has remained unchallenged, in a field where everything else is under constant dispute; it will be shown here, however, that much of the dissent about aspect and its relationship to Aktionsart is caused precisely by these subdivisions of the imperfective. As early as 1982, Bache challenged the subjectivity-based definition of aspect, demonstrating the unavailability of choice in the case of the progressive and even more so in that of the habitual. As will become evident from the following discussion, mainly on the basis of the Greek aspectual system, he is entirely right. Greek expresses morphologically only the main [±PERFECTIVE] opposition, although the habitual – continuous opposition has some syntactic consequences. The habitual is used in cases of regular repetition of events; as in other languages, simple repetition is not a sufficient condition12 for its use, as is shown by example (12); in fact simple repetition requires the perfective:

(11) Το καλοκαίρι πηγαίναμε[-perfective] για μπάνιο κάθε μέρα ‘in the summer we went (=used to go) swimming every day’

(12) Το καλοκαίρι πήγαμε[+perfective] /*πηγαίναμε[-perfective] τρεις φορές στην Άνδρο ‘this summer we went to Andros three times’

The most crucial fact about the Greek habitual from the point of view of this investigation is that it leaves no room for choice: not only is it impossible to use the imperfective for a non-habitually repeated situation, but a habitual situation can only be expressed by the imperfective. The [±PROGRESSIVE] opposition is irrelevant to Greek, though of course logically possible. It needs to be pointed out that the aspectual opposition of Greek is very different to that of English, where the non-progressive includes all other subdivisions. The [±PROGRESSIVE] opposition is thus wrongly, though frequently, equated with the [±PERFECTIVE] one, with dire consequences for the theoretical analysis of aspect. Greek is also different to, say, Spanish, which expresses both oppositions. Languages which possess the progressive impose severe restrictions on its use: it is allowed only with verbs expressing situations consisting of discernible phases (activities and accomplishments):

12 According to Comrie (1976: 26-30) and on the basis of examples such as The temple of Diana used to stand at Ephesus repetition is not even a necessary condition; this, however, seems to be a case of an extension of the use of the habitual to a subset of (long-lasting) imperfective situations, in a language with no other means of distinguishing non-progressive imperfective situations from perfective ones.

Invited Papers 1 1 3 Major Trends in Theoretical and Applied Linguistics

(13) Last night he walked home. All the way he was humming (a tune). (14) He used to love walking, but he doesn’t any more. While he *was loving it, he often walked home in the evenings. (15) Last night I found/ *was findingthe pen I’d lost.

Uses of the progressive with states and achievements constitute a deliberate flouting of the rule, used to convey extra meaning:

(16) All those months that they stayed out of touch their father was dying. (17) They were so arrogant about their skills and now they’re losing the game. I’m loving this!

Bache is therefore entirely justified with respect to the unavailability of choice and therefore absence of subjectivity in the case of the habitual and the progressive; rather than changing the definition of aspect, however, I opt for the rearrangement of the categories involved along a continuum of subjectivity.

5. The Aspect – Aktionsart Continuum

A gradient of subjectivity accommodates all the facts pointed out so far:

• the fact that free choice is not an attribute of all types of aspect • the fact that there are different degrees of freedom of choice within and across languages, for instance - that habituality in Greek leaves no room for choice - that it does in English, though only for the past tense - that progressivity in English is a matter of choice, but this choice is limited to certain types of Aktionsart

According to the discussion in section 1, the two ends of the continuum are rigidly defined in terms of subjectivity: Aktionsart is objective and aspect is subjective. The only aspectual category that can be accommodated at the subjective end of the continuum is [±PERFECTIVE]. The English [±PROGRESSIVE] would be closer to the aspect end, since it is optional, but dependent on Aktionsart. A fully-fledged [±HABITUAL], such as that of Greek, which is determined by objective characteristics of the situation, would be very close to the Aktionsart end, though not coinciding with it for two reasons: (a) that habituality is not inherent in the meaning of verbs and (b) that cross- linguistically habituality is very frequently expressed by the imperfective, although there exist languages with exclusively habitual forms. A habitual

1 1 4 Invited Papers Amalia Moser

such as the English used-to construction, with its optionality combined with the possibility of being used as a continuous, would be closer to the aspect end.13 The continuum, taking into account the categories considered so far, would have an appearance like the one presented in Figure 7:

[-SUBJECTIVITY] [+SUBJECTIVITY] Aktionsart aspect [±PERFECTIVE] ______Greek Habitual English Habitual Progressive

Figure 7. The Aktionsart-aspect continuum

So far, the discussion has been limited to subjectivity, the feature that differentiates aspect and Aktionsart, as well as the various subdivisions of the former, creating the gradient. There has been no mention of (a) feature(s) that unite(s) the two categories. The notion of a continuum in fact presupposes the existence of a common property or properties. It has been seen above (section 2.2.2) that the common property cannot be internal temporal constituency, since the basic aspectual distinction [±PERFECTIVE] is independent from the actual temporal constituency of each situation. At the same time, the interaction of [±PERFECTIVE] and Aktionsart outlined in section 3 suggests that there is some affinity between the imperfective and atelic situations on the one hand and between the perfective and telic situations on the other, in the sense that the members of each pair combine in a more straightforward manner, without any side- effects. It was proposed in Moser (1994) and elaborated in Moser (2009) that what comes into play is a feature of telicity. Telicity here is taken at its most abstract, with a different realization in each of the categories. The telicity involved in Aktionsart is based on reality: Vendler’s accomplishments and achievements both include a real end point in their meaning, whether this end point is realized in actual fact or not; they are terminative. As can be seen in Table 1, accomplishments are actually activity verbs with a complement which signals exactly this end point. Aspectual telicity, on the other hand, which is embodied in perfectivity, is completely notional. It is the logical result of viewing a situation as a whole: this is impossible unless some kind

13 The high frequency of imperfectives used as habituals is not difficult to explain. The temporal structure of an event that consists of a series of clearly discernible phases (such as an activity or accomplishment) is very similar to that of a series of independent events perceived as a set. This is also what allows for the reverse shift in the case of the English habitual.

Invited Papers 1 1 5 Major Trends in Theoretical and Applied Linguistics

of boundary is visualized in order to keep the situation in question distinct from its surroundings. While the perfective brings along a boundary, however abstract, the imperfective does not preclude it; it simply does not require it and does not pay any heed to its presence or absence. The logical requirement for the imperfective aspect is of a different kind: in order for the internal constituency of a situation to be brought to the fore, there has to be an internal constituency. This does not mean that the situation has to consist of discernible phases (this widespread idea is the result of analyses based on languages that do not have the opposition [±PERFECTIVE], but only [±PROGRESSIVE], such as English); it simply needs to have some duration, i.e. not to be punctual. This is why achievements have some difficulty in combining with the imperfective. However, since telicity and atelicity are entirely abstract in the case of aspect, these problems can be and are indeed overcome in the way that was demonstrated in section 3 (examples 7 and 8). Another way of formulating the outcome of this last discussion, which is presented schematically in Table 3, would be to say that telicity (and its absence) is objective in the case of Aktionsart and subjective in the case of aspect.

Table 3. Telicity, aspect and Aktionsart.

+TELICITY - TELICITY

accomplishments activities

Aktionsart achievements states

perfective imperfective Aspect

It follows from the analysis that, in terms of subsective gradience, the prototypical aspectual category is [±PERFECTIVE], being the only aspectual opposition that allows for entirely free choice and therefore the only entirely subjective category in the sense of subjectivity defined above. An interesting aspect of the Aktionsart-aspect continuum, which goes beyond the topic of this paper, is the possibility of applying it to the historical development of the category aspect in Greek, as well as to comparative studies of aspectual systems.

1 1 6 Invited Papers Amalia Moser

6. Conclusion

The analysis of the relationship between aspect and Aktionsart has led to the following conclusions: • aspect and Aktionsart are related but distinct categories • they are related through telicity (abstract in the case of aspect and concrete in the case of Aktionsart) • they are differentiated through subjectivity, in the sense that Aktionsart refers to the objective temporal structure of situations, while aspect is independent of it and therefore is a matter of choice • while Aktionsart has an ontological basis, it is linguistically significant, since it interacts with aspect semantically and syntactically • the two categories form a continuum of subjectivity in the sense outlined above, with Aktionsart at the objective and (prototypical) aspect at the subjective end • the prototypical aspectual category is the perfective-imperfective opposition [±PERFECTIVE], while the categories that have become known as subdivisions of the imperfective are in fact intermediate categories on the aspect-Aktionsart continuum.

As an instance of intersective gradience, it is hoped that the aspect-Aktionsart continuum proposed in this paper has lent some support to the claim that gradience, far from being based on vague concepts, requires rigid definitions of focal categories in the case of subsective gradience and the categories at the two ends of the continuum in the case of intersective gradience.

Invited Papers 1 1 7 Major Trends in Theoretical and Applied Linguistics

References

Aarts, B. 2007. Syntactic Gradience: the Nature of Grammatical Indeterminacy. Oxford: Oxford University Press.

Aarts, B., D. Denison, E. Keizer and G. Popova (eds.) 2004. Fuzzy Grammar: A Reader. Oxford: Oxford University Press.

Athanasiadou, A., C. Canakis and B. Cornillie (eds.) 2006. Subjectification: Various Paths to Subjectivity. Berlin/New York: Mouton de Gruyter.

Bache, C. 1982. Aspect and Aktionsart: towards a semantic distinction. Journal of Linguistics 18: 57-82.

Bella, S. 2005. Cognitive motivation and pragmatic function of the Greek deictics. Journal of Greek Linguistics 6: 39-60.

Bolinger, D. 1961. Generality, Gradience and the All-or-none. The Hague: Mouton.

Bybee, J., R. Perkins and W. Pagliuca. 1994. The Evolution of Grammar: Tense, Aspect and Modality in the Languages of the World. Chicago/London: The University of Chicago Press.

Clairis, C. and G. Babiniotis, in collaboration with A. Moser, A. Bakakou-Orfanou and S. Skopeteas. 2005. Gramatiki tis Neas Elinikis, Domoliturgiki – Epikinoniaki [Modern Greek Grammar]. Athens: Ellinika Grammata.

Comrie, B. 1976. Aspect. Cambridge: Cambridge University Press.

Comrie, B. 1985. Tense. Cambridge: Cambridge University Press.

Davidse, K., L. Vandelanotte and H. Cuyckens (eds.) 2010. Subjectification, Intersubjectification and Grammaticalization. Berlin/New York: Mouton de Gruyter.

1 1 8 Invited Papers Amalia Moser

Delveroudi, R., I. Tsamadou and S. Vassilaki. 1993. Contribution à l’ étude de la modalité en grec moderne: le marqueur να. Collection ERA 642. Paris: UFRL, Laboratoire de linguistique formelle (URA 1028).

De Smet, H. and J.-C. Verstraete. 2006. Coming to terms with subjectivity. Cognitive Linguistics 17: 365-392.

Holton, D., P. Mackridge and I. Philippaki-Warburton. 1997. Greek Grammar: A Comprehensive Grammar of the Modern Language. London: Routledge.

Hopper, P.J. and S. Thompson. 1984. The discourse basis for lexical categories in Universal Grammar. Language 60: 703-752.

Jaszczolt, K. 2007. Representing Time: An Essay on Temporality and Modality. Oxford: Oxford University Press.

Langacker, R.W. 1987/1991. Foundations of Cognitive Grammar. Volumes I and II. Stanford: Stanford University Press.

Langacker, R.W. 1990. Subjectification.Cognitive Linguistics 1: 5-38.

Moser, A. 1988. The History of the Perfect Periphrases in Greek. Doctoral dissertation, University of Cambridge.

Moser, A. 1994. Pion ke apopsis tou rimatos [Ationsart and Aspect in Greek]. Parousia Monograph n. 30. Athens: National and Kapodistrian University of Athens.

Moser, A. 2009. Apopsi ke xronos stin istoria tis Ellinikis [Aspect and Tense in the History of Greek]. Parousia Monograph n. 77. Athens: National and Kapodistrian University of Athens.

Moser, A. forthcoming. From Aktionsart to aspect: grammaticalization and subjectification in Greek.

Nikiforidou, K. 2004. Grammatical Meaning and Construal: A Cognitive Linguistic Approach. Parousia Monograph n. 61. Athens: National and Kapodistrian University of Athens.

Nuyts, J. 2001. Subjectivity as an evidential dimension in epistemic modal expressions. Journal of Pragmatics 33: 383-400.

Rosch, E. 1973. Natural categories. Cognitive Psychology 4: 328-350.

Invited Papers 1 1 9 Major Trends in Theoretical and Applied Linguistics

Rosch, E. 1978. Principles of categorization. In E. Rosch and B.B. Lloyd (eds.), Cognition and Categorization. Hillsdale, NJ: Erlbaum, 27-48. Reprinted in Aarts et al. (2004), 91-108.

Sasse, H.-J. 1991. Aspekttheorie. In H.-J. Sasse (ed.) Aspektsysteme. Arbeitspapier n. 14. Köln: Institut für Sprachwissenschaft, Universität zu Köln, 1-35.

Stein, D. and S. Wright. 1995. Subjectivity and Subjectivisation: Linguistic Perspectives. Cambridge: Cambridge University Press.

Taylor, J.R. 1989. Linguistic Categorization: Prototypes in Linguistic Theory. Oxford: Clarendon Press.

Traugott, E.C. 1995. Subjectification in grammaticalization. In D. Stein and S. Wright (eds.), Subjectivity and Subjectivisation: Linguistic Perspectives. Cambridge: Cambridge University Press, 31-54.

Tsangalidis, A. 1999. Will and Tha: A Comparative Study of the Category Future. Thessaloniki: University Studio Press.

Tsangalidis, A. 2010. (Inter)subjectivity and (inter)subjectification: A reassessment. In K. Davidse, L. Vandelanotte and H. Cuyckens (eds.), Subjectification, Intersubjectification and Grammaticalization. Berlin/New York: Mouton de Gruyter, 29-71.

Tsohatzidis, S. (ed.) 1991. Meanings and Prototypes: Studies in Linguistic Categorization. London/ New York: Routledge.

Veloudis, I. 2001. Ná kai na [Ná and na]. Eliniki Glosologia ’99: Praktika tu 4u Diethnus Sinedriu Elinikis Glosologias [Greek Linguistics ’99: Proceedings of the 4th International Conference on Greek Linguistics]. Thessaloniki: University Studio Press, 243-250.

Vendler, Z. 1957. Verbs and Times. The Philosophical Review LXVI: 143-160. Reproduced in Z. Vendler, 1967, Linguistics in Philosophy, Ithaca: Cornell University Press, 97-121.

Xydopoulos, G. and A. Tsangalidis. 2007. I rimatiki apopsi stin eliniki ke i sxesis E, R, S [Verbal aspect in Greek and the relations E, R, S]. Studies in Greek Linguistics 27 (In memory of A.-Ph. Christidis): 322-336.

1 2 0 Invited Papers Asking the Right Questions in “New School” EFL Curriculum Design

Nicos C. Sifakis

Hellenic Open University [email protected]

Abstract

The paper raises questions that focus on the ‘deep fundamentals’ of EFL curriculum designing, i.e., assumptions that tend to be taken for granted by policy makers involved in making decisions about Greek state-school English language curriculum designing, teaching, learning and assessment. I argue that such decisions should be informed by extensive research into the critical issues of modern-day English language usage inside and outside of the foreign language classroom; the broader foreign language literacy needs, wants and demands of the emerging learner anthropogeography of state school classrooms; stakeholders’ beliefs about the function of the textbook; an informed understanding of terms like ‘competence’, ‘fluency’ and ‘proficiency’; the integration of technology in the foreign languages classroom; teacher development.

1. Introduction

In recent years, there has been much debate about how to bring the Greek state education system in line with 21st century realities, challenges and demands. The Greek Ministry of Education has announced numerous documents, policies and action plans on educational reform initiatives and curricular innovation, perhaps the most influential of all being the action plan entitled “New School”. In this paper, I explore two perspectives of this situation. On the one hand, I consider ways in which the current realities and concerns of a particular subject- matter can be brought to bear on the broader curricular policies put forward. On the other hand, I discuss how the principles and concerns laid out in such policy documents and laws can impact innovation in the pedagogical practices of that subject-matter. The main centre of concern will be foreign language education and, in particular, teaching English as a foreign language (TEFL), otherwise known at times as teaching English to speakers of other languages (TESOL). In what

Invited Papers 1 2 1 Major Trends in Theoretical and Applied Linguistics

follows, I am interested in raising questions that refer to what I call the ‘deep fundamentals’ of the TEFL/TESOL practice and seeing the extent to which these are met in “New School” policy-making. The purpose is to arrive at a series of implications for English language teaching in Greek state schools.

2. Background

The debate about the modernisation of the Greek educational system reflects extensive discussions that have concerned scholars on a global scale for many years. For example, Gunther Kress, in a paper that touches upon the language curricula of the globalization era, vehemently argues for the need for curricula to be ‘relevant and productive’, in the sense that they delineate the teaching and learning of ‘quite new kinds of dispositions, attitudes and skills’ (Kress, 1996: 195). For Kress, the state educational system, in its central role as the determinant policy-maker, is responsible for incorporating parameters that respond to the social and political realities that define the global era, such as heteroglossia, hybridity, multilingualism and the plurality of semiotic forms. In essence, school students should be given access to modes of communication that will be essential for them in the future, in all aspects of their lives. This calls for a new type of literacy, one that empowers students to act relevantly in and through different language modes in a wide variety of continually changing social and geographical domains. Similar concerns have been raised time and again by scholars in disciplines as diverse as geography and history (Philippou, 2007), citizenship education (Keating et al., 2009), and the information and communication technologies (Hartnell-Young & Vetere, 2011). In the TEFL field, such concerns have been equally extensive in recent years, with the debate on curricular innovation referring to research in areas such as the role of English as an international lingua franca (e.g., Matsuda & Friedrich, 2011; McKay, 2003; Nunan, 2003), the building of digital literacies (Pegrum, 2009; Prensky, 2006; Warschauer, 2006, 2011), and the idea of global and European citizenship, (Byram, 2008; Huddleston & Kerr, 2006). All such concerns dismiss the view of foreign language literacy as a means of transmitting a narrow, singular standard language to passive recipients, in favour of one which exposes learners to and empowers them in becoming active users of a range of repertoires, genres and modes of communication, in line with the demands of the ‘new communicative order’ (Street 1999). The new globalization era demands that we abandon our old ways of EFL schooling and open up to the possibilities of a post-modern, post-EFL world (Sifakis & Fay, 2011). In what follows, I will first briefly describe the basic principles of the “New School” action plan. I will then explore a series of what I consider to be key questions that should be considered for those principles to generate change

1 2 2 Invited Papers Nicos C. Sifakis

in Greek state school TESOL practices. In doing so, I will be briefly referring to different aspects of teaching, learning, assessment and teacher education. In the final section, I will outline some guiding principles for the design and implementation of a high quality curriculum.

2.1. The “New School” Action Plan for Greek State Compulsory Education

The “New School” action plan was introduced in 2009 by the Greek Ministry of Education.1 It touches upon many aspects of teaching, learning and assessment and also draws and sets out a broad array of principles for pre- and in-service teacher education and training, curriculum design, implementation and evaluation, teacher evaluation, school evaluation and leadership. With regard to teaching and learning, the “New School” action plan (henceforth, “New School”) promotes the parameters of learner-centredness, digital literacy (there is particular reference to the use of technologies such as interactive whiteboards, e-books, and personal computers as a means of facilitating learning and incorporating more motivating and creative activities for all students), local, national, regional and global citizenship, and multicultural awareness and communication. According to the “New School”, teachers have the freedom to design and adapt materials that can help their students form lifelong learning skills, be responsible citizens, and participate actively in their social and cultural surroundings while exploring their potential as ‘small researchers’, ‘small scientists’, and as people who are able to communicate and express themselves in more than one foreign language. The principles of the “New School” formed the basis for the development of the curricula for all subjects of compulsory state education. The aim of these curricula is to bring to realisation the “New School” principles by incorporating perspectives and practices that promote the pre-existing cross-curricular framework of teaching and learning while exploiting the needs and challenges of differentiated instruction. The new curricula were published in the summer of 2011 and are now (in the 2011-2012 school year) in the process of being piloted in a large number of schools, before they become fully implemented in all Greek state primary and junior secondary schools in September 2012. For various reasons, the “New School” action plan makes clear the high priority it allocates to foreign language education, the most important being the

1 The text of the law is available in Greek on the following website: http://www.minedu.gov. gr/?option=com_content&view=article&id=402&Itemid=785&lang=el (accessed: 19.10.2013).

Invited Papers 1 2 3 Major Trends in Theoretical and Applied Linguistics

importance of foreign language competence in all of the major parameters and the existing low status given to foreign language schooling in the state domain, especially at the junior high school level.

3. Key Questions

This section will explore ways in which the principles and parameters of the “New School” action plan can be linked with the realities and challenges of foreign language education, with particular reference to the teaching and learning of English as a foreign language. My main concern here is to put forward a set of key questions that need to be answered, in my opinion, for EFL education to empower school students with regard to 21st century realities. These questions draw on Stern’s ‘basic building blocks of all language teaching’ (Stern, 1983, 1992). He identifies the building blocks simply as ‘language’, ‘learning’, ‘teaching, and ‘context’. My questions presuppose a comprehensive profiling of all stakeholders involved in the language teaching and learning process in a particular context, as well as an informed awareness of the ‘culture’ of EFL pedagogy. My aim here is to propose a critical orientation of where we stand today, in Greece, with reference to these building blocks (or fundamentals), and how foreign language education (in the form of curriculum design, teaching practice and learner response) can reflect the changing world around us. The questions I would therefore like to pose are the following: What reasons do we have for teaching English? What English do courses and teachers teach? What English do learners learn? How do we teach English? What do we “feel” about English? The emphasis here is on elaborating on the questions in some detail in an attempt to raise awareness rather than provide answers. It is important to ask these questions, as they refer to concepts that are not always apparent or explicitly stated but are often implicit in acts of teaching or in central policy decision-making.

3.1. What Reasons Do We Have for Teaching English?

The question is about the need for the teaching of English in state schools. The term ‘need’ here takes several meanings, two of which I detail. It turns our attention to the primary need for using the language outside school, which then forms the basis of the argument for incorporating the teaching of that language as part of the state school curriculum. It is a truism that English is an important global lingua franca, i.e., the language most people around the world fall back on when they need to communicate with people with whom they do not share any other language. The extensive familiarity with English and its increasing choice as the default language of communication by people who are not its ‘native

1 2 4 Invited Papers Nicos C. Sifakis

speakers’ has resulted in English constituting the ‘primary foreign language’ in most educational contexts around the world (Crystal, 2003). Another way of looking at ‘need’ is by examining how the term is related to foreign language teaching pedagogy in Greece. English has been taught to Greek state schools consistently for more than sixty years, the very first curriculum (‘analytical programme’) for state high schools being published in 1953 (Σηφάκις, 2012: 300). Nevertheless, despite the fact that the communicative approach to language teaching was pioneered by D. A. Wilkins in his 1976 seminal paper on notional/functional syllabuses, and despite the communicative curricula that were published in Greece in the 1980s and espoused that approach, the teaching of English in Greek state schools could be described as far-removed from learners’ communication needs. A much better description of the situation becomes apparent in the application of the acronym TENOR, coined by Abbot (1981), which stood for ‘teaching English for no obvious reason’ and was meant to characterise typical secondary-school learners as belonging in a situation where no obvious learning objective is envisaged. According to Abbot, learners in TENOR situations are, in the main, school- based, where motivation levels are low or vague and learning needs are difficult to define. As far as curricular design is concerned, TENOR situations are broadly specified and the range of language skills at best balanced. It is for these reasons that TENOR curriculum design is typically described as conservative. At the same time, the picture of English language teaching and courses in the private sector is starkly different. Private language schools in Greece offer courses tailored to satisfy students’ needs that are linked to sitting for a particular high-stakes proficiency examination. This means that the curricula designed for those purposes are far more specific to individual goals, albeit goals that are almost always linked to a specific language proficiency level. At the same time, in the state school system, EFL teaching is viewed at best as offering an essential skill (to those students who are willing and ready to accept it) and at worst as teaching another subject-matter.

3.2. Predispositions Towards English

This question addresses schoolchildren’s predisposition towards English and can shed light on the role of English in their lives and their reactions to the need for knowing and using it. By all accounts, Greek people are extremely positively inclined towards English. For example, a Eurobarometer study from 2001 showed that English was acknowledged as the most useful language after the mother tongue for 91% of respondents (Eurobarometer 2001). In the same research, 42% of people asked acceded to having ‘sufficient knowledge’ of English, with 24% believing that they are ‘very good’ users of the language.

Invited Papers 1 2 5 Major Trends in Theoretical and Applied Linguistics

At the same time, data from standardized proficiency examinations draw a rather different picture. For example, according to a report by Cambridge ESOL Exams from 2006, Greece ranked 50th (out of 67 countries), with 57% success rate, in the FCE (First Certificate in English) exams. According to the same report, in the CPE (Cambridge of Proficiency in English) exams Greece ranked 31st (out of 31 countries), with 46% success rate. What the above studies show is that there exist obvious mismatches between what the broader perspective in Greece tends to be about, the knowledge and capability of usage of English and the corresponding efficiency with which the language is actually put to use. In other words, there is a mismatch between people’s confidence in using English, which is high, and their competence in actually using it, which is rather low. There are various reasons for this, the discussion of which is beyond the scope of this paper. Suffice to say, one way of making sense of the confidence metric would be to refer to socio-cultural traits in L2 communication, whereas the competence metric could be interpreted with reference to the fact that most FCE and CPE exam sitters in Greece are far below the age recommended by the hosting examination body.

3.3. What English Do Courses and Teachers Teach?

The question focuses not only on the English delineated by the national curriculum but on the actual language that is taught in Greek state schools. It should be clear that the two are not mutually exclusive, as EFL teachers have access to courseware and additional material from the Internet that they are free to select, adapt and incorporate in their teaching should they be willing to do so. It is possible to distinguish between four different options for deciding on the type of English that state school learners are exposed to. The first option, which I shall call the ‘foreign language’ option, corresponds to the teaching of a particular native speaker variety of English (typically, British or General American English) to be used as a foreign language by Greek speakers. In the literature, this is the typical view of English language teaching in Greece, that is, in as far as Greece and the Greek ELT paradigm is, in Kachru’s (1985) terms, an Expanding Circle context of English language usage that is closely linked to the Teaching English as a Foreign Language (TEFL) paradigm. The second option, which I shall call the ‘exam-oriented’ case, describes the teaching of a particular native speaker variety that has an explicit and dominant orientation towards exam preparation (e.g., Cambridge ESOL exams). After all, it is well-documented that certification of C2 proficiency in EFL, in this country, is as important in finding a job as having basic computer skills (for a review, see Σηφάκις, 2012). While these two options are familiar characterizations and depict an established reality for most Greek state-school EFL students and teachers,

1 2 6 Invited Papers Nicos C. Sifakis

I would like to suggest that there are two further options for deciding on the type of English that is taught: the ‘international’ and the ‘multicultural’ option. Both of these options are influenced by the widespread uses of the language in domains that do not necessarily involve native speakers and it could be argued that including them would allow for a more creative exploitation of the cultural anthropogeography of the Greek state-school classroom. According to the ‘international’ option, teaching mostly focuses on preparing learners for English- medium communication in international contexts, a communicational zone in which interactions between non-native speakers are increasingly dominant. This interactional pattern contrasts with the more familiar native - non-native speaker interactions of the foreign language paradigm (option 1 above). It describes English-medium communication between our Greek learners of English and other non-native speakers. This option, which can be linked to the Teaching English as an International Language (TEIL) paradigm (McKay, 2002, Seidlhofer et al., 2006), corresponds to the international aspect potentially influencing English use by Greeks in general, but as operationalised by Greek learners of English in the foreign language classroom – for example, by means of tele-collaboration projects and other technology-mediated possibilities that increasingly enable pupils to use English internationally without even leaving their home town. Finally, the ‘multicultural’ option corresponds to an interest in the functions of English, largely in interactions between non-native users of the language, in increasingly multicultural societies. It characterises the use of the linguistic and cultural resources available in the increasingly multicultural character of Greek society and in the ELT classroom more particularly. A teaching paradigm with this orientation has been termed Multicultural Awareness Through English (MATE, see Fay et al., 2010). Recent research involving state school ELT teachers in Greece (Sifakis & Fay, 2011) showed that, when asked which of the four descriptions best matches their current, or most recent, teaching situation, respondents showed a clear preference for the ‘foreign language’ option (47%). In fact, when combined with the testing variant of it (option 2), nearly two thirds of the surveyed teachers characterised their practice in terms of the traditional TEFL paradigm. This corresponds to the typical orientation of the Greek TESOL system as belonging to the Expanding Circle. What is more, the fact that only 16% of the respondents identified their teaching situation as an exam-preparation one means that, while the exam-preparation situation is popular in Greece, the state sector is not perceived to be part of it, as exam-preparation is largely believed to be catered for by private institutions. That said, the study noted that a fifth of the teachers were inclined towards the multicultural option and a sixth towards the international one. The conclusion from this small survey was that the multicultural and international options may be becoming a significant

Invited Papers 1 2 7 Major Trends in Theoretical and Applied Linguistics

possibility for TESOL practitioners in Greece but are not yet on a par with the traditional TEFL paradigm.

3.4. What English Do Learners Learn?

This question refers to the English that most state primary and junior high school children learn inside and outside their EFL classroom and the potential links and mismatches that exist. As already mentioned, in the foreign language classroom, learners are exposed to materials that are decided upon by the established curriculum and selected textbook(s) and can incorporate a wide array of additional resources. Outside of the foreign language (FL) classroom, however, learning opportunities abound and can come from domains that are closely related to information and communication technologies, such as social media (e.g., Facebook), gaming (online, offline) and mobile technology (e.g., iPhone, iPod, iPad, PSP, etc). A growing body of research shows that such technologies are natural springboards for young learners and adolescents learning English as a foreign language (e.g., Adams & Brindley, 2007; Richardson, 2006). In fact, it is these technologies, whose mastery as instruments that empower communication between people have formed the backbone of what is now called digital literacy (e.g., Warschauer, 2006, 2011), that have prompted Kress’s plea for ‘relevant and productive’ curricula. Digital literacy has been incorporated in the “New School” action plan. The question is, to what extent are the learning opportunities that exist and are practised by most schoolchildren in Greece today outside the foreign language classroom exploited by the English language teacher. There are further implications of the understanding that young people’s familiarity with English is increasing. As more and more non-native speakers of English use the language to express themselves and communicate with other non-native speakers, they essentially become owners of that language. This means that the ‘native speaker model’ (which, as we have seen in our discussion of the previous question, is prevalent in EFL teachers’ minds) may no longer be that relevant because our learners have already bypassed it. This has important implications for the EFL classroom. As Suresh Canagarajah puts it:

We don’t have to teach each and every variety of English in the world (or the standard Englishes of Inner Circle communities); we simply have to change our understanding of language learning. To begin with, we have hitherto taught English in terms of a ‘target language’. The target has been defined in terms of a ‘native’ variety. […] Now we should teach in terms of a repertoire of language competence. (Canagarajah, 2006: 209-10)

1 2 8 Invited Papers Nicos C. Sifakis

3.5. How Do We Teach English?

On many levels, the ‘how’ and the ‘what’ in language pedagogy are intertwined. Does English language teaching in Greek state schools aim at making learners aware of the fluidity of language and communication? Is the raising of learners’ awareness of emergent and emerging means of communication and the rise of new linguistic forms (Kress, 1996: 195) a central part of the curriculum? The ways in which language is treated in the curriculum can vary according to beliefs held, policy concerns, overall orientation and aim (e.g., language testing). For example, it may be treated analytically or synthetically, as a set of grammatical constructions or discourse structures or sounds. Broader aspects of language use, such as appropriateness, may or may not be attended to, depending on the overall aims and learners’ proficiency levels (Richards & Rodgers, 2001). These issues can be complicated further if we consider the fact that many textbooks interpret the general curriculum differently. Another important side of the ‘how’ in language teaching is learner motivation. We have seen that knowing and using English is an important part of adolescent identity all over the world. In the majority of contexts, this results in English being not a ‘distant’ or ‘foreign’ language but one that young people can readily relate to. This has individual, societal and pedagogical consequences (Ehlich, 2009: 27). As long as different facets of this identity are allowed by the teacher to enter the foreign language classroom, language teaching and learning can be quite motivating. Research has shown that for language learning to work in school settings, teachers have to ensure the promotion of three essential ingredients with regard to learners (Deci & Ryan 1985, 1991): autonomy (which will satisfy learners’ need to participate in decisions that affect them, having some control over what is taught and how); affiliation (learners’ need to connect to and be accepted by other people in their peer groups); and competence (learners’ need to feel they are good at certain activities that are important to them). To that end, technology will certainly play a very significant role, as will learners’ knowledge of other languages:

Previously learned languages can be acknowledged and used within the classroom context by students and teachers alike as bridge languages. The explicit acknowledgement of the existence of previous languages, plus recognition of their status as useful pedagogical tools will naturally ease the new language learning process. (Hufeisen & Jessner, 2009: 126)

If treated with attention to detail and respect to learners’ idiosyncrasies, the Greek state curriculum for the English language can set the scene for the

Invited Papers 1 2 9 Major Trends in Theoretical and Applied Linguistics

‘cultural authentication’ of the ELT classroom (McKay & Bokhorst-Heng, 2008; Sifakis, 2005). It can empower teachers in allowing their learners to participate in a setting where they feel unthreatened by a dominant language (in virtually all cases, Greek), while, at the same time, expressing their own cultural identity. With the assistance of the appropriate integration of information and communication technologies, learners can engage in useful interactions with their peers using the ‘neutral’ vehicle of English (Fay et al, 2010). Such a perspective will have significant repercussions on learner assessment and testing. While the school curriculum aims at setting specific standards for gauging learners’ progress, it is important for the teacher to also attend to individual learners’ needs. The problem with standards is that, while they are (partially) learner-based, they tend to be holistic rather than individualistic (Kohn, 2000). It is crucial that state school assessment can distinguish between establishing and gauging how well learners are doing and what they are doing with the English language (cf. Broadfoot, 1996). Alfie Kohn (ibid.) has eloquently described this by drawing a distinction between viewing assessment as a ‘doing to’ process (one that involves a measurement of learners’ behavior, actions and performance) and viewing it as a ‘working with’ process (in which case the primary focus is on establishing the reasons, motives and values that inform learners’ behavior). The former is in many cases power-driven and imposes compliance, whereas the latter focuses on long-term goal setting and engenders a sense of cooperation between teachers and learners (Shohamy, 2001).

4. Implications for Teacher Education

What impact can reflecting on the above questions have on teacher education practices? The above discussion shows that probably the most important ingredient of successful English language teaching in the Greek state school is the empowerment of teachers to take responsibility for and have full control of their own teaching—essentially, to become autonomous teachers. To achieve autonomy, teachers need more than a basic training in ESOL pedagogy; they need to transform their own perceptions about their professional roles. It has been shown that ESOL teachers’ own perceptions of their professional roles can have a huge impact on the learning and achievement of their learners (Beijaard et al. 2000). Teachers’ own images of their professional roles have been known to be largely determined by their training, their own practical experiences as learners and the local educational contexts (Ben-Peretz et al. 2003). In a similar way, EFL learners’ engagement and motivation for participation in the foreign language classroom have been shown to be linked, among other things, to their teachers’ own sense of professional responsibility (e.g., Abu-Rabia 2004; Lee 1998).

1 3 0 Invited Papers Nicos C. Sifakis

Both of these observations are confirmed by research with Greek ESOL state-school teachers (Sifakis & Sougari, 2007). On the one hand, it has been shown that, within the educational framework, teachers regard themselves as executing their duties by the mere teaching of the syllabus as prescribed by the higher authority, namely the ministry of Education. As they are not engaged in the curriculum design decision-making process, they are limited to viewing the instruction of English as if it were just another subject within the school curriculum. Their beliefs regarding their role in the teaching process are deeply rooted in the status that the teachers hold in the teaching domain and extend beyond any limitations that could be borne as a result of years of teaching experience, age or even professional qualifications. Another study (Sougari & Sifakis, 2010) found that more experienced teachers are more inclined to constrict their teaching to native speaker varieties only, with younger teachers being more open to explore and incorporate other English-speaking cultures in their teaching. What these studies indicate is the need for teachers to be made aware of the importance of incorporating the world in which their learners live into their teaching. They also need to be provided with the instruments through which to achieve this. Current pre-service teacher education and in-service training practices need to come to terms with these concerns in order to overcome limiting assumptions and preconceptions about what constitutes successful English language teaching and learning in the state sector (in this regard, see Sifakis, 2009; Sifakis & Sougari, 2010; for a topical study of how tertiary education in the USA fails to develop students’ capacity for critical thinking and complex reasoning, see Arum & Roksa, 2011). The ultimate aim for autonomous teaching, according to Kumaravadivelu (2001: 540-544; 2003) incorporates the principles of particularity (the need for teachers to theorise appropriately their context), practicality (the acceptance of the primacy of situated teaching over abstract information as a source of context-specific knowledge) and possibility (the awareness that both language and pedagogy can either sustain or challenge existing power relations).

5. Conclusion

In this paper, I have raised a series of questions in an attempt to challenge assumptions about state-school English language teaching policy-making. It is important that these questions are researched extensively to inform relevant decision-making and contribute towards a curriculum that responds to the ‘new kinds of dispositions, attitudes and skills’ of the globalization era (Kress, 1996: 195). The principles outlined in the “New School” action plan are a step in that direction.

Invited Papers 1 3 1 Major Trends in Theoretical and Applied Linguistics

Such a process involves a reflection on the ‘basic building blocks’ of language teaching (Stern, 1983, 1992). What is required is a concept of the nature of language (‘what’ is taught), a view of the learner (‘who’ is involved) and a comprehensive understanding of the nature of language learning (which corresponds to the ‘how’ of teaching). As far as the ‘who’ is concerned, it is important to know as much as possible about the learners involved in a specific teaching situation, rather than unquestionably adopt (and endorse) the profiling of their proficiency and learning strategies put forward by the broad curricular and course-book orientations. This will also lead us to collecting information about another dimension of language learning, namely, the reasons learners have for learning (or, to be more precise, for attending the classes)—the ‘why’. These reasons will enlighten us on many levels and will significantly help clarify ways in which the syllabus can be taught more efficiently. Also, this process will raise issues that relate to other stakeholders that are sometimes ignored and not easily identified (e.g., parents or sponsoring institutions), whose views may also play an important role in teaching practice, if not the entire curriculum. Finding out about learners’ interests and preferred learning strategies and styles will also help teachers work out what motivates them in their learning, which will further inform their own teaching practice. These perspectives should inform all aspects of curriculum designing, namely, content (what students should know, be able to do, and be committed to), assessment (a measurement of what learners are doing at each time and how well they are doing) and context (how the education system is organized). They can be aligned to developing and delivering a high-quality curriculum. Another important element that should not be underestimated is the role of course books in Greek state education. To the extent that they accurately interpret the curricular content and offer teachers opportunities for the development of learner-centred learning episodes, textbooks can be significant instruments within educational systems. However, if used unimaginatively, textbooks can be limiting and can dominate school teaching to the exclusion of other important materials (Chall et al., 1991). This is especially important in ESOL education, where, as we have seen, the role of the language and the impact of technology and global communication have shifted substantially in later years. As Bourdieu and Passeron (1977) caution, we must be wary of those traditional curricular elements that keep exerting an influence even within educational structures that have been reformed (also see Soysal, 2002). It is crucial that state ESOL education in Greece is placed on the alert so that limiting preconceptions that dominate teaching and assessing practices are eliminated and real progress can be made that corresponds to 21st century needs and challenges. It is with these needs and challenges in mind that the above questions have been raised.

1 3 2 Invited Papers Nicos C. Sifakis

References

Abbot, G. 1981 Encouraging communication in English: a paradox. ELT Journal 35(3): 228-230.

Abu-Rabia, S. 2004. Teachers’ role, learners’ gender differences, and FL anxiety among seventh-grade students studying English as a FL. Educational Psychology 24(5): 711-721.

Adams, A. and S. Brindley. 2007. Teaching Secondary English with ICT. Maidenhead, & McGraw-Hill, New York: Open University.

Arum, R. and J. Roksa. 2011. Academically Adrift: Limited Learning on College Campuses. Chicago: University of Chicago Press.

Beijaard, D., N. Verloop and J.D. Vermunt 2000. Teachers’ perceptions of professional identity: An exploratory study from a personal knowledge perspective. Teaching and Teacher Education 16(7): 749-764.

Ben-Peretz, M., N. Mendelson and F.W. Kron 2003. How teachers in different educational contexts view their roles. Teaching and Teacher Education 19(2): 277-290.

Bourdieu, P. and J.-C. Passeron. 1977. Reproduction in Education, Society and Culture. London, Sage.

Broadfoot, P. 1996. Education, Assessment and Society: a Sociological Analysis. Buckingham: Open University Press.

Byram, M. 2008. From Foreign Language Education to Education for Intercultural Citizenship. Clevedon, UK: Multilingual Matters.

Invited Papers 1 3 3 Major Trends in Theoretical and Applied Linguistics

Canagarajah, S. 2006. An interview with Suresh Canagarajah. In R. Rubdy and M. Saraceni (eds.), English in the World: Global Rules, Global Roles. London: Continuum, 200-211.

Crystal, D. 2003. English as a Global Language (2nd ed). Cambridge University Press.

Deci, E. and R. Ryan. 1985. Intrinsic Motivation and Self-Determination in Human Behavior. New York: Plenum.

Deci, E. and R. Flaste. 1991. Why we do what we do: The Dynamic of Personal Autonomy. New York: G.P. Putnam’s Sons.

Ehlich, K. 2009. What makes a language foreign? In K. Knapp and B. Seidlhofer (eds.), Handbook of Foreign Language Communication and Learning. Berlin: De Gruyter Mouton, 21-44.

Eurobarometer 2001. Europeans and Languages. Eurobarometer 54 Special. International Research Associates Europe. The Education and Culture Directorate- General Unit.

Fay, R., V. Lytra and M. Ntavaliagkou. 2010. Multicultural awareness through English: A potential contribution of TESOL in Greek schools. Intercultural Education, 21(6): 581-595.

Hartnell-Young, E. and F. Vetere. 2011. A means of personalising learning: incorporating old and new literacies in the curriculum with mobile phones. Curriculum Journal, 19(4): 283-292.

Huddleston, T. and D. Kerr (eds.). 2006. Making Sense of Citizenship: A Continuing Professional Development Handbook. London: Hodder Education.

Hufeisen, B. and Jessner, U. 2009. Learning and teaching multiple languages. In K. Knapp and B. Seidlhofer (eds.), Handbook of Foreign Language Communication and Learning. Berlin: De Gruyter Mouton, 109-138.

Kachru, B.B. 1985. Standards, codification and sociolinguistic realism: the English language in the outer circle. In R. Quirk and H. Widdowson (eds.), English in the World. Cambridge: Cambridge University Press, 11-34.

Keating, A., D.H. Ortloff and S. Philippou. 2009. Introduction: Citizenship education curricula: the changes and challenges presented by global and European integration. Journal of Curriculum Studies, 41(2): 145-158.

1 3 4 Invited Papers Nicos C. Sifakis

Kohn, A. 2000. The Case Against Standardized Testing: Raising the Scores, Ruining the Schools. Portsmouth, NH: Heinemann.

Kress, G. 1996. Internationalisation and globalisation: rethinking curriculum of communication. Comparative Education 32: 185-96.

Kumaravadivelu, B. 2001. Toward a postmethod pedagogy. TESOL Quarterly, 35(4): 537-560.

Kumaravadivelu, B. 2003. Beyond Methods: Macrostrategies for Language Teaching. New Haven, C.T.: Yale University Press.

Kumaravadivelu, B. 2006. Understanding Language Teaching: from method to postmethod. Mahwah, New Jersey: Lawrence Erlbaum Associates.

Lee, I. 1998. Supporting greater autonomy in language learning. ELT Journal 52(4): 282-289.

Matsuda, A. and P. Friedrich. 2011. English as an international language: A curriculum blueprint. World Englishes, 30(3): 332-244.

McKay, S.L. 2002. Teaching English as an International Language: Rethinking Goals and Approaches. Oxford: Oxford University Press.

McKay, S.L. 2003. EIL curriculum development. RELC Journal, 34(1): 31-47.

McKay, S.L. and W.D. Bokhorst-Heng. 2008. International English in its Sociolinguistic Contexts: Towards a Socially Sensitive EIL Pedagogy. New York: Routledge.

Nunan, D. 2003 The impact of English as a global language on educational policies and practices in the Asia-Pacific Region.TESOL Quarterly 37(4): 589-613.

Pegrum, M. 2009. From Blogs To Bombs: The Future of Digital Technologies in Education. Perth: UWA Publishing.

Philippou, S. 2007. Re-inventing ‘Europe’: the case of the European dimension in Greek-Cypriot geography and history curricula, Curriculum Journal, 18(1): 57-88.

Prensky, M. 2006. Don’t Bother Me Mum, I’m Learning. St Paul: Paragon House.

Invited Papers 1 3 5 Major Trends in Theoretical and Applied Linguistics

Richards, J.C. and T.S. Rodgers. 2001. Approaches and Methods in Language Teaching (2nd ed.). Cambridge: Cambridge University Press.

Richardson, W. 2006. Blogs, Wikis, Podcasts, and Other Powerful Web Tools for Classrooms. Thousand Oaks, CA: Corwin Press.

Seidlhofer, B, A. Breiteneder and M.-L. Pitzl. 2006. English as a lingua franca in Europe: challenges for applied linguistics. Annual Review of Applied Linguistics, 26: 3-34.

Shohamy, E. 2001. Democratic assessment as an alternative. Language Testing, 18(4): 373-391.

Sifakis, N.C. 2005. Teaching EIL – teaching international or intercultural English: What teachers should know. System, 32(2): 237-250.

Sifakis, N.C. 2009. Challenges in teaching ELF in the periphery: the Greek context. ELT Journal, 63(3): 230-237.

Σηφάκις, Ν.K. 2012. Αγγλική Γλώσσα και Παγκοσμιοποίηση: ‘Οψεις της Σύγχρονης Πραγματικότητας στην Ελλάδα, στην Ευρώπη και στον Υπόλοιπο Κόσμο. Αθήνα: Ηρόδοτος.

Sifakis, N.C. and R. Fay. 2011 Integrating an ELF pedagogy in a changing world: the case of Greek state schooling. In A. Archibald, A. Cogo and J. Jenkins (eds.), Latest Trends in ELF Research. Newcastle upon Tyne: Cambridge Scholars Publishing, 285-298.

Sifakis, N.C. and A.-M. Sougari. 2007. Teacher roles and learners’ motivation – a preliminary investigation of Greek state-school EFL teachers’ viewpoints. In E. Agathopoulou, M. Dimitrakopoulou and D. Papadopoulou (eds.), Selected Papers on Theoretical and Applied Linguistics. Thessaloniki: Monochromia, 528-539.

Sifakis, N.C. and A.-M. Sougari. 2010. Between a rock and a hard place: an investigation of EFL teachers’ beliefs on what keeps them from integrating global English in their classrooms. In C. Gagliardi and A. Maley (eds.), EIL, ELF, Global English: Teaching and Learning Issues. Bern: Peter Lang, 301-320.

Sougari, A.-M. and N.C. Sifakis. 2010. Rethinking the role of teachers’ beliefs about their status: a survey of Greek teachers’ inner thoughts. In A. Psaltou-Joycey and M. Matthaioudakis (eds.), Advances in Research on Language Acquisition and Teaching: Selected Papers from the 14th GALA Conference, 415-428.

1 3 6 Invited Papers Nicos C. Sifakis

Soysal, Y.N. 2002 Locating Europe. European Societies, 4(3): 265-284.

Stern, H.H. 1983. Seeing the wood and the trees. In K. Johnson (ed.), The Second Language Curriculum. Cambridge: Cambridge University Press, 207-221.

Stern, H.H. 1992. Issues and Options in Language Teaching. Oxford: Oxford University Press.

Street, B. 1999. New literacies in theory and practice: What are the implications for language in education? Linguistics and Education 10(1): 1-24.

Warschauer, M. 2006. Laptops and Literacy: Learning in the Wireless Classroom. New York: Teachers College Press.

Warschauer, M. 2011. Learning in the Cloud: How and Why to Transform Schools With Digital Media. New York: Teachers College Press.

Invited Papers 1 3 7 Major Trends in Theoretical and Applied Linguistics

1 3 8 Part II Selected Conference Papers Section 1: Phonology - Phonetics

1 3 9 Major Trends in Theoretical and Applied Linguistics

1 4 0 The Acquisition of English Intonation by Native Greek Speakers

Evia Kainada and Angelos Lengeris

Technological Educational Institute of Patras, Aristotle University of Thessaloniki [email protected], [email protected]

Abstract

This study examined the production of English intonation by Greek second- language (L2) learners of English, specifically their production of polar questions and their pitch range in English. Productions of (a) comparable materials in Greek spoken by the same Greek speakers and (b) English materials spoken by native English speakers were used to assess phonological and phonetic native language (L1) transfer when learning an L2. The results showed that Greek speakers used their L1 (Greek) intonation in English polar questions. Greek speakers’ pitch span in English was narrower from both their L1 (Greek) and from the target (English) pitch span.

1. Introduction

The acquisition of L2 prosody, particularly intonation, is a very difficult task and the underlying processes involved in the acquisition of the intonational component of the L2 grammar still remain elusive for researchers and teachers alike. Despite findings showing that deviations in the production of L2 prosody (e.g. stress, rhythm, intonation) may affect listeners’ judgments more than deviations in the production of L2 segmentals (vowels and consonants) (e.g. Derwing et al. 1998; Hahn 2004; Jilka 2000; Kang 2010; Kang et al. 2010; Munro 1995; Munro & Derwing 1999, 2001; Pickering 2001), relatively little research has been done in the acquisition of L2 prosody. For example, apart from impressionistic data and intuitive predictions, no experimental study has examined the learning of English intonation by Greek speakers. The current study is a first step towards filling this gap as it applies a phonologically-motivated qualitative method to analyse the production of English polar questions by Greek speakers and their range of pitch when speaking English. Research at the segmental level has shown that native language affects the learning of L2 vowels (e.g. Cebrian 2006; Flege et al. 1999; Flege & MacKay 2004;

Phonology - Phonetics 1 4 1 Major Trends in Theoretical and Applied Linguistics

Iverson & Evans 2007; Lengeris 2009; Polka 1995) and consonants (e.g. Best et al. 2001; Guion et al. 2000; Hattori & Iverson 2009; Iverson et al. 2003; Mackay et al. 2001). For example, Greek learners of English have difficulty perceiving and producing the English tense-lax /iː/-/ɪ/ distinction (e.g. feel vs. fill) because they have a single vowel category /i/ in their vowel system (Lengeris 2009; Lengeris & Hazan 2010). Likewise, Japanese speakers are very poor at perceiving and producing the English /r/-/l/ distinction (e.g. rock vs. lock) because they focus on second formant frequency (F2) which is important for the perception of the Japanese voiced tap /ɾ/ but is irrelevant for the English /r/-/l/ distinction (Hattori & Iverson 2009; Iverson et al. 2003). Strong evidence that L1 transfer plays an important role in the way learners perceive and produce the L2 intonational patterns is also provided in the literature, but until relatively recently research focused on the errors produced by learners (e.g. Backman 1979; Willems 1982). More recent research (e.g. Mennen 2006) stresses the need to adopt a generally-agreed upon framework for intonational analysis to better examine cross-linguistic similarities and differences in intonation and, ultimately, to extend our segmental-level understanding of second language acquisition to the acquisition of L2 prosody. To this end, this study follows the autosegmental-metrical (AM) theory of intonational phonology (Pierrehumbert 1980; Ladd 2008), which has been applied successfully in recent L2 intonation research (e.g. Atterer & Ladd 2004; Jilka 2000; Jun & Oh 2000; Mennen 2004, 2006). The AM theory distinguishes between the underlying phonological representation of intonation (e.g. tonal inventory) and its phonetic manifestation (e.g. F0 peak alignment) which allows for direct extension of speech learning models originally developed for the learning of L2 segmentals to the learning of L2 intonation (e.g. Flege’s Speech Learning Model, see Flege 1995). Mennen (2004) adopted the AM theory to examine the production of Greek pre-nuclear rises in declaratives by experienced Dutch learners of Greek. At a phonological level, Greek and Dutch use identical pre-nuclear rises in declaratives, but the two languages differ in terms of the phonetic manifestation of the rise. In Greek, the alignment of the peak is realized in the vowel following the accented syllable whereas in Dutch the peak is realized slightly earlier, within the accented syllable. Mennen (2004) found that four out of five Dutch learners of Greek tested in her study transferred their L1 (Dutch) phonetic realization of the pre-nuclear rise in Greek (i.e., they aligned the peak earlier than Greek speakers) and only one Dutch learner showed native-like performance (i.e., her peak alignment values were within the norms for native Greek speakers recorded for control reasons). The present study focuses on one phonological and one phonetic aspect of the acquisition of L2 intonation, namely the production of English tonal targets by native Greek speakers in polar questions and the range of pitch they use when speaking English. This is an interesting L1/L2 pair because, as will be discussed in the

1 4 2 Phonology - Phonetics Evia Kainada and Angelos Lengeris

following section, there are major differences between Greek and English in the choice of tonal targets in polar questions (which are of course accompanied by phonetic differences in their implementation).

1.1. Polar Questions

Within the AM framework, Greek has received significant attention over the last years (e.g. Arvaniti & Baltazani 2005; Arvaniti 2007; Baltazani & Jun 1999; Kainada 2010; Tserdanelis 2003). Greek polar questions in particular have been analysed by Arvaniti, Ladd & Mennen (2006), Arvaniti (2009) and Baltazani (2007). Arvaniti (2009) reports that the autosegmental representation of the polar question melody is L* L+H- L% whereby the rise-fall pitch movement shows two possible alignments with the segmental string depending on where the focus of the question lies. If the focused word is the final word of the utterance, the L* pitch accent (the nucleus of the question) co-occurs with the stressed syllable of the final word and the L+H-L% phrase accent and boundary tone occur on the last syllable of the utterance (see figure 1). If, on the other hand the focused item is not the final word of the utterance, the L* pitch accent appears on the stressed syllable of the focused item, the L+H- phrase accent co-occurs with the stressed syllable of the final word and the L% boundary tone appears on the last syllable of the utterance (see figure 2).

Figure 1. Spectrogram, phonetic transcription and tonal transcription of the Greek polar question /ta lu’luðʝa mi’ɾizune/ “Do flowers smell?” (example replicated from Arvaniti & Baltazani 2005). The superimposed blue line represents the F0 contour. The L* co-occurs with the stressed syllable of the final word and the L+H- L% phrase accent and boundary tone co-occur with the final syllable of the last word, indicating that the focused item of the utterance is the final one.

Phonology - Phonetics 1 4 3 Major Trends in Theoretical and Applied Linguistics

Figure 2. Spectrogram, phonetic transcription and tonal transcription of the Greek polar question /ta lu’luðʝa mi’ɾizune/ “Do flowers smell?” (example replicated from Arvaniti & Baltazani 2005). The superimposed blue line represents the F0 contour. The L* pitch accent aligns with the stressed syllable of the non-final word, the L+H- phrase accent co-occurs with the stressed syllable of the final word and the L% co-occurs with the final syllable of the final item of the utterance, indicating that the focused item of the utterance is not the final one.

While Greek polar questions involve a rise-fall pitch movement at the end of the question, English polar questions are signaled via a range of possible contours. Within the AM framework, Grabe, Post & Nolan (2001) and Grabe (2004) have shown that the three most common ones are: (a) a final rise pitch movement that takes the form of L*H H%; (b) a final (fall-)rise pitch movement that takes the form of H*L H%; and (c) a final fall pitch movement that takes the form of H*L% (see figures 3-5 respectively). Greek and English therefore differ fundamentally in the use of intonation for expressing polar questions (letting aside any differences in the use of tonal targets in prenuclear position in polar questions); where Greek has a rise-fall pitch movement at the end of the question, English may have a rise, a fall-rise or a fall.

1.2. Pitch Range

Languages differ not only in terms of their phonological inventory of intonational targets or the phonetic manifestation of those targets but also in terms of the range of pitch they use. Pitch range can be analyzed as varying along two dimensions, namely pitch level and pitch span (Ladd 2008). Pitch level refers to the height of overall pitch (high, mid or low) and pitch span refers to the range of frequencies used (wide or narrow). For example, German has been found to

1 4 4 Phonology - Phonetics Evia Kainada and Angelos Lengeris

Figures 3-5. Spectrogram, tonal transcription and phonetic transcription of the English polar question “May I leave the meal early?” uttered using the three most common intonation contours in English polar questions, namely L*H H% (final rise), H*L H% (final fall-rise) and H*L% (final fall). The materials are taken from the freely available IViE corpus (Grabe et al. 2001).

Phonology - Phonetics 1 4 5 Major Trends in Theoretical and Applied Linguistics

employ lower pitch level and narrower pitch span than English (Mennen 2007; Mennen et al. 2007). Such cross-linguistic differences may give rise to emerging stereotypes about national and linguistic groups. For example, the use of narrower pitch range may make German speakers sound ‘bored’ or ‘unfriendly’ to English speakers and, reversely, the use of a wider pitch range may make English speakers sound ‘overexcited’ or ‘aggressive’ to German speakers. Additionally to such cross-linguistic differences in the use of pitch range, there is evidence that L2 learners use a narrower pitch range than native speakers of the target language (Backman 1979; Willems 1982). However, given potential cross- linguistic differences in pitch range, if an L2 learner is found to use a narrower pitch range than the one used in the target language, this may simply mean that the learner is using her L1 pitch range. In order to rule out this possibility, it is therefore necessary to also obtain baseline measures of the learner’s L1 pitch range.

1.3. Research Questions

The current study addresses the following research questions: 1. How do Greek learners of English produce English polar questions? a) Do they transfer their L1 (Greek) intonational targets? b) Have they acquired the L2 (English) intonational targets? 2. How do Greek learners of English use pitch range when speaking English? a) Do the two languages differ in terms of pitch range? b) If not, do Greek speakers use a narrower pitch range when speaking English?

2. Method

2.1. Participants

The participants were 7 Greek learners of English from Athens (4 male, 3 female), aged 12-15 years, all intermediate learners of English. None had spent a period of more than one week in an English-speaking country. They were asked to read polar questions in Greek and English. They were also asked to read the Cinderella story in both languages for the purpose of obtaining pitch range measurements. Comparable materials (i.e., English polar questions and the Cinderella story in English) spoken by native English speakers were taken from the freely available IViE Corpus (Grabe et al. 2001). Six English speakers from Cambridge (3 male, 3 female) were used with a mean age of around 15 years old.

1 4 6 Phonology - Phonetics Evia Kainada and Angelos Lengeris

2.2. Speech Materials and Procedure

Participants were recorded using a MicroTrack 24/96 digital recorder at a sampling rate of 44.1 kHz. They read the sentences from script and were not given any instructions as to the placement of focus in each question. Polar questions were matched for number of syllables and position of stress across languages. There were 4 polar questions in Greek and 4 questions in English (see table 1). Fillers were also included containing statements and wh-questions. The sentences were blocked by language and were fully randomized within each language. All sentences were read three times by each Greek speaker. English sentences preceded Greek sentences to avoid any interference from Greek. The participants read a few sentences before the actual recordings to familiarize themselves with the procedure.

Table 1. English sentences read by native English speakers (taken from the IViE corpus) and Greek and English sentences read by Greek learners of English.

English L1 Greek L1 English L2

May I lean on the railings? Τον είδες τον Χαρίδημο; Have you seen Haridimos? May I leave the meal Μπορώ να φύγω μόνος; May I leave the meal early? early? Will you live in Ealing? Θα μείνεις στη Λήμνο; Will you live in Limnos?

Τη γνωρίζετε την Έλενα; Have you been introduced to Helena?

Following Grabe (2004) and in order to obtain representative measurements of pitch range in each language, after reading the sentence materials Greek speakers read the Cinderella story, first in English and then in Greek (the two passages were matched for type and length of sentences across languages). English and Greek polar questions were transcribed in Praat (Boersma & Weenink 2009) using the ToBI system (see Arvaniti & Baltazani 2005 for Greek and Grabe et al. 2001 for English) and the percentage of times each pitch contour appeared on the data was measured (since this is an exploratory study we are interested in whether Greek learners have acquired the English tonal targets in polar questions leaving the examination of the exact phonetic realization of those targets for future research). For pitch level, F0 frequency (Hz) from the Cinderella story was calculated. For pitch span, three measurements, all in semitones (ST), were taken: (a) 80% range (i.e., the difference between the 90th and the 10th percentile); (b) interquartile range (IQR) (i.e., the difference between the 75th and the 25th percentile, in other words the middle 50% of the observations); and (c) +/- 2 standard deviations around the mean.

Phonology - Phonetics 1 4 7 Major Trends in Theoretical and Applied Linguistics

3. Results

3.1. Polar Questions

Figure 6 shows percentage tune selection for signalling English polar questions by native English speakers. The two rise pitch movements (i.e., H*L H% and L*H H%) were used about 57% of times (around 33% and 23% respectively) and the fall pitch movement (i.e., H* L%) was used about 43% of times. A learner of English is therefore required to use at least one of these three tunes to successfully approximate the way English speakers signal polar questions.

Figure 6. Percentage tune selection by English speakers in English polar questions. H* L% corresponds to a final fall pitch movement at the end of the question; H*L H% corresponds to a final (fall-)rise pith movement; and L*H H% corresponds to a rise pitch movement towards the end of the question (see Grabe et al 2001).

Figure 7 shows percentage tune selection for signalling Greek polar questions (blue bars) and English polar questions (grey bars) by native Greek speakers. For representation reasons, the label L* L+H- L% refers to utterances with focus in a non-final word of the question and the label L*+H L* L+H- L% refers to questions with focus in the final word of the question. Greek speakers used the expected rise fall pitch movement to signal Greek polar questions. They predominantly (around 90%) placed the focus of the question on the non-final word, indicating a clear preference for this realisation when no instructions are given to participants regarding focus placement.

1 4 8 Phonology - Phonetics Evia Kainada and Angelos Lengeris

Figure 7. Percentage tune selection by Greek speakers in English polar questions (blue bars) and in Greek polar questions (grey bars). L* L+H- L% refers in this graph to utterances with focus in a non-final word of the question and L*+H L* L+H- L% refers to utterances with focus in the final word of the question.

Importantly, Greek speakers transferred their L1 intonation in polar questions (including focus placement) when producing English polar questions, i.e., they exclusively used the ‘Greek’ rise-fall tune in English without demonstrating any sign that they have acquired either of the tunes used by native English speakers in polar questions (see figure 8).

Figure 8. Spectrogram and tonal transcription of the question “May I leave the meal early?” spoken by a Greek learner of English, showing a complete transfer of the learner’s L1 (Greek) polar question intonation to English polar questions. The word “may” is in focus and carries the L* pitch accent. The final stressed syllable of the question (i.e., the first syllable of “early”) carries an L+H- phrase accent. The boundary tone is an L% as we would expect if this were a Greek polar question.

Phonology - Phonetics 1 4 9 Major Trends in Theoretical and Applied Linguistics

3.2. Pitch Range

Table 2 shows mean values and statistical comparisons for pitch range (three measurements of pitch span and one measurement of pitch level) obtained for Greek learners of English and for native English speakers. Greek and English did not differ in terms of pitch span across measurements (the English pitch span was somewhat wider than in Greek but this difference was not statistically significant). However, Greek speakers’ pitch span in English was narrower than both the native English one and their own native in Greek, indicating that their speech was more monotonous and with less pitch variation in English than in Greek. Furthermore, Greek speakers’ pitch level was significantly lower when speaking English than when speaking Greek (no statistical comparison can be made between Greek and English speakers’ pitch level for pitch level being an absolute measurement).

Table 2 Mean values of pitch range (three measurements of pitch span in ST and one measurement of pitch level in Hz) for Greek speakers in L1 (Greek) and L2 (English) and for English speakers in L1 (English). The results of t-tests (Bonferroni-adjusted) comparing mean values are also given.

L1 Greek L2 English L1 English L1 Greek vs. L2 English L2 English vs. L1 English

80% (ST) 6.82 7.96 4.85 t(6)=4.562, p<.001 t(11)=-5.107, p<.001

IQR (ST) 3.73 4.23 2.50 t(6)=5.741, p<.001 t(11)=-5.122, p<.001

SD4 (ST) 9.75 10.20 8.01 t(6)=3.128, p<.01 t(11)=2.838 p<.01

Level (Hz) 196 180 188 t(6)=3.119, p<.01 n.c.

n.c. = non-comparable data

4. Discussion

The current study is part of a larger project that examines, for the first time in the literature, the production of English intonation by Greek learners of English. The analysis of intonation follows the AM theory for intonational analysis (Pierrehumbert 1980; Ladd 2008). Here we focus on the production of English polar questions by intermediate learners of English and their range of pitch when speaking English. Baseline productions uttered by the same Greek speakers and by native English speakers were used for control reasons. The results showed that learners transferred the Greek tune (i.e., a rise-fall movement at the end of the utterance) in English polar questions. There was no evidence that they have acquired any of the patterns used by native English speakers for signalling polar questions. This L1 transfer effect on the learning of L2 intonation is consistent

1 5 0 Phonology - Phonetics Evia Kainada and Angelos Lengeris

with previous work on L2 intonation (e.g. Jun & Oh 2000; Mennen 2004, 2006) and with previous work on L2 segmentals (e.g. Best et al. 2001; Flege & MacKay 2004; Iverson & Evans 2007; Iverson et al. 2003; Lengeris 2009; Lengeris & Hazan, 2010). Regarding the use of pitch range in the two languages, the comparison of baseline productions by native speakers of Greek and English showed no cross- linguistic differences in the use of pitch span. However, Greek speakers’ pitch span in English was narrower from both their Greek pitch span and from the target (English) pitch span, indicating an L2 production effort strategy (Backman 1979; Willems 1982). This was corroborated by the finding that Greek speakers also showed lower pitch level in English than in Greek. The results of the current study put forward a number of directions for future research on the acquisition of English intonation by Greek speakers. After having established that intermediate learners transfer their L1 (Greek) tonal targets in English polar questions, the next step for this research entails a detailed phonetic analysis of the exact manifestation of those targets (e.g. F0 peak alignment) in English polar questions and an examination of how this compares to the alignment of tonal targets in L1 (Greek). Future plans also include an analysis of phonological targets and their phonetic manifestation in other types of sentences such as wh-questions, polar questions and statements by both intermediate and advanced Greek learners of English. For example, the learning of wh-questions by Greek speakers is another interesting L1/L2 case because Greek and English use completely different intonation patterns in signalling wh-questions; Greek uses a final rise pitch movement at the end of the question while English uses a final fall pitch movement. The ultimate goal of this research is to apply our understanding of the relationship between Greek and English intonation to improve the learning of English intonation by Greek speakers via the use of computer-based training. The training stimuli will comprise of intonation contours spoken by native English speakers and re-synthesized intonation patterns based on each trainee’s Greek-accented English production.

Acknowledgments

We would like to thank the students of the Foreign Language Center “Psychidou” for their participation, as well as an anonymous reviewer for their constructive comments.

Phonology - Phonetics 1 5 1 Major Trends in Theoretical and Applied Linguistics

References

Arvaniti, A. 2007. Greek Phonetics: State of the Art. Journal of Greek Linguistics 8: 97-208.

Arvaniti, A. 2009. Greek intonation and the phonology of prosody: polar questions revisited. Proceedings of the 8th International Conference on Greek Linguistics, 14-29.

Arvaniti, A. and Baltazani, M. 2005. Intonational analysis and prosodic annotation of Greek spoken corpora. In S.-A. Jun (ed.), Prosodic Typology: The Phonology of Intonation and Phrasing. Oxford University Press, 84-117.

Arvaniti, A., Ladd, D.R. and Mennen, I. 2006. Tonal association and tonal alignment: evidence from Greek polar questions and contrastive statements. Language and Speech 49: 421-450.

Atterer, M. and Ladd, D.R. 2004. On the phonetics and phonology of “segmental anchoring” of F0: evidence from German. Journal of Phonetics 32: 177-197.

Backman, N. 1979. Intonation errors in second language pronunciation of eight Spanish speaking adults learning English. Interlanguage Studies Bulletin 4(2): 239-266.

Baltazani, M. 2007. Intonation of polar questions and the location of nuclear stress in Greek. In C. Gussenhoven and T. Riad (eds.), Tones and Tunes, Volume II: Experimental Studies in Word and Sentence Prosody. Berlin: Mouton de Gruyter, 387-405.

Baltazani, M. and Jun, S.-A. 1999. Focus and Topic Intonation in Greek. Proceedings of the XIVth International Congress of Phonetic Sciences, San Francisco, 1305- 1308.

1 5 2 Phonology - Phonetics Evia Kainada and Angelos Lengeris

Best, C.T., McRoberts, G.W. and Goodell, E. 2001. Discrimination of non-native consonant contrasts varying in perceptual assimilation to the listener’s native phonological system. Journal of the Acoustical Society of America 109: 775-794.

Cebrian, J. 2006. Experience and the use of non-native duration in L2 vowel categorization. Journal of Phonetics 34: 372-387.

Derwing, T.M., Munro, M.J. and Wiebe, G. 1998. Evidence in favour of a broad framework for pronunciation instruction. Language Learning 48: 393-410.

Flege, J.E. 1995. Second language speech learning: Theory, findings, and problems. In W. Strange (ed.), Speech perception and linguistic experience: Issues in cross-language research. Baltimore: York Press, 233-277.

Flege, J.E. and Mackay, I.R.A. 2004. Perceiving vowels in a second language. Studies in Second Language Acquisition 26: 1-34.

Flege, J.E., MacKay, I.R.A. and Meador, D. 1999. Native Italian speakers’ production and perception of English vowels. Journal of the Acoustical Society of America 106: 2973-2987.

Grabe, E. 2004. Intonational variation in urban dialects of English spoken in the British Isles. In P. Gilles and J. Peters (ed.), Regional Variation in Intonation. Tuebingen: Niemeyer, 9-31.

Grabe, E., Post, B. and Nolan, F. 2001. Modelling intonational Variation in English. The IViE system. In S. Puppel and G. Demenko (ed.), Proceedings of Prosody 2000, Adam Mickiewitz University, Poznan, 51-57.

Guion, S.G., Flege, J.E., Akahane-Yamada, R. and Pruitt, J.C. 2000. An investigation of current models of second language speech perception: The case of Japanese adults’ perception of English consonants. Journal of the Acoustical Society of America 107: 2711-2724.

Hahn, L.D. 2004. Primary stress and intelligibility: research to motivate the teaching of suprasegmentals. TESOL Quarterly 38: 201-223.

Hattori, K. and Iverson, P. 2009. English /r/-/l/ category assimilation by Japanese adults: Individual differences and the link to identification accuracy. Journal of the Acoustical Society of America 125: 469-479.

Phonology - Phonetics 1 5 3 Major Trends in Theoretical and Applied Linguistics

Iverson, P. and Evans, B.G. 2007. Learning English vowels with different first language vowel systems: Perception of formant targets, formant movement, and duration. Journal of the Acoustical Society of America 122: 2842-2854.

Iverson, P., Kuhl, P.K., Akahane-Yamada, R., Diesch, E., Tohkura, Y. and Kettermann, A. 2003. A perceptual interference account of acquisition difficulties for non- native phonemes. Cognition 87: 47-57.

Jilka, M. 2000. The contribution of intonation to the perception of foreign accent. PhD dissertation, University of Stuttgart.

Jun, S.-A. and Oh, M. 2000. Acquisition of 2nd language intonation. Proceedings of International Conference on Spoken Language Processing, 76-79.

Kainada, E. 2010. The phonetic and phonological nature of prosodic boundaries: evidence from Modern Greek. PhD Thesis, University of Edinburgh.

Kang, O. 2010. Relative salience of suprasegmental features on judgments of L2 comprehensibility and accentedness. System 38: 301-315.

Kang, O., Rubin, D. and Pickering, L. 2010. Suprasegmental measures of accentedness and judgments of English language learner proficiency in oral English. Modern Language Journal 94: 554-566.

Ladd, D.R. 2008. Intonational Phonology. Cambridge University Press, 2nd Edition.

Lengeris, A. 2009. Perceptual assimilation and L2 learning: Evidence from the perception of Southern British English vowels by native speakers of Greek and Japanese. Phonetica 66: 169-187.

Lengeris, A. and Hazan, V. 2010. The effect of native vowel processing ability and frequency discrimination acuity on the phonetic training of English vowels for native speakers of Greek. Journal of the Acoustical Society of America 128: 3757-3768.

Mackay, I.R.A., Meador, D. and Flege, J.E. 2001. The identification of English consonants by native speakers of Italian. Phonetica 58: 103-125.

Mennen, I. 2004. Bi-directional interference in the intonation of Dutch speakers of Greek. Journal of Phonetics 32: 543-563.

1 5 4 Phonology - Phonetics Evia Kainada and Angelos Lengeris

Mennen, I. 2006. Phonetic and phonological influences in non-native intonation: an overview for language teachers. QMUC Speech Science Research Centre Working Papers WP-9.

Mennen, I. 2007. Phonological and phonetic influences in non-native intonation. In Trouvain, J. and Gut, U. (ed.), Non-native Prosody: Phonetic Descriptions and Teaching 2 Practice (Nicht-muttersprachliche Prosodie: phonetische Beschreibungen und didaktische Praxis). Mouton De Gruyter.

Mennen, I., Schaeffler, F. and Docherty, G. 2007. Pitching it differently: a comparison of the pitch ranges of German and English speakers. Proceedings of the 16th International Congress of Phonetic Sciences. University of Saarbruecken.

Munro, M. 1995. Nonsegmental factors in foreign accent. Studies in Second Language Acquisition 17: 17-34.

Munro, M. and Derwing, T. 1999. Foreign accent, comprehensibility, and intelligibility in the speech of second language learners. Language Learning 49 Supplement 1: 285-310.

Munro, M.J. and Derwing, T.M. 2001. Modelling perceptions of the comprehensibility and accentedness of L2 speech: The role of speaking rate. Studies in Second Language Acquisition 23: 451-468.

Pickering, L. 2001. The role of tone choice in improving ITA communication in the classroom. TESOL Quarterly 35: 233-255.

Pierrehumbert, J. 1980. The Phonology and Phonetics of English Intonation. PhD dissertation, MIT.

Polka, L. 1995. Linguistic influences in adult perception of non-native vowel contrasts. Journal of the Acoustical Society of America 97: 1286-1296.

Tserdanelis, G. 2003. Phonetic variation in syntactically ambiguous sentences. Studies in Greek Linguistics 23: 931-942.

Willems, N.J. 1982. English intonation from a Dutch point of view. Dordrecht: Foris Publications.

Phonology - Phonetics 1 5 5 Major Trends in Theoretical and Applied Linguistics

1 5 6 Phonology - Phonetics The Greek Rhotic in /rC/ Sequences: An Acoustic and Electropalatographic Study

Katerina Nicolaidis and Mary Baltazani

Aristotle University of Thessaloniki, University of Ioannina [email protected], [email protected]

Abstract

This paper analyses acoustic and electropalatographic data to examine rhotic production in /rC/ sequences in Greek. The rhotic was typically produced with a single constriction phase and a vocoid following the constriction and preceding the consonant. The duration data showed that the vocoid was longer than the constriction with formants similar to the nuclear vowel but more centralised. The articulatory analysis showed an alveolar rhotic with variation in place and degree of constriction (ranging from fully constricted to more open articulations) due to context and speaker. The data indicate tap production superimposed on a vocalic gesture which is necessary for the ballistic movement to occur.

1. Introduction

Rhotic production has been reported to vary within and across languages. Previous literature has documented variation in manner and place of articulation due to factors such as context, position, speech rate and dialect (Lindau 1985; Inouye 1995; Ladefoged & Maddieson 1996; Recasens & Pallarès 1999; Recasens & Espinosa 2007). Variability in rhotic production has also been reported for Greek. Descriptions of the Greek consonantal system mainly based on impressionistic auditory categorization have classified it as a trill (Φιλιππάκη-Warburton 1992; Nespor 1996) or as a tap (e.g. Arvaniti 1999). Experimental studies have verified tap production but have also shown approximant articulation (Nicolaidis 2001; Baltazani 2005, 2009). Variation in the place of articulation of the rhotic has also been reported ranging between alveolar, retracted alveolar and postalveolar productions (Nicolaidis 2001; Nicolaidis and Baltazani 2011; Baltazani and Nicolaidis (2012)). Further context-based variation has also been reported. In particular, tap productions have been described to occur in initial and intervocalic position (Arvaniti 1999; Nicolaidis 2001; Baltazani 2005, 2009) while in clusters with stops or fricatives there is evidence of the production of a vocoid between

Phonology - Phonetics 1 5 7 Major Trends in Theoretical and Applied Linguistics

the consonant and the rhotic (Baltazani 2009; Nicolaidis and Baltazani 2011). The nature of the latter articulation, which involves the presence of a vocoid, has been debated in the literature. On the one hand, it has been claimed to indicate trill production (Arvaniti 2007). On the other, it has been interpreted to manifest the production of a tap which involves a vowel-like transition between the consonant and the rhotic (Baltazani 2009). Evidence of such realization of the rhotic in clusters, i.e. as a combination of a tap and a vowel-like interval, has also been reported for other languages including Spanish, Catalan, French, Finnish, Hungarian, Hamburg German and Romanian (Navarro Tomás 1918; Rochette 1973; Harms 1976; Jannedy 1994; Bradley & Schmeiser, 2003; Bradley 2004; Colantoni & Steele, 2005; Recasens & Espinosa, 2007; Vago & Gósy 2007; Savu 2012). The current study aims to investigate variation in the production of the Greek rhotic in /r-stop/ and /r-fricative/ sequences. It is part of a larger project which analyses simultaneous acoustic and articulatory data on the Greek rhotic in order to examine variation due to position and context. This paper examines duration and formant frequency data aiming to investigate the realization of both the constriction phase and the vocoid in these /rC/ sequences. Formant frequency data are also examined for the vowels preceding and following the sequences in order to explore their influence on the vocoid realization. The paper also analyses articulatory data in order to examine variability in rhotic production as a function of context and speaker. Overall, the study aims to examine (a) whether the vocoid is systematically present during rhotic production in /rC/ sequences, (b) whether the vocoid is a neutral vowel or affected by context, (c) the range of duration of the vocoid compared to the constriction phase, (d) the influence of the place and manner of articulation of the following consonant on the articulation of the rhotic, (e) the influence of the vocalic environment on the acoustic and articulatory characteristics of the rhotic (see also (b)), and (e) evidence suggesting a tap vs. trill production in /rC/ sequences.

2. Methodology

2.1. Speech Material and Subjects

Words containing a /VrCV/ sequence were recorded with C = /p, t, k, f, θ, x/ and V= /i, e, a, o, u/1. The test words were embedded in the carrier phrase

1 In this paper, the /r/ symbol is used for the Greek rhotic for practical reasons. Similarly, the vowels are transcribed as /i, e, a, o, u/. For a description of the quality of the Greek vowels, see Arvaniti (1999, 2007).

1 5 8 Phonology - Phonetics Katerina Nicolaidis and Mary Baltazani

[i �leksi ___�ine a�pli] ‘The word _ is simple’. Only symmetrical sequences were recorded, i.e. with the same vowel before and after the /rC/ sequence. All words were up to four syllables long. The location of the stress varied so that the majority of items in the speech material were real words (see appendix). Five Greek adult speakers (three female and two male) repeated the speech material five times at a comfortable speaking rate in a sound-treated room in the Phonetics Laboratory of the School of English at Aristotle University. In total, 750 words (6 Cs x 5 Vs x 5 repetitions x 5 speakers) were recorded.

2.2. Data Recording and Analysis

The British EPG system marketed by Articulate Instruments was used to record acoustic and electropalatographic (EPG) data simultaneously. In addition, acoustic data were independently recorded using a Marantz PMD 660 digital recorder with a Røde NT1-A cardiod condenser microphone and were subsequently analysed using PRAAT. For segmentation and annotation of the rhotic, the different phases of its production were identified on the acoustic data. The rhotic typically consisted of two phases, i.e. a constriction phase and a vocoid following the constriction and preceding the obstruent in the /VrCV/ sequences. The constriction phase boundaries were identified at the offset of the formant structure of the preceding vowel and at the onset of following vocoid formants. The vocoid boundaries were identified at the onset and offset of voiced formant structure (Figure 1), or else, in cases of the presence of frication noise or a whispered vocoid, at the onset/offset of voiceless formant

Figure 1. Example of segmentation of the /rC/ sequence in the word [skorpo’xori] (the labels ‘a1’, ‘a2’ correspond to the preceding and following vowels respectively, ‘r1’ to the constriction phase of the rhotic, ‘r2’ to the vocoid phase of the rhotic, ‘c’ to the obstruent following the rhotic).

Phonology - Phonetics 1 5 9 Major Trends in Theoretical and Applied Linguistics

structure (Figure 2). Measurements of F1, F2, F3 formants of both the vocoid and the flanking vowels in the /VrCV/ sequence were taken at the ¼, ½, and ¾ points. In this paper, we report results only for F1 and F2 at the midpoint of the vocoid and the preceding tautosyllabic vowel. Electropalatography records lingual contact with the palate in continuous speech. A thin custom-made artificial palate embedded with electrodes on its surface is worn by the subjects to register linguo-palatal contact. For the British system used in this study, the artificial palate has 62 electrodes on its surface which are distributed in eight rows (Figure 3). The front four rows correspond to the alveolar zone and the back four rows to the palatal zone. The alveolar zone is subdivided to the alveolar and postalveolar regions (rows 1 to 2 and 3 to 4 respectively). The first two columns on the left and right side of the palate are characterized as lateral and the four remaining columns as central (Gibbon & Nicolaidis 1999; Recasens et al 1993).

Figure 2. Example of segmentation of the /rC/ sequence in the word [‘tirfi] (for the labels used, see caption of figure 1).

zones subzones

front alveolar alveolar postalveolar

prepalatal palatal mediopalatal

postpalatal

Figure 3. The artificial palate used in electropalatography for the British system marketed by Articulate instruments (left). Schematic representation of electrodes and division of electrodes into zones and subzones (right).

1 6 0 Phonology - Phonetics

Katerina Nicolaidis and Mary Baltazani

For the articulatory analysis, Articulate Assistant (version 1.18) was used for data display, segmentation and analysis. The place of articulation of the rhotic always occurred in the alveolar zone. Thus, the first EPG frame of maximum contact/constriction in the four front rows was identified for each rhotic (Figure 4). For the majority of tokens, the frame of maximum contact coincided with the frame of maximum constriction. When it did not, the frame of maximum constriction in the alveolar zone was annotated. The percentage frequency of electrode activation of the entire palate at the frame of maximum contact/ constriction over five repetitions was then calculated for each word (see, e.g., Figure 12).

Figure 4. Screen display from Articulate Assistant showing acoustic waveform, spectrogram, EPG palatograms and contact totals display (alveolar total) for the rhotic in the word [‘arpa] produced by speaker MM. The annotation line corresponds to the first frame of maximum contact/constriction in the alveolar zone. The corresponding palatogram is shown at the top right of the display.

3. Results

3.1. Acoustic Analysis

The acoustic analysis showed that the rhotic in /rC/ sequences is typically a tap which consists of two phases: a single constriction phase and a short vocoid appearing between the rhotic constriction and the obstruent (see section 3.2 for details). A tap realization of the rhotic was also reported for /Cr/ clusters in Nicolaidis & Baltazani (2011), where only 5 tokens out of a total of 750 tokens were produced with a second very open approximation phase suggesting

Phonology - Phonetics 1 6 1 Major Trends in Theoretical and Applied Linguistics

possible trill production. A tap realisation was also reported in Baltazani & Nicolaidis (2012) with the exception of 10 trills out of the 375 tokens analysed. These results clearly demonstrate that the Greek rhotic is not a trill but involves a single tap-like gesture superimposed on a vocalic one. Moreover, the acoustic measurements revealed variability in the structure of both rhotic components in /rC/ sequences (Figures 5-7). Specifically, the constriction phase varied between tokens with presence of silence and a burst, suggesting complete articulatory closure (Figure 5), and tokens with frication noise (Figure 6), suggesting incomplete closure, something which is corroborated by the articulatory analysis (section 3.2). Further variability was observed in the vocoid production which ranged from a modal vowel structure (Figures 5 and 6) to a whispered one (Figure 7). Finally, observation of the rhotic data in /rC/ sequences examined in this paper vs. /Cr/ clusters examined in previous work (Nicolaidis & Baltazani 2011) indicates that there was a tendency for more tokens with whispered vocoids or frication noise during the constriction phase in heterosyllabic /rC/ sequences than in homosyllabic / Cr/ clusters. This suggests more assimilatory effects of the following voiceless obstruent in /rC/ sequences.

Figure 5. A token of /r/ in the word [‘fortos] produced by speaker MM with acoustic evidence suggesting complete closure during the constriction phase and modal vowel quality for the vocoid (for the labels used, see caption of Figure 1).

Figure 6. A token of /r/ in the word [‘arpa] produced by speaker TP with frication noise during the constriction phase (for the labels used, see caption of Figure 1).

1 6 2 Phonology - Phonetics Katerina Nicolaidis and Mary Baltazani

Figure 7. A token of /r/ in the word [‘tirfi] produced by speaker MM with acoustic evidence suggesting complete closure during the constriction phase and a whispered vowel quality during the vocoid (for the labels used, see caption of Figure 1).

As far as the duration of the rhotic components is concerned, in /rC/ sequences the vocoid was found to be longer than the constriction phase across experimental conditions. Table 1 presents average durations for the constriction and vocoid for all vocalic contexts across speakers and repetitions. The vocoid in all vocalic contexts is on average 7ms longer than the constriction. Longer duration for the vocoid compared to the constriction phase was also reported in Nicolaidis & Baltazani (2011) for /Cr/ clusters as well as in Baltazani & Nicolaidis (2012) for /r/ in phrase initial position.

Table 1. Average durations for the constriction and vocoid phases of the rhotic in all vocalic contexts.

V context Constriction duration (ms) Vocoid duration (ms)

i 28 37

e 25 31

a 26 31

o 22 30

u 23 32

Average 25 32

Variability both in the duration of the rhotic components and the duration differences between the two components is shown in Figure 8, where results from each of the five speakers are shown (from top left to bottom right, speakers TP, RP, KN, AT, and MM). Speaker AT (bottom left) produced the shortest segments and the smallest differences between the constriction (shown in black) and vocoid duration (grey), speaker TP (top left) produced the longest segments, while for speaker MM (bottom right) there was the biggest durational difference between the constriction and the vocoid.

Phonology - Phonetics 1 6 3 Major Trends in Theoretical and Applied Linguistics

Figure 8. Speaker variability in the duration of the constriction phase (r1=black) and the vocoid (r2=grey). Results for speakers TP, RP, KN, AT, and MM are shown from top left to bottom right.

The F1, F2 formants of the vocoid and the nuclear vowel V1 in /rC/ sequences were compared to determine the eff ect of the latter on the former. The results showed that the vocoid formant values are similar to those of the nuclear vowel (V1), with a clear tendency to be more centralized than V1. However, there is variability in the amount of centralization among vocalic contexts and speakers, as Figure 9 shows. For example, in the context of non- high vowels, speakers TP (top left), RP (mid left) and MM (bottom) produced the vocoid with considerably higher F1 values compared to V1, whereas the other two speakers did not show such large diff erences. Variability was also evident in the context of high vowels, e.g. for MM (bottom) large diff erences in F2 between the vocoids and the nuclear vowels /i/ and /u/ are evident showing considerable centralization of the vocoids; relatively large diff erences in F2 are also evident between the vocoid and /u/ for KN (top right) and the vocoid and /i/ for AT (middle right). The other speakers do not show such large diff erences. As can be seen, therefore, the data show a centralization tendency with notable variability in the F1, F2 values of the vocoid across speakers and vocalic contexts.

1 6 4 Phonology - Phonetics Katerina Nicolaidis and Mary Baltazani

Figure 9. Comparison of V1 (black line) and vocoid (grey line) formant values for the fi ve speakers, from top left to bottom, TP, KN, RP, AT, MM.

The amount of variability in the realization of both the vocoid and V1 can be appreciated in Figure 10, which presents the Euclidean distance between the vocoid and V1 across speakers and vocalic contexts. This distance is the square root of the sum of the squares of the diff erence between the vowel formant frequencies2 of each V1 (nuclear vowel) and corresponding V2 (vocoid). Bigger numbers indicate greater distance between the vocoid and V1 formants, hence a greater amount of centralization for the vocoid. The most consistent trend is variation, since no generalized statement can be made across speakers or vocalic contexts3.

2 ED= √ ((F1V1–F1V2)2 + (F2V1–F2V2)2) where V1=nuclear vowel and V2=vocoid. 3 Note that TP shows less variability across vocalic contexts compared to the other speakers.

Phonology - Phonetics 1 6 5 Major Trends in Theoretical and Applied Linguistics

Figure 10. The Euclidean distance between the vocoid and V1 across speakers and vocalic contexts.

3.2. Articulatory Analysis

The electropalatographic data also showed that there was typically one constriction present during the production of the rhotic in /rC/ sequences which was followed by a vocoid before the obstruent (Figure 11). Thus, /rC/ sequences typically involve a tap-like gesture and not a trill. This gesture varied in terms of the degree of constriction involving instances of complete closure across a row of electrodes and tokens with incomplete closure. Diff erent realizations of incomplete closure were evident which related to the degree of approximation of the lingual gesture to the palate. They ranged from cases with considerable approximation to others involving very open articulations. Such articulatory manifestations related to variation in the acoustic signal with evidence of a stop-like pattern in cases of complete constriction and of approximant formant transitions in cases of relatively open articulations. Presence of noise was also evident in several tokens involving incomplete constriction. Figure 11(a, b) presents two repetitions of the rhotic in the word /’tirpis/ produced by the same speaker with complete and incomplete constriction. Overall, 57% of the rhotics were realized with incomplete constriction (426 out of 749 tokens). Table 2 shows the number of tokens produced with incomplete constriction for the fi ve speakers. Evidence of considerable speaker variability is observed with over 70% of tokens produced with incomplete constriction by speaker KN and just 25% by MM. In addition, more open productions were found in the environment of a following fricative compared to the stop context (250 vs. 176), (see also the analysis of contextual variation below).

1 6 6 Phonology - Phonetics Katerina Nicolaidis and Mary Baltazani

Figure 11 (a, b). Acoustic and electropalatographic data for the rhotic in the word [‘tirpis] produced by speaker MM. In (a) the rhotic is produced with complete constriction and a stop-like pattern on the acoustic data; in (b) there is incomplete constriction and presence of formant structure. The corresponding palatograms at the point of maximum contact/constriction are shown at the right top of the displays.

Table 2. Number of tokens produced with incomplete constriction in /r-stop/ and /r-fricative/ sequences for the five speakers.

KN AT RP MM TP

r-stop 59 41 34 14 28

r-fric 71 62 48 30 39

Total 130 103 82 44 67

Phonology - Phonetics 1 6 7 Major Trends in Theoretical and Applied Linguistics

Contextual influence on the place of articulation of the rhotic was also evident. More fronted productions were found in the environment of the front vowels /i, e/ and more retracted productions in the context of the back vowels. In some consonantal contexts, the rhotic was most retracted when next to /a/ and /o/. Figure 12 shows that constriction is more anterior in the environment of /i/, i.e. on the first and second row of electrodes; constriction gradually retracts in the other vowel contexts with contact on the 2nd-3rd rows in the context of /e/, third row in the context of /a/ and 3rd-4th rows in the context of /o, u/. Thus, while the rhotic is always articulated in the alveolar zone, its place of articulation ranges between alveolar, retracted alveolar, advanced postalveolar or postalveolar depending on the vocalic context. Some evidence of contextual influence on the place of articulation of the rhotic due to the following consonantal context was also found, although large variation was present among speakers and contexts. Figure 13 shows an example of more anterior production of the rhotic in the environment of /t/ compared to the bilabial/irpi/ context, which/erpe/ is more evident/arpa/ mainly in/orpo/ the context of/urpu/ /a, o, u/.

/irpi/ /erpe/ /arpa/ /orpo/ /urpu/

Figure 12. EPG palatograms displaying percentage frequency of electrode activation over five repetitions during the production of the /r/ in /rp/ sequences by speaker MM.

irpi erpe arpa orpo urpu

irti erte arta orto urtu

Figure 13. EPG palatograms displaying percentage frequency of electrode activation over five repetitions during the production of the /r/ in /rp/ and /rt/ sequences by speaker AT.

1 6 8 Phonology - Phonetics Katerina Nicolaidis and Mary Baltazani

An effect of the consonantal context on the degree of rhotic constriction was found. More open productions were evident in the environment of a following fricative compared to the stop context (Table 2). A comparison of the palatograms for the /rp/ sequences in Figure 13 (top) with the /rf/ sequences in Figure 14 produced by speaker AT shows differences in the degree of constriction of the rhotic, i.e. more open articulation in the environment of the fricative /f/ compared to the stop /p/. This was true for the rhotic not only in the context of a lingual consonant, dental or velar, where the tongue is involved both in the articulation of the rhotic and the consonant, but also in the labial consonantal environment which involves an independent articulatory system.

Variation irfi was alsoerfe evident in thearfa articulationorfo of the rhoticurfu as a function of the speaker. Variability was found both within and across speakers. Figure 15 shows differences in the degree of constriction and place of articulation of the /r/ in /rk/ sequences for speakers TP and KN. The former shows more constricted

irfi erfe arfa orfo urfu

Figure 14. EPG palatograms displaying percentage frequency of electrode activation over five repetitions during the production of the /r/ in /rf/ sequences by speaker AT.

/irki/ /erke/ /arka/ /orko/ /urku/

Figure 15. EPG palatograms displaying percentage frequency of electrode activation over five repetitions during the production of the /r/ in /rk/ sequences by speakers TP (top) and KN (bottom).

Phonology - Phonetics 1 6 9 Major Trends in Theoretical and Applied Linguistics

productions compared to the more open articulations of the latter. In addition, more fronted production is evident for TP. Differences in the overall amount of contact in the alveolar and palatal zones between the speakers are also evident with generally more contact in the alveolar zone for TP and in the palatal zone for KN. Finally, there were six tokens which were produced as trills with two constriction phases present and another six tokens which showed a possible second very open approximation gesture (Figure 16).

Figure 16. Acoustic and electropalatographic data for the rhotic in the word [‘ferte] produced by speaker TP. Trill production of the rhotic is illustrated. The annotation line corresponds to the first frame of maximum contact/constriction in the alveolar zone during the first constriction phase. The corresponding palatogram is shown at the top right of the display. The palatograms below the spectrogram show presence of complete closure during both constriction phases.

4. Discussion and Conclusion

Both the acoustic and articulatory analysis showed that the rhotic is typically produced with a single constriction of short duration followed by a vocoid interval. The presence of the single constriction indicates the production of a tap and not a trill in /rC/ sequences. The duration of the constriction ranged between 22 and 28 ms, displaying considerably less variability than the constriction duration range of 11-50ms reported for /Cr/ clusters in Nicolaidis and Baltazani (2011), (cf. Nicolaidis 2001; Arvaniti 2007; Baltazani 2005, 2009). A very limited number of tokens (6 or possibly 12) out of the 749 analysed were produced as trills.

1 7 0 Phonology - Phonetics Katerina Nicolaidis and Mary Baltazani

As mentioned in the introduction, the presence of the vocoid during rhotic production in clusters and sequences has been documented in several languages (see section 1). Different interpretations have been offered for the presence of this vocoid. On the one hand, it has been interpreted to result from lack of temporal overlap between the articulatory gestures for the /r/ and the other consonant in the cluster or sequence (Riera & Romero 2006; Romero 1996, 2008; Bradley 2004). On the other hand, Blecua (2001) argues that the vocoid is an inherent part of the rhotic itself based on the observation that the formant structure of the vocoid is similar to that of the tautosyllabic vowel. While the presence of the vocoid has been documented in clusters and sequences, useful insights on the articulatory behaviour of the rhotic can be gained from the study of the rhotic in phrase initial or intervocalic position where its articulation is not affected by a preceding or following consonant. Data reported in Baltazani & Nicolaidis (2012) show that the production of the rhotic in phrase initial position also involves a vocoid preceding the constriction phase. In addition, preliminary evidence of the presence of the vocoid when the rhotic is produced intervocalically is reported in the same study. Unpublished data on word final /r/ in Greek also document the presence of the vocoid following the constriction. These findings suggest that the rhotic is superimposed on a rhotic- specific vocalic gesture which is needed for the ballistic movement to occur. The vocoid is thus an inherent part of the rhotic itself (cf. Blecua 2001 for clusters and sequences). This interpretation is corroborated by the results of the acoustic analysis. The formant structure of the vocoid was similar but more centralised than the nuclear vowel, suggesting that its quality depends on the nuclear vowel. Similar results have been reported in Nicolaidis & Baltazani (2012) for /r/ in phrase initial position. The similarity between the vocoid and the vowel on the other side of the tap constriction, especially for /Cr/ and /rC/ sequences, has been documented for other languages as well (e.g. Quillis 1993; Bradley 2001; Blecua 2001; Ramírez 2006; Savu 2012). Further support for the tap interpretation is provided by the duration differences found between the vocoid and the constriction. Trills are typically produced with vocalic and consonantal intervals of equal durations (e.g. Recasens 1991), while our findings demonstrate that the vocoid is consistently longer than the constriction suggesting that the rhotic is not a trill. The articulatory data provided evidence of variation in place and degree of constriction due to the context and the speaker. While the rhotic is always articulated in the alveolar zone (the four front rows of electrodes), its place varies due to the vocalic context and ranges between alveolar, retracted alveolar, advanced postalveolar and postalveolar. More advanced articulation is present in the front vowel contexts while more retracted production is overall evident in the back vowel environments. Some evidence of contextual influence on

Phonology - Phonetics 1 7 1 Major Trends in Theoretical and Applied Linguistics

place of articulation from the following consonant was also present, although large variation was found. Rhotic production also varied in terms of the degree of constriction. Presence of complete constriction across one or more rows of electrodes was evident for some tokens while incomplete constriction was found for others. Tokens with incomplete constriction ranged from very constricted to very open articulations. The corresponding acoustic data included tokens with a stop-like pattern when complete constriction was present, tokens with presence of formants throughout the constriction when there was incomplete constriction indicating approximant production, and tokens with presence of noise during incomplete constriction. Similar findings have been reported for /Cr/ clusters and singleton /r/ in different prosodic positions in Greek (Nicolaidis and Baltazani 2011; Baltazani and Nicolaidis (2012)). In total, 57% of the tokens were produced with incomplete constriction and more tokens with an open articulation were found in the context of a fricative compared to a stop. Important speaker variability in the total number of tokens produced with incomplete constriction was however found suggesting individual strategies in production. Inter- and intra-speaker variability was also found in the place of articulation of the rhotic, in the degree of constriction and in the total amount of contact in the alveolar and palatal zones. Overall, the results on the contextual influence, in particular, evidence of V-to-r and C-to-r contextual effects, indicate that the tongue tip/blade coarticulates with neighbouring gestures during the production of the rhotic in /rC/ sequences similarly to findings reported for /Cr/ clusters and singleton /r/ in Greek (Nicolaidis and Baltazani 2011; Baltazani and Nicolaidis (2012)). Coarticulatory effects on the rhotic have also been reported for other languages (e.g. Recasens 1991; Recasens and Pallarès 1999). In conclusion, the acoustic and articulatory data of this study and previous studies (Nicolaidis and Baltazani 2011; Baltazani and Nicolaidis (2012)) provide evidence that the Greek rhotic involves a single constriction produced by a rapid ballistic movement which is superimposed on a vocalic gesture. The presence of the vocoid in /rC/ sequences is part of this rhotic-specific vocalic gesture and is necessary for the ballistic movement to take place.

1 7 2 Phonology - Phonetics Katerina Nicolaidis and Mary Baltazani

References

Arvaniti, A. 2007. Greek phonetics: The state of the art. Journal of Greek Linguistics 8: 97-208.

Arvaniti, A. 1999. Standard Modern Greek. Journal of the International Phonetic Association 29: 167-172.

Baltazani, M. 2005. Phonetic variability of the Greek rhotic sound. Phonetics and Phonology in Iberia (PaPI) 05. Barcelona, Spain. Poster available at http://users. uoi.gr/mbaltaz/en/Papi05.

Baltazani, M. 2009. Acoustic characterization of the Greek rhotic in clusters. Selected Papers from the 18th ISTAL, Aristotle University, Thessaloniki, Greece, 87-95.

Baltazani, M. and K. Nicolaidis 2012. Production of the Greek rhotic in initial and intervocalic position: an acoustic and electropalatographic study. In Z. Gavriilidou, A. Efthymiou, E. Thomadaki & P. Kambakis-Vougiouklis (eds) Selected Papers of the 10th International Conference on Greek Linguistics, Komotini,Greece: Democritus University of Thrace, 141-152.

Blecua Falgueras, B. 2001. Las vibrantes del español: manifestaciones acústicas procesos fonéticos. Ph.D. dissertation. Universidad Autónoma de Barcelona.

Bradley, T.G. 2001. The phonetics and phonology of rhotic duration contrast and neutralization. Ph.D. dissertation. Pennsylvania State University.

Bradley, T.G. 2004. Gestural timing and rhotic variation. In T.L. Face, (ed.), Laboratory Approaches to Spanish Phonology. Berlin: Mouton de Gruyter, 195- 220.

Phonology - Phonetics 1 7 3 Major Trends in Theoretical and Applied Linguistics

Bradley, T.G. and B.S. Schmeiser. 2003. On the Phonetic Reality of Spanish /r/ in Complex Onsets. In P.M. Kempchinsky and C.-E. Piñeros (eds), Theory, Practice, and Acquisition: Papers from the 6th Hispanic Linguistics Symposium and the 5th Conference on the Acquisition of Spanish and Portuguese. Somerville, MA: Cascadilla Press, 1-20.

Colantoni, L. and J. Steele. 2005. Liquid asymmetries in French and Spanish. Toronto Working Papers in Linguistics 24: 1-14.

Gibbon, F. and K. Nicolaidis. 1999. Palatography. In W. Hardcastle and N. Hewlett (eds), Coarticulation. Cambridge: CUP, 229-245.

Harms, R.T. 1976. The segmentalization of Finnish ‘nonrules’. Texas Linguistic Forum 5: 73–88.

Inouye, S. 1995. Trills, Taps and Stops in Contrast and Variation. Ph.D. Dissertation, University of Los Angeles, California.

Jannedy, S. 1994. Rate effects on German unstressed syllables. OSU Working Papers in Linguistics 44, 105–124.

Ladefoged, P. and I. Maddieson. 1996. The Sounds of the World’s Languages. Cambridge: Blackwell.

Lindau, M. 1985. The story of /r/. In V.A. Fromkin (ed.), Phonetic Linguistics: Essays in Honor of Peter Ladefoged. Academic Press, 157-168.

Navarro Tomás, T. 1918. Diferencias de duración entre las consonantes españolas. Revista de Filología Española 5: 367-393.

Nespor, M. 1996. Φωνολογία. Athens: Patakis.

Nicolaidis, K. 2001. An electropalatographic study of Greek spontaneous speech. Journal of the International Phonetic Association 31: 67-85.

Nicolaidis, K. and M. Baltazani. 2011. An electropalatographic and acoustic study of the Greek rhotic in /Cr/ clusters. In Proceedings of the 17th International Congress of Phonetic Sciences, City University of Hong-Kong, 1474-1478.

Quilis, A. 1993. Tratado de fonología y fonética españolas. Madrid: Editorial Gredos.

Ramírez, C.J. 2006. Acoustic and Perceptual Characterization of the Epenthetic Vowel between the Clusters Formed by Consonant + Liquid in Spanish. In M.

1 7 4 Phonology - Phonetics Katerina Nicolaidis and Mary Baltazani

Díaz-Campos (ed.), Selected Proceedings of the 2nd Conference on Laboratory Approaches to Spanish Phonetics and Phonology. Somerville, MA: Cascadilla Proceedings Project, 48-61.

Recasens, D. 1991. On the production characteristics of apicoalveolar taps and trills. Journal of Phonetics 19: 267-280.

Recasens, D. and A. Espinosa. 2007. Phonetic typology and positional allophones for alveolar rhotics in Catalan. Phonetica 63: 1-28.

Recasens, D., E. Farnetani, J. Fontdevila and M.D. Pallarès. 1993. An electropalatographic study of alveolar and palatal consonants in Catalan and Italian. Language and Speech 36: 213-234.

Recasens, D. and M.D. Pallarès. 1999. The study of /ɾ/ and /r/ in the light of the ‘DAC’ coarticulation model. Journal of Phonetics 27: 143-169.

Riera, M. and J. Romero. 2006. V+/l/ and V+/r/ sequences in American English: a preliminary acoustic study. In Proceedings of the 29th AEDEAN International Conference, Jaén, Spain: Universidad de Jaén, Servicio de Publicaciones, 529-536.

Rochette, C.-E. 1973. Les groupes de consonnes en français: étude de l’enchaînement articulatoire à l’aide de la radiocinématographie et de l’oscillographie. Quebec: Les presses de l’Université Laval.

Romero, J. 1996. Articulatory blending in lingual gestures. Journal of Phonetics 24: 99-111.

Romero, J. 2008. Gestural Timing in the Perception of Spanish r+C Clusters. In L. Colantoni and J. Steele (eds), Selected Proceedings of the 3rd Conference on Laboratory Approaches to Spanish Phonology. Somerville, MA: Cascadilla Proceedings Project, 59-71.

Savu, C.F. 2012. An Acoustic Phonetic Perspective on the Phonological Behavior of the Rhotic Tap. Presentation in ConSOLE XX, University of Leipzig.

Vago, R. and M. Gósy. 2007. Schwa vocalization in the realization of /r/. Proceedings of the 16th International Congress of Phonetic Sciences, Saarbrücken Germany, 505-509.

Φιλιππάκη-Warburton, Ε. 1992. Εισαγωγή στη Θεωρητική Γλωσσολογία. Athens: Nefeli.

Phonology - Phonetics 1 7 5 Major Trends in Theoretical and Applied Linguistics

Appendix

The experimental material used for /rC/ sequences. The words in the table below were embedded in a carrier phrase.

/a/ /e/ /i/ /o/ /u/

/p/ ‘arpa erpe’to ‘tirpis skorpo’xori ɣur’punis

/t/ ‘arta ‘ferte ‘sirtis ‘fortos sur’tuko

/k/ ‘marka ‘perces ‘circi ‘orkos kur’kuti

/f/ kar’fara aðer’fe ‘tirfi ‘korfos fur’furi

/θ/ ‘marθa ‘erθete ‘sirθice or’θos sur’θun

/x/ karxa’rias ‘erçete ‘irçisan orxome’nos xurxu’rizo

1 7 6 Phonology - Phonetics Where the Glide Meets the Palatals*

Nina Topintzi and Mary Baltazani

Leipzig University, University of Ioannina [email protected], [email protected]

Abstract

This paper explores the relationship of the glide /J/ to the high front vowel and the palatal consonants. We argue that /i/ and /J/ are both phonemes, which nonetheless undergo contextual neutralization in a large class of nouns under paradigmatic pressure. The proposed phonemic status of the glide has repercussions for the other palatal consonants, i.e. those deriving from velar obstruents or the nasal and lateral coronals (collectively notated as VNLs). We claim that palatals are always derived either through simple or extreme palatalization; in Standard Greek, the former only affects velars, resulting in simple allophony between velars and palatals, whereas the latter takes VNL + vocoid input sequences and returns palatalized segments with simultaneous partial or full absorption of the vocoid into the now palatalized preceding consonant (cf. Bateman 2007).

1. Introduction

This paper revisits a famous conundrum in Greek phonology, namely that of “the glide [j]”, which sometimes behaves as an allophone of /i/ (1) and sometimes as a phoneme distinct from it (2)1.

(1) [i]~[j] alternations pó.ði “foot” pó.ðʝa “feet” ðo.ká.ri “girder” ðo.kár.ʝa “girders” má.ti “eye” má.tça “eyes”

* Many thanks to an anonymous reviewer for constructive feedback. All remaining errors are our own. 1 The phonetic manifestation of the glide is variable, which is why we will occasionally use the generalized notation /J/ or glide when this is not at stake. Notably, this notation is not to be confused with an archiphonemic representation. As an onglide, /J/ generally surfaces as [ʝ] if a voiced consonant precedes it, e.g. pó.ðʝa “feet”; if a voiceless consonant precedes it, then it emerges as [ç], e.g. mátça “eyes”. Finally, /l/ or /n/ + J + V → [ʎ, ɲ], e.g. ku.kʎá “dolls”, pa.ɲá “cloths”. If it is an offglide, then it is realized as an approximant, e.g. γáj.ða.ɾos “donkey”.

Phonology - Phonetics 1 7 7 Major Trends in Theoretical and Applied Linguistics

(2) [i]~[j] contrast á.ði.a “permission” á.ðʝa “empty” sci.á.zo “shade” scá.zo “scare”

Common denominator in most previous accounts of the glide’s behavior is the appeal to diachrony (given that the glide historically evolved from [i]) or to diglossia (since in cases of alternation, [i] is considered more part of the purified katharevusa, whereas the glide as part of the colloquial dhimotiki). While it is true that both factors are influential to an extent, we argue that much of the data can be understood synchronically utilizing grammatical considerations only. In this paper, we focus on onglides only and argue that /J/ must be recognized as a distinct phoneme in Greek2. We also show that in specific morpho-phonological contexts, it may neutralize with /i/ and we provide an analysis. Beyond capturing the distinctive as well as allophonic nature of the glide, we claim that the introduction of the phoneme /J/ allows us to better understand the related class of palatals, the classification of which has also remained elusive so far.

2. Contrast vs. Alternation

The literature on the phonological nature of the Greek glide is considerable. Two main trends can be identified. On the one hand, there are accounts that propose a single phoneme /i/ (Kazazis 1968; Malavakis 1984; Nikolopoulos 1985; Warburton 1976) or an underspecified archiphoneme /I/ (Deligiorgis 1987; Malikouti-Drachman & Drachman 1990) from which both [i] and [J] derive. These accounts fare well with regard to the [i]~[J] alternations of (1), but are less successful when dealing with cases of contrast (2). The opposite situation holds for accounts proposing two phonemes /i/ and /j/ (Holton 1997; Householder 1964; Koutsoudas 1962; Mirambel 1959; Nyman 1981; Setatos 1974). We argue that both approaches are right in some sense, but need to be combined before they capture the empirical results adequately. We thus propose that Greek contains both phonemes /i/ and /J/ which may either surface faithfully (in which case, we get instances of contrast) or neutralize and surface as [i] or [J] depending on the environment. The latter, we claim, is often predictable on mopho-phonological grounds, as we show next. The assumed representation is schematized in (3) and is in line with comparable structures for glides in Karuk, Pulaar and Sundanese as argued by Levi (2008).

2 A separate argument for a phoneme /J/ comes from data with offglides. See Baltazani & Topintzi (2012) for details.

1 7 8 Phonology - Phonetics Nina Topintzi and Mary Baltazani

(3) /i/ /J/ Input

/ i  i, J J Phonological representation

prevoc. postvoc.

[i] [ʝ] [ç] [ʎ] [ɲ] [j] Phonetic representation 2.1. Alternation and Lack Thereof in Morphophonology

As mentioned before, we acknowledge that historical and sociolinguistic factors should be taken into consideration if one is to attain a full account of the [i]~[J] alternation (for recent work in that direction, see Rytting 2005), however we argue that there is more room for grammatical factors coming into play than usually assumed. We presently explore an instance of such sort. In particular, consider the large class of neuter nouns ending in –i vs. those ending in –(i)o. The former, but not the latter, present [i]~[J]3 alternations (4). Representative paradigms follow in (5).

(4) a. Alternations SINGULAR PLURAL máti mátça “eye” peðí peðʝá “child” xoráfi xoráfça “field” b. No alternation SINGULAR PLURAL i. ðomátio ðomátia “room” sçéðio sçéðia “sketch, plan” peðío peðía “field, plain” ii. psóɲo psóɲa “shopping” órɲo órɲa “vulture”

(5) a. paradigm with alternation b. paradigm without alternation singular plural singular plural nom. máti-Ø mátç-a ðomáti-o ðomáti-a gen. matç-ú matç-ón ðomatí-u ðomatí-on acc. máti-Ø mátç-a ðomáti-o ðomáti-a voc. máti-Ø mátç-a ðomáti-o ðomáti-a

3 Please see footnote 1 for an explanation of what /J/ or glide mean.

Phonology - Phonetics 1 7 9 Major Trends in Theoretical and Applied Linguistics

A generalization out of these facts is that nouns with the structure Stem+Ø in the NOM.SG present glide formation in the Plural and in the GEN.PL; on the other hand, the Stem+o ones in the NOM.SG either present hiatus (ðomátio) or glide formation (psóɲo) throughout.

2.2. An OT Analysis

To analyse the facts in question, we will make use of the O(ptimal) P(aradigms) framework as set forth by McCarthy (2005). In this model, faithfulness constraints evaluate candidates that consist of entire inflectional paradigms (OP-Faith), where an inflectional paradigm contains all and only the words based on a single lexeme (2005: 174). Roughly speaking, when an OP constraint dominates the relevant Input-Output (IO)-constraint, then paradigm uniformity emerges at the cost of faithfulness to the input. This is the situation here. As a first step in the analysis, we argue that Greek generally allows hiatus4 due to the ranking in (6), since faithfulness to the underlying specification of [±vocalic] is more important than the avoidance of sequences of vowels. Note that we follow Nevins and Chitoran (2008) in that the feature that distinguishes vowels from glides is [±voc], thus vowels are [+voc], glides are [-voc].

(6) Ident-IO[±voc] >> *VV

Second, glide formation may occur as a response to the constraint in (7) which we claim is active in Greek nominal paradigms, thus yielding the ranking in (8), according to which the form [ðomáti-a] surfaces intact, permitting the word final hiatus, due to the Ident-IO[±voc] constraint, whereas the form /mati+a/ surfaces as [mátç-a], thus violating Ident-IO[±voc] to satisfy the higher ranking OP-Faith-s#5.

(7) OP-FAITH-s#: Words in a paradigm have an identical number of syllables (adapted from Bat-El 2008)

4 Here we refer to word-internal hiatus. For word-external hiatus see Baltazani (2006). 5 Obviously, OP-Faith-s# should not be too high-ranked; if that were the case, then we would have no explanation for the existence of imparisyllabic nouns like máθima – maθímata “lesson-s”. Instead it must be that Dep and Max dominate OP-Faith-s# banning insertion to or deletion from input material, but having no say to other faithfulness violations. In particular, changes to [±voc], regulated by the low-ranked Ident-IO[±voc] are permitted as a way to satisfy paradigmatic faithfulness.

1 8 0 Phonology - Phonetics Nina Topintzi and Mary Baltazani

(8) Suggested ranking for [i]~[J] alternation in Greek morphophonology OP-Faith-s# >> IDENT-IO[±voc] >> *VV

To ensure that our suggested ranking in (8) is correct and assuming a Rich Base (cf. Kager 1999; Prince and Smolensky 1993/2004), our analysis needs to examine all combinations of stems with underlying i/J plus a null or –o suffix as shown in (9). To exemplify, we use the hypothetical stem /koti/.

(9) Nom.Sg. Nom.Pl. (I) /koti+ø/ /koti+a/ (II) /koti+o/ /koti+a/ (III) /kotJ+ø/ /kotJ+a/ (IV) /kotJ+o/ /kotJ+a/

The empirical facts suggest that of those, only patterns (II) and (IV) surface faithfully (cf. 4b.i and 4b.ii, respectively). Patterns (I) and (III) converge to the surface pattern […i+ø] – […J+a], i.e. the one that presents alternation (4a). This state of affairs is correctly generated by the ranking in (8) and exemplified for each of the patterns below. We start by looking at the patterns without any alternation6.

(10) Pattern II: [i-i] – no alternation

/sçéði+o/NOM.SG/sçéði+a/NOM.PL. OP-Faith-s# Ident-IO[±voc] *VV sçé.ði.o ☞ a. ** sçé.ði.a sçé.ði.o b. *! * * sçé.ðʝa sçé.ðʝo c. *! * * sçé.ði.a sçé.ðʝo d. *!* sçé.ðʝa

Candidates (10b) and (10c) are ruled out, because by exhibiting variation in the number of syllables across the paradigm, they cause a fatal violation of OP-Faith. (10d) is eliminated since it turns the input [+voc] /i/ to [J], thus (10a) – with hiatus – is the winner. Note that comparison of (10a) with (10d) gives us

the ranking argument for Ident-IO[±voc] >> *VV. Similarly, an underlying glide is maintained throughout as shown in (11).

6 Note that only the nom.sg. and nom.pL. are included in the tableaux, although OP- Faith of course evaluates the whole paradigm. This is done for reasons of space and convenience. The results are in either case the same (only with extra violations in the case of the whole paradigm).

Phonology - Phonetics 1 8 1 Major Trends in Theoretical and Applied Linguistics

(11) Pattern IV: [J-J] – no alternation7

/psónJ+o/nom.sg /psónJ+a/nom.Pl. OP-Faith-s# Ident-IO[±voc] psó.ɲo ☞ a. psó.ɲa psó.ɲo b. *! * psó.ni.a psó.ni.o c. *! * psó.ɲa psó.ni.o d. *!* psó.ni.a

Patterns I&III do not surface faithfully. Instead, neutralization of the /i/-/J/ contrast occurs so that paradigmatic faithfulness is satisfied, even at the expense of featural faithfulness (compare 12a with 12d and 13b with 13d). Thus, the way an underlying /i/ or /J/ surfaces (i.e. whether Ident-IO[±voc] gets to be violated or not), depends on whether by doing so, the number of syllables across the paradigm can be kept invariant (high-ranking OP-Faith-s#). As shown in the tableaux, the only way to do that is by predictably alternating between the two vocoids. In the NOM.SG – and also in Accusative and Vocative Singular - an [i] emerges, otherwise a glide surfaces.

(12) Pattern I: [i-J] alternation

/peðí+ø/nom.sg /peðí+a/nom.Pl. OP-Faith-s# Ident-IO[±voc] pe.ðí a. *! pe.ði.á péðʝ b. *! ** pe.ðʝá péðʝ c. *!* * pe.ði.á pe.ðí ☞ d. * pe.ðʝá

(13) Pattern III: [i-J] alternation

/peðʝ+ø/nom.sg /peðʝ+a/nom.Pl. OP-Faith-s# Ident-IO[±voc] pe.ðí a. *! ** pe.ði.á péðʝ b. *! pe.ðʝá péðʝ c. *!* * pe.ði.á pe.ðí ☞ d. * pe.ðʝá

7 A full-fledged analysis of the palatals that also relates them to the glide, follows in §3.

1 8 2 Phonology - Phonetics Nina Topintzi and Mary Baltazani

To sum up thus far, the proposed analysis that incorporates both /i/ and /J/ phonemes correctly produces the empirical facts distinguishing among three surface patterns: • hiatus throughout (10) • glide throughout (11) • i-J alternation (12, 13) Which of the patterns will arise is predictable based on two factors, namely, (a) whether the UR contains /i/ or /J/ and (b) the declension class, that is, whether there are zero affixed forms, like in (12, 13) or not, like in (10, 11).

3. Palatals

The section above discussed cases where the glide, surfacing as a palatal (c, ɟ, ʝ, ç, ʎ, ɲ) segment, alternates with the high front vowel in some morpho-phonological environment. Palatals however arise in various contexts8, making their distribution a challenge to understand. We will argue that a generalization about the distribution of the palatals is nonetheless state-able, but its specifics rely on whether the segment in question is an obstruent or a sonorant (nasal, lateral). Anticipating the discussion that follows, we claim that palatal obstruents are either the product of simple or extreme palatalization, whereas palatal sonorants of extreme palatalization only9. We use the abbreviation VNL below to indicate the undergoers of extreme palatalization, i.e. velar obstruents, coronal nasal and lateral. As a vital step towards this goal, consider the instances of obstruent palatals found elsewhere in the language in (14), which at first glance offer conflicting evidence.

(14) a. Palatals as allophones of velars ceɾí *keɾi “candle” kalá “good” cípos *kipos “garden” kúpa “cup” b. Palatals contrast with velars [káli] - [cáli] “beauty” - “binocular” [γatí] - [ʝatí] “cat” - “why” [xóni] - [çóni] “shove”- “snow”

8 Notably, beyond some phonetic studies (Nicolaidis 2003, Arvaniti 2007), palatals have received insufficient phonological/phonetic attention. See however Baltazani & Topintzi (2012). 9 This state of affairs holds for Standard Modern Greek. In other dialects, like some Peloponnesian ones, palatal sonorants are also subject to simple palatalization (see §4 for details).

Phonology - Phonetics 1 8 3 Major Trends in Theoretical and Applied Linguistics

(14a) presents the typical type of palatalization (occasionally abbreviated as ‘pal.’) where velars and palatals appear in complementary distribution, so that velars appear in front of back vowels and palatals before front vowels. We will call this process simple palatalization. The minimal pairs in (14b) though suggest that palatals and velars may contrast with each other. So should palatals be considered allophones of velars (as suggested by (14a)) or separate phonemes themselves (as suggested by (14b))? To add to an already complex distribution, palatals also arise in the morpho-phonological environment of the type discussed in section 2.2, indicated in (15), where the singular patterns with (14a) above, while the plural patterns with (14b).

(15) Palatals in morpho-phonological environments (cf. (4))10 Sing. Pl. lú.ci lú.ca “gutter” pu.ɟí pu.ɟá “reticule” níçi níça “nail”

In what follows, we give a unified and simplifying account of the baffling facts in (14, 15) and argue that palatals are always derived sounds – with the exception of the approximant segment /j/ itself of course – and originate in either underlyingly velar consonants (simple palatalization) or in sequences of VNLs + J (what we term extreme palatalization, following Bateman 2007).

3.1. Velars vs. Palatals in Non-derived Environments

The contrast in (16) is the standard textbook example of allophony in Greek involving simple palatalization.

(16) Velar-palatal allophony: The simple case Velars + back Vs Palatals + Front Vs kalós ‘good’ címa ‘wave’ kóta ‘chicken’ ceɾí ‘candle’ kupí ‘oar’

In OT-terms, this allophony can be understood as a conflict between the informally-labeled-constraint *ki (Sequences of velars followed by front vocoids

10 Below, we fine-tune our IPA transcription for these forms to match as accurately as possible the phonetic facts we will lay out next.

1 8 4 Phonology - Phonetics Nina Topintzi and Mary Baltazani

are banned) and the faithfulness constraint Ident-IO [Place] which aims to preserve underlying Place specifications. In order to get palatalization, the former constraint must dominate the latter.

(17) Simple pal.: /kípos/ → [cípos] “garden” kípos *ki Ident-io-Place

☞ a. cípos *

b. kípos *!

In sequences of velars followed by back vowels, *ki is irrelevant, thus a violation of Ident-io-Place through palatalization is unwarranted. As a result, the winning candidate remains faithful to the input.

(18) Nothing happens, since no trigger for pal.: /kalós/ → [kalós] “good” kalós *ki Ident-io-Place

☞ a. kalós

b. calós *!

While this allophony accounts for a significant amount of data, it leaves a sizeable portion of data unexplained, since the source of the palatal remains famously elusive.

(19) Palatals contrast with velars [káli] - [cáli] “beauty” - “binocular” [γatí] - [ʝatí] “cat” - “why” [xóni] - [çóni] “shove”- “snow”

We agree with previous accounts that palatals here too are derived; unlike other approaches however (e.g. Arvaniti 1999: 169 who posits /i/ as the trigger of palatalization, since she accepts no phonemic glide /J/), we crucially propose that these are the result of the merge of underlying /Velar + J/ into a single palatal sound11, as suggested by the phonetic results of Baltazani & Topintzi (2012), discussed in some detail in the next section. This suggests that in (19), the posited underlying glide gets to be absorbed by the preceding consonant, a situation termed extreme palatalization (after Bateman

11 In an articulatory analysis, Nicolaidis (2003) describes these segments as palatalized, rather than palatals.

Phonology - Phonetics 1 8 5 Major Trends in Theoretical and Applied Linguistics

2007 and references therein). Triggers may delete because “the information contained in the trigger can be recovered from the palatalization on the consonant target” (Bateman 2007: 82-83). Unlike vowels, /i/ and /e/ in this case, whose deletion after palatalization would be highly costly, since they carry along nucleic information, glides are free from this restraint, as they fulfil a non-nucleic role. Consequently, deletion of a glide trigger is typical cross-linguistically. In terms of analysis, we interpret this effect by adding a few constraints to our original palatalization-triggering-ranking *ki >> Ident-Place. The tableau illustrates. So as to avoid burdening it unnecessarily, we only present candidates that have undergone palatalization.

(20) Extreme palatalization by means of fusion

/k1J2áli/ → [c1,2áli] “binocular”

/k1J2áli/ OCP-Pal Ident-IO [±voc] Uniformity a. cjá.li *! b. ci.á.li *! ☞ c. c1,2á.li *

Through the interplay of these constraints, the underlying velar-glide sequence surfaces as a palatal consonant which merges the features of both segments, as indicated by the subscripts 1, 2 of the winner (20c). In particular, it preserves the place of articulation of the glide (palatality) and the [+consonantal] feature of the velar at the cost of the low-ranked anti-coalescence UNIFORMITY constraint. To rule out a candidate like (20a), where some correspondent of the glide is still present in the structure, we utilize OCP-Pal (which bans sequences of palatals12). Changing the glide into [i] (20b) produces a superfluous violation of Ident-IO [±voc] due to a featural value change from [-voc] to [+voc]. Other candidates like

[kJa.li], [ki.a.li] are eliminated due to *ki, whereas deletion of the glide, as in [c1ali] violates MAX. We assume that both are high-ranking constraints and thus not considered here.

3.2. Palatals in Derived Environments

We have thus far accounted for the cases of simple allophony and superficial contrast between velars and palatals in (14). We still need to offer an analysis for

12 But crucially not of palatal+i/e. This can be captured by either utilizing somewhat different place features for vowels and consonants, or by modifying OCP-PAL to something like a conjunction constraint OCP-PAL^ IDENT[-voc] that only gets to be violated by non- vocalic segments of identical palatality.

1 8 6 Phonology - Phonetics Nina Topintzi and Mary Baltazani

the data in (15) where the morpho-phonological concerns considered in §2.2. are also in action. Before presenting that, we need to discuss an important issue which has arisen through the preliminary phonetic analysis of the same data by Baltazani & Topintzi (2012). This paper suggests that while fusion is indeed what occurs in the palatalized data of (14b/19), at morpheme boundaries (as in (15)), less amount of absorption is observed. More concretely, phonetic differences were detected between extreme palatalization in derived and underived environments in the realization of the transition from the palatal consonant to the following vowel: the transition was shorter in underived environments of extreme palatalization (as in cáto) than at morphological boundaries (as in lúca13). These results were interpreted as evidence for fuller absorption of triggers by palatal segments in underived contexts than at morphological boundaries. In short, the palatalization trigger is absorbed by the target in varying degrees in extreme palatalization – more within a morpheme (cáto), less at morpheme boundaries (luc+a). We identify this difference phonologically by representing

extreme palatalization in underived environments as fusion, i.e. /k1J2ato/ → [c1,2ato] and in derived environments as some preservation of the onglide transition through a superscript, i.e. [lucja14]. Note that we use this transcription to differentiate the latter from a form like [lucja], where the transition is presumably interpreted as a segment itself, since the phonetic facts provide no evidence for such form. Proceeding with the analysis, we first demonstrate in (21) that the ranking devised in (20) is unable to capture the fine-grained phonetic details, just outlined. A slight modification is thus in order.

(21) Extreme pal. in morpho-phonological environments: Wrong winner15! Ident- /lúk i +Ø/ /lúk i +a/ OP-Faith-s# Uniformity *VV 3 4 nom.sg 3 4 nom.Pl. io[±voc] lú.c i  a. 3 4 * lú.c3,4a

lú.c3i4 ☞ b. j4 *! lú.c3 a lú.c i c. 3 4 *! * lú.c3i4.a

13 In the experimental material all palatal+vowel sequences were unstressed to avoid the confounding effect of stress on duration and formants. We thank an anonymous reviewer for bringing this matter up. 14 The target words for the comparison between derived and underived environments were selected so as not to differ in stress condition or in the vowel following the palatal consonant. We thank an anonymous reviewer for raising this question. 15 A candidate like [luci, lucːa] is conceivable, but unsupported by the phonetic results of Baltazani & Topintzi (2012), thus it is not considered.

Phonology - Phonetics 1 8 7 Major Trends in Theoretical and Applied Linguistics

(21) illustrates application of the crucial constraints and ranking for the specific forms. The fused candidate (21a) is incorrectly predicted instead of the actual winner with partial retention of the glide, cf. (21b). We propose that the distinguishing factor here is the presence of morphological boundaries. In the forms with fusion, the glide was root internal; in the forms with glide retention, the segment in question is found at the right edge of a root. We consequently propose that the constraint in (22) is in action:

16,17 (22) Max-io[±voc]Rt]: Do not delete the [voc] feature at the right edge of a root

This constraint penalizes the loss of the [voc] feature at the right edge of a root, thus eliminating the candidate with fusion (24a) since the [voc] feature originating in the segment with the index 4 is no longer preserved. Candidate (24d) is easily ruled out due to a violation of quite high-ranking OP-Faith-s#. One new complication arises though; the winning candidate (24b) presents an onglide transition, which however does not itself constitute a real segment as (24c) implies. How can we distinguish between the two and render (24b) as the winner? We claim that only (24c) violates OCP-PAL as it involves a sequence of two palatal segments; (24b) avoids such violation since the glide is not a separate segment, rather a mere transition. Instead it violates the constraint in (23), which must be below OCP-PAL and, as verified in (26), also above UNIFORMITY.

(23) *Secondary Articulation (*SECARTIC): No secondary articulation (cf. Rosenthall 1997)18

16 Parallels to root- as well as rightmost-faithfulness can be found elsewhere too. See for instance Jurgec (2011; §9.4-5) for a brief summary. He discusses both, although with a different focus, that of harmony. 17 Marc van Oostendorp (p.c. to first author) suggests the use of a constraint that penalizes

deletion of root-final segments, e.g. Max-SegRoot] . This would not work, however. Under the assumption that in fusion, there is still a correspondent for the vocoid, then both [lú.c3j4a] and [lú.c3,4a] would equally satisfy Max-Seg Root] . Given the low ranking of UNIFORMITY, fusion would thus be incorrectly chosen. What we consequently need is a constraint that specifically targets the [±voc] feature and asks that it is preserved on the output. 18 We don’t literally mean that the superscripted glide here is a secondary articulation. After all, it is unclear what a palatal secondary articulation to a palatal segment would mean! What we aim to do here instead is to indicate a partially absorbed glide (24b) and distinguish it from both a true segment [j] as in (24c) and a fully absorbed one as in (24a). *SECARTIC is also conceptually close to what the violation seems to be in (24b), thus we decided to use it – as it has already been proposed in the literature – rather than suggest a new, more accurate, constraint.

1 8 8 Phonology - Phonetics Nina Topintzi and Mary Baltazani

(24) Extreme pal. in morpho-phonological environments with input /i/ /luki+Ø, luki+a/ → [luci, lucja]

/lúk3i4+Ø/nom.sg OP- OCP- *Sec Ident- Max-[±voc]Rt] Uniform /lúk3i4+a/nom.Pl. Faith-s# PAL Artic io[±voc] lú.c i a. 3 4 *! * lú.c3,4a

lú.c3i4 ☞ b. j4 * * lú.c3 a lú.c i c. 3 4 *! * lú.c3j4a lú.c i d. 3 4 *! lú.c3i4.a

In a similar fashion, the same winner surfaces even when an underlying glide exists root-finally.19

(25) Extreme pal. in morpho-phonological environments with input /j/ /lukj+Ø, lukj+a/ → [luci, lucja]

/lúk3j4+Ø/nom.sg OP- OCP- *Sec Ident- Max-[±voc]Rt] Uniform /lúk3j4+a/nom.Pl. Faith-s# PAL Artic io[±voc] lú.c i a. 3 4 *! * * lú.c3,4a

lú.c3i4 ☞ b. j4 * * lú.c3 a lú.c i c. 3 4 *! * lú.c3j4a lúc j d. 3 4 *! lú.c3j4a

Importantly, since Max-[±voc]Rt] only evaluates the right edges of roots, it has no say in the evaluation of candidates in words of the type /kJali+Ø/, where the glide appears root-internally. Uniformity – being the lowest-ranked constraint – is now being given the chance to favour a candidate with coalescence,

namely [c1,2ali] (26d). Rival candidates (26a-c) are eliminated by the constraints dominating Uniformity. Tableau (26), an updated version of (20) with all the constraints at play, demonstrates this effect.

19 See the discussion related to (12) and (13) for the reasoning behind checking both these inputs.

Phonology - Phonetics 1 8 9 Major Trends in Theoretical and Applied Linguistics

(26) Extreme palatalization by means of fusion [updated version]

/k1J2áli/ → [c1,2áli] “binocular”

OP- OCP- *Sec Ident- /kJ áli/ Max-[±voc] Uniform 1 2 Rt] Faith-s# PAL Artic io[±voc]

a. c1i2.á.li *!

j2 b. c1 á.li *!

c. c1j2á.li *! ☞ d. c1,2á.li *

4. Discussion: Simple vs. Extreme Palatalization in Relation to Input /i/ and /j/

Recall that according to our analysis, both /i/ and /J/ are phonemes in Greek, but their contrast neutralizes under certain morpho-phonological circumstances. We have also suggested that the contrast between /i/ and /J/ helps us understand the patterns of palatalization arising in the language. Briefly we differentiate between two types of palatalization, dubbed simple and extreme, as summarised below.

(27) I. Simple palatalization a. affects velar obstruents before the triggering front vowels /i, e/→ the velar palatalizes, without any change of the pal. trigger. b. affects velar obstruents before /i, J/ root-finally→ the velar palatalizes and the vocoid palatalization trigger emerges as [i] due to phonotactics and paradigmatic faithfulness (cf. data for the singular in (15)) II. Extreme palatalization20 a. affects velar obstruentsor nasal/lateral coronals (VNL) before /J/ root- internally. Any vowel can follow21 → VNL palatalizes and the pal.trigger /J/ gets to be absorbed by it. Result: palatalization with fusion of the trigger. b. affects VNL consonants before /i, J/ root-finally. Any vowel can follow → VNL palatalizes and the vocoid palatalization trigger leaves a trace in the output by means of an onglide transition in the following vowel.

20 Some more specific discussion for the palatalization of the nasal and palatal coronals follows later in this section. 21 Although /a, o, u/ are the most common segments to follow, /i, e/ can follow as well, in words like [ʝe], [ʝi], the singular vocative and plural nominative forms of [ʝos] “son”, which are all arguably formed through the same process of extreme palatalization.

1 9 0 Phonology - Phonetics Nina Topintzi and Mary Baltazani

To put it in different words, simple palatalization occurs when /i, e/ follow (27Ia) and extreme palatalization when an input /J/ follows (27IIa)22. The distinction is understandably obscured in the morpho-phonological paradigms of the type in (4a) and (15). As we have seen before, the phonemic distinction between /i/ and /J/ is predictably neutralized there, thus producing certain patterns of palatalization when the latter is possible to occur (i.e. with velars and coronal sonorants). Thus, no matter what the input vocoid is, simple palatalization occurs when an [i] emerges (i.e. in the Singular with the exception of the Genitive, cf. (27Ib)) and extreme palatalization when a glide normally follows (i.e. in the Gen. Sg. and the Plural, cf. (27IIb)). Both types of palatalization thus arise in the same paradigm in a predictable manner. These processes can also be visualised by means of a palatalization scale (28), based on the phonetic results of Baltazani & Topintzi (2012), according to the surface realization of the palatalization trigger.

(28) Palatalization scale in Greek based on the amount of absorption of the palatalization trigger by the palatalization target

(27I): no absorption (27IIa): full absor- of trigger ption, i.e. merge (27IIb): partial trigger absorption, i.e. onglide transition

There is additional empirical evidence in support of the proposed analysis and the distinction between simple and extreme palatalization. First, the existence of words like [cianós] *canós “blue” or [cíonas] *cónas “column” – where no morphological boundary exists at the locus of palatalization – alongside words of the [cáli] type indicate that two types of palatalization are in order here. We claim that what differentiates the two patterns is the contrast between /i/ vs. /J/. Indeed, our analysis correctly generates words of the type of [cianós] too.

22 The difference in behaviour between /i/ and /J/ in palatalization is unsurprising. The sounds are acoustically different and present different degrees of constriction. Numerous languages treat them asymmetrically (see Bateman 2007, esp. Chpt 5, for details).

Phonology - Phonetics 1 9 1 Major Trends in Theoretical and Applied Linguistics

(29) Simple pal. and hiatus

/k1i2anos/ OCP-Pal *Sec Artic Ident-IO [±voc] Uniformity *VV

☞ a. c1i2anós *

j2 b. c1 anós *! *

c. c1j2anós *! *

d. c1,2anós *!

A second argument comes from the sonorants /l/ and /n/. These consonants do not palatalize before front vowels in SMG, e.g. [limáni] “harbour”, even though there exist many words with the palatalized counterparts [ʎ, ɲ] both within the same morpheme (30a) and across different ones (30b).

(30) a. Palatalized sonorants in phonologically simple environments23 ʎóno “I melt” ʎanízo “chop” ɲóθo “I feel” ɲáta “youth” b. Palatalized sonorants across morphological boundaries xalí “rug” xaʎá “rugs” pepóni “melon” pepóɲa “melons”

In other dialects, palatalization is more widespread. For instance, in addition to (30), speakers in NW Peloponnese produce [ʎimáɲi] for SMG [limáni] “harbour”. One way to understand this is that coronal sonorants in SMG only undergo extreme palatalization, whereas in other dialects both types of palatalization are in effect. Formally, the distinction is easily captured if /i/ causes simple palatalization with /J/ being responsible for extreme palatalization. This last point is relevant for another issue too; recall that we classified the palatalization of the type in (27IIb) as extreme too. However, one could dispute this characterization, because, unlike the cases in (27Ia), a remnant of the palatalizing trigger still remains, thus possibly rendering itself actually an instance of simple palatalization. There are various reasons we think this is not the case. First, the data of coronal sonorants above in SMG clearly group together the environments in (30a and b), as acoustic evidence in Baltazani & Topintzi (2012) also shows, in a single palatalization pattern to the exclusion of the environment Cor.Son. + i,e that could also potentially undergo palatalization (as it does in

23 For reasons of space, the formal analysis of the coronal sonorants is omitted here, but it would be a mere extension of the present account.

1 9 2 Phonology - Phonetics Nina Topintzi and Mary Baltazani

some dialects) but fails to do so in SMG. If the two environments in (30) were treated differently with regard to palatalization, then it would be pure chance that they pattern together. In contrast, the distinction between extreme vs. simple palatalization, offers a neat generalization of facts: coronal sonorants only undergo extreme palatalization in SMG, but both processes of palatalization in other dialects. Second, extreme palatalization is not uniformly realized across languages anyway. In Luvale and Mandarin for instance, the trigger i is maintained, but j is deleted. In Shilluk, the glide coalesces with the consonant it palatalizes and in Yimas it is optionally maintained or deleted (Bateman 2007: 83 and references therein). What all this suggests is that extreme palatalization allows for variable realization, a fact that Greek seems to exploit. In fact it seems to do so in an interesting way. Recall that a remnant of the trigger remains at morphological boundaries, but is fully absorbed morpheme-internally (or to be precise, root- internally). Perhaps, the presence of the glide enhances the presence of a boundary there, signifying the contrast between edges and internal content of a morpheme. For these reasons, we conclude that the palatalization process occurring across morphological boundaries and induced by OP-Faith-s# is an instance of extreme palatalization too.

5. Conclusion

In this paper we have argued in favour of both a phonemic contrast between /i/ and the glide /J/, as well as of their contextual neutralization. We have illustrated the latter by exploring the morphophonology of the nominal paradigms of the large class of nouns ending in –i and –io. We have claimed that while vowel hiatus is generally admitted, as in [peðío] - [peðíu] “field”, paradigm uniformity – requiring that the number of the syllables of the noun remains constant across the paradigm – may cause its resolution, e.g. [peðí] - [peðʝú] *[peðiú] “child”. An extension of our account has been the analysis of palatals. Like earlier descriptions, we agree that palatals are derived sounds, but unlike them, we suggest that palatals are the product of two different palatalization processes, coined simple and extreme palatalization. Simple palatalization always fully retains the palatalizing front trigger vowel (i or e), whereas extreme palatalization either fully absorbs the trigger glide into the newly palatalized preceding consonant (as in cáli ‘binocular’) or retains traces of the underlying vocoid by means of an onglide transition at morpheme boundaries (as in lúca). The claims and analysis above not only reconcile the superficially conflicting nature of palatals as both allophones and as distinct phonemes, but also meld neatly with the analysis of the Greek /i/ - /J/ contrast and neutralization.

Phonology - Phonetics 1 9 3 Major Trends in Theoretical and Applied Linguistics

References

Arvaniti, A. 1999. Illustrations of the IPA: Modern Greek. Journal of the International Phonetic Association 19: 167-172.

Arvaniti, A. 2007. Greek Phonetics: The State of the Art. Journal of Greek Linguistics 8, 97-208.

Baltazani, M. 2006. Focusing, Prosodic Phrasing, and Hiatus Resolution in Greek. In Laboratory Phonology 8 ed. by L. Goldstein, D. Whalen and C. Best, 473-494. Berlin: Mouton de Gruyter.

Baltazani, M. and N. Topintzi. (to appear). On some phonetic and phonological properties of the Greek glide. In Z.Gavriilidou, A. Efthymiou, E. Thomadaki, P. Kambakis-Vougiouklis (eds.), Selected papers of the 10th International Conference in Greek Linguistics, pp. 153-166. Komotini/Greece: Democritus University of Thrace.

Bat-El, O. 2008. Morphologically Conditioned V-∅ alternation in Hebrew: Distinction among Nouns, Adjectives and Participles, and Verbs. In Current Issues in Generative Hebrew Linguistics, S. Armon-Lotem, G. Danon, and S. Rothstein (eds), John Benjamins, Amsterdam & Philadelphia, 27-59.

Bateman, N. 2007. A Crosslinguistic Investigation of Palatalization. Doctoral Dissertation. University of California, San Diego.

Bhat, D.N.S. 1978. A general study of palatalization. In Universals of language, Vol. 2: Phonology, J.H. Greenberg (ed.), Stanford University Press, Stanford, 47-91.

Deligiorgis, I. 1987. On Rules of Syllabification and Syllabic Restructuring. CLS 23, 76-90.

1 9 4 Phonology - Phonetics Nina Topintzi and Mary Baltazani

Gilley, L.G. 1992. An Autosegmental Approach to Shilluk Phonology. A publication of Summer Institute of Linguistics and UT at Arlington.

Holton, D., P. Mackridge and I. Philippaki-Warburton. 1997. A Comprehensive Grammar of the Modern Greek Language. Routledge, London.

Householder, F.W. 1964. Three Dreams of Modern Greek Phonology. In Papers in Memory of George C. Papageotes (issued as Supplement to Word 20:3) ed. by R. Austerlitz, 17-27.

Jurgec, P. 2011. Feature Spreading 2.0: A Unified Theory of Assimilation. Doctoral Thesis, University of Tromsø. Available at: http://ling.auf.net/lingBuzz/001281

Kager, R. 1999. Optimality Theory. Cambridge: Cambridge University Press.

Kazazis, K. 1968. Sunday Greek. CLS 4, Department of Linguistics, University of Chicago, 130-140.

Koutsoudas, A. 1962. Verb Morphology of Modern Greek. Bloomington: Mouton.

Levi, S. 2008. Phonemic vs. derived glides. Lingua 118: 1956–1978.

Malavakis, T.I. 1984. Φωνηεντικές Συνέχειες: Διφθογγοποίηση, Ουρανικοποίηση και Φωνηματική Κατάταξή τους [Vowel Sequences: Diphthongization, Palatalization and Phonemic Classification]. Studies in Greek Linguistics 4: 1-16.

Malikouti-Drachman, A. and G. Drachman. 1990. Φωνολογική κυβέρνηση και προβολή: Αφομοιώσεις, ανομοιώσεις’ [Phonological government and projection: assimilations, dissimilations]. Working Papers in Greek Grammar, University of Salzburg, 1-20.

McCarthy, J.J. 2005. Optimal paradigms. In L. Downing, T.A. Hall and R. Raffelsiefen (eds.), Paradigms in Phonological Theory. Oxford: Oxford University Press, 170-210. Available at: http://works.bepress.com/john_j_mccarthy/48

Mirambel, A. 1959. La Langue Grecque Moderne: Description et Analyse. Librairie Klincksieck, Paris.

Nevins, A. and I. Chitoran. 2008. Phonological representations and the variable patterning of glides. Lingua 118: 1979-1997.

Nicolaidis, K. 2003. Μια ηλεκτροπαλατογραφική μελέτη των ουρανικών συμφώνων της Ελληνικής (An electropalatographic study of palatals in Greek). In

Phonology - Phonetics 1 9 5 Major Trends in Theoretical and Applied Linguistics

Theophanopoulou-Kontou, D., C. Lascaratou, M. Sifianou, M. Georgiafentis and V. Spyropoulos, Σύγχρονες Τάσεις στην Ελληνική Γλωσσολογία (Current Trends in Greek Linguistics). Athens: Patakis, 108-127.

Nikolopoulos, G. 1985. Επανεξέταση του τόνου και της συνίζησης στα Νέα Ελληνικά [Re-examination of stress and diphthongization in Modern Greek]. Studies in Greek Linguistics 5: 21-40.

Nyman, M. 1981. Paradigms and transderivational constraints: Stress and yod in Modern Greek. Journal of Linguistics 17: 231-246.

Prince, A. and P. Smolensky. 1993/2004. Optimality Theory: Constraint Interaction in Generative Grammar. Malden, MA, and Oxford, UK: Blackwell.

Rosenthall, S. 1997. The distribution of prevocalic vowels. Natural Language and Linguistic Theory 15(1): 139-180.

Rytting, A.C. 2005. An iota of difference: Attitudes to yod in lexical and social contexts. Journal of Greek Linguistics 6: 151-185.

Setatos, M. 1974. Φωνολογία της Κοινής Νεοελληνικής [Phonology of Modern Greek Koine]. Papazisis, Athens.

Warburton, I. 1976. On the boundaries of morphology and phonology: A case study from Modern Greek. Journal of Linguistics 12: 258-278.

1 9 6 Phonology - Phonetics The Production of English Aspirated Stops in Foreign Language Acquisition

Eleni Tsiartsioni

Aristotle University of Thessaloniki [email protected]

Abstract

This study examines the effect of teaching on the production of L2 English aspirated stops among Greek EFL (English as a Foreign Language) learners. Two groups of speakers were recorded, an experimental group who received the pronunciation teaching intervention and a control group who did not receive special pronunciation instruction. The data were analysed acoustically and a native-likeness criterion (Birdsong 2007, Flege et al. 1992) was used to investigate individual variability. Overall, this study reports a positive effect of pronunciation instruction on the production of aspirated stops, however the great individual variability suggests that systematic target-like production of stops constitutes a challenging goal. A tentative effect of learners’ gender is also obtained, as the female speakers produce greater VOT than the male ones after teaching.

1. Introduction

This paper is part of a larger research project on the acquisition of features of the English speech rhythm and stop voicing system among Greek learners of English. In particular, the present study examines the production of English word-initial aspirated stops by Greek students of ten to sixteen years old who learn English in a foreign language (FL) / formal setting context. One of the differences between the Greek and English phonetic systems lies in the production of /p, t, k/ sounds. In English these sounds are produced as voiceless aspirated with long voice onset time (VOT, Lisker & Abramson 1964), whereas in Greek they are produced as voiceless unaspirated with relatively short VOT (Arvaniti 1987, 2001, Botinis et al 2000, Fourakis 1986, Nicolaidis 2002, Panagopoulos 1972). Given these differences, an interesting question is how these segments are realised by Greek learners of English and whether pronunciation instruction can have an effect on speakers’ production.

Phonology - Phonetics 1 9 7 Major Trends in Theoretical and Applied Linguistics

Previous research on the production of L2 aspirated stops by speakers whose L1 has unaspirated stops (for example, speakers of L1 Dutch, Italian, French and Portuguese) showed that the context of acquisition appears to be related to the degree of accuracy in L2 production. Studies which were conducted in a second language / naturalistic context of language acquisition (namely, in an environment where the learners live in a target-language community) showed that the L2 speakers generally produce longer VOT in L2 than in their L1, however, the L2 VOT is shorter than that of the target language (Caramazza et al. 1973, Flege & Eefting 1987, Major 1987, Williams 1979). Studies which were conducted in foreign contexts of language acquisition (i.e. in environments where the learners are not exposed to the target language in their everyday communication) reported that the L2 learners had great difficulty producing target-like aspirated stops, as they tended to transfer L1 voiceless unaspirated stops into L2 (Riney & Takagi 1999, Poedjianto 2003). However, exceptional cases of speakers who produced target-like VOT have been reported in foreign language acquisition studies. Riney and Takagi (1999), in particular, investigated the correlation between Global Foreign Accent (GFA, determined on the basis of native English listeners’ scores) and VOT for /p, t, k/ among L1 Japanese learners of English as a foreign language. The results showed that, generally, the L2 English VOT was very similar to that of L1 Japanese. A significant correlation was also found between GFA and VOT scores; in particular, more target-like GFA scores were associated with more target-like VOVT values. The individual speakers’ analysis showed that two L2 speakers produced VOT that was similar to L1 English. Riney and Takagi (1999) concluded that VOT is one aspect of pronunciation that some L2 speakers over the age of 12 can produce accurately. Approximation to target-like VOT in an FL context was reported by Lowie (2009) and Lowie and Reitsma (2009) regarding a group of Dutch children starting English at the age of 4 and being taught by a native English teacher. That study concluded that in FL contexts approximation to L2 VOT values can be achieved provided that early starting age and native speaker input are ensured. Another key issue in the present study is the effect of pronunciation instruction. The previous literature on pronunciation teaching shows that this is an area that has not been adequately explored in applied linguistics (Goodwin et al. 1994, Pardo 2004). It is, therefore, not surprising that teachers have reported being unsure about the importance and effectiveness of teaching pronunciation to their students (Pardo 2004). A summative research of twenty-five studies on the effect of pronunciation instruction, however, showed that there is a positive effect of well-organised pronunciation teaching, as long as carefully planned techniques are used (Pardo 2004). The positive effect of pronunciation teaching has been documented both in naturalistic settings (as in the vast majority of studies reviewed by Pardo 2004) and in foreign language contexts of language

1 9 8 Phonology - Phonetics Eleni Tsiartsioni

acquisition, as well (Ekstrand 1982, Olson & Samuels 1982, Thogmartin 1982). With regard to the learning of L2 aspirated stops, Reis and Nobre-Oliveira (2008) and Aliaga-García and Mora (2008) reported an increase in L2 VOT after teaching among Brazilian-Portuguese, Spanish and Catalan learners of English respectively, however, the L2 VOT did not reach the target values, even after teaching. Another parameter that has been explored in L2 phonological acquisition research is that of speakers’ gender. With regard to this issue, no conclusive results have been obtained. Whereas certain studies found that female speakers tended to speak with a more target-like accent (Tahta et al. 1981, Thompson 1991), other studies reported no effect of speakers’ gender on L2 pronunciation (Flege & Fletcher 1992, Purcell & Suter 1980, Snow & Hoefnager-Höhle 1977, Suter 1976). In view of the aforementioned literature, the aims of the present study are: 1. to investigate the production of L2 aspirated stops in a foreign / formal context of language acquisition and, in particular, to examine the effect of a specific pronunciation teaching intervention among Greek learners of English, 2. to investigate the potential effect of learners’ gender in the acquisition of English VOT.

2. Methodology

2.1. Subjects

In the present study two groups of subjects were recorded, an experimental group (n=36), which received pronunciation instruction, and a control group (n=36) which followed the regular English classes at school. An equal number of male and female speakers participated in each group, in order to investigate the potential effect of learners’ gender. Each group comprised students of three different ages: twelve ten-year old students, twelve thirteen-year old students and twelve sixteen-year old students. Given the aims of the present paper (section 1), the potential effect of speakers’ age is not addressed in the present paper. However, it is acknowledged that the effect of learners’ age on L2 phonological acquisition is an interesting issue, therefore this will be addressed in future publications. Data for L1 Greek were also obtained from thirty L1 Greek children. L1 English data were obtained from fifteen Greek-American bilingual speakers who were born in the USA and lived in Greece at the time of the experiment. The L1 speakers were of similar ages to the L2 speakers.

Phonology - Phonetics 1 9 9 Major Trends in Theoretical and Applied Linguistics

2.2. Speech Materials / Recording Procedure / Data Analysis

Recordings of students’ speech samples were made twice, once before and once after the teaching intervention (time 1 and time 2 respectively). For L1 and L2 English, the speakers were recorded on the words ‘Pat’, ‘tab’, ‘cap’ embedded in the carrier phrase ‘______is the next word’. For L1 Greek, the words ‘πάσα’ /ˈpasa/ (‘pass’), ‘Τάσα’ /ˈtasa/ (‘Tasa’, a female name), ‘κάσα’ /ˈkasa/ (‘box’), were recorded in the carrier phrase ‘_______ λέω πάλι’, /_____ ˈleo ˈpali/, (‘_____ I say again’). Six repetitions of each word were measured for each speaker. A total of 3402 items were measured for the present study (for L1 Greek: 30 speakers*3 words*6 repetitions=540 items, L1 English 15 speakers*3 words*6 repetitions=270 items; L2 English: 72 speakers*3 words*6 repetitions*2 recordings=2592 items). The recordings were analysed acoustically with the use of waveforms and digital spectrograms generated by the speech analysis software PRAAT (Boersma & Weenink 2007). The VOT duration was measured from the beginning of the release burst to the first positive peak in the periodic portion of the waveforms. For the data of the experimental group at time 2, the ‘native-likeness criterion’ (Birdsong 2007, Flege et al. 1992) was used, based on the plus / minus one standard deviation from the average native English VOT. The percentage of stop tokens that belonged to this range was calculated. Finally, a series of t-tests were conducted in order to investigate the statistical significance of the difference between: (a) L1 Greek and L1 English VOT, (b) L1 and L2 VOT, (c) L2 VOT between time 1 and time 2, (d) the VOT of the experimental and the control group and (e) the VOT of the male and the female speakers. 2.3. Pronunciation Teaching Intervention

The framework of pronunciation teaching which was used in the present study was adopted from Celce-Murcia et al. (1996: 36), who proposed five teaching stages for pronunciation teaching which range from controlled to free activities. These stages include the following: (1) description and analysis of features, (2) listening discrimination activities, (3) controlled practice, (4) guided practice and (5) communicative practice. The students of the experimental group received fifty-one lessons of pronunciation instruction on English stops and speech rhythm. Twenty-two of these lessons were devoted to the teaching of word-initial phonologically voiceless and voiced stops. Each pronunciation lesson lasted 10-15 minutes and it was embedded in the English classes at school. The lessons were conducted by the author, who is a non-native speaker of English and was also the main English teacher of the experimental groups. The demonstration of the phenomenon of aspiration (stage 1) started with the comparison of the pronunciation of Greek and English words. Segmentally

2 0 0 Phonology - Phonetics Eleni Tsiartsioni

similar Greek and English words were chosen (e.g. ‘Pie’-Πάει’ /'paI/), the former produced with aspirated word-initial stops, while the latter produced with unaspirated stops. A multimodal activity was then used with visual and tactile reinforcement, as the students were asked to observe the puff of air associated with the production of English aspirated stops, as they were holding sheets of paper in front of their mouths. During the following stages (stages 2-5) the production of aspirated stops was practised in parallel with the production of devoiced word-initial /b, d, g/. The activities included, among others, listening discrimination of minimal pairs presented both in isolation (e.g. ‘coat’ - ‘goat’) and in context (e.g. ‘He had the peach to himself’ - ‘He had the beach to himself’), reading-aloud dialogues with emphasis on the particular sounds, information-gap activities, role-plays and creation of dialogues. Nursery rhymes (for the 10-year-old students) and limericks (for the 13- and 16-year old students) with a number of occurrences of the particular sounds were also used in class.

3. Results

This section presents the results on the production of word-initial voiceless stops in L1 and L2 production. Figure 1 presents the results for the VOT duration in L1 Greek and L1 English averaged across the speakers of the three ages. The fi gure indicates that, in accordance with the literature, Greek /p, t, k/ are voiceless unaspirated with short VOT values, whereas English /p, t, k/ are produced as voiceless aspirated with long VOT. The diff erence in the VOT of the two languages is statistically signifi cant for all places of articulation (independent-sample t-tests, p<0.001 for all places of articulation).

Figure 1. Mean VOT (ms) and standard deviation in L1 Greek and English.

Phonology - Phonetics 2 0 1 Major Trends in Theoretical and Applied Linguistics

The results on L2 English production reveal that before teaching both the experimental and the control groups produced unaspirated voiceless stops with short VOT (fi gure 2). No signifi cant diff erence is observed between the VOT of the experimental and the control groups at time 1 (independent-sample t-tests, p>0.05). It appears that the speakers generally produced Greek-like VOT at time 1, since the VOT of L2 English at time 1 is signifi cantly diff erent from L1 English (independent-sample t-tests, p<0.001 both for the experimental and the control group) and no diff erent from L1 Greek (independent-sample t-tests, p>0.05 for the experimental and the control group, compare fi gures 1 and 2). The results at time 2 indicate that there is a statistically signifi cant increase in VOT duration for the experimental, but not for the control group (paired- sample t-tests comparing VOT at time 1 and 2, p<0.001 for the experimental group; p>0.05 for the control group). As a result, at time 2 the VOT of the experimental group is diff erent from that of the control group to a signifi cant extent (independent-sample t-tests, p<0.001). The comparison between L1 and L2 production reveals that after teaching the VOT of the experimental group is signifi cantly diff erent from L1 Greek (p<0.05) and no diff erent from L1 English (p>0.05). The VOT values of the control groups at time 2 remain signifi cantly diff erent from L1 English (p<0.05) and no diff erent from L1 Greek (p>0.05), suggesting no change in VOT for the particular group.

Figure 2. Mean VOT (ms) and standard deviation in L2 English in the experimental and control groups before and after teaching.

Figure 2 presents the results for the male and the female speakers before and after teaching, averaged across the three places of articulation. The results on speakers’ gender indicate that before teaching no signifi cant diff erence

2 0 2 Phonology - Phonetics Eleni Tsiartsioni

is observed between the VOT of the male and female speakers, either in the experimental or in the control group (independent-sample t-tests p>0.05 for the experimental and the control groups). However, at time 2 the VOT of the female speakers is signifi cantly greater than that of the male ones for the experimental group (p<0.001). No diff erence is found for the control group with regard to speakers’ gender.

Figure 3. Mean VOT (ms) in L2 English for the male and female speakers before and after teaching.

Figure 2 indicates that considerable standard deviation is prevalent in L2 data, especially with regard to the VOT of the second recording of the experimental group. This fi nding suggests that inter-speaker variability is evident in this group of speakers, since not all speakers performed in a uniform way. In order to explore individual variability in detail, the criterion of ‘native- likeness’ (Birdsong 2007, Flege et al. 1992) was adopted. This criterion was based on the plus / minus one standard deviation from the native English average VOT scores and, according to this, the percentage of stop tokens that belonged to the native-like range was estimated. Table 1 presents the native-like VOT range obtained as a result of the application of the native-likeness criterion.

Table 1. The VOT range for voiceless stops based on the plus / minus one standard deviation from the average L1 English VOT values (in ms).

Native-likeness criterion

/p/ /t/ /k/

47,8-77,2 59,2-88,7 60,5-87,5

Phonology - Phonetics 2 0 3 Major Trends in Theoretical and Applied Linguistics

In the experimental and the control group at time 1 and in the control group at time 2 there are no speakers whose average VOT falls within the native-like range, as they all produce short VOT values, similar to L1 Greek, except for one ten-year-old speaker who produces target-like VOT for /k/ at time 1. After the teaching intervention, the implementation of this criterion led to the identifi cation of four types of voiceless stop production in the experimental group. These four types of stop production are the following: (a) native-like production (according to the native-likeness criterion), (b) longer VOT values than those determined by the native-likeness criterion, (c) intermediate VOT between L1 Greek and L1 English (estimated as the VOT values that fell between L1 Greek average VOT plus 1 standard deviation and L1 English average VOT minus 1 standard deviation) and (d) Greek-like VOT. From these categories the fi rst three can be interpreted to denote an improvement in VOT production, whereas the last one shows no change from L1 Greek VOT. In needs to be noted that type (b) is interpreted to denote improvement, since similarly high VOT scores have been reported in the literature for adult monolingual English speakers (Allen et al. 2002, Deterding & Nolan 2007). Figure 4 presents the percentage of stop tokens that belong to each type of stop production in the experimental group after teaching.

Figure 4. The percentage of stop tokens that belong to each type of voiceless stop production in the experimental group at time 2. The percentages are averaged across all places of articulation and all age groups.

Figure 4 indicates that a greater percentage of stop tokens is intermediate between the VOT of L1 Greek and L1 English than the tokens that belong to each of the other three types of production. Also a considerable percentage of tokens are indicative of longer VOT than that of the L1 English speakers of the

2 0 4 Phonology - Phonetics Eleni Tsiartsioni

present study, whereas approximately one fourth of stop tokens belong to the native-like range. The least frequent type of stop production is that of Greek-like unaspirated stops. The greatest percentage of stop tokens, i.e. 83,2%, belongs to the types (a), (b) and (c), which are considered to denote an improvement in production.

4. Discussion

The present study examined the VOT for word-initial voiceless stops produced by Greek learners of English before and after a specific pronunciation instruction. The first aim of the study was to investigate L2 production and, in particular, the effect of teaching in a foreign language context. The analysis of the results shows that, before the teaching intervention, interference from L1 Greek was prevalent in L2 production, as both groups tended to transfer the production of voiceless unaspirated stops into L2 English. In particular, the L2 VOT was not significantly different from that of L1 Greek at time 1. Our finding that Greek learners of English tend to transfer the L1 VOT in L2 production does not agree with previous studies that reported intermediate VOT values between L1 and L2 in the speech of the L2 speakers (Caramazza et al. 1973, Flege & Eefting 1987, Major 1987, Williams 1979). However, one major difference between our study and the earlier ones is that the latter examined speakers who acquired the L2 in a naturalistic setting, as they were exposed to the target language in their every day communication. The extensive exposure to the target language and to native-speaker input can explain the approximation to the target-like VOT values for the speakers examined previously; however, the foreign language context of acquisition may not provide learners with adequate native-like input and exposure to L2 in order to help them approximate the L2 VOT values, as a result speakers tend to resort to L1 transfer. As for the effect of teaching, at time 2 a significant increase was observed for the VOT of the experimental, but not of the control groups. Also the VOT of the experimental groups was not significantly different from English at time 2, whereas the VOT of the control groups was not significantly different from Greek, similarly to time 1. These results suggest that well-organized and planned pronunciation teaching can improve students production, even if it occurs in a foreign language environment with all the limitations that characterise it, for example lack of native English teacher, limited exposure and practice outside classroom, limited interaction in the target language in everyday life and limited amount of time devoted to English lessons at schools. It needs to be noted, however, that even though the statistical analyses revealed English-like VOT for the experimental groups at time 2, the standard deviation, as well as the implementation of the native-likeness criterion revealed

Phonology - Phonetics 2 0 5 Major Trends in Theoretical and Applied Linguistics

great variability in VOT production. In particular, the use of the native-likeness criterion proved a useful tool for the investigation of individual variability, as it helped identify four interlanguage patterns, one of which was closer to L1 Greek, two closer to L2 English and one was intermediate between the two. This finding suggests that, even though approximately one fourth of stop tokens were realised as target-like after teaching, the speakers’ production was still not systematic. This lack of systematic production can be related to the teaching intervention adopted, together with the foreign context of language acquisition. Perhaps the particular number of lessons might not have been adequate enough for aspiration to be produced systematically in a target-like way and that longer and more intense practice, as well as greater exposure to the language are required before students’ VOT production shows systematic improvement. Regarding the role of speakers’ gender, our results provide preliminary evidence of female students’ advantage over the male ones in the production of aspirated stops after the teaching intervention. In particular, although no difference between the male and female speakers appeared before teaching, after the teaching intervention the female speakers produced significantly longer VOT values than the male speakers. However, it needs to be noted that both the male and the female speakers produced longer aspiration at time 2 than at time 1, which suggests that speakers of both genders benefited from pronunciation instruction. Providing an explanation of the tentative advantage of the female students in the production of the particular feature appears quite challenging, as the examination of students’ background questionnaire fails to provide a clear answer. However, a tentative explanation comes from the teachers’ observation that, as a general trend, the female students, and in particular those who participated in the present research, tended to be more diligent and more attentive during the lessons than the male ones. Also according to the teacher’s observations, the female students generally tended to show greater enthusiasm and interest in the pronunciation lessons than the male students. This could potentially explain the advantage of the female over the male speakers, therefore it might be useful for future studies to investigate students’ gender in relation to attitudinal and motivational factors before firmer conclusions are obtained on the issue.

5. Conclusion

Overall, this study has established the need for teaching the phenomenon of aspiration to Greek learners of English, since without teaching Greek learners of English tend to transfer the Greek pattern of unaspirated stop production. The results indicate that the students can benefit from pronunciation instruction, as they generally produced longer VOT values after the teaching intervention than

2 0 6 Phonology - Phonetics Eleni Tsiartsioni

before teaching. Also a considerable percentage of stop tokens were produced in a native-like fashion, which indicates that native-like production of aspirated stops is an achievable goal even for students in a foreign context of language acquisition. Nevertheless, native-like performance was not systematic, possibly due to constraints related to the particular teaching intervention and the foreign context of language acquisition. Finally, an advantage of the female over the male students was reported and this was tentatively interpreted in relation to attitudinal and personality factors.

Phonology - Phonetics 2 0 7 Major Trends in Theoretical and Applied Linguistics

References

Aliaga-García, C. and J.C. Mora 2008. Assessing the effects of phonetic training on L2 sound perception and production. In A.S. Rauber, M.A. Watkins and B.O. Baptista (eds.), Proceedings of the Fifth International Symposium on the Acquisition of Second Language Speech Florianópolis, Brazil: Federal University of Santa California, 25-28 November 2007. Brazil, 10-27.

Allen, J.S., Miller, J.L. and D. DeSteno 2002. Individual talker differences in voice-onset-time. Journal of the Acoustical Society of America 113: 544- 552.

Arvaniti, A. 1987. Timing of Consonant Clusters: A Study of Initial Consonant Clusters in Modern Greek. M.Phil. Dissertation, University of Cambridge.

Arvaniti, A. 2001. Comparing the Phonetics of Single and Geminate Consonants in Cypriot and Standard Greek. In Proceedings of the Fourth International Conference on Greek Linguistics. Thessaloniki: University Studio Press, 37-44.

Birdsong, D. 2007. Nativelike pronunciation among late learners of French as a second language. In O.-S. Bohn and M.J. Munro (eds.), Language Experience in Second Language Speech Learning: In Honour of James Emil Flege. Amsterdam, Philadelphia: John Benjamins Publishing Company, 99-116.

Botinis, A., M. Fourakis and I. Prinou. 2000. Acoustic structure of the Greek stop consonants. Glossologia 11-12: 167-199. Athens: Leader Books.

Caramazza, A., G. Yeni-Komshian, E. Zurif and E. Carbone 1973. The acquisition of a new phonological contrast: The case of stop consonants in French- English bilinguals. Journal of the Acoustical Society of America 54: 421-428.

2 0 8 Phonology - Phonetics Eleni Tsiartsioni

Celce-Murcia, M., D. Brinton and J. Goodwin. 1996. Teaching pronunciation: A Reference for Teachers of English to Speakers of Other Languages. Cambridge: Cambridge University Press.

Deterding, D. and F. Nolan 2007. Aspiration and voicing of Chinese and English plosives. In Proceedings of the 16th International Conference of Phonetic Sciences, Saarbrücken, 6-10 August 2007: 385-388.

Ekstrand, L.H. 1982. English without a book revisited: The effect of age on second language acquisition in a formal setting. In S. Krashen, R. Scarcella, M.H. Long (eds.), Child Adult Differences in Second Language Acquisition. Roley, M.A.: Newbury House, 136-158.

Flege, J.E. and W. Eefting 1987. The production and perception of English stops by Spanish speakers of English. Journal of Phonetics 15: 67-83.

Flege, J.E. and K.L. Fletcher 1992. Talker and listener effects on degree of perceived foreign accent. Journal of the Acoustical Society of America 91: 370- 389.

Flege, J.E., M. Munro and L. Skelton 1992. Production of the word-final /t/-/d/ contrast by native speakers of English, Mandarin, and Spanish. Journal of the Acoustical Society of America 92 (1): 128-143.

Fourakis, M. 1986b. A timing model for word-initial CV syllables in Modern Greek. Journal of the Acoustical Society of America 79 (6): 1982-1988.

Goodwin, J., D. Brinton and M. Celce-Murcia 1994. Pronunciation assessment in the ESL/EFL curriculum. In J. Morley (ed.), Pronunciation Pedagogy and Theory: New Views, New Directions. Alexandria, VA: TESOL, 3-16.

Lisker, L. and A.S. Abramson 1964. A cross-language study of voicing in initial stops: Acoustical measurement. Word 20: 384-422.

Lowie, W. 2009. Early Birdies: The Sky’s the Limit: An Investigation for the Effects of Early Bilingual Education on the Acquisition of an L2 Sound System. M. A. Thesis. University of Groningen.

Lowie, W. and S. Reitsma 2009. The early bird catches the worm (?): The acquisition of an English sound system in Dutch primary school children. In Proceedings of the Phonetics Teaching and Learning Conference 2009, UCL London, 51-54.

Phonology - Phonetics 2 0 9 Major Trends in Theoretical and Applied Linguistics

Major, R.C. 1987. English voiceless stop production by speakers of Brazilian Portuguese. Journal of Phonetics 15: 197-202.

Nicolaidis, K. 2002. Durational variability in vowel-consonant-vowel sequences in Greek: The influence of phonetic identity, context and speaker. In M. Makri- Tsilipakou (ed.), Selected Papers on Theoretical and Applied Linguistics from the 14th International Symposium, April 20–22 2000, Thessaloniki, 280–294.

Olson, L.L. and Samuels, S.J. 1982. The relationship between age and accuracy of foreign language pronunciation. In S. Krashen, R. Scarcella and M.H. Long (eds.), Child Adult Differences in Second Language Acquisition. Roley, M.A.: Newbury House, 84-92.

Panagopoulos, E. 1972. Temporal differences distinguishing plosives in English and Greek: Voice Onset Time. In Occasional Papers 13. Essex: University of Essex Language Center, 77-91.

Pardo, D.B. 2004. Can pronunciation be taught? A review of research and implications for teaching. Revisita Alicantina de Estudios Inglesses 17: 6-38.

Piske, T., MacKay, I. and J.E. Flege 2001. Factors affecting degree of foreign accent in an L2: A review. Journal of Phonetics 29 (2): 191-215.

Poedjianto, N.H. 2003. The production of Indonesian English: Evidence from voicing contrast. In Proceedings of the 15th International Conference of Phonetic Sciences, Barcelona, 1089-1092.

Purcell, E. and R.W. Suter 1980. Predictors of pronunciation accuracy: A re- examination. Language Learning 30: 271-287.

Reis, M.S. and D. Nobre-Oliveira 2008. Effects of perceptual training on the identification and production of the English voiceless plosives aspiration by Brazilian EFL learners. In A.S. Rauber, M.A. Watkins and B.O. Baptista (eds.), Proceedings of the Fifth International Symposium on the Acquisition of Second Language Speech Florianópolis, Brazil: Federal University of Santa California, 25- 28 November 2007, 398-407.

Riney, T.J. and N. Takagi 1999. Global foreign accent and voice onset time among Japanese EFL speakers. Language and Learning 49 (2): 275-302.

Snow, C.E. and M. Hoefnagel-Höhle 1977. Age differences in the pronunciation of foreign sounds. Language and Speech 20: 357-365.

2 1 0 Phonology - Phonetics Eleni Tsiartsioni

Suter, R.W. 1976. Predictors of pronunciation accuracy in second language learning. Language Learning 26: 233-253.

Tahta, S., Wood, M. and K. Lowenthal 1981. Age changes in the ability to replicate foreign pronunciation and intonation. Language and Speech 24 (4): 363-372.

Thogmartin, C. 1982. Age, individual differences in musical and verbal aptitude, and pronunciation achievement by elementary school children learning a foreign language. International Review of Applied Linguistics 20 (1): 66-72.

Thompson, I. 1991. Foreign accents revisited: the English pronunciation of Russian immigrants. Language Learning 41: 177-204.

Williams, L. 1979. The modification of speech perception and production in second-language learning. Perception and Psychophysics 26: 95-104.

Phonology - Phonetics 2 1 1 Major Trends in Theoretical and Applied Linguistics

2 1 2 Section 2: Syntax - Morphology - Semantics

2 1 3 Major Trends in Theoretical and Applied Linguistics

2 1 4 The Proto-Kartvelian and Proto-Indo- European Common Typological Feature: An Active Alignment (?)

Rusudan Asatiani

Tbilisi State University, Georgia [email protected]

Abstract

The split-ergativity of the Kartvelian languages defined by the TAM categories and verb classes’ restrictions is represented as a result of hierarchically organized changes of linguistic structures. Various levels of hierarchies reflecting dynamic synchronic processes of linguistic structuring are interpreted as the stages of minor-syntactic constructions’ development based on the ‘appearance’ of certain grammatical categories (Telicity, Transitivity, TAM, S/O). The hierarchies reveal that the Proto-Kartvelian alignment system is semantically oriented an active one. Taking into consideration the reconstruction of Proto-Indo-European alignment as an active one (Gamkrelidze & Ivanov 1994/95) it seems possible to speak about a typological closeness of those proto-languages that further developed in different ways giving nowadays either the nominative (IE) or ergative (K) languages.

1. Introduction

Comparative-historical linguistics makes it possible to represent diachronic changes of conceptualization and linguistic structuring of reality. Changes of a language system mirror changes within various social-cultural spheres of human being such is art, religion, ethnic traditions, economy, civilization, culture, etc. On the basis of the comparative-historical methodology, a reconstructed proto-language and it’s diachronic development helps us to get complete information about the historical existence of the speakers including ecological environment (fauna, flora, geographic surrounding, climate) and human habitation, and migration in the environment as well as culture in the broadest sense (including both, material and spiritual culture). In this respect the most decisive are syntactic constructions, which reflect various models of linguistic structuring and create some picture of linguistic “world view”.

Syntax - Morphology - Semantics 2 1 5 Major Trends in Theoretical and Applied Linguistics

Syntactic constructions are built on the basis of a verb and its argument structure. Number of arguments (resp. Nouns) is defined by a verb valency and can be maximum four; e.g. in Georgian some constructions with two dative nouns can be interpreted as an argument structure showing four nouns formally linked with a verb:

(1) mi-m-i-k’er-a ma-n me ɣil-i kaba-s//-ze PV-IO.1-OV-sew-AOR.S.3.SG s/he-ERG 1.SG.DAT button-NOM dress-DAT//[DAT]-on ‘S/he sewed me a button on (my) dress.’

Various types of formal relations between a verb and its arguments are cross-linguistically specific and bilateral: a verb can define arguments’ form (resp. cases) while arguments themselves can trigger a verb form (resp. person- triggering markers).

Verb N1, N2, N3...

Thus, arguments’ case patterns and verb concord models (resp. alignment type of a concrete language) define the basic frame of a simple syntactic construction that can be reconstructed on the basis of morphological reconstructions reflecting the grammatical relations between a verb and arguments.

2. The Proto-Kartvelian Alignment System

2.1. The Main Models of Case Patterns in the Kartvelian Languages

In the Kartvelian languages (Georgian, Megrelian, Laz, Svan) structures of simple syntactic constructions are defined by the verb valency and three main models of arguments’ case patterns.

2 1 6 Syntax - Morphology - Semantics

Rusudan Asatiani

I. Nominative pattern: {A-argument is represented in Nominative (*-i/-0), B-argument – in Dative (*-s), C-argument – in Dative (*-s)}1;

(2) Georgian: k‘ac-i saxl-s a-g-eb-s man-NOM house-DAT NV-build-THM.-PRES.S.3.SG “A man builds a house.”

(3) Megrelian: k’oč(i) ‘ude-s o-g-an-s man.NOM house-DAT NV-build-THM.-PRES.S.3.SG “A man builds a house.”

(4) Laz: k’oči-k oxori k’od-um-s man-ERG house.NOM NV-build-THM.-PRES.S.3.SG “A man builds a house.”

(5) Svan: māre kor-s a-g-em man.NOM house-DAT NV-build-THM(PRES.S.3.SG) “A man builds a house.”

II. Ergative pattern: {A – in Ergative (*-ma/m)2, B – in Nominative (*-i/0), C – in Dative (*-s)};

(6) Georgian: k‘ac-ma saxl-i a-a-g-o man-ERG house-NOM PV-NV-build-AOR.S.3.S “A man built a house.”

(7) Megrelian: k’oč(i)-k(i) ‘ude da-g-u man-ERG house.NOM PV-build-AOR.S.3.SG “A man built a house.”

(8) Laz: k’oči-k oxori do-k’od-u man-ERG house.NOM PV-build-THM.-AOR.S.3.SG “A man built a house.”

1 The reconstructed case markers here and below are given in accordance to the works by Oniani (1978, 1989) and Machavariani (2002). 2 A reconstruction of the ergative case is the most problematic as all Kartvelian languages show various allomorphs: Georgian: /-ma/~/-m/∞/-n/ Megrelian: /-k/ Laz: /-k/ Svan: /-d////-em/~/-em/∞/-nem/ Conventionally, we indicate it as */ma/.

Syntax - Morphology - Semantics 2 1 7 Major Trends in Theoretical and Applied Linguistics

(9) Svan: māre-m kor ad-g-e man-ERG house.NOM PV-build-AOR.S.3.SG “A man built a house.”

III. Dative pattern: {A – in Dative (*-s), B – in Nominative (*-i/0), [C disappears as an argument of a verb and transforms into a Prepositional Phrase]};

(10) Georgian: k‘ac-s saxl-i a-u-g-i-a man-DAT house-NOM PV-SINV.3.CV-build-PF-OINV.3(SIVN.3.SG) “A man has built a house.”

(11) Megrelian: k‘o-s ude du-g-ap-u(n) man-DAT house.NOM PV(SINV.3)CV-build-THM-OINV.3(SIVN.3.SG) “A man has built a house.”

(12) Laz: k’oči-k oxori u-k’od-un man-ERG house.NOM (SINV.3)CV-build-THM-OINV.3(SIVN.3.SG) “A man has built a house.”

(13) Svan: māra-s kor x-o-g-em-a man-DAT house.NOM SINV.3- CV-build-THM-IONV.3(SINV.3.SG) “A man has built a house.”

2.2. Restrictions of the Case Patterns

The case-marking patterns are determined by the verb (conjugation) class in interaction with its tense/aspect/mood (resp. TAM categories) properties. Following the Georgian grammatical tradition TAM categories based on their morphosyntactic features are classified into three series: Series I: Present (Indicative and Subjunctive), Future (Indicative and Subjunctive), Imperfect, Conditional; Series II: Aorist Indicative, Subjunctive-II; Series III: Present Perfect, Past Perfect, Subjunctive III. The Kartvelian verbs are divided into four conjugation classes: class I mainly involves active transitive verbs (including derived causatives); class II involves dynamic-passives; class III involves active- intransitive verbs; and class IV contains constructions with an experiencer (resp. affective) and static-passive (including, so called, medio-passive) verbs (Compare with Shanidze (1973) and Harris (1981)). Conjugation classes can be defined by the features of Transitivity, Dynamicity, Telicity.

2 1 8 Syntax - Morphology - Semantics Rusudan Asatiani

Table 1. Conjugation classes

Verb classes Transitivity Dynamicity Telicity I-class: + + + Affected objects II-class: _ + + Achievements III-class: _ + _ Active processes IV-class: All others (Agentive states, Passive states, _ _ _ Experiencer)

The conjugation classes together with the series determine different case patterns shown below in the tables 2-4.

Table 2. Georgian and Svan

Series I Series II Series III

Arguments A B C A B C A B C

Class I NOM DAT DAT ERG NOM DAT DAT NOM (PP)

Class II NOM ------DAT NOM ------DAT NOM ------DAT

Class III NOM ------DAT ERG ------DAT DAT ------(PP)

Class IV NOM ------DAT NOM ------DAT NOM ------DAT (static) (experiencer) DAT ------NOM DAT ------NOM DAT ------NOM

Table 3. Megrelian

Series I Series II Series III

Arguments A B C A B C A B C

Class I NOM DAT DAT ERG NOM DAT DAT NOM (PP)

Class II NOM ------DAT ERG ------DAT NOM ------DAT

Class III NOM ------DAT ERG ------DAT DAT ------(PP)

Class IV (static) NOM ------DAT ERG ------DAT NOM ------DAT

(experiencer) DAT ------NOM DAT ------NOM DAT ------NOM

Syntax - Morphology - Semantics 2 1 9 Major Trends in Theoretical and Applied Linguistics

Table 4. Laz

Series I Series II Series III

Arguments A B C A B C A B C

Class I ERG NOM DAT ERG NOM DAT DAT NOM (PP)

Class II NOM ------DAT NOM ------DAT NOM ------DAT

Class III ERG ------DAT ERG ------DAT DAT ------(PP)

Class IV (static) NOM ------DAT NOM ------DAT NOM ------DAT (experiencer) DAT ------NOM DAT ------NOM DAT ------NOM

2.3. Verb Concord in Kartvelian

The Kartvelian language has two types of verbal person affixes denoting the person agreement; so called, the V-type and M-type affixes. Georgian: V-type M-type

sing. pl. sing. pl. I v- v- -t m- gv- II - - -t g- g- -t III -s,-a,-o -n,-en,-an, h-,s-,ø- h-,s-,ø- (-t) -nen, -es ø- ø- Megrelian: V-type M-type

sing. pl. sing. pl. I v- v- -k m- m- -t, -n,-(n)en,-an,-es II - - -k g- g- -t, -n,-en,-(n)an,-es III -s,-n,-o -an,-a, ø- ø- -n,-na(n), -es Laz: V-type M-type

sing. pl. sing. pl. I v- v- -t m- m- -t, -n,-(n)en,-an,-es II - - -t g- g- -t,-n,-(n)en,-an,-es III -s,-n,-u -n,-(n)an, ø- ø- -es

2 2 0 Syntax - Morphology - Semantics Rusudan Asatiani

Svan: V-type M-type

sing. pl. sing. pl. I-excl. (x)û-/m (x)û-/m- -d m-(η)- gû - incl. l-(η)- -d n-(η)- II x- x- -d ǯ- ǯ -x III l-/-s,-a,-e l- -x *h-,x- *h- -(x)

Traditionally the V-type affixes are considered to be the subject markers, while the M-type are qualified as the object markers. However, this is not always the case: In the Perfect3 tense forms and also with the affective (psychological) verbs, the subject appears with the M-type and the object with the V-type affixes. For that reason most Georgian scholars qualify these forms as the inversive ones. To avoid such inconsistency of functional interpretations, it seems better to analyze these markers without any functional qualification, simply by their relation to cases. 1. Noun in the Dative always triggers the M-type affixes; 2. Noun in the Ergative always triggers the V-type affixes; 3. Noun in the Nominative triggers either (a) V-type (if there is no ergative linked with the verb as well), or (b) M-type (in case there is an ergative linked with the verb), or (c) Zero (if both ergative and dative appear in the construction).

2.4. Interpretation of the Case Patterns and Verb Concord Models

On the basis of semantic-functional analysis of above presented models some regularities can be observed: 1. The first or second personal pronouns (I or II) in all Kartvelian languages have no case endings. As for a verb concord, they behave similarly as Ergative, Nominative or Dative nouns. Consequently, we can speak about

3 The Kartvelian Perfect demonstrates an additional semantic nuance: “apparently”, “it seems“, “probably”. It represents the following aspectual situation: The speaker sees the result of an action, (s)he does not pay any attention to Ag (or (s)he is not sure; or (s)he does not actually know; or (s)he merely forgets, who was the Agent of the action), but because of the actually presented result (Patient), (s)he says, what “apparently” happened; e.g. dauxat’avs ‘It seems that (s)he has drown’, ucxovria ‘Apparently (s)he has lived’, augia ‘Apparently, (s)he has built it’, and so on.

Syntax - Morphology - Semantics 2 2 1 Major Trends in Theoretical and Applied Linguistics

I/II Ergative (always triggers V-type person markers), I/II Dative (always triggers M-type person markers) and Nominative (triggers either M-type, or V-type, or ø-.) 2. Constructions with an affective verbs and experiencer subject (class IV- experiencer) are the most consecutive; they always exhibit the Dative system: an experiencer (A) that functionally might be qualified as the subject is represented by the Dative case and a stimulus (C) – as the object, consequently, represented by the Nominative case; 3. The ergative is the only case strictly linked to one and only one grammatical function: if ergative, then the subject. Any other case is polyfunctional: the Nominative refers either to a subject or a (direct)-object; the Dative represents an indirect object, a subject-experiencer and class I or class III verbs’ subject in series-III, also, a direct object in series I (except Laz). 4. The ergative construction appears if two conditions are met – the ergative construction arises if and only if: (i) a verb is of class I (transitive) or class III (intransitive-atelic); (ii) a verb is in the series II. Thus, there is a split-ergativity (Dixon 1979, Harris 1981) in the Kartvelian languages: Nominative and Ergative constructions are distributed and defined by the verb-classes and TAM categories. The restrictions are functioning in the Kartvelian languages variously: • Megrelian keeps to only (ii): Any kind of verbs in series II trigger the ergative case pattern and, consequently, follow the V-type verb concord models, while any kind of verbs in series I show the nominative construction; • Laz – only (i): Any kind of transitive (or broadened along the activity cat- egory4 intransitive-atelic) verbs trigger the ergative case pattern and, consequently, follow the V-type verb concord models, while all other kind of verbs show the nominative construction; and this regularity is canonical for all series; • Georgian and Svan case patterns are defined by the both, (ii) and (iii), restrictions: Only transitive or broadened along the activity category intransitive- atelic verbs, and only in series II show the ergative construction, while all other cases show the nominative construction. 5. Interpretations of morphosynstactic variations according to syntactic functions are more complicated (see 6.) and cannot be defined simply by syntactic functions: there are no unequivocal, simple correspondences

4 See the work by Winfried Boeder, who analyses the Georgian “ergative construction” as not being the typical one but as a construction broadened along the activity category (Boeder 1979); also, the work by Irine Melikishvili (Melikishvili 2012).

2 2 2 Syntax - Morphology - Semantics Rusudan Asatiani

between the syntactic functions and cases. It seems better to analyze the formal models according to the semantic roles that can be differentiated by the semantic feature ‘Free Will of Arguments’. The feature ‘Free Will’ is regarded as the main characteristic feature for describing and distinguishing the semantic roles: • Agent – +[FW] (an argument who acts and controls an action);

• Addressee/Experiencer/Agentunknown – -[FW] (an argument who does not control an action); • Patient – Ø (an argument who is under an action and its controlling is logically excluded; so, for the patient the feature ‘Free Will’ is redundant). According to this feature the appearance of different verb concord models can be described by the following hierarchically organized rules I>II>IIIa/>IIIb/>IIIc/>IIId: I An argument whose free will is not included in a situation (or it is unknown whether its free will is included or not) triggers the M-type affixes. (Semantically such are: Addressee, Experiencer, and an actually ‘unknown’ Ag of Perfect tense forms); II An argument that acts according to its free will triggers the V-type affixes (such is Ag). III An argument left after the application of I>II rules, triggers either a) The V-type, if it is only argument linked with a verb (P); or b) The V-type, if other argument’s free will is not meant in a situation (constructions: {P-Ad}, or {P-Exp}, or {P-unknownAg}); or c) The M-type, if other argument’s free will is meant in a situation (construction Ag-P); d) Zero, if both other arguments (with +[FW] and with –[FW]) are linked with a verb (construction : {Ag-P-Ad}). Taking into account all above given discussion, a reconstruction of ergative alignment is well-grounded, yet, it is difficult to define one function andto reconstruct one morpheme for the ergative case (see reference 2). Yet, there exists another restriction as well: (iii) an argument must be the third person represented either by the third Personal Pronoun or a Noun. The last restriction functions in every Kartvelian language and this regularity leads us to suppose that Kartvelian languages, first of all, strive for formal markedness of I/II versus III persons dichotomy and the arguments for such an interpretation are the following (Asatiani 1999): 1. Case patterns: I and II pronouns are undifferentiated according to the cases, while III (pro)nouns show very developed case systems: there are seven functionally and formally differentiated cases in Georgian and Laz, nine cases – in Megrelian, six cases – in Svan; 2. Verb concord: I/II person singular always triggers prefixes in verb forms, while III person either singular or plural – mostly suffixes(except Svan);

Syntax - Morphology - Semantics 2 2 3 Major Trends in Theoretical and Applied Linguistics

3. Variation of III person allomorphs: Allomorphs of I/II are pure and mostly defined by various phonetic processes, while III person allomorphs are rich; they are morphologically restricted and defined by various features: animateness, active- passive opposition, tense-mood differences, functionally defined competitive situations and so on; 4. Number agreement: I/II person always agrees with a verb in number, while III person is semantically and functionally more restricted from this point of view; 5. Main function of polypersonality of the Kartvelian verb is the obligatory formal markedness of I/II persons: the person markers of I/II persons (despite their functions – as subjects so objects) are always presented in verb forms in combinations with III person; 6. Category of version: if an action is directed/oriented to I/II person, a verb has *-i- prefix; while in case an action is directed/oriented to III person, a verb has either *-i- (when III person is subject) or *-u- (when III person has functional qualification of indirect object); 7. Category of direction: if an action is directed towards I/II persons, verb forms are denoted by one kind of prefixes (so called, preverbs) expressing direction, while other kind of preverbs are used in case an action is directed towards III person. The arguments show that in the Kartvelian languages the tendency of formalization of person dichotomy is in privileged position and the dominant category which defines the process of linguistic structuring of events is the deictic anchoring: first of all, the grammaticalization of I/II versus III dichotomy is decisive. Appearance of nominative or ergative constructions are restricted and characteristic mostly for III person subsystem, first of all, defined by the semantic roles and the semantic feature ‘Free Will of Arguments’. The tense-mood, telicity, transitivity and functional S/O categories play an important role as well. We offer to represent the alignment systems of Kartvelian languages as hierarchically organized subsystems reflecting the restrictions and morphosyntactic features that define the appearance of case patterns and verb-concord models.

LAZ

I stage: I/II III (arguments are undifferentiated)

II stage: Free Will ø(P)

III stage: +FW (Ag) -FW (Ad/Agun/Exp)

Noun cases: ERG DAT NOM Verb concord: V-type M-type V-type

2 2 4 Syntax - Morphology - Semantics Rusudan Asatiani

MEGRELIAN

I stage: I/II III (arguments are undifferentiated)

II stage: Free Will ø(P)

III stage: +FW (Ag) -FW (Ad/Agun/Exp)

IV stage: +aorist -aorist +aorist -aorist

V stage: S O S O

Noun cases: ERG NOM DAT ERG NOM NOM DAT/NOM Verb concord: V-type V-type M-type V-type M-type V-type M/V-type

GEROGIAN and SVAN

I stage: I/II III (arguments are undifferentiated)

II stage: Free Will ø(P)

III stage: +FW (Ag) -FW (Ad/Agun/Exp)

IV stage: +aorist -aorist +aorist -aorist

V stage: S.int/tel O S.int/tel O .

Noun cases: ERG NOM DAT NOM DAT NOM DAT/NOM Verb concord: V-type V-type M-type V-type M-type V-type M/V-type

The hierarchies are defined according to the priority given to marked categories during the surface realizations. They reflect dynamic synchronic processes of linguistic structuring of relations existed between a verb and its arguments. We suppose that the hierarchies can explain diachronic linguistic processes as well: Various levels of hierarchies could be interpreted as the stages of

minor-syntactic constructions developed on the basis of appearance of certain grammatical categories (S/O, TAM, telicity):

I-II stages – I/(>II):III – Morphosyntactic models are defined through the dichotomy of communicative act participants – Georgian/Megrelian/Laz/Svan;

III stage – [+FreeWill](Ag):[-FreeWill](Ad):ø(P) – Morphosyntactic models are defined through the opposition of semantic roles (Ag:Ad:P) – Laz;

Syntax - Morphology - Semantics 2 2 5 Major Trends in Theoretical and Applied Linguistics

IV stage – [+Aorist]:[-Aorist] – Morphosyntactic models are defined through the opposition of TAM categories – Georgian/Megrelian/Svan;

V stage – Subject:Object – Morphosyntactic models are defined through the syntactic functions – Georgian/Megrelian/Svan.

Supposedly, development of constructions might be the result of some changes of Kartvelians’ world cognition that is reflected by the changes of certain linguistic structures creating a picture of linguistic “world view”. Cognitive broadening of a world goes from a speaker (resp. I person, Ego Space) via a listener (resp. II person, close to Ego Space) to other world around (resp. III person, Alter Space). Entities of a world that are conceptually more close to a speaker are conventionally excluded from an Alter Space and included into the Ego Space in accordance with the Silverstain’s hierarchies (Silverstain 1976): I Person/>II Person (I/II stage)>Humans (III stage) >Animates />Actives (IV stage) and developed in accordance with the specific linguistic cognition of the Kartvelians reflecting by the rising of grammaticalized semantic-functional oppositions: Active-transitive and Intransitive-telic verbs are opposed Inactive- intransitive and Intransitive-atelic verb forms (IV stage). Consequently, step by step, first of all active constructions and afterwards broadened ergative constructions have arisen (V stage):

I stage: Alter Space Ego Space

II stage:

Alter Space I/II person

III stage:

Alter Free will – WHO/humans Space - WHAT )

IV stage: AS – 2 2 6 inanimate Syntax - Morphology - Semantics Animate>Active/ />inactive >Transitive/telic

V stage:

Inact /Intr. Sintr /Telic Inactive

Str/

Aint/ Sintr/ O Telic atelic Ergative nominati Atr/ Active Aint- Atelic Sintr

Nominative

I stage: Alter Space Ego Space

II stage:

Alter Space I/II person

III stage:

Alter Free will – WHO/humans Space - Rusudan Asatiani (WHAT )

IV stage: AS – inanimate Animate>Active/ />inactive >Transitive/telic

V stage:

Inact /Intr. Sintr /Telic Inactive

Str/

Aint/ Sintr/ O Telic atelic Ergative nominati Atr/ Active Aint- Atelic Sintr

Nominative

Figure 1.

3. Proto-Indo-European as a Language of Active Alignment

Thomas Gamkrelidze and Viacheslav Ivanov have reconstructed the Proto-Indo- European language as a language of active alignment (Gamkrelidze, Ivanov 1994/95). Very briefly, their arguments for such a reconstruction are based on the analysis of following data: • The formation of *-os and *-om genitive and their correlation with the binary classification of nouns into active and inactive; • The nominative *-os and accusative *-om as original markers of active and inactive noun classes; • Doublet verb lexemes as a reflexes of the binary semantic classification of nouns into active and inactive; • The two series of verbal endings, *-mi and *-Ha, associated with active and inactive arguments; • The suffix *-nth’s interpretation as a marker of membership in the active class.

Syntax - Morphology - Semantics 2 2 7 Major Trends in Theoretical and Applied Linguistics

These and some other arguments lead the authors to the conclusion (Gamkrelidze, Ivanov 1994/95: 238):

“It is clear that a basic structural-semantic principle of early Indo- European was a binary classification of nouns into active and inactive based on the character of their denotata. The active class included nominal formations referring to people, animals, trees, and plants, i.e. those whose denotata are alive, while the inactive class referring to objects lacking life cycles.”

Afterwards they are developing their reasoning: Naturally active denotata are cognitively broadened and inanimate objects viewed by speakers as expressing the active principle or endowed with the capacity for action are structurally unifying: *-s, *-os turns into the marker of active class (correspondently, of an structurally active argument of a verb) and *-om, *-t’, *-th, *-kh, *-Ø becomes the marker of the inactive class (correspondently, of an structurally inactive argument of a verb): Active: *-s/*-os Active (A): *-s/*-os Genitive: *-om → Inactive (In): *-om, Ø Structural inactive: *-[o]m Structural-syntactic inactive (A { in): *-m/*-om

Further development of constructions in the Indo-European languages is presented as a process of changes of linguistic “world view”: Semantically oriented Active Constructions have been transformed into the functionally oriented Nominative Constructions5. If the authors’ diachronic analysis of the Proto-Indo-European alignment system changes are concurred then some general results could be established.

5 Cf. Gamkrelidze, Ivanov (1994/95: 273): “The breakdown of the active/inactive dichotomy in nouns and the rise of the transitivity opposition in verbs bring about the rise of subject-object relations and triggers the functional identification of the originally inactive argument of a one-place verb with the historically active argument of one-place (intransitive) and two-place (transitive) verbs. Thus the inactive argument of a one-place intransitive verb is syntactically opposed to the former structural inactive with two-place verbs, which becomes the case (accusative) of the direct object of a transitive verb. From the Proto-Indo-European active type shown in figure 4 we get the typical morphosyntactic structure of the accusative type, shown in figure 5” (see Figure 2 in this article).

2 2 8 Syntax - Morphology - Semantics Rusudan Asatiani

A-os A S int. Active -os

Str. O-[o]m -[o]s In*-[o]m e in A-[os] Accusative S intr. -om Inactive In-om .

Nominative

Figure 2. 4. Conclusions

(a) The Proto-Indo-European alignment system can be qualified as an Active/ Inactive one;

(b) The Proto-Kartvelian alignment system (actually, V stage of its development) is qualified as an Active/Inactive one;

(d) This common typological feature raised in various backgrounds has differently developed in Proto-Indo-European and Proto-Kartvelian.

Active Construction

Nominative Construction Ergative Construction PIE PK

(e) The different resulted systems supposedly are implied by various cognitive processes in progress reflecting their specific way of linguistic structuring. PIE: Alter Space Ego Space

S O NOM ACC Active Syntax - Morphology - Semantics 2 2 9 In*

PK: Ego Space Alter Space

Inact Sintr Stran / O Atrans ERG NOM Gg G Aintr

Active Construction

Major Trends inNominative Theoretical Construction and Applied Linguistics Ergative Construction PIE PK

PIE: Alter Space Ego Space

S O NOM ACC Active

In*

PK: Ego Space Alter Space

Inact Sintr Stran / O Atrans ERG NOM Gg G Aintr

Figure 3.

2 3 0 Syntax - Morphology - Semantics Rusudan Asatiani

References

Asatiani, R. 1999. Markedness and the Dominant Category. The Proceedings of 3rd and 4th Symposium on Language, Logic and Computation. Amsterdam: ILLC Publications.

Asatiani, R. 2008. Directions of Diachronic Developments of the Kartvelian Languages. Bulletin (Moambe), vol. II, no.1. Tbilisi: Georgian Acad. Press, 135- 139.

Boeder, W. 1979. Ergative Syntax and Language Change: The South Caucasian Languages. In P. Plank (ed.). London, 435-480.

Dixon, R.M.W. 1979. Ergativity. Cambridge: Cambridge Un. Press.

Gamkrelidze, T. and Ivanov, V. 1994/95. Indo-European and the Indo-Europeans. A Reconstruction and Historical Analysis of a Proto-Language and a Proto- Culture, 2 vols. English version by J. Nichols, with a Preface by R. Jakobson, Berlin-New York: Mouton de Gruyter.

Harris, A.C. 1981. Georgian Syntax. A study in relational grammar. Cambridge: Cambridge University Press.

Harris, A.C. 1985. Diachronic syntax: the Kartvelian case. Syntax and Semantics, 18. New York: Academic Press.

Harris, A.C. 1991. Mingrelian. The Indigenous Languages of the Caucasus, Vol.1, The Kartvelian Languages, Delmar-New York: Caravan Books, 315-394.

Holisky, D.A. 1991. Laz. The Indigenous Languages of the Caucasus, Vol.1, The Kartvelian Languages, Delmar-New York: Caravan Books, 395-472.

Syntax - Morphology - Semantics 2 3 1 Major Trends in Theoretical and Applied Linguistics

Kartozia, G. et al. 2010. megrulis lingvist’uri analizi [Linguistic analysis of Megrelian]. Tbilisi: Meridian.

Machavariani, G. 1966. k tipologicheskoj kharakteristike obshekartveljskogo jazyka [Typological Characteristics of the Proto-Kartvelian Language]. Voprosy jazikoznanija, No.4. Moskva. 60-70.

Machavariani, G. 2002. kartvelur enata šedarebiti gramat’ik’a (lekciebis k’ursi) [Comparative grammar of the Kartvelian Languages (Lecture course)]. Tbilisi.

Melikishvili, I. 2012-forthcoming. Two Sets of Georgian Person Markers as the Expression of the Opposition Active/Inactive. In: Special Issues on Georgian in the Typological perspective, The Journal of language Typology and Universals/ Sprachtypologie und Universaliedforschung (STUF).

Oniani, A. 1989. kartvelur enata šedarebiti gramat’ik’is sak’itxebi (saxelta morpologia) [The issues on the Kartvelian language’ comparative grammar (Noun morphology). Tbilsi.

Oniani, A. 1998. svanuri ena [Svan Language]. Tbilisi: PU press.

Shanidze, A. 1973. kartuli enis gramat’ik’is sapuʒvlebi, I, [Fundamentals of the Grammar of the Georgian Language, I]. Tbilisi: TSU Press.

Sirverstain, M. 1976. Hierarchy of Features and Ergativity. In: R.M.W. Dixon (ed.): Grammatical Categories in Australian Languages. Canberra: Australian Institute of Aboriginal Studies.

Topuria, V. 1931. svanuri ena, I. zmna [Svan Language, I. A verb]. Tbilisi.

2 3 2 Syntax - Morphology - Semantics A Conceptual Metonymy Account of Count and Non-Count Nouns: A Study of Modern Greek Nouns from the Domains of Eating and Drinking

Ifigeneia Athanasiadou and Martha Lampropoulou

Aristotle University of Thessaloniki [email protected], [email protected]

Abstract

The aim of the paper is to examine how conceptual metonymy influences the way food and drink nouns are conceptualized as count or non-count in specific utterance contexts. The nouns examined are the following: tsai (tea) , gala (milk), kafes (coffee), nero (water), fakes (lentils), fasolia (beans), arakas (peas), sitari (wheat), vromi (oats), halvas, and zeles (jelly). The sentences are either constructed examples or borrowed from the Hellenic National Corpus (HNC). The paper is grounded on Idealized Cognitive Models (Lakoff 1987) and on the model of conceptual metonymy (Radden and Kövecses 1999).

1. Introduction

The paper is an attempt to identify countability preferences as motivated by the use of conceptual metonymies. Countability should not be claimed to be a “black or white” issue. Instead of being either countable or not, most food and drink nouns in Modern Greek oscillate between a countable and non-countable conceptualization depending on the respective Idealized Cognitive Model. Substance-mass nouns, like coffee or water, typically receive amount quantification, since the real-world physical substance they refer to can only be conceptualized as extending to an unbounded, continuate stretch of space (Hirtle 1982: 76-77). Yet, this is not always the case since mass-to-count countability shifts are often observed, leading to a non-count noun being used in a count-noun syntactic frame. For instance, even if water (‘nero’) is a non- count noun, in the ICM of a container (glass, bottle, etc.), water can turn into a countable noun, namely, “two waters” (‘dio nera’), metonymically referring to two glasses of water or two bottles of water.

Syntax - Morphology - Semantics 2 3 3 Major Trends in Theoretical and Applied Linguistics

2. Theoretical Background

Structural and transformational approaches consider the language faculty as an independent component of the human mind with its own self-contained structure and function (Taylor 1995: 16). One would wonder, though, how language could possibly be independent of what it stems from. As Geeraerts (2006: 5) very aptly puts it: “language use is undeniably related to human experience and perception of the outside world and linguistic meaning, among other things, is encyclopedic, in that, it results from our world knowledge”. The major principles (Geeraerts 2006: 7-20) that characterize the theoretical background of the paper are presented below. In the first place, linguistic meaning is perspectival, i.e. the same situation may be conceived differently by different speakers since every single speaker uses his/her own idiolect sourcing from his/her perspective, e.g. by asking for “a water” (‘ena nero’) a speaker may implicitly refer to a glass of water or a bottle of water. Secondly, linguistic meaning is dynamic and flexible. Etymological research revealed that the noun tsai (tea) was firstly introduced in Greece in the 18th century (Andriotis 1995: 378). Similarly, the word ‘tea’ and the practice of drinking tea was first introduced in England in 1644, while in 1738 the word carried the meaning “afternoon meal at which tea is served” (Online Etymological Dictionary). Thirdly, language use uncovers social and cultural identities, in that, language use reflects the world as it is experienced by the norms of every single society. For example, oats (‘vromi’) is a plural-only noun in English while the equivalent word is singular-only in Greek. English speakers focus on the discrete oat grains while Greek speakers highlight its collective denotation (Athanasiadou 2009:39). This paper is based on two crucial factors of the Cognitive Linguistics framework, namely, grammatical patterns are triggered by conceptual and pragmatic factors and, more specifically, metonymy and metaphor may have some impact on grammatical form (Panther and Thornburg 2009: 4-5).

3. Conceptual Metonymy

A rather popular definition of conceptual metonymy is provided by Lakoff (1980: 35): “Using one entity to refer to another that is related to it” [e.g. The ham sandwich is waiting for his check (= the client who ordered the ham sandwich)]. During metonymic processes, the mapping between the two entities is realized within a single conceptual domain. In other words, the name of one entity is used to refer to another entity that is contiguous to it. Thus, metonymy can be considered as having a reference function. Metonymy is based on contiguity, that is actual proximity or association (Geeraerts 2006: 13). This means that the substitution of one term for another depends on a kind of association or relation between the two terms.

2 3 4 Syntax - Morphology - Semantics Ifigeneia Athanasiadou and Martha Lampropoulou

For Croft (1993), contiguity reveals that encyclopedic knowledge is organized into domains and, as far as metonymy is concerned, two domains interact, a matrix domain and a subdomain. For Panther and Thornburg (1999), contiguity shows that our knowledge is classified in terms of frames and scenarios. And, for Lakoff (1987: 288), contiguity can be explained by means of Idealized Cognitive Models and conceptual domains, as will be discussed below. For Radden and Kövecses (1999), which is the approach adopted in the paper, metonymy occurs when an idealized cognitive model is activated. In other words, Idealized Cognitive Models involve any possible associations stimulated in our mind. Suppose we use the word ‘tea’ and we think of the ‘tea ICM’. The latter may, among many other things, refer to “a cup of tea” (Container ICM) or to a “gathering of friends to drink tea” (Event ICM). ICMs, the outcome of a one-to-one correspondence mapping between concepts, entities, and things, could be classified into three broader categories (Radden and Kövecses 1999: 24-28): a) sign ICMs and sign metonymies, e.g. using the word ‘dollar’ for money in general b) reference ICMs and reference metonymies, e.g. using the word ‘cow’ to refer to the real-world cow; both the word ‘cow’ and the concept of the animal function as reference for the actual cow one perceives visually c) concept ICMs and conceptual metonymies, e.g. employing the word ‘bus’ to refer to bus-drivers in general. The concept ICMs constitute the focus of the paper.

3.1. Why Metonymy for Countability

Since metonymy is based on substitution and on a contiguous relationship between two terms, it seems that it could reveal the countability preferences one has in mind when he or she uses count or non-count nouns. In other words, it will uncover the target term speakers have in mind, namely, a discrete unit or a continuous substance, when, for example, asking for “a glass of water” or “water”; in this respect, Langacker (1993:30) emphasizes the need to be accurate and argues that metonymy allows for an efficient reconciliation of two conflicting factors: that of being sure that the addressee’s attention is directed towards the intended target and that of our natural inclination to think and talk explicitly about those entities that have the greatest cognitive salience for us.

3.2. Idealized Cognitive Models: How Metonymy Accounts for Count and Non-count Nouns

Depending on the metonymic relationships being used, one would observe that countability preferences vary according to the corresponding setting. However,

Syntax - Morphology - Semantics 2 3 5 Major Trends in Theoretical and Applied Linguistics

commonalities in the way food and drink count and non-count nouns are conceptualized and expressed occur due to the scenarios they are part of. Such scenarios mostly have to do with: a) ordering at restaurants, taverns, cafeterias b) purchasing goods c) means of collecting goods or referring to their properties or to the type of goods in a generic sense.

4. Nouns and Countability

In order to clarify the notion of countability, it would be helpful to explore the way nouns, or else nominal predications, are presented in the framework of Cognitive Linguistics. According to Langacker (1987: 183), nominal predications designate things whereas relational predications designate atemporal relations or processes. Things not only refer to physical objects in the real world but to abstract cognitive events as well. In detail, a thing occupies a region in some domain. A region is defined schematically as a set of interconnected entities (Langacker 1991: 15). Cases where the profiled region is conceptualized as being bounded ina primary domain are characterized grammatically as prototypical count nouns (Langacker 1987: 189). Prototypical count nouns refer to physical objects in the real-world. Something is construed as being a bounded region because the conceptualizer imposes a “virtual boundary” in framing a conceived situation (Langacker 1987: 196). On the contrary, non-count nouns, while occupying a region in some domain, this region is characterized by internal homogeneity and construed as a continuous uniform whole. Prototypical non-count nouns refer to physical substances in the real-world. The specification of a boundary is vital to the distinction between count and non-count nouns. Overall, the ontological distinction between the two principal types of things, namely, objects and substances, is reflected in the grammatical distinction between count and non- count nouns. Radden and Dirven (2007: 72) refer to a type-of relation. Since a physical substance can be conceptualized as a discrete object, the respective noun denoting the continuous substance can be pluralized like any other count noun referring to a real-world physical object. A bounded discrete amount of a substance can stand for a variety or type of that substance. A singular mass noun, when pluralized, refers to different varieties or types of that substance. However, notice that substances construed as objects do not occupy space in the way prototypical objects do but that they are delimited by the containers they are placed in. Bearing in mind the aforementioned approaches with respect to the conceptualization and use of food and drink nouns, the analysis that follows

2 3 6 Syntax - Morphology - Semantics Ifigeneia Athanasiadou and Martha Lampropoulou

attempts an approach to countability by means of a metonymic model (Radden and Kövecses 1999).

5. Analysis

The formation of ICMs results in the use of metonymies. The relationships that give birth to metonymies are called metonymy-producing relationships. The metonymies produced are various and may overlap. However, two major configurations can be distinguished (Radden and Kövecses 1999: 30): (i) the formation of a whole ICM via its parts. For example, one may say: “the lightning of the Christmas tree” instead of “the lightning of the candles on the Christmas tree” In this case, the Whole is the Christmas tree, while the Part is the set of candles. Despite the fact that speakers are aware that the candles are lightened, they refer to the tree as being lightened itself. And (ii) the formation of a part of an ICM via its whole. For example, one may utter: “those are cool wheels you have there” instead of “you have got a cool high speed car” In this example, the Part is the wheels, while the Whole is the car. The speaker uses the Part, namely, the wheels, instead of the Whole, that is the car. Or Some commonly used words may function metonymically, such as hand for control: e.g. Things got out of hand. Similarly, the hand is the Part which stands for the Whole, which is the word ‘control’. These two configurations apply to all metonymic relationships. Hence, despite the fact that a metonymic relationship is either a Whole for Part or a Part for Whole metonymy, the particular metonymic relationship should be further classified with regard to the metonymic type it complies with. For instance, it should be specified whether it is an Event ICM or a ‘Category and Member’ ICM. An ICM type that could cover most of the examples in the paper seems to be the Event ICM since it comprises all the main activities typically encountered in ordinary food and drink contexts: ordering, purchasing, and collecting goods “see Section 3.2”

5.1. Event ICM (Radden and Kövecses 1999: 32)

In the utterance: I am drinking milk, one can observe the following metonymic configuration:

Syntax - Morphology - Semantics 2 3 7 Major Trends in Theoretical and Applied Linguistics

It is a ‘Part for Whole’ metonymy and more specifically an Event ICM. The event taking place involves the common scene of: having breakfast in the dining room or the kitchen and drinking a glass of milk or having two mugs of milk. By using a bare substance-mass noun, that is milk, the speaker focuses on the event and not on the amount of the liquid he or she is having, namely, the glass or the mugs. The use of the word ‘milk’ functions metonymically. Thus, the Part is the subevent of drinking something; in particular, milk: I am drinking milk. Contrary to the Part, the Whole event comprises many more details that can easily be represented in the speaker’s mind without being specified. The use of the singular shows that the speaker focuses on the event, not on the amount of the liquid s/he is drinking.

(1) Ένα νερό στο 6. ‘A water for 6’1.

(2) Μπομπ, φέρε μου, σε παρακαλώ, ένα τσάι. ‘Bob, bring me a tea, please’.

In examples (1) and (2) a restaurant or a café ‘frame’ is being indicated and an Event ICM is being constructed, that of ordering something to drink. We do not clarify the amount of the substance to be served, yet, the speaker highlights a countable conceptualization of both drinks. Obviously, whenever we order “a water” (‘ena nero’), or “a tea” (‘ena tsai’) a bounded amount of the respective substance is being implied. But, do we mean a glass of water, a bottle of water, or still something else? Only by finding ourselves in the relevant ‘frame’ can we mentally access the denoted container, be it a glass or a bottle. In (1) even if we use the amount quantifier “a” (‘ena’), we still do not know the exact container the water is to be placed in. The “type of serving” of the water (Tsoulas 2007:2) is not explicitly mentioned but evokes from an Event-ICM. Hence, in the particular example, a Containment ICM could also be activated under the Event ICM “see Section 5.2” In (2) we can assume various scenarios. For instance, the scenario of being in a tea-house and asking to buy some tea, or being in a field seeded with tea (a tea plantation) and collecting tea. Both Event ICMs facilitate the speaker to conceptualize a package or a handful of tea, respectively. In other words, a bounded amount of tea is implied and the Event ICMs trigger the singular use of

1 The Greek example sentences are translated in their English equivalents, except for the ones preceded by a question mark which are a word-for-word translation of the original utterance.

2 3 8 Syntax - Morphology - Semantics Ifigeneia Athanasiadou and Martha Lampropoulou

the word ‘tea’ which, in turn, enables the speaker to omit a lot of the scenario’s details.

(3) Τις φακές τότε τις κρατούσαμε μέσα σε λάδι για να μην πιάνουνε σκουλήκι - όπως τις καθαρίζαμε, τα χέρια μας λαδώνονταν μέχρι τους αγκώνες (HNC). ‘We used to keep lentils in oil so that worms wouldn’t eat them and as we were cleaning them, grease would drip from our palms till our elbows’.

The substance-mass noun lentils (‘fakes’) illustrates an Event ICM, that is the way of producing and preserving the plant lentils. Notice that lentils is used in a plural count-noun sense since the lentils are conceptualized as discrete grains in this case, they are thus not an ingredient of a lentil soup or of another kind of food. The Event ICM of preserving and cleaning the separate lentil grains enables and probably requires the use of the plural.

(4) O Ρίτσαρντ Χόλμπρουκ, 57, που είδε χθες τον κ. Πάγκαλο για το Κυπριακό - είναι πολύ ευφυής, υπερβολικά φιλόδοξος, πίνει άπειρους καφέδες την ημέρα, έχει κακούς τρόπους και θέλει πάντοτε να επιβάλλεται στους συνομιλητές του (HNC). ‘Richard Holbrook, 57, who met Mr. Pagkalos yesterday to discuss the Cypriot issue – is very intelligent, drinks numerous coffees per day, has bad manners and always wants to keep control of his interlocutors’.

In (4), we have the Event ICM (Lakoff and Johnson 1980:15-20) since coffees stands for the drinking of many coffees on a daily basis. In other words, the substance stands for the numerous events that are characterized by its consumption. It is also a ‘Part for Whole’ metonymy: Part = the subevent of having coffees, Whole = the scenario of consuming many coffees on a daily basis.

(5) Το ‘Βυζάντιον’ είχε κι αυτό πάτωμα εκ χώματος, έμπαζε από παντού, πλημμύριζε με τη βροχή, κι ο Μπάμπης σταμάταγε τους καφέδες, τα τσάγια και τα χαμομήλια και[...] στέγνωνε το πάτωμα εξαπολύοντας τσουβάλια πριονίδι[...] (HNC). ‘‘Vyzantion’ had a muddy floor, and water used to come in from everywhere, so Babis would stop boiling the coffees, the teas and the chamomiles and […]dried the floor by throwing bags of sawdust[…]’.

Example (5) is a ‘Part for Whole’ metonymic configuration as well as an Event ICM. It is a ‘Part for Whole’ metonymy since only the unbounded liquid substance is denoted instead of the scene of the preparation and serving of the drinks in their containers. The content, such as the coffee or the tea substance, metonymically leads to the preparation of three different kinds of drinks and their serving in containers, such as a cup of coffee or a cup of tea or a tea- pot, and results in building an Event ICM. The action verb ‘stop’, which refers to

Syntax - Morphology - Semantics 2 3 9 Major Trends in Theoretical and Applied Linguistics

the ceasing of the preparation of drinks, enhances the sense of an event taking place, making the Event ICM more prominent. Nevertheless, whenever the speaker chooses to direct his or her attention to the individual food and drink nouns, and not on the entire event, the Containment ICM is being activated, as discussed below.

5.2. Containment ICM (Radden and Kövecses 1999: 41)

In the containment ICM, what is of great importance is the notion of a container that functions as the vehicle through which we mentally access the respective substance, that is the target. Langacker (1991: 30) claims that “the greater the conceptual contrast between vehicle and target, the better is the relationship to be exploited metonymically”. Content for container: The milk tipped over (Norrick 1981: 58). Containment ICM: Someone is shopping in a supermarket when, by accident, the milk container tips over (Norrick 1981: 58). Content: milk Container: milk carton (implied)

(6) [….]θα φέρει κάποια υδατική ισορροπία μεταξύ Δυτικής Ελλάδας, η οποία έχει πολλά νερά και λίγες πεδιάδες, και Ανατολικής, η οποία έχει περισσότερους κάμπους και λίγα νερά (HNC). ‘[….] it will bring a kind of water balance between Western Greece, an area with many waters and few plains, and Eastern Greece, an area with much more plains and few waters’.

In (6), the singular non-count noun water (‘nero’) is used in a plural count- noun sense, namely, waters (‘nera’). It is a ‘Whole for Part’ metonymy because the substance here acquires its generic sense. The reference is the element of water, which is the Whole, stands for a limited amount found in lakes or rivers. Thus, lakes and rivers is the implied Part. Additionally, the ‘Content for Container’ metonymy motivates the use of the plural, that is waters (‘nera’). Content is the water while Container is a lake or a river. The ‘Content for Container’ metonymic model (Radden and Kövecses 2009: 41) is usually employed where water as an unbounded physical substance metonymically extends to the bounded water containers, such as lakes or rivers.

(7) Ο κανονισμός της Ευρωπαϊκής Ένωσης προβλέπει ότι για να δοθεί επιδότηση πρέπει να θεριστούν τα σιτάρια μέχρι τις 30 Ιουνίου ή να συντηρηθούν μετά τις 30 Ιουνίου (HNC).

2 4 0 Syntax - Morphology - Semantics Ifigeneia Athanasiadou and Martha Lampropoulou

‘The regulation of the European Union stipulates that in order to grant a subsidy, the wheats should be harvested till the 30th of June or stored and kept after the 30th of June’.

In (7), we could schematically be transferred to a scene of fields where wheat is cultivated. By using the singular substance-mass noun wheat (‘sitari’) as a plural-count noun, namely, wheats (‘sitaria’), we metonymically refer either to the plurality of fields seeded with wheat or to the wheat kernels that need to be harvested. Thus, the use of the plural activates two ICMs: 1) The first involves the number of fields that could be visualized inthe speaker’s mind. This is a ‘Whole for Part’ metonymy: the Whole is the fields seeded with wheat, and the Part is the wheat kernels. Apart from being a ‘Whole for Part’ metonymy, it is also a ‘Container for Content’ ICM: container-fields, content-wheat kernels. The speaker illustrates the harvesting of the fields that in fact stands for the harvesting of the wheat kernels. 2) The second metonymic mapping is observed when the discrete kernels of wheat, and not the plurality of fields, are visualized by the speaker. A ‘Part for Whole’ metonymy and the ‘Content for Container’ ICM are constructed: content- wheat kernels, container-fields. The speaker refers directly to the kernels of wheat. Overall, in example (7), the speaker determines what he needs to refer to, either the Content (wheat) or the Container (fields). As observed in (6) and (7), the ‘Content for Container’ metonymically leads to different kinds of bounded containers. It is also worth noting that the singular substance-mass noun arakas (peas) has no grammatical plural form (? arakades). However, arakas could also be pluralized, motivated by the appropriate Event ICM or ‘Content for Container’ ICM, when referring to packs of the product used for its containment:

(8) Πού έβαλες τους αρακάδες; ‘Where did you put the peas?’

However, there are types of physical substances where the ‘Content for Container’ ICM is not so frequently applied. For instance, halvas and jelly (‘zeles’), two types of sweets are not that easily conceptualized in a kind of container due to their thick liquid nature:

(9) Λαϊκοί οργανοπαίκτες, τραγουδιστές και χορευτικά συγκροτήματα απ’ όλη την Ελλάδα και φυσικά λαγάνες, χαλβάς, ελιές, φασολάδα και άφθονο κρασί (HNC). ‘Folk musicians, singers and dancing groups from all over Greece and, of course, laganes, halvas, olives, beans soup and plenty of wine’.

Syntax - Morphology - Semantics 2 4 1 Major Trends in Theoretical and Applied Linguistics

(10) Τέλος, βάζουμε φέτες ροδάκινο πάνω από την τελευταία στρώση κρέμας και ρίχνουμε και το ζελέ (Google). ‘Finally, we place slices of peach on top of the last cream layer and we pour the jelly’.

However, both nouns can be encountered in the plural, even if the substances denoted are mostly conceptualised as a uniform whole. For instance:

(11) Έβαλες τους ζελέδες στο ψυγείο; ? ‘Did you put the jellies in the fridge?’

(12) Φτιάχνουμε διάφορους χαλβάδες, σκέτους, με μύγδαλο… ? ‘We produce various halvades, pure, with almond…’

In (11) and (12), the ‘Content for Container’ ICM is being used. The relevant containers, enclosing jellies or halvades, such as bowls or pots, are being backgrounded. Instead, the uniform mass substance is being foregrounded. It has to be noted here that halvas and zeles (jelly) are not real Greek words; they are, actually, loan words. Yet, despite their origin and their non-count nature, they can receive the plural inflection in the ‘Content for Container’ ICM.

5.3. Category and Member ICM (Radden and Kövecses 1999: 34)

Radden and Kövecses exemplify this type by means of the following examples:

‘Generic for Specific’: “Boys don’t cry” ‘Specific for Generic’: “A/ The spider has eight legs”

The authors elaborate on the first example sentence by explaining that the statement about the boys can be used in a particular occasion when a boy is crying and, thus, the statement is perceived specifically. However, in the second example sentence, the authors claim that “specific tokens may be used to stand for generic types” since “the definite article the or the indefinite a are used to refer to spiders in general” (Radden and Kövecses, 1999: 34). In the domain of food and drinks, the type ‘Specific for Generic’ can be observed when reference is made to a specific plant or vegetable which stands for a type of drink or food extending to an entire category with certain properties. The use of the singular is mostly employed in order to refer to these types of drinks or foods having specific properties, as illustrated in the following examples:

2 4 2 Syntax - Morphology - Semantics Ifigeneia Athanasiadou and Martha Lampropoulou

(13) Το τσάι είναι πλούσιο σε βιταμίνες C, B, και D (Epistimi kai Zoi). ‘Tea is rich in vitamins C, B, and D’.

Generic: The statement is valid for every tea-plant. Specific: The utterance becomes specific when spoken at a specific moment in time for a specific tea-plant or drink representing every-tea plant or tea-drink.

(14) Η ξεφλουδισμένη φακή είναι αρκετά θρεπτική για τον άνθρωπο (Anthanassa) ‘Peeled lentils are very nutritional for man’.

(15) Tο φασόλι δεν είναι ανάγκη να λιπαίνεται με αζωτούχα χημικά λιπάσματα (Epistimi kai Zoi). ‘Beans do not need nitrogenous chemical fertilizers’.

(16) Η βρώμη: ενισχύει το νευρικό σύστημα και ελέγχει τις λειτουργίες του (HNC). ‘Oats: they boost the neural system and control its functions’.

From a semantic point of view, Wierzbicka (1985: 329) underlines that although beans is a plural-mostly noun and oats is a plural-only noun, both types of nouns are conceptualized only in a collective, not in an individual, sense. In Modern Greek, the semantic meaning of the noun ‘faki’ (lentils), bearing a collective sense, remains the same, either in the singular, namely, faki (lentils), or in the plural, that is fakes (lentils). Fasoli (bean) can also have a plural form, namely, fasolia (beans), unlike vromi (‘oats’), which cannot have a plural form (? vromes). The physical size of the real-world referents of the relevant nouns could explain this observation since individual beans are bigger than oat grains (Wierzbicka 1988: 535). In examples (14)-(16), the ‘Category and Member ICM’ is projected. Each one of the relevant nouns, namely, faki (lentils), fasoli (bean), and vromi (oats), is employed in the singular to refer to a distinct type of food. These nouns function as a member of a category highlighting beneficial health effects, (14) and (16), or the required conditions for cultivating the plant (15).

5.4. Production ICM (Radden and Kövecses 1999: 39)

In the particular type of metonymic relationship, the producer, be it the owner of a business or a brand name, usually replaces the product itself. In such cases, the brand name or the business owner is famous and widely known. Furthermore, the brand name is encountered mostly in the singular, which is retained even if the speaker implies a number of products of this brand.

Syntax - Morphology - Semantics 2 4 3 Major Trends in Theoretical and Applied Linguistics

‘Producer for Product’ ICM: Agoazo agno me 1,20 to litro. ? ‘I buy Agno with 1,20 per litre’. ‘Producer for Product’ ICM: The producer is the dairy products company ‘Agno’, which metonymically stands for its product, namely, milk. Producer: The dairy products company ‘Agno’ Product: milk

(17) Γέμισε το ψυγείο “Υφαντή” (commercial). ? ‘Fill the refrigerator with ‘Ifanti’’.

(18) Αγόρασε και ένα “ΝΟΥΝΟΥ”. ? ‘And, also, buy a ‘NOUNOU’ ’.

In (17), Ifanti, which is the producer’s or the brand’s name, metonymically triggers either a specific type i.e. turkey fillet or other types of the product i.e. ham or sausages. The specific quantity required is not explicitly mentioned and the use of the singular implies an unbounded mass quantity of the product. Once more in (18), the ‘Producer for Product’ conceptual metonymy is in use. However, in this case, the brand name ‘NOUNOU’, which is the producer and is preceded by the numeral ena (one), metonymically stands for a specific product. For Radden and Kövecses (1999: 39) the ‘producer’ can either be replaced by the brand’s name (‘NOUNOU’) as in (18) or the producer’s name (‘Ifanti’) as in (17).

(19) Αγόρασε δύο “NOYNOY”. ? ‘Buy two ‘NOUNOU’’.

‘NOUNOU’, cannot receive a plural inflection, even if we actually refer to two cartons or bottles of milk. This is often observed, especially when the brand’s or the manufacturer’s name is not in Greek, and hence the noun remains in the singular. Even though the brand name remains in the singular, the use of the number quantifier dio (two) extends to two instances of the same product.

6. Concluding Remarks

In all of the examples presented in the paper, there is a scenario, or else, a frame of eating or drinking, which is a primitive and non-culturally specific human experience. The types of food and types of drinks exemplified in the study can either be treated as count or non-count according to the Idealized Cognitive Model employed in the relevant context.

2 4 4 Syntax - Morphology - Semantics Ifigeneia Athanasiadou and Martha Lampropoulou

More specifically, the Event ICM and the ‘Content for Container’ ICM seem to motivate both the use of the singular and the plural regardless of the fact that some of the foods and drinks constitute discrete particles while others form a homogeneous mass. For instance, concerning the Greek food nouns fasolia (beans) and faki (lentils), there is not a particular grammatical number preference since both the singular and the plural forms can well be used interchangeably, hence fasoli-fasolia and faki-fakes. Similarly, turning to the domain of drinks, types of drinks, such as nero (water), and kafes (coffee), undoubtedly constitute a continuous mass, yet, when the frame or else, the ICM calls for a container, then, they can be perfectly enumerated giving rise to the use of plural inflection, hencenero-nera and kafes- kafedes. However, the ‘Specific for Generic’ ICM, which is a subtype of the ‘Category and Member’ ICM, and the ‘Producer for Product’ ICM seem to favor mostly the use of the singular, as shown in the analysis of the example utterances, “see Section 5.3” and “Section 5.4”, respectively. We cannot overlook the fact that linguistic categories influence non-linguistic behavior, such as non-verbal thought processes. According to the Sapir-Whorf Hypothesis, language can influence thought and underlie our conceptualizations (Lucy and Gaskins 2003). Speakers across cultural groups may construe and use a noun as count or non-count, depending on how their language encodes the real-world referent of the relevant noun. It appears that, not only across cultures but also within a cultural group, speakers display countability preferences based on criteria of functionality and economy of language use in accordance with a certain type of ICM. Speakers of Modern Greek use a noun as count, arakades (peas), or non-count, arakas (peas), according to the functional needs of the specific utterance context. All of the example nouns from the domain of food and drinks examined in the study were shown to be encountered in both the singular and the plural in a certain type of ICM. This fact reveals that Greek speakers are flexible enough in noun countability shifts and that they can accommodate the noun’s countability status to the corresponding utterance context. Metonymy appears to be an efficient means of communicating meaning since, especially in the domain of food and drinks, it achieves accuracy, clarity, relevance, and economy of use, at the same time2.

2 We would like to thank Kyriaki Kourouni for the reviewing of the example sentences presented in the paper and Angeliki Athanasiadou for her invaluable advice and support.

Syntax - Morphology - Semantics 2 4 5 Major Trends in Theoretical and Applied Linguistics

References

Andriotis, N.P. 1995. Etymologiko leksiko tis koinis neoellinikis [Etymological Dictionary of Standard Modern Greek]. 3rd edition. Aristotle University of Thessaloniki: Institute for Modern Greek Studies.

Anthanassa. Retrieved 20 August 2012 from

Athanasiadou, I. 2009. Count and Non-count Nouns in Greek and in English. A thesis submitted for the degree of MA by Dissertation. University of Essex. United Kingdom.

Croft, W. 1993. The role of domains in the interpretation of metaphors and metonymies. Cognitive Linguistics 4(4): 335-370.

Egkiklopedia Epistimi kai Zoi [Encyclopedia ‘Science and Life’]. Vol. 18. Chatziiakovou (ed.). Thessaloniki and Athens: Chatziiakovou. 83,194.

Geeraerts, D. 2006. Cognitive Linguistics: Basic Readings. Berlin: Mouton de Gruyter.

Hellenic National Corpus (HNC). Retrieved 10 March 2011 from < http://hnc.ilsp.gr>

Hirtle, W.H. 1982. Number and inner space: A study of grammatical number in English. Quebec: Les Presses de l’Université Laval.

Lakoff, G. and Johnson, M. 1980. Metaphors we live by. Chicago: The University of Chicago Press.

Lakoff, G. 1987. Women, Fire, and Dangerous Things. Chicago: The University of Chicago Press.

2 4 6 Syntax - Morphology - Semantics Ifigeneia Athanasiadou and Martha Lampropoulou

Langacker, R.W. 1987. Foundations of Cognitive Grammar, Vol. I. California: Stanford University Press.

Langacker, R.W. 1991. Foundations of Cognitive Grammar, Vol.II. California: Stanford University Press.

Langacker, R.W. 1993. Reference-point constructions. Cognitive Linguistics 4(1): 1-38.

Lucy, J.A. and Gaskins, S. 2003. Interaction of Language Type and Referent Type in the Development of Nonverbal Classification Preferences. In D. Gentner and S. Goldin-Meadow (eds.). Language in Mind: Advances in the Study of Language and Thought. Cambridge MA:.MIT Press, 465-492.

Norrick, N.R. 1981. Semiotic Principles in Semantic Theory. Amsterdam / Philadelphia: John Benjamins Publishing Company.

Online Etymological Dictionary. Retrieved 13 March 2011 from

Panther, K.-U. and Thornburg, L. 1999. The Potentiality for Actuality Metonymy in English and Hungarian. In K.-U. Panther and G. Radden (eds.). Metonymy in Language and Thought. Amsterdam/ Philadelphia: John Benjamins Publishing Company, 333-357.

Panther, K.-U. and Thornburg, L. 2009. On figuration in grammar. In K.-U. Panther, L.L. Thornburg, and A. Barcelona (eds.). Metonymy and Metaphor in Grammar. Amsterdam/Philadelphia: John Benjamins Publishing Company, 1-44.

Radden, G. and Dirven, R. 2007. Cognitive English Grammar. Amsterdam/ Philadelphia: John Benjamins Publishing Company.

Radden, G. and Kövecses, Z. 1999. Towards a theory of Metonymy. In K.-U. Panther and G. Radden (eds.). Metonymy in Language and Thought. Amsterdam/ Philadelphia: John Benjamins Publishing Company, 17-59.

Taylor, J.R. 1995. Linguistic Categorization. New York: Oxford University Press.

Tsoulas, G. 2007. On the grammar of number and mass terms in Greek. Retrieved 23 March 2011 from

Syntax - Morphology - Semantics 2 4 7 Major Trends in Theoretical and Applied Linguistics

Wierzbicka, A. 1985. Oats and wheat: The fallacy of arbitrariness. In J. Haiman (ed.). Iconicity in Syntax. Amsterdam/Philadelphia: John Benjamins Publishing Company, 311-340.

Wierzbicka, A. 1988. The semantics of grammar. Amsterdam/Philadelphia: John Benjamins Publishing Company.

2 4 8 Syntax - Morphology - Semantics Basic Illocutions of the Modern Greek Subjunctive

Maria Chondrogianni

University of Westminster [email protected]

Abstract

In this paper we present the basic illocutions associated with the Modern Greek Subjunctive. The interface between Morphosyntax, Phonology and Pragmatics is very important in this work: each basic illocution is described in terms of prosodic contour (with relevant Praat illustrations); the optional or necessary presence of the associated negation μη(ν); the use of segmental markers, which provide cues on how a certain utterance is to be interpreted; grammatical tense restrictions, where appropriate, including number and person restrictions; aspectual restrictions, where appropriate; as well as the potential answer provided by an addressee to a question, or a question-like utterance.

1. Introduction

In this paper we discuss the basic illocutions associated with the Modern Greek (MG) Subjunctive mood which form part of the grammar system. Our aim is to provide a systematic description of the basic illocutions of MG, as expressed in main clauses, based on their formal characteristics i.e. based on distinct markers that have an illocutionary impact. Markers considered include the verb mood; the relevant negation particle; the prosodic contour; the number, person and tense where applicable; the addressee’s response, where applicable; and any additional segmental markers that apply. Our primary objectives, as part of a wider research on the basic illocutions of the MG system, included the specification of the relationship between form (e.g. grammar, verb mood, prosodic contour) and function (basic illocution); and the application of the basic illocution systematic hierarchical classification of propositional and behavioural basic illocutions, as proposed by Hengeveld et al. (2007) in MG. Our secondary objectives involved establishing the formal organisation of the MG verb mood system; as well as establishing the MG primary intonation patterns that operate at the level of utterance.

Syntax - Morphology - Semantics 2 4 9 Major Trends in Theoretical and Applied Linguistics

The methodology we followed included the identification of morphosyntactic tools speakers have at their disposal to express their intention i.e. establishing the MG verb moods through a series of tests involving morphology; use of particles; negation; clitic placement; and participation to subordination. In addition, it included the identification of phonological tools speakers have at their disposal to express their intention. In order to establish the MG intonation patterns, we considered different approaches in MG Phonology. One of these approaches was GR ToBi (Arvaniti and Baltazani 2006, accessible at http:// idiom.ucsd.edu/~arvaniti/grtobi.html), a tool for the intonational, prosodic and phonetic representation of Greek spoken corpora, designed to capture Athenian Greek and focusing on a prosodic analysis of phrase based structures. We also considered approaches aiming to explore the relationship between intonation and sentence type interpretation (from a production and perception point of view) such as Kotsifas (2009) and Chaida (2008). Our original research hypothesis included 6 intonation patterns, as outlined in Table 1 below. Note that the adopted intonation patterns were later revised and reduced to 5.

Table 1. MG intonation patterns as part of our original methodology

Intonation Additional Final Utterance Type Pattern Name characteristics Boundary Assertions INT1 Broad focus Low Directives? Assertions Narrow focus INT2 Low Directives? Alternative to INT1 Content INT3 High Interrogatives Polar INT4 High-Low Interrogatives Directives? INT5 Low-High Curses Assertions-in- disguise (Rhetorical INT6 Questions)? ? Miratives? Wishes?

Our data were collected introspectively, drawn from the linguistic intuitions of the author and compared against examples from reference grammars. The examples were also checked by an informal group of informers (6 male and 6 female, aged 12-72, based in different geographical areas of Greece, including 5 male and 4 female from Athens, Thessaloniki, Crete and Patras) and members of the Greek diaspora (1 male and 2 female based in London and in New York, USA). The informers were contacted on a regular basis over a period of 9 months, (and less frequently towards the final stages of this work), and were

2 5 0 Syntax - Morphology - Semantics Maria Chondrogianni

individually asked to check the provided examples against their intuition. The sets of examples (covering all MG moods) were separated by mood and function, while brief scenarios were offered for specific cases (e.g. miratives). We were interested in the informants’ production instinct (rather than their perception): although for an illocution to be effective it needs to be recognized by an addressee, the addressee might not pay the necessary attention, might not want to recognize the illocution or might lack necessary background information to do so (Levelt 1989). In that sense, illocution is a speaker-centered notion (Levelt 1989, p. 59). Furthermore, sets of examples were compared to internet-based uses. The internet is an easily accessible source of linguistic expressions in use; it is very common for linguists to create web-based informal mini-corpora of specific linguistic phenomena. For example, the interchangeable distribution of μη with and without a final ‘v’ was checked against an internet–based sample of prohibitive uses. For the mini-internet searches, blogs and chat groups were mainly accessed, where the language used is closest to the way current MG users speak; particles and segmental markers were used as the main key-words. In addition, during the evaluation process, we explored the prosodic contour of instances of uses in each grammatical mood. Prosodic contour, as mentioned earlier, is one of the criteria that allow us to establish a particular illocution. Using a single speaker (the author), we performed a production experiment; by recording examples for each grammatical mood in studio conditions, using a laptop and a portable microphone, we took advantage of Praat’s mono recording tool. 85 examples were recorded in total (some in multiple versions) which were edited and analysed using Praat. The evaluation of the results, i.e. the comparison of the Praat illustrations of intonation patterns across different uses and verb moods, made us revise our research hypothesis (and reject the possible 6th intonation pattern originally considered for assertions in disguise, miratives and wishes) and helped us establish the 5 distinct intonation patterns which formally contribute to distinguish different MG illocutions. This approach allowed us to fulfill one of our objectives, i.e. to relate Pragmatics with Phonology, as both form part of the grammatical component. Each grammatical mood, once identified and established, was analysed considering its propositional and behavioural functions. The results were then classified according to their function. The term grammatical mood is used in this work as the category which includes “all grammatical elements operating on a situation/proposition, that are not directly concerned with situating an event in the actual world, as conceived by the speaker” (Hengeveld 2004). Basic illocution (also Sentence Type, or Speech Act prototype) is understood as “a coincidence of grammatical structure and conventional conversational use” (Sadock & Zwicky 1985). Basic illocutions are expressed by the speaker in various forms, using syntactic, morphological and phonological means.

Syntax - Morphology - Semantics 2 5 1 Major Trends in Theoretical and Applied Linguistics

Below we discuss the MG Subjunctive and its propositional and behavioural functions.

2. The MG Subjunctive

Subjunctive is taken to mean a particular grammatical mood with its own semantic identity (modal value, modality). As modality does not form part of this research1, we define the Modern Greek Subjunctive as a distinct morphosyntactic category by the presence of the particle να (see also Chondrogianni 2009). Subjunctive exhibits a richness of uses, hence a one-to-one relationship between the form and function cannot be established. Following Hengeveld et al. (2007), the MG Subjunctive used we describe below have been distinguished into propositional and behavioural uses. Propositional uses are associated with assertive and questioning illocutions. Assertive subtypes consist of declarative and mirative uses, whilst questioning subtypes consist of polar and content interrogatives. Behavioural uses involve speech acts that intend to influence or affect the behaviour of the Addressee and/or others. Behavioural (positive and negative) uses include imperative subtypes (orders), hortative subtypes (exhortations), admonitive subtypes (warnings) and supplicative subtypes (requests for permission). As we show below, the MG Subjunctive’s propositional uses consist of wishes, curses, miratives- expressions of disapproval, wondering and expression of uncertainty. Subjunctive exhibits a rich variety of behavioural uses: the particle να is often explicitly used in order to mitigate the illocutionary force of an utterance; it introduces mitigated directives, mitigated prohibitions and mitigated requests, including supplicatives/ requests for permission.

3. Propositional Uses of the MG Subjunctive

3.1. Introduction

In this section we discuss propositional subjunctive uses, including wishes and curses, wondering and expression of disapproval.

1 Modality involves the “modification of the content of speech acts” (Hengeveld 2004). This research concentrates on “identifying sentences as specific types of speech acts” (i.e. the illocution, as defined by Hengeveld 2004), rather than the modification of the content of speech acts.

2 5 2 Syntax - Morphology - Semantics Maria Chondrogianni

3.2. Wishes

Wishes in subjunctive denote uses which are not meant to influence the addressee’s behaviour. They express the Speaker’s desire for a particular state of affairs (which might or might not already be the case) for which the Speaker wishes to happen or to be extended in the future. In that sense, wishes might be possible to be fulfilled, or are currently unfulfilable. In addition, wishes might involve fixed expressions, i.e. wishes that are usually expressed in a formulaic way at ceremonial events including weddings, christenings, anniversaries, funerals etc. In some such cases the verb might be omitted; if it is included in the utterance, though, it is always in subjunctive. Wishes share the same intonation pattern with assertions2; they are uttered using an intonation pattern INT1 on INT2 (see also Chondrogianni 2012), as part of a schematic intonation pattern illustration we have adopted. INT1 is characterised by broad focus and a high level of the accented syllable (although the nucleus might create variations of this pattern). When narrow focus and a rise from low to the accented syllable applies, as is the case of wishes introduced by μακάρι (see 3.2.1 below), they are uttered in INT 2 (as focality affects the way the utterance is expressed.).

(1) Να τoν βλέπαμε συχνότερα! SUBJ him see-1P.PS.IPF more often If only we saw him more often!

Example (1), in the first person plural, expresses a wish for something that is currently not the case (irrealis), with a past imperfect subjunctive used. A Praat illustration of its prosodic contour is presented below. Example (2) is a typical example of an MG wish3; it expresses self-exhortation. Such utterances are usually expressed in the 1st person singular, in past imperfect tense. It might also be used as a condition if followed by a result, where ‘να’ would have a conditional function. Conditionals do not form part of this research.

(2) Να ήμουν πλούσιος! SUBJ was-1SG.PS.IPF rich Were-I rich (If only I were rich!).

2 We recognise that INT1 assertions usually finish with a fall; also, we have observed INT2 assertions finish with a rise. Example (1) ends with a rise and gives the impetus for further research in the wishes’ category. 3 Wishes might also be expressed in Hortative, introduced by ας. See also Chondrogianni (2009).

Syntax - Morphology - Semantics 2 5 3 Major Trends in Theoretical and Applied Linguistics

0.99

-0.9591 0.845 2.474 Time (s)

500

300 vlepame

) ton z Na syxnotera

H 200 (

h 150 c t i

P 100 70 50 0.845 2.474 Time (s) Figure 1. PRAAT illustration of wish (INT1)

Examples (3) and (4) illustrate some fixed expressions of wishes; (3) may typically be uttered in a religious setting as a good-will wish, for example as an expression of gratitude to a benefactor’s family. Example (4) is usually addressed to the parents of a child, at birth, christening or other important event.

(3) Nα συχωρεθούν τα πεθαμένα σας! SUBJ forgive-3PL.PRF.PASS the dead your May the dead members of your family be forgiven.

(4) Να σας ζήσει! SUBJ to you live-3SG.PRF May (your child) live long.

3.2.1. Additional Segmental Marking

Wishes might be marked by special particles, such as μακάρι, which offers the addressee a clue as to how the particular utterance is to be interpreted. When

2 5 4 Syntax - Morphology - Semantics Maria Chondrogianni

uttered, the focal point is on the segmental marker (INT2), as we can also see in fig. 2 below. Such wishes might be fulfilable, now or in the future as in example (5) below, or unfulfilable.

(5) Μακάρι να γίνει καλά. WISH SUBJ become-3SG.PRF well I wish he/she gets well.

0.99

-0.8861 0.884 2.314 Time (s)

500

300 Makari ) z na

H 200

( yini

kala

h 150 c t i

P 100 70 50 0.884 2.314 Time (s)

Figure 2. PRAAT illustration of a wish using the segmental marker μακάρι (INT2).

3.2.2. Summary of the Characteristics of the Basic Illocution Wish

Wishes in MG are expressed in Subjunctive (or in Hortative), marked by the Subjunctive particle να (and the optional negation μη(ν )) and the intonation pattern INT1. When introduced by μακάρι, the segmental marker is the focal point of the utterance (INT2). Any person or number might be used, while aspectual differences affect the fulfilability of unfulfilability of a wish.

Syntax - Morphology - Semantics 2 5 5 Major Trends in Theoretical and Applied Linguistics

Table 2. Summary of the characteristics of the basic illocution wish

Type Propositional

Function Wishes Subjunctive (particle να, optional negation μη(ν), optional Grammatical Mood segmental marker μακάρι) (also Hortative ας ) Present (fulfilable) Tense Past (unfulfilable) Imperfective (Present, Past) Aspect Perfective (Present) Person 1st, 2nd and 3rd

Number Singular or Plural

Intonation Pattern INT1 (INT2 when introduced by μακάρι) Addressee’s response N/A

3.3. Curses

Curses are a form of negative wish. In main clauses they might be expressed in any number or person, most often using perfective aspect, as we can see from examples (6), (7) and (8).

(6) Να μην ξημερωθείς. SUBJ NEG ‘see the day downed’-2SG.PR.PRF.PASS You may not see another day.

(7) Να καoύν στην κόλαση. SUBJ burn-3PL.PR.PRF.PASS in hell May they burn in hell.

(8) Να μη δει ξανά άσπρη μέρα. SUBJ NEG see-2SG.PR.PRF again white day May he not experience again a happy day.

3.3.1. Additional Segmental Marking in Curses

The use of the segmental marker που, followed by a Subjunctive, adds a temporary value of immediacy to a curse. This is the case of examples (9) and (10) below.

(9) Που να μη σε είχα συναντήσει ποτέ. UNWISH SUBJ NEG you have-1SG.PS met never May I had never met you.

2 5 6 Syntax - Morphology - Semantics Maria Chondrogianni

(10) Που να σπάσεις το πόδι σου! UNWISH SUBJ break-2SG.PR.PRF the leg your Break your leg!

Curses are marked by intonation pattern INT5. This pattern starts with a small fall, followed by a rise (and possibly a high plateau), and followed by a fall (and a potential small rise at the end). The boundary is low-high. INT 5 is illustrated in Fig. 3 below, where we observe a small fall after ‘pou’, a rise after ‘na’ with a high plateau, a fall following the first (accented) syllable of pόdi‘ ’ and a minor rise at ‘sou’.

0.7314

-0.99 0.7579 2.162 Time (s)

700 500 ) z 300 spasis to podi H ( Pou h na

c 200

t sou i

P 150 100 70 0.7579 2.162 Time (s)

Figure 3. PRAAT illustration of a cursed marked by που (INT5).

3.3.2. Summary of the Characteristics of the Basic Illocution Curse

Curses are marked by the distinct intonation pattern INT5 and the optional use of the segmental marker που. They occur in Subjunctive, introduced by the particle να, and the optional negation μη(ν). Present tense with perfect aspect characterise their most common uses, as well as 2nd and 3rd person. In the first person they are similar to an oath, often with a complement sentence provided as a means of context. Their characteristics are summarised in Table 3 below.

Syntax - Morphology - Semantics 2 5 7 Major Trends in Theoretical and Applied Linguistics

Table 3. Summary of the characteristics of the basic illocution curse

Type Propositional

Function Curses (Negative Wishes) Subjunctive (particle να, optional negation μη(ν), optional Grammatical Mood segmental marker που) Tense Present Perfective Aspect [ Imperfective not excluded but uncommon] Person 2nd and 3rd [1st possible, ‘oath’]

Number Singular or Plural

Intonation Pattern INT5

Addressee’s response N/A

3.4. Miratives of Disapproval

When a speaker expresses his/her admiration, surprise or delight through an utterance, effectively they share some information with their addressee. In this type of utterances, it is not so much the content of the utterance itself that is being transmitted, but rather the emotional reaction of the speaker with respect to this content. (Hengeveld et al. 2007). Utterances in this category demonstrate a mixture of declarative and interrogative properties. Approval (positive surprise) is usually expressed in indicative; mirative utterances in subjunctive usually express the speaker’s disapproval towards the addressee’s taste or choice among others, as in example (11) below.

(11) Να φορέσεις σορτς στο γάμο! Πού ακούστηκε! SUBJ wear.2SG.PRF shorts to the wedding! (Where was heard!) To wear shorts at the wedding! This is unheard of!

3.4.1. Summary Characteristics of Miratives of Disapproval

Miratives of disapproval are expressed in Subjunctive, introduced by the subjunctive particle να, and optionally by the negation μη(ν), usually in Present Perfect; they are marked by intonation INT3, a pattern typical for, but not restricted to, content interrogatives. Their characteristics are summarised in Table 4 below.

2 5 8 Syntax - Morphology - Semantics Maria Chondrogianni

Table 4. Summary characteristics of the basic illocution mirative

Type Propositional

Function Mirative uses [Indicative: approval] Grammatical Mood Subjunctive (disapproval, particle να, optional negation μη(ν)) Present Tense [Past and Future possible for Indicative] Perfective Aspect [ Imperfective not excluded] Person 2nd and 3rd [1st possible]

Number Singular or Plural

Intonation Pattern INT3

Addressee’s response N/A

3.5. Wondering

In most languages subjunctive has a built-in uncertainty/doubt element. Examples (12)-(14) below present question-like utterances, signalling to the addressee the speaker’s doubt, uncertainty or speculation, often also expressed as self-directed expressions where the addressee genuinely does not know the answer. Such expressions might appear as polar or content interrogative-like, where no response is expected from the addressee.

(12) Να βρέχει; SUBJ vrehi-3SG.PR.IPF Is it raining (I wonder)?

Wondering is marked by the combination of the subjunctive with intonation pattern INT4, a pattern also shared with polar interrogatives (which can only be expressed in indicative, where the addressee is expected to confirm the truth of the proposition). Following a gradual fall, we observe a low plateau followed by a rise, with a rise-fall boundary4. In examples (13) and (14) below, the speaker’s wondering is expressed through a deliberative question, in the first person plural and third person

4 Although the rise-fall boundary might not be evident in Fig. 4, its overall picture is consistent with other INT4 examples observed (see also Chondrogianni 2012).

Syntax - Morphology - Semantics 2 5 9 Major Trends in Theoretical and Applied Linguistics

0.99

-0.6384 0.8375 1.822 Time (s)

700 500 )

z 300 ( H

c 200 t Na i vrehi P 150

100 70 0.8375 1.822 Time (s)

Figure 4. PRAAT illustration of wondering in subjunctive (INT4).

singular respectively, in a content interrogative-like utterance marked by INT3 intonation.

(13) Τι να κάνουμε; What SUBJ do.2PL.IPF What shall we/can we do?

(14) Τι να έγινε ο Γιάννης; What SUBJ become.3SG.IPF the Yannis? What might have happened to Yannis?

3.5.1. Additional Segmental Marking in Wondering

A speaker might strengthen the wondering element of their utterance through the use of the segmental marker άραγε, as in example (15), which, similarly to the uncertainty marker ίσως below, can combine with either the indicative or the

2 6 0 Syntax - Morphology - Semantics Maria Chondrogianni

subjunctive. The choice of mood is guide by modal criteria; through the use of subjunctive the speaker is less inclined to believe the possibility of the truth of the content of the clause (irrealis). The segmental marker might be placed at the beginning of the utterance, or at the end.

(15) Άραγε να βρέχει; WOND SUBJ rain-3SG.PR.IPF May be raining, I wonder?

3.5.2. Summary Characteristics of the Basic Illocution Wondering

Wondering in MG is expressed either in Subjunctive with or without the use of a particular wondering marker (it can also be expressed in Indicative with compulsory use of the marker άραγε), marked by intonation INT4 and the use of 3rd person, with 1st person deliberative uses also noted. Their characteristics are summarised in Table 5 below.

Table 5. Characteristics of the basic illocution wondering

Function Wondering [Indicative (segmental marker άραγε, optional particle θα, optional negation δεν)] Grammatical Mood Subjunctive (particle να, or combination of segmental marker άραγε and να, optional negation μην) Scope Propositional

Tense Present/Past

Aspect Perfective/Imperfective 3rd (1st for deliberative questions) Person 2nd person acceptable for indicative uses only Number Singular or Plural

Intonation Pattern INT4

3.6. Expression of Strong Sense of Uncertainty in Subjunctive through the Use of the Segmental Marker ίσως

A speaker might chose to enhance the built-in uncertainty in subjunctive through the use of the marker ίσως, as shown in example (16).

Syntax - Morphology - Semantics 2 6 1

Major Trends in Theoretical and Applied Linguistics

(16) ΄Ισως να έφυγε. UNC SUBJ leave-3SG.PR.PRF Perhaps he left.

Such uses are marked by the use of intonation pattern INT1, also used in assertions, with a Low boundary, illustrated in Figure 5 below. Note that when ίσως is used in an indicative uncertainty construction, it acts as a focal point and thus the intonation pattern INT2 applies.

0.9899

-0.99 1.091 2.178 Time (s)

700 500 )

z 300 Isos

( H na

h efiye

c 200 t i P 150

100 70 1.091 2.178 Time (s) Figure 5. PRAAT illustration of reinforced uncertainty.

3.6.1. Summary of the Basic Illocution of Uncertainty, Introduced by ίσως

In many ways, wondering in subjunctive, as described in section 3.5 above, involves to some extent the speaker’s uncertainty about the validity of the described State of Affairs. Explicitly denoted uncertainty is marked bythe particle ίσως, which, in subjunctive, always takes a fixed position preceding the subjunctive να. Its characteristics are summarised in Table 6 below.

2 6 2 Syntax - Morphology - Semantics Maria Chondrogianni

Table 6. Characteristics of the basic illocution uncertainty.

Function Expression of uncertainty [Indicative (segmental marker ίσως, optional particle θα, optional negation δεν; segmental marker usually precedes the verb, but Grammatical Mood position after the verb acceptable)] Subjunctive (segmental marker ίσως in fixed pre-verb position, particle να, optional negation μην) Scope Propositional

Tense Present/Past [Future in indicative acceptable by some speakers]

Aspect Perfective/Imperfective

Person Any (3rd very common)

Number Singular or Plural

Intonation pattern INT1(Subjunctive) [INT2 in Indicative- focus on uncertainty marker]

4. Behavioural Uses

We present below a series of uses which involve an attempt by the speaker to alter the addressee’s behaviour, including mitigated directives, mitigated prohibitions and supplicatives. The use of subjunctive here lessens (mitigates) the utterance’s illocutionary force, allowing the addressee some sense of saving face.

4.1. Mitigated Directives

The typical marker of MG subjunctive να acts as a mitigator in behavioural uses; mitigated directives are one of its typical uses (as opposed to imperative non- mitigated uses). Where Perfective is used, as in example (17), directives are to be fulfilled in the immediate future, while imperfective uses (example 18) provide general advice on matters the Speaker considers important.

(17) Να σιδερώσεις τα ρούχα σου. SUBJ iron-2SG.PRF the clothes your You should iron your clothes (now).

(18) Να σιδερώνεις τα ρούχα σου. SUBJ iron-2SG.PR.IPF the clothes your You should be ironing your clothes.

Mitigated directives are marked by intonation INT2, as also illustrated in Figure 6 below, with the verb acting as the focal point of the utterance.

Syntax - Morphology - Semantics 2 6 3 Major Trends in Theoretical and Applied Linguistics

0.99

-0.9455 0.7456 2.204 Time (s)

700 500 ) z 300 sideronis ( H ta h

c 200 t i

P Na rouha sou 150

100 70 0.7456 2.204 Time (s)

Figure 6. PRAAT illustration of mitigated directive (INT2).

4.1.1. Summary of Characteristics of Mitigated Directives

Mitigated directives are expressed in present subjunctive 2nd person singular or plural. INT1 and INT2 intonation patterns apply. Negative uses are excluded from this category, as they are considered separately under the mitigated prohibitions category. Their characteristics are summarised in Table 7 below.

Table 7. Characteristics of mitigated directives

Type Behavioural

Function Mitigated Directives Subjunctive (particle να) Grammatical Mood Excludes negative uses Present (fulfilable) Tense Past (unfulfilable) Aspect Perfective/ Imperfective

Person 2nd

Number Singular or Plural

Intonation Pattern INT1/INT2

Addressee’s response N/A

2 6 4 Syntax - Morphology - Semantics Maria Chondrogianni

4.2. Mitigated Directives- Encouragement

A speaker might opt to use a mitigated directive marked by intonation INT4, seemingly giving the impression to the addressee that they have a choice whether to comply or not. Unpleasant suggestions, in particular, trigger the speaker to encourage the addressee to perform an action through a subjunctive question-like utterance, in the 2nd person singular or plural. The use of INT4 further mitigates the impact of the directive, as in example (19), allowing the addressee to save face while encouraged to perform the action.

(19) Να βγάλεις το παλτό σου ? SUBJ remove-2SG.PR.PRF the coat your? Shouldn’t you take your coat off?

The summary of the characteristics of this category can be seen in Table 8 below.

Table 8. Characteristics of the basic illocution mitigated directive-encouragement

Type Behavioural

Function Mitigated Directives- encouragement

Grammatical Mood Subjunctive (particle να, optional negation μη(ν)) Tense Present

Aspect Perfective

Person 2nd

Number Singular or Plural

Intonation Pattern INT4

Addressee’s response [Consent]

4.3. Mitigated Prohibitions

In our view, when μη(ν) is used independently, i.e. without being preceded by the subjunctive particle να, it acts as the MG prohibitive marker (see also Chondrogianni 2011). By opting to choose a structure where να precedes μη(ν) (in a 2nd person singular or plural utterance) the speaker makes a choice to mitigate the impact of a prohibition, as is the case of examples (21) and (22). The presence of the particle να is absolutely necessary for a prohibition to be mitigated. Intonation INT2 applies, with the prohibitive marker (negation) acting as the focal point.

Syntax - Morphology - Semantics 2 6 5 Major Trends in Theoretical and Applied Linguistics

(20) Να μη μιλάς όταν δεν σε ρωτάνε. SUBJ PRH speak-2SG.IPF when NEG you ask-3PL.IPF You shouldn’t speak when you are not asked.

(21) Να μη μιλάτε στον οδηγό. SUBJ PRH speak-2PL.IPF to the driver You shouldn’t speak to the driver.

The characteristics of this category are summarised in Table 9 below.

Table 9. Characteristics of mitigated prohibitions

Type Behavioural

Function Mitigated Prohibitions

Grammatical Mood Subjunctive (particle να, compulsory negation μη(ν)) Tense Present Imperfective Aspect (Perfective possible) Person 2nd

Number Singular or Plural

Intonation Pattern INT2

Addressee’s response [Consent]

4.4. Supplicatives- Requests for Permission

Requests for permission in Modern Greek subjunctive often have an interrogative like intonation, but not an interrogative function: the speaker does not ask for the propositional content of the question to be assigned a value true or false. In a way, the process of a question is reversed. Examples (22) and (23) remind us of the use of modal ‘may’ in English. Such utterances are expressed in the 1st person singular or plural. Intonation INT4 applies.

(22) Να ρωτήσω κάτι; SUBJ ask.1SG.PRF something? May I ask something?

(23) Να πλύνω τα πιάτα; SUBJ wash.1SG.PRF the dishes? May I wash the dishes?

2 6 6 Syntax - Morphology - Semantics Maria Chondrogianni

The supplicatives’ main characteristics are summarized in Table 10 below.

Table 10. Characteristics of supplicatives- requests for permission

Type Behavioural

Function Supplicatives- Requests for permission

Grammatical Mood Subjunctive (particle να, optional negation μη(ν)) Tense Present (Past further mitigates the utterance)

Aspect Perfective

Person 1st

Number Singular or Plural

Intonation Pattern INT4

Addressee’s response Consent

5. Conclusion

In this paper we discussed the pragmatic functions associated with the Modern Greek Subjunctive. We differentiated between propositional and behavioural uses and presented a comprehensive classification of each particular use. We showed that Subjunctive propositional uses include wishes, which might be fulfilable or unfulfilable, optionally introduced by the segmental marker μακάρι; curses, marked by a distinct intonation pattern, optionally introduced by που; wondering uses, optionally introduced by the segmental marker άραγε; mirative uses (of disapproval) , marked by intonation; and expressions of uncertainty introduced by the segmental marker ίσως. Subjunctive behavioural uses, as presented above, include mitigated directives, marked by intonation and use of the 2nd person; mitigated prohibitions, where the presence of negation μη(ν) is obligatory; mitigated directives-encouragement, marked by intonation, with the expectation of a consent response; and supplicative uses (requests for permission) , expressed in the 1st person singular or plural, marked by intonation.

Syntax - Morphology - Semantics 2 6 7 Major Trends in Theoretical and Applied Linguistics

References

Arvaniti, A. 2007. Greek Phonetics: The State of the Art, Journal of Greek Linguistics 8: 97-208.

Arvaniti, A. and Baltazani, M. 2005. Intonation analysis and prosodic annotation of Greek spoken corpora. In Sun-Ah, J. (ed.) Prosodic Typology: The Phonology of Intonation and Phrasing, 84-117, Oxford: Oxford University Press.

Auwera, J. van. 2006. Why languages prefer prohibitives. Journal of Foreign Languages 161, Vol.1., 2-25.

Baltazani, M. 2007. Introduction to Phonetics, Translation into Greek and adaptation of Ladefoged, P. A Course in Phonetics, Athens: Patakis.

Chondrogianni, M. 1997. The Subjunctive in Modern Greek: A Functional Grammar Perspective. In Butler, C.S., Connolly, J.H., Gatward, R.A. and Vismans, R.M. (eds.) A Fund of Ideas: Studies in Language and Language in Use, Amsterdam: IFOTT.

Chondrogianni, M. 2009. The Indicative in Modern Greek. In Tsangalidis, A. (ed.) Selected papers from the 18th International Symposium on Theoretical and Applied Linguistics (18th ISTAL, May 2007), 123-130, Thessaloniki: Monochromia

Chondrogianni, M. 2011 The pragmatics of the modal particles να, θα, ας and μη(ν). In Chatzopoulou, K., Ioannidou, A. and Yoon, S. (eds.) Proceedings of the 9th international conference on Greek linguistics, 322-332, Chicago, Illinois, USA, University of Chicago.

Chondrogianni, M. 2011. The Pragmatics of Prohibitive and Hortative in MG. In Kitis, E., Lavidas, N., Topintzi, N. and Tsangalidis, T. (eds.) Selected papers from the 19th International Symposium on Theoretical and Applied Linguistics (19th ISTAL, April 2009) 135-142, Thessaloniki: Monochromia.

2 6 8 Syntax - Morphology - Semantics Maria Chondrogianni

Chondrogianni, M. 2012 Basic illocutions of the MG indicative. In Gavriilidou, Z., Efthymiou, A., Thomadaki, E. and Kambakis-Vougiouklis, P. (eds.) Selected papers of the 10th ICGL, 223-234, Democritus University of Thrace, Komotini, Greece.

Christidis, A.P. 1986. On the Modern Greek Deictic Particle na, Studies in Greek Linguistics 6, Proceedings of the 5th Annual Conference on Greek Linguistics, May 1985, 221-241, Thessaloniki: Kyriakidis.

Clairis, C., and Babiniotis, G. 1999. Grammar of Modern Greek: The verb, Athens: Ellinika Grammata.

De Groot, C. 2010. Mood in Hungarian. In Rothestein, B. and Tieroff, R. (eds.) Mood in the languages of Europe, Amsterdam: Benjamins.

Economidou-Kogetsidi, M. 2002. Request strategies in English and Greek: Observations from an Airline’s call centre, Nottingham: Linguistic Circular 17.

Giannakidou, A. 2007. Time for mood: the subjunctive revisited. Ms., University of Chicago.

Hannay, M. and Hengeveld, K. 2009. Functional Discourse Grammar: Pragmatic aspects. In Verschueren, J. and Östman, J. (eds.) Handbook of Pragmatics, Amsterdam: Benjamins.

Hengeveld, K. 2004. Illocution, Mood and Modality. In Booij, G.E., Lehman, C., Mugdan, J. and Skopeteas, S. (eds.) Morphology: An International Handbook on Inflection and Word-Formation, Vol. 2, Berlin: De Gruyter.

Hengeveld, K. and Mackenzie, J.L. 2008. Functional Discourse Grammar: A typologically-based theory of language structure, Oxford: Oxford University Press.

Hengeveld, K., Nazareth Bechara, E., Gomes Camacho, R., Regina Guerra, A., Peres de Oliveira, T., Penhavel, E., Goreti Pezatti, E., Santana, L., de Souza, E.R.F. and Teixeira, M.L. 2007. Basic illocutions in the native languages of Brazil. In Mattos Dall’Aglio Hattnher, M. and Hengeveld, K. (eds.) Advances in Functional Discourse Grammar, Special issue of Alfa–Revista de Lingüística 51.2: 73-90.

Holton, D., Mackridge, P. and Philippaki-Warburton, I. 1997. Greek: A comprehensive grammar of the Modern Greek language, London: Routledge.

Syntax - Morphology - Semantics 2 6 9 Major Trends in Theoretical and Applied Linguistics

Joseph, B.D. and Janda, R.D. 1999. The Modern Greek Negator μη(ν)(-) as a Morphological Constellation. In Babiniotis, G. (ed) Greek Linguistics: Proceedings of the 3rd International Conference on Greek Linguistics, 341-351, Athens: Ellinika Grammata.

Joseph, B.D. and Philippaki-Warburton, I. 1987. Modern Greek, Guilford: Biddles Ltd.

Levinson, D. 2005. Aspect in Negative imperative and genitive of negation: a unified analysis of two phenomena in Slavic languages, MS, Stanford University.

Nespor, M. and Vogel, I. 1986. Prosodic Phonology, Dordrecht: Foris.

Noonan, M. 1985. Complementation. In Shopin, T. (ed.) Language Typology and Syntactic Description, Vol. II, Cambridge: Cambridge University Press.

Pavlidou, T. 1991. Cooperation and the choice of linguistics means: some evidence from the use of Subjunctive in Modern Greek, Journal of Pragmatics 15, 11-42.

Platzack, C. and Rosengren, I. 1998. On the subject of imperatives: A minimalist account of the imperative clause, The Journal of Comparative Germanic Linguistics 1, 177-224.

Philippaki-Warburton, I. 1994. The Subjunctive mood and the syntactic status of the particle na in Modern Greek, Folia Linguistica XXVIII/3-4, 299-328.

Philippaki-Warburton, I. 1998. Functional categories and Modern Greek syntax, The Linguistic Review 15, 159-186.

Philippaki-Warburton, I. and Spyropoulos, V. 2004. A change of mood: the development of the Greek mood system, Linguistics 42.2, 791-817.

Philippaki-Warburton, I. and Veloudis, I. 1985. The subjunctive in complement clauses, Studies in Modern Greek Linguistics 5, Proceedings of the 5th Annual Meeting of Greek Linguistics, May 1984, Thessaloniki: Kyriakidis.

Quer, J. (ed). 2009. The distribution and interpretation of indicative and subjunctive, Special issue, Lingua 120.

Risselada, R. 1990. Illocutionary function and functional illocution, Working papers in Functional Grammar 34, Amsterdam: University of Amsterdam.

2 7 0 Syntax - Morphology - Semantics Maria Chondrogianni

Sadock, J.M. and Zwicky, A.M. 1985. Speech Act Distinctions in Syntax. In Shopen, T. (ed.) Language Typology and Syntactic Description, Vol. I, Cambridge: Cambridge University Press.

Tsangalidis, A. 1999. What is the subjunctive? Verbal categories in Modern Greek dictionaries and grammars, Studies in Modern Greek Linguistics 20: 543-554, Thessaloniki: Kyriakidis.

Tsangalidis, A. 2002. Criteria for Modalhood: the case of the Modern Greek modal particles, Studies in Modern Greek Linguistics 21: 759-770, Thessaloniki: Kyriakidis.

Tzartzanos, A. 1946/1989. Modern Greek Syntax, 2 vols, Thessaloniki: Kyriakidis.

Veloudis, I. 1987. *mi figo, mi figis, *mi figi…/ na mi figo, na mi figis, na mi figi and the grammaticalisation of directness, Studies in Greek Linguistics 8: 293-310, Thessaloniki: Kyriakidis.

Veloudis, I. and Philippaki-Warburton, I. 1983. The subjunctive in Modern Greek, Studies in Greek Linguistics 4: 151-168, Thessaloniki: Kyriakidis.

Syntax - Morphology - Semantics 2 7 1 Major Trends in Theoretical and Applied Linguistics

2 7 2 Pride Concepts

Aikaterini Delikonstantinidou

Aristotle University of Thessaloniki [email protected]

Abstract

Pride is generally perceived as an emotion and treated as an emotion concept. The conceptual domain of Pride contains one concept that serves as a cognitive reference point, or its prototype and a variety of related concepts. The connotations accompanying each one of these concepts seem to depend on a balance between the value scale of the causes of Pride and of the Pride scale itself. This intrinsic structure facilitates the categorization of its conceptual metaphors that originate from the emotion concept Pride and are frequently motivated by the physiological effects and behavioral reactions associated with the particular emotion. However, as metaphors are conceptual in nature, beyond their linguistic manifestations we can detect their non-linguistic realizations in other areas of human experience. The effects and reactions attributed to pride also provide the metonymic basis for all the conceptual metonymies related to both the prototypical Pride and to its related concepts. Pride proves to be an extremely complex emotion concept that is culturally non-specific and whose conceptualizations vary greatly, diachronically and interculturally.

1. Introduction

The words we use to construct our inner thoughts and those we use to communicate our thoughts to the others correspond to cognitive models in our minds. Cognitive models are composed of conceptual metaphors; understanding one conceptual domain in terms of another conceptual domain, conceptual metonymies; directing attention to one conceptual domain through another conceptual domain related to it, and literal concepts, which, in their turn, consist of several cognitive models that can be prototypical or non-prototypical (Kövecses 2010). The concept of pride is perceived and treated by linguists as an emotion concept and as such it is characterized by a complex conceptual structure. Although, pride, as all emotions, is an abstract concept, it is not devoid of conceptual content. On the contrary, it is grounded in human experience since

Syntax - Morphology - Semantics 2 7 3 Major Trends in Theoretical and Applied Linguistics

it acquires an embodied nature in order to serve its communicative purposes. The conceptual metaphors, the conceptual metonymies and the related concepts of pride constitute cognitive models that represent conceptually pride as an emotion.

2. Conceptual Structures

Pride, though, beyond its conceptualization as an emotion, can be perceived as a conceptual domain that contains many concepts; conceit, vanity, self-esteem, dignity which correspond to the notions justifi ed pride and false pride. All these concepts relate to pride in a direct way, since we tend to defi ne pride by means of the other concepts, which in turn are defi ned by means of pride. Therefore, on the one hand, pride can be seen as a superordinate category that defi nes a conceptual domain where all the other related concepts lie. However, there has to be a concept in this category of concepts that serves as a ‘cognitive reference point’ or a prototype. This prototype for pride can be termed as “balanced pride as an immediate response” (Kövecses 1986); it is generally considered as justifi ed pride, defi ned as a feeling of self-respect and personal worth, while other forms of pride are systematically defi ned in relation to it. On the other hand, since each of the related concepts of pride constitutes a conceptual domain on its own, one could argue that the relationship between the primal concept of pride and its related concepts is that of overlap; there is partial overlay or correspondence of each conceptual domain with the broader conceptual domain of pride. Of course, we should always keep in mind that the boundaries of each concept, and subsequently of each conceptual domain, are rather fuzzy; for this reason, any claim for an unambiguous and precise defi nition of a concept and of where one conceptual domain ends and where another one begins, would seem rather naïve. In relation to what has been said so far, the two approaches could be schematized thus:

Figure 1. Containment: the superordinate category subsumes related concepts

2 7 4 Syntax - Morphology - Semantics

Aikaterini Delikonstantinidou

Figure 2. Overlap: the superordinate category is partially overlaid by related concepts

Since cognitive models are the conceptual representations of particular emotions, they can be prototypical, related to prototypical members of the category pride, and non-prototypical, thus deviating from the prototypical model(s). According to the prototypical models of emotions; there have to be certain causes that lead to the emotion, and the emotion produces certain responses to the ‘bearer’ of the emotion, some of which are socially acceptable while some others are not. The typical causes of pride are: achievements; solving a problem or winning a race, possessions; having money or possessing a valuable object, belonging to a group; a nation or a team, appearances; pretty face or a gorgeous body, physical and mental capabilities, skills or properties; how strong one is or how healthy one is, moral qualities; one’s honesty or integrity, and social position, social status or class. In the case of pride the various kinds of pride ‘match with’ particular causes, without that being absolute or binding of course. Therefore, justified pride (balanced pride as immediate response, self-esteem, dignity) often ‘goes together’ with achievements, vanity with appearances, while conceit with social status. Additionally, as Zoltán Kövecses suggests, there appear to be values attached to the particular causes of pride; the cause of pride can have built- in social values or the proud person assigns individual values to it. Thus, a value scale is created that is oriented upwards and accompanies the pride scale (how proud a person is). If, for example, someone is proud of his/her virtue, and since virtue has socially built-in high value, then that person is justifi ably proud of it. The value scale can be imagined as having a threshold associated with it; the causes of pride that are above that threshold having high value are proper causes for pride, whereas the opposite stands for those causes that are below the threshold. If one’s pride is justifi ed by the causes of it then we can talk of justifi ed pride, of kinds of pride that are regarded positively, as self-esteem and dignity; in those cases the value scale and pride scale are balanced. However, both when the causes of pride are below the threshold on

Syntax - Morphology - Semantics 2 7 5 Major Trends in Theoretical and Applied Linguistics

the value scale and when one’s pride is in a point higher on the pride scale than the point that corresponds to the point on the value scale, then one’s pride is unjustified; in those cases we regard pride negatively, as conceit or vanity (Kövecses 1986).

3. Conceptual Metaphors

3.1. Prototypical Pride1

Those scales are important when it comes to the conceptualization of pride in metaphors. Not only do they provide an understanding of the distinction between ‘good’ and ‘bad’ pride but, as we shall see further, they provide a cognitive basis for the image schema UP-DOWN that plays a crucial role in the conceptualization of emotion concepts. In addition, the UP-DOWN schema relates to the behavioral reaction of erect posture and of ‘head held unnaturally high’ that is associated with pride and with the MORE IS UP generic-level metaphor we find inherent in many conceptual metaphors. Image-schemas constitute another form of embodiment; the fusion of our bodily, subjective experience with abstraction. In the conceptualization of pride the CONTAINER schema, the UP-DOWN schema, the VOLUME schema and the FORCE schema play a very crucial role. The FORCE schema is based on the notion that when one forceful entity -the self- interacts with another forceful entity –the emotion- that interaction is the cause for the production of certain responses. Besides, the causes of emotions are perceived as forces according to the generic-level conceptual metaphor CAUSES ARE FORCES. The VOLUME schema is based on the behavioral reaction ‘chest-out’ that is associated with pride, and with the notion of BIGNESS that is closely related to pride. Finally, the CONTAINER image schema pertains to the most well-known conceptual metaphor concerning emotions: EMOTIONS ARE FLUIDS IN A CONTAINER. Since emotions are conceptualized as fluids in a container and given that THE BODY IS CONTAINER FOR THE EMOTIONS is another well-known metaphor concerning emotion concepts, PRIDE IS A FLUID IN A CONTAINER metaphor and PRIDE IS A FLUID IN THE HEART metaphor appear to be two logical conceptual metaphors for the prototypical meaning of pride and they are considered general

1 Image-schemas & conceptual metaphors will be appearing in capitals as it is largely the case in linguistic treatments.

2 7 6 Syntax - Morphology - Semantics Aikaterini Delikonstantinidou

emotion metaphors (Kövecses 1986, 2005). In reality, these two metaphors find a logical basis when we consider that the increased heart rate is one of the physiological effects of pride -the heart being container with blood in it- and that one of the behavioral reactions of pride is ‘chest-out’, which involves chest as a container. Therefore:

Source domain: Source domain: The body as a container The heart as a container Target domain: Pride Target domain: Pride PRIDE IS FLUID IN A CONTAINER PRIDE IS A FLUID IN THE HEART

(1a) His good performance filled (2a) Her heart swelled with pride. him with pride. (2b) The news filled her heart with (1b) Her pride rose as she watched pride. her children perform.

In the first metaphor the body is seen as a container, while in the second metaphor the container is the heart. These metaphors emphasize the aspect of intensity: full of, swelled up, the change of intensity: rise, the causes of the intensity: performance, a sight, a win, the news, and the fact that it has a beginning: fill. They also contain the metaphor MORE IS UP that relates to the image schema UP-DOWN: rise. Of course, the CONTAINER image-schema is the most prominent one in those conceptual metaphors. As far as the ontological correspondences –or mapping- are concerned, we understand that knowledge about the constituent elements of the source domain is mapped onto the content of the target domain as follows:

• the container is the body/heart • the container with the fluid/substance is the person who has pride • the fluid/substance in the container is pride • the pressure of the fluid/substance in the container is the intensity of the pride • the cause of the pressure is the cause of pride • the fluid moves UPWARDS when pride becomes more intense (MORE IS UP)

In another conceptual metaphor that provides an evaluation of pride, that is a metaphor that emphasizes the positivity of pride providing a highly positive evaluation for this concept, is the metaphor PRIDE IS LIGHT. The source domain here is the light, whereas the target domain is again pride.

(3) He was beaming with pride.

Syntax - Morphology - Semantics 2 7 7 Major Trends in Theoretical and Applied Linguistics

In this metaphor the ontological correspondences go as follows:

• the source of light is the proud person • the light is pride • the intensity of light is the intensity of light

Again intensity appears as a central aspect of pride, conceptually related to force and thus, to the cause of the emotion, of pride in our case. At this point, we should note that a categorization for the conceptual metaphors concerning pride could not be based on a hierarchy of its intensity alone because there is no psychological theory, cognitive theory or even linguistic data to support and validate such a categorization. Rather the categorization that seems to be valid, and the one that appears here, is the one that draws its validation from the ‘conceptual tools’: the value scale (referring to the causes of pride) and pride scale. In this way, as we have seen, balanced pride as an immediate response, self-esteem and dignity compose one category, where the value scale and pride scale are balanced, whereas conceit and vanity compose the other category, where that is not the case.

3.2. Positive Pride

Let us examine now two conceptual metaphors of pride where the emotion concept appears not with its prototypical meaning but with the senses of self- esteem and dignity, which are both regarded positively in our familiar social context. Self-esteem and dignity are associated and determined by a person’s achievements, possessions, capabilities and moral qualities and less –or not at all- by appearances and/or social status. For pride as self esteem, there are two basic conceptual metaphors that provide us with the conceptualization of pride as something that a person has had for some time and that can act in given situations in a particular manner or that can be treated in a particular way. Thus: PRIDE IS A PERSON

(4a) Her pride was deeply injured. (4b) His pride revolted against the treatment he received. (4c) His national pride was roused/awakened by the stupid comments.

A person can be proud in this way when he/she belongs to a particular group, has valuable possessions or outstanding moral qualities. In these cases self-esteem is a synonym for pride and the word self-esteem could replace the word pride in the examples given. As we see in the examples above a person

2 7 8 Syntax - Morphology - Semantics Aikaterini Delikonstantinidou

can be injured, but can also awaken or even revolt against the cause of pain and agitation to be relieved. In this conceptual metaphor pride can ‘act’ as a person when criticism or humiliating treatment aims to reduce it. As far as the ontological correspondences are concerned, we could claim that:

• The person is pride • The physical harm done to the person is emotional harm to the proud person • Physical revolt against something is the proud person’s emotional revolt • For the person to awaken is for pride to begin or continue to function

Within the same framework, pride can be not just a person, but a superior one as well. In PRIDE IS A SUPERIOR metaphor, pride can dictate to the self what the self can or cannot do, exercising its influence upon the proud person and determining its various actions. Let us illustrate this with some examples:

(5a) His pride prevents him from doing anything dishonorable. (5b) His pride did not allow him to accept the aid.

Pride in the context of this conceptual metaphor continues to be considered positively even if it does not allow a person to act according to his/her wishes, since it is believed that it is reasonable and justified. In the second basic conceptual metaphor of pride as self-esteem, PRIDE IS AN OBJECT and as an object it can exist or not, it can function or not and can be treated badly. However, due to its high value it can occasionally dictate, or interfere with, a person’s actions thus acting also as A SUPERIOR (given the PRIDE IS A SUPERIOR metaphor). The examples go as follows:

(6a) He lost his pride and began to implore her not to leave him. (6b) He tore her pride to shreds. (6c) His pride could not allow him to do that.

Dignity is another related concept to the emotion concept of pride that can be employed as a synonym instead of pride in certain contexts and that is positively connotated. The most common conceptual metaphor for dignity is DIGNITY IS AN OBJECT because dignity is typically conceptualized as having the same function as self-esteem in determining a person’s actions. Moreover, dignity can replace pride in the famous expressions: keeping one’s pride and losing one’s pride, where pride is seen as something that if one has it will prevent him/her from doing certain things, but if one has not, then one will do things he/ she wouldn’t do otherwise -PRIDE IS A SUPERIOR metaphor coupled with PRIDE IS AN OBJECT metaphor (Kövecses 1986). The sentences below exemplify these very facts:

Syntax - Morphology - Semantics 2 7 9 Major Trends in Theoretical and Applied Linguistics

(7a) His grey hair and solemn manner lent him pride/dignity. (7b) He disposed of his dignity.

According to the examples, dignity is not simply an object, but a valuable and desirable one too. In addition, the examples show that dignity is something over which a person exercises control; that control, though, can be lost.

3.3. Negative Pride

Up to that point we have seen pride as a positive aspect of a person’s character. However, pride in the sense of conceit and vanity is considered to be one of the worst vices of the human kind. Linguistically and cognitively, pride is considered negative when the value of the cause of pride is below the threshold associated with the value scale or/and when the pride scale and the value scale are not in balance as we have analyzed earlier; in particular, in the cases of conceit and vanity, the pride scale is higher than the value scale. In contexts where pride has the sense of conceit the conceptual metaphors we run across originate from the physiological effects and behavioral reactions associated with pride. The behavioral reactions: ‘head held unnaturally high’ and ‘chest thrust out’ provide the motivation for the metaphors A CONCEITED PERSON IS UP/HIGH and A CONCEITED PERSON IS BIG. Besides, the up/high position motivates conceptually the UP-DOWN image-schema we find in many conceptual metaphors, and bigness is typically associated with the notion of pride, as we have already seen in the analysis of the VOLUME image-schema. Consider the examples:

A CONCEITED PERSON IS UP/HIGH

(8) He’s been on his high horse ever since he was elected.

A CONCEITED PERSON IS BIG

(9) He has a swollen opinion of herself.

If normal sizes correspond to normally proud people, then greater sizes correspond to people with greater pride. The erect posture becomes head held unnaturally high and chest out becomes chest unnaturally thrust out, because as pride increases the behavioral reactions that accompany pride increase. Furthermore, the concept of SUPERIORITY that applies to the conceptual metaphors for pride as conceit relates directly to the UP/HIGH position of the conceited person.

2 8 0 Syntax - Morphology - Semantics Aikaterini Delikonstantinidou

The concept of SUPERIORITY is coupled with the physiological effects of pride; interference with accurate perception and interference with normal functioning to motivate the conceptual metaphor PRIDE IS AN OPPONENT/SUPERIOR. This metaphor exemplifies the negativity of pride in our cultural model, given that it produces undesirable physiological effects and is dangerous to people. PRIDE IS AN OPPONENT/SUPERIOR

(10a) Conceit robbed him of the ability to judge things properly. (10b) She was blinded by pride.

With the source being any kind of assault done by the opponent and with the target being pride, this conceptual metaphor is constituted by the following ontological correspondences:

• The opponent/superior is pride • Winning is controlling pride • Losing or being assaulted (in any way) is the inability to control pride • The energy/resources needed to control pride are the mental capabilities of the person, necessary to help him/her control his/her pride • Surrendering is for the person to allow himself/herself to be proud

The OPPONENT metaphor focuses on the issue of control and the danger of loss of control to the proud person himself/herself. The metaphor PRIDE IS A DANGEROUS ANIMAL can be imagined as a special case of the OPPONENT metaphor. In the western culture human passions are typically conceptualized as beasts. The widespread conceptual metaphor PASSIONS ARE BEASTS INSIDE A PERSON provides the conceptual foundation for the metaphor PRIDE IS A DANGEROUS ANIMAL. Thus:

(11a) She was swallowed up by conceit. (11b) He was devoured by conceit.

In this metaphor the dangerous animal stands for pride with the sense of conceit, while being harmed by the animal corresponds to the person being exceedingly proud. Not unlike animals, which prey upon other animals voraciously completely destroying them, conceited people demonstrate aggressive behavior when it comes to the causes of their pride; normally their possessions and/or their social status. Similarly, as with the verb ‘swallow’, the verb ‘devour’ indicates that pride destroys one’s personality and often leads conceited people to their eventual doom. As has been pointed out earlier, the cause of pride as an immediate response is a person’s achievement, while the primary causes for self-esteem and dignity

Syntax - Morphology - Semantics 2 8 1 Major Trends in Theoretical and Applied Linguistics

are possessions, belonging to a group, a set of achievements, moral qualities and capabilities of various kinds. Previously, we saw that conceit originates from possessions and social status and now we shall examine vanity as the kind of pride that originates from appearances -though it can be also caused by the intellectual capabilities of a person. Vanity has been traditionally related to the conceptual domain of desire, and thus, the most prevalent conceptual metaphor for pride with the sense of vanity is PRIDE IS A SENSUAL PERSON (Kövecses 1986). In the examples that follow, vanity is conceptualized as having an appetite and as having desires that need to be satisfied:

(12a) She indulged her vanity by looking in the mirror. (12b) Her vanity was eventually gratified.

In another metaphor concerning vanity, VANITY IS AN INFLATED OBJECT. The VOLUME image-schema is dominant in this conceptual metaphor that brings to the fore again the notion of BIGNESS.

(13a) His sarcastic remarks pricked her vanity. (13b) His vanity has been punctured.

Vanity in the two examples is an inflated object that is pricked and punctured by criticism. The ‘harm’ against one’s vanity, or the reduction of it comes from (harsh) criticism against the proud person. The inflated object is filled with air; air has, in folk tradition, no real substance, therefore, vanity has no real ‘substance’, has not real grounds on which to be based and the causes of it are actually insignificant; thus the vain person is superficial and his/her personality is ‘hollow’ -filled with nothing but air. Finally, PRIDE IS AN OPPONENT in another conceptual metaphor concerning pride as vanity as the examples below prove:

(14a) She yielded to her vanity. (14b) Her vanity has been appeased. (14c) He gave in to his pride.

The OPPONENT metaphor appears again here as in the case of conceit and brings to the surface the same ontological correspondences. When the person yields or gives in to vanity he/she loses control of it and his/her actions began being determined by pride rather by his/her reason. The judgment of the vain person then becomes ‘clouded’ and pride becomes ‘the enemy’ within the self. It should be noted that the expression ‘appease’ relates closely to the association of vanity with desire as we have seen.

2 8 2 Syntax - Morphology - Semantics Aikaterini Delikonstantinidou

The vain person perceives appearances as having more value than they actually have and thus, the pride scale and the value scale in the case of vanity, as in the case of conceit are not balanced. The behavioral reaction linked with vanity is the undue preoccupation with one’s self; a self that desires more recognition that is due to him/her. By this point it has become clear that even if the concept of pride exists and is understood independently of any conceptual metaphors, metaphors provide ways of talking about the ontology of pride.

4. Non-linguistic Realizations

Metaphors are conceptual in nature, therefore, beyond their linguistic manifestations or ‘metaphorical linguistic expressions’, conceptual metaphors are realized in many other areas of human experience other than language; these non-linguistic manifestations of conceptual metaphors are called the realizations of conceptual metaphors (Lakoff & Johnson 1980). Our culture abounds with non-linguistic realizations of pride; literature, movies, myths and symbols, even historical figures and cartoons are realizations of conceptual metaphors concerning pride. Let us mention some famous examples. Faust (or Faustus) is the protagonist of a classic German legend that served as the basis for many literary, artistic, cinematic and musical works. Pride, as a vice and a sin, is personified in the legend as the devil’s representative Mephistopheles who is called by Faust in his quest to gain further knowledge and magic powers. Eventually, Faust, is damned due to his hybris; his excessive pride. The conceptual metaphor PRIDE IS AN OPPONENT/SUPERIOR finds its non- linguistic realization in this story, where pride as the Devil is both an opponent and a superior of Faust. Another mythical figure that constitutes realization of the conceptual metaphor PRIDE IS A PERSON is Narcissus. Narcissus or Narkissos in Greek mythology was exceptionally proud and disdained those who loved him. Ever since psychoanalysts studied narcissism as a psychological disorder, the word is used in many contexts as a synonym for either conceit or vanity. Today in the domain of cognitive linguistics, the concept of self-love is inherent both in conceit and in vanity. PRIDE IS A (SUPERIOR) PERSON conceptual metaphor finds its realization also in the figure ofCleopatra ; a historical person that came to be acknowledged as a symbol of pride, of vanity in particular. Politics as a domain is rife with conceptual metaphors. Pride is frequently associated negatively with politics, political thought and discourse and with certain countries’ foreign policies. Pride with the sense of hybris, as we have seen, often indicates being out of touch with reality and overestimating one’s

Syntax - Morphology - Semantics 2 8 3 Major Trends in Theoretical and Applied Linguistics

competence or capabilities, especially for people in positions of power. The conceptual metaphor PRIDE IS A PERSON has found its non-linguistic realization in the faces of several politicians throughout the years. Aesop’s fables frequently warn people against unjustified and great pride. The moral message of most of these popular fables, where frequently PRIDE IS AN OPPONENT, is no other than this: pride comes before a fall. Many symbols of pride derive from the animal kingdom: the eagle, the perdix, the peacock, and the lion. Although, animals in this case are not the realization of the conceptual metaphor PRIDE IS A DANGEROUS ANIMAL, they are associated with pride on the basis of the behavioral reactions pride is linked with; erect posture, chest-out, forms of walking. These behavioral reactions of pride have also given rise to non-verbal manifestations of the concept pride. Many gestures associated with pride are metonymically, rather than metaphorically, motivated due to such reactions.

5. Conceptual Metonymies

Metonymies are similar to metaphors, given that they are conceptual in nature, and they are revealed by metonymic linguistic expressions. As we have seen earlier, conceptual metonymies direct attention to an entity through another entity related to it. The entity that directs attention to, or provides mental access, to another entity is called the vehicle entity, whereas the kind of entity to which attention, or mental access is provided, is called the target entity. The most basic feature, though, of conceptual metonymies, is the fact that the metonymically related vehicle and target entities are ‘close’ to each other in conceptual space; the two entities are contiguously related, or are in each other’s proximity. Therefore, metonymy contrasts with metaphor in that it is based on the relationship of contiguity, and not on the relationship of similarity. When entities form a coherent whole in our experience of the world and they co-occur repeatedly, they are tightly linked in experience and thus can be used to indicate –provide mental access to- other entities within the same idealized cognitive model (ICM) (Lakoff 1987). Metonymies related to the concept of pride originate from the idealized cognitive model of emotion that includes; the physiological effects of an emotion, the behavioral reactions or expressive responses of an emotion, the bearer of the emotion, and so on. More specifically, conceptual metonymies of pride can be categorized as those that are motivated by the physiological effects of pride and those that are motivated by the behavioral reactions or expressive responses of pride. The physiological effects of pride include: redness in the face, increased heart rate, interference with accurate perception and interference with normal mental functioning. Therefore, given the metonymic principle: THE PHYSIOLOGICAL

2 8 4 Syntax - Morphology - Semantics Aikaterini Delikonstantinidou

EFFECTS OF AN EMOTION STAND FOR THE EMOTION, metonymies can be organized as follows:

• Redness in the face (15) He flushedwith pride • Increased heart rate (16) His heart was throbbing with pride • Interference with accurate perception (17) His conceit prevented him from seeing clearly • Interference with normal mental functioning (18) Success went into his head.

It can be noted at this point that redness in the face and increased heart rate apply to what we have characterize justified pride (balanced pride as immediate response, self-esteem, dignity), while interference with accurate perception and interference with normal mental functioning apply to conceit and vanity. The behavioral reactions or expressive responses of pride include: erect posture, chest out, brightness of the eyes, smiling, boasting, head held unnaturally high, forms of walking, ostentatious behavior and thinking one is unique. Given the metonymic principle: THE BEHAVIORAL REACTIONS OF AN EMOTION STAND FOR THE EMOTION, metonymies could be organized in the following manner:

• Erect posture (19) After winning the race, he walked to the rostrum with his head held up high. • Chest out (20) After winning the race, he swelled with pride. • Brightness of the eyes (21) There was pride in his eyes as he looked at his son. • Smiling (22) He was smiling proudly after winning the race. • Chest unnaturally thrust out (23) You should not be so chesty. • Boasting (24) He’s always broadcasting his own achievements. • Head held unnaturally high (25) She is going around with her nose in the air. • Forms of walking (26) He swaggered down the street after winning the fight. • Ostentatious behaviour or theatrical behavior (27) He’s just a show-off. • Thinking one is unique (28) You needn’t think your shit doesn’t stink.

Syntax - Morphology - Semantics 2 8 5 Major Trends in Theoretical and Applied Linguistics

Similarly as in the case of the physiological effects, some of the behavioral reactions as erect posture, chest out, brightness of the eyes and smiling characterize justified pride, whereas the rest of the reactions are linked with conceit and vanity.

6. Conclusion

In the present paper we have examined the conceptual structure of PRIDE; its network of meanings, the conceptual metaphors concerning both the prototypical and the non-prototypical meanings of pride, some non-linguistic realizations of the concept of pride, corresponding either in conceptual metaphors or in physiological effects and behavioral reactions associated with pride, and the conceptual metonymies motivated by the same effects and reactions. Pride is an extremely complex and interesting emotion concept given its rich conceptual domain and its intrinsic structure of the two scales: the value scale and the pride scale, which determines ‘cognitively’ the connotations of the concepts related to pride as well as of its prototypical concept; this structure facilitates its comprehension as a concept and the categorization of its conceptual metaphors. Only in the 20th century pride began being conceptualized as an emotion, and as such it is studied by linguists today and perceived by the majority of people. People, though, even in our modern times, distinguish between “good” and “bad” pride based on their reason and moral judgments of what is and what is not acceptable, justified behavior, even if they are not familiar with the conceptual structure of pride as such.

2 8 6 Syntax - Morphology - Semantics Aikaterini Delikonstantinidou

References

Kövecses, Z. 1986. Metaphors of anger, pride, and love: a lexical approach to the structure of concepts. Amsterdam; Philadelphia: J. Benjamins, 39-60.

Kövecses, Z. 2005. Metaphor in culture: universality and variation. Cambridge; New York: Cambridge University Press, 35-43.

Kövecses, Z. 2010. Metaphor: a practical introduction. Oxford; New York: Oxford University Press, 2010. 63-73, 107-117, 121-124.

Lakoff, G. 1987.Women, Fire, and other Dangerous Things: What Categories Reveal about the Mind. University of Chicago Press.

Lakoff, G. and M. Johnson. 1980. Metaphors We Live By. University of Chicago Press.

Syntax - Morphology - Semantics 2 8 7 Major Trends in Theoretical and Applied Linguistics

2 8 8 Case Attraction in Free Relative Clauses of Ancient Greek: A Study of the Syntax– Morphology Interface

Konstantinos Kakarikos

[email protected]

Abstract

Syntax and morphology often show asymmetries in their interfaces. A typical instance of this asymmetry is illustrated cross-linguistically by case attraction in the free relative clauses. More specifically, structures expressing agreement of the head and the relative pronoun in number and gender but not in case coexist with structures expressing also case agreement but without any syntactic justification. In this paper we examine case attraction in ancient Greek asan instance of the syntax – morphology interface. We claim that it is primarily a morphological phenomenon due to the special way by which morphology interprets syntactic information under certain conditions, such as those defined by an established case feature hierarchy.

1. Introduction

Asymmetric relations between syntax and morphology are an issue of great interest in the study of grammatical interfaces. This phenomenon is typically illustrated, among other things, by case attraction in the syntax of the free relative clauses of ancient Greek. More specifically, whereas the feature of case is taken to be syntactically controlled in structures involving a subject – verb or an object – verb agreement relation, this norm is often violated in structures with free relative clauses. In practical terms there are two different construction types. The first is represented by free relatives which show agreement between the relative pronoun and its omitted antecedent in the features of gender and number but not in the feature of case (cf. 1). The second concerns structures in which the agreement relation counts also the feature of case, but in a way not predicted by syntax (cf. 2).

Syntax - Morphology - Semantics 2 8 9 Major Trends in Theoretical and Applied Linguistics

(1) στυγὼν proi ἣi μ’ ἔτικτεν hating pro-sg-fem-acc who-sg-fem-nom me gave birth “hating the one who gave me birth” (Eur. Alc.338)

(2) ἡ πόλις proi ὧνi (for ἅ-ACC) ἔλαβεν ἅπασιν μετέδωκεν the city pro-pl-neut-gen what-pl-neut-gen received everybody gave “the city gave everybody everything of what it received” (Isocr. 4.29.3)

Examples like (2) illustrate the well-known phenomenon of case attraction and are very common in languages with rich case systems (cf. Latin, Gothic, Old and Middle High German, Modern Greek, Romanian etc), (Grosu 1994: 125). In theoretical terms, they are described in the context of the syntax – morphology interface which acknowledges the distinction between the syntactic and morphological case and assumes: (a) that their relation is not always in a one to one correspondence and (b) that morphological case has its own status which is affected by conditions lying outside the syntactic component (Español- Echevarría & Ralli 2000, Spencer 2006, Sigurðsson 2009, Spyropoulos 2011). In view of these assumptions we claim that case attraction in free relative constructions of ancient Greek is a surface phenomenon which emerges in the course of derivation of the nominal forms and their insertion into certain terminal positions of a syntactic structure. In this procedure morphology acts as a separate, autonomous component of the grammar and its principal role is to interpret the abstract information defined by syntax and expressed in terms of syntactic features (Spencer 2006). This interpretation is only possible through the postulation of a separate set of morphological features which have a central role in the spell-out mechanism of morphology and act in parallel to their syntactic counterparts (Sigurðsson 2009) (see section 3.1. for details).

2. The Free Relative Clauses of Ancient Greek

2.1. The Morpho-Syntax of the Free Relative Clauses

A free relative clause is a sort of embedded clause introduced with a relative pronoun such as ὅς – ἥ – ὅ “who, which”, ὅσος – ὅση – ὅσον “so much”, οἷος – οἷα – οἷον “such as” etc and functions either as an argument or as a nominal complement to the matrix clause which it depends on (Humbert 1960, Schwyzer [1950] 2002, Smyth 197610 among others). The connection of the relative clause to the matrix host is possible through the anaphoric relation between the relative pronoun and its head in the matrix clause. An interesting point and

2 9 0 Syntax - Morphology - Semantics Konstantinos Kakarikos

crucial for the nature of the free relatives is that, in contrast to other types of relative clauses, such as the restrictive ones, the host of the relative pronoun is not overtly realized. As such it assembles the properties of an empty category which has been identified to the small pro (Stavrou & Philippaki-Warburton 1987, Chila-Markopoulou 1990/1991). From a structural viewpoint free relatives may be defined as complement phrases (CPs) introduced with a wh-phrase occurring in their Spec, CP position and attached to the DP of the XP or matrix clause as it is shown in (3)1.

(3) XP X DP D0 CP pro Spec C wh-phrase C0 …….

As mentioned already, the specificity of the construction in (3) lies in the relation of the wh-phrase to its pro head in terms of their grammatical properties and particularly the property of case, causing a sort of asymmetric conflict between the syntactic information defining the relation of the matrix and relative clauses and the lexical forms realizing it in the terminal positions D0 and Spec, CP. In the following section we will sketch out the profile and the parameters of the phenomenon by considering the relevant data from Classical and Hellenistic Greek.

2.2. Case Attraction in Classical and Hellenistic Greek

Asymmetric relations in the feature of case resolved via case attraction are a frequent phenomenon in both Classical and Hellenistic Greek (cf. Goodwin 1900, Humbert 1960, Monteil 1963, Smyth 197610, Mandilaras 19985, Schwyzer [1950] 2002 and Blass 1905, Green 1912, Robertson 1914, Mayser 1934 respectively). However the data available in the literature are not enough to cover all aspects of the phenomenon and draw full conclusions. Thus, our approach will be grounded on indicative examples illustrating only the most common aspects of it2.

1 For a similar view see also Alexiadou & Varlokosta 2007 and Spyropoulos 2011. 2 All examples concern instances of case attraction in simple free relative clauses. Thus, instances involving (i) attraction in structures with a predicate or (ii) inverse attraction are not discussed here.

Syntax - Morphology - Semantics 2 9 1 Major Trends in Theoretical and Applied Linguistics

I. Classical Greek a. Genitive vs. Nominative → Genitive

(4) οὐδὲν ἄν πράξαιμ’ ἄν proi ὧνi (for ὅ-NOM) οὐ σοὶ φίλον nothing could do pro-gen which-gen not you liked “I could never do anything of which you do not like” (Soph. OT.862)

b. Genitive vs. Dative → Genitive

(5) τινὲς ὀλίγοι proi ὧνi (for οἷς-DAT) ἐγὼ ἐντετύχηκα some less pro-gen who-gen I have met “a few of those whom I have met with” (Pl. Res.531e)

c. Genitive vs. Accusative → Genitive

(6) ἀμελήσας proi ὧνi (for ἅ-ACC) μὲ δεῖ πράττειν having neglected pro-gen what-gen I ought to do “having neglected what I ought to do” (Xen. Cyr.5.1.8.)

d. Dative vs. Nominative → Dative

(7) τοῦτο ὅμοιόν ἐστι proi ᾧi (for ὃ-NOM) νῦν δὲ καὶ ἐλέγετο this like is pro-dat which-dat now and said “this is like that which was said just now” (Pl. Phaedr.69a)

e. Dative vs. Accusative → Dative

(8) πᾶν τὸ τεκὸν τροφὴν ἔχει proi ᾧi (for ὅ-ACC) ἄν τέκῃ any parent food provides pro-dat what-dat gives birth “any parent provides food to what he gives birth” (Pl. Men.237e)

f. Accusative vs. Nominative → Accusative

(9) εἶδoν καὶ τὸ κάτ’ Εὐρύλοχον καὶ proi ὅi (for ὅ-NOM) κράτιστον ἦν saw and Eurylochus’s (troops) and pro-acc what-acc the most brave was “they saw Eurylochus’s (troops) and what was the most brave” (Thuc. 3.108.1)

g. Prepositional Genitive3 vs. Nominative → Prepositional Genitive

(10) βλάπτεσθαι ἀφ’ proi ὧνi (for ἅ-NOM) ἡμῖν παρεσκεύασται to be harmed by pro-gen what-gen by us prepared “to be harmed by what had been prepared by us” (Thuc. 7.67.4)

3 The term prepositional refers to structures in which an oblique case (i.e. genitive, dative or accusative) depends on a preposition.

2 9 2 Syntax - Morphology - Semantics Konstantinos Kakarikos

h. Prepositional Genitive vs. Dative → Prepositional Genitive

(11) παρ’ proi ὧνi (for οἷς-DAT) βοηθεῖς οὐκ ἀπολήψει χάριν from pro-gen who-gen help not get thanks “you will get no thanks from those who you help” (Aesch. 2.117)

i. Prepositional Genitive vs. Accusative → Prepositional Genitive

(12) ἀφ’ proi ὧνi (for ὅ-ACC) προσαιτεῖ ἀπὸ τούτων διάγει from pro-gen what-gen begs from these lives “from what he begs from that he lives” (Dem. 8.26)

j. Prepositional Dative vs. Accusative → Prepositional Dative

(13) σύν proi οἷςi (for οὕς-ACC) μάλιστα φιλεῖς with pro-dat whom-dat the most love “with those whom you love the most” (Xen. Anab.1.9.25)

II. Hellenistic Greek a. Genitive vs. Nominative → Genitive

(14) ἀπὸ ἀρουρῶν ις΄ ἤ proi ὅσουi (for ὅσον-NOM) ᾖ τὸ τέταρτον μέρος from acres 16 or pro-gen what-gen is the forth part “from 16 acres or so much land of what equals to the one forth of it” (Mayser 1934: 106)

b. Genitive vs. Dative → Genitive

(15) μάρτυρα proi . . . ὧνi (for οἷς-DAT) τε ὀφθήσομαί σοι witness pro-gen which-gen and will be seen by you “witness for … which I will be seen by you” (Acts, 26.16)

c. Genitive vs. Accusative → Genitive

(16) ἀπήγγειλαν οὐδὲν proi ὧνi (for ἅ-ACC) ἐώρακαν said nothing pro-gen what-gen had seen “they said nothing of what they had seen” (Luke ix.36)

d. Dative vs. Accusative → Dative

(17) ἀκολούθως proi οἷςi (for ἅ-ACC) γέγραφεν Εἰρηναῖος according to pro-dat what-dat has written Eirenaeus “according to what Eirenaeus has written” (Mayser 1934: 105)

e. Prepositional Genitive vs. Nominative → Prepositional Genitive

(18) περὶ δὲ proi ὧνi (for ἅ-NOM) συντετέλεσται εἴς με, Διοφάνην διαγνῶναι for pro-gen which-gen had happened to me Diophanes decide “for what had happened to me, let Diophanes decide” (Mayser 1934: 102)

Syntax - Morphology - Semantics 2 9 3 Major Trends in Theoretical and Applied Linguistics

f. Prepositional Genitive vs. Dative → Prepositional Genitive

(19) ἵνα μή λύπην σχῶ ἀφ’ proi ὧνi (for οἷς-DAT) ἔδει με χαίρειν in order not regret have for pro-gen which-gen should me be happy “in order to feel no regret for those which should made me happy” (Cor. II. 2.3.)

g. Prepositional Genitive vs. Accusative → Prepositional Genitive

(20) ἐρωτῶ περὶ proi ὧνi (for ἅ-ACC) δέδωκάς μοι ask with regards pro-gen what-gen have given me “I ask you with regards to what you have given me” (John xvii.9)

h. Prepositional Dative vs. Accusative → Prepositional Dative

(21) οὐκ ἀρκεσθέντες ἐφ’ proi οἷςi (for ἅ-ACC) ἦσαν διαπεπραγμένοι; not satisfied by pro-dat what-dat had done? “were they not satisfied by what they had done?” Mayser( 1934: 105)

The examination of the data presented reveals that all instances of case attraction occur in fixed environments strictly defined by generalizations concerning first the direction of the attraction process and second the type of the case forms involved. More concretely, the only possible combinations of the conflicting case forms concern exclusively a head in a nominal or prepositional case and a relative pronoun in a nominal case. On the other hand, the resolution of case conflicts is always accomplished with the attraction of the case of the relative to that of its head in accord with the patterns illustrated in table (1) below. More specifically, (i) the direct cases attract into the oblique (cf. 4, 7, 9, 14), (ii) the dative attracts into the genitive (cf. 5, 15), (iii) the accusative attracts into the genitive and dative (cf. 6, 8, 16, 17) and (iv) the nominal cases attract into the prepositional ones (cf. 10, 11, 12, 13, 18, 19, 20, 21).

Table 1. Case conflict resolution in Ancient Greek Free Relatives

Head Relative Resolution

gen nom/dat/acc gen

dat nom/acc dat

acc nom acc

prep-gen nom/dat/acc prep-gen

prep-dat nom/acc prep-dat

The generalizations of table (1), although theoretically valid, are not necessarily in accord with the real tendencies of the language. This means, as far as our evidence can tell, that some case conflict combinations are either not

2 9 4 Syntax - Morphology - Semantics Konstantinos Kakarikos

resolved via case attraction, even when this is possible, or occur more frequently than others. Thus, whereas the conflict between an accusative and a genitive or a dative is a common structure, the case conflict between a genitive and a dative or between a nominative and a prepositional case is rare. On the other hand, whereas masculine and feminine nominatives do not attract into other cases, neuter nominatives do (Harbert 1983, Grosu 1994, Español-Echevarría & Ralli 2000) (see section 4.2. for details).

3. The Analysis

3.1. Theoretical Assumptions: Morphological Vs. Syntactic Features

The starting point for the analysis of case attraction lies on the assumption that morphology is an independent module of the grammar which holds a post-syntactic position and interacts with syntax in several aspects of the language faculty (Halle & Marantz 1993). This interaction may be conceived as a process by which morphology interprets the abstract information of syntax and realizes them by means of inflectional forms (Spencer 2006, Sigurðsson 2009). A necessary prerequisite for such a mechanism to operate is the access of morphology to syntax and also the existence of a morphological system able to administer the syntactic information. In practical terms, it has been claimed (Spencer 2006) that this is possible through two different in nature and independently existent sets of grammatical features: the syntactic and the morphological. The syntactic features are considered to be properties of the phrasal nodes, and therefore are abstract in nature; the morphological features are properties of the word nodes and accordingly are formal representations referring to the concrete values of their syntactic counterparts. The evidence for the existence of these two types of morphosyntactic features may be found first in the fact that there are clear indications for features having exclusively a morphological function without any implication to syntax, such as the inflectional class which regulates the formation of inflectional forms (Spencer 2006, Sigurðsson 2009) or conversely features with a purely syntactic function which have no referent in morphology, like the feature of definiteness which is realized syntactically by certain grammatical words (i.e. articles, pronouns, numerals etc) and not by morphological suffixes (Spencer 2006). On the other hand, one should also have in mind instances where a grammatical feature

Syntax - Morphology - Semantics 2 9 5 Major Trends in Theoretical and Applied Linguistics

may play both a syntactic and a morphological role, in the sense it defines an agreement relation and also regulates the formal marking of words by triggering the rules of suffixation, such as the feature of number etc (Spencer 2006). As a result, the information of syntax assigned in the phrasal nodes corresponds to the morphological information realized in the word nodes of a construction in order to activate the spell-out inflectional rules which will derive the inflectional forms proper for these positions in accord with the requirements of the phrasal nodes. When the syntax – morphology set of correspondences works in a one to one way fashion, the rules of morphosyntax will be defined over a single type of feature; however this feature will be distributed separately over both phrasal and word levels of a structure. This does not seem to happen when the syntax – morphology mapping is not straightforward thus leading to morphosyntactic inconsistencies, such as those attested in case attraction phenomena.

3.2. The Morphological Agreement

A consequence of the distinction between the abstract syntactic and the concrete morphological features proposed is the reformation of the way we understand the notion of agreement in general. Specifically, if morphological features are distinct from their syntactic counterparts, then the relations concerning them should be equally different from the relations referring to the syntactic features. In such a context one may speak for two different types of agreement, the syntactic and the morphological (Sigurðsson 2009). Their existence is evident by the fact that in many languages of the world the syntactic correlations are not always or not necessarily, overtly expressed by morphology and therefore the selections of morphology are independent of the underlying syntactic structure4. Given this, syntactic agreement is an abstract process which establishes the correlation between any syntactic elements that are merged. On the other hand, morphological agreement is a realizational process responsible for the representation of this correlation. This is possible through a mechanism which copies information from a lexical item A to a lexical item B. A necessary prerequisite for this to happen is that all items involved should belong to the same structure or domain and have the same co-reference index. Thus,

4 Consider the lack of overt morphological agreement between the subject and the verb in English (e.g. They would be elected) or the partial finite verb agreement in German (e.g. Sie würden gewählt warden “they would be elected”) and the Swedish partial plural agreement in the predicate participle valda (e.g. Die skulle bli valda “they would be elected), (for the examples and the details see Sigurðsson 2009).

2 9 6 Syntax - Morphology - Semantics Konstantinos Kakarikos

morphological agreement is nothing but a special mechanism which is part of the morphological spell-out operations and has no syntactic motivation. Moreover, the information copied between the agreeing lexical items A and B is not of the same value in the sense it is not equally meaningful for both items. This means that the information of the lexical item A which has a particular meaning to it, may be redundant for the lexical item B (Sigurðsson 2009). The structural formulation of a morphological agreement process brings in mind the phonological assimilation processes and is given in (22)

(22) Lex. item A[a, b, c]i, Lex. item B[a, d, e]i ⇒ Lex. item A[a, b, c]i, Lex. item B[a, b, c]i

In the case of the free relative clauses of ancient Greek the different case features of the pro head (i.e. the lexical item A) and the relative pronoun (i.e. the lexical item B) are syntactically justified since they conform to the requirements of the matrix and relative clause respectively. However they should not be necessarily maintained in the morphological representation of the construction thus allowing for morphology to reorganize the information regarding case by copying the case feature of the pro head to the relative pronoun. Yet, the new information of the relative pronoun has no meaning to it, since it contradicts the syntactic requirements of the relative clause as a whole. The definition of agreement as also a morphological process which exists in parallel to syntactic agreement give clues to our understanding of case attraction phenomena as those presented in section 2.2. On the other hand, it raises questions regarding the evidence speaking for the morphological character and the directionality of the phenomenon. Both issues will be discussed in the next section.

4. The Interpretation of the Data

4.1.Case Attraction in the free Relative Clauses of Ancient Greek

The evidence for the morphological character of case attraction is to be found first in the structure of the free relatives itself. As mentioned already (cf. section 2.1) free relative clauses depend on a covert head (pro) which has a completely defined syntactic profile. This profile is recognized intuitively by the natural speaker of the language (Grosu 1994: 126). If this is true, the role of the morphological agreement as described here is exactly to reflect the hidden syntactic information (Suñer 1983 mentioned by Chila-Markopoulou

Syntax - Morphology - Semantics 2 9 7 Major Trends in Theoretical and Applied Linguistics

1990/1991). Second, there are often instances cross-linguistically in which the asymmetries in the syntax – morphology interface are resolved exclusively by strategies like case syncretism as it is shown in the examples (23) – (25) below (Groos & van Riemsdijk 1981, Harbert 1983, Chila-Markopoulou 1990/1991, Vogel 2003, Spyropoulos 2011). However case syncretism concerns only inflectional forms and therefore is considered to be a morphological rather than a syntactic phenomenon.

(23) εἶδον καὶ τὸ κάτ’ Εὐρύλοχον καὶ proi ὅi (for ὅ-NOM) κράτιστον ἦν saw and Eurylochus’s (troops) and pro-acc what-acc the most brave was “they saw Eurylochus’s (troops) and what was the most brave” (Thuc.3.108.1)

(24) να τραγουδήσουν proi όσεςi (for όσες-ACC) κεράσαμε σφηνάκια should sing pro-nom (those girls) who-nom we treated shots “those (girls) we treated shots should sing” (Spyropoulos 2011)

(25) Ich erzähle proi wasi (for was-nom) immer gefällt I tell pro-acc what-acc ever please “I tell what ever pleases me” (Vogel 2003)

The third piece of evidence relates closely to the factors conditioning case attraction by setting out the direction of the whole process. These factors are also morphological in nature since they concern the morphological features, namely the linguistic primitives expressing the internal constitution of the inflectional forms and triggering the spell-out operations which lead to their derivation. This latter role of features depends crucially on their internal relations and geometry defined in terms of a hierarchical ordering. Feature hierarchies are generally established for both grammatical features (i.e. number, gender, case … etc) and their specific values (i.e. singular …, masculine …, nominative … etc). For highly inflected languages such as ancient Greek, they play a central role as controllers of the maximum number of the inflectional forms which are grammatically derivable in a language and their organization into a small and economical set of inflectional patterns (Kakarikos 2010). On the other hand, their establishment has strong connections with syntax since there are analogies between the syntactic relations of features and their morphological realization. Focusing on the case feature, this analogy is to be found in the syntactic case relations and the way they are expressed by morphological cases. The idea is that peripheral syntactic case relations such as the oblique meanings (i.e. instrument, location etc) are ranked lower in the hierarchy; in contrast, core syntactic case relations such as subject or object are ranked higher. This hierarchy which has a universal status is presented in (26), (Blake 20012).

2 9 8 Syntax - Morphology - Semantics Konstantinos Kakarikos

(26) subject < direct object < indirect object < oblique meanings < prepositional complements

In morphological terms, the meaning of the analogy lies exactly in the fact that the syntactically complex case relations are realized by equally complex morphological representations which subsequently appear in the lower positions of the relevant morphological case feature hierarchy in accord with the pattern illustrated in (27) (Grosu 1994: 122).

(27) nominative < accusative < dative < genitive < prepositional cases

The definition of case attraction as a morphological process strictly conditioned by case hierarchy provides a full account not only of the structures showing the phenomenon but also of those which are not. On the other hand, it gives answers as regards the type and frequency of case conflict resolution practices described in section 2.2. More concretely, case attraction takes place in two successive stages. The first involves the checking of the morphological case features of thehead and the relative pronoun filling out the 0D and Spec, CP positions of (3) as to their compatibility with the established case hierarchy; the second involves the activation or blocking of the morphological agreement copying process depending on weather the morphological case features of the correlating items do satisfy case hierarchy or not. The central idea is that case attraction takes place only in structures contradicting case hierarchy.

Thus, in structure (28)

(28) proi ὧνi (for οἷς-DAT) ἐντυγχάνω πολὺ μάλιστα ἄγαμαί σε pro-gen who-gen meet the most like you “of those who I meet I like you the most” (Pl. Prot.361e)

the morphological genitive of the head and the morphological dative of the relative pronoun are not compatible to the case hierarchy, since genitive ranks lower than dative and therefore dominates over it. As a result, morphology realizes the genitive at the expense of the dative by the morphological agreement process given in (29) which copies the morphological case feature of the head to the relative pronoun.

(29) pro[Pl. Masc. Gen]i, wh-phrase[Pl. Masc. Dat]i ⇒ pro[Pl. Masc. Gen]i, wh-

phrase[Pl. Masc. Gen]i

On the other hand, in a structure like (30)

Syntax - Morphology - Semantics 2 9 9 Major Trends in Theoretical and Applied Linguistics

(30) εἰδέναι τὴν δύναμιν proi ἐφ’ οὕςi ἄν ἴωσιν to know the power pro-gen against which-acc they-proceed “to know the power of those against which they proceed” (Xen. Anab.5.1.8)

the morphological genitive of the head and the prepositional accusative of the relative fit to case hierarchy, since the prepositional accusative is heavier than the genitive and therefore dominates over it. The compatibility of the two cases to the established hierarchy blocks the attraction process.

4.2. The Reliability of Case Hierarchy

As we see, the morphological agreement based on case feature hierarchy is able to account for all relevant data in a proper way. However there are also structures which either challenge case hierarchy or have an impressively small representation in our data. The counter-hierarchical structures occur in examples like (1) repeated here as (31), and in (32).

(31) στυγὼν proi ἥi (instead of ἥν-ACC) μ’ ἔτικτεν hating pro-acc who-nom me gave birth “hating the one who gave me birth” (Eur. Alc.338)

(32) ξυνέλαβον proi ὅσοιi (instead of ὅσους-ACC) ἦσαν ἐν τῇ Ἀττικῇ arrested pro-acc who-nom were in Attica “they arrested those who were in Attica” (Thuc. 2.6.2)

In general terms, structures like these are used as evidence for the optional character of case attraction in relative clauses. This view is also supported by the rich inflection of the relative pronouns in ancient Greek which has been described as the main reason for the non-obligatory case attraction in contrast to the obligatory character of the phenomenon in Modern Greek which has a poorer system of inflection (Suñer 1983, Chila-Markopoulou 1990/1991). With respect to this, one may argue that non-obligatory case attraction, although true for the restrictive relative clauses, does not hold for their free relative counterparts (Grosu 1994: 144-145). The main reason for this distinction is that in contrast to the former which show both hierarchically and non-hierarchically controlled structures without certain restrictions, counter- hierarchical free relatives have always a fixed structure which concerns exclusively masculine and feminine nominatives (Harbert 1983). Such a remark could mean that nominative case is not a possible target of case attraction and also that this is an idiosyncratic feature of Greek totally irrelevant to the character of case hierarchy as such (Grosu 1994: 145). This interpretation finds

3 0 0 Syntax - Morphology - Semantics Konstantinos Kakarikos

extra support to the fact that counter-hierarchical structures like (31) and (32) are not actually conditioned by the feature of the (nominative) case but rather are subject to the feature of the animate (masculine and feminine) gender, as we conclude by the fact that neuter nominatives do conform to case hierarchy without exception (cf. the examples 4, 7, 9, 14, 18). The same conclusion may also be drawn from a different context. More specifically, if it is true, at least on the basis of the linguistic evidence available, that structures like (31) and (32) do not represent targets of case attraction, it follows that a possible resolution of the case conflict via attraction would lead to a highly marked structure. However, it has been proposed (Grosu 1994: 144) that minimally marked structures, when slightly different and semantically equivalent, are more preferable to their highly marked variants. This explanation may also hold for more complicated structures like those involving a conflict between a nominal and a prepositional case. These structures represent even more marked positions in the markedness hierarchy and therefore they are more sensitive to any deviations from the hierarchy. As a result, the possibility for the existence of structures parallel to those represented by the examples given above, is rather small. That is probably the reason for the rarity of examples illustrating the conflict between, say, a nominative and a prepositional-dative or a prepositional-accusative (Grosu 1994: 145).

5. Conclusions

In this paper we examined the case relations in the free relative constructions of ancient Greek. We claimed that these phenomena are morphological andnot syntactic in nature. Therefore, their interpretation is possible through a mechanism by which morphology interprets the syntactic information in order to resolve its asymmetric relation in its interface with syntax. This mechanism, the morphological agreement, is a structural element of the morphological operations responsible for the derivation of the inflectional forms of the language. This approach has the advantage that it is in accord with the current views about the syntax – morphology interface, it supports the autonomy of morphology without precluding its relation with syntax under certain conditions and finally, it reveals the importance of the internal organization of morphological features, such as case, through their hierarchical ranking to the interpretation of syntactic information.

Syntax - Morphology - Semantics 3 0 1 Major Trends in Theoretical and Applied Linguistics

References

Alexiadou, A. and S. Varlokosta. 2007. On the Structure and matching Effects of Free Relatives in Greek. In A. Alexiadou (ed.), Studies in the Morpho-Syntax of Greek. Cambridge: Cambridge Scholar Publishing, 222-250.

Blake, B. 20012. Case. Cambridge: Cambridge University Press.

Blass, F. 1905. Grammar of New Testament Greek. London – New York: McMillan and Co Limited.

Chila-Markopoulou, D. 1990/1991. Προβλήματα Διαχρονικής Σύνταξης: Οι Ελεύθερες Αναφορικές Προτάσεις στη Μεσαιωνική και Νέα Ελληνική. Glossologia 9-10: 13-42.

Español-Echevarría, M. and A. Ralli. 2000. Case Mismatches in Greek. Evidence for the Automony of Morphology. Acta Linguistica Hungarica 47: 179-203.

Goodwin, W.W. 1900. A Greek Grammar. Boston: Ginn and Company.

Green, S.G. 1912. Handbook to the grammar of the Greek Testament. London: The Religious Track Society.

Groos, A. and H. van Riemsdijk. 1981. Matching Effects in Free Relatives: a parameter of core grammar. In A. Beletti, L. Brandi and L. Rizzi (eds.), The Theory of Markedness in Generative Grammar. Pisa: Scuola Normale Superiore di Pisa, 171-216.

Grosu, A. 1994. Three studies in Locality and Case. London: Routledge.

Halle, M. and A. Marantz. 1993. Distributed Morphology and the Pieces of Inflection. In K. Hale and S.J. Keyser (eds.), The view from Building 20: Essays in Honor of Sylvain Bromberger. Cambridge MA: The MIT Press, 111-176.

3 0 2 Syntax - Morphology - Semantics Konstantinos Kakarikos

Harbert, W. 1983. On the nature of the matching parameter. The Linguistic Review 2: 237-284.

Humbert, J. 1960. Syntaxe grecque. Paris: Klincksieck.

Kakarikos, K. 2010. Feature hierarchy and Nominal Inflection. Evidence from Ancient Greek. In A. Ralli, G. Booij, S. Scalize and A. Karasimos (eds.), Proceedings of the 7th Mediterranean Morphology Meeting. Morphology and diachrony Nicosia, Cyprus: 10–13 September 2009. University of Patras, Universiteit Leiden, Università degli Studi di Bologna. Retrieved in October 2010 from: http://www. Philology.upatras.gr/LMGD/el/research/downloads/MMM7Proceedings. pdf, 18- 38.

Madilaras, V. 19985. Η Δομή της Αρχαίας Ελληνικής Γλώσσας. Athens: Kardamitsa.

Mayser, E. 1934. Grammatik des papyri aus der Ptolemäerzeit. Vol. II, Band 3 (Satzlehre). Berlin – Leipzig: Walter de Gruyter and Co.

Monteil, P. 1963. La phrase relative en grec ancien. Paris.

Robertson, A.T. 1914. A Grammar of the Greek New Testament in the light of historical research. Hodder and Stoughton, New York: G.H. Dovan Company.

Schwyzer, E. [1950] 2002. Η Σύνταξη της Αρχαίας Ελληνικής Γλώσσας (Trans. G.E. Papatsimpas and P. Chairopoulos). Athens: Papadimas.

Sigurðsson, H. 2009. Remarks on Features. In K. Grohmann (ed.), Explorations of Phase Theory: Features and Arguments. Berlin: Mouton de Gruyter, 21-52.

Smyth, H.W. 197610. Greek Grammar. Cambridge, Mass: Harvard University Press.

Spencer, A. 2006. Syntactic vs. Morphological Case: implications for morphosyntax. In L. Kulikov, A. Malchukov and P. De Swart (eds.), Case, Valency and Transitivity. Amsterdam: John Benjamins, 2-21.

Spyropoulos, V. 2011. Case Conflict in Greek Free relatives: Case in Syntax and Morphology. In A. Galani, G. Hicks and G. Tsoulas (eds.), Morphology and its Interfaces. Amsterdam: John Benjamins, 21-56.

Stavrou, M. and I. Philippaki-Warburton. 1987. Η Παράμετρος Εναρμόνισης και οι Ελεύθερες Αναφορικές στην Ελληνική Γλώσσα. Studies in Greek Lingustics 8: 311- 322.

Syntax - Morphology - Semantics 3 0 3 Major Trends in Theoretical and Applied Linguistics

Suñer, M. 1984. Free relatives and the matching parameter. The Linguistic Review 3: 363-387.

Vogel, R. 2003. Surface Matters. Case conflict in the free relative constructions and case theory. In E. Bradner and H. Zinsmeister (eds.), New perspectives in Case Theory. Standford: CSLI Publications, 269-300.

3 0 4 Syntax - Morphology - Semantics Τα διαθεσιακά επιρρήματα: Επιμέρους υποκατηγορία προτασιακών επιρρημάτων της Νέας Ελληνικής

Φρύνη Κακογιάννη Ντοά

Πανεπιστήμιο Κύπρου [email protected]

Abstract

This paper deals with Greek attitudinal phrasal adverbs. These adverbs express the speaker’s attitude to what s/he is saying (evaluation of it, shades of certainty or doubt about it) and they can be grouped according to their semantic and morphosyntactic features. We begin by describing the data collection and processing methods. Then, we present the criteria that allowed the identification of these adverbs which are their particular morphosyntactico-semantic properties. We finally present a classification model associated with appropriate syntactic and semantic properties.

1. Εισαγωγή και διάρθρωση εργασίας

Μελετώντας τα σύγχρονα σχολικά εγχειρίδια (Παπαναστασίου, 2010, Χατζησαββίδης & Χατζησαββίδου. 2009, Τριανταφυλλίδης, 2001) και γραμματικές της ΝΕ (πχ. Holton et al., 2000· Κλαίρης & Μπαμπινιώτης, 2010) αλλά και άλλων διεθνών γλωσσών (Grevisse & Goosse, 1995, Arrivé et al.,1986), φαίνεται ότι το επίρρημα αποτελεί ένα μέρος του λόγου γα το οποίο υπάρχουν περαιτέρω προοπτικές έρευνας, εφόσον βασικό κριτήριο της αναγνώρισης και της ανάλυσης τους παραμένει κυρίως το σημασιολογικό τους περιεχόμενο. Άλλωστε, είναι αποδεκτό και στη διεθνή βιβλιογραφία ότι το επίρρημα γενικότερα, από την οπτική γωνία της γραμματικής κατηγορίας ή/και της συντακτικής λειτουργίας, αποτελεί ένα μέρος του λόγου για το οποίο οι μελετητές καταφέρνουν δύσκολα να διακρίνουν και να περιγράψουν το σύνολο των μορφολογικών και σημασιοσυντακτικών ιδιοτήτων τους. Την ιδιαίτερη αυτή δυσκολία τονίζει ο Ηenning Nølke (1990: 3) αναφέροντας ότι: «η συγκεχυμένη ποικιλομορφία των ιδιοτήτων των επιρρημάτων εξακολουθεί να αντιστέκεται στις προσπάθειες συστηματοποίησής τους από λεξικογράφους, μεταφραστές, διδάσκοντες ή γλωσσολόγους». Παρ’όλα αυτά, εξειδικευμένες μελέτες για διάφορες γλώσσες, όπως τα αγγλικά (Greenbaum, 1969), τα γερμανικά (Bartsch, 1977) και τα γαλλικά (Molinier, 1984·

Syntax - Morphology - Semantics 3 0 5 Major Trends in Theoretical and Applied Linguistics

Gross, 1990· Molinier & Lévrier, 2000), καταφέρνουν να διακρίνουν και να περιγράψουν το σύνολο των μορφολογικών και σημασιοσυντακτικών ιδιοτήτων τους που επιτρέπουν την ανάλυση και ταξινόμησή τους με βάση κριτήρια που δεν περιορίζονται μόνο στο σημασιολογικό περιεχόμενο. Τα κριτήρια αυτά μπορούν να είναι μορφοσυντακτικά, λειτουργικά, σημασιοσυντακτικά ή πραγματολογικά. Η διεύρυνση αυτή είχε ως συνέπεια την καταγραφή και λεπτομερή ανάλυση νέων –απόντων από τα γλωσσικά εγχειρίδια– επιρρηματικών στοιχείων και κατηγοριών όπως, για παράδειγμα, τα «προτασιακά» και «ενδοπροτασιακά» επιρρήματα και την κατασκευή μιας ολοκληρωμένης γραμματικής επιρρηματικών στοιχείων (Molinier & Lévrier, 2000). Όσον αφορά τη ΝΕ, μελέτες της τελευταίας 25ετίας (Νάκας1, 1987· Παυλίδου, 1989· Μουστάκη, 1995· Αλεξιάδου, 1997· Κλαίρης και Μπαμπινιώτης, 2001· Βογιατζή, 2003, 2004, 2006· Κακογιάννη Ντοά, 2008· Γούτσος, 2009· Βογιατζή & Κακογιάννη Ντοά, 2009· Κακογιάννη Ντοά & Βογιατζή, 2010· Κακογιάννη Ντοά & Βογιατζή, 2011) περιγράφουν τις ιδιαίτερες ιδιότητες και λειτουργίες των επιρρημάτων ακολουθώντας προσεγγίσεις παρόμοιες με τις ήδη υπάρχουσες για άλλες ευρωπαϊκές γλώσσες. Με βάση τις παραπάνω διαπιστώσεις συνεχίζουμε την έρευνα μας πάνω στα επιρρηματικά στοιχεία της ΝΕ (Source Project, 2012-2013) που έχει στόχο την συστηματική μορφοσυντακτική και σημασιολογική ανάλυση, την εξαντλητική συλλογή και καταγραφή των ελεύθερων και παγιωμένων δομών τους, τη συστηματική και τυποποιημένη γλωσσολογική περιγραφή αλλά και τη συστηματική κατηγοριοποίηση τους. Τελικός στόχος αυτής της έρευνας είναι η δυνατότητα παροχής τυποποιημένου υλικού για τους ερευνητές, συγγραφείς γραμματικών αλλά και κατασκευαστές διδακτικών μεθόδων. Στο παρόν άρθρο παρουσιάζουμε τα αποτελέσματα ανάλυσης μιας υποκατηγορίας προτασιακών επιρρημάτων, τα οποία αποκαλούμε διαθεσιακά επιρρήματα, λόγω του ότι ισοδυναμούν συνήθως με μια παρενθετική πρόταση του ομιλητή που εκφέρει ένα σχόλιο ως προς το διατυπωμένο μήνυμα. Αφορμή επιλογής των επιρρημάτων που παρουσιάζουμε αποτελούν δύο ευρείας χρήσης Γραμματικές της Ελληνικής (Holton et al., 2000, Κλαίρης & Μπαμπινιώτης, 2010), οι οποίες κατατάσσουν στην κατηγορία των προτασιακών επιρρημάτων επιρρήματα όπως βέβαια, δυστυχώς, ευτυχώς, ίσως, σίγουρα, φυσικά2. Ωστόσο, σύμφωνα με την έρευνα μας, φαίνεται να τοποθετούνται τελικά κάτω από το ίδιο υπερώνυμο λεκτικές μονάδες που διαφέρουν σημαντικά όσον αφορά τα μορφοσυντακτικά χαρακτηριστικά και τις σημασιοσυντακτικές λειτουργίες τους. Αναφέρουμε λοιπόν αρχικά το θεωρητικό πλαίσιο της έρευνάς μας και τη μέθοδο συλλογής και διαχείρισης των γλωσσικών δεδομένων που εξετάζουμε, έπειτα αναλύουμε

1 Ο Νάκας, εμπνευσμένος από τον Greenbaum (1969) και Bartsch ( 1976) αναφέρεται εκτεταμένα

στα χαρακτηριστικά των διαθεσιακών επιρρημάτων (ΠΑΡΑΔΙΟΡΔΙΑΘ) της ΝΕ, σελ. 204-222. 2 Οι Γραμματικές της Ελληνικής (Holton et al. 2000, Κλαίρης & Μπαμπινιώτης 2010), διακρίνουν ανάμεσα σε προτασιακά και συνδετικά (conjunctive) ή προτασιακά και κειμενικά επιρρήματα, αντίστοιχα.

3 0 6 Syntax - Morphology - Semantics Φρύνη Κακογιάννη Ντοά

τα κριτήρια εντοπισμού τους, και συνακόλουθα εκθέτουμε τις μορφοσυντακτικές και σημασιολογικές ιδιότητες τους σχετικά με τη λειτουργία τους μέσα στην πρόταση. Τέλος, βάσει των μορφοσυντακτικών και σημασιολογικών ιδιοτήτων τους, προτείνουμε τη ταξινόμησή τους που μας επιτρέπει να διαπιστώσουμε ότι δεν πρέπει να τοποθετούνται κάτω από το ίδιο υπερώνυμο λεκτικές μονάδες που διαφέρουν όσον αφορά τις σημασιοσυντακτικές λειτουργίες τους.

2. Θεωρητικό πλαίσιο

Η μορφοσυντακτική και σημασιολογική ανάλυση των προτασιακών επιρρημάτων της ΝΕ που παρουσιάζουμε πραγματοποιήθηκε με βάση τις θεωρητικές και μεθοδολογικές αρχές του Λεξικού-Γραμματική (Gross, 1975, 1990· Molinier et Lévrier, 2000) που στηρίζεται στη μετασχηματιστική θεωρία του Harris (1964, 1968). Μέσα σε αυτό το πλαίσιο, η ελάχιστη σημασιολογική ενότητα πραγματώνεται στο επίπεδο της βασικής, στοιχειώδους πρότασης (elementary sentence) και όχι στο επίπεδο της λέξης. Η βασική πρόταση (υποκείμενο- ρήμα- συμπλήρωμα) αποτελεί συγχρόνως την κύρια συντακτική ενότητα στην οποία αναφέρεται η γραμματική πληροφορία. Σύμφωνα με το θεωρητικό πλαίσιο που ακολουθούμε, χρησιμοποιούμε τον όρο «επίρρημα» για να δηλώσουμε αφενός τη γραμματική κατηγορία των λέξεων (σε αυτή την περίπτωση αναφερόμαστε στο μέρος του λόγου το οποίο οι συμβατικές γραμματικές ονομάζουν επίρρημα) και, αφετέρου, τη σημασιοσυντακτική λειτουργία που δύνανται να επιτελέσουν τα διάφορα στοιχεία της γλώσσας μέσα σε συγκεκριμένα συντακτικά περιβάλλοντα. Θεωρούμε, λοιπόν, επιρρήματα:

(1) μονολεκτικές λεξικές μονάδες μη αναλύσιμες, π.χ.: πράγματι, όντως, κακώς, κτλ. (2) μονολεκτικές λεξικές μονάδες που παράγονται από επίθετα ή ουσιαστικά, π.χ.: ευτυχώς, δυστυχώς, περίεργα, περιέργως κτλ. (3) πολυλεκτικές λεξικές μονάδες, δηλαδή φράσεις με πιθανή την εναλλαγή ενός ή περισσοτέρων συστατικών, π.χ.: όλως περιέργως, πολύ περιέργως, προς μεγάλη έκπληξη του Πέτρου, προς ευχάριστη έκπληξη του Γιώργου κτλ.

3. Συλλογή των γλωσσικών δεδομένων

Η συλλογή των γλωσσικών δεδομένων για την απομόνωση των μονολεκτικών προτασιακών επιρρημάτων έγινε με δύο μεθόδους. Αρχικά, καταρτίσαμε μια λίστα 12.000 μονολεκτικών επιρρημάτων από το ηλεκτρονικό αντίστροφο λεξικό της ΝΕ (Αναστασιάδη-Συμεωνίδη, 2003) από όπου απομονώσαμε τα προτασιακά επιρρήματα, βάσει κριτηρίων αναγνώρισης μέσα στην πρόταση (προαιρετικός χαρακτήρας, ερώτηση, παράφραση κτλ.). Στην συνέχεια, για τα πολυλεκτικά επιρρηματικά στοιχεία, αντίστοιχα και μη, έγινε η αναζήτησή τους με αποδελτίωση των λεξικών και γραμματικών ευρείας

Syntax - Morphology - Semantics 3 0 7 Major Trends in Theoretical and Applied Linguistics

χρήσης, αλλά και αυθεντικών κειμένων δημοσιογραφικού, λογοτεχνικού, εκπαιδευτικού και διαδικτυακού λόγου και εξειδικευμένες σελίδες3 (ΕΘΕΓ, Η-Κόμβος). Ο κατάλογος που έχει καταρτιστεί μέχρι του παρόντος περιλαμβάνει περίπου 360 προτασιακά επιρρήματα. Πιο συγκεκριμένα, έχουμε συλλέξει περίπου 145 μονολεκτικές δομές (π.χ. γενικά, γενικώς κτλ.) και 215 πολυλεκτικές δομές (π.χ. σε γενικές γραμμές, κατά τη γνώμη μου κτλ.). Ωστόσο, μόνο 128 από αυτά τα επιρρήματα ανήκουν στην κατηγορία των διαθεσιακών επιρρημάτων. Τα υπόλοιπα ανήκουν στην κατηγορία των μεταγλωσσικών προτασιακών επιρρημάτων, αντικείμενο μελέτης υπό δημοσίευση (Κακογιάννη Ντοά, υπό δημοσίευση).

4. Τα διαθεσιακά προτασιακά επιρρήματα

4.1. Γενικά κριτήρια αναγνώρισης

Βασικά κριτήρια αναγνώρισης των διαθεσιακών προτασιακών επιρρημάτων είναι ότι α) ισοδυναμούν συνήθως με μια παρενθετική πρόταση, εφόσον ο ομιλητής εκφέρει ένα σχόλιο ως προς το διατυπωμένο μήνυμα, β) ως προτασιακά επιρρήματα, τροποποιούν ολόκληρη την πρόταση (1), (γ) έχουν τη δυνατότητα εμφάνισης του επιρρήματος στην αρχική θέση αρνητικής πρότασης (2) και (δ) δεν έχουν τη δυνατότητα εστίασης του επιρρήματος μέσα στη δομή Είναι Επίρρημα που Πρόταση (3).

(1) [Χθες στη Βουλή παρουσιάστηκε το φαινόμενο να μην ασκηθεί ο κοινοβουλευτικός έλεγχος] κάτι που βέβαια σημαίνει ότι ούτε οι βουλευτές ήρθαν να... κατακεραυνώσουν τους υπουργούς (2) [Χθες στη Βουλή παρουσιάστηκε το φαινόμενο να μην ασκηθεί ο κοινοβουλευτικός έλεγχος] κάτι που βέβαια δεν σημαίνει ότι οι βουλευτές ήρθαν να... κατακεραυνώσουν τους υπουργούς (3) *[…]είναι βέβαια που σημαίνει ότι ούτε οι βουλευτές ήρθαν να... κατακεραυνώσουν τους υπουργούς

Εξάλλου δεν έχουν σταθερή θέση μέσα στην πρόταση όπως στα πιο κάτω παραδείγματα:

(4) Προς μεγάλη θλίψη των Αθηναίων, όταν κατέκτησαν οι Πέρσες την Ακρόπολη, έκαψαν την ιερή ελιά της Αθήνας

3 Βλ. http://hnc.ilsp.gr/find.asp και http://www.komvos.edu.gr/dictionaries/corpora/ Corpora.htm.

3 0 8 Syntax - Morphology - Semantics Φρύνη Κακογιάννη Ντοά

(5) Όταν κατέκτησαν οι Πέρσες την Ακρόπολη, έκαψαν την ιερή ελιά της Αθήνας, προς μεγάλη θλίψη των Αθηναίων (6) Όταν κατέκτησαν οι Πέρσες την Ακρόπολη, προς μεγάλη θλίψη των Αθηναίων, έκαψαν την ιερή ελιά της Αθήνας (7) Όταν κατέκτησαν οι Πέρσες την Ακρόπολη έκαψαν, προς μεγάλη θλίψη των Αθηναίων, την ιερή ελιά της Αθήνας

Τέλος, συνήθως μπορούν να παραφραστούν κυρίως με τη δομή, Το ότι Πρόταση είναι Επίθετο (Ε + γεγονός). Για παράδειγμα,

(8) Ευτυχώς ο Πέτρος δεν άργησε = Το ότι ο Πέτρος δεν άργησε είναι ευτυχές γεγονός

Καταγράψαμε λοιπόν ως τέτοια 128 επιρρήματα που παρουσιάζουν ποικίλες μορφοσυντακτικές δομές και παραλλαγές. Εμφανίζονται ως α) μονολεκτικά, β) πολυλεκτικά, γ) λόγιες δομές (τω όντι, ασυζητητί κτλ.), δ) με τις δύο καταλήξεις, σε -α και σε -ως (αναμφίβολα, αναμφιβόλως, βέβαια, βεβαίως, απαραίτητα κτλ.), ε) παραγόμενα από επίθετα που δηλώνουν αλήθεια, βεβαιότητα, πιθανότητα (αναπόφευκτο, σίγουρο, φυσικό κτλ.), ζ) έχουν τη δυνατότητα εμφάνισης στον υπερθετικό βαθμό (βεβαιότατα, πιθανότατα κτλ.) και η) έχουν τη δυνατότητα συνδυασμού με το όχι και το και (όχι αναγκαστικά, και βέβαια κτλ.) (βλ. εικόνα 1).

4.2. Ειδικότερες μορφοσυντακτικές ιδιότητες

Παρατηρούμε ότι τα προαναφερθέντα επιρρήματα παρουσιάζουν ειδικότερες μορφοσυντακτικές και σημασιολογικές ιδιότητες που μας προέτρεψαν να τα επαναξιολογήσουμε. Για παράδειγμα, τα διαθεσιακά που δηλώνουν συνήθεια όπως παραδείγματος χάριν, συνήθως, ως συνήθως, κλπ., περιλαμβάνουν δομές με τις δύο καταλήξεις: σε -ά και -ως (αιώνια, αιωνίως, γενικά, γενικώς, μόνιμα, μονίμως κτλ.), λόγιες επιρρηματικές δομές (ανέκαθεν) αλλά και πολυλεκτικές μορφές, (κατά γενικό κανόνα, τις περισσότερες φορές, όπως συνήθως). Όσον αφορά τις συντακτικές τους ιδιότητες, αν και ως προτασιακά, έχουν τη δυνατότητα εμφάνισης στην αρχική θέση καταφατικής ή αρνητικής πρότασης και εμφανίζονται σε προτάσεις όλων των τύπων (κατάφαση, άρνηση, ερώτηση) και όλων των χρόνων, αποκλείονται στην καταφατική προστακτική (9) και στην υποτακτική (10), ιδιότητα που εμφανίζεται σε επιρρήματα όπως ασφαλώς, βέβαια κτλ. (11, 12):

(9) * Συνήθως φάε Μαρία! (10) * Συνήθως ας του το έλεγες! (11) Ασφαλώς, φάε Μαρία! (12) Βέβαια ας του το έλεγες!

Syntax - Morphology - Semantics 3 0 9 Major Trends in Theoretical and Applied Linguistics και και , όχι και βέβαια και ναι φυσικά όχι αναγκαστικά σίγουρα όχι Συνδυασμός με το ναι το και βεβαιότατα πιθανότατα κανονικότατα σιγουρότατα Υπερθετικός Υπερθετικός βαθμός αδιαμφισβήτητα αναμφισβήτητα αναπόφευκτα κανονικά παράξενα περίεργα φυσικά Από επίθετα βέβαια βεβαίως μοιραία ομολογουμένως σίγουρα υποχρεωτικά και σε -ως –α και ανέκαθεν ασυζητητί τω όντι Λόγια από θαύμα από τύχη άλλο δίχως όλως περιέργως όπως συνήθως καλά πάλι πέραν πάσας αμφιβολίας μου προς μεγάλη έκπληξη μου προς μεγάλη χαρά ως εκ θαύματος ως συνήθως Πολυλεκτικά Δείγμα διαθεσιακών επιρρημάτων Δείγμα διαθεσιακών 1. Πίνακας ασφαλώς γενικά δυστυχώς ευτυχώς κακώς οπωσδήποτε προφανώς συνήθως φυσικά Μονολεκτικά

3 1 0 Syntax - Morphology - Semantics Φρύνη Κακογιάννη Ντοά

Επίσης, αυτά τα επιρρήματα συνοδεύονται συχνά μέσα στην πρόταση από επιρρηματικά στοιχεία που δηλώνουν χρόνο (13), και μπορούν να αποτελούν απάντηση σε ερώτηση σε συνδυασμό με το ναι και το όχι (14).

(13) Συνήθως η Μαρία τρώει στην ταβέρνα (τις Δευτέρες + το μεσημέρι) (14) Ναι, συνήθως, τα αποτελέσματα είναι διαθέσιμα εντός 2-3 ημερών

Τα διαθεσιακά του τύπου περιέργως, ευτυχώς, δυστυχώς που δηλώνουν θετική η αρνητική αξιολόγηση παρουσιάζουν επίσης μονολεκτικές αλλά και πολυλεκτικές δομές όπως ευτυχώς, δυστυχώς, περίεργα, παράξενα, κακώς, από τύχη, για κακή (του + της) τύχη, για καλή (του + της) τύχη, όλως παραδόξως, προς μεγάλη λύπη, κτλ.). Έχουν επίσης τη δυνατότητα να αποτελούν απάντηση σε ερώτηση σε συνδυασμό με το ναι και το όχι (15), εμφανίζονται συχνά σε ερωτηματικές προτάσεις (16) και αποκλείονται στη καταφατική προστακτική και υποτακτική (17,18).

(15) Το ύφος του άλλαξε; (Όλως παραδόξως + προς μεγάλη μου έκπληξη + ευτυχώς+ …), (ναι + όχι) (16) Γιατί όλως περιέργως + προς μεγάλη μου έκπληξη, οι περισσότεροι διαμαρτύρονται στην Ελλάδα; (17) * Ευτυχώς φάε Μαρία! (18) * Ευτυχώς ας του το έλεγες!

Εξάλλου, σύμφωνα με τον Νάκα (1987: 221), τα διαθεσιακά γενικότερα, ενώ συνάπτονται με προτάσεις κρίσεως, δεν συνάπτονται με προτάσεις επιθυμίας. Μια άλλη ομάδα διαθεσιακών είναι αυτά που δηλώνουν τροπικότητα όπως φυσικά, βέβαια, όντως, μοιραία, αναπόφευκτα, ομολογουμένως, ασυζητητί, ενδεχομένως, ίσως, όντως, πιθανώς προφανώς, σίγουρα, δίχως άλλο, πέραν πάσας αμφιβολίας, στα σίγουρα, ασυζητητί, μάλλον, οπωσδήποτε, πιθανόν κτλ. Τα περισσότερα παράγονται από επίθετα που δηλώνουν αλήθεια, βεβαιότητα, πιθανότητα (πχ. αναπόφευκτο, σίγουρο, φυσικό κτλ.) και παρουσιάζουν μια παραφραστική συνωνυμία μεταξύ τους. Για παράδειγμα:

(19) Προφανώς, δεν έκανε τα μαθήματά του = Είναι προφανές ότι δεν έκανε τα μαθήματά του

Έχουν τη δυνατότητα να εμφανίζονται στον υπερθετικό βαθμό (βεβαιότατα, πιθανότατα), να έπονται επιρρηματικών στοιχείων όπως πολύ, εντελώς κτλ. (π.χ. πολύ πιθανά, πολύ σωστά, εντελώς σίγουρα κτλ.) αλλά και να ακολουθούν τον σύνδεσμο και (20, 21).

(20) Και βέβαια δεν θα έλθω αύριο (21) Και φυσικά δεν θα έλθω αύριο

Syntax - Morphology - Semantics 3 1 1 Major Trends in Theoretical and Applied Linguistics

Όσον αφορά τις συντακτικές τους ιδιότητες, ως προτασιακά επιρρήματα έχουν φυσικά τη δυνατότητα εμφάνισης στην αρχική θέση αρνητικής πρότασης, στην αρχική θέση θετικής πρότασης, καθώς και τη δυνατότητα μετατόπισης τους σε άλλες θέσεις, όπως στα παραδείγματα (22), (23) και (24):

(22) Προφανώς, θα χιονίσει στον Όλυμπο (23) Θα χιονίσει προφανώς στον Όλυμπο (24) Θα χιονίσει στον Όλυμπο, προφανώς

Εμφανίζονται σε προτάσεις όλων των τύπων (κατάφαση, άρνηση, ερώτηση) χρόνων και εγκλίσεων (25, 26). Δεν αποκλείονται δε από καταφατική προστακτική και υποτακτική όπως συμβαίνει με τις δύο προηγούμενες ομάδες (συνήθειας, θετικής η αρνητικής αξιολόγησης), (27, 28).

(25) Σίγουρα θα έλθεις; (26) Σίγουρα πήγες μάθημα σήμερα; (27) Πρέπει , βεβαίως, να αποχωρήσεις το συντομότερο από το παιγνίδι! (28) Ας του το έλεγες, βεβαίως!

Έχουν όμως, όπως και τα διαθεσιακά που δηλώνουν θετική ή αρνητική αξιολόγηση (ευτυχώς, δυστυχώς κτλ.), τη δυνατότητα να αποτελέσουν απάντηση σε ερώτηση σε συνδυασμό με το ναι και το όχι (27).

(29) Θα χιονίσει στον Όλυμπο; Σίγουρα, ναι

Κάποια από αυτά, ανάλογα με τη θέση στην οποία βρίσκονται παρουσιάζουν αμφισημία, όπως στο παράδειγμα (28), όπου έχουμε επίρρημα που εκφράζει τροπικότητα, ενώ στο (29) το επίρρημα τροποποιεί το ρήμα της πρότασης.

(30) Φυσικά, χιόνισε στον Όλυμπο (31) Ο Παύλος έπαιξε φυσικά τον ρόλο του

Τέλος παρατηρήσαμε ότι μια μικρή ομάδα επιρρημάτων, εκ πρώτης όψεως τροπικά, (ηλιθιωδώς, ανέντιμα, εγωιστικά κτλ.) πληρούν τα γενικά κριτήρια αναγνώρισης των προτασιακών διαθεσιακών επιρρημάτων (ο ομιλητής εκφέρει ένα σχόλιο ως προς το διατυπωμένο μήνυμα, υπάρχει τροποποίηση ολόκληρης της πρότασης, δυνατότητα εμφάνισης του επιρρήματος στην αρχική θέση αρνητικής πρόταση, μη δυνατότητα εστίασης του επιρρήματος μέσα στη δομή Είναι Επίρρημα που Πρόταση, μη δυνατότητα εστίασής του μέσα στη δομή Είναι Επίρρημα που Πρόταση. Τα επιρρήματα αυτά προσανατολίζονται προς το ανθρώπινο υποκείμενο και περιλαμβάνουν περίπου 40 επιρρήματα μόνο σε μονολεκτικές δομές, οι οποίες αναφέρονται πάντα στο υποκείμενο της πρότασης. Για παράδειγμα:

3 1 2 Syntax - Morphology - Semantics Φρύνη Κακογιάννη Ντοά

(32) Ηλιθιωδώς, η Αλεξία δεν είπε την αλήθεια (33) *Είναι ηλιθιωδώς που η Αλεξία δεν είπε την αλήθεια

Το αντίστοιχο επίρρημα το οποίο τροποποιεί το ρήμα της πρότασης έχει τη δυνατότητα εστίασής του μέσα στη δομή Είναι Επίρρημα που Πρόταση και είναι απόλυτα συμβατό με το ρήμα όπως στο ακόλουθο παράδειγμα:

(34) Χαμογελούσε ηλιθιωδώς ο Μιχαλάκης

Τα επιρρήματα λοιπόν αυτά δεν τροποποιούν το ρήμα της πρότασης, εφόσον δεν είναι συμβατά με το ρήμα. Εξάλλου, η ακόλουθη πρόταση

(35) Ηλιθιωδώς, η Αλεξία δεν είπε την αλήθεια

μπορεί να παραφραστεί με τους εξής τρόπους

(36) Η Αλεξία, τέτοια ηλίθια που είναι, δεν είπε την αλήθεια (37) Το ότι η Αλεξία δεν είπε την αλήθεια, είναι ηλίθιο εκ μέρους της (38) Ήταν ηλίθιο εκ μέρους της Αλεξίας που δεν είπε την αλήθεια

Αξίζει να σημειωθεί ότι τα διαθεσιακά που προσανατολίζονται προς το ανθρώπινο υποκείμενο δεν εμφανίζονται σε προτάσεις όλων των τύπων ούτε σε όλους τους χρόνους και εγκλίσεις. Αποκλείονται, παραδείγματος χάριν, στην ερώτηση (39) και στην πρόταση επιθυμίας (40).

(39) ?Ηλιθιωδώς, η Αλεξία είπε την αλήθεια; (40) * Ηλιθιωδώς, ας πάρεις τα χρήματα!

Τέλος, και αυτά μπορούν να αποτελούν απάντηση σε ερώτηση σε συνδυασμό με το ναι και το όχι.

(41) Η Αλεξία δεν είπε την αλήθεια; Ναι, ηλιθιωδώς

5. Συμπεράσματα

Η παρούσα έρευνα κατέδειξε ότι τα προαναφερθέντα διαθεσιακά επιρρηματικά στοιχεία έχουν αρκετές ομοιότητες αλλά και μικροδιαφορές όσον αφορά τις μορφοσυντακτικές και σημασιολογικές τους ιδιότητες. 1) Πληρούν τα βασικά κριτήρια αναγνώρισης των προτασιακών επιρρημάτων (παρενθετική πρόταση, τροποποίηση ολόκληρης της πρότασης, αρχική θέση αρνητικής πρότασης, εστίασης στη δομή Είναι Επίρρημα που

Syntax - Morphology - Semantics 3 1 3 Major Trends in Theoretical and Applied Linguistics

Πρόταση, δυνατότητα μετατόπισης και 2) περιλαμβάνουν κοινές μορφοσυντακτικές ιδιότητες (ίδιου τύπου δομές, απάντηση σε ερώτηση σε συνδυασμό με το ναι και το όχι). Από την άλλη, 1) δεν παρουσιάζουν όλα τον ίδιο τύπο παραφραστικής συνωνυμίας. Επιπλέον, 2) τα διαθεσιακά που δηλώνουν συνήθεια (συνήθως, γενικά κτλ.) και αυτά που δηλώνουν θετική ή αρνητική αξιολόγηση (περιέργως, ευτυχώς, δυστυχώς) αποκλείονται από δευτερεύουσες προτάσεις στην προστακτική και υποτακτική. 3) Τα διαθεσιακά που δηλώνουν τροπικότητα όπως φυσικά, βέβαια, όντως και αυτά που αναφέρονται στο υποκείμενο της πρότασης (ηλιθιωδώς, ανέντιμα κτλ.) έχουν τη δυνατότητα α) εμφάνισης στον υπερθετικό βαθμό (βεβαιότατα, πιθανότατα, ηλιθιωδέστατα), β) παρουσιάζουν αμφισημίες με τα αντίστοιχα τροπικά και τέλος, γ) έπονται επιρρηματικών στοιχείων όπως πολύ, εντελώς (πολύ πιθανώς, εντελώς ηλιθιωδώς κτλ.). Ως εκ τούτου, τα διαθεσιακά προτασιακά επιρρήματα μπορούν να κατανεμηθούν σε τέσσερις σημασιοσυντακτικές υποκατηγορίες. Περιλαμβάνουν αυτά που δηλώνουν α) συνήθεια (συνήθως, ως συνήθως κτλ., β) θετική ή αρνητική αξιολόγηση (περιέργως, ευτυχώς, δυστυχώς κτλ.), γ) τροπικότητα (φυσικά, βέβαια, όντως, μοιραία, αναπόφευκτα, ομολογουμένως, ασυζητητί κτλ.), και τέλος, δ) προσανατολισμό προς το ανθρώπινο υποκείμενο (ηλιθιωδώς, ανέντιμα, εγωιστικά κτλ.) (βλ. Πίνακα, 2).

Πίνακας 2. Κατηγορίες των διαθεσιακών επιρρημάτων

Προτασιακά διαθεσιακά

Υποκατηγορίες Επιρρήματα Καταχωρήσεις

Συνήθειας συνήθως, γενικά κτλ. 13

Αξιολόγησης δυστυχώς, ευτυχώς κτλ. 31

Τροπικά εμφανώς, βέβαια, φυσικά κτλ. 44

Προσανατολισμού προς το υποκείμενο ηλιθιωδώς, ανέντιμα κτλ. 40

128

Από την παρούσα έρευνα προκύπτει ότι υπάρχουν σημασιολογικές και συντακτικές διαφορές ακόμη και εντός της ίδιας (ενδεικτικής) υποκατηγορίας επιρρηματικών στοιχείων (των διαθεσιακών επιρρημάτων), και επομένως δεν είναι δυνατό να τα τοποθετούμε κάτω από το ίδιο υπερώνυμο (π.χ. προτασιακό). Η προτεινόμενη λοιπόν ταξινόμηση των προτασιακών διαθεσιακών επιρρημάτων, αποτέλεσμα της συνέργειας της μορφολογίας, της σύνταξης και της σημασιολογίας, μας επιτρέπει να διαπιστώσουμε την ύπαρξη ενδιάμεσων κατηγοριών που μπορεί να διευκολύνει την κατανόηση του γνωστικού πεδίου των επιρρηματικών στοιχείων λαμβάνοντας υπόψη την πραγματική χρήση της γλώσσας.

3 1 4 Syntax - Morphology - Semantics Φρύνη Κακογιάννη Ντοά

Βιβλιογραφία

Alexiadou, A. 1997. Adverb placement. A case study in antisymmetric syntax. Amsterdam/ Philadelphia: John Benjamins Publishing Company.

Αναστασιάδη-Συμεωνίδου, Ά. 2003. Αντίστροφο Λεξικό της Νέας Ελληνικής. Θεσσαλονίκη, Ινστιτούτο Νεοελληνικών Σπουδών. [Ίδρυμα Μανόλη Τριανταφυλλίδη].

Arrivé, M., Gadet F. & Galmiche M. 1986. La grammaire d’aujourd’hui : Guide alphabétique de linguistique française. Paris: Flammarion.

Bartsch, R. 1972. Adverbialsemantik. Linguistische Forschungsberichte 6. Frankfurt, Main.

Γούτσος, Δ. 2009. Μόρια, δείκτες λόγου και κειμενικά επιρρήματα: Η οριοθέτηση των γλωσσικών κατηγοριών με τη χρήση ΗΣΚ. Πρακτικά 8ου Διεθνούς Συνεδρίου Ελληνικής Γλωσσολογίας, 30 Αυγούστου-2 Σεπτεμβρίου, Πανεπιστήμιο Ιωαννίνων, 754-768. Ηλεκτρονική δημοσίευση στην ιστοσελίδα: http://www.linguist-uoi.gr/cd_web/case2.html.

Greenbaum, S. 1969. Studies in English Adverbial Usage. London, Longman.

Grevisse, M. & Goosse A. 1995. Nouvelle grammaire française. Bruxelles: De Boeck Duculot.

Gross, M. 1975. Méthodes en syntaxe, Paris, Hermann.

Gross, M. 1990. Grammaire transformationnelle du français. 3-Syntaxe de l’adverbe, Paris, ASSTRIL.

Harris, Z. S. 1964. Elementary Transformations. Philadelphia, University of Pennsylvania.

Harris, Z. S. 1968. Mathematical Structures of Language. New York, Wiley.

Syntax - Morphology - Semantics 3 1 5 Major Trends in Theoretical and Applied Linguistics

Holton, D., Mackridge, P. & Ε. Φιλιππάκη-Warburton. 2000. Γραμματική της Ελληνικής Γλώσσας. Αθήνα, Εκδόσεις Πατάκη.

Kakoyianni Doa, F. 2008. Adverbes de phrase: étude contrastive et perspectives didactiques. Thèse de doctorat, Toulouse, Université de Toulouse Le Mirail.

Kakoyianni Doa F. & S. Voyatzi. 2010. Les adverbes français dans un corpus d’apprenants chypriotes grecs : recensement et problèmes d’ambigüités, in: P. L. Popović, C. Krstev, D. Vitas et G. Pavlović-Lažetić, éds, Actes du 29ème Colloque International sur le Lexique et la Grammaire (LGC’10), Belgrade, 14-18 septembre 2010, 161-171.

Kakoyianni Doa F. & S. Voyatzi. 2011. Évaluer la compétence d’emploi de l’adverbe en FLE . Colloque international sur la didactique des langues. Les compétences en progression : défi pour la didactique des langues, 23 et 24 septembre 2011, Université de Poitiers.

Κλαίρης, Χ. και Γ. Μπαμπινιώτης. 2010. Γραμματική της Νέας Ελληνικής. Δομολειτουργική – Επικοινωνιακή. Αθήνα, Ελληνικά Γράμματα.

Molinier, Ch. & F. Lévrier. 2000. Grammaire des adverbes. Description des formes en –ment, Genève-Paris, Droz.

Μουστάκη, Α. 1995. Επίρρημα ρήματος ή επίρρημα πρότασης; Η θεωρία της συναναφοράς. Μελέτες για την ελληνική γλώσσα 15, 340-351.

Νάκας, Θ. 1987. Τα Επιρρηματικά της Νέας Ελληνικής (Προβλήματα Υποκατηγοριοποίησης), Διδακτορική Διατριβή, Αθήνα, Εθνικό και Καποδιστριακό Πανεπιστήμιο Αθηνών.

Nølke, H. (εκδ.) 1990. Classification des adverbes, Langue Française 88, Paris, Larousse.

Παπαναστασίου, Γ. 2010. Εφαρμοσμένη γραμματική της δημοτικής και συντακτικό. Τόμος δεύτερος. Αθήνα, Εκδόσεις Πρωτεύς.

Παυλίδου, Θ. 1989. Τα διστακτικά επιρρήματα. Μελέτες για την ελληνική γλώσσα 9, 527-546.

Τριανταφυλλίδης, Μ. 2001. Νεοελληνική Γραμματική της Δημοτικής. Θεσσαλονίκη, Ινστιτούτο Νεοελληνικών Σπουδών [Ίδρυμα Μανόλη Τριανταφυλλίδη].

Voyatzi, S. 2003. Description morphosyntaxique des adverbes simples en vue d’un système d’analyse automatique des textes grecs. 22ème Conférence Internationale sur le Lexique et la Grammaire (LGC’03), Thessalonique, 17-22 septembre 2003, Grèce.

3 1 6 Syntax - Morphology - Semantics Φρύνη Κακογιάννη Ντοά

Voyatzi, S. 2003. Description par automates des adverbes de date du grec moderne. 6ème édition de l’atelier INTEX, Sofia, 28-30 mai 2003, Bulgarie.

Voyatzi, S. 2004. Les adverbes figés du grec moderne : les classes GPAC etGPCA. 23ème Conférence Internationale sur le Lexique et la Grammaire (LGC’04), Deauville, 11-14 novembre 2004, France.

Voyatzi, S. 2006. Description morphosyntaxique et sémantique des adverbes figés en vue d’un système d’analyse automatique des textes grecs. Thèse de doctorat, Paris, Université de Marne-la-Vallée.

Χατζησαββίδης, Σ & Α. Χατζησαββίδου. 2009. Γραμματική Νέας Ελληνικής γλώσσας. Αθήνα, Οργανισμός εκδόσεως διδακτικών βιβλίων.

Syntax - Morphology - Semantics 3 1 7 Major Trends in Theoretical and Applied Linguistics

3 1 8 The Role of Repetition in the Rise of Concessivity

Haritini Kallergi

Aristotle University of Thessaloniki [email protected]

Abstract

The paper deals with three types of lexical repetition that are involved in structures of asyndetic clause linkage with a concessive meaning. The central question is: how do these apparently similar but also different patterns come to have concessive interpretations? An attempt to answer this question is made on the following basis: the structures involving lexical repetition are treated as a group of constructions (in the Fillmorean sense of the term) and analyzed as the products of a process of development (whether diachronic or synchronic) towards concessivity. With respect to this process, the meanings and functions of repetition, as well as various scenarios of pragmatic enrichment of these meanings play a significant role.

1. Preliminaries

The following examples, which may be found in oral, colloquial speech of Modern Greek (MG)1, seem to invite concessive readings of the type although p > q:

(1) griɲazes miɲazes, to fajes olo sto telos nag-3.IPFV.PST [echo] it eat-3.SG.PFV.PST all in-the end “you were nagging and all, but you ate it all in the end”

1 The data were collected either from the spontaneous speech of speakers of Northern Greek dialects of Modern Greek or from Google searches (especially in informal blogs). In the latter case, it is not easy to retrieve information about the dialect of the speakers, whose utterances have been used. Regarding example (1) (echo-reduplication), Konstantinidou (2004: 346) reports that it is a dialectal feature of people living in settlements or neighbourhoods of Thessaloniki, made by refugees from Asia Minor and the Pontus (e.g. Brusa, Ionia). For the other two constructions (examples 2 and 3), there is no available information on origin, as they are typically not referred to as such in the literature.

Syntax - Morphology - Semantics 3 1 9 Major Trends in Theoretical and Applied Linguistics

(2) griɲazes ksegriɲazes, to fajes olo sto telos nag-3.IPFV.PST kse-nag-3.IPFV.PST it eat-3.SG.PFV.PST all in-the end “despite your nagging, you ate it all in the end”

(3) griɲazes griɲazes, to fajes olo sto telos nag-3.IPFV.PST nag-3.IPFV.PST it eat-3.SG.PFV.PST all in-the end “you were nagging and nagging, but you ate it all in the end”

As is evident, all three examples above involve lexical repetition: in example (3), the repetition of the verb is exact, whereas in (1) and (2) the second occurrence of the verb is somewhat changed. In fact, the repetition of example (1) has been observed in other languages and has often been discussed under the terms “echo reduplication” or “echo-construction” (see, e.g., Keane 2005; Stolz 2008; Kirchner 2010). Repetition of example (2), on the other hand, seems to be a case specially encountered in MG and involves suffixation of the second word by the prefixkse- (for the semantics of kse-, see, e.g., Efthymiou 2001, 2002). Despite their differences with respect to the formal properties of lexical repetition, the three examples above exhibit a number of similarities with respect to their concessive reading. First, they all have a concessive interpretation without the involvement of a concessive marker. In fact, they express concessivity, in fulfilling the following basic requirements: they involve a relation between two propositions (p and q), which may be said to co-occur and to be incompatible with each other, on the basis of world knowledge (Halliday & Hasan 1976; König 1988). This relation is signaled by the juxtaposition of the repeated items (that designate proposition p) to another clause (that designates proposition q), without the mediation of an item that explicitly marks this relation as concessive. Note that a concessive marker may appear in cases (2) and (3) above. For example:

(4) (Molonoti) griɲazes griɲazes/ ksegriɲazes, to fajes olo sto telos “(although) you nagged and nagged, you ate it all in the end”

However, its presence is not necessary. In the case of echo-construction (case 3 above), adding a concessive marker is rather inappropriate:

(5) ?(Molonoti) griɲazes miɲazes, to fajes olo sto telos.

Second, in all three cases the lexical repetition is necessary in order for the concessive reading to arise. That is, instances like those in (6) are impossible.

(6) *Griɲazes/*miɲazes/*ksegriɲazes, to fajes olo sto telos “you nagged/you nagged [echo]/you kse-nagged, you ate it all in the end”

3 2 0 Syntax - Morphology - Semantics Haritini Kallergi

Third, in the majority of the cases above, the repetition is bound to sentence- initial position (similarly to other concessive constructions, see, e.g., Nikiforidou 1990). Thus, with the exception of repetition with kse-, reversing the order of clauses is unacceptable:

(7) *To fajes olo, griɲazes griɲazes/ miɲazes “you ate it all, you nagged and nagged / you nagged [echo]”

but

(8) To fajes olo, griɲazes ksegriɲazes. “you ate it all, despite your nagging (you nagged kse-nagged)”

Fourth, although all three cases of repetition have a range of other functions in MG, they are all subject to special restrictions when they have a concessive interpretation. For example, a word may be repeated three or more times, when the repetition stands in a causal relationship with a juxtaposed clause, as in (9):

(9) xartça, xartça, xartça, piksame sto xartomani lit. ‘papers, papers, papers we got stuck in the paper-pile’ “with all these papers, we got stuffed in paper”.

However, repetition in a concessive relation with the subsequent clause is restricted to two instances:

(10) xartça, xartça, ?xartça, aðji ine i kaði anaciklosis ‘papers, papers, ?papers, empty be-3.PL the bins recycling-GEN’ “despite (having) all these papers, the recycling bins are empty”.

Also, when all three types of repetition express contempt/rejection, the conjunction ce ‘and’ may interfere between the repeated words (examples 11- 13), whereas this is not possible when these types of repetition are involved in concessive relations (example 14):

(11) Papás ce papas, mas eprikses! lit. ‘priest and priest you swelled us’ “(you tell us he is a) priest all the time, you got us tired!”

(12) Ti papás ce ksepapás, oli ta iðja ine. lit. ‘what priest and kse-priest they are all the same’ “you may say he is a priest, but they are all the same”;

Syntax - Morphology - Semantics 3 2 1 Major Trends in Theoretical and Applied Linguistics

(13) Ti Kundera ce Mundera mu tsabunas! ‘what Kundera and [echo] me talk.about-2.SG.PR’ “Kundera schmundera! What are you talking to me about?”

but

(14) ?Griɲaze ce griɲaze/miɲaze/ksegriɲaze, to faje olo. ‘s/he nagged and nagged/echo/kse-nagged, he ate it all’ “although s/he nagged, she ate it all.”

Finally, the three types of repetition are characterized by special intonation when juxtaposed to a clause in a concessive relation. Specifically, they are accompanied by a continuation rise, whereas some constituent of the clause that follows carries narrow focus (e.g., griɲakse, griɲakse, to faje OLO sto telos). It appears that the particular melody characterizes all examples from (1) to (3) above and contributes significantly to their concessive meaning2.

2. Constructions

In view of the characteristics and restrictions discussed above, the three cases of juxtaposition that involve lexical repetition can be treated as constructions, in the specific sense of Construction Grammar of Fillmore et al. (1988). Fillmore et al. 1988: 501, 502) define the construction as a pair of syntactic-semantic rules that gives direct pragmatic interpretation, i.e. without the mediation of the proposition expressed by the structure. In “trespassing” semantics to have a direct pragmatic interpretation, constructions à la Fillmore et al. are idioms in a very broad sense. Such idioms may extend beyond the phrase and often consist of a “skeletal” structure that may be filled by a large number of lexical items. Our cases of juxtaposition of a repetitive structure to a clause (X X / X kse-X / X [m]X, Y) may be considered concessive constructions because: i) they are instances of a specific meaning attached to a specific form, i.e. a pairing of a skeletal form such as the above with a concessive meaning (remember the non-

2 Note that this melody is not unique to these cases of concessive relation, but it may accompany other concessive constructions of the type ‘although X, Y’(e.g. concessive constructions with na, see Nikiforidou 1990: 210). Also, the intonation pattern in question may characterize causal constructions (for more on the intonation of repetition in cause- relation utterances in technical terms, see Kallergi 2013). It is rather claimed here that in our cases, the repetition needs this intonation to be interpreted as the protasis of the clause that follows (the apodosis).

3 2 2 Syntax - Morphology - Semantics Haritini Kallergi

necessary presence of a concessive marker), ii) they are idiosyncratic, in that they are not strictly predictable from the semantics of repetition and the clause that follows. They are also language-specific (e.g. they are not present in English)3; iii) the “skeletons” in question have open lexical slots. In particular, X may be filled by almost any verb, noun or adjective4; and iv) they have special formal features, such as distinctive intonation (see also Taylor 2002 for phonological features of constructions). Also, in accordance to Nikiforidou’s (1990) treatment of X,Y structures (where X a clause introduced by na) as constructions in the above sense, an additional criterion is that the structures involving repetition do not have a concessive reading outside the specific X,Y configurations. A legitimate question might now be: how do structures like X X, Y or X kse-X, Y come about as concessive constructions? Already by assuming that they are constructions (viz. types of formal idioms), we suggest that they have probably undergone some process through which they have acquired their special characteristics. Fillmore et al. (1988: 534) assume that constructions arise from syntax. But this might not explain much in our case, since in the structures in question we are faced with juxtaposition, i.e. with a very primitive form of syntax. Yet, even if the various types of repetition that we examine are juxtaposed to some clause, they have independent semantics, in which we may seek for the “seeds” of concessive relations (taking into account the semantics of the clause that follows and general principles of interpreting the order of clauses, e.g. iconicity, see Haiman 1986). In arguing for grammar as “emergent”, Hopper (1987) suggests that constructions arise, in fact, through language use. Through use, chunks of language become “structuralized”, in that they gradually achieve “cross-textual

3 However, constructions involving the morphological alteration of the initial consonant as in the X -[m]X construction (encountered as echo-word formation, distortive reduplication etc., Stolz 2008; Konstantinidou 2004) are found in a wide range of languages, mostly of Eastern Europe and Asia. Note, however, that it is not certain whether in these languages this kind of formation is involved in concessive constructions (rather than simply meaning “X and the like”, see Keane 2005). Also, in English, a parallel construction with concessive readings would involve the so-called schm-reduplication: fat-schmat, as long as she’s happy (Stolz 2008: 115). Again, the sentence as a whole can be marginally considered as expressing a concessive relation (instead, the first part means “Who cares if she’s fat?” see, e.g. Zwicky & Pullum 1987). Regarding the concessive readings of simple reduplication/repetition, I know only one example in the literature: Riau Indonesian kecil kecil punya cewek itu ‘small small have girl this’ “even though he’s small, he’s got girlfriends” (Gil 2005: 57). 4 Some examples with nouns and adjectives are: krisi ksekrisi/krisi misi, oli ekso ine ‘crisis kse-crisis/ crisis [echo] everybody out be-3.PL’ “despite the (economic) crisis, everybody goes out”; arostos ksearostos, θa pao ‘ill kse-ill will go-1.SG.PFV’ “even if I’m ill, I’ll go”. Admittedly, the echo-construction is less flexible with large classes of items (whether verbs, nouns or adjectives) and it is also restricted phonologically, since words beginning with a vowel or /m/ are dispreferred as bases (see also Keane 2005; Stolz 2008).

Syntax - Morphology - Semantics 3 2 3 Major Trends in Theoretical and Applied Linguistics

consistency” (ibid: 150), i.e. a kind of productivity. Specifically with respect to concessive markers, König (1988) takes a “combining” position in suggesting that “the development of a genuine concessive meaning and its differentiation from related meanings is motivated by general principles of language use and interpretive augmentation”. In other words, a concessive construction may be rooted both in the other meanings that the form of the construction may have, as well as in language use. Note that the latter is not restricted to earlier language use, i.e. in the use of the structures in a diachronic perspective. As Hopper & Traugott (1993: 169) argue, “combining clauses in grammatical ways is grounded in rhetorical production strategies” (emphasis mine). This means that the concessive constructions under examination may be cases of clause combining (without having acquired complete grammatical status), which occur in everyday discourse and on a synchronic basis (for types of clause linkage in general and on a cross-linguistic basis, see, e.g. Cristofaro 2003). In the next two sections, we will examine in some more detail the aforementioned assumptions on the evolution of the concessive constructions in question.

3. Semantics of Repetition and the Evolution of Concessive Constructions

In the previous section, it was suggested that the “seeds” of the concessive meaning of the constructions in question may be sought in other meanings that these forms have. In this section, we will examine the semantics of the three types of repetition and briefly discuss the various possibilities that this semantics give for the rise of concessivity. It is important to mention here that, cross-linguistically, certain semantic domains concerning the proposition p have been attested as sources for the rise of concessive meanings in grammaticalization. König (1988) discusses the origin of concessive markers in notions such as contempt, conditionality (i.e. conditional clauses as predecessors of concessive clauses), “free choice quantification” (viz. the meanings of ‘all’, ‘whatever’ etc.), and the emphatic assertion of the truth of p. The latter two notions relate to epistemic necessity and the notion of “totality of possible worlds”, which have also been discussed as sources of concessive markers (see van der Auwera & Plungian 1998 for concessive grams out of modal markers, Sidiropoulou 1990 for emphatic assertion or epistemic necessity as paths from causation to concession, and Veloudis 1997 for the connection of MG concessive markers with words that denote totality). Let us now examine whether these notions are present in the types of repetition that we are dealing with.

3 2 4 Syntax - Morphology - Semantics Haritini Kallergi

3.1. X [m]X Repetition

This type of repetition (also known as echo-reduplication) has been discussed by Daniel & Moravcsik (2005) as a device for the expression of associative plural, i.e. the meaning of ‘X the like’. In this sense, the second constituent rather vaguely encloses all the entities associated to X, which may be dismissed as irrelevant for the current purposes of the speaker. Thus, it is fairly easy for ‘X and the like’ to also mean ‘X and whatever else relevant’ (i.e. free choice quantification). Moreover, it is easy to consider what is not overtly mentioned as unimportant and, in this sense, to use the construction in order to express contempt, as has been observed in various languages, including MG (Konstantinidou 2004, Keane 2005, Stolz 2008). In fact, Stolz (2008: 126-127) argues that the incomplete morphophonological copying of X in the second constituent metaphorically implies that whatever is referred to by [m]X is a “bad copy” of X, which does not deserve to be mentioned. By means of [m]X, X may also be mocked. In section 4, we will see how the notions of free choice quantification and contempt may encourage the concessive reading of this construction. 3.2. X kse-X Repetition

The prefixkse- primarily signals removal of meaning and cancellation (Efthymiou 2001, 2002), in the sense of Eng. un- or de- (e.g. kse-γrafo ‘un-write-1.SG.PR’ “delete, erase from a list”, kse-pupuʎazo ‘un-feather-1.SG.PR’ “pluck”). In this sense, also, a large amount of derivatives with kse- denote the opposite (or reverse) meaning from the meaning of their base: paγono ‘freeze’ - kse-paγono– ‘un-freeze’ “defreeze”; dino ‘dress’- kse-dino ‘un-dress’ “undress”. In a metaphorical extension, removal and cancellation may concern the value of X, especially when its replica carrying kse- immediately follows it. In this way, kse-X becomes, as with echo-reduplication, a way for the speaker to derate X. This idea is supported by the fact that, in this function, kse- is extremely productive and forms words that do not exist independently in MG, nor can they stand alone (i.e. without being juxtaposed to X): papas ksepapas “priest or whatever”, krio ksekrio “cold or whatever” but *ksepapas ‘kse-priest’, *ksekrio ‘kse-cold’. Furthermore, the use of kse- is strictly colloquial as it is prone to sound offensive or pejorative. More importantly, however, the creation of opposites by means of derivational kse- may be extended in its productive use. Cases like ksepapas and ksekrio may be conceived as virtual opposites of papas and krio respectively. The juxtaposition of opposites in papas ksepapas and krio ksekrio creates a situation, whereby the totality of possible worlds is covered (see Sidiropoulou 1990; Veloudis 1997, 2005). By taking into account the two extreme possibilities (say, of cold and “non-cold”) the speaker exhausts a whole area of possibilities (the totality of possible worlds). In other words, X kse-X involves the semantic ingredients of epistemic necessity.

Syntax - Morphology - Semantics 3 2 5 Major Trends in Theoretical and Applied Linguistics

3.3. X(,) X Repetition

Word repetition has a very wide range of functions, in some of which we can identify sources of concessivity. One such function is emphasis (see, e.g. Kakridi- Ferrari 1998 for MG). By placing emphasis, as, e.g., in Erçete! Erçete! “S/he’s coming! S/he’s coming!”, the speaker asserts the truth of the proposition, and this semantically relates to epistemic necessity. Second, and as we have already mentioned in section 1, repetition may be used to signal contempt and it may also express irony or indignation about something said (usually, continuously) by another interlocutor or referent in the context:

(15) Aman! Siɣnomi, siɣnomi, oli tin ora, mas kurasan! ‘INTERJ sorry sorry all the time us tire-3.SG.PFV.PST’ “Well, enough! By saying they’re sorry all the time, they got us tired!”

Also, similarly to echo-reduplication (although much rarely, and certainly not in MG), repetition or reduplication may express the associative plural, as in Thai: nǝn nǝn ‘money and that sort of thing’ (Moravscik 1978). In fact, this function of quantification is mentioned under the term “similarity” by Moravscik, but it appears to be strikingly similar to the function of echo-reduplication that encourages connotations of contempt and dismissiveness on the part of the speaker. Finally, repetition, especially of verbs, often denotes iterativity or continuation (see Kakridi-Ferrari 1998; Kallergi (2013) and references therein). In juxtaposition to another clause, this type of repetition is often employed to signal a causal or conditional relationship to the proposition of that clause. For instance:

(16) aŋxonotan, aŋxonotan, epaθe eɈefaliko ‘get.stress-3.SG.IPFV.PST get.stress-3.SG.IPFV.PST get-3.SG.PFV.PST stroke’ “he was stressed all the time, (so) he had a stroke”,

In (16), (continuous) stress is interpreted as the cause of the stroke. Also, with verbs in the imperative, as in leje leje, θa ton pisis ‘say-2.SG.IMP say-2.SG.IMP will him convince-2.SG.IPFV’ “by telling him (something) all the time, you will convince him”, repetition is semantically equivalent to a conditional (“if you tell him something all the time, you will convince him”)5.

5 In Kallergi (2013), I argue that both such instances of repetition may be regarded as cases of total reduplication that express the iterative (or durative, see Stolz et al. 2011) and that especially reduplicated verbs in the imperative may possibly be viewed as functionally equivalent to MG “gerunds” or “converbs”, which also have conditional readings (see, e.g., Tsimpli 2000; Moser 2001).

3 2 6 Syntax - Morphology - Semantics Haritini Kallergi

In the next section, we will attempt to explain how these meanings can be pragmatically exploited to render the concessive reading of the constructions under study. For the relation of causality, conditionality and concessivity, the reader may also see, e.g., Sidiropoulou (1990), Athanasiadou (1997), Couper- Kuhlen & Kortmann (2000) and Kitis (2002).

4. Pragmatics and the Evolution of Concessive Constructions

Having examined the functions and meanings of the types of repetition in question, we may now see whether these give rise to the concessive reading of our constructions and, if so, how. It turns out, however, that the various scenarios of pragmatic enrichment towards concessivity may work independently of or in parallel with the meanings mentioned in section 3. There are two (interestingly, opposite) scenarios of pragmatic enrichment that are based on some of the meanings we have seen to be associated with repetition. A highly plausible scenario involves the pragmatic “strengthening” of the proposition in p. The meanings/functions of epistemic necessity (‘whether X or not’), free choice quantification (‘any kind of X’) and the emphatic assertion of the truth of p (‘it is certainly the case that X’) reinforce the truth conditions under which the proposition in p holds. Given that the premises of q (viz., p) become very strong and that q is incompatible with p but also true, q is interpreted to occur despite p (see also Veloudis 1997 for a similar analysis). Strengthening of p may also work at the level of implicatures. As König (1988: 162) argues, “there is no point in emphasizing the truth of the ‘antecedent’ proposition unless the truth or the significance of this proposition as an argument is somehow called into question”. Through the violation of the principle of Relevance (Grice’s “R principle”, viz. ‘say no more than you must’), the addressee understands that the speaker’s reinforcing of p (which is already known to be true) prepares the ground for q as the competitor of p. This mechanism of strengthening works similarly to the ‘yes, but…’ moves in dialogues. In other words, it is important for this scenario that the first part of the concessive relation “acknowledges” the truth of p, against which the truth of q is presented (Sperber & Wilson 1995, Couper-Kuhlen & Thompson 2000). To this end, repetition plays an important role, whether it occurs across speakers or within the utterance of the same speaker. Repetition as a direct quotation (an “echo”) of an interlocutor’s statement (A: X, B: X, but Y) or as a mocking representation of a previous statement (as in example 15 above) sets the background for an evaluation (Sperber & Wilson 1995) or an attack (Couper-Kuhlen & Thompson 2000). In this vein, Couper-Kuhlen & Thompson

Syntax - Morphology - Semantics 3 2 7 Major Trends in Theoretical and Applied Linguistics

(2000) argue that concession develops according to the following rhetorical6 pattern:

(17) The “Cardinal Concessive” (adapted from Couper-Kuhlen & Thompson 2000: 382) 1st move (stating)> A: X (A states something) 2nd move (conceding)> B: X’ (B acknowledges validity of A’s statement) 3rd move (countering)> B: Y (B goes on to claim the validity of a potentially contrasting statement or point)

Couper-Kuhlen & Thompson additionally suggest that speaker B introduces his/her contrasting statement by means of “implicative prosody” and “semantic partitioning” (2000: 381). The former roughly refers to the intonation of repetition mentioned in section 1: the continuation rise (here concerning X’) “implies” that new, interesting information is coming. The latter refers to the fact that, by making an acknowledgement, speaker B has the opportunity to set up a direct semantic contrast between an element of X and an element of Y. Semantic partitioning is even more strongly evident in a relevant scenario of pragmatic enrichment towards concessivity, the “falsified expectations” scenario. In this scenario, the starting point is the meaning of iterativity and its connotations of causality and conditionality, which were mentioned with respect to repetition in section 3. To illustrate, let us take example (16) above: aŋxonotan, aŋxonotan, epaθe eɈefaliko “he got stressed all the time, (so) he had a stroke”. As has been mentioned before, on the basis of Relevance as well as world knowledge, the hearer infers a cause-effect relationship between the iteration of the event in X (‘he got stressed’) and the event in Y (‘he had a stroke’). However, if instead of epaθe eɈefaliko, Y involves a proposition like telia eγrapse ‘perfectly write-3.SG.PFV.PST’ “he got on perfectly with his exams”, the expectation about the cause-effect relationship is falsified due to the semantic contrast between ‘he got stressed’ and ‘(he got on) perfectly’ (which is additionally reinforced by narrow focus on telia7). In assuming that the speaker is still being relevant, and that both X and Y are true, the hearer interprets the relation between the two as concessive (or as “anti-causal”, according to an alternative label for the concessive, see Latos 2009 and Rudolph 1996: 48-52).

6 In fact, Couper-Kuhlen & Thompson (2000: 383) mention that conceding as discourse/ textual relation and especially of this type (by first acknowledging the opposite view) has “been a concern of rhetoricians” as an optimal way to counter and persuade. 7 The truth of this contrastive proposition (that s/he did well with his/her exams) has to be emphasized for the contrast to be sustainable. Thus, apart from narrow focus intonation, telia is preposed (it comes before the verb), whereas with the default (and more “neutral”) word order (eγrapse telia), emphatic stress would perhaps not be adequate for the concessive effect.

3 2 8 Syntax - Morphology - Semantics Haritini Kallergi

The alternative scenario to strengthening p (and the mechanisms we discussed so far), is to derate p. Put in better terms, the speaker may undermine the capability of p to resist or defy the truth of q. This can be done by presenting p as weak or irrelevant. The notion of contempt and rejection which has been referred to in respect to repetition (especially of the X [m]X and X kse-X types) may be exploited as a strategy to reject p as a strong competitor of q and to present q as true in any case. König (1988: 157) suggests that the notion of contempt directly relates to notions of dissonance and conflict, without further discussing this relation. Halliday & Hasan (1976: 254) also mention the notion of irrelevance (premises that are irrelevant and, thus, cannot affect q) in their discussion of the meaning of “dismissal” and its importance in “adversative (concessive) relations” (parenthesis mine). Extending this connection a bit further, we may assume that the meaning of contempt (interestingly attested with all types of repetition under study), may “ignite” the concessive reading of the constructions in question, even if it is only pragmatically exploited in order for the speaker to dismiss p as irrelevant to the truth of q (‘q is true no matter what you have in p > despite p’).

5. Conclusion and Issues for Future Research

This brief study has shown that formal repetition seems to play a significant role in the rise of concessivity. Via the meanings of the three types of repetition under examination (simple repetition, echo-reduplication and repetition with kse- suffixation), and especially the meanings of emphasis, iterativity (> causality, conditionality) and contempt, repetition was seen to relate to: i) the semantics that stand as synchronic sources for the rise of forms with concessive meanings (and which may largely coincide with the diachronic sources of concessive markers in the literature), and ii) the discourse strategies from which concessive interpretations may arise synchronically. Moreover, by treating such forms as constructions (as in Fillmore et al. 1988), we have, on the one hand, isolated the three cases of interest as a group with common characteristics (the lexical repetition and its meanings) and an idiosyncratic character, and, on the other hand, we have placed these instances among a larger inventory of types of clause combination with grammatical meanings. More specifically, in treating structures like X [m]X, Y; X kse-X, Y and X X, Y as constructions, we have assumed an evolutionary process from free syntax and language use to more restricted forms of expression. This process may hold for other types or strategies of asyndetic clause linkage (especially for subordination, see Cristofaro 2003). Thus, an issue that arises (and which has been taken up, but, due to limitations of space, is not included in the current presentation) is whether

Syntax - Morphology - Semantics 3 2 9 Major Trends in Theoretical and Applied Linguistics

the evolution towards constructions expressing a grammatical meaning can be considered a case of “grammaticalization across clauses” (Hopper & Traugott 1993, 2003). Hopper & Traugott (1993) have proposed three subsequent stages of grammaticalization of clause combination, based on a number of morphosyntactic, semantic and pragmatic criteria: parataxis > hypotaxis > subordination (Hopper & Traugott 1993: 179). If the concessive constructions in question reflect instances of grammaticalization, where should they be placed along this cline of grammatical clause linkage? In a similar vein, it remains to be seen whether the formal and structural differences between these three types of constructions (some of which have been mentioned in section 1) reflect differences in status with respect to the relevant levels of grammaticalization. There is, for example, evidence that the constructions involving echo-reduplication and repetition with kse- can be regarded as instances of hypotaxis, whereas the construction involving simple repetition is located at the stage of parataxis. This evidence, however, cannot be discussed in the current paper. A final issue that may be worth mentioning is that, so far, we have been dealing with repetition of verbs, hence with “legitimate” cases of clause linkage; The question is what happens when X (the repeated item in the above constructions) is not a verb, hence we cannot strictly talk about clauses, as, e.g., in jatros (kse) jatros, ðio paketa tin imera kapnizi ‘doctor (kse-)doctor, two packets the day smoke-3.SG.PR’ “he may be a doctor, but he smokes two packets of cigarettes a day”. In all likelihood, the first part of the concessive relation jatros( (kse)jatros), which is a case of repetition of a noun, must be treated separately from a model of grammaticalization that is based on criteria of “clause linkage”. Such a treatment, however, would certainly require a separate and more extended study.

List of Abbreviations

1 1st person 2 2nd person 3 3rd person GEN genitive IMP imperative INTERJ interjection IPFV imperfective PL plural PR present PFV perfective PST past SG singular

3 3 0 Syntax - Morphology - Semantics Haritini Kallergi

References

Athanasiadou, A. 1997. Concession and adversativeness: areas of convergence and divergence. In B. Smieja and M. Tasch (eds.), Human contact through language and linguistics. Peter Lang, 3-16.

Couper-Kuhlen, E. and B. Kortmann. 2000. Cause, Condition, Concession, Contrast: Cognitive and Discourse Perspectives. Berlin: Mouton de Gruyter.

Couper-Kuhlen, E. and S. Thompson. 2000. Concessive Patterns in Conversation. In E. Couper-Kuhlen and B. Kortmann (eds.), Cause, Condition, Concession, Contrast: Cognitive and Discourse Perspectives. Berlin: Mouton de Gruyter, 381- 410.

Cristofaro, S. 2003. Subordination. Oxford: Oxford University Press.

Efthymiou, A. 2001. Το νεοελληνικό πρόθημα ξε-: οι έννοιες της απομάκρυνσης και της αλλαγής κατάστασης. [The Modern Greek prefixkse- : the notions of removal and change of state]. Μελέτες για την Ελληνική Γλώσσα. Πρακτικά της 21ης ετήσιας συνάντησης του Τομέα Γλωσσολογίας, Φιλοσοφική Σχολή. ΑΠΘ, 202-213.

Efthymiou, A. 2002. Σημασιολογικές παρατηρήσεις για τα Ν.Ε. προθήματα ξε-, εκ- και από-. [Semantic observations on the Modern Greek suffixes kse-, ek- and apo-]. Μελέτες για την Ελληνική Γλώσσα. Πρακτικά της 22ης ετήσιας συνάντησης του Τομέα Γλωσσολογίας, Φιλοσοφική Σχολή. ΑΠΘ, 199-209.

Fillmore, C.J., P. Kay and M.C. O’Connor. 1988. Regularity and Idiomaticity in Grammatical Constructions: The case of let alone. Language 64: 501-538.

Haiman, J. 1986. Constraints on the form and meaning of the protasis. In Traugott, E.C. (ed.), On conditionals. Cambridge: Cambridge University Press, 215–227.

Syntax - Morphology - Semantics 3 3 1 Major Trends in Theoretical and Applied Linguistics

Halliday, M.A.K. and R. Hasan. 1976. Cohesion in English. London, New York: Longman.

Hopper, P. 1987. Emergent grammar. Berkeley Linguistic Society 13, 139-157.

Hopper, P. and E.C. Traugott. 1993. Grammaticalization. Cambridge: Cambridge University Press.

Kakridi-Ferrari, M. 1998. Επανάληψη: Η λειτουργία της ως βασικού μηχανισμού στη γλώσσα. [Repetition: its function as basic mechanism in language]. PhD dissertation. National and Kapodistrian University of Athens.

Kallergi, H. Kallergi (2013). Reduplication at the word level: the Greek facts in a typological perspective. PhD dissertation. Aristotle University of Thessaloniki.

Keane, E. 2005. Phrasal Reduplication and Dual Description. In B. Hurch (ed.), Studies on Reduplication. Berlin and New York: Mouton de Gruyter, 239-262.

Kirchner, J.S. 2010. Minimal Reduplication. PhD dissertation. University of California, Santa Cruz.

Kitis, E. 2002. On the Modern Greek conditional connective an or towards restoring the image of the Greek culture. In M. Makri-Tsilipakou (ed.), Selected Papers on Theoretical and Applied Linguistics. From the 14th International Symposium April, 2000. Thessaloniki: University Studio Press, 154-174.

König, E. 1988. Concessive Connectives and Concessive Sentences: Cross- linguistic Regularities and Pragmatic Principles. In J.A. Hawkins (ed.), Explaining Language Universals. Oxford, New York: Basil Blackwell Ltd., 145-166.

Konstantinidou, M. 2004. Παραμορφωτική αναδίπλωση λέξεων: ένα διαλεκτικό σχήμα λόγου και ο λειτουργικός του ρόλος. [Distortive reduplication of words: a dialectical figure of speech and its functional role]. Μελέτες για την Ελληνική Γλώσσα: Πρακτικά της 24ης Ετήσιας Συνάντησης του Τομέα Γλωσσολογίας του τμήματος Φιλολογίας του Αριστοτελείου Πανεπιστημίου Θεσσαλονίκης, Φιλοσοφική Σχολή. ΑΠΘ, 345-356.

Latos, A. 2009. Concession on Different Levels of Linguistic Connection: Typology of Negated Causal Links. Retrieved 9 February 2012 from http://www.ncl.ac.uk/ linguistics/assets/documents/5Latos.pdf.

Moravcsik, E. 1978. Reduplicative Constructions. In J.H. Greenberg (ed.), Universals of Human Language, Vol III. Stanford: Stanford University Press, 297-334.

3 3 2 Syntax - Morphology - Semantics Haritini Kallergi

Moravcsik, E. and M. Daniel. 2005. Associative Plural. In M. Haspelmath, M.S. Dryer, D. Gil and B. Comrie (eds.), The World atlas of Language Structures 1. New York: Oxford University Press, 150-154.

Moser, A. 2001. Μετοχές, γερούνδια, -οντας: χρόνος και άποψη. [Participles, Gerunds, -ondas: tense and aspect.] Γλωσσολογικές Έρευνες για την Ελληνική ΙΙ: Πρακτικά του 5ου Διεθνούς Συνεδρίου Ελληνικής Γλωσσολογίας, Σορβόννη, 13-15 Σεπτεμβρίου 2001, 111-114.

Nikiforidou, K. 1990. Eναντιωματικές προτάσεις, πραγµατολογία και γραµµατικές ιδιαιτερότητες. [Concessive clauses, pragmatics and grammatical idiosyncrasies]. Μελέτες για την Ελληνική Γλώσσα: Πρακτικά της 10ης Ετήσιας Συνάντησης του Tοµέα Γλωσσολογίας του τμήματος Φιλολογίας του Αριστοτελείου Πανεπιστημίου Θεσσαλονίκης, Φιλοσοφική Σχολή. Αριστοτέλειο Πανεπιστήµιο Θεσσαλονίκης, 205-223.

Rudolph, E. 1996. Contrast: Adversative and concessive relations and their expressions in English, German, Spanish, Portuguese on sentence and text level. Berlin and New York: Walter de Gruyter.

Sidiropoulou, M. 1990. Αιτία-υπόθεση/Εναντίωση: παρατηρήσεις για το σημασιολογικό χαρακτήρα των συνδετικών. [Cause-condition/concession: remarks on the semantics of connectives]. Μελέτες για την Ελληνική Γλώσσα: Πρακτικά της 10ης Ετήσιας Συνάντησης του Tοµέα Γλωσσολογίας του τμήματος Φιλολογίας του Αριστοτελείου Πανεπιστημίου Θεσσαλονίκης, Φιλοσοφική Σχολή. Αριστοτέλειο Πανεπιστήµιο Θεσσαλονίκης, 285- 294.

Sperber, D. and D. Wilson. 1995. Relevance: Communication and Cognition. [Second Edition]. Oxford, Berlin: Blackwell.

Stolz, T., C. Stroh and A. Urdze. 2011. Total Reduplication: the Areal Linguistics of a Potential Universal. Studia Typologica 8. Berlin: Akademie Verlag.

Stolz, T. 2008. Total reduplication vs. echo-word formation in language contact situations. In P. Siemund, and N. Kintana (eds.), Language Contact and Contact Languages. Amsterdam, Philadelphia: John Benjamins B.V., 107-132.

Taylor, J.R. 2002. Cognitive Grammar. Oxford: Oxford University Press.

Tsimpli, I.M. 2000. Gerunds in Greek. Journal of Greek Linguistics 1, 133-169.

van der Auwera, J. and V.A. Plungian. 1998. Modality’s Semantic Map. Linguistic Typology 2, 79-124.

Syntax - Morphology - Semantics 3 3 3 Major Trends in Theoretical and Applied Linguistics

Veloudis, I. 1997. Από την καθολικότητα-συνολικότητα στην αντίθεση-εναντίωση [From totality to contrast-concession]. Studies in Greek Linguistics 17, 417-423.

Veloudis, Ioannis. 2005. Η άρνηση. Αθήνα: Πατάκης.

3 3 4 Syntax - Morphology - Semantics Τα ρήματα σε -άμαι της Νεοελληνικής Κοινής

Γεωργία Κατσούδα

Ακαδημία Αθηνών [email protected]

Abstract

In this paper we deal with the modern Greek deponent verbs in -άμαι (θυμάμαι “I remember”, κοιμάμαι “I sleep”, λυπάμαι “I’m sorry”, φοβάμαι “I fear”), which constitute an idiomorphic verbal subsystem of the Β΄ conjugation, and the problems of their categorization and lemmatization in modern Greek grammar books and lexicography. More specifically, in order to propose solutions, 1. we present the morphological change of these verbs, in the frame of the total reconstruction of the Greek verb, and 2. we take into account data of modern Greek dialects and data given from modern corpora.

1. Πρόλογος και δομή εργασίας

Τα νεοελληνικά ρήματα σε -άμαι (κοιμάμαι < κοιμῶμαι, λυπάμαι < λυποῦμαι, φοβάμαι < φοβοῦμαι, θυμάμαι < ἐνθυμοῦμαι) αποτελούν ένα ρηματικό υποσύστημα της Β΄ συζυγίας, το οποίο οι σύγχρονες ν.ε. γραμματικές και λεξικά αντιμετωπίζουν διαφορετικά ως προς α) την κατηγοριοποίησή του (βλ. Τριανταφυλλίδης [1941] 1996, Holton-Mackridge-Warburton 1999, Κλαίρης-Μπαμπινιώτης 2005, Κατσούδα 200911, Χατζησαββίδης-Χατζησαββίδου 2009), β) τη λημματογράφησή του και τις παρεχόμενες γραμματικές πληροφορίες (βλ. ενδεικτικά ΝΕΛ 1995, ΛΚΝ 2007, ΛΝΕΓ 20022, 20083 στο λήμμα λυπώ και λυπάμαι). Οι παραπάνω διαφορές οφείλονται εν μέρει στη μέχρι σήμερα απουσία εξειδικευμένης διαχρονικής και συγχρονικής μελέτης αυτής της ρηματικής υποκατηγορίας που δεν μεταπλάστηκε σε -ιέμαι /-iéme/ στη Νεοελληνική Κοινή (ΝΕΚ), όπως συνέβη σε μεγάλο αριθμό ρημάτων (βλ. Κατσούδα 2007), αλλά απέκτησε κατόπιν μεταπλαστικής διαδικασίας ένα ιδιότυπο ν.ε. κλιτικό παράδειγμα. Στόχος του παρόντος άρθρου είναι ο εντοπισμός των μεθοδολογικών προβλημάτων που προκύπτουν από συγκεκριμένες κατηγοριοποιήσεις και λεξικογραφικές πρακτικές των θυμάμαι, κοιμάμαι, λυπάμαι, φοβάμαι. Γι’ αυτό, στην προσπάθειά μας τόσο να αναδείξουμε αυτά τα προβλήματα όσο και να προτείνουμε λύσεις, οι οποίες μπορούν να βρουν εφαρμογή στη γραμματική θεωρία και στη λεξικογραφία, κρίνουμε απαραίτητο

Syntax - Morphology - Semantics 3 3 5 Major Trends in Theoretical and Applied Linguistics

να λάβουμε υπ’ όψιν τα συμπεράσματα της διαχρονικής μελέτης του υποσυστήματος, τα δεδομένα από νεοελληνικές διαλεκτικές ποικιλίες και τα δεδομένα σύγχρονων ηλεκτρονικών σωμάτων κειμένων. Γι’ αυτό, α) θα παρουσιάσουμε τη μορφολογική μεταβολή αυτών των τεσσάρων ρημάτων, σε σχέση με την αναδόμηση που υπέστησαν συναφείς ρηματικές τάξεις, β) θα μελετήσουμε μορφολογικώς το υποσύστημα σε νεοελληνικές διαλέκτους και ιδιώματα και γ) θα αναζητήσουμε συγκεκριμένους τύπους στο corpus του ΕΘΕΓ και του ΣΕΚ. Στη συγκεκριμένη εργασία θα ακολουθήσουμε το μοντέλο ρηματικής ανάλυσης της Ράλλη (2005: 126-130), μια δομική ανάλυση του ρήματος σε δύο συστατικά μέρη, το ρηματικό θέμα και το κλιτικό επίθημα, με χρήση μηδενικών επιθημάτων, εφόσον αυτή καταστεί απαραίτητη.

2. Η μορφολογική μεταβολή των α.ε. κοιμ-ῶμαι & ἐνθυμ-/λυπ-/φοβ-οῦμαι

Τα αποθετικά1 α.ε. ρήματα κοιμῶμαι, ἐνθυμοῦμαι, λυποῦμαι και φοβοῦμαι κλίνονταν σύμφωνα με τα περισπώμενα ρήματα: το κοιμῶμαι κατά τα -άομαι/ -ῶμαι ρήματα2, ενώ τα υπόλοιπα τρία κατά τα -έομαι/ -οῦμαι. Δηλαδή:

Πίνακας 1. Τα τέσσερα αποθετικά στην α.ε.

[-Συνοπτικό] [+Παρόν]

1.Εν. κοιμ-ῶμαι φοβ-οῦμαι 1.Πληθ. κοιμ-ώμεθα φοβ-ούμεθα

2.Εν. κοιμ-ᾷ φοβ- ῇ/εῖ 2.Πληθ. κοιμ-ᾶσθε φοβ-εῖσθε

3.Εν. κοιμ-ᾶται φοβ-εῖται 3.Πληθ. κοιμ-ῶνται φοβ-οῦνται

[-Συνοπτικό] [+Παρελθόν]

1.Εν. ἐ-κοιμ-ώμην ἐ-φοβ-ούμην 1.Πληθ. ἐ-κοιμ-ώμεθα ἐ-φοβ-ούμεθα

2.Εν. ἐ-κοιμ-ῶ ἐ-φοβ-εῖ 2.Πληθ. ἐ-κοιμ-ᾶσθε ἐ-φοβ-εῖσθε

3.Εν. ἐ-κοιμ-ᾶτο ἐ-φοβ-εῖτο 3.Πληθ. ἐ-κοιμ-ῶντο ἐ-φοβ-οῦντο

1 Και τα τέσσερα ρήματα ήταν αποθετικά ήδη από την α.ε., καθώς διέθεταν μεσοπαθητική μορφολογία, χωρίς αντίστοιχο ενεργητικό τύπο και ήταν από συντακτική άποψη αμετάβατα ή δέχονταν άμεσο αντικείμενο (για την αποθετικότητα βλ. Θεοφανοπούλου 1982: 59, Lavidas- Papangeli 2007: 97, Λαβίδας-Παπαγγελή 2009: 926), π.χ. κοιμήσατο χάλκεον ὕπνον Il. 11. 241,

ἐνθυμούμεθα ὀρθῶς τὰ πράγματα Th. 2,40, τὰς ἐσχάτας λυποῖτο λύπας Pl.Grg 494a, φόβους φοβεῖσθαι 9 Pl.Prt.360b, (βλ. Liddell-Scott 1996 ). 2 Στην ιωνική διάλεκτο το ρήματα κλίνεται κατά τα έω/ -ῶ, π.χ. κοιμέονται (Hp.Judic.43). Πβ. ΛΚΝ 20077.

3 3 6 Syntax - Morphology - Semantics Γεωργία Κατσούδα

Και τα τέσσερα ρήματα διέθεταν στην α.ε. ενεργητικά μεταβιβαστικά ρήματα αντίστοιχης με αυτά κλίσης: κοιμάω/ῶ, ἐνθυμέω/ῶ, λυπέω/ῶ, φοβέω/ῶ. Στους πρώτους μεταχριστιανικούς χρόνους, στην Παλαιά Διαθήκη και στους αιγυπτιακούς παπύρους, γίνεται πλέον αισθητή η σύγχυση ανάμεσα στα -άω και -έω ρήματα [+Ενεργ. Φ.]3, με επικράτηση ΘΦ <ου> αντί <ω> για τα ρήματα σε -άω στο 1.Πληθ./3.Πληθ. [-Συνοπτικό] [+Παρόν/+Παρελθόν], το 1.Εν. [-Συνοπτικό] [+Παρελθόν] και τις μετοχές [-Συνοπτικό] [+ Παρόν] και <α> αντί <ει> για τα ρήματα σε -έω στο 2.Εν./3.Εν. και 2.Πληθ. [-Συνοπτικό] [+ Παρόν/+Παρελθόν], (Mandilaras 1973: 61-62, Horrocks 2006: 458-459, Κατσούδα 2007: 149). Αποτέλεσμα αυτής της διασάλευσης, ήταν η δημιουργία του παρακάτω κοινού κλιτικού παραδείγματος (βλ. πίνακα 2), το οποίο κατά την ύστερη μεσαιωνική εποχή ακολούθησαν πολλά ρήματα σε -έω. Σημειώνουμε, ωστόσο, ότι οι τύποι σε -εῖς, -εῑ και -εῖτε εξακολούθησαν να εμφανίζονται παράλληλα (Horrocks 2006:459):

Πίνακας 2.. Το κοινό κλιτικό παράδειγμα των ρ. σε -άω και -έω

[-Συνοπτικό] 1. Εν. -ῶ [-Συνοπτικό] 1.Εν. -ουν [+Παρόν] [+Παρελθόν] 2.Εν. -ᾷς 2.Εν. -ας

3.Εν. -ᾷ 3.Εν. -α(ν)

1.Πληθ. -οῦμε(ν) 1.Πληθ. -οῦμε(ν)

2.Πληθ. -ᾶτε 2.Πληθ. -ᾶτε

3.Πληθ. -οῦσι/-οῦν(ε) 3.Πληθ. -ουν(ε)

Η σύγχυση ανάμεσα στις δύο τάξεις συνηρημένων ρημάτων επέκταθηκε και στις μεσοπαθητικές δομές [-Συνοπτικό] [+Παρόν], όπως φαίνεται από δημώδη μεσαιωνικά κείμενα. Αρχικά τα ρήματα σε -άομαι/ῶμαι απέκτησαν δομές 1.Εν. και 1.Πληθ./3.Πληθ. σε -οῦμαι, -ούμεθα και -οῦνται αντιστοίχως (βλ. παραδείγματα 1-2), διατηρώντας τις δομές -ᾶμαι, -ᾶται και -ᾶσθε στο 2.Εν./3.Εν. και 2.Πληθ. (Horrocks 2006: 460), π.χ.4:

(1) καὶ ἐγὼ ὡς διὰ τὴν πανεύνοστην ὕπνον οὐδὲν κοιμοῦμαι Διγ. (Jeffr.) 843 (2) Ὕπνον γλυκύ, γλυκύτατον ὁμάδι νὰ κοιμοῦνται Φλώρ. (Ortolá), 1708

3 Εφεξής θα χρησιμοποιούμε τις συντομογραφίες Ενεργ. Φ.= Ενεργητική Φωνή, Μεσοπαθ. Φ.= Μεσοπαθητική Φωνή, ΘΦ= Θεματικό Φωνήεν. 4 Τα παραδείγματα μεσοπαθητικών συνηρημένων ρημάτων των μεσαιωνικών χρόνων προέρχονται από το σώμα κειμένων του Thesaurus Linguae Graecae (TLG), ακολουθώντας, όμως, το σύστημα συντομογραφήσεων του Λεξ. Κριαρά (1969-2008). Ωστόσο, δημιουργήσαμε δικές μας συντομογραφίες (βλ. βιβλιογραφία) για εκείνες τις ελάχιστες περιπτώσεις, όπου μια έκδοση του TLG δεν συμπεριλαμβάνεται στο Λεξ. Κριαρά.

Syntax - Morphology - Semantics 3 3 7 Major Trends in Theoretical and Applied Linguistics

ενώ προσαρμόστηκαν στο διαμορφούμενο κλιτικό παράδειγμα και ρήματα σε -έομαι/ -οῦμαι, συγκαταλεγομένων των ἐνθυμ-/λυπ-/φοβ-οῦμαι (βλ. παραδείγματα 3α-13):

(3α) καὶ ὁρκοπατεῖς τὰ ὤμοσες, τὴν δούλωσιν ἀρνᾶσαι Λίβ. διασκευή α 4166 (3β) Τὰ εὐγενικὰ τοῦ Χάλεπε κοράσια οὐκ ἐνθυμᾶσαι Διγ. (Jeffr.) 236 (3γ) Λέγει με: «Χαῖρε ἀπὸ τοῦ νῦν, τίποτα μὴ λυπᾶσαι Λίβ. διασκευή α 1167 (4α) εἰ δὲ κατὰ βάθους | ὁρμήσῃ, ταχέως κυνηγᾶται Φυσιολ. (Sbord.), 25,4-5 (4β) Πάλιν δὲ τοῦτο ὁ πατὴρ οὕτως ἀπιλογᾶται Φλώρ. (Ortolá) 176 (4γ) Παραπονᾶται το συχνά, ποσῶς οὐδὲν τοῦ ἀρέσει Πόλ. Τρωάδ. (Παπαθωμ.- Jeffr.) 4343 (4δ) ἡ ἀρχόντισσα ἡ εὐγενικὴ κακὸν νὰ μοῦ θυμᾶται Πόλ. Τρωάδ. (Παπαθωμ.- Jeffr.) 3662 (4ε) νὰ μὴν λυπᾶται τίποτε τὰ εἶπεν ὁ πατήρ του Ιμπ. 159 (5α) Ἀγοῦροι μου, ἂν θυμᾶσθε, | τὸ πῶς σᾶς ὑπεξέβαλα ἀπὸ πολλῶν πολέμων Διγ. (Jeffr.) 500-501 (5β) πῶς ἐσεῖς διὰ νὰ φοβᾶσθε τὴν κατάραν τῆς μητρός σας Διγ. Άνδρ. 33230 (5γ) ἐγὼ ἔχω καὶ τὸ ἀδέλφι σας καὶ ὡς δι’ αὐτὴν μὴ λυπᾶσθε Διγ. (Jeffr.) 163.

Από τα παραδείγματα (1)-(5γ) φαίνεται ότι τα κοιμ-/(ἐν)θυμ-/λυπ-/φοβ-οῦμαι απέκτησαν κοινές δομές, ακολουθώντας το παρακάτω διαμορφούμενο κλιτικό παράδειγμα συνηρημένων ρημάτων στην όψιμη μεσαιωνική γλώσσα:

Πίνακας 3. Ο ενεστώτας των τεσσάρων αποθετικών στην ΟΜΕ

[-Συνοπτικό] [+Παρόν] [+Μεσοπαθ. Φ] Όψιμη Μεσαιωνική Ελληνική (ΟΜΕ)

Ενεστώτας

-οῦμαι5 κοιμ-/(ἐν)θυμ-6/λυπ-/φοβ-οῦμαι

-ᾶσαι κοιμ-/(ἐν)θυμ-/λυπ-/φοβ-ᾶσαι

-ᾶται κοιμ-/(ἐν)θυμ-/λυπ-/φοβ-ᾶται

-ούμεθα κοιμ-/(ἐν)θυμ-/λυπ-/φοβ-ούμεθα

-ᾶσθε κοιμ-/(ἐν)θυμ-/λυπ-/φοβ-ᾶσθε

-οῦνται κοιμ-/(ἐν)θυμ-/λυπ-/φοβ-οῦνται 56

5 Το κλιτικό παράδειγμα με λίγες διαφορές μαρτυρείται και στις πρώτες νεοελληνικές γραμματικές και αφορά μόνο τα συνηρημένα ρήματα σε -ῶ/ -ᾶς , Σοφιανός: -οῦμαι, -ᾶσαι, -ᾶται,-ούμεσθεν, -ᾶσθεν, -ῶνται, Germano: -οῦμαι, -ᾶσαι, -ᾶται, -ούμεστεν, -ᾶστε, -οῦνται, Portius: -οῦμαι, -ᾶσαι, -ᾶται, -ούμεσθεν, -ᾶσθε, οῦνται. Για τα συνηρημένα σε -ῶ/-εῖς, βλ πίνακα 4 (i)-(ii) και υποσημείωση 8. 6 Σέ δημώδη μεσαιωνικά κείμενα το ρήμα ἐνθυμοῦμαι εμφανίζεται και χωρίς την πρόθεση ἐν.

3 3 8 Syntax - Morphology - Semantics Γεωργία Κατσούδα

Κοινές δομές απέκτησαν και στον μεσοπαθητικό [-Συνοπτικό] [+Παρελθόν] με ΘΦ <ου> στο 1.Εν., 1.Πληθ./3.Πληθ. και <α> στο 2.Εν./3.Εν., βλ. παραδείγματα (6)-(10):

(6) καὶ ἦτον εὔμορφος ὁ τόπος, ὁποῦ ἐκοιμούμουν Διγ. Άνδρ. 37619 (7) κι ἂν ἐθυμάσου τὰ ἔπαθες εἰς τὴν Πελαγονίαν Χρον. Μορ. Η 5552 (8α) τὸν θά-|νατον τὸν ἐδικόν του ἐθυμᾶτον καὶ ὅλον ἦτον θλιμμένος νυκτὸς Διήγ. Αλ. Ε-F (Konst.), 120, 6-7 (8β) Λοιπὸν ἐλυπᾶτον ἡ ψυχή του Διγ. Άνδρ. 23824 (9) Καὶ ἀπ’ ἐκεῖ ὁποῦ ἐκοι|μούμεσθεν ἐσηκώθη κρυφᾶ ἀπὸ τὸ στρῶμα Διγ. Άνδρ. 36934 (10) μεχρὶ καὶ τοῦ προγεύματος τὴν ὥραν ἐκοιμοῦνταν Αχιλλ. (Smith) Ο 603

Πρέπει, επίσης, να σημειώσουμε ότι την ίδια περίoδο οι ενεργητικοί μεταβιβαστικοί συνηρημένοι τύποι υποχωρούν7 και τη θέση τους καταλαμβάνουν μεταπλασμένοι τύποι σε -ίζω: το ήδη αρχ. κοιμίζω αντικαθιστά το κοιμῶ, το μεσαιων. θυμίζω (< μεταγ. ἐνθυμίζομαι) αντικαθιστά το ἐνθυμῶ και το μεσαιων. φοβίζω το φοβῶ. Ακόμη και το ρήμα λυποῦμαι σχημάτισε μεταβιβαστικό ρήμα λυπίζω που δεν κατόρθωσε όμως να εκτοπίσει το λυπῶ στη ν.ε (βλ. Κριαρά 1969-2008: αντίστοιχα λήμματα). 3. Το υποσύστημα στις νεοελληνικές διαλεκτους/ ιδιώματα

Το μεσοπαθητικό κλιτικό σύστημα των συνηρημένων ρημάτων συνέχισε τις περαιτέρω μεταβολές, που ξεκίνησαν από [-Συνοπτικό] [+Παρελθόν], [+Ενεργ. Φ.] κατά την ύστερη μεσαιωνική εποχή, π.χ. (ἐ)κράτει(ε), (ἐ)κράτειες, (ἐ)κρατειέτε.8 Αν και οι τελευταίοι τύποι δεν επικράτησαν, οδήγησαν σταδιακά στη διαμόρφωση του μεσοπαθητικού κλιτικού παραδείγματος συνηρημένων ρημάτων της σύγχρονης Νέας Ελληνικής (Horrocks 2006: 461-462), όπως μαρτυρούν και οι πρώτες νεοελληνικές γραμματικές του Σοφιανού και του Portius9:

7 Σε δημώδη κείμενα είναι σπάνιοι οι μεταβιβαστικοί τύποι, αν και βρίσκουμε ενεργητικές μετοχές σε -ώντας (π.χ ἔκλαιε πικρῶς ἐνθυμώντας Αλ. Ε-F (Konst.), 78, 8), ο σχηματισμός των οποίων όμως χαρακτηρίζει τα αποθετικά ρήματα (Λαβίδας-Παπαγγελή 2009: 930). 8 Σύμφωνα με τη δομιστική ανάλυση του Μπαμπινιώτη (1972:184-188), οι δομές -ι-έμαι δημιουργήθηκαν από ουρανικόληκτα ρηματικά θέματα με το ενοποιημένο ΘΦ e/i. Το στοιχείο της ουράνωσης θεωρήθηκε αρχικά τμήμα του ΘΦ. Κατόπιν, αφού επεκτάθηκε σ’ όλο το κλιτικό παράδειγμα, με μετατόπιση των ορίων του μορφήματος πέρασε στο τμήμα του θέματος, ώστε τα δύο θεμάτα [± Συνοπτικό] να καταστούν ισοσύλλαβα και φωνηεντόληκτα, π.χ. νικ-έται [nicéte] > νικ-ιέται [nicéte] > νικι-έται [nicéte]. 9 Τα κλιτικά παραδείγματα (i)-(ii) [-Συνοπτικό] [+Παρελθόν] [Μεσοπαθ. Φ.] από τις γραμματικές Σοφιανού και Portius αντιστοίχως αφορούσαν μόνα τα ρήματα -ῶ (-εῖς, -εῖ), ενώ για το αντίστοιχο κλιτικό παράδειγμα των συνηρημένων σε -ῶ (-ᾶς, -ᾶ), βλ. υποσημείωση 5.

Syntax - Morphology - Semantics 3 3 9 Major Trends in Theoretical and Applied Linguistics

Πίνακας 4. Οι δομές [-Συνοπτικό] [+Παρόν] των μεσοπαθητικών συνηρημένων ρ.

[-Συνοπτικό] [+Παρόν] [-Συνοπτικό] [+Παρόν] [-Συνοπτικό] [+Παρόν] Πρώιμη Νέα Ελληνική Πρώιμη Νέα Ελληνική Σύγχρονη Νέα Ελληνική (iii) (i) (ii) (βλ. σύγχρονες ν.ε. (βλ. Σοφιανός) (βλ. Portius) γραμματικές) -οῦ-μαι -ειοῦ-μαι -ι-έ-μαι -ειέ-σαι -ειέ-σαι -ι-έ-σαι -ειέ-ται -ειέ-ται -ι-έ-ται -ού-μεσθα -ειού-μεσθεν -ι-ό-μαστε/-ιού-μαστε -ειέ-στε -ειέ-στε -ι-έ-στε/ -ι-ό-σαστε -οῦ-νται -ειοῦ-νται -ι-ού-νται

Σε πολλές νεοελληνικές διαλεκτικές ποικιλίες τα τέσσερα αυτά ρήματα προσαρμόστηκαν μερικώς ή και ολικώς σε ένα από τα παραπάνω κλιτικά παραδείγματα. Λόγω όμως έλλειψης χώρου, θα αρκεστούμε στην παράθεση παραδειγμάτων προερχομένων από το αρχείο του ΙΛΝΕ. Συγκεκριμένα: α. Οι μεσαιωνικοί τύπου 2.Εν./3.Εν. [-Συνοπτικό] [+Παρόν] σε -άσαι και -άται αντιστοίχως και -άτο(ν) στο 3.Εν. [-Συνοπτικό] [+Παρελθόν] εμφανίζονται ευρύτατα σε ιδιώματα και σε διαλέκτους. Ενδεικτικώς10:

(11α) κιουμ-άσαι [cumáse], κιουμ-άται [cumáte] Καλαβρία (Μπόβα) (11β) φο’-άσαι [foáse], φο’-άται [foáte] Πόντος (Κερασούντα, Σάντα) (11γ) ελυπ-άτον [elipáton] Κύπρος.

Στο αρχείο του ΙΛΝΕ δεν εντοπίσαμε για καμία γεωγραφική ποικιλία τύπους σε -είσαι, -είται, και -είστε για το 2.Εν./3.Εν. και 2.Πληθ. β. Σε αρκετά ιδιώματα -νότια και βόρεια– τα ρήματα θυμάμαι, λυπάμαι και φοβάμαι, ακολούθησαν τις κυρίαρχες για τη ΝΕΚ μορφολογικές δομές σε -ι-έμαι, π.χ.:

(12α) θυμι-έμαι [θimɲéme] Ηλεία (Ήλις), Πρέβεζα (Πάργα), Κως, θυμι-έμι [θimɲémi] Ιωάννινα (Αρτοπούλα), Γρεβενά (Δεσκάτη) (12β) θυμι-έσαι [θimɲése] Παξοί, Αίγινα, θυμι-έσι [θimɲési] Ιωάννινα (Κόνιτσα) (12γ) θυμι-έται [θimɲéte] Θεσπρωτία (Αυλότοπος), αθυμι-έτι [aθimɲéti] Αιτωλοακαρνανία, θυμι-έτι [θimɲéti] Καστελόριζο (12δ) θυμι-όμαστε [θimɲόmaste] Πρέβεζα (Πάργα) (13α) φοβι-έμαι [fovjéme] Πρέβεζα (Πάργα) (13β) φοβι-έσι [fovjési] Ιωάννινα (Λάκκα)

10 Για λόγους οικονομίας, παραθέτουμε σε κάθε αριθμημένο παράδειγμα διαλεκτικούς τύπους, οι οποίοι είτε προέρχονται από την ίδια περιοχή και ανήκουν στο ίδιο κλιτικό παράδειγμα (βλ. παραδείγματα 11α-11β) είτε διαφοροποιούνται γεωγραφικώς, αλλά έχουν κοινά τα χαρακτηριστικά του χρόνου και του προσώπου (βλ. παραδείγματα 12α-17β).

3 4 0 Syntax - Morphology - Semantics Γεωργία Κατσούδα

(13γ) φοβι-ούντι [fovjúdi] Ήπειρος (13δ) φοβι-ότανε [fovjόtane] Πρέβεζα (Πάργα) (14α) λυπι-έμαι [lipçéme] Κεφαλληνία, λυπι-έμι [lipçémi] Ιωάννινα (14β) λυπι-έσαι [lipçése] Απουλία (Μαρτάνο) (14γ) λυπι-έται [lipçéte] Κεφαλληνία, Καστοριά (Βογατσικό), λυπι-έτι [lipçéti] Έβρος (Μάνη) (14δ) λυπι-έστε [lipçéste] Κεφαλληνία (14ε) λυπι-ούνται [lipçúde] Πρέβεζα (Ριζοβούνιον), λυπι-ούντι [lipçúdi] Ιωάννινα (Κόνιτσα), Κοζάνη (Δαμασκηνιά) (14στ) λυπι-όταν [lipçόtan] Θεσπρωτία (Αυλότοπος).

γ. Εύκολα διαπιστώνει κανείς μελετώντας το ρηματικό σύστημα ιδιωμάτων και διαλέκτων, ότι το ΘΦ μεταβάλλεται σε κάποια πρόσωπα σύμφωνα με το αντίστοιχο ΘΦ άλλων προσώπων με παραδειγματική ισοπέδωση και χωρίς φωνολογικές εξαρτήσεις (πρβλ. Ralli 2007)11. Εντύπωση κάνει η διαλεκτικώς μαρτυρημένη γενίκευση του <α>, αναλογικά προς τα 1.Εν./2.Εν./3.Εν. & 2.Πληθ. [-Συνοπτικό] [+Παρόν] εις βάρος των θεματικών φωνηέντων <ο> και <ου>, που μαρτυρεί την τάση επικράτησης αλλόμορφου {-a} για το ρηματικό θέμα [-Συνοπτικό] μετά από μετατόπιση των ορίων του ρηματικού και κλιτικού μορφήματος. Εντοπίσαμε τύπους με ΘΦ <α> για το 3. Πληθ. [-Συνοπτικό] [+Παρόν] (βλ. 15α-15β), για το 3. Εν. [-Συνοπτικό] [+Παρελθόν] (βλ. 16α-16β) και για το 3.Πληθ. [-Συνοπτικό] [+Παρελθόν] (βλ. 17α-17β):

(15α) φοβά-νται [fováde] Μεσσηνία (Ροδιά/ ΙΛΝΕ Λεντεκάδα), φουβά-d’ [fuvád] Λάρισα (Κρυόβρυση) (15β) κοιμά-dαι [cimáde] Μεσσηνία (Τριφυλία) (16α) φουβά-ταν [fuvátan] Φθιώτιδα (16β) κοιμά-τανι [cimátani] Φθιώτιδα, κ’μά-ταν [kmátan] Αιτωλοακαρνανία (Ναύπακτος) (17α) φοβά-νταν12 [fovádan] Πρέβεζα (Γοργόμυλος) (17β) κοιμά-dανι [cimádani] Αιτωλοακαρνανία (Μπούκκα)

11 Η ισοπέδωση της αλλομορφίας στο θεματικό φωνήεν κινήθηκε προς διάφορες κατευθύνσεις στις ν.ε. διαλεκτικές ποικιλίες, φαινόμενο που χρήζει ιδιαίτερης μελέτης. Ενδεικτικά, εντοπίσαμε τύπους με ΘΦ <ου> στο 3.Εν. [-Συνοπτικό] [+Παρόν], π.χ. φο’ούται [foúte] Κάρπαθος (Όλυμπος/ ΙΛΝΕ Έλυμπος), με <ο> στο 1.Εν./ 2.Εν./3.Εν. [-Συνοπτικό] [+Παρόν], π.χ. φουβόμι [fuvόmi], φουβόσι [fuvόsi], φουβότι [fuvόti] Λάρισα (Σκλήθρα). 12 Τόση εντύπωση προκαλεί ο τύπος που ο συλλογέας του υλικού σημείωσε sic μετά τον τύπο.

Syntax - Morphology - Semantics 3 4 1 Major Trends in Theoretical and Applied Linguistics

4. Αιτίες μη περαιτέρω μεταπλασμού των 4 αποθετικών ρημάτων στη ΝΕΚ.

Ενώ ρήματα, όπως εκείνα στα παραδείγματα (3α), (4α), (4β), (4γ), είτε προσαρμόστηκαν στο κυρίαρχο κλιτικό παράδειγμα -ι-έμαι της σύγχρονης ν.ε. (π.χ. αρνιέσαι, απιλογιέσαι, παραπονιέται, κυνηγιέται) είτε υπό την λόγια επίδραση διατηρήθηκαν σε -ούμαι/ -είσαι/ -είται κ.ο.κ. (π.χ. αρνούμαι, απολογούμαι, παραπονούμαι), τα ρήματα θυμ-/ κοιμ-/ λυπ-/ φοβ-ούμαι δεν προσαρμόστηκαν σε κανένα από τα παραπάνω δύο παραδείγματα. Αφού σχηματίστηκε και το 1.Εν. σε -άμαι (θυμ-/ κοιμ-/ λυπ-/ φοβ-άμαι) κατά το 2.Εν./3.Εν. με παραδειγματική ισοπέδωση (leveling), διαμορφώθηκε στη ΝΕΚ το παρακάτω ιδιαίτερο κλιτικό παράδειγμα:

Πίνακας 5. Τα τέσσερα αποθετικά στη ΝΕΚ

[-Συνοπτικό][+Παρόν] [-Συνοπτικό][+Παρελθόν] θυμά-/κοιμά-/λυπά-/φοβά-μαι & θυμ-/κοιμ-/λυπ-/ θυμ-/κοιμ-/λυπ-/φοβ-όμουν(α) φοβ-ούμαι θυμά-/κοιμά-/λυπά-/φοβά-σαι θυμ-/κοιμ-/λυπ-/φοβ-όσουν(α)

θυμά-/κοιμά-/λυπά-/φοβά-ται θυμ-/κοιμ-/λυπ-/φοβ-όταν(ε) θυμ-/κοιμ-/λυπ-/φοβ-όμαστε & θυμ-/κοιμ-/λυπ-/ θυμ-/κοιμ-/λυπ-/φοβ-όμασταν & θυμ-/κοιμ-/λυπ-/ φοβ-ούμαστε φοβ-ούμασταν & θυμ-/κοιμ-/λυπ-/φοβ-όμαστε θυμά-/κοιμά-/λυπά-/φοβά-στε & θυμ-/κοιμ-/λυπ-/ θυμ-/κοιμ-/λυπ-/φοβ-όσασταν & θυμ-/κοιμ-/λυπ-/ φοβ-όσαστε φοβ-όσαστε θυμ-/κοιμ-/λυπ-/φοβ-ούνται & θυμ-/κοιμ-/λυπ-/ θυμ-/κοιμ-/λυπ-/φοβ-όνταν, θυμ-/κοιμ-/λυπ-/φοβ- φοβ-όνται όντουσαν, θυμ-/κοιμ-/λυπ-/φοβ-ούνταν

Το ερώτημα που τίθεται είναι γιατί αυτή η ρηματική υποκατηγορία δεν προσαρμόστηκε στο κλιτικό παράδειγμα του πίνακα 4 (iii) τόσο στη ΚΝΕ όσο και σε διαλεκτικές ποικιλίες εκτός Ελλαδικού χώρου, σε περιοχές δηλ. μακριά από την επίδραση της Κοινής (βλ. κεφ. 3.α); Ο Χατζιδάκις (1905: Α΄, 47) θεωρεί ότι τα ρήματα σε -ούμαι, -άσαι, -άται προχώρησαν σε επιπλέον μορφολογική μεταβολή, για να διακρίνεται το 2.Πληθ. [-Συνοπτικό] [+Παρόν] [+Ενεργ. Φ.] από τον αντίστοιχο 3.Εν. [-Συνοπτικό] [+Παρόν] [+Μεσοπαθ. Φ.], δηλ. κρατᾶτε= κρατᾶται αλλά κρατᾶτε ≠ κρατιέται. Μια τέτοια διαφοροποίηση δεν απαιτούσαν τα αποθετικά ρήματα, γι’ αυτό και διατηρήθηκαν, π.χ. παραπονούμαι (-ᾶσαι, -ᾶται), ἀρνοῦμαι (-ᾶσαι, -ᾶται)13, θυμοῦμαι. Ο Χατζιδάκις, συνδέει δηλαδή την αποθετικότητα των ρημάτων με τη διατήρηση του μεσαιωνικού κλιτικού παραδείγματος.

13 Οι συγκεκριμένοι τύποι μαρτυρούνται και στον Jannaris 1897: 218.

3 4 2 Syntax - Morphology - Semantics Γεωργία Κατσούδα

Όμως, επειδή, εκτός από τα θυμά-/ κοιμά-/ λυπά-/ φοβά-μαι, άλλα αποθετικά ρήματα, όπως παραπονούμαι, αρνούμαι κ.ά., υπέστησαν περαιτέρω μεταβολή στη ΝΕΚ, υποστηρίζουμε ότι η αποθετικότητα δεν ήταν ο μοναδικός λόγος διατήρησης του μεσαιωνικού κλιτικού προτύπου. Λέξεις του βασικού λεξιλογίου –των οποίων η κειμενική εμφάνιση είναι ιδιαίτερα μεγάλη– δύσκολα υπόκεινται σε περαιτέρω αλλαγές (βλ. Bynon 1977: 42-43). Χαρακτηριστικά αναφέρουμε ότι στη λίστα Swadesh το κοιμάμαι εμπεριέχεται στις 100 συνηθέστερες λέξεις (http://en.wikipedia.org/wiki/ Swadesh_list#Shorter_lists). Όμως στην Ελληνική δεν έχουν γίνει μέχρι σήμερα λεξικοστατιστικές μελέτες για το βασικό λεξιλόγιο της από άποψη περιεχομένου14. Ωστόσο, τα αριθμητικά δεδομένα από τη διαδικτυακή αναζήτηση του 1. Εν. των θυμάμαι, κοιμάμαι, λυπάμαι, φοβάμαι στο www.google.gr –παρά τα μεθοδολογικά προβλήματα που χαρακτηρίζουν τέτοιες αναζητήσεις (βλ. Τράπαλης-Οικονομίδης 2007: 4)– μαρτυρούν την ευρεία κειμενική χρήση τους: λυπάμαι (1.120.000 εμφανίσεις), φοβάμαι (2.333.000 εμφανίσεις), κοιμάμαι (690.000 εμφανίσεις), θυμάμαι (3.320.000 εμφανίσεις). Το παραπάνω πιθανόν να εξηγεί τον μη περαιτέρω μεταπλασμό τους. Πάντως, και η αποθετικότητα των εν λόγω ρημάτων πρέπει να έπαιξε με τη σειρά της σημαντικό ρόλο, αφού ενίσχυσε τη μορφολογική τους απομόνωση και συνέβαλε στη διατήρηση του παλαιότερου κλιτικού παραδείγματος.

5. Τα τέσσερα αποθετικά στην αναδομημένη Β΄ Συζυγία της ΝΕΚ

Σύμφωνα με τους Μπαμπινιώτη (1972) και Κατσούδα (2007), στο πλαίσιο της ολικής αναδόμησης του α.ε., η δυναμική {a}, μια από τις δυναμικές, σύμφωνα με την δομιστική μελέτη των ρηματικών μεταπλασμών, οδήγησε σταδιακά στη δημιουργία της πολυπληθούς ρηματικής κατηγορίας -άω, που απορρόφησε ρήματα διαφορετικών ρηματικών τάξεων της α.ε., π.χ. πετά-ννυ-μι> πετά-ω, κρατ-έω> κρατά-ω, μεθύ-ω> μεθά-ω. Σε αντιδιαστολή προς τα συμφωνόληκτα θέματα των ρημάτων σε -ω, που αποτελούν την Α΄ συζυγία, ή την Κλιτική Τάξη 1 (ΚΤ1) σύμφωνα με την Ράλλη (2005: 133)15, τα -άω ρήματα της Β1 συζυγία –ή της Κλιτικής Τάξη 2 (ΚΤ2) σύμφωνα με τη Ράλλη (2005: 133-136)– διαθέτουν φωνηεντόληκτα θέματα και χαρακτηρίζονται από αλλομορφική ποικιλία Χ(α) ~ Χφωνήεν, για τις δομές

14 Οι ελληνικές λεξικοστατιστικές έρευνες μελετούν συνήθως τη συχνότητα εμφάνισης των λέξεων. Π.χ. στις 100 πιο συχνές λέξεις του ΕΘΕΓ ανήκουν λέξεις, όπως και, εγώ, θα, μπορώ, λέγω αλλά και λέξεις όπως κυβέρνηση, πολιτική, Ελλάς, δίδω, χρόνιος κ.ά. 15 Σύμφωνα με τη Ράλλη (2005: 132), ο νόμος προσδιορισμού συνεμφανιζόμενων χαρακτηριστικών όπου ένα ρηματικό θέμα στερείται ή μη συστηματικής αλλομορφικής ποικιλίας, καθιστά προβλέψιμη μια ρηματική τάξη.

Syntax - Morphology - Semantics 3 4 3 Major Trends in Theoretical and Applied Linguistics

[-Συνοπτικό] [+Ενεργητ. Φ.] και [+Συνοπτικό] [+Ενεργητ. Φ.] αντιστοίχως, π.χ. αγαπά-ω/αγαπ-ώ: αγάπα(γ)-α/αγαπ-ούσα: αγάπη-σα. Τη Β2 συζυγία –ή την Κλιτική Τάξη 2β (ΚΤ2β) σύμφωνα με τη Ράλλη ό.π. – αποτελούν ρήματα με αλλομορφική ποικιλία Χ ~ Χφωνήεν, για τις δομές [-Συνοπτικό] [+Ενεργητ. Φ.] και [+Συνοπτικό] [+Ενεργητ. Φ.] αντιστοίχως, π.χ. θεωρ-ώ: θεωρ-ούσα: θεώρη-σα. Τα 4 ρήματα, όπως φάνηκε από τη διαχρονική τους μελέτη, αν και [-λόγια] δεν μεταπλάστηκαν τελικώς σε -ι-έμαι στη ΝΕΚ πιθανόν για τους λόγους που είδαμε παραπάνω. Εντούτοις, η δυναμική {a}, που δημιούργησε τη Β1 συζυγία, είναι εμφανής στο συγκεκριμένο υποσύστημα, καθώς το <α> επικρατεί σε τέσσερα πρόσωπα του κλιτικού παραδείγματος [-Συνοπτικό] [+Παρόν] και απ’ ό,τι φαίνεται, και από το διαλεκτικό υλικό (βλ. παραπάνω 3γ.), αυτό τείνει να επικρατήσει και στους υπόλοιπους τύπους του [-Συνοπτικού] [+Παρόν/+Παρελθόν]. Αν συγκρίνουμε μάλιστα το κλιτικό παράδειγμα του υποσυστήματος [-Συνοπτικό] [+Παρόν] [+ Μεσοπαθ. Φ.] με το κλιτικό παράδειγμα της Β1 συζυγίας, [-Συνοπτικό] [+Παρόν] [+ Ενεργ. Φ.] (βλ. Πίνακα 6), παρατηρούμε παρόμοια κατανομή της αλλομορφικής ποικιλίας του θέματος [-Συνοπτικό]. Εξάλλου, παρατηρούμε την ίδια αλλομορφική ποικιλία του ρηματικού κλιτικού επιθήματος, συγκρίνοντας τις δομές [-Συνοπτικό] [+Παρελθόν] [+Μεσοπαθ. Φ.] του υποσυστήματος με τις αντίστοιχες της Β1 συζυγίας.

Πίνακας 6. Σύγκριση δομών των τεσσάρων αποθετικών με τη Β1 συζυγία

[+ Ενεργ. Φ.] [+Μεσοπαθ. Φ.] [+ Μεσοπαθ. Φ.] [-Συνοπτικό] [-Συνοπτικό] [-Συνοπτικό] [+Παρόν] [+Παρόν] [+Παρελθόν] [+Παρελθόν]

Ι ΙΙ Ι ΙΙ

αγαπά-ω/ αγαπ-ώ θυμά-μαι/ θυμ-ούμαι αγαπι-όμουν θυμ-όμουν

αγαπά-ς θυμά-σαι αγαπι-όσουν θυμ-όσουν

αγαπά-(ει) θυμά-ται αγαπι-όταν θυμ-όταν αγαπι-όμασταν/ θυμ-όμασταν/ θυμ- αγαπά-με/ αγαπ-ούμε θυμ-όμαστε αγαπι-όμαστε όμαστε αγαπι-όσασταν/ θυμ-όσασταν/ θυμ- αγαπά-τε θυμά-στε/ θυμ-όσαστε αγαπι-όσαστε όσαστε αγαπι-όνταν/ αγαπι- θυμ-όνταν/ θυμ- αγαπά-νε/ αγαπ-ούν(ε) θυμ-ούνται όντουσαν/ αγαπι- όντουσαν/ θυμ-ούνταν ούνταν

Από τα παραπάνω, υποστηρίζουμε ότι το υποσύστημα των τεσσάρων αυτών ρημάτων είναι αποτέλεσμα της ίδιας δυναμικής {a} που οδήγησε στη δημιουργία φωνηεντόληκτων θεμάτων [-Συνοπτικό]. Επομένως, το <α> που εμφανίζεται στον 1. Εν./2. Εν./3.Εν. και 2.Πληθ. δεν ανήκει στο κλιτικό επίθημα, όπως υποστηρίζουν οι Κλαίρης-Μπαμπινιώτης (2005: 522), αλλά ανήκει στο ρηματικό θέμα. Δεδομένου ότι το κλιτικό παράδειγμα

3 4 4 Syntax - Morphology - Semantics Γεωργία Κατσούδα

των ρημάτων -άμαι εμφανίζει ίδια αλλομορφική κατανομή με τα ρήματα σε -άω, (Χ(α) ~ Χ(φωνήεν), π.χ. αγαπ(ά)-ω ~ αγάπη-σα // θυμά-μαι κ. θυμ-ούμαι ~ θυμή-θη-κα) το ρηματικό υποσύστημα των ρημάτων σε -άμαι μπορεί να ενταχθεί ως υποομάδα της Β1 συζυγίας (ή της ΚΤ2 σύμφωνα με τη Ράλλη 2005).

6. Το υποσύστημα σε ν.ε. γραμματικές και λεξικά

6.1. Ως προς τη γραμματική κατηγοριοποίησή τους

α) Σύμφωνα με τη Γραμματική Τριανταφυλλίδη ([1941] 1996: 337, 340) η συγκεκριμένη υποομάδα ρημάτων εντάσσεται στη Β2 συζυγία της παθητικής φωνής. Μετά το κλιτικό παράδειγμα του θεωρώ-θεωρούμαι (-είσαι, -είται κ.ο.κ.) βρίσκουμε το αντίστοιχο αυτού του υποσυστήματος. Αν και ήδη έχουν συνταχθεί αρκετά σύγχρονα εγχειρίδια γραμματικής, η συγκεκριμένη κατηγοριοποίηση υιοθετείται και από σύγχρονα γλωσσικά εγχειρίδια της δευτεροβάθμιας εκπαίδευσης16, β) Αντιθέτως, η γραμματική Κλαίρη-Μπαμπινιώτη (2005: 522), Holton-Mackridge-Warburton (1999: 146-148), Χατζησαββίδης (2009: 251-253), Χατζησαββίδης-Χατζησαββίδου (2009), χαρακτηρίζουν τα τέσσερα ρήματα ως αποθετικά της Β΄ συζυγίας, χωρίς να αξιοποιούνται πλήρως όλα αυτά τα μορφολογικά χαρακτηριστικά του υποσυστήματος, που θα οδηγούσαν στην ένταξή του σε συγκεκριμένη ρηματική τάξη. Οι παραπάνω κατηγοριοποιήσεις δεν ικανοποιούν ιστορικά, μορφολογικά αλλά και παιδαγωγικά-διδακτικά κριτήρια. Συγκεκριμένα, οι εν λόγω προτάσεις κατηγοριοποίησης: 1. από διαχρονική άποψη, αδυνατούν να αποτυπώνουν σωστά το αποτέλεσμα των μεταπλαστικών διαδικασιών που υπέστη η υποκατηγορία των -άμαι ρημάτων. Βάσει όσων είπαμε παραπάνω, το θέμα των φοβούμαι, λυπούμαι, (εν)θυμούμαι υπέστη μεταπλασμό. Μετά από έναν αλλοκατηγοριακό όμως μεταπλασμό17, όπως συνέβη εν προκειμένω, το μεταπλαστικό προϊόν αλλάζει κλιτική κατηγορία. Έτσι, από τη στιγμή που τα τέσσερα ρήματα μεταπλάστηκαν με εμφανή τη δυναμική {a} σε τέσσερα πρόσωπα (1.Εν./2.Εν./3. Εν. & 2.Πληθ.) (βλ. παραπάνω), δεν εξακολουθούν να ανήκουν στην τάξη των -ώ (-είς, -εί)/-ούμαι ρημάτων. Αυτό ισχύει πολύ περισσότερο για το ρήμα κοιμάμαι, το οποίο και προ μεταπλασμού δεν ανήκε στην κλιτική τάξη των -ώ (-είς, -εί)/-ούμαι ρημάτων. 2. Δεν περιγράφουν σε συγχρονικό επίπεδο την αλλομορφική ποικιλία του

16 Η κατηγοριοποίηση των συζυγιών και του υποσυστήματος των τεσσάρων αποθετικών ρημάτων στο βιβλίο του μαθητή της Νεοελληνικής Γλώσσας Β΄ Γυμνασίου (: 51) βασίζεται στη γραμματική Τριανταφυλλίδη, βλ. http://www.pi-schools.gr/books/gymnasio/. 17 Για τον όρο, βλ. Κατσούδα 2007: 50-51.

Syntax - Morphology - Semantics 3 4 5 Major Trends in Theoretical and Applied Linguistics

υποσυστήματος. Όπως ήδη υποστηρίξαμε, το υποσύστημα των ρημάτων -ά-μαι έχει παρόμοια αλλομορφική κατανομή με τα ρήματα της Β1 συζυγίας (ΚΤ2 σύμφωνα με τη Ράλλη). Θα μπορούσε κάποιος να προβάλει ως αντεπιχείρημα ότι τύποι 1.Εν. (θυμ-/ κοιμ-/λυπ-/φοβ-ούμαι) και 3.Πληθ. (θυμ-/κοιμ-/λυπ-/φοβούνται) ακολουθούν το κλιτικό πρότυπο της Β2 συζυγίας (ΚΤ2β σύμφωνα με τη Ράλλη). Αυτό εύκολα μπορούμε να το αντικρούσουμε: καταρχάς, οι τύποι -ούμαι και -ούνται δεν είναι οι μοναδικοί. Αντιθέτως είναι παράλληλοι τύποι που συνυπάρχουν με τους αντίστοιχους -άμαι και -όνται, οι οποίοι δεν απαντούν αντιστοίχως στο κλιτικό παράδειγμα του θεωρούμαι. Επίσης, μην ξεχνάμε ότι ακόμη και η ίδια η Β1 συζυγία διαθέτει παράλληλους τύπους στο 1.Πληθ. και στο 3.Πληθ., οι οποίοι συμπίπτουν μορφολογικώς με τους αντίστοιχους της Β2 συζυγίας, αγαπά-με // αγαπ-ούμε (= θεωρ-ούμε), αγαπά-νε // αγαπ-ούν (= θεωρ-ούν). 3. Έχουν μειονεκτήματα και από παιδαγωγική-διδακτική άποψη, επειδή οι μαθητές διδάσκονται ότι το χαρακτηριστικό φωνήεν της πρώτης τάξης είναι το <α> ενώ της δεύτερης το <ει>. Επειδή, σύμφωνα με όσα υποστηρίξαμε στο κεφ. 5, το υποσύστημα των τεσσάρων αυτών ρημάτων [+Μεσοπαθητ. Φ.] εμφανίζει την αλλομορφική κατανομή Χ(α) (θυμά-μαι/θυμ-ούνται) ~ Χφωνήεν (θυμή-θηκ-α), καλό είναι να ενταχθεί ως υποκατηγορία στην Β1 συζυγία (ΚΤ2 τάξη στη Ράλλη), όπως έχει ήδη προταθεί από την Κατσούδα (200911: 194). Τα οφέλη μιας τέτοιας κατηγοριοποίησης είναι πολλά, γιατί 1. διαχρονικώς, αποτυπώνεται καλύτερα το αποτέλεσμα της μορφολογικής μεταβολής που υπέστησαν αυτά τα 4 [-λόγια] ρήματα 2. σε συγχρονικό επίπεδο περιγράφεται καλύτερα η αλλομορφική κατανομή, 3. παιδαγωγικώς-διδακτικώς βοηθά τους μαθητές να εντάξουν το υποσύστημα σε ένα γενικότερο σύστημα που εμφανίζει το χαρακτηριστικό {a} στο ρηματικό θέμα [-Συνοπτικό].

6.2. Ως προς τη λημματογράφησή τους

Το σημαντικότερο πρόβλημα που αντιμετωπίζουμε ως προς τη λημματογράφηση των ρημάτων σε -άμαι είναι ποια είναι η λέξη-κεφαλή του λήμματος. Τα ΝΕΛ (1995), ΛΚΝ (20077) όσο και το ΛΝΕΓ (20022) εντάσσουν το λυπάμαι/ λυπούμαι σε λήμμα όπου λέξη-κεφαλή τίθεται ο τύπος λυπώ. Μια τέτοια λημματογράφηση δεν μπορεί όμως να ισχύσει, καθώς το ρ. λυπάμαι, όπως άλλωστε είδαμε και παραπάνω, διατήρησε την αποθετικότητά του από την α.ε. μέχρι και σήμερα. Το λυπώ είναι το μεταβιβαστικό ρήμα του λυπάμαι και λειτουργεί σημασιολογικά όπως ακριβώς τα κοιμίζω, θυμίζω και φοβίζω των αντίστοιχων αποθετικών ρημάτων κοιμάμαι, θυμάμαι και φοβάμαι, τα οποία λημματογραφούνται χωριστά. Εξάλλου, το λήμμα, μιας κλιτής λέξης περιλαμβάνει όλο το σύνολο των τύπων που ανήκουν σε ένα κοινό κλιτικό παράδειγμα: τόσο το λύνω, λύνεις, λύνει, όσο το λύνομαι, λύνεσαι κ.ο.κ. ανήκουν στην Α΄ συζυγία (ΤΚ1 στη Ράλλη). Το παραπάνω δεν ισχύει στην περίπτωση των ρημάτων λυπώ και λυπάμαι, γιατί, εκτός του ότι και τα δύο είναι μεταβατικά, δεν ανήκουν στην ίδια ρηματική τάξη. Και τα δύο ανήκουν στη Β΄ συζυγία,

3 4 6 Syntax - Morphology - Semantics Γεωργία Κατσούδα

αλλά το πρώτο κλίνεται κατά τη Β2 (ΤΚ2β στη Ράλλη) και το δεύτερο, κατά τη Β1 συζυγία (ΤΚ2 στη Ράλλη). Το ΛΝΕΓ (20083) διορθώνει την προηγούμενη έκδοσή του (20022), συντάσσοντας δύο διαφορετικά λήμματα για το ρήμα λυπώ και το λυπάμαι. Δεδομένου ότι τα ρήματα σε -ά-μαι διαθέτουν και διτυπίες, πρόσθετο πρόβλημα της λημματογράφησης είναι ποιος τύπος θα τεθεί ως λέξη-κεφαλή.Όπως φαίνεται στον παρακάτω πίνακα, τα νεοελληνικά λεξικά τοποθετούν το ίδιο τύπο ως λέξη-κεφαλή μόνο στο λήμμα κοιμάμαι:

Πίνακας 7. Η λημματογράφηση των τεσσάρων αποθετικών σε λεξικά της ΝΕΚ

ΝΕΛ (1995) θυμούμαι & θυμάμαι κοιμάμαι & κοιμούμαι φοβούμαι & φοβάμαι

ΛΚΝ (20077) θυμάμαι & θυμούμαι κοιμάμαι & κοιμούμαι φοβούμαι & φοβάμαι

ΛΝΕΓ (20022) θυμάμαι & θυμούμαι κοιμάμαι & κοιμούμαι φοβούμαι & φοβάμαι (υποτεταγμένο) (υποτεταγμένο) (υποτεταγμένο) ΛΝΕΓ (20083 και θυμάμαι & θυμούμαι κοιμάμαι & κοιμούμαι φοβάμαι & φοβούμαι 20124) (υποτεταγμένο) (υποτεταγμένο) (υποτεταγμένο)

Η λημματογράφηση των ΝΕΛ (1995), ΛΚΝ (20077) και ΛΝΕΓ (20022) δεν στηρίζεται σε δεδομένα των σύγχρονων σωμάτων κειμένων (ΕΘΕΓ, ΣΕΚ). Αναζητήσαμε τους τύπους -άμαι και -ούμαι στο σώμα κειμένο του ΕΘΕΓ και του ΣΕΚ. Τα αποτελέσματα της αναζήτησης αποδεικνύουν την αριθμητική υπεροχή των τύπων σε -άμαι:

Πίνακας 8. Δεδομένα από ΕΘΕΓ (Ε) και ΣΕΚ (Σ)

Τύποι Ε % Σ % Τύποι Ε % Σ %

κοιμάμαι 54 97, 73 47 90,38 λυπάμαι 647 79,28 84 79,28

κοιμούμαι 3 2,27 5 9,62 λυπούμαι 169 20,72 54 20,72

θυμάμαι 1806 97,41 386 89,55 φοβάμαι 1079 75,87 239 79,66

θυμούμαι 48 2,59 45 10,45 φοβούμαι 343 24,13 61 20,34

Βάσει των παραπάνω δεδομένων, θεωρούμε ότι πρέπει να λημματογραφούνται οι τύποι σε -άμαι και να ακολουθούν οι διτυπίες σε -ούμαι ή να αποτελούν υποτεταγμένα λήμματα, όπως ήδη κάνει το ΛΝΕΓ (20083 και 20124).

7. Συμπεράσματα-Λύσεις

Η παρακολούθηση της μορφολογικής μεταβολής μέσα στο πλαίσιο της αναδόμησης των ρημάτων σε -ῶ, τα δεδομένα από διαλεκτικές ποικιλίες και υλικό από τα σύγχρονα σώματα κειμένων μάς προσφέρουν μια ολοκληρωμένη εικόνα του ν.ε. υποσυστήματος

Syntax - Morphology - Semantics 3 4 7 Major Trends in Theoretical and Applied Linguistics

αποθετικών ρημάτων της Β΄ συζυγίας, που επιτρέπει να προτείνουμε λύσεις για την σωστότερη από ιστορική, μορφολογική και παιδαγωγική άποψη κατηγοριοποίηση και λημματογράφησή του στις σύγχρονες ν.ε. γραμματικές και λεξικά. Επομένως, τα εγχειρίδια γραμματικής θα πρέπει να συσχετίζουν το ρηματικό υποσύστημα των ρημάτων σε -άμαι με τα ρήματα σε -άω. Αν δεν ακολουθούν την ταξινόμηση της Ράλλη (2005), αλλά διατηρούν κατηγοριοποιήσεις σε συζυγίες και τάξεις, καλό θα ήταν να θεωρήσουν αυτά τα αποθετικά ρήματα σε υποκατηγορία της Β1 συζυγίας (ρήματα σε -άω), κατηγοριοποίηση που έχει ήδη προταθεί από την Κατσούδα (200911: 194). Αλλά και λεξικογραφικώς, οφείλουμε να λάβουμε υπ’ όψιν μας συγκεκριμένα κριτήρια για την ορθότερη λημματογράφησή τους. Βάσει των ποσοτικών δεδομένων που προκύπτουν από τα σύγχρονα σώματα κειμένων, πρέπει να προταχθούν ως λέξεις-κεφαλές οι τύποι σε -άμαι που υπερέχουν αριθμητικώς και να υποταχθούν οι αντίστοιχοι σε -ούμαι, οι οποίοι χαρακτηρίζονται από σαφώς μικρότερη κειμενική εμφάνιση. Από την άλλη, συγκεκριμένα σημασιοσυντακτικά χαρακτηριστικά, όπως η αποθετικότητά τους και η μεταβατικότητά τους, αλλά και μορφολογικά κριτήρια επιβάλλουν τη χωριστή λημματογράφηση των τεσσάρων αυτών ρημάτων από τα αντίστοιχα μεταβιβαστικά ρήματά τους.

3 4 8 Syntax - Morphology - Semantics Γεωργία Κατσούδα

Βιβλιογραφία

Bynon, T. 1977. Historical linguistics. Cambridge: Cambridge University Press.

Θεοφανοπούλου, Δ. 1982. Τα μέσα ρήματα της Νέας Ελληνικής: μερικές προκαταρκτικές παρατηρήσεις στο σύστημα των διαθέσεων. Μελέτες για την Ελληνική Γλώσσα, Πρακτικά της 2ης Ετήσιας Συνάντησης του Τμήματος Γλωσσολογίας της Φιλοσοφικής Σχολής του Αριστοτελείου Πανεπιστημίου Θεσσαλονίκης. (4-6 Μαΐου 1981), 51-78. Θεσσαλονίκη: Αφοί Κυριακίδη.

Germano, G. [1622] 1907. Vocabolario italiano et greco. Ρώμη: H. Pernot.

Holton, D., P. Mackridge και Ε. Φιλιππάκη-Warburton. 1999. Γραμματική της Ελληνικής Γλώσσας (μετάφρ. Β. Σπυρόπουλος). Αθήνα: Πατάκης.

Horrocks, G. 2006. Ελληνικά. Ιστορία της γλώσσας και των ομιλητών της (μετάφρ. Μ. Σταύρου και Μ. Τζεβελέκου). Αθήνα: Εστία [Greek: A History of the Language and its Speakers, Longman 1997].

Iνστιτούτο Νεοελληνικών Σπουδών (Ίδρυμα Mανόλη Tριανταφυλλίδη). 20077. Λεξικό της Kοινής Nεοελληνικής. Θεσσαλονίκη: Aριστοτέλειο Πανεπιστήμιο Θεσσαλονίκης. =ΛΚΝ

Jannaris, A. 1897. An Historical Greek Grammar. Hildesheim- Zürich-New York: Georg Olms.

Jeffreys, E. 1998. Digenis Akritis. The Grottaferrata and Escorial versions [Cambridge Medieval Classics 7, 238-374]. Cambridge: Cambridge University Press. = Διγ. (Jeffr.)

Κατσούδα, Γ. 2007. Ρηματικοί Μεταπλασμοί (Μεταδομήσεις ενεστωτικού και αοριστικού θέματος), Διδακτορική Διατριβή. Αθήνα: Βιβλιοθήκη της Σοφίας Σαριπόλου. Εθνικό και Καποδιστριακό Πανεπιστήμιο Αθηνών. Φιλοσοφική Σχολή.

Syntax - Morphology - Semantics 3 4 9 Major Trends in Theoretical and Applied Linguistics

Κατσούδα, Γ. 200911. Σύγχρονη πρακτική γραμματική τσέπης. Αθήνα: Άγκυρα.

Kλαίρης, X. και Γ. Μπαμπινιώτης (Σε συνεργασία με τους A. Mόζερ, A. Mπακάκου- Oρφανού, Σ. Σκοπετέα). 2005. Γραμματική της Nέας Eλληνικής, Δομολειτουργική- Eπικοινωνιακή. Αθήνα: Ελληνικά Γράμματα.

Kριαράς, E. 1969-2011. Λεξικό της Mεσαιωνικής Eλληνικής Δημώδους Γραμματείας, τ. A΄-ΙΖ΄ Θεσσαλονίκη.

Kριαράς, E. 1995. Νέο Ελληνικό Λεξικό της Σύγχρονης Eλληνικής Δημοτικής Γλώσσας. Αθήνα: Eκδοτική Aθηνών. =ΝΕΛ

Λαβίδας, Ν. και Δ. Παπαγγελή. 2009. Αποθετικότητα, μη παρεμφατικότητα και χρόνος στην Ελληνική. Στα Πρακτικά του 8ου Διεθνούς Συνεδρίου Ελληνικής Γλωσσολογίας (Ιωάννινα, 29 Αυγούστου - 2 Σεπτεμβρίου 2007), 926-940 (http://www.linguist-uoi.gr/ cd_web/docs/greek/024_lavidasICGL8_OK.pdf).

Lavidas, N. και D. Papangeli. 2007. Deponency in the Diachrony of Greek. In the Proceedings of the British Academy 145, 97-126.

Liddell, H.G. και R. Scott. 19969. Greek-English Lexicon (with a revised supplement) (Ninth edition with Revised Supplement by H. Stuart Jones και R. McKenzie). Oxford: Clarendon Press.

Mandilaras, B. 1973. The verb in the Greek non-literary papyri. Athens: Ministry of Culture and Science.

Mπαμπινιώτης, Γ. 1972. Tὸ Ρῆμα τῆς Ἑλληνικῆς (Δομικαὶ ἐξελίξεις καὶ συστηματοποίησις τοῦ ρήματος τῆς Ἑλληνικῆς [Ἀρχαίας καὶ Nέας]). Ἀθήνα.

Mπαμπινιώτης, Γ. 1998, 20022, 20083, 20124. Λεξικό της Nέας Eλληνικής Γλώσσας. Aθήνα: Κέντρo: Λεξικολογίας. =ΛΝΕΓ

Ortolá-Salas, F.J. 1998. Florio y Platzia Flora: una novela bizantina de época paleólogica [Nueva Roma 6, 108-202 Madrid: Universidad de Cádiz. = Φλώρ. (Ortolá)

Portius, S. [1638] 1889. Grammatica linguae grecae vulgaris. Εκδ. W. Meyer. Paris: Vieweg.

Ράλλη, Α. 2005. Μορφολογία. Αθήνα: Πατάκης.

3 5 0 Syntax - Morphology - Semantics Γεωργία Κατσούδα

Ralli, A. 2007. On the role of Allomorphy in Inflectional Morphology: Evidence from Dialectal Variation. In G. Sica (ed.), Open problems in Linguistics and Lexicography. Milano: Polimetrica.

Σοφιανός, Ν. 18742. Γραμματική της κοινής των Ελλήνων γλώσσης = Grammaire du grec vulgaire et trduction en grec vulgaire du traité de Plutarque sur l’education des enfants. Εκδ. É. Legrand. Παρίσι: Maisonneuve.

Τράπαλης, Γ. και Α. Οικονομίδης. 2007. Κριτήρια επιλογής παραδειγμάτων στα μονόγλωσσα λεξικά. Στα Πρακτικά του 7ου Διεθνούς Συνεδρίου Ελληνικής Γλωσσολογίας (8-10 Σεπτεμβρίου 2005), Πανεπιστήμιο York, 1-13 (http://icgl7.icte.uowm.gr/ Trapalis&Oikonomidis.pdf).

Tριανταφυλλίδης, Μ. [1941] 1996. Νεοελληνική Γραμματική (της δημοτικής). (ανατύπωση της έκδοσης του ΟΕΣΒ (1941) με διορθώσεις). Ίδρυμα Μανόλη Τριανταφυλλίδη [Ινστιτούτο Νεοελληνικών Σπουδών].

Χατζησαββίδης, Σ. και Α. Χατζησαββίδου 2009. Γραμματική της Νέας Ελληνικής γλώσσας Α΄, Β΄, Γ΄ Γυμνασίου. Παιδαγωγικό Ινστιτούτο.

Χατζησαββίδης, Σ. 2009. Γραμματική της Νέας Ελληνικής. Θεωρητικές Βάσεις και Περιγραφή. Τόμος Α΄ (Φωνητική-Φωνολογία-Μορφολογία). Θεσσαλονίκη: Βάνιας.

Χατζιδάκις, Γ. 1905. Μεσαιωνικὰ καὶ Νέα, τ. Α΄. Ἀθῆναι.

Σώματα κειμένων

www.ilsp.gr (Eθνικός Θησαυρός Ελληνικής Γλώσσας = ΕΘΕΓ)

www.sek.edu.gr (Σώμα Ελληνικών Κειμένων= ΣΕΚ).

Syntax - Morphology - Semantics 3 5 1 Major Trends in Theoretical and Applied Linguistics

3 5 2 The Syntax-IS Interface: On the Functional Discrepancies Between Clitic Left Dislocation and ‘Bare Left Dislocation’ in Modern Greek

Axiotis Kechagias

UCL & University of Salford [email protected]

Abstract

Only in CLLD does the dislocated DP perform the discourse function [Topic] partitioning the utterance into [Topic] + [Comment]; cliticless non-focal dislocated DPs are solely fronted [Ground] information that partition the utterance into [Ground] + [New]. In other words, the two constructions correspond to two different information packaging strategies and grammaticalise two distinct Information Structure interface rules.

1. Preliminaries

Until the late nineties there had been a consensus in the literature that Greek patterned along languages like Italian and Spanish in that left dislocation of non- focal objects is not permissible unless a pronominal clitic resumed the dislocated DP (1-3), unlike English, where this ‘un-resumed’ fronting is also a possibility (4):

(1) tin brizola *(tin) efaγa Greek the steak it-CL ate-1s

(2) la bistecca *(la) ho mangiato Italian the steak it-CL have-1s eaten

(3) el filete *(lo) comì Spanish the steak it-CL ate-1s ‘The steak, I ate’

(4) the steak I ate English

Syntax - Morphology - Semantics 3 5 3 Major Trends in Theoretical and Applied Linguistics

Nonetheless—despite the proliferation of these views—during the last decade there has been an overt shift in the literature of Greek. In particular, researchers like Alexopoulou and Koliakou (2002), Tsimpli and Roussou (2006) and lately Gryllia (2009), have correctly pointed out that the aforementioned assumption is actually too strong as the grammaticality of utterances like these below indicate: a non-focal object is—after all—possible to occur cliticless in a preverbal position:

(5) ti lisi vrike mono o Nikos the solution found only the Nick ‘Only Nick found the solution’

(6) ton proθipurγo θa sinodefsi o ipurγos aminas the prime-minister-acc will accompany the minister-nom of-defense ‘The prime minister, the minister of defense will accompany’

As far as the terminology is concerned, the construction that involves a resumptive pronominal clitic is almost invariably referred to as ‘Clitic Left Dislocation’ (CLLD) in the relevant literature, while the cliticless construction is sometimes referred to as Topicalisation (e.g. in Alexopoulou & Kolliakou 2002). Although I will be coming back to this shortly, note at this early point the inconsistent pattern that arises here regarding naming: the former construction bears a label that describes (part of) its structural specification, whereas the latter construction bears a label that merely reflects its interpretive burden. In the light of these observations the following question needs to be tackled: What exactly regulates the presence/absence of a pronominal clitic in non-focal dislocations of the type presented above? Or—in other words—what are the functional discrepancies between the two constructions? The rest of the paper is organized as follows: In section 2 I review and criticize certain major aspects of the relevant literature, while in section 3 I present the current analysis. Section 4 constitutes a short note about the syntax of the two constructions under examination and the Information Structure (IS) interface.

2. An Overview of the Literature

Rather expectedly after what has been discussed in section 1, while there is a good amount of work about CLLD both on syntactic and functional/ interpretive grounds, still very little has been said about ‘Topicalisation’ since the construction was assumed to be a non-option in Greek until quite recently. But before I start reviewing the literature’s stand on the issue, let me—at this point—make an important announcement for the course of the discussion:

3 5 4 Syntax - Morphology - Semantics Axiotis Kechagias

From now on I will be using the term Bare Left Dislocation (BLD) to exclusively refer to and describe the type of non-focal dislocation illustrated in example (5) and (6), thus abandoning the misleading term ‘topicalisation’, since as we will see in what follows it is CLLDed DPs that perform the discourse function topic and not BLDed ones. Coming into CLLD, there has always been a consensus that dislocated-to- the-left-periphery doubled DPs are ‘topics’, topichood being non infrequently defined in terms of ‘old’ or ‘presupposed’ information (see Philippaki- Warburton 1985). In this view, in (7) below the CLLDed DP ‘ti Maria’ is meant to be a topic because it corresponds to the given part of the utterance it shows up in:

(7) A: Pios filise ti Maria ? who kissed the Maria? ‘Who kissed Maria’ B: [ti Maria] ti filise o Nikos (CLLD) the Maria her-CL kissed the Nick-nom ‘Maria, Nick kissed her’

Nonetheless, this direct association of topichood with old information runs into problems, since CLLDed DPs need not always correspond to old or presupposed information. The case below illustrates this:

(8) A: Ti jinete ekso? Ti ine aftos o θorivos? (CLLD) what happens outside? What is this the noise? ‘What is going on outside? What’s this noise?’ B: [kapion perastiko] ton kiniγane kati skilia some passer-by him-CL chase-3p some dogs ‘Some passer-by is being chased by some dogs’

Furthermore, even if this type of analysis were on the right track, it would fail to account—in a minimal at least way—for the functional discrepancies between CLLD and BLD, since even BLDed DPs typically (but not always) constitute old or presupposed information:

(9) A: pios sinoδevi ton proθipurγο stin Italia? (BLD) who accompanies the prime-minister to-the Italy? ‘Who accompanies the prime-minister to Italy?’ B: [ton proθipurγo] sinoδevi o ipurγos aminas the prime-minister.acc accompanies the minister of-defense ‘The prime-minister, the minister of defense is accompanying to Italy’

Syntax - Morphology - Semantics 3 5 5 Major Trends in Theoretical and Applied Linguistics

For reasons like these, Alexopoulou & Kolliakou (2002) have suggested that CLLDed DPs are rather ‘links’ along the lines of Vallduvi (1992), that is, they indicate the entity (the ‘file-card’) where new information should be added and stored in an information-updating system of communication. Thus, links are necessarily part of the ground partition of the utterance upon which new information is added. The other core property of links is that their referent is supposed to be in a relation of non-monotone anaphora with the discourse along the lines of Hendriks & Dekker (1995): a link’s referent is either selected out of a discourse prominent set of entities, or is just related to some other discourse prominent entity (subselectional and relational anaphora in their terms respectively). Crucially, Alexopoulou & Kolliakou’s (2002) main claim is not only that CLLDed DPs are links but that linkhood in Greek is formally realized exclusively as CLLD, their conclusion being that BLDed (‘topicalised’ in their terms) DPs cannot be links. These assumptions, however, are both too strong and too weak. First, they are too strong, since there seem to be cases of felicitous CLLD constructions where neither subselectional nor relational anaphora can be established between the dislocated DP and the discourse. The example in (8) above is such a case. On the other hand, they are too weak since— if linkhood truly constitutes ground information plus non-monotone anaphora—BLDed DPs can equally (a) constitute ground material (so they function as anchor for new information giving rise to focus domains smaller than the whole utterance) and (b) stand in a relation of non-monotone anaphora with some discourse entity or set of entities. The following cases are quite revealing:

(10) A: Afto to provlima itan poli δiskolo... this the problem was very difficult ‘This problem was very difficult…’ B: praγmati, na fandastis [ti lisi] vrike mono o Aris indeed, SUBJ imagine-2s the solution found only the Ares ‘Indeed, the solution, only Ares found’

(11) A: Pios sinoδepse tus filus su sto parti? who accompanied the friends your to-the party? ‘Who accompanied your friends to the party?’ B: δe θimame jia olus; pandos [to Niko] nomizo sinoδefse i Maria not remember about all though the Nick-acc think-1s escorted the Maria-nom ‘I don’t remember about all of them; Nick, however, I think that Mary escorted’

In the examples above, the fronted DP objects ‘the solution’ and ‘Nick’ are ground information, functioning as anchor for new information (the subject in 10; the verb and the subject in 11). Moreover, the referents of the linguistic expressions ‘the solution’ and ‘Nick’ can also be said that fulfill the non-

3 5 6 Syntax - Morphology - Semantics Axiotis Kechagias

monotone anaphora requirement for linkhood, since in (10) the fronted object is clearly related with the discourse salient DP ‘the problem’, whereas in (11) the object is meant as being selected out of a discourse salient set (i.e. the set of ‘your friends’). Thus, it seems that Alexopoulou & Kolliakou’s distinction between [+Link] and [-Link] for CLLDed and BLDed DPs respectively cannot fully account for the functional discrepancy between CLLD and BLD (and there should be one), even if we assume that CLLDed DPs do indeed constitute ‘links’. The last analysis I would like to review here is Gryllia’s (2009) recent contribution which constitutes the first explicit attempt towards a full understanding of the discrepancies between CLLD and BLD. In a nutshell, the kernel of this account is that both BLDed and CLLDed DPs are topics but ‘of a different level’. Gryllia exploits the idea that there are two different kinds of topics, in particular topics of ‘sentential level’ (s-topics) and topics of ‘the discourse level’ (d-topics)1. An s-topic is supposed to be the expression whose referent a mere sentence is about, and is obligatorily fronted in the preverbal domain along the lines of Reinhart (1981), while a d-topic is the referent of some linguistic expression that appears to be the topic of a stretch of discourse that is larger than a simple sentence (cf. van Dijk 1977, Reinhart 1981). Although d-topics need not be marked grammaticality in Greek, they are non-infrequently marked via movement to the beginning of the clause. This is shown below:

(12) a. Το 1899, ο θios Vania ekane premiera stin Mosxa the 1899 the Uncle Vania made premiere in-the Moscow ‘In 1899 Uncle Vania premiered in Moscow’

b. o Stanislavski skinoθetise [D-TOP tin parastasi] the Stanislavksi directed the performance ‘Stanislavski directed the performance’ OR

b’. [D-TOP tin parastasi] skinoθetise ο Stanislavksi the performance directed the Stanislavksi c. Afti γnorise meγali epitixia it (=the performance) met big success ‘It was a great success’

In the example above—according to Gryllia’s analysis—the DP object [the performance] is meant to be a d-topic since the discourse continuation provided

1 Note that such a hypothesis is not unmotivated: several researchers have expressed the idea that there are different kinds of topics that do not necessarily behave alike with respect to each other, in terms of syntax, phonology and morphology. See Neeleman & van de Koot 2008, Vermeulen 2007, Frascarelli & Hinterhölzl 2007, Bianchi & Frascarelli 2010 among others. Cf. also Kechagias 2011.

Syntax - Morphology - Semantics 3 5 7 Major Trends in Theoretical and Applied Linguistics

by the utterance in (c) is also about the referent of that DP. In what (b) differs from (b’) is that in the latter case the d-topic has moved to the preverbal domain. In this light, Gryllia (2009) then puts forward her proposal which is summarized as follows:

“[...] a preverbal object has to be taken up by a clitic in cases that it cannot be interpreted as a discourse-topic. (Gryllia 2009: 72)”

Despite the appeal that such an analysis may have at first sight, it runs into serious problems both on theoretical and empirical grounds. In what follows I will highlight only some of these. To begin with, the aforementioned statement seems to make three very specific predictions:

(13) i. A BLDed DP has to be either a (fronted) d-topic, or a d-topic and an s-topic at the same time; that is, it cannot be a mere s-topic by any means. ii. A fronted DP should be able to occur cliticless (i.e. in a BLDed fashion) when performing the function d-topic. iii. For a DP to be interpreted as an s-level topic only, it has to show up in a CLLDed fashion.

In what follows, I will show that none of the aforementioned predictions is actually borne out by the data. Let me first show why (i) cannot be true. For the sake of the argument consider first the following example:

(14) a. Praγmatopiiθike xtes o polianamenomenos aγonas ton 100 metron was-held yesterday the well-expected race the-gen 100 meters-gen ‘The well-expected 100m race was held yesterday’ b. [tin proti θesi] katelave o aθlitis apo tin Elaδa the first place occupied the athlete from the Greece ‘The first place, the athlete from Greece won” c. Prokite jia enan neo aθliti pu... is for a new athlete that… ‘He’s a new athlete that...’

According to (i) above, the fronted object in (b) must either be a mere fronted d-topic or an s-topic and a d-topic at the same time. However, this assumption is problematic. On the one hand, I cannot see how the DP ‘the first place’ can be a d-topic since by considering sentence (a) one would say that the d-topic is a ‘100m race’. That the DP ‘the first place’ cannot be a d-topic is further supported by the fact that when sentence (c) is added in as a discourse continuation in Gryllia’s terms, BLD in (b) is perceived as an entirely felicitous option, despite the fact that neither sentence (a) nor sentence (c) now is about

3 5 8 Syntax - Morphology - Semantics Axiotis Kechagias

the referent of the fronted linguistic expression ‘the first place’ in (b). (Indeed sentence (c) is about the athlete introduced by (b). In other words, the BLDed DP ‘the first place’ cannot be a d-topic contrary to what Gryllia’s analysis predicts. In the same vein, let me now show why the second of the aforementioned predictions is equally problematic. In order to do so, consider the following discourse stretch:

(15) a. O Arise exi fisiko talendo stis ksenes γ loses the Ares has natural talent in-the foreign languages ‘Ares has a natural talent in foreign languages’ b. iδika ta aglika *(ta) milai aptesta especially the English it-CL speak-3s fluently ‘English, especially, he speaks it fluently’ c. Ala endaksi, ta aglika ine sxetika efkoli γlosa… but OK, the English is relatively easy language ‘But OK, English is a relatively easy language…’

Recall that, according to the second of the predictions above, an object is taken up by a clitic only when it cannot be understood as the topic of a stretch larger than a mere sentence (i.e. d-topic). In that sense, the fronted DP ‘English’ in (b) should indeed be a d-topic since the discourse continuation provided by (c) is also about the referent of that DP. However, not only CLLD is not banned, but actually in this particular example CLLD looks like the only option. In other words, BLD is for some reason ruled out as a possibility, despite the fact that the DP ‘English’ would be a d-topic according to the definition of d-topichood Gryllia adheres to. The third prediction is not verified by the data either. According to (iii) above, for a DP to function as a mere s-topic (under the view adopted by Gryllia) it has to be taken up by a clitic; however, it seems that this is not true either. Consider the example below:

(16) a. Pios sinoδepse tin Eleni xtes sto parti? who escorted the Eleni yesterday to-the party? ‘Who escorted Helen yesterday to the party?’ b. [tin aderfi tis] sinoδepse siγura o Nikos... the sister her escorted-3s definetely the Nick-nom ‘Her sister, definitely Nick escorted …’ c. …tin Eleni δe θimame, me sinxoris the Eleni not remember-1s me-CL forgive-2s ‘As for Eleni, I don’t remember, sorry’

Syntax - Morphology - Semantics 3 5 9 Major Trends in Theoretical and Applied Linguistics

In clause (b) the DP object ‘her sister’ shows up dislocated in a BLDed fashion. According to Gryllia’s analysis, one would expect that this is because that DP is meant to be the topic of the particular discourse stretch it appears in. However, considering (a) and (c) this does not seem to be the case—the discourse topic seems to be ‘Eleni’. Thus, the fronted DP in (b) is not the topic of the discourse; in other words, it can only be the topic of the sentence it occurs in, that is (b). But if a fronted topic is taken up by a clitic when it does not function as a d-topic, then yet again Gryllia’s account makes the wrong predictions: a DP can occur dislocated in the left periphery clitic-less, even when it clearly does not constitute a d-topic, that is, when it is a plain s-topic. For these reasons then, I assume that the distinction between CLLD and BLD cannot be captured through this sentence-level and discourse-level topichood. In what follows I will put forward an alternative account.

3. The Current View

3.1. The Analysis

The analysis I am putting forward here is based on the idea that CLLDed and BLDed DPs participate in two distinct Information Structure mappings. In particular, CLLDed DPs are fronted Topics which associate with the IS category Comment, whereas BLDed DPs are fronted Ground that associates with the IS category New. Therefore, as far as the latter construction is concerned, I assume that the terms ‘Topicalisation’ or ‘Discourse-Topic fronting’ should be abandoned for the more descriptively accurate Ground Fronting. The idea is schematically illustrated in (17) below:

(17) CLLD : [TOPIC] ←→ [COMMENT] BLD : [GROUND] ←→ [NEW (or FOCUS)]

But before I show how this binary mapping works and the predictions it makes, let me first clarify what I take the information structure categories Topic and Ground to be. Regarding topic, I am essentially adopting and adapting Reinhart’s (1981) view, according to which an XP functions as topic when its referent/denotatum is meant to be ‘what a sentence is about’ in a given context. However, there are a couple of problematic aspects with this view that will make us eventually adjust it to some extent. The first problem has to do with the fact that under this definition of topichood we automatically lead ourselves to the commitment that all sentences should have up to one topic. However, this is

3 6 0 Syntax - Morphology - Semantics Axiotis Kechagias

not always the case. In (18) below it is absolutely plausible to assume that the sentence is both about ‘him (=Ares)’ and about ‘his car’.

(18) A: δen boro na su δoso to amaksi mu, jiati δe rotas ton Oresti? not can-1s SUBJ to-you give-1s the car my why not ask-2s the Orestes ‘I can’t give you my car. Why don’t you ask Orestis?

B1: [TOP aftos] [TOP to aftokinito tu] δen to δini me tipota... he the car his not it-CL give-3s with nothing

B2: [TOP to aftokinito tu] [TOP aftos] δen to δini me tipota... the car his that-one-nom not it-CL gives with nothing ‘That one wouldn’t give his car, no matter what…’

In the same spirit, if a topic was simply ‘what a sentence is about’ in a given context one could wonder why, for instance, a focused element cannot be doubled by a clitic in contexts where the referent of the focused item can easily be interpreted as the entity a sentence is about. However, it is a well known property of doubling in Greek that doubled DPs cannot bear focal stress (see Tsimpli 1995, and references therein):

(19) A: leo na foreso afto to mavro pukamiso1 apopse; ti les? say-1s SUBJ wear-1s. this the black shirt tonight; what say-2s ‘I reckon I am gonna wear this black shirt tonight; what do you think?’

B: ne, AFTO1 na (*to) valis; su pai poli. Yes this SUBJ it-CL put-2s to-you suits-3s well ‘Yes, you should were that one; it suits you very well’

For these reasons I take topichood to be the function under which the referent of a linguistic expression is understood as being in a particular semantic-o-pragmatic relation with a predicate, namely one in which the predicate is understood as a comment about that referent in a given context. To put it differently, the predicate-comment is seen as a property ascribed on some entity, the topic. Crucially, for us topichood is the particular relation of aboutness between two categories, say a DP and a verb, rather than an absolute ‘aboutness’ label on some category: A syntactic block α is a [Topic] as long a block β is interpreted as a [Comment]; and similarly, a syntactic block γ is [Ground] as long as block δ is interpreted as [New]. (See also Neeleman & van de Koot 2008, Slioussar 2007).

(20) Topic as a Holistic Function [XP]—[COMMENT] → XP=Topic

Syntax - Morphology - Semantics 3 6 1 Major Trends in Theoretical and Applied Linguistics

Regarding now the function Ground, I am essentially retaining Vallduvi’s (1992) Information-Updating definition: Ground is that part of the utterance upon which new information is added. Ground acts as an anchor for focus. I also assume that Ground material need not be old or given information in absolute terms. Rather ground material are linguistic expressions that for contextual, situational or psychological reasons qualify as more activated in a speaker’s mind as compared to the expressions that make up the rest of the utterance (see Chafe 1976, Slioussar 2007). Ground is—once again—not assumed to be a strict categorical label, but rather it corresponds to a holistic function: What makes YP a ground is the fact that the rest of the utterance is interpreted as new information, introducing new (i.e. non-evoked) referents and thus updating the discourse.

(21) Ground-hood as a holistic function [YP]—[NEW] → YP=Ground

In what follows I show how the aforementioned dichotomy can account in a straightforward manner for most (if not all) of the data examined so far.

3.2. The Data

To begin with, consider first the following example in (22) below which is originally due to Alexopoulou & Kolliakou (2002:196):

(22) a. To kratiko θeatro ksekinise ti ximerini sezon me tin Erofili tu Xortatsi the state theatre begun the winter season with the Erofili of Xortatsi ‘The state theatre began its winter season with Erofili by Hortatsis’

b. [TOP tin parastasi] [COM tin skinoθetise o Karolos Kun] the performance it-CL directed the Karolos Kun

b’. [GND tin parastasi] [FOC/NEW skinoθetise o Karolos Kun] the performance directed the Karolos Kun

In the minimal context provided by sentence (a), both the IS mappings presented in (20) and (21) are available: By uttering (b) the speaker actually sees what the predicate denotes as a comment about the dislocated DP; a property is ascribed on that DP; in other words the predicate is a comment and as such a clitic shows up, since then the DP the performance has to be a topic. On the other hand, when the speaker goes for (b’), what follows the dislocated DP is not seen as a comment about that DP; rather what follows simply introduces a new referent into the discourse, namely the director of

3 6 2 Syntax - Morphology - Semantics Axiotis Kechagias

‘the performance’, updating thus the communication process2. The fact that both strategies qualify as legitimate options in this context should not come as a surprise. Topics are necessarily part of the ground partition of an utterance (i.e. doubled DPs cannot participate for instance in utterances answering all-new information questions), while the comment is the part of the utterance that typically contains new information or the focus. The fact that topics are necessarily part of the ground partition of an utterance can be rather straightforwardly accounted for by assuming that we typically comment things on after they have been first inserted into the discourse, or things that their referents are meant as salient/accessible enough (cf. Chafe 1976) either by them being physically present or situationally evoked by virtue of being, for example, related to some other discourse salient entity. Crucially, however, in a given context, it is not that all ground DPs function as topics, while, on the other hand, topics are necessarily ground material. The idea that what follows a BLDed DP is not a comment, and by that, that the dislocated DP itself is not a topic, but rather fronted ground material, is verified by cases like the one below:

(23) a. Praγmatopiiθike xtes o polianamenomenos aγonas ton 100 metron was-held yesterday the well-expected race the-gen 100 meters-gen ‘The well-expected 100m race was held yesterday’

2 As a anonymous reviewer points out “cliticless constructions of this type belong to a particular ‘formal register’ (e.g. news, announcements), and that one could claim that this type of register favours the precedence of cliticless non-focal material as journalists wish to keep the audience’s level of interest high. As such, such data cannot be used as prototypical “neutral” examples that give support to the claims of the author”. Although the observation is absolutely valid for minimal pairs such as the one illustrated above in (22) for instance, there are cases where the cliticless construction is contained in utterances without any formal flavour. For the sake of the argument, consider for instance the following two cases that arguably belong to an every-day ‘common’ register: (i) i proetimasies jia to parti pane kala. O Kostas kani ta fajita, ke i Maria θa feri ta pota; the preparations for the party go well The Kostas makes the foods and the Maria will bring the drink ‘The preparations for the party go well; Kostas makes the food and Maria will bring the drinks;’ [ti musiki] exi analavi o aδerfos mu. the music has undertaken the brother my ‘My brother has been in change of the music’ (ii) A: to provlima itan panδiskolo! the problem was very-difficult ‘The problem was very difficult!’ B: praγmati; [ti lisi] vrike mono o Nikos! indeed the solution found only the Nikos ‘Indeed, only Nick found the solution!’ (examples from Kechagias 2012)

Syntax - Morphology - Semantics 3 6 3 Major Trends in Theoretical and Applied Linguistics

b. [tin proti θesi] (#tin) katelave o aθlitis apo tin Elaδa the first place (it-CL) occupied the athlete from the Greece ‘The first place the athlete from Greece won” c. prokite jia enan neo aθliti pu... ‘He is a new athlete that...’

In the minimal context provided by sentences (a) and (c), CLLD in (b) looks as a rather unnatural choice. Indeed, the speaker by uttering (b) does not in fact intend to make a comment about the DP ‘the first place’; rather, the DP [the first place] constitutes ground information (its referent/denotatum is situationally evoked since by uttering ‘the 100m race’ a whole network of entities is activated among which ‘the first place’) that rolls back all the way up to a sentence initial position so that new information is inserted into the discourse. Rather expectedly and naturally what appears as a discourse continuation in (c) is about the referent of the phrase ‘the athlete from Greece’—which has just been presented as new information in sentence (b)—and not about the dislocated DP ‘the first place’. Another advantage of the line of argumentation I have been following is that it can actually account for the constrained distribution of BLD. Consider for instance the following two examples:

(24) iδika ta aglika *(ta) milai aptesta especially the English it-CL speak-3s fluently ‘English, especially, he speaks it fluently’

(25) tin aδelfi mu *(ti) latrevo the sister mine her-CL adore-1s ‘My sister I adore’

What differentiates the cases above from what we have seen so far is that— irrespective of what precedes or/and what follows these utterances—they are actually perceived as ungrammatical3: the dislocated DP cannot be left cliticless, or, in other words, BLD is not an option. Although I would like to leave this issue open for a more detailed discussion in the future, I believe that this behaviour is actually predicted by the alternative account I have been sketching out: if we are correct in that BLD involves rolling up of ground material so that new information is inserted into the discourse, then lack of focus referents post- verbally blocks BLD. In other words, when what follows the dislocated phrase fails to introduce new discourse referents (and presumably neither a verbal

3 And it is due to cases like these that originally BLD was dubbed ungrammatical in Greek.

3 6 4 Syntax - Morphology - Semantics Axiotis Kechagias

predicate nor a manner adverb can introduce referents), then it is necessarily interpreted as an attribute/property—that is a comment—about the dislocated entity—the topic. It is not a coincidence that Ground Fronting becomes very productive with orders that involve post-verbal subjects, since S-V inversion is a mechanism that many languages exploit when the subject is to be inserted into the discourse as new information (see Lambrecht 2000, Belletti 2001, and for Greek Kechagias 2011 and references there in).

4. A Brief Note on the Syntax-C/I Interface

As it was made clear earlier, I take topic and ground to be relational (i.e. non absolute) pragmatic functions that are furthermore not bound onto any specific region in the clause. (26) illustrates this for topic, whereas (27) illustrates this for ground:

(26) A: Pes mu kati endiposiako pu exis kani… tell-2s-imp me-gen something impressive that have-2s done ‘Tell me something impressive you have done...’ B. Kapote kerδisa enan Maraθonio 40 xiliometron once won-1s a Marathon 40 km-gen ‘I once won a 40km Marathon race’

A: Siγa to prama; ki eγo to kano [TOP afto] not the thing and I it-CL do this

A’: Siγa to prama; [TOP afto] to kano ki eγo not the thing this it-CL do and I ‘Big deal; this, even I can do’

(27) a. Praγmatopiiθike xtes o polianamenomenos aγonas ton 100 metron was-held yesterday the well-expected race the-gen 100 meters-gen ‘The well-expected 100m race was held yesterday’

b. o aθlitis apo tin Elaδa katelave [GND tin proti θesi] the athlete from the Greece won the first place

b’. [GND tin proti θesi] katelave o aθlitis apo tin Elaδa the first place won the athlete from the Greece ‘The first place the athlete from Greece won’

It was also pointed out that the mapping rules in (20) and (21) are Information Structure rules, and not syntactic representations. The conceptual basis of the mapping rule in (21) lies in two well-known processing advantages associated with early mention of ground information: First, the earlier ground information occurs in the sentence, the easier it is to link it to the previous discourse.

Syntax - Morphology - Semantics 3 6 5 Major Trends in Theoretical and Applied Linguistics

Second, new information is easier to integrate if the ground information that facilitates contextualization has been processed. As far as the mapping rule in (20) is concerned, as it was mentioned earlier, it is natural to typically comment something after it has been inserted into the discourse. When topic dislocations (26A’) and ground dislocations (27b’) occur in the language (and possibly across languages), they merely facilitate a transparent mapping between syntactic structures and continuous blocks of Information Structure (a la Neeleman & van de Koot 2008). Nonetheless, because the syntax and the information structure are independent modules of the grammar, this transparent mapping not infrequently can be suspended (i.e. topic and ground objects occur in-situ, since the syntax and the information structure need not be always isomorphic). Alternatively, one can assume that such orders are generated freely by the narrow syntax through generalized EPP or Edge Features (a la Chomsky 2005, 2006) and the dislocated constituents receive an interpretation at the Interface.

3 6 6 Syntax - Morphology - Semantics Axiotis Kechagias

References

Alexopoulou, T. and D. Kolliakou. 2002. On Linkhood, Topicalisation and Clitic Left Dislocation. Journal of Linguistics 38.2, 193-245.

Belletti, A. 2001. Inversion as Focalisation. In A.C. Hulk and J-I. Pollock (eds.), Subject inversion in Romance and the theory of Universal Grammar. New York: OUP, 60-90.

Bianchi, V. and M. Frascarelli. 2010. Is topic a root phenomenon?. Iberia: An International Journal of Theoretical Linguistics 2.1, 18-65.

Chafe, W. 1976. Givenness, contrastiveness, definiteness, subjects, topics, and points of view. In C. Li (ed.), Subject and Topic. New York, 25-56.

Chomsky, N. 2005. On Phases. Ms, MIT.

Chomsky, N. 2006. Approaching UG from below. Ms, MIT.

Frascarelli, M. and Hinterhölzl, R. 2007. Types of topics in German and Italian. In S. Winkler and K. Schwabe (eds.), On Information Structure, Meaning and Form. Amsterdam: John Benjamins, 87-116.

Gryllia, S. 2009. On the nature of preverbal focus in Greek. A theoretical and experimental approach. PhD. Diss. University of Leiden.

Hendriks, H. and P. Dekker. 1995. Links without locations: information packaging and non-monotone anaphora. In P. Dekker and M. Stokhof (eds.), Proceedings of the 10th Amsterdam Colloquium. Institute of logic, language and computation, Department of Philosophy, Universiteit van Amsterdam, 339-358.

Syntax - Morphology - Semantics 3 6 7 Major Trends in Theoretical and Applied Linguistics

Kechagias, A. 2011. Regulating word order in Modern Greek: Verb-Initial and non Verb-Initial orders and the Conceptual-Intentional interface. Ph.D Dissertation, UCL. UK.

Kechagias, A. 2012. An experimental analysis of object non-focal left dislocation in Greek. (Ms.) UCL.

Lambrecht, K. 2000. When subjects behave like objects: An Analysis of the merging of S and O in sentence focus constructions across languages. Studies in Language 24.3, 611-682.

Neeleman, A. and H. van de Koot. 2008. The nature of discourse templates. Journal of comparative Germanic linguistics 11, 137-189.

Philippaki-Warburton, I. 1985. Word Order in Modern Greek. Transactions of the Philological Society 83, 113-143.

Reinhart, T. 1981. Pragmatics and linguistics: an analysis of sentence topics. Philosophica 27, 53-94.

Roussou, A. and I.-M. Tsimpli. 2006. On Greek VSO again! Journal of Linguistics 42, 317-354.

Slioussar, N. 2007. Grammar and Information Structure: A study with reference to Russian. PhD dissertation. Utrecht institute of Linguistics OTS.

Tsimpli, I.-M. 1995. Focusing in Modern Greek. In K. Kiss (ed.), Discourse Configurational Languages. Oxford: Oxford University Press, 176-206.

van Dijk, T. 1977. Text and Context. London: Longman.

Vallduvi, E. 1992. The Informational Component. New York: Garland.

Vermeulen, R. 2007. Japanese wa-phrases that aren’t topics. In R. Breheny and N. Velegrakis (eds.), UCL Working Papers in Linguistics 19, 183-201.

3 6 8 Syntax - Morphology - Semantics Distributivity and Genericity in Greek: The Case of kathe with the Definite Article

Dimitra Lazaridou-Chatzigoga

University of Crete & Queen Mary, University of London [email protected], [email protected]

Abstract

This paper discusses the co-occurrence of the universal distributive quantifier kathe with the definite article (o, i, to) in Greek. Following Beghelli & Stowell (1997) and Tunstall (1998) I argue that while both bare kathe-phrases and kathe- phrases with the definite article (henceforth o kathe-phrases) exhibit strong distributivity, o kathe-phrases are linked to obligatory and total distribution, while bare kathe-phrases are linked to optional and partial distribution. As far as genericity is concerned, kathe can be generic, while o kathe cannot, because it is presuppositional. The contribution of the definite article in this construction is associated with weak familiarity and the count-as-unique condition.

1. Introduction1

This paper concerns the interaction between determiners and quantifiers in the Greek N(oun) P(hrase), as evidenced in the construction [definite article (o,i,to) + kathe ‘every/each’ + Noun], focusing on two semantic properties, distributivity and genericity. Here is an example:

1 I would like to thank especially Louise McNally, as well as Elena Anagnostopoulou, Josep M. Brucart, M. Teresa Espinal, Winfried Lechner, Josep Quer, Melita Stavrou, Evangelia Vlachou, the people at the Athens reading group in Linguistics and the two anonymous reviewers, who all helped me with insightful comments and suggestions. Needless to say that any remaining errors are mine. This paper draws on my PhD thesis (Lazaridou- Chatzigoga 2009b), completed amb el suport del Comissionat per a Universitats i Recerca del Departament d’Innovació, Universitats i Empresa de la Generalitat de Catalunya.

Syntax - Morphology - Semantics 3 6 9 Major Trends in Theoretical and Applied Linguistics

(1) O kathe fotoghrafos tha feri pende fotoghrafies apo tin siloghi tu. the kathe photographer fut bring.3sg five photographs from the collection his ‘Each photographer will bring five photographs from his collection.’

It is not uncommon to encounter the definite article co-occurring with other determiners in Greek, as is evident, for instance, in demonstratives, possessives and with the universal quantifier olos ‘all’ (see among others Marmaridou- Protopapa 1984, Lyons 1999, Alexiadou et. al 2007 and Giannoulopoulou 2007). A striking difference of the construction in question is the fact that this co-occurrence seems redundant2, since the universal distributive quantifier kathe is an item that can also form an NP combining with a Noun with no other determiner present. This does not hold either for demonstratives or for possessives.

2. Morphological, Syntactic and Semantic Characteristics of (o) kathe

2.1. Morphological and Syntactic Characteristics

In this paper I focus on the semantic characteristics of (o) kathe, but let me just briefly point out the main morphological and syntactic characteristics of (o) kathe: a) kathe is not inflected, b)(o) kathe does not occur without a Noun, and c) (o) kathe is always preceding the Noun:

(2) (To) kathe pedhi pire ena dhoro. the kathe child got.3sg a present ‘Every3 child got a present.’ (3) *(To) kathe pire ena dhoro. the kathe got.3sg a present (4) *Pedhi (to) kathe pire ena dhoro. child the kathe got.3sg a present

2 For another co-occurrence of the definite article with a determiner, namely, the co- occurrence with opjosdhipote ‘FC any’ see Lazaridou-Chatzigoga (2007, 2009a,b). For the construction in question through a cross-linguistic point of view, see Etxeberria & Giannakidou (2008) for Basque and Matthewson (2001) for Lilloet Salish. This aspect of the phenomenon remains for future research. 3 For expository reasons, I translate here (o) kathe with every, but as will be obvious in the discussion, kathe is to be translated as every, while o kathe as each.

3 7 0 Syntax - Morphology - Semantics Dimitra Lazaridou-Chatzigoga

From (4) we can deduce that (o) kathe is not a floating quantifier, that is, its syntactic position is fixed. In contrast, other universal quantifiers like olos, can float, as they can appear in various positions within a sentence, as seen below:

(5) Ola ta pedhia mpikan stin taksi/ Ta pedhia mpikan ola stin taksi. all the children entered in.the class/ the children entered all in.the class ‘All the children entered the class/ The children all entered the class.’

2.2. Semantic Characteristics

Before entering into the discussion of distributivity and genericity with respect to (o) kathe, let me highlight two distinctive semantic properties of this quantifier. The first one has to do with the scope properties of (o) kathe. (O) kathe has the tendency to receive wide scope with respect to other quantifiers, a fact in accordance with observations regarding similar quantifiers in other languages4. In sentences like the following, which are scopally ambiguous, the first reading is the preferred one, according to which kathe musikos ‘every musician’ scopes over the indefinite mia klimaka ‘a scale’:

(6) (O) kathe musikos kseri mia klimaka. the kathe musician knows a scale ‘Every musician knows a scale.’ 6’ ∀x (musician(x) → ∃y (scale(y) ∧ know (x,y))) 6’’ ∃y (scale(y) ∧ ∀x (musician(x) → know (x,y)))

The second distinctive property is that kathe can quantify over atomic individuals, as we saw above in (2), or plural individuals in the sense of Link (1983), as we see below. Kathe here quantifies over the plural individual formed by tris katikus ‘three citizens’, which serves as the minimum unit kathe quantifies over. This is though not possible for o kathe, as we see below:

(7) Iposhethikan na fitepsun ena dentro ya kathe tris katikus/*tus kathe tris katikus. promised.3pl subj plant a tree for kathe three citizens/the kathe three citizens ‘They promised to plant a tree for every three citizens.’ The quantification here is only possible over groups of three, which means that we have no access to the sub-parts of the plural individual, a fact that is

4 See Ioup (1975) for a Quantifier Hierarchy regarding scope: each > every > all > most >

many > several > somepl > a few.

Syntax - Morphology - Semantics 3 7 1 Major Trends in Theoretical and Applied Linguistics

predicted under my analysis, which presupposes access to each and every entity in the kathe-set.

3. (O) kathe: Distributivity and Genericity

In the descriptive literature only the emphatic nature of the construction is pointed out (Holton, Mackridge & Philippaki-Warburton 1999: 311), and no clear explanation is given to the apparent optionality of the co-occurrence of the definite article with kathe. Building though on previous analyses (Giannakidou 1999, 2004, Tsili 2001, Exteberria & Giannakidou 2008) I will refine the semantic characteristics of o kathe drawing on distributivity and genericity. Before entering into the discussion, let me illustrate the phenomenon with more examples5:

(8) Videoskopuse to kathe vima tis. filmed.3sg-IMPFVE the kathe step her ‘She filmed each and every step of hers.’ (9) I Ino fotoghrafise to kathe zoo tu zoologhiku kipu. the Ino photographed.3sg the kathe animal the-GEN zoologic garden ‘Ino photographed each animal of the zoo.’ (10) Kathe hrono ipologhizete oti 5 ekatomiria theates vlepun, kata meson oro, tis 1200 parastasis tu ‘holiday on ice’, pu dinonte se 70 polis apo 15 hores [...] I kathe parastasi periodevi ya tria hronia. ‘Every year it is calculated that 5 millions spectators watch, on average, the 1200 performances of ‘Holiday on Ice’ that are given in 70 cities in 15 countries [...] Each performance is on tour for two years.’

The discussion illustrates the two main points of this paper: a) the fact that bare kathe-phrases and o kathe-phrases are not interchangeable in all contexts and b) the issue of the contribution of the definite article in this construction. In order to address the first issue I will be concerned with the properties of distributivity and genericity in subsections 3.1 and 3.2, respectively. In order to answer the second issue, I will discuss the notion of definiteness in section 4.

5 I have left outside the scope of my paper occurrences of o kathe N, which I think deserve a different treatment, being similar to minimizers and pointing to the end of a scale: Ehume varethi ton kathe asheto stis tileorasis ‘lit. we are bored of the kathe ignorant on the TV’. On these occurrences see Margariti (2007).

3 7 2 Syntax - Morphology - Semantics Dimitra Lazaridou-Chatzigoga

3.1. (O) kathe and Distributivity

In the literature we encounter two main refinements of distributivity, one proposed by Beghelli & Stowell (1997; B&S henceforth) and the other proposed by Tunstall (1998). B&S (1997) discuss distributivity with respect to all, every and each and propose two differentiations. The first concerns the difference between strong distributivity, attested in NPs with every and each, and pseudo-distributivity/ weak distributivity, attested in NPs with all. This type of distributivity depends on whether the NP can have a collective interpretation or not. (O) kathe is not compatible with collective predicates (Dowty 1987) like mazevome ‘gather’ in contrast to olos ‘all’, so (o) kathe is related to strong distributivity, while olos to pseudodistributivity.

(11) Ola ta pedhia mazeftikan stin avli. all the children gathered in.the yard ‘All the children gathered in the yard.’ (12) *(To) kathe pedhi mazeftike stin avli. the kathe child gathered in.the yard

A further subdivision concerns what B&S (1997) call optional distributivity, as with every, and obligatory distributivity, as with each. As far as Greek is concerned, I follow Tsili (2001), who argues that kathe is optionally distributive, while o kathe obligatorily distributive. The application of the tests that B&S (1997) use for every/each to the Greek items highlights the parallel distribution between every and kathe and each and o kathe-phrases:

a. O kathe-phrases cannot appear in universal collective readings:

(13) *Katevalan tin kathe prospathia ya na lithi to thema.6 they.strained the kathe intent to solve the issue

b. O kathe-phrases cannot be modified by shedhon ‘almost’, in contrast to kathe-phrases:

(14) Mia musikos mantepse sosta shedhon {kathe traghudhi/ *to kathe traghudhi}. a musician guessed.3sg correctly almost kathe song/ the kathe song

6 As an anonymous reviewer points out this might be due to a more general phenomenon concerning periphrases like katavalo prospathia ‘lit. strain intent’or simiono proodho ‘lit. mark progress’, which are hostile to the occurrence of the definite article.

Syntax - Morphology - Semantics 3 7 3 Major Trends in Theoretical and Applied Linguistics

c. There are constructions similar to floating and binominal each, expressed via o kathenas ‘lit. the eachone’, which crucially involve the definite article:

(15) Ta pedhia efaghan dhio mila to kathena. the children ate.3pl two apples the each.one ‘The kids ate two apples each.’

Despite the construction above, a noted difference concerns the fact that (o) kathe cannot float, in contrast toeach . For this reason, we cannot adopt accounts that have been offered for the co-occurrence of other universal quantifiers with the definite article that appeal to the floating nature of quantifiers like all. Such an account has been put forward by Brisson (1998), who argues that all is not quantificational, but rather a kind of exhaustivity marker. A second refinement of distributivity proves to be crucial for(o) kathe. Tunstall (1998) discusses distributivity and claims that every and each require multiple, or distributive, event structures, where the members of their restrictor set are associated with a number of different subevents7. In the case of every she argues that there is a condition that the event must be at least partially distributive and in the case of each there is a condition that the event must be totally distributive. Turning to example (9), repeated here, we see how the above distinction is applied to the event structures associated to the following nominals:

(16) I Ino fotoghrafise [kathe /to kathe] zoo tu zoologhiku kipu. ‘Ino photographed kathe/ the kathe animal of the zoo.’

For (16), the scenario (a) below is ruled out, because we cannot have a collective interpretation with (o) kathe8. The event of taking photos of all the

7 The conditions for every and each (Tunstall 1998: 99 and 100 respectively) are defined as follows: 1. The Event Distributivity Condition A sentence containing a quantified phrase headed by every can only be true of event structures which are at least partially distributive. At least two different subsets of the restrictor set of the quantified phrase must be associated with correspondingly different subevents, in which the predicate applies to that subset of objects. 2. The Differentiation Condition A sentence containing a quantified phrase headed by each can only be true of event structures which are totally distributive. Each individual object in the restrictor set of the quantified phrase must be associated with its own subevent, in which the predicate applies in that object, and which can be differentiated in some way from the other subevents. 8 An anonymous reviewer points out that the scenario (a) is not ruled out for kathe, This would be though a scenario expressed in Greek with the universal quantifierolos,-i,-o ‘all’, which receives a collective interpretation.

3 7 4 Syntax - Morphology - Semantics Dimitra Lazaridou-Chatzigoga

animals in the zoo has to be partially differentiated with kathe, while totally differentiated with o kathe. In the version with kathe, (16) can be claimed to be true in a setting where Ino took four photos in a real small zoo, with a total of five animals, and say, for instance, the giraffe and the hippo were photographed together, while the remaining three animals, the zebra, the monkey and the fox were each photographed separately (scenario c below). This cannot be a possible scenario for o kathe, since each animal must be associated with each individuated event of photo-taking. The partially distributive scenarios (b) and (c) are ok for kathe, but are ruled out for o kathe-phrases, for which only the (d) scenario of complete/total distribution is legitimate:

a. zebra b. zebra monkey monkey e1 fox e fox giraffe giraffe e hippo hippo e2

COLLECTIVE PARTIAL DISTRIBUTION

c. zebra e1 d. zebra e1 monkey e2 monkey e2 fox e3 e fox e3 e giraffe giraffe e4 hippo e4 hippo e5

PARTIAL DISTRIBUTION TOTAL DISTRIBUTION

I argue that the difference between kathe and o kathe is that whereas kathe requires that there be at least two distinct subevents (for every object that is acted upon in one subevent we simply need to find one other object that is acted upon in another subevent), o kathe requires all the subevents to be distinct (for every affected object we must check that all other objects are in another subevent). This fact, found implicit in Giannakidou’s account (2004), is made clear if we add ala ohi ksehorista ‘but not separately’. This modification marks that the subevents are not totally distributed to each and every animal in the zoo, and is thus deviant when we have o kathe:

(17) I Ino fotoghrafise to kathe zoo tu zoologhiku kipu, the Ino photographed the kathe animal the-GEN zoologic garden, # ala ohi ksehorista. but not separately ‘Ino photographed each animal of the zoo, # but not separately.’

Syntax - Morphology - Semantics 3 7 5 Major Trends in Theoretical and Applied Linguistics

Support for this claim concerning every and each comes from an observation made by Vendler (1967:78), who argued that “[There is] a marked difference in emphasis: every stresses completeness or, rather, exhaustiveness...; each on the other hand, directs one’s attention to the individuals as they appear, in some succession or other, one by one. Such an individual attention is not required in vain: you have to do something with each of them, one after the other”. What I claim here (to be discussed in detail in section 4) is that what you have to do with each of the individuals in question is apply the count-as-unique condition to each and every one them, one after the other. Furthermore, that this condition is associated to the definite article in Greek, while in other languages, such as English, this process may be lexicalized.

3.2. (O) kathe and Genericity

If we now turn to genericity, we observe that the distribution of every and each (Gil 1992; B&S 1997) is parallel to the distribution of kathe and o kathe. Giannakidou (1999, 2004) has argued that kathe can be generic, while o kathe cannot9, and the following examples (adaptions from Gil 1992) corroborate this claim, with the importance of discourse being underlined:

(18) Afu afierose tis teleftees tris dhekaeties sti meleti tis leksikis simasiologhias, after devoted.3sg the last three decades to.the study the-GEN lexical semantics i Ino ekane mia spudhea anakalipsi. the Ino made a startling discovery ‘After devoting the last three decades to the study of lexical semantics, Ino made a startling discovery.’ a. Kathe ghlosa ehi pano apo ikosi leksis ya hromata. kathe language has over twenty words for colours ‘Each language has over twenty words for colours.’ b. Oles i ghloses ehun pano apo ikosi leksis ya hromata. all the languages have.3pl over twenty words for colours ‘All languages have over twenty words for colours.’ c. ?I kathe ghlosa ehi pano apo ikosi leksis ya hromata. the kathe language has over twenty words for colours

9 This generalization is initially surprising, because genericity in Greek is typically expressed with definite noun phrases, mostly plurals, like i falenes ine thilastika ‘lit. the whales are mammals’, but also singulars as in i fanela ine thilastiko ‘lit. the whale is a mammal’. Due to space, we cannot elaborate more on this point, but we think that the fact that (o) kathe is necessarily singular might be one crucial parameter.

3 7 6 Syntax - Morphology - Semantics Dimitra Lazaridou-Chatzigoga

d. I ghloses ehun pano apo ikosi leksis ya hromata. the languages have over twenty words for colours ‘Languages have over twenty words for colours.’

(19) I Ino molis anakalipse dheka aghnostes mehri tora ghloses sta ipsipeda the Ino just discovered ten unknown hitherto languages in.the highlands tis Papua sti Nea Guinea. the.gen Papua in.the New Guinea ‘Ino has just discovered ten hitherto unknown languages in the Papua highlands in New Guinea.’ a. ?Kathe ghlosa ehi pano apo ikosi leksis ya hromata. kathe language has over twenty words for colours b. ?Oles i ghloses ehun pano apo ikosi leksis ya hromata. all the languages have.3pl over twenty words for colours c. I kathe ghlosa ehi pano apo ikosi leksis ya hromata. the kathe language has over twenty words for colours ‘Each language has over twenty words for colours.’ d. ?I ghloses ehun pano apo ikosi leksis ya hromata. the languages have over twenty words for colours

In (18) we have a context, according to which Ino studies lexical semantics and makes a discovery about the color terms languages use to have. Since no explicit mention of languages is made in the context, Ino’s discovery can be presented with a generic NP headed by kathe, oles ‘all’ or a definite plural (the usual way to express genericity in NPs in Greek), but crucially not with o kathe. Moving now to a context, where Ino has discovered ten previously unknown languages and then studies the color terms they have, we see that the situation in (19) is reversed: kathe, oles ‘all’ and the definite plural are deviant, while o kathe is completely grammatical, referring to each and every one of the ten languages discovered by Ino. I argue that additional support for the non-genericity of o kathe comes from the following type of contexts: a) individual-level predicates, b) there-insertion contexts and c) intensional contexts. First, (o) kathe-phrases can appear with individual-level predicates (Carlson 1977), that is, predicates that express permanent properties of individuals, as below:

(20) Kathe ghata ehi tesera podhia/ I kathe ghata ehi tesera podhia. ‘Every cat has four legs / Each cat has four legs.’

Both sentences are grammatical, but the claim with the definite article can only be true of some particular set of cats previously introduced in the discourse

Syntax - Morphology - Semantics 3 7 7 Major Trends in Theoretical and Applied Linguistics

and cannot be generic. When reference is made to kinds, of which there are no instances in the actual world, we observe that o kathe becomes unacceptable. This is due to the fact that kathe can make claims both about the actual and possible worlds, while o kathe can only refer to the actual world, and more restrictively to the universe of discourse, as argued by Etxeberria & Giannakidou (2008), from whom I quote the following example:

(21) Kathe monokeros ehi ena kerato/ #O kathe monokeros ehi ena kerato. ‘Every unicorn has one horn.’/‘ Each unicorn has one horn.’

Following Etxeberria & Giannakidou, I claim that when we have o kathe the claim can only be made about a specific set of unicorns, e.g. only if we imagine an illustration in a book that is present physically at the time of conversation, since o kathe cannot escape reference to the universe of discourse. To account for the intuition, I follow Kratzer’s (1981) theory on modality and claim that o kathe-phrases seem to be associated with a realistic modal base. Second, there-insertion contexts, which in general disallow strong quantifiers (Milsark 1974) like every or kathe, may favor a kind interpretation under certain conditions (McNally & Van Geenhoven 1998). In this type of contexts, kathe is licensed, while o kathe is ungrammatical:

(22) Sto dhiadhiktio iparhi kathe/*i kathe idhous pliroforia ya tin evrizonikotita. in.the internet exists kathe/ the kathe kind.GEN information for the.ACC broadband ‘In internet, there is every kind of information on broadband.’

Third, kathe-phrases with intensional verbs like psahno ‘look for’ give rise both to opaque and transparent readings, while o kathe-phrases only to transparent ones:

(23) Epsaksa na vro kathe lathos sto hiroghrafo (ala den ipirhe kanena)/ to kathe lathos sto hirografo (#ala den ipirhe kanena). ‘I looked for kathe/ the kathe error in the manuscript (but there wasn’t any)’

On the basis of the above observations, I argue that o kathe is presuppositional (Heim & Kratzer 1998), that is, it presupposes a non-empty domain, while kathe makes no claim with respect to its domain, it may be empty or not. Giannakidou (2004) also underlines the presuppositionality of o kathe casting it into an account of veridicality on the basis of its behaviour with respect to negative polarity items, which can appear in the argument of the NP of kathe, but not of o kathe, as the following example from Giannakidou (1999: 396) shows:

3 7 8 Syntax - Morphology - Semantics Dimitra Lazaridou-Chatzigoga

(24) a. Kathe fititis pu ghnorizi tipota shetika me tin ipothesi, as milisi tora. every student that know.3sg anything about with the case subj talk.3sg now ‘Every student who knows anything about the case should speak now.’ b. *O kathe fititis pu ghnorizi tipota shetika me tin ipothesi, as milisi tora the every student that know.3sg anything about with the case subj. talk.3sg now

4. The Proposal: the Semantics of the Definite Article with kathe

In the light of the above distribution of o kathe-phrases with respect to distributivity and genericity, it should be clear by now that the definite article does have a contribution when appearing with kathe. Kathe- and o kathe-phrases are not interchangeable in all contexts, given that a) o kathe requires obligatory distribution and b) o kathe cannot be generic. In this section I will focus on my proposal of the nature of this contribution. A first piece of evidence when addressing this issue is the fact thatthe construction [definite article + kathe + Noun] does not seem to build a definite NP (Giannakidou 2004) on the basis of the following strong evidence: The definite article appearing here cannot spread, so it cannot give rise to determiner spreading or polydefiniteness (see Alexiadou and Wilder 1998, Kolliakou 2004 and Campos & Stavrou 2005 among others):

(25) To kathe pedhi tha pari mia karamela. the kathe child fut get a candy ‘Each child will get a candy.’ (26) *To kathe to pedhi tha pari mia karamela. the kathe the child fut get a candy (27) *To kathe to mikro to pedhi tha pari mia karamela. the kathe the small the child fut get a candy

The definite article co-occurring with kathe does nevertheless have a semantic repercussion in the interpretation of the nominal. On the basis of the construction in question and the co-occurrence of the definite article with the Free Choice Item opjosdhipote ‘FC any’ in Greek, in Lazaridou-Chatzigoga (2009b) I proposed a reconsideration of the notion of definiteness. My proposal was based on (weak) familiarity, as in Roberts (2003), and on a redefinition of the notion of uniqueness as the count-as-unique condition, inspired by the ontological insights of the philosophical work of Badiou (1988). Here is the definition I advanced:

Syntax - Morphology - Semantics 3 7 9 Major Trends in Theoretical and Applied Linguistics

(28) Definition of definite

Given a context C, use of a definite iNP presupposes that it has as

antecedent discourse referent xi , which is: a) weakly familiar in C (Roberts 2003) and b) counted-as-unique among discourse referents in C in being contextually

entailed to satisfy the descriptive content of NPi

I argue that the contribution of the definite article ino kathe-phrases amounts to the two conditions of definiteness as defined above. (Weak) familiarity is satisfied in o kathe-phrases, as can be seen in examples (10) and (19) above10. The Count-as-unique (count-U) condition is formalized as follows:

(29) The count-U condition [∀x: P(x)] [count-U (x)] to be read: for every x that has the property P, x is counted-as-unique

Under the count-U condition I envisage an operation that takes place whenever the definite article is used in such a construction. This part of the meaning of the definite article is like a presupposition, and could be seen as part of the procedural meaning in the context of recent advances in Relevance Theory (Wilson and Sperber 2012). In o kathe the procedure is as follows: The speaker makes a plea for the hearer to make sure that she fully distributes the counting to each and every one of the members of the kathe-set. The property in question is already interpreted in a distributive way, but the definite article adds to this that the distribution may not be partial, but it needs to be total. Each member of the kathe-set is counted-as-unique, in the sense that it is given its own separate event structure, which may not overlap with the one or the other member of the set. Following the received view on quantifiers in natural language (Barwise & Cooper 1981), kathe denotes a generalized quantifier: (30) [[kathe]]= λP λQ [∀x: P(x)] [Q(x)]

If we apply the above formalization to (9), repeated here for convenience, we end up with the following semantic representation for the o kathe-phrase:

(31) I Ino fotoghrafise to kathe zoo tu zoologhiku kipu. ‘Ino photographed each animal of the zoo.’

10 A more detailed account of how o kathe satisfies (weak) familiarity is beyond the scope of this paper, but it is part of ongoing research with corpora. The focus here is on the count-U condition, given that this is the part of the meaning that formalizes the most significant differences between kathe and o kathe.

3 8 0 Syntax - Morphology - Semantics Dimitra Lazaridou-Chatzigoga

(32) [[kathe zoo]]= λQ [[∀x: animal (x)] [Q(x)] and the definite article adds the count-U condition:[ ∀x: animal (x)] [count-U (x)]

The proposal advanced here emphasizes the counting process of each and every entity involved, representing the total distribution and the non-generic nature of the construction. The Count-U condition predicts that we need to have access to the instances, to the individuals one by one, so no reference to the kind is possible when the definite article is present. And this is indeed what we found in the section discussing the non-genericity of o kathe. I associate the contribution of the definite article in o kathe with the regular meaning of the definite article, contrasting in this with an analysis that treats the definite article as an overt domain restrictor and claims that o kathe is a complex quantifier (see Giannakidou 2004), essentially giving to the Greek definite article an added meaning.

5. Conclusions

In this paper I have argued that the construction [definite article +kathe + Noun] is related to strong and obligatory distribution in the sense of Beghelli & Stowell (1997) and that it requires total distribution in the sense of Tunstall (1998). Furthermore, o kathe cannot refer to a kind, thus it cannot be generic, and it is presuppositional (Heim & Kratzer 1998). The contribution of the definite article in this construction amounts to the two conditions of definiteness, weak familiarity and the Count-U condition, as defined in Lazaridou-Chatzigoga (2009b).

Syntax - Morphology - Semantics 3 8 1 Major Trends in Theoretical and Applied Linguistics

References

Alexiadou, A., L. Haegeman and M. Stavrou. 2007. Noun phrase in the generative perspective. Berlin: Mouton de Gruyter.

Alexiadou, A. and C. Wilder 1998. Adjectival modification and multiple determiners. In Alexiadou, A. and C. Wilder (eds.), Possessors, predicates and movement in the DP, Amsterdam. 303-332.

Badiou, A. 1988. L’Etre et l’événement. English translation: Being and event. Trans. O. Feltham. London, Continuum, 2005.

Barwise, J. and R. Cooper. 1981. Generalized quantifiers in natural language. Linguistics and Philosophy 4(1): 159-219.

Beghelli, P. and T. Stowell. 1997. Distributivity and Negation. In A. Szabolcsi (ed.), Ways of Scope Taking. Dordrecht: Kluwer.

Brisson, C. 1998. Distributivity, Maximality, and Floating Quantifiers. PhD Dissertation, Rutgers University.

Campos, H. and M. Stavrou 2004. Polydefinite constructions in Modern Greek and in Aromanian. In Tomic (ed.), Balkan Syntax and Semantics, John Benjamins. 137-144.

Carlson, G. 1977. Reference to kinds. New York: Garland. PhD thesis, UMass, Amherst.

Dowty, D. 1987. Collective Predicates, Distributive Predicates, and All. In F. Marshall et al. (eds.), ESCOL ’86: Proceedings of the Third Eastern States Conference on Linguistics. Columbus: Ohio State University.

Etxeberria, U. and A. Giannakidou. 2008. Contextual domain restriction across languages: Definiteness, indefiniteness and the structure of QP, ms.

3 8 2 Syntax - Morphology - Semantics Dimitra Lazaridou-Chatzigoga

Giannakidou, A. 1999. Affective dependencies. Linguistics and Philosophy 22: 367–421.

Giannakidou, A. 2004. Domain restriction and the arguments of quantificational determiners. In Semantics and Linguistic Theory 14. Cornell Linguistics Club: Ithaca, NY. 110-128.

Giannoulopoulou, G. 2007. Η ανάδυση του οριστικού άρθου στην ελληνική και την ιταλική: μία συγκριτική μελέτη. In Proceedings of the 27th Annual Meeting of the Department of Linguistics. Thessaloniki, Aristotle University of Thessaloniki. 78-89.

Gil, D. 1992. Scopal quantifiers: Some universals of lexical effability. In M. Kefer and J. van der Auwera (eds.), Meaning and grammar cross-linguistic perspectives, 303-345.

Heim, I. and A. Kratzer. 1998. Semantics in generative grammar. Malden, MA: Blackwell Publishers.

Holton, D., Mackridge, P. and I. Philippaki-Warburton. 1999. Γραμματική της ελληνικής γλώσσας. Athina: Patakis.

Ioup, G. 1975. The Treatment of Quantifier Scope in a Transformational Grammar. PhD Dissertation, The City University of New York, NY.

Kolliakou, D. 2004. Monadic definites and polydefinites: their form, meaning, and use. Journal of Linguistics 40: 263-333.

Kratzer, A. 1981. The Notional Category of Modality. In H.J. Eikmeyer, and H. Rieser (eds.), Words, Worlds, and Contexts. Berlin, 38-74.

Lazaridou-Chatzigoga, D. 2007. Free Choice Items and definiteness: Evidence from Greek. In E. Puig-Waldmueller (ed.), Proceedings of Sinn und Bedeutung 11. Barcelona: Universitat Pompeu Fabra, 403-417.

Lazaridou-Chatzigoga, D. 2009a. Greek Generic Noun Phrases Involving the Free Choice Item opjosdhipote and the Definite Article, Proceedings of the 2007 Workshop in Greek Syntax and Semantics at MIT, MITWPL 57 (MIT Working Papers in Linguistics), 123-137.

Lazaridou-Chatzigoga, D. 2009b. On definiteness and the co-occurrence of the definite article with other determiners in Modern Greek. Unpublished PhD thesis. Universitat Autònoma de Barcelona.

Syntax - Morphology - Semantics 3 8 3 Major Trends in Theoretical and Applied Linguistics

Link, G. 1983. The logical analysis of plurals and mass terms: A lattice-theoretical approach. In R. Bauerle, C. Schwartze and A. von Stechow (eds.), Meaning, Use and Interpretation of Language. De Gruyter, 302-323.

Lyons C. 1999. Definiteness. Cambridge University Press.

Margariti, A.M. 2007. Quantification and intonation in Modern Greek. In E. Agathopoulou, M. Dimitrakopoulou and D. Papadopoulou (eds.), Selected Papers on theoretical and applied linguistics. Department of Theoretical and Applied Linguistics, School of English, Aristotle University of Thessaloniki, v.I.

Marmaridou-Protopapa, A.S.S. 1984. The study of reference, attribution and genericness in the context of English and their grammaticalization in M. Greek noun phrases. Unpublished PhD thesis. Darwin College. Cambridge.

Matthewson, L. 2001. Quantification and the nature of crosslinguistic variation. Natural Language Semantics 9: 145-189.

McNally, L. and V. Van Geenhoven. 1998. Redefining the weak/strong distinction. Paper presented at CSSP 3, Paris. ms, Universitat Pompeu Fabra and Radboud University.

Milsark, G. 1974. Existential Sentences in English. PhD thesis. MIT.

Roberts, C. 2003. Uniqueness in definite noun phrases,Linguistics and Philosophy 26, 287-350.

Tsamadou-Jacoberger, I. 2002. Le nom en grec moderne: marqueurs et opérations de détermination. Editions L’Harmattan.

Tsili, M. 2001. The quantificational phrases(o) kathe NP, (o) kathenas > universality and distributivity (In Greek). Studies in Greek Linguistics 21, 783-794.

Tunstall, S. 1998. The Interpretation of Quantifiers: Semantics and Processing. PhD thesis. UMass.

Vendler, Z. 1967. Linguistics in Philosophy. Ithaca, NY: Cornell University Press.

Wilson, D. and D. Sperber. 2012. Meaning and Relevance. Cambridge: Cambridge University Press.

3 8 4 Syntax - Morphology - Semantics Adjectival Participles Bearing on Unaccusativity Identification. Evidence from Modern Greek

Elisabeth Mela-Athanasopoulou

Aristotle University of Thessaloniki [email protected]

Abstract

The adjectival properties of past participles have long been studied in the literature (Levin & Rappaport 1986, Markantonatou 1995, Embick 2004, Alexiadou 2001, Meltzer-Asscher 2011, et al). The attributive use of past participles will be postulated, in this paper, as an unaccusativity diagnostic for English and Modern Greek (MG). It will be shown that nouns premodified by adjectival past participles can be subjects of unaccusative (yet, not unergative) verbs, e.g. fallen leaves, expired passport, wilted flowers, etc. In Mela- Athanasopoulou (2007), I argue that past participles of unaccusative verbs, in MG, turn into adjectives through a morphological rule of conversion. In this paper, I will discuss mismatches (clashes) between predictions of past participle-to- adjective conversion and other unaccusativity diagnostics, on the basis of data from MG and English. Whereas the formation of Adjectival Past Participles (APPs) from unaccusative verbs is rather marginal in English, it is quite productive in MG. I will use telicity as a cross-linguistic semantic constraint for the formation of APPs from atelic verbs. Moreover, the distinction between adjectival stative and verbal eventive (resultative) passive participles will also be raised.

1. The Distinction between Adjectival and Verbal Passives

As is well known, adjectival passive participles and verbal passive participles are morphologically identical, i.e. they are represented with the same affix, {-ed} and {-men-}, for English and MG. respectively. In ‘have’ languages (perfective aspect: have en), such as English, however, the phenomenon of APPs is rather marginally productive, especially with unaccusative verbs. Moreover, APPs are closer to passive than to perfect participles. According to Bresnan (1982b), past

Syntax - Morphology - Semantics 3 8 5 Major Trends in Theoretical and Applied Linguistics

participles of intransitive verbs convert into adjectivals and appear as modifying past participles, only if their subjects are themes, that is, they undergo the motion or change of state (hence resultative) specified by the verbs they derive from. Within the GB framework (Chomsky 1981) the APP formation properties are as follows (cited from Levin & Rappaport, p. 624): a) affixation of the passive morpheme {-ed}, b) change of category: [+V, -N] changes to [+V, +N] c) Suppression of the external role of the base verb the APP derives from, d) externalization of an internal role of the base verb, e) absorption of case, and f) elimination of the [NP, VP] position. The main concern here is the identification of the external argument of the base verb the APP premodifies or is predicated of, i.e. the externalization of the internal argument of the base verb. With regard to unergative verbs (e.g. run, swim, cough, cry), subjects are assigned an external argument, namely the thematic role of Agent; whereas with unaccusative verbs (e.g. die, fall, vanish, arrive), subjects lack this active participation. Actually, these surface subjects are non-volitionary passive participants in the action of the verb. The thematic role of agent i.e. external argument is missing with unaccusative verbs. And according to GB theory, such verbs as they fail to assign an external theta role, they also fail to assign accusative case to an object position (Mela-Athanasopoulou, 2007). In GB theory, the surface subject of an unaccusative verb is an object in the deep structure. In other words, an unaccusative verb has an underlying object (which will become the surface subject later in the derivation of the sentence) but no underlying subject, as in (1) (Mela-Athanasopoulou 2007): 1.

3 8 6 Syntax - Morphology - Semantics Elisabeth Mela-Athanasopoulou

Therefore, NP2 ta loulouδiaNom/Pl. will fail to be assigned case ta loulouδiaAcc/

Pl. unless it moves to a subject position where it will receive nominative case from the Infl. position. On the other hand, with unergative verbs (run, cry, cough, resign, swim, etc.), subjects are assigned an external argument, namely the thematic role of Agent as in (2): 2.

Moreover, the argument of the verbs (verbs of manner of motion) bears the agent role as well as the theme role. Thus, the past participles of (3a-3f) and (4a- 4f) are unlikely to occur as adjectivals since they are oriented towards the verb’s agent, that is, the subject, which is a volitionary active participant in the action of the verb and it is normally +animate.

3a. I faγomeni γata 4a. *the eaten cat The cat that has eaten Artfem/sgPastPartfem/sg Nfem/sg I γata pou efaγe The cat which has eaten 3b. to *chasmourimeno moro 4b. *the yawned baby The baby who has yawned Artntr/sg PastPartntr/sg Nntr/sg To moro pou chasmourithike The baby who has yawned 3c. to *petaγmeno kanarini 4c. *the flown canary The canary that has flown Artntr/sg PastPartntr/sg Nntr/sg To kanarini pou petakse The canary that has flown 3d. o *piomenos oδiγos 4d. *the drunk driver The driver who has drunk Artmsc/sPastPartmsc/sg Nmsc/sg O oδiγos pou ipie also: the driver who is drunk The driver who has drunk (lexicalized) 3e. o*hyposhemenos politikos 4e. *the promised politician The politician who has promised Artmsc/sgPastPartmsc/sg Nmsc/sg O politicos pou iposheθike also: the politician who has been The politician who has promised promised (passive) 3f. o *treγmenos aθlitis 4f. *the run athlete The athlete who has run Artmsc/sg PastPartmsc/sg Nmsc/sg O aθkitis pou etrekse The athlete who has run

Syntax - Morphology - Semantics 3 8 7 Major Trends in Theoretical and Applied Linguistics

All the English items of (4a-4f) are ill formed as Adjectival Past Participles since none of them derive from unaccusative verbs (hence the unaccusative diagnostic). One can claim that in ‘have’ languages, such as English, the phenomenon of Adjectival Past Participle formation is non-existent with unergative verbs and little productive with unaccusative ones. Contra Lieber’s view (1980) that adjectival past participles are formed from verbal (perfect and passive) participles by affixation of a null adjectival morpheme {-ed}, Bresnan (1982) proposed that adjectival participles are closer to passive than to perfect participles. Regarding unaccusative verbs, Markantonatou (1995) argues that past participles of such verbs in MG may convert to adjectives, e.g. pesmenos ‘fallen’, pethamenos ‘dead’. In Bresnan’s view then, (3a) i faγomei γata ‘the eaten cat’, should mean ‘the cat was eaten’ rather tan ‘the cat which has eaten’; hence, the ambiguity of the past participle: either adjectival or verbal (4d – the drunk driver). Similarly in MG, the items of (3b-3f) are ill-formed as adjectivals. Items (3d-3f) may occur with an ambiguous meaning. Anagnostopoulou (2003), among others (e.g. Levin & Rappaport 1986) offers some criteria to disambiguate passive participles of any verb ±transitive, but no mention is done on either unergative or unaccusative verbs, with this respect.

2. Adjectival Past Participles Bearing on Unaccusativity Definition

Now returning to the adjectival past participle as an unaccusativity diagnostic, the participles of both MG (3a-3f) and English (4a-4f) data are based on monadic verbs with a single external argument as mentioned earlier in this paper, whereas the adjectival past participles of (5a-5f) are based on monadic verbs with a single internal argument, i.e. they have no underlying subject and fail to assign an external θ-role.

5a. Ta pesmena fila 6a. The fallen leaves Artntr/pl Adjntr/pl Nntr/pl 5b. ta prismena poδia 6b. The swollen feet Artntr/pl Adjntr/pl Nntr/pl 5c. to γermeno δenδro 6c. *the leant tree Artntr/sg Adjntr/sg Nntr/sg 5d. o kaθismenos papous 6d. *the seated grandfather

Artmsc/sAdjmsc/sg Nmsc/sg 5e. I afixθentes taksiδiotes 6e. *the arrived travelers

Artmsc/plAdjmsc/pl Nmsc/pl 5f. I prokipsases δiskolies 6f. *the arisen difficulties

Artfem/plAdjfem/pl Nfem/pl

3 8 8 Syntax - Morphology - Semantics Elisabeth Mela-Athanasopoulou

From the above picture we can see that there are instances (6c-6f) when a putative unaccusative verb, in English, may not give rise to an adjectival past participle unless it is combined with an adjectival modification of time or manner, e.g. *an appeared novel vs. a recently appeared novel, *a seated grandfather vs. a well seated grandfather, *an emerged problem vs. a newly emerged problem, etc. In other words, in English (but not necessarily in MG) adjectival past participles of unaccusative verbs require obligatory modification. Moreover, one has to take into account the telicity requirement for adjectival past participle formation. Past participles of verbs which are not telic (i.e. do not reach an inherent end-point) cannot be used attributively (6c-6f). Those unaccusative verbs of ‘existence’ are stative and atelic and, as such, their adjectival past participles cannot be used attributively, e.g. the *existed solution, the *remained documents. In MG, though, adjectival past participles from atelic unaccusative verbs can be used attributively but not predicatively, e.g. ta prokipsanta provlimata ‘the emerged problems’ vs. *ta provlimata ine prokipsanta ‘the problems are emerged’, I enapominantes politiki ‘the remained politicians’ vs. *I politiki ine enapominantes ‘the politicians are remained’. On the other hand, telic unaccusative verbs such as melt, freeze, collapse, etc. can yield adjectival past participles which are resultative, e.g. i liomeni sokolata ‘the melted chocolate’, I soriasmeni skini ‘the collapsed tent’. For example, the adjectival past participles, ‘melted’ and ‘collapsed’, mean that the chocolate is melted and the tent is collapsed as a result of a melting and collapsing event, respectively (Cleris & Babiniotis, 1999). It is not surprising that the acceptability of unmodified adjectival participles of both transitive and intransitive verbs frequently depends on the noun to be premodified as is shown in (7a-b) and (8a-b):

7a. pliromenos erγatis 8a. pliromenos δolofonos vs. *paid worker paid killer

7b. milimenos sinaδelfos 8b. milimenos politicos vs. *spoken to colleague spoken to politician

The informativeness of the adjectival participles of pliromenos and milimenos in (8a-b) (both used attributively and predicatively) is higher than in (7a-b). See Ackerman and Godlberg (1996) for an account for the infelicity of unmodified Adjectival past participles such as those in (7a-b) or forms such as *a combed child vs. a well-combed child, by means of the Non-redundancy constraint (Ackerman & Goldberg (1996: 21)).

Syntax - Morphology - Semantics 3 8 9 Major Trends in Theoretical and Applied Linguistics

3. Conclusion

In this paper, I have discussed the formation of adjectival past participles in MG as an unaccusativity diagnostic. On the basis of data from MG I have shown that if an adjectival past participle is ill-formed or rejected then the corresponding verb cannot be unaccusative. Moreover, I have raised the issue of telicity as a requirement for Adjectival past participle occurrence and I have shown that adjectival participles cannot be formed from atelic (stative) verbs in English, while they can be marginally formed from such verbs in MG. The whole issue, however, deserves more in-depth research.

3 9 0 Syntax - Morphology - Semantics Elisabeth Mela-Athanasopoulou

References

Ackerman, F. and A. Goldberg. 1996. Constraints on Adjectival Past Participles. In A. Goldberg (ed.), Conceptional Structure. Discourse and Language, 17-30. Stanford, CA: CSLI Publications.

Alexiadou, A. 2001. Functional Structure in Nominals: Nominalization and Ergativity. Amsterdam: John Benjamins.

Anagnostopoulou, E. 2003. Participles and Voice. In A. Alexiadou, M. Rathert & A. von Stechow (eds.), Perfect Explorations. Berlin: Mouton de Gruyter.

Bresnan, J. 1982. The Mental Representation of Grammatical Relations. MIT.

Chomsky, N. 1981. Lectures on Government and Binding. Dordrecht: Foris

Cleris, C. and G. Babiniotis. 1999. Neoelliniki Grammatiki. To rima. (Modern Greek Grammar. The Verb). Athens: Ellinika Grammata.

Embick, D. 2004. On the structure of resultative participles in English. Linguistic Inquiry 35: 335-392.

Levin, B. and M. Rappaport. 1986. The formation of Adjectival Passives. Linguistic Inquiry 17(4): 623-661.

Lieber, R, 1980. The Organization of the Lexicon. PhD dissertation, MIT.

Markantonatou, S. 1995. Modern Greek deverbal nominals: an LMT approach. Journal of Linguistics 31: 267-299.

Mela-Athanasopoulou, E. 2007. Adjectival Participles of Unaccusative verbs. Evidence from Modern Greek. http:icg17.icte.uown.gr/Mela.pdf.

Syntax - Morphology - Semantics 3 9 1 Major Trends in Theoretical and Applied Linguistics

Meltzer-Asscher, A. 2011. Adjectival passives in Hebrew. Evidence for parallelism between the adjectival and verbal systems. Natural Language & Linguistic Theory 29: 815-855.

3 9 2 Stranded Quantifiers, Reconstruction and QR

Masaki Ohno

Showa University [email protected]

Abstract

Sportiche’s (1988) theory of quantifier stranding does not in principle exclude the possibility of quantifiers being stranded in theta-positions. This prediction is not borne out, however, as in *The students were arrested all. Bošković (2001, 2004) states that quantifiers cannot be stranded in theta-positions. In this paper I argue that the object part of Bošković’s generalization is derived from the interaction between reconstruction and QR.

1. Issue: Stranded Quantifiers in Object Position

Quantifiers are not stranded in the internal argument position, or roughly speaking, object position. In (1a-c), the position of the stranded object-related quantifiers seems to be responsible for the ungrammaticality of each sentence.

(1a) *The students arrived all. (1b) *The teachers were arrested all. (1c) *John kicked the students all.

Bošković (2001, 2004) states that quantifiers cannot be stranded in theta- positions. Let us call this Bošković’s generalization (hereafter, BG). Although Bošković argues that his generalization holds of quantifiers in general, irrespective of whether they modify the subject or the object, here I focus on the object part of BG. BG is supported by the contrast in (2) as well. In (2a) the quantifier is stranded by wh-movement in its base-position, resulting in the ungrammaticality. (2b) is reported to be marginally acceptable. This is perhaps because the sentence does not contain a stranded quantifier in the base position.

(2a) *The books, which I will have to read all, are interesting. (2b) ?The books, which I will all have to read, are interesting. (Doetjes 1992)

Syntax - Morphology - Semantics 3 9 3 Major Trends in Theoretical and Applied Linguistics

The contrast in (3) also illustrates the same point. Stranding a quantifier through relativization results in the ungrammaticality as in (3a).

(3a) *the books that John read all (3b) ?the books that all were read by John (3c) the books that were all read by John (Bowers 1993)

In French, however, it looks like object-related quantifiers can be stranded in their theta-positions.

(4) Je les ai vus tous. I them(cl) have seen all ‘I have seen them all’ (Doetjes 1992)

It is not clear, though, whether (4) holds as a counterexample to BG. We could analyze tous as being at VP-adjoined position (right adjoined position) by clitic movement on the way to its final landing site. Or it may be thattous is located at a higher position (e.g., Spec-AgrP) than its base position, but the surface position of tous is obscured by the following clitic movement and overt movement of vus. In this view, it amounts to saying that in fact tous is stranded at Spec of a functional projection higher than VP. If this is tenable, then we can assume that the object part of BG still holds in (4). Suppose that BG is true. Then why is it that quantifiers cannot be stranded in object position? To this problem, I point out that an obligatory application of reconstruction, combined with Quantifier Raising (QR), provides an answer. Specifically, I argue that sentences containing stranded object quantifiers cannot meet the contradictory requirement of these two operations simultaneously, leading to their ungrammaticality.

2. Stranded Quantifiers and Reconstruction

Consider the following examples.

(5a) All the boys aren’t happy. (OKall > not), (OKnot > all) (5b) The boys aren’t all happy. (*all > not), (OKnot > all)

The sentence in (5a) is ambiguous with respect to scope of all and negation. On one reading, called a wide scope reading of all (all > not), it says that for each of the boys it is true that he is not happy (“x [student (x) → ¬happy (x)]). On the other reading, called a wide scope reading of negation (not > all), it says that it is not the case that all the boys are happy (¬”x [student

3 9 4 Syntax - Morphology - Semantics Masaki Ohno

(x) → happy (x)]). Suppose that reconstruction of the (subject) noun phrase to its base position is optionally applied at LF. Then this explains the scope ambiguity in (5a). In contrast, the sentence in (5b) is not ambiguous. It has only a wide scope reading of negation. The initial observation would be that sentences containing a stranded quantifier cancel scope ambiguity. In other words, only a surface scope reading is available to them. I assume that reconstruction of the (subject) noun phrase is obligatorily applied at LF to the position where its stranded quantifier is located. In (5b), the stranded quantifier all requires the A-moved subject the boys to be reconstructed to its base position (i.e., the predicate internal subject position), resulting in a wide scope reading of negation (not > all). Dowty and Brodie (1984) observes that the sentence (6a) below is ambiguous, whereas the sentence (6b) is not.

(6a) The students all didn’t leave. (OKall > not), (OKnot > all) (6b) The students didn’t all leave. (*all > not), (OKnot > all) (Dowty and Brodie 1984)

I take it from the examples above that the position of a stranded quantifier indicates that the A-moved host DP is interpreted as a whole QDP there at LF. Let us describe this requirement as follows.

(7) Reconstruction At LF, host DPs must be reconstructed and interpreted as QDPs in the position where their stranded quantifiers are located.

In (6b), the students is obligatorily reconstructed to the position where all is located in accordance with the rule in (7). I assume that all in (6b) occupies a thematic position, or put differently, it is located at the predicate internal subject position.

(8) The students didn’t [VP all leave].

This results in a wide scope reading of negation in (6b) (and a wide scope reading of all in the case of (6a)). Notice that reconstruction of A-moved DPs to their base position is optional. Thus a wide scope reading of negation in (6a) appears after the obligatory application of reconstruction of the host DP followed by the optional application of reconstruction of the whole QDP to its base position. Regarding optionality of reconstruction, I follow Fox’s (2000) Scope Economy, which disallows semantically vacuous applications of covert operations.

Syntax - Morphology - Semantics 3 9 5 Major Trends in Theoretical and Applied Linguistics

(9) Scope Economy (Fox 2000) Scope-shifting operations cannot be semantically vacuous.

The economy consideration provides an account of the contrast in (10a,b) below.

(10a) The contestants all can win. (OKall > can), (*can > all) (10b) The contestants can all win. (*all > can), (OKcan > all) (Dowty and Brodie 1984)

The rule in (7) predicts that all has a scope over can in (10a), whereas can has a scope over all in (10b). This prediction is borne out. In addition, (10a) has the possibility of undergoing an optional application of reconstruction of all the contestants to its base position. However, this possibility is ruled out by Scope Economy as it has no semantic effect on the sentence.

3. Type Mismatch Problems and Quantifier Raising

Syntactic trees are interpreted node by node by application of semantic rules. Let us see a simple case.

(11) S, t

John VP

kicked > Mary

A two-place predicate denotes a function which maps individuals to a function which maps individuals to truth values. In (11), the two-place predicate kicked takes Mary as its argument by function application and maps to a one-place predicate. This one-place predicate, namely the VP node, is a function which maps any individual that kicked Mary to a truth value “True”. In (11), the VP takes John as its argument by function application and maps to a truth value. With this in mind, consider the sentence below where the subject is a quantificational NP.

(12) Every student kicked Mary

Let us assume that kicked Mary is a one-place predicate which is true of any individual which kicked Mary. From the way the sentence in (11) is interpreted, we may expect that kicked Mary takes every student as its argument by function application.

3 9 6 Syntax - Morphology - Semantics

Masaki Ohno

But this is not possible because quantificational NPs cannot denote individuals. There are several facts that quantificational NPs do not denote an individual. Here is one example discussed in Heim and Kratzer (1998). In the inference below, any argument of type e is expected to make it valid if the argument makes the premise true. This is so when the denotation of came yesterday morning (a set of individuals who came yesterday morning) is a subset of the denotation of came yesterday (a set of individuals who came yesterday), and a sentence whose subject denotes an individual is true iff that individual is a member of the set denoted by the VP.

(13) Subset-to-superset inference x came yesterday morning. \x came yesterday.

The inference in (14) below is expected to be valid if the premise is true, because if [[John]] is a member of [[came yesterday morning]], then [[John]] is also a member of [[came yesterday]].

(14) John came yesterday morning. \John came yesterday.

However, there is a case in which the premise is true but the conclusion is false. QNPs like no boy, few students, when substituted for x in (13), make the inference invalid.

(15) No boy came yesterday morning. \No boy came yesterday.

The inference in (15) is invalid even if the premise is true. There may be some boy who came yesterday evening. A plausible conclusion therefore is that the denotation of QNPs is not of type e. A solution for interpreting the sentence in (12) is to assume that actually the QNP in subject position is a higher order function that takes as its argument the VP-predicate, rather than the other way around. More generally, it is assumed that generalized quantifiers denote a function which maps a predicate to truth value (<t>>).

(16) Every student kicked Mary S, t

Every student VP <t>> kicked Mary >

Syntax - Morphology - Semantics 3 9 7 S, t

QNP <t> VP

Q <, <t>> NP kicked Mary

Every student

kicked every student > <t>

Every student kicked Mary Every student kicked S, t Mary S, t Every student VP Every student VP Major Trends in Theoretical <t>> and Applied Linguistics <t>> kicked Mary kicked > Mary >

Since QNPs denote <t>, given that the semantic type of NPs is , and quantifiers and NPs semantically combine by function application, it follows that the meaning of quantifiers is <, <t>>. S, t (17) S, t QNP <t> VP QNP <t> VP Q <, <t>> NP kicked Mary Q <, <t>> NP kicked Mary Every student Every student

Now, let us consider how the sentence in (18) below is interpreted in which a QNP appears in object position.

(18) John kicked every student.

It is clear, by now, that the sentence in (18) cannot be interpreted as it stands. The object QNP cannot combine with a predicate by function application because of a type mismatch. The LF syntactic structure in (19) is not interpretable unless we assume a semantic approach with flexible type mechanisms.

(19)

VP VP kicked every student kicked > every student<t> > <t> In Heim and Kratzer (1998), it is proposed that type mismatch problems induced by the object QNP are resolved in terms of application of Quantifier Raising (QR), with several additional rules such as ‘traces left by QR denote an individual’ and ‘sister nodes of Q-raised DP denote the set of individuals’. Ultimately, QR creates the interpretable LF structure where it provides the transitive verb in (19) with an argument of type e, and the Q-raised DP with an argument of type .

4. Reconstruction and QR

This section provides an account of the problem why quantifiers cannot be stranded in object position. I argue that the interaction of reconstruction and QR is responsible for the ungrammaticality of sentences with a stranded object quantifier. Consider (20).

3 9 8 Syntax - Morphology - Semantics Masaki Ohno

(20) *John kicked the students all.

The previous section reviewed that QDPs in object position always incur a type mismatch problem. This interpretability problem occurs because the meaning of quantificational expressions is a higher order predicate which takes as its argument a predicate from individuals to truth values (<,t>>). The logical form reconstruction of a host NP requires that the students in (20) move back to its base position, as in (21).

(21) John kicked t [ all the students]

But the resulting LF structure is not interpretable, as it induces a type mismatch. Generally, QR can save type mismatch problems. But in the case of (20) it cannot, because of a contradictory requirement of reconstruction and QR. The position of the quantifier in (20) indicates that the whole QDP should be interpreted at the Q-stranded position, i.e., the object position. If reconstruction does not occur, it leads to a failure of quantification of a stranded quantifier. If reconstruction occurs, the result is a type mismatch and semantic computation is stuck. Therefore object quantifiers cannot be stranded in base positions.

5. Conclusion

In this paper I first reviewed that stranded quantifiers require their host DPs to undergo reconstruction to the position where their modifying quantifiers are stranded. Then I introduced Heim and Kratzer’s version of QR application. I argued that a contradictory requirement of reconstruction and QR provides an account of the ungrammaticality of sentences with stranded object quantifiers. One consequence is that the proposed account predicts that subject quantifiers can be stranded in theta-positions, which departs from part of Bošković’s generalization.

Syntax - Morphology - Semantics 3 9 9 Major Trends in Theoretical and Applied Linguistics

References

Bošković, Ž. 2001. Floating quantifiers and theta-role assignment. NELS 31: 59- 78.

Bošković, Ž. 2004. Be careful where you float your quantifiers.Natural Language and Linguistic Theory 22: 681-742.

Bowers, J. 1993. The syntax of predication. Linguistic Inquiry 24: 591-656.

Doetjes, J. 1992. Rightward floating quantifiers float to the left. The Linguistic Review 9: 313-332.

Dowty, D. and B. Brodie. 1984. The semantics of “floated” quantifiers in a transformational grammar. In M. Cobler et al. (eds.), Proceedings of the 3rd West Coast Conference on Formal Linguistics (WCCFL 3), 75-90. Stanford, CA: Stanford Linguistics Association.

Fox, D. 2000. Economy and Semantic Interpretation. MIT Press.

Heim, I. and A. Kratzer. 1998. Semantics in Generative Grammar. Blackwell

Sportiche, D. 1988. A theory of floating quantifiers and its corollaries for constituent structure. Linguistic Inquiry 19, 425-449.

4 0 0 Syntax - Morphology - Semantics On Scalar Predicative PPs in Spanish1

Isabel Oltra-Massuet and Isabel Pérez-Jiménez

Universitat Rovira i Virgili, CCHS-Consejo Superior de Investigaciones Científicas [email protected], [email protected]

Abstract

We assert the cross-categorial significance of the notion of scalarity, focusing on bare PPs headed by sin/con ‘without/with’. Con-PPs are always non-gradable. With sin-PPs, the mereological properties of the N-complement of P determine the gradability of the PP: mass Ns yield gradable PPs; count Ns generate non- gradable PPs. We investigate the locus of scalar meaning in PPs and how this is compositionally determined. We conclude that the quantificational properties of P, together with the algebraic part structure and the type of scale of the N-complement of P are involved in determining the scalar semantics of bare PPs.

1. Introduction

The goal of this paper is to analyze the scalar properties of so-called bare PPs headed by sin/con ‘without/with’, illustrated in the shaded cells in Table 1, in their function as postnominal modifiers within a DP. Bare PPs are PPs that have a bare (singular or plural) noun in complement position (Grønn et al. 2010, Le Bruyn et al. 2011). Con-headed PPs in (1b) always express non-gradable properties and cannot combine with degree modifiers. In the case ofsin , we argue that the mereological properties of the noun selected by the preposition mold the scalarity of the property expressed by the entire PP. Degree quantification is possible when the noun complement of sin is a mass noun, as in una habitación muy sin luz in (1a) which roughly means ‘a very lightless room’. However, when sin selects a count noun such as corbata ‘tie’, the PP is non-gradable and combination with a degree quantifier becomes ungrammatical.

1 The investigation underlying this work has been partially funded by MICINN projects FFI2009-071114/FILO and FFI2010-22181-C03-01. We thank K. Arregi, I. Bosque, O. Fernández Soriano, C. Kennedy, L. McNally, H. de Swart, and the members of the LyCC group, especially E. Castroviejo and V. Demonte, for comments and discussion of (not always shared) ideas presented in this paper.

Syntax - Morphology - Semantics 4 0 1 Major Trends in Theoretical and Applied Linguistics

Table 1. Bare PPs headed by sin/con ‘without/with’

N-Complement: mass noun N-Complement: count noun una habitación *un hombre una habitación un hombre sin muy sin luz muy sin corbata (1a) sin luz sin corbata without a room very w/o a man very w/o a room w/o light a man w/o tie light tie *un hombre una habitación *una habitación un hombre muy con con con luz muy con luz (1b) con corbata corbata with a room w/ a room very w/ a man w/ tie a man very light light w/ tie

w/o: without; w/: with

This initial paradigm poses the questions in (2).

(2a) What are the factors that determine the gradability of sin-/con-headed PPs? (2b) What is the locus of scalar semantics in the structure of bare PPs, specifically of those headed by sin? What is the contribution of the preposition and of the noun it selects to the scalar properties of the whole? (2c) How is the notion of scalarity expressed in the syntax and semantics of bare PPs? More specifically, how is scalar meaning compositionally determined in these PPs?

Although the cross-categorial nature of scalarity has been acknowledged in the literature (Sapir 1944, Bolinger 1972), there is no systematic study on the scalar structure of PPs – with the exception of directional PPs (see e.g. Grolla 2004, Den Dikken 2010, Svenonius 2010, among others) –, and, to the best of our knowledge, the type of PP that is the topic of this paper has not been investigated with respect to scalarity. We believe that the study of bare PPs that express properties can help answer questions with far-reaching consequences (that will not be tackled here) such as whether we can establish cross-categorial generalizations about the interaction among different types of syntactico- semantic properties: countability-gradability/scalarity, scalarity-telicity, countability-telicity, or whether there is a cross-categorial universal typology of scale structures.

2. Theoretical Background

In this section, we outline the main theoretical assumptions that will be relevant for our proposal.

4 0 2 Syntax - Morphology - Semantics Isabel Oltra-Massuet and Isabel Pérez-Jiménez

2.1. Scalarity and Gradation

We take a scalar predicate to be a predicate that expresses a scale, where a scale is defined as in (3).

(3) A scale is “a dense linearly ordered set of points, or ‘degrees’, where the ordering is relativized to a dimension.” (Kennedy 1999:43)

Thus, a scale must include three elements: a set of degrees [D], with a linear ordering [R], in a specific dimension [Δ]. Scales associated with scalar predicates (e.g. adjectives) vary in structure. An adjective like alto ‘tall’ denotes a set of degrees of height which are linearly ordered as represented in (4). This scale has neither a minimal nor a maximal value and is therefore an open scale. On the other hand, an adjective like despierto ‘awake’ also denotes a set of linearly ordered degrees of awakeness, but in this case, the scale has a minimal value, as in (5), i.e. there is a minimal positive point of awakeness below which one can no longer predicate the awakeness of something. It is therefore a lower-closed scale.

(4) (5)

Scalar predicates can combine with degree modifiers, (6a), can appear in comparative structures, (6b), and some of them accept measure phases (Kennedy & McNally 2005 [K&M henceforth]), (6c).

(6a) muy alto ‘very tall’, poco despierto ‘little awake’ (6b) más alto que Pedro ‘taller than Pedro’, más despierto que Juan ‘more awake than Juan’ (6c) 3 metros de alto ‘3 meters of tall’

We assume that these modifiers are generated in a degree phrase (DegP) that merges over the projection occupied by the scalar predicate, e.g. the adjective in (7) (cf. Kennedy 1999, Corver 1997).

(7) DegP

very AP

For languages like English, K&M (2005) have shown that different degree modifiers are sensitive to the scale structure of the predicate they combine with. Spanish lacks such a systematic analysis, but as we will show, the scale structure

Syntax - Morphology - Semantics 4 0 3 Major Trends in Theoretical and Applied Linguistics

of the complement of sin is crucial for the interpretation of the entire PP. In order to determine the scale structure of the PPs, we use the correlation established in K&M (2005) for adjectives, according to which, relative adjectives are linked to open scales while closed scales are associated with absolute adjectives. For instance, an adjective like tall is relative, because its meaning is contextually determined in relation to some standard degree of tallness for a comparison class, so that a sentence like (8a) can be true or false depending on the context provided by the comparison class; therefore, it can give rise to implications like (8b). However, a sentence like (9), with an absolute adjective, expresses that Juan has some degree of the property ‘being awake’ that is different from zero, independently of the context.

(8a) Juan es alto. ‘Juan is tall’ Situation: Juan is 1.50 m tall. i. Comparison class: 10 year old children. Standard degree for 10 year old children: 1.33 m. → The sentence is true. ii. Comparison class: 40 year old men. Standard degree for 40 year old men: 1.65 m. → The sentence is false. (8b) Juan es más alto que Pedro, aunque los dos son bajos (para ser jugadores de baloncesto). ‘Juan is taller than Pedro, although both are short (as basketball players)’ (9) Juan está despierto. ‘Juan is awake’

For example, we can determine that the PP sin luz ‘without light’ in, for instance, una habitación sin luz ‘a room without light’ expresses a gradable contextually-dependent property, i.e. a relative property. We know that the property is relative because we can establish comparisons of the type tenemos una habitación más sin luz que otra, pero ambas tienen luz ‘we have a room more without light than another, but both have light’. Consequently, the predicate sin luz ‘without light’ will be associated with an open scale.

2.2. The Mereological Structure of Nouns

We claim that the mereological structure of the noun complement of sin – specifically, the count/mass distinction – is crucial to determine the scale structure of the PP it heads. Count nouns, like tie, denote entities that cannot be further subdivided while keeping their nature and name; so, a part of a tie is not a tie. Conversely, mass nouns have cumulative reference; they denote substances that “can be

4 0 4 Syntax - Morphology - Semantics Isabel Oltra-Massuet and Isabel Pérez-Jiménez

divided ad infinitum while keeping their nature and name, such as water” (Bello 1847:§123, from Bosque 1999:8, our translation), so that a part of water is also water. They differ in the type of quantifiers they take. For instance, only count nouns accept numeral quantifiers, as in Table 2 (1); or, only mass nouns allow indefinite quantifiers like singular mucho in (2). Nouns like color, which accept both types of quantifiers, illustrate the fact that certain nouns are ambiguous, so we can have much color and also three colors.

Table 2. Mass vs. Count nouns. Distinguishing criteria (not exhaustive).

Mass Noun Count Noun

1. With numeral quantifiers and multipliers *tres {importancias, humos} tres {corbatas, colores} three importances, smokes three {ties, colors} *muchas importancias2 muchas {corbatas, colores} many importances many {ties, colors} 2. With indefinite quantifiers (non-cardinals) poca/mucha {energía, importancia, luz} *poca corbata little/much {energy, importance, light} little tie poco/mucho {humo, color} little/much {smoke, color} 3. With the adjectival quantifier medio *media {energía, importancia} media corbata half energy, importance half tie 4. In the construction una cantidad de N una cantidad de {energía, humo} *una cantidad de corbata a quantity of energy, smoke a quantity of tie 2

2.3. The Structure of Postnominal Modifiers

With respect to postnominal modifiers, we assume that they are generated in Romance as predicates in clause structure. Specifically, as complements inside a predicate phrase (PredP), (Campos & Stavrou 2004, see also Bowers 2001).3 Hence, the structure for una habitación sin luz ‘a room without light’ is the one in (10). The PP expresses a property that is predicated of a subject.

2 This type of quantification is possible with mass nouns, but only with a concomitant recategorization process, as in muchas energías ‘many energies’ or muchas luces ‘many lights’ (see Bosque 1999). 3 See Alexiadou et al. (2007) for a discussion of alternative proposals.

Syntax - Morphology - Semantics 4 0 5 Major Trends in Theoretical and Applied Linguistics

(10)

The syntactic structure for una habitación muy sin luz ‘a room very without light’ is represented in (11). The degree modifier is generated in a degree phrase that merges above the complement of Predº (Bowers 2001), i.e. above the bare PP in the case at hand.4

(11)

4 A Degree projection is also postulated within the structure of locative and directional PPs in Den Dikken (2010) and Svenonius (2010).

4 0 6 Syntax - Morphology - Semantics Isabel Oltra-Massuet and Isabel Pérez-Jiménez

3. Scalarity in Sin-Headed Bare PPs

In this section, we explore the scalar properties of sin-headed bare PPs. As we have already mentioned, the mereological structure of the noun complement of sin (whether count or mass) determines the gradability properties of the entire PP. When sin selects a bounded or count noun like corbata ‘tie’, (12) – remember that these nouns are incompatible with the singular quantifier mucho/mucha ‘much-masc/fem’, (13) – the PP conveys a non-gradable negative property (that can be paraphrased as ‘absence of’), and combination with a degree quantifier is not possible, (14).

(12) un hombre sin corbata, un vestido sin cinturón a man w/o tie a dress w/o belt (13) *sin mucha corbata, *sin mucho cinturón w/o much tie w/o much belt (14) *un hombre muy sin corbata, *un vestido más sin cinturón que otro a man very w/o tie a dress more w/o belt than other

On the contrary, when sin selects an unbounded or mass noun as complement, (15a), the PP expresses a gradable property, (15b).

(15a) un ambiente sin humo, una habitación sin luz a environment w/o smoke a room w/o light (15b) un ambiente muy sin humo, una habitación muy sin luz a environment very w/o smoke a room very w/o light

Within the class of mass nouns, we show in the next subsections that the subclassification in (16) is crucial to explain the scalar properties of sin-headed bare PPs. Recall that all these nouns are compatible with the singular quantifier mucho/mucha ‘much-masc/fem’, or some other construction from the set of diagnoses in Table 2 for plurals, such as una cantidad de ‘a quantity of’.

(16) Mass nouns (16a) Scalar nouns (nouns expressing a scale) Nouns expressing a linear scale Dimension nouns: volumen ‘volume’ (mucho volumen ‘much volume’) Quality nouns: importancia ‘importance’ (mucha importancia ‘much importance’) Nouns expressing a non-linear scale: color ‘color’ (mucho color ‘much color’) (16b) Stuff-denoting nouns: luz ‘light’ (mucha luz ‘much light’) (16c) Plural nouns: piedras ‘stones’ (una cantidad de piedras ‘a quantity of stones’)

Syntax - Morphology - Semantics 4 0 7 Major Trends in Theoretical and Applied Linguistics

3.1. Scalar Nouns

Within the class of scalar nouns, we establish a difference between nouns denoting non-linear scales like color, and nouns denoting a linear scale, which can be further subdivided into dimension nouns like volume and quality nouns like interest, as in (16) above. Nouns like volumen ‘volume’ (or energía ‘energy’) can express a physical dimension, a magnitude that can be measured on a numerical scale. Under this interpretation, they are scalar nouns that express a set of linearly ordered values, with a minimal point and without a maximal value, (17)-(18). These nouns admit measure phrases that select a specific value on the scale, (19).

(17a) volume: VOLUME ‹ › volume: VOLUMEVol., ‹ Vol.Deg., < › Δ Vol., Vol.Deg.,D R < Δ D R energy: ENERGY ‹ › (17b) energy: ENERGYEnerg., ‹ Energ.Deg., < › ΔEnerg., Energ.D Deg.,R < Δ D R

(18) 3 Minimal point: 0 m / 30 joules Minimal point: 0 m / 0 joules

(19) 5m3 de volumen, 200 gigajulios de energía

5m3 of volume, 200 gigajoules of energy

However, when these (mass and scalar) nouns are selected by sin, the property expressed by the PP is non-gradable, (20), which seems to go against our initial generalization that a bare PP is gradable whenever the noun complement of sin is a mass noun. To account for these cases, we must explore the interaction between the specific scalar properties of dimension nouns and the scalar properties of sin.

(20a) *una esfera muy sin volumen (= casi un círculo) a sphere very w/o volume almost a circle

(20b) *un rayo muy sin energía (= un rayo de pocos julios) a ray very w/o energy a ray of few joules

Sin has been analyzed as a (negative) quantifier, since it licenses negative polarity items in many languages (also in Spanish, (21)).

(21) *Vino con nadie – Vino sin nadieNPI ‘He came *with/without anybody’ (See Bosque 1980 for Spanish; Espinal 2002 for sense in Catalan; de Swart & Sag 2002, Corblin et al. 2004 for sans in French)

4 0 8 Syntax - Morphology - Semantics Isabel Oltra-Massuet and Isabel Pérez-Jiménez

Our claim is that when sin heads a PP with a scalar noun, it projects the external argument onto the minimal value on the scale associated with the noun. In the case of esfera sin volumen ‘sphere without volume’, (22), this minimal value corresponds to degree zero, because volumen denotes a lower-closed scale, (23). A sphere without volume is, therefore, a sphere with a volume of 0m3, i.e. a circle.

(22)

(23)

sin volumen Sin, (24), behaves as a measure phrase, (25), or degree quantifier, (26): it combines with predicates expressing a scale and sets a specific value on that scale: the minimal value.5 (See K&M 2005 on measure phrases).

(24) [[sin]] = λGλx.∃d [d = Min(SG) (C) Λ G(d)(x)] (25) [[5 m3]] = λGλx.∃d [d >5 m3 Λ G(d)(x)] (26) Common denotation for degree quantifiers/measure phrases [[Deg(P)]] = λGλx.∃d [R(d) Λ G(d)(x)] (K&M 2005:367) (R corresponds to the restriction over degrees that distinguishes the semantics of degree morphemes among themselves)

Once sin sets a value on the degree scale, combination with a degree quantifier as in *una esfera muy sin volumen ‘a sphere very without volume’ becomes impossible. As with adjectives, (27), once some scalar property has been assigned some specific value by means of a measure phrase, modification by a degree quantifier is not possible.

(27a) very old – degree quantifier (27b) two years old – measure phrase (27c) *very two years old

5 Note that we assert that sin combines with scalar predicates. The impossibility that sin combines with adjectives must be analyzed as a strictly syntactic restriction.

Syntax - Morphology - Semantics 4 0 9 Major Trends in Theoretical and Applied Linguistics

Consider now the grammatical example in (28a). In this case, the noun volumen does not denote a dimension of the object, a magnitude, but a quality or property of the haircut.6 These cases should be treated on a par with (29), where we have abstract nouns like importancia ‘importance’ or interés ‘interest’ that appear in gradable bare PPs.

(28a) una melena muy sin volumen (≈ una melena con poco volumen) a haircut very w/o volume (≈ a haircut w/ little volume) (28b) una persona muy sin energía (≈ una persona con poca energía) a person very w/o energy (≈ a person w/ little energy) (29) un asunto muy sin importancia; una noticia muy sin interés a issue very w/o importance; a news very w/o interest

As claimed in the literature, these quality nouns have a reading in which they measure the degree to which some quality is applied to a certain entity (NGRALE: § 12.1i, § 22.13m; Zwarts 1992: 154).7 In this interpretation, these nouns are scalar. The scale they denote is open, (30)-(31). The preposition sin, by virtue of its semantics, forces a process of scalar coercion on that scale. When sin combines with a noun that denotes an open scale, the property expressed at the PP level denotes a bounded open scale (Kennedy 2007, n. 28), i.e. a scale that approaches a zero value without including it.

(30)

(31)

6 In these cases, the external argument of PredP co-determines the dimension expressed by the noun complement of P, and hence, the scalar structure of the property expressed by the PP. Note that the interpretation of multidimensional gradable adjectives like seco ‘dry’ as relative or absolute adjectives also varies depending on the entity they are predicated of (camisa seca ‘dry shirt’ vs. clima seco ‘dry weather’, K&M 2005:49). 7 They can also express the name of a property, (i). Compare this meaning with the degree meaning, (ii): “Ana possesses certain degree of intelligence that is higher than the degree of intelligence of Pedro”. (i) La inteligencia de Ana la hizo famosa. ‘Ana’s intelligence made her famous’ (ii) La inteligencia de Ana es mayor que la de Pedro. ‘Ana’s intelligence is greater tan Pedro’s’

4 1 0 ( ) Syntax - Morphology - Semantics muy sin {importancia/volumen} ‘very without {importance/volume}’

Isabel Oltra-Massuet and Isabel Pérez-Jiménez

Consequently the scale of the property formed at the PP level is also open; note the implication patterns in (32).

(32a) Este asunto parece más sin importancia que ese para el público general,

this issue seems more w/o importance than that for the general public general, aunque ambos pueden ser importantes. general although both can be important (32b) una melena más sin volumen que otra, aunque ambas tienen a haircut more w/o volume than another, though both have bastante volumen plenty.of volume

As an open scale where no value has been set, it is possible to add a degree modifier. Inmuy sin importancia, or muy sin volumen in (33), muy ‘very’ indicates that the subject possesses the property in a degree that is close to the minimal value.

(33) ( ) muy sin {importancia/volumen} ‘very without {importance/volume}’

Now consider nouns like color ‘color’, sabor ‘taste’, olor ‘smell’, which have a mass interpretation, (34a), and a count interpretation, (34b). As expected, when color is a count noun, it gives rise to non-gradable PPs, as in (35b). When it is a mass noun, it gives rise to a gradable PP, as in (35a).

(34a) mucho color much color (34b) dos, tres, muchos … colores two three many colors

(35a) una foto muy sin color (= descolorida) a photo very w/o color (= faded) (35b) líquido para obtener [fotos sin color] (= en blanco y negro) → *muy sin color liquid to obtain photos w/o color (= in black and white) → *very w/o color

It has been claimed in the literature (Fleischhauer & Gamerschlag 2010) that nouns like color denote non-linear scales, i.e. scales with a dimension and a set

of values, whose parameter of ‘ordering relation’ is not specified, (36).

(36a) color: COLOR ‹Color., Color. Val., Rc› [Rc = ordering determined by context] Δ D R

Syntax - Morphology - Semantics 4 1 1 Major Trends in Theoretical and Applied Linguistics

(36b)

[From: E. F. Schubert, 2006, Light-Emitting Diods, CUP]

Given that the scale structure of the noun is inherited at the PP level and non- linear scales cannot be graded, because gradability requires an ordering relation between the values, what is the source of gradability in (35a)? We claim that sin selects a linear scalar dimension associated with one of the properties expressed in the meaning of the noun, e.g. ‘intensity’ or ‘brightness’. The intensity scale or the brightness scale are linear and open, so that sin coerces them into bounded open scales. Consequently, the PP can be graded.

3.2. Stuff -denoting Nouns

When the complement of sin is a stuff -denoting noun, the bare PP is gradable, (37).

(37a) un ambiente muy sin humo a environment very w/o smoke (37b) una habitación muy sin luz a room very w/o light (37c) un rostro muy sin maquillaje a face very w/o make-up (37d) un café muy sin azúcar a coff ee very w/o sugar

Again, the mereological (cumulative) structure of the N complement of sin gives rise to the expression of a gradable property at the level of PP; roughly, the property

4 1 2 Syntax - Morphology - Semantics Isabel Oltra-Massuet and Isabel Pérez-Jiménez

of tener humo ‘having smoke’ or tener luz ‘having light’. There is a homomorphic relation between the referential properties of the noun and the scale structure expressed by the PP. Given that these nouns are unbounded and cumulative, the scale associated with the property expressed by the PP is open. In fact, this scale is a bounded open scale, given the semantics of sin, and degree modification is possible.8

3.3. Plural Nouns

As for plural nouns, since they also have cumulative reference (Quine 1960, Link 1983), they give rise to gradable sin-headed bare PPs, as expected, (38).9

(38a) un pueblo muy sin cuestas a town very w/o hills (38b) una piel muy sin impurezas a skin very w/o impurities (38c) muy sin arrugas very w/o wrinkles

4. Scalarity in Con-Headed Bare PPs

PPs introduced by the preposition con are non-scalar and non-gradable, independently of the mereological and scalar properties of the noun, (39).10

8 However, cases like those in (i) are not possible. An important difference between (i) and the examples in the text, which could open a way to explain this contrast, is that in (i) there is a relation of inalienable possession between the nouns in specifier and complement of PredP respectively.

(i) a. *un pozo muy sin agua a well very w/o water b. *una paella muy sin arroz a paella very w/o rice 9 There are also examples judged as ungrammatical by the speakers consulted (*un árbol muy sin hojas ‘a tree very without leaves’, *una escuela muy sin estudiantes ‘a school very without pupils’). There also seems to be a relation of inalienable possession in these cases. 10 There are also cases like un chicle muy con sabor a frutas ‘a chewing gum very with taste to fruit’, muy con olor a jazmín ‘very with smell of jasmine’ (Google) that are judged as grammatical by the same native speakers that reject *un chicle muy con sabor ‘a chewing gum very with taste’, *muy con olor ‘very with smell’. Although we have no account for this contrast at present, it is obvious that the presence of a complement of the noun plays a central role.

Syntax - Morphology - Semantics 4 1 3 Major Trends in Theoretical and Applied Linguistics

(39a) count noun un chico con corbata – *un chico muy con corbata a boy w/ tie a boy very w/ tie

(39b) mass noun – stuff-denoting noun una habitación con luz – *una habitación muy con luz a room w/ light a room very w/ light

(39c) mass noun – scalar noun: dimension una melena con volumen – *una melena muy con volumen a haircut w/ volume a haircut very w/ volume

(39d) mass noun – scalar noun: quality un asunto con importancia – *un asunto muy con importancia a issue w/ importance a issue very w/ importance

(39e) noun denoting a non linear scale un vino con sabor – *un vino muy con sabor a wine w/ taste a wine very w/ taste

(39f) plurals una piel con impurezas – *una piel muy con impurezas a skin w/ impurities a skin very w/ impurities

The crucial question is where the difference between sin and con lies. In our analysis, sin as a negative operator has access to the semantic structure of its complement and can modify it. Con is not an operator. We believe that this is the research path to account for the – semantic and possibly also structural – differences between sin and con. This will be our next step in the investigation of these data.

5. Conclusions

In this paper we have analyzed the scalar properties of some bare predicative PPs in Spanish, and we have established that the factors involved in determining the gradability of these PPs are: a) the quantificational nature of the preposition, and b) the semantic properties relative to the mereological structure and the scalarity of the noun complement of P, in the case of sin-headed PPs. Hence, a first conclusion that can be drawn on the cross-categorial status of the notion of scalarity is that the scalar properties of bare PPs that express properties parallel those exhibited by adjectives. There are bare PPs that express scalar properties and also bare PPs that express non-gradable properties. As with adjectives, the

4 1 4 Syntax - Morphology - Semantics Isabel Oltra-Massuet and Isabel Pérez-Jiménez

external argument that bare PPs are predicated of can determine the scalar properties of the properties expressed at the PP level (multidimensional PPs). We have further established certain preliminary distinctions in the realm of scalar nouns that deserve a more thorough analysis. We have distinguished between scalar nouns denoting a linear scale (open or closed) and scalar nouns denoting a non-linear scale, a property that has syntactic as well as semantic consequences. There are many open issues, and other questions remain still unexplored. Among the former, a common semantics for sin and con is still to be developed, as well as a unified semantics forsin that accounts for its combination with both count and mass nouns. Since sin can combine with any type of noun, be it a proper name as in una fiesta sin Juan ‘a party without Juan’, or any kind of count noun, as in una persona sin {hogar, trabajo, enemigos} ‘a person without home, work, enemies’, the proposal that the semantics of sin operates on the scale structure of its complement entails the assumption that all nouns are scalar entities. One could assume a trivial scalarity for count nouns that, for instance, could express a binary scale [0, 1].11 Another possibility would be to assume GrØnn et al.’s (2010) and Le Bruyn et al.’s (2011) proposal that bare nouns appearing as complements of P contain an existential quantification. Whereas in the presence of a count noun we would predicate the existence of certain individual, in the case of mass nouns it would quantify over the existence of certain degree of the property expressed in the noun. The negative quantification of sin would operate on the existence of something in both cases, an individual x or a certain degree d. Whereas in the former we would obtain ¬$x, in the latter quantification would be over the degree of a gradable property, ¬$d. In this second case, the effect of negative quantification would be a scale reversal, so that we would obtain a certain minimal degree below the standard of the comparison class that would allow for further degree modification. With count nouns, we would get the absence of the individual. Among the unexplored questions, we can find the impossibility of having double quantification in the phrases under study, (40), or a conspicuous crosslinguistic contrast. In languages like Catalan, sin-headed bare PPs can never be graded, (41).

(40) *un asunto muy sin mucha importancia a issue very w/o much importance

11 Thanks to H. de Swart and C. Kennedy for suggesting the lines of reasoning summarized in the text. See Kennedy (2012) for an analysis of all nouns as measure functions.

Syntax - Morphology - Semantics 4 1 5 Major Trends in Theoretical and Applied Linguistics

(41) *un assumpte molt sense importància a issue very w/o importance

The paradigm we have described in this paper cannot be found in other Romance languages, like French or Italian, or in Germanic languages, like English or German. It is thus necessary to investigate the property – apparently unique to Spanish – that allows the existence of these cases.

4 1 6 Syntax - Morphology - Semantics Isabel Oltra-Massuet and Isabel Pérez-Jiménez

References

Alexiadou, A., L. Haegeman and M. Stavrou. 2007. Noun phrase in the generative perspective, Berlin, Mouton de Gruyter.

Bello, A. [1847] 1981. Gramática de la lengua castellana, edited by R. Trujillo, Tenerife.

Bolinger. D. 1972. Degree words, The Hague, Mouton.

Bosque, I. 1980. La preposición sin, Lingüística Hispánica, 3: 71-85.

Bosque, I. 1999. El nombre común, in I. Bosque and V. Demonte (eds.), Gramática descriptiva de la lengua española, Madrid, Espasa, 1:3-75.

Bowers, J. 2001. Predication, in M. Baltin and C. Collins (eds.), The handbook of contemporary syntactic theory, Oxford, Blackwell: 299-333.

Campos, H. and M. Stavrou. 2004. Polydefinites in Greek and Aromanian, in O. Tomic (ed.), Balkan syntax and semantics, Amsterdam, John Benjamins: 137-173.

Corblin, F., V. Déprez, H. de Swart and L. Tovena. 2004. Negative concord, in F. Corblin and H. de Swart (eds.). Handbook of French Semantics, Stanford, CSLI Publications: 427-461.

Corver, N. 1997. The internal syntax of the Dutch extended adjectival projection, Natural Language and Linguistic Theory, 15: 289-368.

Den Dikken, M. 2010. On the functional structure of locative and directional PPs, in G. Cinque and L. Rizzi (eds.), Mapping spatial PPs, Oxford, Oxford University Press: 74-126.

Espinal, M.T. 2002. La negació, in J. Solà et al. (eds.) Gramàtica del català contemporani. Barcelona: Empúries, Vol. 3:2727-2797.

Syntax - Morphology - Semantics 4 1 7 Major Trends in Theoretical and Applied Linguistics

Fleischhauer, J. and T. Gamerschlag. 2010. The interaction of verbs and arguments in scalar composition, handout of the talk given at Workshop on the lexicon-syntax interface: Verb meaning, event semantics and argument structure, UAB, December 2-3 2010.

Grolla, E. 2004. Prepositions, scales and telicity: A case study, in Proceedings of the 23rd West Coast Conference on Formal Linguistics, Somerville, MA, Cascadilla Press: 293-303.

Grønn, A., B. Le Bruyn, H. de Swart and J. Zwarts. 2010. Bare PPs across languages, Ms. Utrecht University.

Kennedy, C. 1999. Projecting the adjective. The syntax and semantics of gradability and comparison, New York & London, Garland.

Kennedy, C. 2012. The Composition of Incremental Change. In V. Demonte and L. McNally (eds), Telicity, Change, State: A Cross-categorical View of Event Structure, Oxford, Oxford University Press, 103-121.

Kennedy, C. 2007. Vagueness and grammar: The semantics of relative and absolute gradable adjectives, Linguistics and Philosophy, 30.1:1-45.

Kennedy, C. and L. McNally. 2005. Scale structure, degree modification, and the semantics of gradable predicates, Language, 81: 345-381.

Le Bruyn, B., H. de Swart and J. Zwarts. 2011. Mass-count distinctions in bare PPs, Ms. Utrecht Institute of Linguistics OTS, Utrecht University.

Link, G. 1983. The logical analysis of plurals and mass terms: A lattice theoretical approach, in R. Bäuerle, C. Schwarze and A. von Stechow (eds.), Meaning, use, and the interpretation of language, Berlin, de Gruyter: 302-323.

RAE y Asociación de Academias de la Lengua Española. 2009. Nueva gramática de la lengua española, Madrid, Espasa. [NGRALE]

Quine, W.V.O. 1960. Word and object, Cambridge, MA, The MIT Press.

Sapir, E. 1944. Grading: A study in semantics, Philosophy of Science, 11: 93-116.

Svenonius, P. 2010. Spatial P in English, in G. Cinque and L. Rizzi (eds.), Mapping spatial PPs, Oxford, Oxford University Press: 127-160.

4 1 8 Syntax - Morphology - Semantics Isabel Oltra-Massuet and Isabel Pérez-Jiménez

De Swart, H. and I.A. Sag. 2002. Negation and Negative Concord in French. Linguistics and Philosophy 25: 373-417.

Zwarts, J. 1992. X´-syntax/X´-semantics. On the interpretation of functional and lexical heads, Utrecht, Led, OTS.

Syntax - Morphology - Semantics 4 1 9 Major Trends in Theoretical and Applied Linguistics

4 2 0 Syntax - Morphology - Semantics Measuring the Productivity of Noun- Deriving Suffixes across Languages: Greek -tita vs. English -ness1

Theofanis Papoutsis

University of Athens [email protected]

Abstract

This paper investigates the productivity of the Greek noun-forming suffix -tita, in comparison to its English counterpart -ness. To this aim, it exploits empirical data drawn from corpora and employs an array of corpus-based productivity measures. The analysis shows that, while both suffixes are considerably productive in relation to the derivational system of the relevant languages, -ness is interlingually more productive than -tita. Furthermore, both suffixes attach productively to adjectival bases; yet, bases belonging to other word-categories (e.g. nominal, pronominal, etc.) are also productively employed. Moreover, the combinations of adjectival bases with -tita and -ness exhibit varying degrees of productivity.

1. Introduction

The Greek suffix -tita2 and the English -ness are similar, in that they both attach to adjectival bases and form abstract nouns which denote the same quality, property or state as that of the adjective itself (e.g. kakós3 ‘bad’ > kakótita

1 I wish to thank Dr Katerina Stathi for her fruitful comments and suggestions as well as her encouragement. Thanks are also extended to Dr Dionysis Goutsos for providing me with the raw data from the Corpus of Greek Texts. Any errors are my own responsibility. 2 The Greek data appear in broad phonemic transcription (with the lexical stress indicated over the relevant vowel). Since inflectional endings are irrelevant for the present discussion, they are not separated from the preceding derivational suffixes. 3 Adjectives in Greek, contra English, inflect for gender (masculine – feminine – neuter), case (nominative – genitive – accusative) and number (singular – plural). In this paper, the masculine-nominative-singular form is given only. Ditto for adjectival suffixes.

Syntax - Morphology - Semantics 4 2 1 Major Trends in Theoretical and Applied Linguistics

‘badness’ – bad > badness). But what exactly renders -tita and -ness interlingually equivalent?4 After all, in both languages, there are other noun-deriving suffixes that are tacked onto adjectives and thus are in rivalry with -tita and -ness. In Greek, these include -osíni and -ia (Ralli 2005: 149, Xydopoulos 2008: 184), while in English, they include -ity and -ce/-cy (Plag 2003: 88, 91-92). However, the use of both -tita and -ness is subject to the fewest possible structural constraints. More specifically, although -ness can be suffixed to virtually any adjective, -ity and -ce/-cy impose certain selection requirements. Thus, –ity is exclusively tacked onto adjectives of Latinate origin, prime examples being adjectives bearing the suffixes –able/-ible, -al, –ic or ending in the phonetic string [ɪd] (e.g. readability, centrality, historicity, solidity), while -ce/-cy attaches to adjectives in -ant/-ent (e.g. efficient > efficiency) and -ate (e.g. adequate > adequacy) (see Baayen & Lieber 1991, Plag 2003 for details). The Greek nominalising suffixes present a similar picture, in that -osíni is exclusively tacked onto non-suffixally derived adjectives which end in either the [-learned] inflection -os (e.g. anixtómjalos ‘open-minded’ > anixtomjalosíni ‘open-mindedness’) or the [+learned] -on (e.g. isxiroγnómon ‘stubbornness’), -ia is added only to [+learned] adjectives in -ίs (e.g. evγenís ‘polite’ > evγénia ‘politeness’), while -tita is suffixed to practically any adjectival base, excluding adjectives in -on and -ίs though (cf. Papoutsis 2012). Moreover, both -tita and -ness are intuitively felt to be significantly more productive than their rivals. Simply put, if a Greek or English speaker wishes to derive an abstract noun from an adjective, then s/he will resort to -tita and -ness as a first choice, respectively. This intuitive feeling about the high productivity of -ness has been verified and quantified by a multitude of empirical evidence, drawn from corpora (see, among others, Baayen & Lieber 1991: 824-826, Baayen & Renouf 1996, Plag 2003, Plag et al. 1999), whereas regarding the productivity of -tita (as well as of any other Greek morphological process), there is practically no empirical evidence to this day. Thus, the aim of this corpus-based study was threefold: (a) to verify and quantify our intuitions about the high productivity of both -tita and -ness; (b) to compare and contrast the productivity degrees of an interlingual pair of noun- forming suffixes; and (c) to apply the quantitative approach to morphological productivity to Greek data for the first time (and gauge any practical problems arising from this application).

4 The correspondence between -tita and -ness is discussed here from the perspective of morphology, not from that of translation, meaning that a given -tita word will not necessarily translate as a -ness word and vice versa. This is illustrated by egirótita, the English translation equivalent being an -ity formation, i.e. validity.

4 2 2 Syntax - Morphology - Semantics Theofanis Papoutsis

2. Data Collection, Processing and Categorisation

The Corpus of Greek Texts (henceforth CGT) served as a source of data on – tita. The CGT comprises approximately 30 million word tokens of present-day Greek language production (1990-onwards), including both written and spoken discourse (ratio 9/1). Being the most balanced and representative corpus, currently available for Greek (cf. Goutsos 2010), the CGT can fruitfully be exploited for morphological research purposes.5 Obviously, comparing affixes in different corpora raises issues of comparability regarding both the corpus size and the contained text-types (Baayen & Lieber 1991: 820). Unfortunately, no reliable parallel corpora for the Greek-English pair are freely available. Therefore, I opted for the freely available 425-million-word Corpus of Contemporary American English (henceforth COCA), which is a greatly balanced and representative collection of contemporary English (including both oral and written discourse) (Davies 2011) and thus is suitable for morphological research purposes.6 More specifically, in order to have an English corpus of comparable size to that of the CGT, the data on -ness were gathered from an approximately 30-million-word reduced version of the COCA (time period: January 2009-August 2010) (cf. Davies 2010). Initially, raw data were extracted – by means of string searches – from the CGT word-frequency lists and the online version of the COCA. These raw data were then analysed manually, in order to exclude any irrelevant lexical items. Wherever necessary, I checked words in their context in the corpora used and consulted current dictionaries (namely, Babiniotis (2002) and Institute for Modern Greek Studies (1998) for Greek; Stevenson (2010) and Merriam-Webster (2010) for English). Subsequently, the data were lemmatised, that is, all inflectional suffixes marking number and case were collapsed with their base forms (i.e. nominative- singular for Greek, singular for English).7 The end-product was a word-frequency list for -tita and one for -ness, respectively. Finally, a lexical item ought to meet two basic requirements, so as to be considered as a token with -tita or -ness. Firstly, it needed to belong to the specific morphological process both morphologically and semantically. Hence, words such as business and wilderness were excluded from further scrutiny, since they are clearly not nominalisations. Secondly, the base should be either an independent word in Modern Greek or Modern English or a bound

5 The CGT is freely available at: http://sek.edu.gr/index.php. 6 The COCA is fully searchable at: http://corpus.byu.edu/coca/. 7 Greek nouns inflect for number (singular – plural) and case (nominative – genitive – accusative).

Syntax - Morphology - Semantics 4 2 3 Major Trends in Theoretical and Applied Linguistics

stem contained in at least one other formation. For example, the item posótita ‘quantity’ was included in the Greek data, as its stem occurs in other formations such as posó(n) ‘amount’, pósos ‘how (much)’, etc.

3. Productivity Measures

The corpus-based productivity measures used in this study originate in the work of Baayen and his colleagues (see, among others, Baayen 1992, 1993, 2008, Baayen & Lieber 1991, Baayen & Renouf 1996, Plag et al. 1999). In particular, three basic statistical measures were used. (i) The total number of tokens with a given affix (Naff), i.e. how many times an affix occurs in a corpus. (ii) The number of types with an affix (Vaff), which shows the extent of use of a morphological category. (iii) The number of hapax legomena (or hapaxes for short) with that aff same affix (n1 ), i.e. the words that occur only once in the corpus. This number indicates how often a morphological process is employed to form new words, that is, words not previously evidenced in a corpus. Hapaxes are intimately tied to the corpus in which they appear and should not be thought of as necessarily true neologisms. However, there is compelling empirical evidence (Baayen & Lieber 1991, Baayen & Renouf 1996, Plag 1999, 2003, 2005, 2006) that a considerable number of neologisms do appear among the hapax legomena. But what is the rationale behind regarding hapaxes as a reliable indication of the productivity degree of an affix? Crucially, the importance of hapaxes is that if language users wish to express a new concept, then they normally resort to productive word-formation rules (WFRs), for it is only a productive process that can give rise to new lexemes (Lüdeling & Evert 2005: 361). Furthermore, it has been empirically found that a productive process is characterised by a large number of low-frequency lexemes (and many hapax and dis legomena) as well as a smaller amount of high-frequency words. Conversely, an unproductive process is characterised by very few low-frequency words (and even fewer hapax and dis legomena) and simultaneously a multitude of high- frequency items (Baayen & Lieber 1991, Lüdeling & Evert 2005). Now, the principal derived measure of productivity is the so-called ‘productivity in the narrow sense’, namely the quotient of the number of hapaxes

n1 and the number of all tokens N:

aff aff (1) P = n1 / N

Crucially, P is a probabilistic tool, which shows us the possibility of coming across new types with a certain affix, when we have sampled N tokens. A high value of P suggests a very productive affix, while a low value of P indicates the opposite (Baayen & Lieber 1991: 809-811, Plag 2003: 57). A great number of

4 2 4 Syntax - Morphology - Semantics Theofanis Papoutsis

hapaxes or/and the absence of many highly frequent words result in an increase of P. Conversely, the value of P is low, when there is a small number of hapaxes or/and a large number of high-frequency lexemes (Baayen & Lieber 1991: 815,

Plag 2003: 57). These interrelations between P, n1 and N perfectly match our intuitions about productivity, given that less productive WFRs are characterised by high-frequency words (Baayen & Renouf 1996, Plag 1999: 93ff.). However, there are cases where P appears to give affix rankings which fail to match our intuitions. To cater for such problematic rankings, another derived measure has been proposed; the so-called ‘global productivity’ (P*). The value of P* is prima facie the result of a complex mathematical equation (see Baayen 1993: 192-193), but Plag (1999: 32) demonstrates that P* actually equals the already familiar number of hapaxes:

aff (2) P* = n1

Note that P and P* stand in a complementary relation; P is rather better applicable to discriminating productive from unproductive processes, whilst P* is perhaps more appropriate for ranking productive processes (Baayen 1993: 194).

4. Results

4.1. Overall Productivity

First, let us explore the overall productivity of -tita and -ness, that is without taking into consideration the word-category of the base.

Table 1. Productivity measures of –tita and –ness

Suffix N V n1 n2 P -tita 122,788 1,208 343 151 0.0027934

-ness 18,935 1,345 585 199 0.0308951

As shown in Table 1, -ness surpasses -tita in almost every measure. In particular,

the English suffix has a greater extent of useV , more hapax and dis legomena (n1,

n2), and a higher value of P. The only index in which the Greek suffix exceeds the English one is the total number of tokens N. These figures lead us to conclude that -tita is less productive than -ness. However, this certainly does not mean that

Syntax - Morphology - Semantics 4 2 5 Major Trends in Theoretical and Applied Linguistics

-tita is not a productive suffix for the Greek language. In other words, just like -ness is very productive with regard to the English derivational system, so is -tita in relation to the Greek one. This claim can be substantiated by examining the word-frequency distribution of both suffixes, which is presented in Figure 1. Recall that a great number of low-frequency types are typically indicative of productive WFRs. Indeed, the frequency distribution of both suffixes concerned is highly skewed to the left. Thus, we observe that the largest portion of the types in -tita and -ness have frequency 1-20 (i.e. 71.52% and 89.66%, respectively).

0 < r ≤ 2 1000 2 < r ≤ 6

800 6 < r ≤ 10

10 < r ≤ 20

600 20 < r ≤ 50

50 < r ≤ 100 400 100 < r ≤ 500

Number Number of types (V) 0 < r ≤ 2

1000

200 r > 500 2 < r ≤ 6

800

6 < r ≤ 10 0

-tita -ness 10 < r ≤ 20

600 Frequency intervals (r) 20 < r ≤ 50

50 < r ≤ 100 400 Figure 1. Summary of the frequency distribution of -tita and -ness 100 < r ≤ 500 Number Number of types (V)

200 r > 500 Moreover, focusing on the head of the frequency distribution, especially the

first 15 ranks0 (see Figures 2&3), we see that the 585 hapaxes in -ness account -tita -ness for 43.4% of all different Frequencytypes V intervals, the (r) 199 types that occur twice represent

14.8% of 400 V, etc. Therefore, roughly one in two types in -ness occurs only once

in a 30-million-word corpus; a ratio which strongly indicates high productivity. Similarly, the300 343 hapaxes in -tita account for 28% of all different types V, the

151 types200 that occur twice represent 12.5% of V, etc. Simply put, if one in three

types in -tita occurs only once in a 30-million-word corpus, then this WFR has to Number Number of types (V) be considered100 productive, too.

0 400

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

300 Frequency rank (r)

200

Number Number of types (V)

100

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Frequency rank (r)

Figure 2. The head of the frequency distribution of -tita

Syntax - Morphology - Semantics 4 2 6

Theofanis Papoutsis

800

600

400

Number Number of types (V) 200

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Frequency rank (r)

Figure 3. The head of the frequency distribution of –ness

4.2. Productivity Ranking in Relation to the Word- Category220 of the Base r = 1 200 r = 2 180 Now, let us examine the productivity of -tita and -ness, considering2 < r the≤ 10 word-

160 category of the base. 10 < r ≤ 100 140 r > 100

Table 2.120

Productivity100 measures of -tita in relation to the word-category of the base Number of types (V) of types Number 80 Word-category of the base N V n n P 60 1 2 Adjectival base 110,033 1,181 333 151 0.0030263 40 Nominal base 2,622 17 6 0 0.0022883 20

Pronominal0 base 10,134 10 4 0 0.0003947 -rόs/-arόs/- -inόs -simos/-imos -ikόs/-akόs -tόs -os/-ίs -ios -eos erόs/-irόs Frequency intervals (r) As can be seen in Table 2, when combined to an adjectival base, -tita ([[XADJ]-

tita]N) exhibits by far the highest productivity indexes, that is P=0.0030263, 333 hapaxes, and 151 types occurring twice. The medial position in

productivity ranking is occupied by the combination of nominal base and -tita

([[XN]-tita]N), since P=0.0022883 and n1=6 (e.g. laótita ‘people-ness’ (< laós ‘people’), usiótita ‘essence-ness’ (< usía ‘essence’)). Note that the eligibility

of both adjectives and nouns as bases for -tita has already been mentioned in the relevant literature (see, among others, Ralli 2005, Xydopoulos 2008). Interestingly, according to the data, pronouns can also function as possible

bases. Although [[X ]-tita] ranks third and last, with P=0.0003947 and only PRO N four hapaxes (e.g. eaftótita ‘self-ness’ (< eaftós ‘self’), eδikótita ‘mine-ness’ (< e δikós (mu) ‘mine’)), yet it can be deemed as characteristic of the derivational power that -tita exhibits in present-day Greek.

Syntax - Morphology - Semantics 4 2 7 Major Trends in Theoretical and Applied Linguistics

As regards -ness, it has long been observed that it can be attached to virtually any adjective, while it is often tacked onto nouns, pronouns and phrases. Because of its great combinability, -ness is currently considered as the most productive suffix of the English language (Plag 2003: 92). Note that, on the basis of the COCA data, adverbials can also constitute eligible bases.

Table 3. Productivity measures of -ness in relation to the word-category of the base

Word-category of the base N V n1 n2 P Phrasal base 21 12 10 1 0.4761904 Adjectival base 18,749 1,288 544 195 0.4226884 Nominal base 67 31 23 2 0.3432835 Adverbial base 29 7 5 0 0.1724137 Pronominal base 69 7 3 1 0.0434782

In Table 3, we notice that the combination of phrase and -ness scores extremely high in terms of P, which is against our intuition that it is more likely to form a new word in -ness using an adjective rather than a phrase. This incongruity between intuition and P perhaps lies in the exceptionally small number of types N, which is known for positively influencing the value of P. However, if solely the number of hapaxes is taken into account, then the emerging productivity ranking of the attested combinations is in better line with our intuitive feeling.

Thus, the combination [[XADJ]-ness]N is characterized by the highest productivity degrees, having a plethora of hapax (e.g. deadness, pinkness) and

dis legomena (e.g. keenness, waterproofness). The combination [[XN]-ness]N ranks second, having a significant number of words that occur only once (e.g.boyness , brunetteness). Interestingly, besides common nouns, proper names can also function as possible bases, be they person names (e.g. Baldwin-ness, Gaga- ness) or place names (e.g. Hong-Kong-ness, Ticuaniness). The third position is

occupied by the combination [[XPHRASE]-ness]N, having ten hapax legomena (e.g. no-matter-what-ness, down-to-earth-ness). The last two positions are taken by

the combinations [[XADV]-ness]N and [XPRO]-ness]N, with five (e.g. apart-ness, now- ness) and three hapaxes (e.g. me-ness, mine-ness), respectively.

4.3. Productivity Ranking in Relation to Adjectival Bases

This section focuses on the productivity of the attested combinations of adjectival bases and -tita or -ness, which constitute by far the greatest portion of the data. Table 4 presents the attested combinations of adjectival suffixes and -tita in decreasing order of P, while Figure 4 summarizes their frequency distribution.

4 2 8 Syntax - Morphology - Semantics Theofanis Papoutsis

Table 4. Productivity measures of -tita in relation to adjectival bases

Adjectival base N V n1 n2 P -alós/-ilós/-olós 106 8 4 0 0.0377358

-simos/-imos 4,110 95 34 7 0.0082725

-ikós/-akós 27,964 657 200 97 0.007152

-arós/-erós/-irós/-rós 1,837 30 11 4 0.005988

800

-inós 967 9 2 0 0.0020682

-tós 11,374 75 17 13 0.0014946 600

-os/-ís 43,108 256 57 24 0.0013222

-íos 400 5,376 24 5 1 0.00093

Number Number of types (V)

-éos 200 4,315 14 2 3 0.0004634

-tírios 9,189 3 1 0 0.0001088 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 -aléos 21 2 0 0 0.0000

-plós 344Frequency rank4 (r) 0 2 0.0000

-teros 1,316 3 0 0 0.0000

-opós 6 1 0 0 0.0000 220

r = 1 200 r = 2 180 2 < r ≤ 10

160 10 < r ≤ 100

140 r > 100 120 100 Number of types (V) of types Number 80 60 40 20 0 -rόs/-arόs/- -inόs -simos/-imos -ikόs/-akόs -tόs -os/-ίs -ios -eos erόs/-irόs Frequency intervals (r)

Figure 4. Summary of the frequency distribution of -tita in relation to adjectival bases

Syntax - Morphology - Semantics 4 2 9

Major Trends in Theoretical and Applied Linguistics

First of all, notice that -tita in combination with adjectives in -alós/-ilós/-olós has the highest value for P in Table 4, which is in sharp conflict with our intuition, since the suffix -alós/-ilós/-olós is no longer used to form new words. Therefore, it would

be better to take only n1 and V into account; hence, having only four hapaxes (e.g. xamilótita ‘lowness’) and very few types (e.g. dropalótita ‘shyness’), -alótita/-ilótita/- olótita ranks fairly low. Secondly, -arós/-erós/-irós/-rós and -inós in combination with -tita score surprisingly high in terms of P. This does not correspond to our intuitive feeling, considering that the particular adjective-forming suffixes are themselves rather

moderately productive. So, if we use n1 as a ranking criterion, these combinations score very low in productivity, a fact which is in better line with our intuition. Their extremely low productivity degree is also evidenced in the occurrence of a considerable number of high-frequency words (e.g. nosirótita ‘morbidity’ (63), triferótita ‘tenderness’ (204), staθerótita ‘stability’ (1,419), kaθimerinótita ‘dailyness’ (837)).8 Therefore, the combinations of adjectives in -simos/-imos and -ikós/-akós with -tita occupy the top position in productivity ranking. -simótita/-imótita is more productive in the narrow sense (since P=0.0082725), but -ikótita/-akótita has more hapax (e.g. aliθiakótita ‘truthfulness’) and dis legomena (e.g. δiavrotikótita ‘corrosiveness’) than the former (e.g. xrisimopiisimótita ‘usability’ (1), ekθesimótita ‘showability’ (2)). The high productivity degree of these co-occurrences is also evidenced in the presence of a large number of low-frequency items (e.g. ereθisimótita ‘irritability’ (3), aγrotikótita ‘agriculturalness’ (4)). Note that this ranking seems to be related to our intuitive feeling that -simos/-imos and -ikós/- akós are themselves highly productive adjective-deriving suffixes. The medial positions in our productivity ranking are occupied by the combination of -tita with deverbal adjectives in -tós as well as the combination of -tita with non-suffixally derived adjectives which bear the inflectional endingsos - and -ís.9 Among the items occurring once and twice, we find: anaklitótita ‘reversibility’ (1), prosforótita ‘suitability’ (1), amvlίtita ‘bluntness’ (1), γraptótita ‘writtenness’ (2), plusiótita ‘richness’ (2), platίtita ‘wideness’ (2). Note also the occurrence of a sizeable number of low-frequency formations (e.g. ektelestótita ‘enforceability’ (3), proorótita ‘prematureness’ (4)), which points to a non-negligible degree of productivity. Turning to less productive combinations, -iótita scores extremely low in all productivity measures and shows many words at the high frequencies (e.g.

8 The number in parentheses following a word indicates the token frequency of the particular word in the corpora used. 9 The inflectional suffix-ís (spelt <ύς> in Greek) mentioned here is the ending that appears in [+learned] adjectives such as amvlίs ‘blunt’. This ending should not be confused with the inflectional suffix -ís (spelt <ής> in Greek) which was mentioned in section 1 and which appears in examples such as evγenίs ‘polite’. In other words, adjectives with these endings belong to separate paradigms (see Ralli 2005 for details).

4 3 0 Syntax - Morphology - Semantics Theofanis Papoutsis

eoniótita ‘eternity’ (207)). Further, the co-occurrence of adjectives in -éos and -tírios with -tita is of negligible productivity. After all, only fourteen types in -eótita (e.g. anangeótita ‘necessity’ (927)) and three in -tiriótita (e.g. iperδrastiriótita ‘hyperactivity’ (14)) are attested in the entire CGT. Finally, -tita is completely unproductive in combination with adjectives in -aléos, -plós, -teros and -opós. Moreover, the vast majority of the few types that are attested in the CGT are of high frequency (e.g. polaplótita ‘multiplicity’ (154), iδieterótita ‘peculiarity’ (1,062)). Note that, on an intuitive basis, these adjective- forming suffixes seem to be rarely used to coin any new words in Greek. As for the English suffix, Table 5 lists all the attested combinations of –ness and adjectival base in order of decreasing P, whilst Figure 5 summarises their frequency distribution.

Table 5. Productivity measures of -ness in relation to adjectival bases

Adjectival base N V n1 n2 P -ern 1 1 1 0 1.00 -ose 1 1 1 0 1.00 -(i)an/-ean 18 10 6 3 0.3333333 -able/-ible 33 14 11 0 0.3333333 -ate 13 6 3 1 0.2307692 -al 25 9 5 1 0.2 -some 18 8 3 3 0.1666666 -ary 25 8 4 1 0.16 -ed 507 133 76 19 0.1499013 -ic/-ical 22 6 2 1 0.1363636 -ish 183 32 16 5 0.0874316 -y 1,393 229 113 46 0.0811198 -ese 13 2 1 0 0.076923 -ous 1,057 108 60 20 0.0567644 -ly 397 41 22 5 0.0554156 -ful 495 61 21 8 0.0424242 -less 675 70 21 12 0.0311111 -ive 1422 99 40 12 0.0281293 no suffix 11,924 433 132 56 0.0110701 -ing 475 9 5 0 0.0105263 -ant 18 2 0 0 0.0000000 -esque 5 2 0 1 0.0000000 -ward 29 4 0 1 0.0000000

Syntax - Morphology - Semantics 4 3 1 Major Trends in Theoretical and Applied Linguistics

r = 1

140 r = 2

2 < r ≤ 10 120 10 < r ≤ 100 r > 100 100

80 60 Number of types (V) of types Number 40 20 0 -(i)an/-e -able/-ibl -ed -ish -y -ous -ly -ful -less -ive simplex -ing an e Frequency intervals (r)

Figure 5. Summary of the frequency distribution of –ness in relation to adjectival bases

To begin with, three observations should be made. Firstly, if for an affix P=1,

then that affix is practically dead. Thus, the combinations -ernness and -oseness are totally unproductive, since each has only one type with frequency one

(southernness , comatoseness). Secondly, -ateness and -eseness score unexpectedly high in terms of P (surpassing even unambiguously more productive combinations

such as -liness, -fulness, -lessness), considering that the Latinate suffixes -ate and -ese are not – on an intuitive basis – currently employed to form new adjectives. However, using n as a ranking criterion, both -ateness and -eseness occupy the 1 lowest positions, having three and one hapaxes (e.g. passionateness, Japaneseness), respectively. Thus, this ranking is in better line with our intuition. Thirdly, -al, -some,

-ary , -ic/-ical in combination with -ness appear to be highly productive in terms of P, even exceeding combinations such as -ishness, -iness. However, if we assume

n1 as our sole ranking criterion, -alness, -ariness, -someness and -ic/-icalness are ranked fairly low, having five, four, three and two hapaxes (e.g. spectralness, fragmentariness, bothersomeness, periodicalness), respectively. Thus, -(i)anness/-eanness and -ableness/-ibleness are by far the most productive combinations in terms of P, with the latter having more hapaxes (e.g. disagreeableness) than the former (e.g. Italianness, Victorianness). Crucially, both combinations do not show any high-frequency items. Next, the combination

-edness is highly productive, since P=0.1499013, n1=76 (e.g. designedness,

unpeopledness) and n2=19 (e.g. open-handedness, pointedness), and sinceit shows a multitude of types at the low frequencies (e.g. small-mindedness (4), woundedness (4)). -ishness and -iness are also significantly productive combinations. Although -ishness is more productive in the narrow sense, -iness is globally more productive, as it has more types appearing once (e.g. hellishness,

4 3 2 Syntax - Morphology - Semantics Theofanis Papoutsis

pepperiness) or twice (e.g. childishness, legginess) and shows more types at the low frequencies (e.g. devilishness (3), achiness (3)). This ranking possibly reflects the intuition that the native suffixes-ed , -ish, -y are often used to coin new words in English. The combinations -ousness, -liness, -fulness, -lessness occupy the middle positions of the ranking. Interestingly, -ness is rather more productively suffixed to words in the Latinate -ous (e.g. miraculousness (1), perilousness (1)) rather than words in the native -ly (e.g. dailiness (1), ungodliness (1)). Moreover, -ness combines with the antonymic suffixes -ful and -less equally productively, as indicated by the identical number of hapaxes (e.g. colorfulness, skillfulness, emotionlessness, worklessness) and the similar frequency distribution (e.g. cheerfulness (9), weightlessness (9)). The combination of -ness with adjectives in -ive (e.g associativeness (1), restiveness (2)) and -ing (e.g. amazingness (1), vivifyingness (1)), as well as with non-suffixally derived adjectives (e.g. goldness (1), spareness (2)) are marginally productive. Interestingly, -iveness shows a sizeable number of types at the low-frequencies (e.g. obsessiveness (4)), whereas no-suffix-ness has many high- frequency types (e.g. darkness (872)). Finally, the co-occurrence of -ness with the adjectival suffixes-ant , -esque and

-ward is totally unproductive, in terms of both P and n1 (e.g. unpleasantness (12), picturesqueness (3), backwardness (17)). This appears to be related to our intuitive feeling that -ant, -esque and -ward do not give rise to any new formations in current English.

5. Conclusions

This final section discusses the main conclusions that can be drawn, concerning the suffixes examined and the application of the quantitative approach to productivity to Greek data. We saw that -tita is a highly productive suffix for the Greek derivational system. Similarly, -ness is an extremely productive suffix, as regards the English derivational system. However, on an interlingual comparison, the latter surpasses the former in productivity degree. How can this discrepancy be possibly interpreted? A mathematical explanation is that the appearance of a great number of high-frequency words in -tita leads to a low value for P. Conversely, the huge number of hapaxes in -ness results in a high value for P. A linguistic, albeit rather technical, explanation might be sought at the different textual make-up of the corpora used. The CGT and the COCA obviously differ not only in the contained texts, but also in the proportions of the contained texts them. Consider as an illustration the case of academic texts; the CGT contains both PhD theses and journal articles (14.67% of total corpus

Syntax - Morphology - Semantics 4 3 3 Major Trends in Theoretical and Applied Linguistics

size), whereas the COCA contains only PhD theses (20% of total corpus size) (cf. Goutsos 2010, Davies 2011). Differences such as these might possibly affect the results obtained by the application of the productivity measures.10 A related linguistic hypothesis is that the use of nominalizations might be more preferable in Greek than in English.11 This preference may lead to the appearance of very frequent words, which in turn accounts for the low value for P. In any case, all of the above explanations are open to further discussion. Another conclusion is that both -tita and -ness do not attach to adjectival bases only, but also to other categories of bases. Perhaps as an outcome of their high degree of productivity, the Greek suffix can combine with nouns and pronouns, while its English counterpart is affixed to nouns, pronouns, adverbs and phrases. A further conclusion is that -tita and -ness are not suffixed equally productively to each and every adjective. In other words, some combinations of adjectival bases and the two noun-forming suffixes under study are extremely productive, some others are less productive and still others are completely unproductive. It might be argued then that the productivity degree of the combinations of -tita and -ness with any adjective is related to the productivity degree of the adjective-forming suffix itself. Recall, for instance, that -simótita/- imótita and -ikótita/-akótita are characterized by a high degree of productivity, whilst, according to our intuition, -simos/-imos and -ikós/-akós themselves are very productive as well. Similarly, the complete unproductiveness of -antness, -esqueness, -wardness appears to reflect the fact that the suffixes -ant, -esque, -ward are rarely used for adjective-formation in Present-Day English. In any case, this intuitive observation needs to be further scrutinized, in order to be substantiated by more empirical evidence. As regards the reliability of the productivity measures used, it can be maintained that the index of ‘productivity in the narrow sense’ (P) can – to a lesser or greater extent – capture our intuitions on the productivity of -tita and -ness. Furthermore, in the few cases where P is in disagreement with our intuition,

‘global productivity’ (P*=n1) proves to be a more reliable measure. Additionally, it should be highlighted that the results obtained from applying the productivity measures to language data are indissolubly related to the corpora used (cf. Lűdeling & Evert 2005: 356, 361). In other words, if data are drawn from certain other corpora, there might be slightly variant productivity rankings. However, neither -tita nor -ness are anticipated to prove to be unproductive suffixes.

10 These problems might be eliminated if there were freely available parallel corpora for the Greek-English pair. 11 Thanks to Dr Dionysis Goutsos (p.c.) for bringing this possibility to my attention.

4 3 4 Syntax - Morphology - Semantics Theofanis Papoutsis

A practical issue needs to be raised. The CGT is currently the most balanced and representative corpus available for Greek; yet, it is neither lemmatised nor part-of-speech-tagged. This fact, along with the unambiguously rich inflectional system of Greek, poses certain difficulties, when the quantitative approach is to be applied to the category of verbs, namely the category which has by far the richest inflection in Greek. More specifically, Greek verb forms inflect for person (1st – 2nd – 3rd), number (singular – plural), tense (+past), aspect (+perfective), mood (+imperative) and voice (active – passive). Therefore, if one wished to explore the productivity of roughly synonymous verb-forming suffixes such as -áro (e.g. sutáro ‘to shoot’) and -ízo (e.g. kapnízo ‘to smoke’),12 s/he would have to collect all the relevant tokens by hand, taking into consideration the morphological manifestation of the aforementioned grammatical categories. Clearly, such an endeavour would be extremely time-consuming (if not daunting!). Thus, the availability of lemmatized and morphologically annotated corpora is deemed vital to the study of the productivity of WFRs in languages with so-called rich inflectional morphology. Despite these practical difficulties, though, one of the most important advantages of the quantitative approach to morphological productivity is the fact that if another researcher uses the very same set of data and follows the very same methodology, then s/he is bound to come up with the same results.

12 As it is often the case with the productivity of synonymous suffixes, selection requirements come into play; thus, -áro is typically tacked onto loan bases (cf. Mela- Athanasopoulou 2002: 443), -ízo attaches to native ones.

Syntax - Morphology - Semantics 4 3 5 Major Trends in Theoretical and Applied Linguistics

References

Baayen, H.R. 1992. Quantitative aspects of morphological productivity. In G. Booij and J. van Marle (eds.), Yearbook of Morphology 1991. Dordrecht: Kluwer, 109-149.

Baayen, H.R. 1993. On frequency, transparency and productivity. In G. Booij and J. van Marle (eds.), Yearbook of Morphology 1992. Dordrecht: Kluwer, 181-208.

Baayen, H.R. 2008. Corpus linguistics in morphology: morphological productivity. In A. Lűdeling and M. Kytö (eds.), Corpus Linguistics: An International Handbook, vol. 2. Berlin: Mouton De Gruyter, 899-919.

Baayen, H.R. and R. Lieber. 1991. Productivity and English derivation: a corpus- based study. Linguistics 29: 801-844.

Baayen, H.R. and A. Renouf. 1996. Chronicling the Times: productive lexical innovations in an English newspaper. Language 72(1): 69-96.

Babiniotis, G. 2002. Λεξικό της νέας ελληνικής γλώσσας. Athens: Centre for Lexicology.

Davies, M. 2010. corpus.byu.edu: five announcements/new resources. Retrieved 8 September 2010 from [email protected]

Davies, M. 2011. The Corpus of Contemporary American English as the first reliable monitor corpus of English. Literary and Linguistic Computing 25: 447-65.

Goutsos, D. 2010. The Corpus of Greek Texts: a reference corpus for Modern Greek. Corpora 5(1): 29-44.

Institute for Modern Greek Studies. 1998. Λεξικό της κοινής νεοελληνικής. Thessaloniki: Institute for Modern Greek Studies.

4 3 6 Syntax - Morphology - Semantics Theofanis Papoutsis

Lüdeling, A. and S. Evert. 2005. The emergence of non-medical –itis: corpus evidence and qualitative analysis. In S. Kepser and M. Reis (eds.), Linguistic Evidence: Empirical, Theoretical, and Computational Perspectives. Berlin: Mouton de Gruyter, 315-333.

Mela-Athanasopoulou, E. 2002. English loan words in Modern Greek. Studies in Greek Linguistics 22: 440-450.

Merriam-Webster. 2010. Merriam-Webster Online Dictionary. http://www. merriam-webster.com

Papoutsis, T. 2012. On the combinability and productivity of the Greek noun- forming suffixes -tita and -sίni: evidence from corpora. In Proceedings of the 6th Athens Postgraduate Conference of the Faculty of Philology, University of Athens, vol. A. Athens: University of Athens, 267-277.

Plag, I. 1999. Morphological Productivity: Structural Constraints on English Derivation. Berlin & New York: Mouton de Gruyter.

Plag, I. 2003. Word-Formation in English. Cambridge: Cambridge University Press.

Plag, I. 2005. Productivity. In K. Brown (ed.), Encyclopedia of Language and Linguistics, vol. 10. Boston: Elsevier, 121-128.

Plag, I. 2006. Productivity. In B. Arts and A. McMahon (eds.), The Handbook of English Linguistics. Oxford: Blackwell, 537-556.

Plag, I., C. Dalton-Puffer and H.R. Baayen. 1999. Morphological productivity across speech and writing. English Language and Linguistics 3(2): 209-228.

Ralli, A. 2005. Μορφολογία. Athens: Patakis. .

Stevenson, A. (ed.). 2010. Oxford Dictionary of English. Oxford: Oxford University Press.

Xydopoulos, G.I. 2008. Λεξικολογία: Εισαγωγή στην ανάλυση της λέξης και του λεξικού. Athens: Patakis.

Syntax - Morphology - Semantics 4 3 7 Major Trends in Theoretical and Applied Linguistics

4 3 8 Result Clauses in Modern Greek and Spanish: A Contrastive Study1

Antonio R. Revuelta Puigdollers

Universidad Autónoma de Madrid [email protected]

Abstract

This paper examines the behaviour of result subordinators in Modern Greek and Spanish. Apart from the more frequently studied subordinators (που and ώστε) the research includes others, such as με αποτέλεσμα να and οπότε, which have received little or no attention until now. The article discusses how the choice of subordinator affects the selection of mood (indicative or subjunctive), the scope of negation and interrogation, the presence or absence of correlative intensifiers in the main clause, and the clause layer the subordinate clause can operate upon (SoA, proposition, speech act).

1. Introduction

Most grammars on Modern Greek and Spanish concentrate on a limited range of result subordinators.2 The purpose of this paper is to extend the number of subordinators and to study their differences in behaviour. Apart fromπου and ώστε (the most frequently studied subordinators), Modern Greek has other connectors that are used to express consequence, while Spanish exhibits the subordinator por lo que/cual, (which has not received much attention either)

1 For the grant I enjoyed during 2009 I would like to express my gratitude to the Alexander S. Onassis Public Benefit Foundation. This paper has been written within the frame of the research project “Funciones y marcas del griego moderno” (HUM2007-61974) financed by the Spanish Ministry of Science and Innovation. I want to express my gratitude to H. Ferguson for making my English more understandable. 2 For result subordinators in Spanish see Álvarez (1995, 1999); for the Greek data see Holton et alii (1997: 465), Cleris, Ch. & Babiniotis, G. et alii (1998-2001 II 2: 107-110), Mackridge (1985: 256, 272-273), Tzartzanos (1963 B: 94-96) and AKN.

Syntax - Morphology - Semantics 4 3 9 Major Trends in Theoretical and Applied Linguistics

in addition to que and de manera/modo que (the most frequently studied), as examples 1 and 2 show:3

(1) Ήταν παρατηρητικός και διεπίστωσε ότι όλα είχαν τον ίδιο αριθμό, οπότε αντελήφθη ότι είναι πλαστά (HNC 909808) ‘Era observador y constató que todos llevaban el mismo número, por lo que se dio cuenta de que eran falsos’ ‘He was a good observer and noticed that all of them had the same number, and therefore he realized that they were false’

(2) Το 1895 η μητέρα της πέθανε ξαφνικά, με αποτέλεσμα η Βιρτζίνια Γουλφ να πάθει τον πρώτο νευρικό κλονισμό. (ΤΟ ΒΗΜΑ: B13785C101 9/2/2003: 61-62) ‘En 1895 su madre murió de repente y a consecuencia de ello Virginia Woolf sufrió su primera crisis nerviosa’ ‘In 1985 her mother suddenly died, and as a consequence Virginia Wolf suffered her first nervous breakdown’

In this paper I am going to concentrate on the behaviour of all of these result subordinators (που, ώστε, με αποτέλεσμα να, οπότε in Greek and que, de modo/ manera que, por lo cual/que in Spanish) as regards mood, negation, interrogation, the presence of correlative intensifiers and the level of the clause.4

2. Mood Selection

As for mood, result clauses seem to present three possible constructions: a) they can exhibit the moods of declarative independent sentences: indicative and conditional (potential and counterfactual); b) they can exhibit subjunctive. In the following sections I will examine the combination of the subordinators with these different moods.5

3 Most Greek examples are taken from the Hellenic National Corpus (HNC) developed by the Institute for Language and Speech Processing. The Spanish examples are my own or translations of the Greek examples. 4 The so-called result για να has been excluded from this study: οι περισσότεροι [ήταν] πολύ φανταχτεροί για να είναι ωραίοι (“most of them [were] too showy to be handsome”, Mackridge 1985, 293). This is a special purpose clause. For this use see Mackridge (1985: 293-294), Tzartzanos (1963 B: 139) and Veloudis (2010: 287). I thank the unknown reviewer for reminding me of this construction. 5 For mood in Spanish and Greek see Revuelta (2011); for the Spanish data see Pérez Saldanya (1999); for Greek see Hesse (2003). See also the references mentioned in note 1.

4 4 0 Syntax - Morphology - Semantics Antonio R. Revuelta Puigdollers

2.1. Ώστε (de manera/modo que)/ που (que) 2.1.1. Indicative and Conditional

Greek ώστε and που and Spanish de manera/modo que and que can take the moods of declarative sentences. Indicative shows that the content of the subordinate clause is an actual consequence of the main clause:

(3) Η ηγεσία της Ιαπωνίας ήταν όμως τόσο αφερέγγυα και τόσο διχασμένη, ώστε οι Αμερικανοί έτρεφαν (IND) πλέον σοβαρές αμφιβολίες για τις προθέσεις του Τόκιο. (HNC 77277) ‘The Japanese leadership was so unreliable and so split that the Americans had serious doubts about Tokyo’s intentions’ a. The Japanese leadership was very unreliable and very split (main) b. The Americans had serious doubts about Tokyo’s intentions (subordinate)

(4) Era tan malo que lo odiaban (IND) ‘He was so evil that they hated him’ a. He was evil (main) b. They hated him (subordinate)

Conditional, on the other hand, points out that the content of the subordinate clause is either a potential (see ‘POT’ in example 5) or counterfactual (see ‘COUNTER’ in example 6) consequence:

(5) Η άλλη τραγική ειρωνεία είναι ότι στα μπραντς των εστιατορίων […] πετιούνται τόσο μεγάλες ποσότητες τροφής που θα μπορούσαν (POT) να θρέψουν μεγάλο μέρος από τους άστεγους και άπορους της Ασίας. (HNC 88585) ‘La otra ironía es que en los brunch de los restaurantes […] se tiran cantidades de alimento tan grandes que podrían (POT) alimentar a una gran parte de los sin techo y necesitados de Asia’ ‘The other irony is that the amount of food that is thrown away from restaurant brunches is so huge that it could feed a large part of Asia’s homeless and needy people’

(6) ο Ιωσήφ παρέμενε τόσο ακίνητος, ώστε το σώμα του θα μπορούσε (COUNTER) να έχει φτάσει να συγχέεται με τις πέτρες. (HNC 190563) ‘Joseph se quedó tan inmóvil que su cuerpo podría (COUNTER) haber llegado a confundirse con las piedras’ ‘Joseph remained so still that his body could have been confused with the rocks’

Syntax - Morphology - Semantics 4 4 1 Major Trends in Theoretical and Applied Linguistics

2.1.2. Subjunctive

Subjunctive is selected as mood when the consequence is not a real, but a possible consequence. The speaker does not commit to the truth of the state of affairs. That is to say, the sentence expresses uncertainty (the consequence may or may not take place). Spanish usually replaces the subordinators with como para (como para plus infinitive or como para que plus subjunctive). When the subjects of both main and subordinate clause are co-referential the subjunctive is replaced by the infinitive:

(7) Ήταν τόσο αφελής, ώστε να τον παρασύρουν κατώτεροί του, τους οποίους εμπιστευόταν. (HNC 1702458) ‘Era tan ingenuo como para que le engañaran (SUBJ) sus subordinados, en los que confiaba’ ‘He was so naive that his subordinates, whom he trusted, could fool him’ a. He was (main) b. Perhaps his subordinates fooled him (subordinate).

2.2. Με αποτέλεσμα να

The Greek construction με αποτέλεσμα να takes only the subjunctive, but unlike previous cases the construction always introduces factive consequences: the result is taken for granted in all circumstances, and therefore the Spanish translation in these cases demands the indicative. There is no exact equivalent in Spanish, and usually the best translation is to coordinate both clauses and add a consecuencia de ello (‘as a consequence’) as a connector to explicitly state the relationship between both states of affairs:

(8) Το 1895 η μητέρα της πέθανε ξαφνικά, με αποτέλεσμα η Βιρτζίνια Γουλφ να πάθει (SUBJ) τον πρώτο νευρικό κλονισμό. (ΤΟ ΒΗΜΑ: B13785C101 9/2/2003: 61-62) ‘En 1895 su madre murió de repente y a consecuencia de ello Virginia Woolf sufrió (IND) su primera crisis nerviosa’ ‘In 1985 her mother suddenly died, and as a consequence Virginia Wolf suffered her first nervous breakdown.’ a. Her mother suddenly died (main) b. Virginia Wolf suffered her first nervous breakdown (subordinate)

This “anomalous” behaviour of the subjunctive in Greek remains to be explained. Nevertheless, the mere fact that there is no possibility of using the indicative clarifies that the subjunctive mood can have a factive interpretation.

4 4 2 Syntax - Morphology - Semantics Antonio R. Revuelta Puigdollers

The use of subjunctive in factive subordinates is common in Spanish in other contexts (for example in complement sentences), but not in Modern Greek.6

2.3. Οπότε / por lo que/por lo cual

The subordinators οπότε and por lo que/por lo cual can take the moods of main declarative sentences (indicative and conditional), but subjunctive is excluded:

(9) Ήταν παρατηρητικός και διεπίστωσε ότι όλα είχαν τον ίδιο αριθμό, οπότε αντελήφθη (IND) ότι είναι πλαστά (HNC 909808) ‘Era observador y constató que todos llevaban el mismo número, por lo que se dio (IND) cuenta de que eran falsos’ ‘He was a good observer and noticed that all of them had the same number, and therefore he realized that they were false’ a. He was a good observer and noticed that all of them had the same number (main) b. He realized that they were false (subordinate)

(10) Αυτό σήμαινε ότι, όταν ο χρόνος θα άλλαζε από 1999 σε 2000, πολλοί υπολογιστές θα αποθήκευαν τα ψηφία 00, οπότε θα μετέβαιναν (COND) σε μια προγενέστερη κατά έναν αιώνα ημερομηνία. (HNC 211315) ‘Esto significa que, cuando el tiempo cambiara de 1999 a 2000, muchos ordenadores almacenarían los dígitos 00, por lo que entrarían (COND) en una fecha anterior en un siglo.’ ‘This means that, when the date would change from 1999 to 2000, many computers would record the digits 00, and therefore they would enter into a date one century before.’

2.4. Mood: General View

The following table summarizes the differences and similarities of the subordinators as far as mood is concerned.

6 See Ridruejo (1999: 3229-3233) for the use of subjunctive in Spanish factive complement clauses. In these factive contexts Modern Greek selects που plus indicative: see Christidis (1981) and Roussou (2006). For a recent discussion of the subjunctive, see the articles in Lingua 119 and a summary in Quer (2009).

Syntax - Morphology - Semantics 4 4 3 Major Trends in Theoretical and Applied Linguistics

Table 1. Mood restrictions.

Conditional (Potential/ Subjunctive Indicative Subordinator Counterfactual: Θα (Spanish Infinitive with (Real) + past) co-referential subjects)

που/que + + + (Uncertainty) Class I ώστε/de manera que + + + (Uncertainty)

Class II με αποτέλεσμα να/ø + (Factuality)

Class III οπότε/por lo cual/que + +

All subordinators allow the use of indicative and conditional (potential or counterfactual) both in Greek and Spanish, except με αποτέλεσμα να, which demands only subjunctive. The subjunctive can appear with all of them except οπότε. In Spanish the subjunctive always expresses uncertainty; this applies to Greek too, except for the subordinator με αποτέλεσμα να, which always exhibits a factual interpretation. As a consequence, in Greek it is possible to distinguish three classes of result subordinators according to their mood.

3. Scope of Negation

Negation allows us to further distinguish the behaviour of the subordinators. In the following subsections I will investigate how the negation of the main clause affects the main clause itself and its subordinate clause.7

3.1. Που / que

Under negation of the main clause the subordinators που and que mandatorily take subjunctive. In Spanish the subjunctive is replaced by the infinitive when the subjects of both main and subordinate clause are co-referential. In both languages the negation of the main clause additionally affects the content of the subordinate clause, as the paraphrases of example 11 show (see b):

7 For negation scope in Spanish and Greek, in result and other subordinate clauses see Revuelta (2011).

4 4 4 Syntax - Morphology - Semantics Antonio R. Revuelta Puigdollers

(11) Σε καμία περίπτωση το χαμόγελό του δεν εξετράπη τόσο, που να μας δώσει (SUBJ) λαβή για περαιτέρω συμπεράσματα. (HNC 268840) ‘En ningún caso su sonrisa fue tan excesiva como para darnos (INF) motivo a ulteriores conclusiones’ ‘In no case was his smile so broad as to give us a reason to come to further conclusions’ a. His smile was not broad (main: negative) b. He did not give us a reason to come to further conclusions (subordinate: negative)

Since the main clause, which is the natural cause of the result clause, does not take place, the event referred to by the subordinate clause does not take place either. The subordinate clause seems to be within the scope of the negation and is affected by it. The choice of subjunctive in the subordinate clause (in Spanish subjunctive or infinitive) reflects this fact. Apart from selecting infinitive when subjects in the subordinate and main clauses are co-referential, Spanish has the particularity that the subordinator most frequently chosen under negation is como para (como para que plus Subjunctive or como para plus Infinitive) rather than que.

3.2. Ώστε / de manera/modo que (como para)

Under negation ώστε allows the interpretation of the subordinate clause both within and out of the scope of the negation. As in the previous cases, the content of the subordinate clause can be within the scope of the negation and can be affected by it (see b in the examples below). The mood has to be subjunctive (infinitive in Spanish when the subjects are co-referential):

(12) Δεν είμαι τόσο αφελής ώστε να πιστεύω (SUBJ) ότι ένα τραγούδι θα μπορούσε να επηρεάσει τις εκλογές. (HNC 72954) ‘I am not so naive as to think that a song could influence the elections’ a. I am not so naive (main: negative) b. I do not think that a song could influence the elections (subordinate: negative)

(13) No era tan malo como para que lo odiasen (SUBJ) ‘He was not so evil as to be hated by them’ a. He was not so evil (main: negative) b. He was not hated by them (subordinate: negative)

Syntax - Morphology - Semantics 4 4 5 Major Trends in Theoretical and Applied Linguistics

When the subordinate clause is negative, the negation is cancelled by the main clause’s negation, since two negations are equivalent to a positive assertion, as in the example below:

(14) Κανένα μαρτύριο δε θεωρείται τόσο βάρβαρο, καμία τιμωρία τόσο σκληρή ώστε να μην επιβάλλεται (SUBJ) σε γυναίκες. (HNC 1120558) ‘Ningún tortura es considerada tan bárbara, ningún castigo tan duro como para que no se aplique (SUBJ) a las mujeres’ ‘No torture is considered so savage, no punishment so hard that it cannot be applied to women’ a. No torture is considered so savage, no punishment so hard (main: negative) b. All tortures (= no no torture), all punishments can be applied to women (subordinate: positive)

As with the previous subordinator (see που), apart from selecting infinitive when subjects in the subordinate and main clauses are co-referential, Spanish has the particularity that the subordinator most frequently used is como para (como para que plus Subjunctive / como para plus Infinitive). The subordinator de manera/modo que cannot be used in these cases. But the subordinator ώστε ―and de manera/modo que in Spanish― can have a second construction. It may fall out of the scope of the negation and refer to the positive result of a negative cause. The mood selected in this case is either indicative or conditional, but subjunctive is excluded:

(15) Η ιστορική μνήμη δεν είναι από τις αρετές της φύλης μας […] ώστε είναι (IND) κέρδος πάντα να την τονίσουμε (Mackridge 1985: 272) ‘A historical memory is not one of the virtues of our race […] so that it is always profitable to stress it’ a. A historical memory is not one of the virtues of our race (main: negative) b. It is always profitable to stress it (subordinate: positive)

(16) No era tan malo, de modo que lo amaban (IND) ‘He was not that evil, so they loved him’ a. He was not that evil (main: negative) b. They loved him (subordinate: positive)

3.3. Με αποτέλεσμα να

The subordinator με αποτέλεσμα να always falls out of the negation’s scope. It always expresses the positive result of a negative cause, but the mood is subjunctive despite its factive sense:

4 4 6 Syntax - Morphology - Semantics Antonio R. Revuelta Puigdollers

(17) Παραδόξως, η ναυτική μας ιστορία δεν έχει πλήρως ερευνηθεί, με αποτέλεσμα να υπάρχουν (SUBJ) ακόμα παρερμηνείες. (HNC 765018) ‘Paradójicamente, nuestra historia naval no ha sido investigada completamente, por lo que todavía existen (IND) interpretaciones erróneas’ ‘Paradoxically, our naval history has not been completely researched, and as a consequence there are still many wrong interpretations’ a. Our naval history has not been researched yet (main: negative) b. There are still many wrong interpretations (subordinate: positive)

Only when explicitly negated by μή(ν), does the subordinate clause express a negative consequence, as in the following example:

(18) Υπάρχουν στιγμές που το Διαδίκτυο γενικότερα έχει (παροδικά) προβλήματα, με αποτέλεσμα να μην είναι δυνατή η πρόσβαση σε συγκεκριμένους υπολογιστές. (HNC 206734) ‘Hay momentos en los que Internet en general tiene (ocasionales) problemas y como resultado de ello no es posible el acceso a determinados ordenadores’ ‘There are times when the internet has (temporary) problems, and as a consequence there is no access to certain computers’

3.4. Οπότε / por lo que/cual

The same applies to οπότε in Greek and por lo cual / por lo que in Spanish. Their content is always out of the scope of the negation and they likewise express the positive result of the negative event referred to by the main clause. Like ώστε and de manera/modo que they exclude subjunctive:

(19) δυστυχώς δεν υπήρχαν τα απαραίτητα χρήματα, οπότε έπρεπε (IND) να γίνει κάποια ιεράρχηση των προτεραιοτήτων (HNC 678695) ‘Desgraciadamente no había el dinero necesario, por lo que hubo (IND) que ordenar las prioridades’ ‘Unfortunately there was not enough money, and as a consequence it was necessary to establish priorities’ a. There was not enough money b. It was necessary to establish priorities

Only when explicitly negated, can the subordinate clause be negative, as in the following example:

(20) Γενικά, όμως, οι τσιγγάνοι στιγματίζονται, οπότε δεν τους προσεγγίζουν οι κατάλληλες υπηρεσίες. (HNC 32524)

Syntax - Morphology - Semantics 4 4 7 Major Trends in Theoretical and Applied Linguistics

‘Generalmente, sin embargo, los gitanos están estigmatizados, por lo que los servicios adecuados no se acercan a ellos’ ‘Generally, however, Gypsies are stigmatized, and for that reason appropriate agencies do not approach them’

3.5. Negation: General View

The following table gives a general view of both the scope of the main clause’s negation and the interplay of the scope of negation with mood selection.

Table 2. Negation scope and mood restrictions.

Class Subordinator Negation Mood

I που / que inside subjunctive (uncertainty) inside subjunctive (uncertainty) II ώστε / de manera/modo que outside indicative, conditional

με αποτέλεσμα να outside subjunctive (factuality) III οπότε, por lo cual/que outside indicative, conditional

When within the scope of negation, the subordinate clause takes subjunctive: the subjunctive reflects the fact that the subordinate clause is affected by the negation and its state of affairs is supposed to not take place. Only με αποτέλεσμα να behaves differently from the other subordinators, since it expresses factuality and not uncertainty, and it is out of the scope of negation. According to their behaviour under negation, the subordinators can be placed into three different classes, which differ from those established by mood selection.

4. Correlative Intensifiers τόσο( ... που/ώστε ...)

Some subordinate clauses express that the consequence is the result of the degree to which the main clause’s state of affairs takes place. This degree is conveyed by correlative intensifiers, which can modify nominal phrases, adjectives, adverbs or verbs in the main clause, and are combinable only with certain subordinators, as the table below shows: It is important to distinguish between intensifiers—such as those in the table—and complex subordinators such as de manera/modo que. Correlative intensifiers can be placed in different positions within the main clause and

4 4 8 Syntax - Morphology - Semantics Antonio R. Revuelta Puigdollers

are not rigidly attached to the subordinator. On the other hand, in many cases complex subordinators are made up of a former intensifier (e.g., de manera) and a subordinator (e.g., que) from a diachronical point of view, but synchronically they behave as indivisible units (de manera/modo que are not separable).

Table 3. Correlative intensifiers allowed by the subordinators.

Correlative intensifier Subordinator Modified constituent

NP Adj Adv Verb

έτσι / así/tan (‘so’) που/ώστε/que + + + (?)

τόσο / tan/tanto (‘so much’) που/ώστε/que + + +

τόσος, -η, -ο / tanto/s (‘so much/so many’) που/ώστε/que +

τέτοιος / tal (‘such’) που/ώστε/que +

Cada (‘such’, lit. ‘every’) que + με / κατά {τέτοιο … τρόπο | … τρόπο τέτοιο} / de tal manera que (‘in such a way that’, που/ώστε/que + ‘so that’) εις τρόπον ώστε / de tal manera que (‘in που/ώστε/que + such a way that’) στο βαθμό / hasta el punto/tal punto (‘in που/que + such a degree that’)

4.1. Use of Correlative Intensifiers

The use of intensive antecedents or correlative intensifiers is limited in Greek to the subordinate clauses headed by που and ώστε, and in Spanish to those headed by que and como para. In the case of που and que their presence is compulsory:

(21) Το έδαφος είναι τόσο άγονο που η σοδειά μετά βίας φτάνει το ένα τρίτο του κανονικού. (HNC 29791) ‘El suelo es tan improductivo que la cosecha con dificultad llega a un tercio de lo ordinario.’ ‘The soil is so unproductive that the crop hardly reaches a third of the usual.’

(22) Οι σινοαμερικανικές σχέσεις είναι αυτή τη στιγμή τόσο πολύπλοκες, ώστε οι δύο πλευρές δεν συμπίπτουν ούτε καν στην περιγραφή της επίσκεψης. (HNC 84493) ‘Las relaciones chinoamericanas son en este momento tan complejas que las dos partes no coinciden ni siquiera en la descripción de la visita.’ ‘The Chinese-American relationship is so complex right now that both sides do not even agree on the assessment of the visit.’

Syntax - Morphology - Semantics 4 4 9 Major Trends in Theoretical and Applied Linguistics

4.2. Lack of Correlative Intensifier

The absence of antecedent is optional with ώστε and compulsory with με αποτέλεσμα να and οπότε in Greek (for these last two subordinators see any of the examples in previous sections) and de manera/modo que and por lo cual/ que in Spanish:

(23) Η ιστορική μνήμη δεν είναι από τις αρετές της φύλης μας […] ώστε είναι (IND) κέρδος πάντα να την τονίσουμε (‘A historical memory is not one of the virtues of our race […] so that it is always profitable to stress it’, Mackridge 1985: 272)

(24) No era malo, de modo que lo amaban (IND) (‘He was not bad, so that they loved him’)

4.3. Correlative Intensifiers: General View

The combinability with antecedents allows us to classify the subordinators into three different categories. One subordinator που( ) demands the compulsory presence of a correlative intensifier, whereas two others με( αποτέλεσμα να and οπότε) preclude the use of such a constituent. Only ώστε allows both their presence and absence.

Table 4. Restrictions on correlative intensifiers in main sentence.

Class Subordinator Correlative intensifier I που que + + II ώστε de manera/modo que - με αποτέλεσμα να - III οπότε por lo que/cual -

5. Level of the Clause

The level of the clause and the type of entity put into relationship by the subordinators (states of affairs, propositions, speech acts) is another factor that has to be taken into account when classifying clauses.8 For example, clauses headed

8 See Dik & Hengeveld (1997) for the levels and layers of the clause.

4 5 0 Syntax - Morphology - Semantics Antonio R. Revuelta Puigdollers

by που, ώστε and με αποτέλεσμα να (but not οπότε) can establish a material cause- consequence connection between the states of affairs (henceforth SoAs) referred to by the main and subordinate clause: the SoA (a second order entity) in the main clause is a possible cause of the SoA described by the subordinate clause:

(25) Η εξώπορτα ενός σπιτιού ήταν τόσο βουτηγμένη στο αίμα, ώστε ο Wehrly δεν κατάφερε να την ανοίξει. (‘The doorway of a house was so drenched with blood that Wehrly failed to open it’, HNC 412739)

In other cases the main clause of ώστε and οπότε (but not of που nor με αποτελεσμα να) expresses not the cause, but rather the reason why a decision is taken, as in the following example:

(26) Σύμφωνα όμως με αξιωματούχους της υπηρεσίας δεν κατέστη δυνατόν να βρεθεί εναλλακτική λύση, οπότε αποφασίστηκε να μην πραγματοποιηθεί καμία αλλαγή. (‘According to some officers of the service it was impossible to find an alternative solution, and for that reason it was decided to make no change’, HNC 17088)

On the other hand, ώστε and οπότε can establish relationships between the propositional content conveyed by both sentences (propositions are third order entities).9 In the following example the first sentences are presented as the evidence from which the speaker draws the conclusion expressed by the clause headed by οπότε:

(27) Αν και κάποιοι είπαν, πως χθες βράδυ ο πρωθυπουργός ανεβαίνοντας στο βήμα της ΔΕΘ σκόνταψε στα σκαλιά και παραλίγο να έπεφτε! Οπότε, σίγουρα θα έβλεπε τα πάνω - κάτω... (‘Although some people say that yesterday when getting on the platform the Prime Minister tripped over the steps and nearly fell down! Therefore he would see everything upside down’, HNC 661819)

In a fourth level there are cases where ώστε or οπότε express that the speaker’s speech act (and not its content or SoA) is a reaction to or a consequence of something previously said by his/her interlocutor:

(28) Ώστε δεν έδωσες το δώρο στη Χρυσούλα; (‘So you did not give the present to Chrysoula?’, Holton et alii 465)

9 See Mackridge (1985: 272-273). Although he speaks of ‘logical consequence’ he refers to consequences in the SoA level that establish a reason-result consequence rather than a material cause-result consequence.

Syntax - Morphology - Semantics 4 5 1 Major Trends in Theoretical and Applied Linguistics

In these last two cases, when the relationships are established between propositions or speech acts, the connectors appear rather at the beginning of an independent clause and work rather as discourse markers and not as subordinators. This fact is sometimes reflected in the punctuation, and the connectors appear after a full stop, but this is not always the rule. The following table depicts this situation.

Table 5. Sentence entity affected by the result clause.

Class Connector Status Relationship Level material cause- που que subordinator SoA consequence I material cause- με αποτέλεσμα να subordinator SoA consequence material cause- SoA subordinator consequence que / de reason SoA II ώστε manera/modo que subordinator/ evidence Proposition discourse marker reaction Speech act

subordinator reason SoA

III οπότε por lo que/cual subordinator/ evidence Proposition discourse marker reaction Speech act

6. Interrogation

Interrogation interplays with both mood choice and level of the clause. When expressing material consequence on the SoA level, result clauses headed by ώστε, που and με αποτέλεσμα να can appear within the scope of interrogation. Those cases cover all uses of the last two subordinators (που and με αποτέλεσμα να), but not all of the first ώστε( ), since, as we have seen in previous sections, ώστε can introduce relationships other than material consequence:

(29) Μήπως διαφημίζουμε πράγματα που δεν ανταποκρίνονται στην πλήρη αλήθεια και αυτό κάποτε γίνεται αντιληπτό, με αποτέλεσμα να το πληρώνουμε στο τέλος πολύ ακριβά; (HNC 1765749) ¿Acaso no estaremos publicitando cosas que no se corresponden con la verdad y ello resulta a veces perceptible y como consecuencia de ello lo pagamos al final muy caro? ‘Aren’t we advertising things that do not correspond with the whole truth, and this fact is perceptible, and as a consequence at the end we are paying a very high price for it?’

4 5 2 Syntax - Morphology - Semantics Antonio R. Revuelta Puigdollers

The subordinator με αποτέλεσμα να always demands the presence of subjunctive. But interrogation may optionally trigger subjunctive in Greek when the subordinate clause is headed by ώστε and που; the same applies in Spanish to que and como para (que), as the following Greek examples and their translations into Spanish prove (the first example is in subjunctive and the second in indicative):10

(30) Είσαι τόσο ασυνείδητος ώστε να μη σου λένε (SUBJ) τίποτα όλα αυτά; (HNC 1189335) ‘¿Tienes tan poca conciencia como para que no te digan (SUBJ) nada todas estas cosas?’ ‘Are you so lacking in scruples that all these things mean nothing to you?’

(31) Η κατάσταση είναι τόσο κρίσιμη, κ. Χατζιδάκι, ώστε εσείς […] “στρατεύεστε» (IND) τώρα εναντίον του νεοναζισμού; (HNC 1331655) ¿Es tan grave la situación, Sr. Hatzidakis, que usted […] “milita” (IND) ahora contra el Nazismo? ‘Is the situation so serious, Mr. Hatzidakis, that you […] are now “becoming militant” against Nazism?’

In the first of the two examples (30) both the main and subordinate clauses are under interrogation and therefore there is uncertainty as to the reality of the state of affairs referred to by them. The subjunctive of the subordinate clause is a formal reflection of such an uncertainty. In the second (31) example the result is just taken for granted (indicative) and only the reason referred to by the main sentence is questioned. Unlike the above-mentioned subordinators, οπότε cannot appear within interrogative sentences to express result.11 That is so because this subordinator cannot establish material cause-consequence relationships between SoAs. On the other hand, both ώστε and οπότε can be out of the interrogative scope and introduce the question:

(32) Ώστε δεν έδωσες το δώρο στη Χρυσούλα; (‘So you did not give the present to Chrysoula?’, Holton et alii 465)

10 See Revuelta (2011). 11 Other uses of οπότε are compatible with interrogation, as for example, when it means ‘when’ and introduces an explicative relative adverbial clause: Πιστεύετε ότι η Τουρκία είναι υποχρεωμένη να αναγνωρίσει την Κύπρο πριν από την 3η Οκτωβρίου (οπότε θα αρχίσουν οι ενταξιακές διαπραγματεύσεις); (‘Do you think that Turkey is obliged to recognize Cyprus before the 3rd of October (time when the negotiations for the adhesion will take place)?’, HNC 27110).

Syntax - Morphology - Semantics 4 5 3 Major Trends in Theoretical and Applied Linguistics

(33) Ο Ρούφους Σιούελ στα 32 του είναι ένας ανερχόμενος Βρετανός αστέρας, οπότε γιατί να μη γίνει Μποντ; (‘Rufus Sewell at his 32 years is one rising British star, therefore why cannot he become Bond?’, HNC 338339)

In these cases the connectors operate not on the SoA level, but on the speech act level. Subjunctive is excluded in these contexts. The following table depicts this situation, where both the scope of interrogation and the interplay between this factor and mood choice are reflected.

Table 6. Interrogation scope and mood restrictions.

Class Subordinator Interrogation’s Mood scope indicative, conditional, που que within I subjunctive με αποτέλεσμα να within subjunctive indicative, conditional, within II ώστε subjunctive de manera/modo que outside indicative, conditional III οπότε por lo que/cual outside indicative, conditional

7. General View

If we take into account all the factors analysed in this paper, we obtain a provisional classification of the Greek data such as the one depicted in the following table.

Table 7. General view: Restrictions imposed by each subordinator.

Subordinator Mood Negation Intensifier Level Interrogation material IND, COND, SUBJ inside που + cause- inside (uncertainty) (SUBJ) consequence material cause- inside consequence, IND, COND, SUBJ (SUBJ), ώστε +/- reason- inside /outside (uncertainty) outside consequence, (IND, COND) evidence, reaction material outside με αποτέλεσμα να SUBJ (factuality) - cause- inside (SUBJ) consequence reason- outside consequence, οπότε IND, COND - outside (IND, COND) evidence, reaction

4 5 4 Syntax - Morphology - Semantics Antonio R. Revuelta Puigdollers

The table considers the behaviour of the subordinators within the different contexts and the interplay among them. As we have seen, negation correlates with a second factor, the mood: when the subordinate clause is within the scope of the main clause’s negation, subjunctive is required; when the subordinate clause is out of negation’s scope indicative or conditional are the rule. The subordinator με αποτέλεσμα να escapes this rule, because subjunctive is its only mood. The Greek data have parallel constructions in Spanish syntax. Although not completely reflected in the table, interrogation’s scope interplays with mood and sentence level: when the subordinate clause is within the scope of the interrogation, subjunctive is possible (but not when it is outside) and the relationship is established between SoAs (but not between propositions or speech acts). The presence of correlative intensifiers is restricted to the same contexts, too. Although the table just reflects the Greek data, throughout this article many of the differences between the Greek and Spanish systems have been described. Some of these differences are the following: the lack of a subordinator like με αποτέλεσμα να in Spanish; the substitution of infinitive for subjunctive when subjects in both clauses are co-referential; the need in Spanish to replace the subordinators que and de manera/modo que with como para (infinitive or que + subjunctive) under negation in most contexts when the subordinate sentence is within the negation’s scope. One of the consequences of this asymmetry is the fact that Spaniards tend to ignore the construction με αποτέλεσμα να when speaking Greek, whereas they have no difficulty learning the use ofπου and ώστε. Apart from the asymmetry of both systems the lack of information in most grammars about με αποτέλεσμα να and οπότε plays a fundamental role.12 On the other hand, Greek speakers tend to ignore the Spanish construction como para. Despite the existence of previous works (Álvarez 1995, 1999) a more complete description of the Spanish data along the lines presented here is needed. Once the conditions of use of the subordinators in both languages are fully described, it will be possible to establish the correspondences between them. The present work is just a first attempt.

12 The grammar by Cleris & Babiniotis et alii (1998-2001) is the only one that mentions this construction.

Syntax - Morphology - Semantics 4 5 5 Major Trends in Theoretical and Applied Linguistics

References

Álvarez, A.I. 1995. Las construcciones consecutivas. Madrid, Arco Libros.

Álvarez, A.I. 1999. Las construcciones consecutivas. In I. Bosque and V. Demonte (eds.), vol. 3, 3739-3804.

Bosque, I. and V. Demonte (eds.) 1999. Gramática descriptiva de la lengua española. 3 vols. Madrid, Editorial Espasa.

Christidis, A.P. 1981. ότι/πως -που: επιλογή δεικτών συμπληρωμάτων στα νέα ελληνικά. Studies in Greek Linguistics 2, 113-178.

Cleris, C. and G. Babiniotis et alii 1998-2001. Γραμματική της νέας Ελληνικής. Δομολειτουργική-επικοινωνιακή. 4 vols. Athens, Ελληνικά Γράμματα.

Dik, S.C. 1997. The Theory of Functional Grammar. 2 vols., ed. K. Hengeveld. Berlin, Gruyter.

Hesse, R. 2003. Syntax of the Modern Greek Verbal System. The Use of the Forms, Particularly in Combination with θα and να. Copenhague, Museum Tusculanum Press.

Holton, D., P. Mackridge and I. Philippaki-Warburton 1997. Greek: a comprehensive grammar of the modern language. London, Routledge.

ΛΚΝ = Λεξικό της Κοινής Νεοελληνικής 1998. Αριστοτέλειο Πανεπιστήμιο Θεσσαλονίκης, Thessaloniki.

Mackridge, P. 1985. The Modern Greek Language. Oxford, Oxford University Press.

Pérez Saldanya, M. 1999. El modo en las oraciones relativas y adverbiales. In I. Bosque and V. Demonte (eds.), vol. 2, 3253-3323.

4 5 6 Syntax - Morphology - Semantics Antonio R. Revuelta Puigdollers

Quer, J. 2009. Twists of mood: The distribution and interpretation of indicative and subjunctive. Lingua 119, 1779-1787.

Revuelta Puigdollers, A.R. 2011. Causal clauses in Spanish and Modern Greek and subjunctive licensing. Studies in Greek Linguistics 31, 408-419. Thessaloniki.

Ridruejo, E. 1999. Modo y modalidad. El modo en las subordinadas sustantivas. In I. Bosque and V. Demonte (eds.), vol. 2, 3209-3251.

Roussou, A. 2006. Συμπληρωματικοί δείκτες Athens, Εκδόσεις Πατάκη.

Tzartzanos, A.A. 1946, 1963. Νεοελληνική Σύνταξις Vols. A and B. Thessaloniki, Aδερφοί Kυριακίδη.

Veloudis, G. 2010. Από τη σημασιολογία της ελληνικής γλώσσας. Όψεις της επιστημικής τροπικότητας. Thessaloniki, Ινστιτούτο Νεοελληνικών Σπουδών (Ίδρυμα Μανόλη Τριανταφυλλίδη).

Syntax - Morphology - Semantics 4 5 7 Major Trends in Theoretical and Applied Linguistics

4 5 8 A vs. en in Spanish Locatives1

Juan Romeu

CCHS-CSIC [email protected]

Abstract

In this paper, we analyze the internal syntactico-semantic structure of the P(reposition)s a and en in Spanish locative constructions. After providing a cartographic approach to Ps, where these elements are decomposed into more projections than Path and Place (Jackendoff 1983), in line with Svenonius (2010), we claim that the difference between a and en is that a lexicalizes a Terminal feature in Spanish, which en is not able to lexicalize. This feature gives the interpretation that the location of the Figure2 is not the only one with which it establishes a relationship in a single locative event.

1. Introduction

In Spanish, although the preposition en is most common in locative constructions like in (1a), the preposition a is also used in certain locative constructions, like in (1b) and (1c):

(1a) Juan está {en/*a} la iglesia. Juan is en/a3 the church ‘Juan is in the church.’

1 The research underlying this work has been partly supported by Grant FFI2009-07114 from the Spanish Ministerio de Ciencia e Innovación and by Grant JAE Predoc from the Consejo Superior de Investigaciones Científicas. 2 In Talmy (1975), Figure is opposed to Ground. The Figure is the entity whose location is being described. The Ground corresponds to the location where the Figure is located or the location with respect to which the Figure moves. Figure and Ground correspond to Trajector and Landmark in other frameworks. In John is in Barcelona, John is the Figure and Barcelona is the Ground. 3 We leave a and en in Spanish in the glosses because we don’t want the reader to get confused between the meaning of these Spanish Ps and the Ps in English into which these Ps are normally translated, i.e. at and in, respectively.

Syntax - Morphology - Semantics 4 5 9 Major Trends in Theoretical and Applied Linguistics

(1b) La casa está {a/*en} la derecha de la iglesia. The house is a/en the right of the church ‘The house is on the right of the church.’

(1c) Tesalónica está {a/en} el norte de Grecia. Thessaloniki is a/en the north of Greece ‘Thessaloniki is at/in the north of Greece.’

The most remarkable point in these examples is the use of the preposition a in locative constructions. This P is normally used in directional constructions in Spanish, unlike en, whose use in directional constructions is restricted to cases with entrar (‘go in, enter’) and other verbs of that kind:

(2a) Juan fue {a/*en} la iglesia. ‘Juan went to the church.’

(2b) Los niños llegaron {a/*en} casa. ‘The kids arrived home.’

Another important point is that, as we have seen in (1a), the use of a is more restricted than the use of en in locative constructions. As we can observe in (1), a is not possible when combined with entities like church. In this paper, we will show that a needs to combine with Grounds that correspond to parts related or connected to another location, like derecha in (1b), which needs to be interpreted as a dependent subpart of the church. As we will see, in case it is not intended to interpret this dependency, the element en is used. This gives the two possibilities in (1c). In the next section, we will explain the most relevant properties of the cartographic model, which we are going to use to analyze these contrasts. In section 3, we will show the locative cases in which the preposition a can appear and we will explain that these cases are those in which two locations can be interpreted in a single locative event by different means. With this, we will show that a cartographic approach is the best way to explain the minimal contrasts that we are analyzing here.

2. A Cartographic Approach to Ps

Before analyzing the differences betweena and en, we will present the so-called cartographic model, which we consider to be the most adequate model to deal with very fine contrasts, as those between a and en in Spanish.

4 6 0 Syntax - Morphology - Semantics Juan Romeu

2.1. Background

Once it was assumed that Ps are lexical categories (Chomsky 1970, Emonds 1972, Jackendoff 1973), it became necessary to split the structure of spatial Ps into more projections to explain subtle properties of Ps. Jackendoff (1983) decomposed Ps into Path and Place. Since then, following a cartographic model (cf. Belletti 2004, Cinque 1999, 2006, Rizzi 1997, 2004, a.o.), many scholars (cf. Koopman 2000, Den Dikken 2010a, Svenonius 2006, 2010, Pantcheva 2011, a.o.) have proposed different decompositions of P into more fine-grained projections, in order to explain constructions for which the Path-Place decomposition was not enough. Here we will work in this way, arguing that the best way to understand the contrasts between a and en in locative constructions is by means of a cartographic approach.

2.2. A Cartographic Approach to Spatial Ps

Although there have been many attempts to capture the structure that underlies spatial constructions in languages (cf. Den Dikken 2010a, 2010b, Pantcheva 2011, a.o.), we think that they are not accurate enough to explain the contrasts we are analyzing in this paper. From our point of view, the most adequate model in the literature is the one proposed by Svenonius (2006, 2010). Svenonius decomposes Place into different syntactic projections, each one corresponding to a semantic feature, in line with Pollock’s (1989) “one feature, one head” claim. For Svenonius (2010), “each component of the semantic decomposition corresponds to a syntactic projection”. In this sense, he establishes a whole set of functional projections that can explain complex spatial examples such as ten meters behind the house (Svenonius 2010:134). Although we are not going to present here the whole structure proposed by Svenonius, we will provide an example of one of the projections into which he suggests Place has to be split. This projection is going to be useful for us later in this paper. Svenonius (2006) claims that, to explain the difference between examples like the ones in (3), it is necessary to have at least one projection more than Place in the structure:

(3a) There was a kangaroo in the front of the car. (3b) There was a kangaroo in front of the car. Svenonius (2006:49-50)

Syntax - Morphology - Semantics 4 6 1 Major Trends in Theoretical and Applied Linguistics

Based on Jackendoff (1996) and Marr (1982), he claims that in cases like (3b) there must exist a projection called Ax(ial)Part. This projection makes front behave as a special noun. Although it preserves some properties of nouns, like, as we will see, the article in Spanish, it lacks other properties of nouns like the possibility of pluralization in English: *There were kangaroos in fronts of the cars. Svenonius (2006) gives many other tests that show that AxParts are different from Ns.4 The important thing is that “AxPart […] is a function from eigenplaces to subparts of them” (Svenonius (2010:132), where eigenplace corresponds to “the space occupied by the Ground” (Svenonius 2010:132), as in Wunderlich (1991), or, in other words, a Region. Thus, AxPart gives a location related to a Region. This new location corresponds to a subpart of the Ground, which can be internal or external (Wunderlich 1991, Svenonius 2006)5. For instance, in (3b), front represents a part related to the car, in this case the external part in front of it:

(4) front

In the schema in (4), the grey square corresponds to the external front part of the car.6 This part is interpreted as connected to the location of the car and not as an independent location by means of AxPart. Following this method, we will try to examine if the difference betweena and en has to do with the presence of a feature in the structure. This way, we will demonstrate that the preposition a in Spanish, even in locative constructions,

4 Pluralization is possible in certain cases, as a reviewer points out: (i) La casa está a orillas del río the house is a riverside.pl of.the river ‘The house is at the river side’ We consider such examples to be idiosyncrasies, which are possible because of the nominal origin of AxParts (Svenonius 2006:61-65). The idea is that even in cases like this one, not all the nominal properties are active. For instance, in (i) it is not possible to add an adjective. 5 For external cases, Svenonius (2010) uses the concept of ‘aura’, meaning that the noun refers to a part of the Ground, which corresponds to an external part of it, but belonging to it in some sense. 6 Of course, the part considered as the front will depend on the frame of reference, such as the viewer, the intrinsic properties of an object, etc. See Svenonius (2010:131) and references therein.

4 6 2 Syntax - Morphology - Semantics Juan Romeu

lexicalizes7 a feature, which can be related to directionality (concretely Terminal, as we will see in next section). In contrast to Fábregas (2007) or Real Puigdollers (2010), we argue that a is not just a locative P. Nevertheless, in contrast to authors like Demonte (2011), who considers a to be directional, we argue that the preposition a lexicalizes also locative features. Thus, we will see in this paper that a in Spanish can be considered a locative and directional P at the same time because it lexicalizes features related to both location and directionality. Following Svenonius’ method, we suggest that there must be a feature in the structure above Place that explains the difference between en and a in Spanish. We call this feature Terminal. The name ‘Terminal’ comes from Hale’s (1986:239-240) opposition between central coincidence and terminal coincidence:

(5) Central coincidence: “the centre of the figure coincides with the centre of the place” Terminal coincidence: “the location of the figure corresponds to its trajectory […], which can be viewed as ending […] or beginning […] at the place”

The reason why we use ‘Terminal’ for this feature is because it gives the possibility to a Figure to establish a dynamic relationship with the Ground, making it possible to have a trajectory. In fact, Hale & Keyser (2002) relate central coincidence to stativity and terminal coincidence to dynamicity. Nevertheless, as we will see later, the Terminal feature doesn’t obligatorily imply dynamicity. Its presence is just the first step for dynamicity to be possible. Then, what is the role of Terminal? We claim that it takes a location in an event and it makes it be understood as displaced with respect to other location. It gives the possibility of having two locations at stake in a single event. For this, it is necessary that at least two locations are available in the event. The Figure can then establish a locative relationship with these locations or, at least, it must be possible to have the interpretation of such a relationship. In directional constructions, it is easy to interpret that the Figure stays in two different locations because it moves from one to the other. The main point here is to demonstrate that, even though in locative and, hence, stative cases, the Figure doesn’t move from one point to the other, it is possible to interpret it

7 In this paper we take ‘lexicalization’ to be a post-syntactic process through which a lexical item gives phonological content to a part of the syntactic structure. This is possible by means of phrasal spell-out (see McCawley 1968, Starke 2001, Fábregas 2007, Svenonius 2010, Pantcheva 2011).

Syntax - Morphology - Semantics 4 6 3 Major Trends in Theoretical and Applied Linguistics

as establishing a relationship with both. This is possible by means of Terminal. To do this, it is necessary that two locations are interpreted in a locative-stative event. In the next section, we will show how this is possible by different means. We will also show that a lexicalizes Terminal and that is why, when it appears in locative constructions, it needs two locations to be somehow interpreted in the event and that the Figure is interpreted as establishing a relationship, despite its lack of movement.

3. A vs. en

With the Terminal feature in our minds, we are on the right track to understand the contrasts between a and en that we presented before and that we repeat now:

(6a) Juan está {en/*a} la iglesia. Juan is en/a the church ‘Juan is in the church.’ (6b) La casa está {a/*en} la derecha de la iglesia. The house is a/en the right of the church ‘The house is on the right of the church.’ (6c) Tesalónica está {a/en} el norte de Grecia. Thessaloniki is a/en the north of Greece ‘Thessaloniki is at/in the north of Greece.’

3.1. Are These Ps Locative?

If we apply Svenonius’ (2010) tests for locatives to a and en, these Ps seem to behave as locatives. For instance, Svenonius argues that locative Ps can combine with verbs such as remain, stay or be located (estar, permanecer…):

(7a) John stayed in the city. (7b) *John stayed to the city.

(8a) Juan permaneció {en/*a} el colegio. Juan remained en/a the school (8b) Juan permaneció {en/a} el fondo de la clase. Juan remained en/a the bottom of the classroom (8c) *Juan permaneció hasta el colegio. Juan remained up-to the school

4 6 4 Syntax - Morphology - Semantics Juan Romeu

As we can see, directional Ps such as to in English or hasta in Spanish ((7b) and (8c), respectively) can’t combine with the relevant verbs. We also observe that en is always possible, as in (8a) and (8b), whereas a seems to be possible only in some cases ((8b)). Taking into account this and the fact that in other Romance languages like French or Catalan the alleged equivalent of a can appear in locative cases like the one in (8a), it is essential to examine the locative cases in which a is possible. This is what we will do now.

3.2. In Which Locative Cases Can a Show Up?

Here we present some more cases in which a is possible in a locative construction:

(9a) La casa está a la derecha de la iglesia. the house is a the right of the church (9b) El puerto está a la orilla del mar. the harbour is a the shore of sea (9c) Tesalónica está al norte de Grecia. Thessaloniki is a-the north of Greece (9d) El agua está a 50 grados. the water is a 50 degrees (9e) Juan está al piano. Juan is a-the piano

What do all these examples have in common? According to Fábregas (2007), we can classify all these cases in three groups depending on the properties of the noun a combines with. First of all, there are those cases in which the noun denotes an (oriented) boundary of objects (Fábregas 2007):

(10) lado ‘side’, borde ‘border’, límite ‘limit’, margen ‘margin’, fondo ‘end, bottom’, término ‘terminal’, vera ‘side of a river’, entrada ‘entrance’, salida ‘exit’, frente ‘front’, norte ‘north’, derecha ‘right’… Fábregas (2007:178-179)

These cases are the ones represented in (9a) through (9c). The second group, represented in (9d), is formed by nouns which correspond to points inside scales:

(11) máximo ‘maximum’, mínimo ‘minimum’, principio ‘beginning’, final ‘end’, mitad ‘middle’ (Fábregas 2007:179)

The last group ((9e)) has to do with pragmatics:

Syntax - Morphology - Semantics 4 6 5 Major Trends in Theoretical and Applied Linguistics

(12) piano ‘piano’, teléfono ‘telephone’, sol ‘sun’, volante ‘steering wheel’… (Fábregas 2007:179-180)

As we will see later, for Fábregas (2007), all these cases have in common that “the figure is in contact with (at least) one point of the boundary of the ground”. For us, as we will show later, what all these examples have in common is that the location of the Figure is not the only one at stake in the event, but it is related to other location. So, taking an example like (9c) from the first group, we see how el norte needs to be related to the centre of Greece. Recalling what we said about AxParts, this is exactly a case where a noun is not behaving as a noun referring to a place but as a subpart of a place. So the first group of locative cases with a corresponds to cases in which the preposition doesn’t combine directly with N, but with an AxPart. Before we said that, for Terminal to be possible, it needs two locations in the event. In AxPart cases we have, precisely, those two locations: the subpart (el norte in (9c)) and the referential Ground (Grecia in (9c)). However, it is also necessary that the two locations are connected. This is what AxPart makes possible. If the two locations are not interpreted as connected or, in other words, if el norte is not interpreted as a subpart of Grecia, but as an independent location, then el norte is not an AxPart and, thus, it is not possible to interpret two locations in the event. In that case, Terminal would not be possible, and only en is possible. This is borne out by the following example:

(13) Tesalónica está (en/*a) el maravilloso norte de Grecia. ‘Thessaloniki is in the marvelous North of Greece.’

If el norte behaves as a normal N and not as an Axpart, then there is only one location at stake in the event (el norte de Grecia) and not two (el norte as a subpart of Grecia and Grecia as the referential location). The difference is that in the former de Grecia doesn’t behave as a referential location for el norte, but just as a modifier. This can also be demonstrated by means of a Measure Phrase. As in (9c), where a is present, there are two locations, the distance between them can be “measured”. On the contrary, in (13) there is no distance because it is a case of central coincidence, so the PP can’t be measured:

(14a) Tesalónica está 15 kilómetros al norte. (14b) *Tesalónica está 15 kilómetros en el norte. ‘Thessaloniki is 15 kilometres towards the north.’

Fábregas (2007:178) says that these examples can be explained assuming that “locative a denotes a place relationship where the figure is in contact with (at least) one point of the boundary of the ground”. It is true that in the cases in which there is a contact with the boundary, a is needed. But the main reason

4 6 6 Syntax - Morphology - Semantics Juan Romeu

is that in locative cases with a there are two locations at stake. This includes boundary cases, in which the two locations correspond to the boundary and the referential Ground, but also other cases in which it is not so clear that the other location corresponds to a boundary like in the following examples:

(15a) La biblioteca está a la vuelta de mi casa. ‘The library is around my house.’ (15b) La iglesia está a mitad de camino. ‘The church is half-way’ (15c) El vaso está más a la derecha que al centro. ‘The glass is more to the right than to the centre.’

We claim that the common property of all these examples is that it is possible to interpret two locations, regardless of the stativity of the event, and, although the Figure is just located in one of them, this location is interpreted as being displaced from the referential one. This is possible by means of Terminal. Thus, although Terminal doesn’t imply real dynamicity, its presence is the first step to obtain dynamicity. The interpretation of two locations is necessary in order to have movement in the event. 8 The second group can be explained by assuming that a scale is like a path formed by different separated points. So in the following examples the two necessary locations for Terminal correspond to the point of the scale being chosen and an idealized neutral point of the scale:

(16a) El agua está a 100 grados. the water is a 100 degrees (16b) La ventana está a 7 metros. the window is a 7 meters

In these cases, the point of the scale that is chosen is interpreted as a reached point. The scale allows the presence of Terminal, because any point belonging to a scale can be related at least to a referential point. Finally, although we won’t say much about the last group, i.e. the group that has to do with pragmatic factors, consider the following example:

(17) Juan está al volante. Juan is a-the steering-wheel ‘Juan is driving.’

8 In this sense, Brucart (2010) suggests that the preposition a is ‘more dynamic’ than the preposition en and that in locative cases with a there is an implicit path.

Syntax - Morphology - Semantics 4 6 7 Major Trends in Theoretical and Applied Linguistics

As we see, here the interpretation is that Juan is driving and not that Juan is in a locative relationship with the steering wheel. A possible explanation of these cases might be that in all these cases the Figure has to approach the Ground somehow. In a case like (17), Juan has to move his arms towards the steering wheel. This dynamic action may give rise to the presence of Terminal and, thus, to the presence of a in Spanish. However, these cases need a better explanation which we hope to return to in further research. In sum, we have seen that in all the cases where a appears it is necessary to interpret the Figure as being spatially related to two different locations in a single event, even in locative cases. This is possible by means of Terminal. Now, let’s see how this is represented in the structure.

3.3. The Different Structure of a and en in Spanish.

So far, we have said that, whenever a Terminal feature is present, only a and not en is able to lexicalize the structure in Spanish, and, conversely, that when there is no Terminal feature present, only en and not a can lexicalize the structure. So, assuming that both a and en lexicalize a Place feature, the representation of the structure is the following one (abstracting away from irrelevant structure):

(18) TerminalP

Terminal PlaceP

a/*en

Place …

en/*a

What this structure represents is that, when only Place and not Terminal is present in the structure, en is the lexical item chosen in Spanish to lexicalize it. On the contrary, when both Place and Terminal are present, a is the lexical item chosen.9

9 The brace represents that a lexicalizes both Place and Terminal at the same time, by means of phrasal spell-out. This could also be explained as head movement of a from Place to Terminal.

4 6 8 Syntax - Morphology - Semantics Juan Romeu

Some remarks are needed. First of all, we can establish that the question of whether a is a locative or a directional element is irrelevant. The important issue is that a lexicalizes different features. As we have seen in this paper, it lexicalizes at least Place and Terminal. The fact that it lexicalizes Terminal makes a perfectly adequate for use in directional constructions, because such constructions obligatorily need a Terminal projection in their structure. At the same time, it doesn’t prevent a from appearing in locative constructions, as long as there are two locations at stake in the event with which the Figure is spatially related. In this sense, we disagree with “coercing” analyses like the one in Fábregas (2007), or the one in Real Puigdollers (2010), who proposes The Extended Structural Ambiguity Hypothesis based on The Structural Ambiguity Hypothesis in Gehrke (2008). The analyses of Fábregas and Real Puigdollers are summarized in the following quotations, which correspond to Fábregas (2007:177) and Real Puigdollers (2010:129), respectively

A is a locative preposition which can get a directional reading embedded under the right syntactic configuration, just like English over.

For any spatial preposition that can be interpreted as locative, it is only locative. Any ambiguity between a directional and a locative meaning is structural.

We claim that the behaviour of a is not dependent on the structure in which it is embedded, i.e. by means of coercion. Rather, we claim that a lexicalizes a different chunk of the structure than en. Thus, even in the locative cases, they don’t lexicalize the same structure. We also suggest that it is the structure that the preposition a lexicalizes what could “coerce” the structure in which it is embedded in some cases, as can be seen in the following contrast:

(19a) Juan corrió en la iglesia (obligatorily locative) ‘Juan ran in the church’ (19b) Juan corrió a la iglesia (obligatorily directional) ‘Juan ran to the church’

We suggest that in (19a) the verb is just an activity verb, whereas in (19b) the verb is encoding a transition, because the presence of a in the latter means that there is a Terminal feature in the structure. We intend to give a better explanation of this in the future. Another important question has to do with the presence of Terminal in other languages. Is it possible to find examples in other languages in which Terminal triggers a different choice of a lexical item? The answer should be ‘yes’, and

Syntax - Morphology - Semantics 4 6 9 Major Trends in Theoretical and Applied Linguistics

it seems that there are in fact such examples. In certain locative constructions in English, to is the element chosen. Interestingly, in those cases Axparts are present, as in (20):

(20a) The Palace Museum is to the north of Tiananmen Square. (20b) My house is to the left of the elevator.

Similarly, the presence of accusative case in directional constructions in German might possibly be explained by the presence of Terminal, as in the following examples from Den Dikken (2010a:112):

(21a) er rannte in dem Laden → Locative interpretation he ran in the-dat store (21b) er rannte in den Laden → Directional interpretation he ran in(to) the-acc store

Furthermore, a reviewer points out that, for cases like the ones we have seen in this paper the preposition an is used in German, indicating the location of the Figure at the Ground’s border: am Meer (‘at the seaside’), an der Grenze (‘at the frontier’), am Steuer (‘at the steering-wheel’), am Klavier (‘at the piano’). It could be that an in German lexicalizes Terminal. This could also be the explanation for cases like the following in North Sámi, from Svenonius (2011:6):

(22) Nieida bissánii Romssii girl remained Tromsø.ill ‘the girl remained in Tromsø.’

In this case, the presence of the illative case could be due to the fact that the verb remain in North Sámi implies that there is another location where the Figure should be instead of the location where it actually is. Of course, this needs a deeper examination that we leave for further research.

4. Brief Conclusions

In this paper, we have seen that, in order to properly understand the behavior of lexical items, it is important to uncover their fine-grained internal structure. By doing so, we have shown that taking the preposition a to be locative or directional is nonsensical. The key fact is that a lexicalizes a Terminal feature which en is not able to lexicalize in Spanish. This Terminal feature explains the

4 7 0 Syntax - Morphology - Semantics Juan Romeu

apparent double role of a as a locative and directional preposition, giving a very simple explanation of why a can appear in certain locative constructions while at the same time it can appear in directional constructions: Terminal makes it possible to relate a Figure to two locations in a single event. This way, what at first sight were considered to be two different possible lexical items (a and en) for a single head, correspond in fact to two different structures. This conclusion would not have been possible without having taken a closer, microscopical and submorphemic look, which a cartographic approach allows. Following a cartographic approach, we are able to understand the fine underlying structure that lexical items lexicalize and we can get rid of superficial classifications and labeling, as for example, the classification of lexical items into categories. We suggest that the best way to work with lexical items is taking into account the minimal features they lexicalize.

Syntax - Morphology - Semantics 4 7 1 Major Trends in Theoretical and Applied Linguistics

References

Belletti, A. 2004. Aspects of the low IP area. In L. Rizzi (ed.), The Structure of CP and IP. Oxford: Oxford University Press, 16-51.

Chomsky, N. 1970. Remarks on Nominalization. In R. Jacobs and P. Rosenbaum (eds.), Readings in English Transformational Grammar. Waltham, MA: Ginn, 184- 221.

Cinque, G. 1999. Adverbs and functional heads. Oxford/New York: Oxford University Press.

Cinque, G. 2006. Restructuring and functional heads. Oxford: Oxford University press.

Demonte, V. 2011. Los eventos de movimiento en español: construcción léxico- sintáctica y microparámetros preposicionales. In J. Cuartero Otal, L. García Fernández and C. Sinner (eds.), Estudios sobre perífrasis y aspecto. München: peniope, 16-42.

Den Dikken, M. 2010a. In G. Cinque and L. Rizzi (eds.), The cartography of Syntactic Structure, vol.6. Oxford: Oxford University Press, 74-126.

Den Dikken, M. 2010b. Directions from the GET-GO. On the syntax of manner-of- motion verbs in directional constractions. Catalan Journal of Linguistics 9: 23-53.

Emonds, J. 1972. Evidence that indirect object movement is a structure preserving rule. Foundations of Language 8: 546–561.

Fábregas, A. 2007. An Exhaustive Lexicalisation Account of Directional Complements. Nordlyd: Tromsø Working Papers on Language and Linguistics 34(2): 165-199.

4 7 2 Syntax - Morphology - Semantics Juan Romeu

Gehrke, B. 2008. Ps in Motion: On the Semantics and Syntax of P Elements and Motion Events. PhD dissertation, Utrecht University.

Hale, K. 1986. Notes on World View and Semantic Categories: Some Warlpiri Examples. In P. Muysken and H. van Riemsdijk (eds.), Features and Projections. Dordrecht: Foris, 233-254.

Hale, K. and S.J. Keyser. 2002. Prolegomenon to a theory of argument structure. Cambridge, MA: MIT Press.

Jackendoff, R. 1973. The Base Rules for Prepositional Phrases. In S.R. Anderson and P. Kiparsky (eds.), Festschrift for Morris Halle. New York: Holt, Rinehart & Winston: 345-356.

Jackendoff, R. 1983. Semantics and cognition. Cambridge, MA: MIT Press.

Jackendoff, R. 1996. The proper treatment of measuring out, telicity, and perhaps even quantification in English. Natural Language and Linguistic Theory 14: 305– 354.

Koopman, H. 2000. Prepositions, postpositions, circumpositions and particles: The structure of Dutch PPs. In H. Koopman (ed.), The syntax of specifiers and heads. London: Routledge, 204–260.

Marr, D. 1982. Vision: A Computational Investigation into the Human Representation and Processing of Visual Information. New York: W.H. Freeman.

McCawley, J.D. 1968. Lexical insertion in a transformational grammar without Deep Structure. In B.J. Darden, C.-J.N. Bailey and A. Davidson (eds.), Papers from the fourth regional meeting of the Chicago Linguistic Society. Chicago: University of Chicago.

Pantcheva, M. 2011. Decomposing Path. The nanosyntax of directional expressions. PhD dissertation, Tromsø University.

Pollock, J.-Y. 1989. Verb Movement, Universal Grammar, and the Structure of IP. Linguistic Inquiry 20: 365–424.

Real Puigdollers, C. 2010. A microparametric approach on goal of motion constructions: properties of adpositional systems in Romance and Germanic. Catalan Journal of Linguistics 9: 125-150.

Syntax - Morphology - Semantics 4 7 3 Major Trends in Theoretical and Applied Linguistics

Rizzi, L. 1997. The Fine Structure of the Left Periphery. In L. Haegeman (Ed.), Elements of Grammar. Handbook of Generative Syntax. Dordrecht: Kluwer, 73-137

Rizzi, L. (Ed.) 2004. The Structure of CP and IP – The Cartography of Syntactic Structures vol 3. Oxford: Oxford University Press.

Starke, M. 2001. Move Reduces to Merge: A Theory of Locality. PhD dissertation, University of Geneva.

Svenonius, P. 2006. The emergence of axial parts. In P. Svenonius and M. Pantcheva (eds.), Nordlyd, Tromsø. Working Papers in Language and Linguistics: 33.1, Special Issue on Adpositions. Tromsø: Tromsø University, 49-77.

Svenonius, P. 2010. Spatial P in English. In G. Cinque and L. Rizzi (eds.), The cartography of Syntactic Structure, vol.6. Oxford: Oxford University Press, 127- 160.

Svenonius, P. 2011. Directed Manner of Motion in North Sámi. Presented in the Workshop on Verbal Elasticity. Universitat Autònoma de Barcelona.

Talmy, L. 1975. Semantics and syntax of Motion. In J. Kimball (ed.), Syntax and Semantics vol. IV. New York: Academic Press, 181-238.

Wunderlich, D. 1991. How do prepositional phrases fit into compositional syntax and semantics?. Linguistics 29: 591-621.

4 7 4 Syntax - Morphology - Semantics In Situ, Ex Situ and (Non) Echo Questions*

Anna Roussou, Christos Vlachos and Dimitris Papazachariou

University of Patras [email protected], [email protected], [email protected]

Abstract

In this paper we consider echo and non-echo questions in Greek, realized either ex-situ or in-situ. We show that there are syntactic differences, depending on the position where the wh-phrase is realized, and that non-echo questions involve different types of quantification, namely exhaustive for in-situ and non- exhaustive for ex-situ. On the other hand, echo questions have no quantificational properties. We next provide novel evidence for the prosodic properties of the two types of questions, arguing that irrespectively of the position where the wh- element is realized, there is a distinct prosodic pattern that disambiguates echo from non-echo questions.

1. Introduction

The formation of constituent, or wh-, questions is typically assumed to fall into one of the two patterns, exemplified by English, as in (1a), and Chinese, as in (1b).1

(1a) Who did you see? (1b) Ni kanjian-le shei? you see-ASP who “Who did you see?”

* We would like to thank Joseph Emonds, Maria Giakoumelou, Kleanthes Grohmann, Ianthi- Maria Tsimpli, and the ISTAL 20 participants, for useful comments and discussion. Research for this paper has been supported by the Research Committee of the University of Patras (C. Caratheodory program, Grant number C.581). 1 For a more elaborate discussion on the wh-parameter see, Richards (1997) and Pesetsky (2000) among others.

Syntax - Morphology - Semantics 4 7 5 Major Trends in Theoretical and Applied Linguistics

The English construction in (1a) is characterized as ‘wh-ex situ’, and involves displacement of the argument who from the object position of the verb to the left periphery of the clause. On the other hand, the Chinese construction in (1b) exhibits ‘wh-in situ’ form, on the assumptions that the argument shei has remained in its canonical object position (from Huang 1982: 253, (159)). In generative grammar terms, wh-elements in English-type languages are displaced via a movement operation (or its equivalent in Minimalism), leaving a variable in the original position of extraction. This variable is bound by the wh- element which is interpreted as an operator (see Chomsky 1977; Katz & Postal 1964; Baker 1970; and Pesetsky 1987, for early discussions). ‘Wh-movement’ takes place in (narrow) syntax, and its output is read off at both PF (displacement, gap) and LF (operator-variable). The formation of Chinese (1b) has received a number of different accounts. For example, according to Huang (1982) the element shei also moves to the left periphery, but covertly. For Nisigauchi (1990) there is no movement involved, but in-situ quantification by an abstract Question (Q) operator in the left periphery. Another approach argues for feature-movement (Chomsky 1995): what moves is the wh-feature only; thus the wh-element is realized in situ in PF, while the relevant feature appears in the scope position at LF. Finally, for Kayne (1998) all instances of movement are overt, and therefore the wh-in situ moves to the left periphery, followed by massive remnant movement of the rest of the clause to a higher position. In all these approaches shei realizes the variable (see also Cheng 1991), at least at the level of PF. Irrespectively of their differences, the core assumption remains that wh-constructions give rise to an operator-variable chain. Variation arises as to how this chain is realized. Greek is a typical wh-movement language, like English (see Agouraki 1990; Tsimpli 1990, 1995; Anagnostopoulou 1994; Kotzoglou 2005). It also exhibits wh-in situ, which is typically assumed to be restricted to an echo interpretation (EQ) (Tsimpli 1998), while ex-situ is a true, information seeking question (IQ). Consider the following examples (the echo reading is indicated with capitalizing):

(2a) Pjon idhes? (IQ) who-ACC saw-2SG “Who did you see?” (2b) Idhes PJON? (EQ) saw-2SG who-ACC “You saw WHO?”

As (2) shows, both ex-situ and in-situ are possible, but they are associated with different interpretations. IQs are true questions, since they ask forthe attribution of a value (out of a set of options) to the wh-element (the variable).

4 7 6 Syntax - Morphology - Semantics Anna Roussou, Christos Vlachos and Dimitris Papazachariou

On the other hand, EQs ask for a repetition of a given value. As Carnie (2006: 340) puts it “...echo questions are not requests for new information; instead they are requests for confirmation of something someone has heard”.2 On the basis of (2), it could be argued that different syntactic realizations of the wh-element go along with different interpretations, thus maintaining a one to one (1:1) mapping between syntax and interpretation (and also PF- realization). However, the picture is not so straightforward. As has been argued by Sinopoulou (2009) and Vlachos (2010, 2012), the structure in (3b) may also be interpreted as a true question (IQ). Similarly, we argue that the structure in (3a) may also be interpreted as an echo question (EQ). These patterns are illustrated in (3):

(3a) PJON idhes? (EQ) who-ACC saw-2SG “WHO did you see?” (3b) Idhes pjon? (IQ) saw-2SG who-ACC “You saw who?”

One additional property to be taken into consideration is that the two constructions bear different intonation, depending on whether they are IQ or EQ, which as we will show in this paper holds irrespectively of the in-situ vs. ex-situ realization. In the light of (2) and (3), Greek exhibits a ‘quadratic’ pattern of form- meaning mapping: wh-ex situ can be IQ or EQ, and the same holds for wh-in situ. Interestingly, English as well, seems to allow for this ‘quadric’ pattern, as opposed to the ‘dyadic’ pattern assumed so far. For example, Pires & Taylor (2007) discuss IQ in-situ constructions in English, while Sobin (2010) comes back to the discussion of EQs, considering ex-situ cases as well. There are a number of questions that arise at this point. First, what are the differences between the ex-situ and the in-situ counterparts? Second, do these differences arise in narrow syntax or are they computed directly at the interfaces? Third, what are the implications of this (apparent) optionality for grammar? In the present paper, we address these questions and provide novel empirical evidence from the PF-properties of echo and non-echo questions. We argue that there is no one to one correlation between syntax and meaning, and

2 For more elaboration on various aspects of echo questions, see Bolinger (1978, 1987); Sobin (1978, 1990, 2010); Parker & Pickeral (1985); Ginzburg & Sag (2001); and Fiengo (2007), among others.

Syntax - Morphology - Semantics 4 7 7 Major Trends in Theoretical and Applied Linguistics

that PF also plays a role in disambiguating among different interpretations. In this respect, there is no optionality in the strict sense, and there is no need to encode these differences in narrow syntax. The paper is structured as follows. Section 2 discusses the basic properties of IQs and EQs with respect to their syntax and interpretation. Section 3 provides evidence from PF, based on prosody. Section 4 concludes the discussion.

2. Wh-questions: Syntax and Interpretation

Let us start by considering the question of optionality, which seems to be the case for ex-situ and in-situ, echo and no echo questions. In Minimalism (Chomsky 1995 onwards), Agree is the basic operation between two elements α and β, with α being the Probe and β the Goal. This operation, broadly speaking, involves feature matching between the Probe and the Goal. It is possible for the Probe to bear one more feature, namely the EPP (thus generalizing the Extended Projection Principle), in which case, Agree is followed by copying the Goal and merging it with the Probe (internal Merge). The traditional Movement operation then reduces to the basic operations of Agree and (Internal) Merge, triggered by the EPP feature. The English wh-construction in (4a) illustrates this mechanism (similar for Greek):

(4a) [CP Which book C[EPP/wh] did [TP you did [vP write which book]]]?

The assumption is that C has an uninterpretable wh-feature which agrees with the interpretable wh-feature of the Goal which book; the EPP on C triggers Merge of which book with C, yielding displacement of which book. It can then be assumed that the in-situ construction in (4b) involves Agree, but there is no EPP feature associated with C, and therefore no Internal Merge takes place (see for example, Denham 2000, among others):

(4b) [CP Op C[wh] [TP You [vP wrote which book]]]? [non-echo]

Descriptively then, optionality could be taken to reduce to the optional presence of the EPP feature. Note though that ‘optionality’ of this sort is attested with echo questions as well (cf., (2b) and (3a)). Can we then claim that the EPP is optional here as well? What is important is the fact that irrespectively of where the wh-phrase is realized, i.e. ex-situ or in-situ, the two basic readings, i.e. non-echo and echo, pertain. So in this respect, displacement driven by the EPP does not alter the core interpretation that is relevant at LF. On the other hand, it has an effect on

4 7 8 Syntax - Morphology - Semantics Anna Roussou, Christos Vlachos and Dimitris Papazachariou

PF, as the position where the wh-phrase is realized differs. Let us then assume that the EPP is nothing more but an instruction for lexicalization, and as such not necessarily a syntactic feature. If this is the case, then the fact that the wh-phrase can be realized either in the left periphery (ex-situ) or in-situ does not on its own distinguish between non-echo and echo readings. Instead it is a broader option available not only for wh-phrases, but for other non-wh arguments and adjuncts in a given grammar. For example, while English has a relatively rigid SVO order, Greek further allows for VSO, in a neutral context. In other words, syntax makes available different positions where elements may occur. The relative freedom exhibited across languages relates to PF and is subject to parametric variation. As we will see immediately below, each option, at least in the case of non-echo questions, comes with a certain cost.

2.1. Distribution

We can now focus on the distributional properties of (non-) echo questions, taking our empirical evidence from Greek. First, it has to be mentioned that the in-situ pattern is not restricted to main clauses but is found in embedded contexts as well (we enclose the wh-in situ IQs/EQs in parentheses in order to show that simultaneous manifestation with wh-ex-situ is prohibited):

(5a) Pjos/PJOS [ipan [oti efije (pjos/PJOS)]]? who-NOM said-3PL that left-3SG (who-NOM) “Who did they say left?” (5b) Pu/PU [ipan [oti tha pai (pu/PU)]]? where said-3PL that will go-3SG (where) “Where did they say that s/he will go?”

In (5a) we have a wh-argument, while in (5b) there is a wh-adjunct; both allow for ex- and in-situ with a non-echo and an echo reading in either position. Notice that wh-in situ (IQs and EQs) in indirect questions, i.e. complements to interrogative verbs like rotao (ask), is possible only to the extent that the (interrogative) complementizer an (if/whether) is present, as shown in (6):

(6) Rotisan [*(an) ipe ti/TI]? asked-3PL if said-3SG what-ACC “They asked *(if/whether) she said what/WHAT?”

We assume that an is necessary in order to satisfy the selectional requirements of the matrix verb and also to license the wh-in situ by lexicalizing the relevant Q operator.

Syntax - Morphology - Semantics 4 7 9 Major Trends in Theoretical and Applied Linguistics

As has been argued in the literature (Sinopoulou 2009; Vlachos 2010 for Greek), while ex-situ (non-)echo questions obey islands, so extraction is blocked out of certain configurations, in-situ ones do not. This is illustrated in (7) and (8), with an adjunct and a relative clause respectively:

(7a) Efije [xoris na xeretisi pjon/PJON ]? left-3SG without prt greet-3SG who-ACC “She left without greeting who/WHO?” (7b) *Pjon/PJON efije [xoris na xeretisi __ ]? who-ACC left-3SG without prt greet-3SG “*Who/WHO did she leave without greeting?” (8a) Katigilan ton epistimona pu anakalipse pja/PJa usia? accused-3PL the scientist that discovered-3SG which substance “They accused the scientist that discovered which/WHICH substance?” (8b) *Pja/PJA usia katigilan ton epistimona pu anakalispe?” which substance accused-3PL the scientist that discovered-3SG “*Which/WHICH substance did they accuse the scientist that discovered?”

Relative and adjunct clauses are classified as strong islands (as opposed to embedded interrogatives which are weak islands). As we observe in the above examples, ungrammaticality arises with the wh-ex situ version in both echo and non-echo questions (for more evidence and discussion see Vlachos 2012). 2.2. Interpretation

Shifting our attention to the interpretation of the constructions at hand, let us start with IQs, while EQs enter the discussion at the end. Consider the two ‘micro-discourses’ (pre-established linguistic environments) in (9), under which only wh-ex situ is felicitous (the sign ‘#’ shows infelicity to context).3

(9a) Out-of-the-blue: Anna, ti jinete (# ti)? Anna what-ACC is-happening (what) “Anna, what’s happening?” (9b) Aggressively non-D-linked: Pjos sto kalo irthe (# pjos sto kalo)? who-NOM to-the-good came-3SG (who-NOM to-the-good) “Who on earth came?”

3 Data revolving around (9) were first discussed in Sinopoulou (2009).

4 8 0 Syntax - Morphology - Semantics Anna Roussou, Christos Vlachos and Dimitris Papazachariou

(9a) shows that wh-in situ instances may not facilitate ‘out-of-the-blue’ readings, contrary to their wh-ex situ counterparts. As the name suggests, out- of-the-blue questions do not require a micro-discourse. A similar effect can be observed with sto kalo (on earth) phrases, which attach to the wh-element (cf., (9b)). As Pesetsky (1987) argues, such phrases are considered to force an ‘aggressively non-D(discourse)-linked’ reading, in that “...the appropriate answer is presumed not to figure in previous discourse (p. 111)” (see also den Dikken & Giannakidou 2002, for a more recent discussion). Out-of-the-blue and aggressively non-D-linked contexts enforce a question- reading where there is no presupposition as to what the value of the wh- element is. In the literature on the semantics of questions, the term that is used to describe this property of (wh-)questions, and will be adopted here, is ‘non- exhaustive quantification’ (see Kartunnen 1977; Groenendijk & Stokhof 1984; Heim 1994; and Beck & Rullman 1999, among others). In general terms, we may understand the non-exhaustive quantification of the wh-ex situ questions in (9) as follows: a satisfactory answer to each of these questions does not rely (and, cannot, after all) on a presupposed set of alternative values from which the wh- elements may draw; hence, anything goes. A piece of corroborating evidence for this reasoning may come from (10).

(10a) Who for example was at the party last night? (10b) Pjos ja paradhighma itan sto parti exthes to vradi? who-NOM for example was-3SG to-the parti yesterday the night “Who for example was at the party last night?”

Beck & Rullman (1999) show that a number of linguistic expressions, among them the phrase for example, are ‘markers of non-exhaustivity.’ So, the English wh-ex situ question in (10a) (theirs (90), p. 286), is compatible with this marker, and yields a non-exhaustive reading. This is also the case with the Greek cognate in (10b). Arguably, a wh-in situ IQ yields a ‘(strong) exhaustive quantification,’ in that the wh-element (necessarily) selects a member from a pre-established set of alternative values, entailing the exclusion of the other members of the set. This interpretation sharply contrasts with that of a wh-ex situ IQ, as becomes clear in (11).

(11a)*Itan sto parti exthes to vradi pjos ja paradhighma? was-3SG to-the parti yesterday the night who-NOM for example “Who for example was at the party last night?” (11b) *Ja paradhighma itan sto parti exthes to vradi pjos? for example was-3SG to-the parti yesterday the night who-NOM “Who for example was at the party last night?”

Syntax - Morphology - Semantics 4 8 1 Major Trends in Theoretical and Applied Linguistics

The in-situ pjos (who) in the ungrammatical (11a) is incompatible with the marker of non-exhaustivity ja paradhighma (for example) (compare with (10b)). (11b) additionally shows that the observed ungrammaticality is due to the incompatible interpretations of the wh-in situ and the marker, and has nothing to do with the surface position of the marker per se, since the same result obtains even if the marker is fronted (or, is in any other linear arrangement, for that matter). Now, given that wh-in situ IQs are indeed available, the question is what makes them felicitous. In-situ IQs necessarily lean on the presence of a micro- discourse. Consider the dialogue in (12).4

(12a) Speaker A: My father, my mother and I went to the store to buy eggs, milk and coffee. My mother bought the eggs. (12b) Speaker B: Ke o pateras su aghorase ti? and the father-NOM yours-CL bought-3SG what-ACC “And what did your father buy?”

Speaker A describes an event of buying that involves three agents, i.e., the speaker’s father, mother, and the speaker herself, as well as, three entities, i.e., eggs, milk and coffee (cf., (12a)). Speaker B, who is familiarized with both the set of agents and that of entities, by virtue of (12a), may make the wh-in situ IQ in (12b). In a strong sense, the value of ti (what) must range over the set of entities already present in the micro-discourse (with the exclusion of the ‘eggs’ for obvious reasons). Keeping a similar mode of exposition, let us turn to echo questions. Consider again the data we have discussed so far, in the context of EQs, as in (13)-(14).

(13) Out-of-the-blue: # Anna, TI jinete (TI)? Anna what-ACC is-happening (what) Intended echo: “#Anna, WHAT’S happening?” (14a) Aggressively non-D-linked: # PJOS STO KALO irthe (PJOS STO KALO)? who-NOM to-the-good came-3SG (who-NOM to-the-good) Intended echo: “*WHO ON EARTH came?”

4 The dialogue is first examined in the context of Spanish wh-in situ by Uribe-Etxebarria (2002: 222, (14a,b)), and is attributed to Jimenez (1997).

4 8 2 Syntax - Morphology - Semantics Anna Roussou, Christos Vlachos and Dimitris Papazachariou

(14b) #PJOS ja paradhighma itan sto parti exthes to vradhi who-NOM for example was-3SG to-the parti yesterday the night (PJOS)? who-NOM Intended echo: “*WHO for example was at the party last night?” (15a) Speaker A: My father, my mother and I went to the store to buy eggs, milk and coffee. My mother bought the eggs. (15b) Speaker B: # TI aghorase o pateras su (TI)? what-ACC bought-3SG the father-NOM yours-CL (what-ACC) “# Your father bought WHAT?”

Either ex-situ or in-situ, EQs are infelicitous in out-of-the-blue and aggressively non-D-linked environments, as shown in (13) and (14a) respectively. It follows that EQs, unlike wh-ex situ IQs, do not facilitate ‘non- exhaustive’ readings. This is further verified by the ungrammaticality of EQs with the non-exhaustivity marker ja paradhighma (for example) (cf., (14b)). Moreover, the dialogue in (15s) and (15b) shows that a pre-established set of alternative values is not enough to make an EQ felicitous, irrespective of the ex-situ or in-situ form. Therefore, EQs, unlike wh-in situ IQs, do not yield ‘exhaustive’ readings, either. An EQ is felicitous so long as the value of the wh-element is prominently figured (to be qualified shortly) in the micro-discourse. Witness againthe dialogue in (15a), repeated below for convenience, with the continuation given in (15b’):

(15a’) Speaker A: My father, my mother and I went to the store to buy eggs, milk and coffee. My mother bought the eggs. (15b’) Speaker B: TI aghorase o pateras su (TI)? what-ACC bought-3SG the father-NOM yours (what-ACC) “Your father bought WHAT?”

In principle, (15a’) makes available a set of values from which a wh-element may draw. These values range over ‘eggs,’ ‘milk’ and ‘coffee,’ but only ‘eggs’ acquires a prominent status among the other members of the set, which are still under ‘negotiation,’ so to speak, since the relevant agents (i.e., ‘father’ and ‘speaker A’) have not been mapped to the corresponding entities (i.e., ‘milk’ and ‘coffee’). The saliency of the ‘eggs’ is what makes the EQ felicitous in (15b’), in that ti (what) targets the most prominent value in the micro-discourse.

Syntax - Morphology - Semantics 4 8 3 Major Trends in Theoretical and Applied Linguistics

In other words, ti (what) presupposes no set of alternative values, either non- exhaustively or (strongly) exhaustively, but just a single value, here the ‘eggs’. To say that EQ-contexts do not facilitate sets of alternative values for the wh- elements, amounts to saying that the wh-element lacks quantification in these contexts, since the notion ‘(wh-) quantification’ entails the notion ‘set of values’. For ease of comparison with IQs, we will coin the relevant EQ reading ‘individual’ (borrowing Tsimpli’s 1998 terminology). To summarize so far, IQs and EQs behave alike with respect to island sensitivity in their ex-situ manifestation, and show island insensitivity in their in-situ manifestation. Ex-situ IQs can be uttered in an out-of-the blue context, and be aggressively non-D-linked. These two properties were used as evidence for the claim that ex-situ wh-phrases (IQs) show non-exhaustive quantification, in the sense that they do not require a presupposed set of alternatives values that will be assigned to the wh-variable. On the other hand, in-situ IQs cannot be out-of-the-blue questions and cannot be aggressively non-D-linked. Instead, they always require the availability of a presupposed set of values, and in this respect they are taken to manifest (strong) exhaustive quantification. EQs cannot also be out-of-the blue and aggressively non-D-linked, either in ex-situ or in- situ. Moreover, a presupposed set of alternative values does not suffice, and in this respect, they do not show any kind of exhaustive quantification. At this point we observe the following: the optional realization in-situ or ex-situ may be constrained by syntax (islands) and may affect the quantificational reading of the wh-element. The latter in particular is clearly the case of IQs, and does not seem to hold for EQs (but see Sobin 2010 for English). In other words, we could argue that what looks like optionality comes with a certain cost (syntax and/or interpretation). Having provided a (brief) discussion of the basic distributional and interpretation properties of the various wh-constructions, we next turn to their the prosodic (PF-) properties.

3. The PF-properties of (non-) Echo Questions

Let us next concentrate on the intonation of IQs and EQs. In order to examine the relevant prosodies, we have conducted an experiment that emulated the natural production of the configurations under consideration, in laboratory conditions.5 Specifically, informants were asked to participate in two pre-constructed written

5 We are grateful to the 20 research participants from the University of Patras, for the completion of this experiment.

4 8 4 Syntax - Morphology - Semantics Anna Roussou, Christos Vlachos and Dimitris Papazachariou

dialogues that facilitated the production of all the four possible configurations, in a context that elicits these structures as naturally as possible. One dialogue was constructed to evoke IQs, and a different dialogue targeted EQs. Each dialogue involved two participants, one of whom was the informant at hand and the other an imaginary interlocutor. A preliminary, yet crucial result of the experiment is that informants were able to produce casually all the four different utterances. This provides independent evidence in favour of the existence of all the four types of wh-questions in Greek. Another result pertains to the intonation contours that each configuration manifests, to which we return shortly, starting from the intonation of IQs, and continuing with that of EQs. The description of the spoken data is implemented through the PRAAT program for speech analysis and synthesis, while for the annotation of the intonation contours, we use GrToBI (i.e., Greek ToBI), as has been modified by Arvaniti & Baltatzani (2005). GrToBI is a tool for annotating the intonation and prosodic structure of spoken instances of Greek, within the autosegmental-metrical framework of intonational phonology established by Pierrehumbert (1980). Arvaniti (2001) has extensively studied and described the intonation structure of the majority of Greek wh-ex situ IQs, as a sequence of L*+H L- !H%. Specifically, L*+H is the pitch accent, L- is the phrase accent, and !H% is the boundary tone. According to Arvaniti, the pitch accent appears on the wh-element, at the beginning of the wh-question, and is either high (H*), if the utterance starts immediately with a monosyllabic stressed wh-word, or rising (L+H), if the beginning of the utterance is another word, which is unstressed and surfaces before the wh-element (see also Arvaniti & Ladd 2009, for discussion). Typical instances of the latter are prepositional phrases that head wh-words, like apo pu (from where), whereby the preposition apo (from) phonetically precedes the adverb pu (where). The phrase accent is a L- tone, which is realized either phonetically, if there is not enough segmental material, or, on a lexically stressed syllable following the wh-element, when the wh-question is not a short one. Finally, there is always a rising at the final syllable of the utterance, therefore a H% boundary tone. Arvaniti also argues that in most of the cases, the H boundary tone is not really high, therefore she characterizes it as a down-stepped tone. With the previous clarifications in mind, consider, first, the intonation structure of the wh-ex situ IQ in (16), which has been produced by the informants of the present experiment as in Figure 1.

(16) Ke pja nomizis oti idhe? (wh-ex situ IQ) and who-ACC think-2SG that saw-3SG “And who do you think that s/he saw?”

Syntax - Morphology - Semantics 4 8 5 Major Trends in Theoretical and Applied Linguistics

The intonation pattern in Figure 1 perfectly agrees with Arvaniti’s pattern. In particular, the pitch accent is also a L*+H (as the rising continues to the next syllable after the stressed one), the phrase accent is a L- which is realized on the next stressed syllable from pja (who), and the boundary tone is a !H%, realized on the last syllable of the utterance.

Figure 1. wh-ex situ constituent question

Consider now the intonation contour of the wh-in situ counterpart in (17), as illustrated in Figure 2.

(17) Ke nomizis oti idhe pja? (wh-in situ IQ) and think-2SG that saw-3SG who-ACC “And who do you think that s/he saw?”

As we may observe, the intonation contour of pja (who), at the right edge of the utterance, is that of a typical wh-ex situ IQ, with this crucial diff erence though: the same intonation contour that expands over the whole sentence, in the case of the wh-ex situ IQ in Figure 2, concentrates on the wh-element, vis-à-vis the wh-in situ IQ in (17). So, as regards the ‘question’ intonation, (16) manifests a ‘spreading’ melody, so to speak, while (17) a ‘shrinking’ one. Regarding the part of the utterance that precedes pja (who) in (17), i.e., ke nomizis oti idhe (and you think that he saw), it is reasonable to argue that it constructs as an intermediate intonation phrase with two pitch accents, that is, L*+H & L+H*, and a L- phrase accent (cf., Figure 2). The most important phonetic cue in favour of this hypothesis is the pitch gap of more than 100 Hz between the end of the intonation contour of the preceding part and the beginning of the wh-element. It could also be noted that the preceding intermediate intonation phrase presents a typical intonation contour of the pre-focus element.

4 8 6 Syntax - Morphology - Semantics Anna Roussou, Christos Vlachos and Dimitris Papazachariou

Figure 2. wh-in situ constituent question

Turning to the intonation contour of EQs, in comparison with that of IQs, witness the panel in Figure 3, which is the melody of the wh-ex situ EQ in (18).

(18) PJA idhe? (wh-ex situ EQ) who-ACC saw-3SG “He saw WHO?”

Figure 3. wh-ex situ echo question

Syntax - Morphology - Semantics 4 8 7 Major Trends in Theoretical and Applied Linguistics

The echo question presents a sequence of L* L- H%. The L* is a low plateau pitch accent on the stressed syllable of pja (who), the L- appears as a low plateau on the next stressed syllable (in case there is one), and the intonation contour is completed with a really high H% during the last syllable of the utterance. Crucially, this intonation pattern is clearly distinct from that of the wh-ex situ IQs in (16). Finally, the wh-in situ EQ counterpart in (19) has the melody in Figure 4.

(19) Idhe PJA? (wh-in situ EQ) saw-3SG who-ACC “S/he saw WHO?”

Figure 4. wh-in situ echo question

Interestingly, if we compare the melody of the wh-ex situ EQ in (18), with that of its in situ counterpart in (19), we fi nd a ‘spreading vs. shrinking’ eff ect, similar to the one we encountered in the case of IQs above. In particular, the ‘spreading’ melody of a typical wh-ex situ EQ (cf., Figure 3) concentrates on the in situ pja (who) in (19). Again, the part that precedes the wh-element manifests as an intermediate intonation phrase. The most important phonetic cue for this assumption is the pitch gap of more than 60 Hz between the end of the intonation contour of the preceding part and the beginning of the wh-element. What we observe in the above graphs is that IQs and EQs show a spreading contour in their ex-situ realization, and a shrinking one in their in-situ realization.

4 8 8 Syntax - Morphology - Semantics Anna Roussou, Christos Vlachos and Dimitris Papazachariou

However, the contour is distinct between IQs and EQs. In short, the same syntactic form (in-situ or ex-situ) corresponds to different contours; thus any ambiguity that may arise in syntax, since the same form may in principle be compatible with two different readings, is resolved at PF.

4. Conclusions

In the present paper we have provided a first account to the differences and similarities between echo and non-echo questions. First, either type of question may exhibit an ex-situ or an in-situ variant. Ex-situ is subject to locality (islands), while this is not the case for in-situ. The latter suggests that this ‘option’ involves binding in the classical sense. At least in the case of non-echo (IQ) questions, we assume that the Q operator in C binds the wh-element in situ, through Agree. The wh-phrase then lexicalizes the position of the variable. In the ex-situ version, the wh-phrase once again agrees with Q in C, while the wh-phrase itself lexicalizes the relevant operator (see also Manzini & Savoia 2011). In either case, there is an instruction for lexicalization (an alternative way of understanding the EPP, which thus becomes redundant), in one of the two positions of the wh-chain. This idea is quite natural under the assumptions that: a) nothing, in principle, should block the lexicalization of the relevant positions in a wh-chain, and b) variables may or may not have a lexicalization (see Manzini & Roussou 2011). These two ‘options’ are illustrated in (20) below:

(20a) [C Pjon Q [I idhes [V pjon]]]? who-ACC saw-2SG who-ACC “Who did you see?”

(20b) [C Q [I idhes [V pjon]]]? saw-2SG who-ACC “Who did you see?”

In either case, Q binds a variable corresponding to the internal argument of the verb in the particular example, forming an Operator-variable chain. Further elaboration of this approach will not be addressed here, due to space limitations (but see Vlachos 2012 for a discussion). Binding in the Greek case escapes locality effect (at least for island configurations). It should be mentioned though that this is not necessary the case of all wh-in situ constructions (see, e.g., Mathieu 1999 for French). Second, it was shown that the melody of IQs is L*+H (or H*) L- !H%, while that of EQs is L* L- H% (for space limitations, we do not repeat the relevant panels). In the light of this, we maintain that intonation may disambiguate an otherwise ambiguous utterance (see Jespersen [1933]2006; Sportiche 1998),

Syntax - Morphology - Semantics 4 8 9 Major Trends in Theoretical and Applied Linguistics

and suggest, in turn, that PF disambiguates between the two utterances by assigning clearly distinct intonation contours. Of course, this is part of the story, and not the whole story, because, so far, we have said nothing about the relevant interpretational differences between ex-situ and in-situ forms examined in section 2.2. In particular, as we may recall, in non-echo questions (IQ), wh- elements map to non-exhaustive quantification when ex-situ but to exhaustive quantification when in situ IQs. On the other hand, echo questions (EQ) yield no quantification, giving rise to an individual reading. Here, we propose that the observed ‘fluctuation’ in the interpretation of wh-elements is due to how syntactic form affects prosody, and how the latter, in turn, affects meaning. In this respect, the picture that emerges, subject to elaboration in future work, is that LF must be visible to PF and vice versa, as has indeed been argued by Brody (1995), and Vergnaud & Zubizarreta (2005), among others. This of course raises the issue of what ‘narrow syntax’ amounts to. We leave this topic open to future research.

4 9 0 Syntax - Morphology - Semantics Anna Roussou, Christos Vlachos and Dimitris Papazachariou

References

Agouraki, Y. 1990. On the projection of maximal categories: the case of CP and FP in Modern Greek. UCL Working Papers in linguistics 2: 183–200.

Anagnostopoulou, E. 1994. Clitic Dependencies in Modern Greek. Doctoral dissertation, University of Salzübrg.

Arvaniti, A. 2001. The intonation of wh-questions in Greek. Studies in Greek Linguistics 21: 57–68.

Arvaniti, A. and M. Baltazani. 2005. Intonational analysis and prosodic annotation of Greek spoken corpora. In S.-A. Jun, (Ed.), Prosodic Typology: The Phonology of Intonation and Phrasing, 84–117. Oxford: Oxford University Press.

Arvaniti, A. and D.R. Ladd. 2009. Greek wh-questions and the phonology of intonation. Phonology 26: 43–74.

Baker, C.L. 1970. Notes on the description of English questions: the role of an abstract Q morpheme. Foundations of Language 6: 197–219.

Beck, S. and H. Rullmann. 1999. A flexible approach to exhaustivity in questions. Natural Language Semantics 7: 249–298.

Bolinger, D. 1978. Asking more than one thing at a time. In H. Hiz, (Ed.), Questions, 107–150. Dordrecht: D. Reidel.

Bolinger, D. 1987. Echoes reechoed. American Speech 62: 261–279.

Brody, M. 1995. Lexico-Logical Form: A Radically Minimalist Theory. Cambridge, MA.: MIT Press.

Syntax - Morphology - Semantics 4 9 1 Major Trends in Theoretical and Applied Linguistics

Carnie, A. 2006. Syntax. A Generative Introduction. Oxford: Blackwell.

Cheng, L.L.-S. 1991. On the Typology of Wh-questions. Doctoral dissertation, MIT.

Chomsky, N. 1977. On wh-movement. In P. Culicover, T. Wasow, and A. Akmajian, (Eds.), Formal Syntax, 71–132. New York: Academic Press.

Chomsky, N. 1995. The Minimalist Program. Cambridge, MA.: MIT Press.

Denham, K. 2000. Optional wh-movement in Babine-Witsuwit’en. Natural Language and Linguistic Theory 18: 199–251.

Den Dikken, M. and A. Giannakidou. 2002. From hell to polarity: “aggressively non-D-Linked” wh-phrases as polarity ttems. Linguistic Inquiry 33: 31–61.

Fiengo, R. 2007. Asking Questions: Using Meaningful Structures to Imply Ignorance. Oxford: Oxford University Press.

Ginzburg, J. and I.A. Sag. 2001. Interrogative Investigations. The Form, Meaning and Use of English Interrogatives. Stanford, California: CSLI Publications.

Groenendijk, J. and M. Stokhof. 1984. Studies in the Semantics of Questions and the Pragmatics of Answers. Doctoral dissertation, University of Amsterdam.

Heim, I. 1994. Interrogative semantics and Karttunen’s semantics for know. In R. Buchalla and A. Mittwoch, (Eds.), The Proceedings of the Ninth Annual Conference, 128–144. Israel Association for Theoretical Linguistics.

Huang, C-T.J. 1982. Logical Relations in Chinese and the Theory of Grammar. Doctoral dissertation, MIT.

Jespersen, O. [1933]2006. Essentials of English Grammar. London: Routledge.

Jiménez, M.L. 1997. Semantic and Pragmatic Conditions on Word Order in Spanish. Doctoral dissertation. Georgetown University.

Karttunen, L. 1977. Syntax and semantics of questions. Linguistics and Philosophy 1: 3–44.

Katz, J. and P. Postal. 1964. An Integrated Theory of Linguistic Descriptions. Cambridge, MA.: MIT Press.

4 9 2 Syntax - Morphology - Semantics Anna Roussou, Christos Vlachos and Dimitris Papazachariou

Kayne, R. 1998. Overt vs. covert movement. Syntax 1: 128–191.

Kotzoglou, G. 2005. Wh-Extraction and Locality in Greek. Doctoral dissertation, The University of Reading.

Manzini, R. and A. Roussou. 2011. “Movement”: a view from the left. Paper presented at the Workshop on Parallel Domains, University of Southern California.

Manzini, R. and L.M. Savoia. 2011. Wh-in situ and wh-doubling in northern Italian varieties. Linguistic Analysis 37: 79–113.

Mathieu E. 1999. French wh-in-situ and the intervention effect. UCL Working Papers in Linguistics 11: 441–472.

Nishigauchi, T. 1990. Quantification in the Theory of Grammar. Dordrecht: Kluwer.

Parker, F. and J. Pickeral. 1985. Echo questions in English. American Speech 60: 337–347.

Pesetsky, D. 1987. Wh-in-situ: movement and unselective binding. In E. Reuland and A. ter Meulen, (Eds.), The Representation of (In)Definiteness, 98–129. Cambridge, MA.: MIT Press.

Pesetsky, D. 2000. Phrasal Movement and Its Kin. Cambridge, MA.: MIT Press.

Pierrehumbert, J. 1980. The Phonetics and Phonology of English Intonation. Doctoral dissertation, MIT.

Pires, A. and H.L. Taylor. 2007. The syntax of wh-in situ and Common Ground. Paper submitted to: P. Masullo, (Ed.), Romance Languages: Structure, Interfaces, and Microparametric Variation. Amsterdam: John Benjamins.

Richards, N. 1997. What Moves Where When in Which Language? Doctoral dissertation, MIT.

Sinopoulou, O. 2009. Απλες ερωτησεις με ερωτηματικη λεξη in situ: η περιπτωση των Ελληνικων. [Simple questions with wh-in situ: the case of Greek] Proceedings of the 8th International Conference on Greek Linguistics, 1118–1132.

Sobin, N. 1978. On echo questions in English. In D. Lance and D. Gulstad, (Eds.), Papers from the 1977 Mid-America Linguistics Conference, 247–259. Columbia: University of Missouri, University Extension Division.

Syntax - Morphology - Semantics 4 9 3 Major Trends in Theoretical and Applied Linguistics

Sobin, N. 1990. On the syntax of English echo questions. Lingua 81: 141–167.

Sobin, N. 2010. Echo questions in the Minimalist Program. Linguistic Inquiry 41: 131–148.

Sportiche, D. 1998. Partitions and Atoms of Clause Structure: Subjects, Agreement, Case and Clitics. London: Routledge.

Tsimpli, I.-M. 1990. The clause structure and word order of Modern Greek. UCL Working Papers in Linguistics 2: 226–255.

Tsimpli, I.-M. 1995. Focusing in Modern Greek. In E.K. Kiss, (Ed.), Discourse Configurational Languages, 176–206. Oxford: Oxford University Press.

Tsimpli, I.-M. 1998. Individual and functional readings for focus, wh- and negative operators: evidence from Greek. In G. Horrocks, B. Joseph and I. Philippaki- Warburton, (Eds.), Themes in Greek Linguistics 2: 197–227. Amsterdam: John Benjamins.

Uribe-Etxebarria, M. 2002. In situ questions and masked movement. Linguistic Variation Yearbook 2: 215–255.

Vergnaud, J.-R. and M.-L. Zubizarreta 2005. The representation of focus and its implications: towards an alternative account of some ‘intervention effects.’ In H. Boekhuis, N. Corver, R. Huybregts, U. Kleinhenz, and J. Koster, (Eds.), Organizing Grammar. Studies in Honour of Henk van Riemsdijk, 641–678. Paris: Mouton De Gruyter.

Vlachos, C. 2010. Wh-in situ: the case of Greek. In V. Torrens, L. Escobar, A. Gavarró, and J. Gutiérrez, (Eds.), Movement and Clitics: Adult and Child Grammar, 84–111. Cambridge: Cambridge Scholars Publishing.

Vlachos, C. 2012. Wh-constructions and the division of labour between syntax and the interfaces. Doctoral dissertation, University of Patras.

4 9 4 Syntax - Morphology - Semantics Semantic Clusters Combined with Kinematics: The Case of English and Modern Greek Motion Verbs

Marietta Sionti1, Leonardo Claudino2, Yiannis Aloimonos2, Carolyn P. Rose3, Stella Markantonatou4

1University of Athens, 2University of Maryland, 3Carnegie Mellon University, 4Institute for Language and Speech Processing/‘R.C.Athena’ [email protected], [email protected], [email protected], [email protected], [email protected]

Abstract

We combine corpus driven linguistic knowledge with experimentally obtained sensorimotor data in an effort to better specify the minimum conceptual representation of a motion event that distinguishes it from all other events, which still are largely presented in a vague and not objectively calculated quantitative method. We use American English and Modern Greek data as a case study, in order to focus on the clustering of motor actions and its correspondence to previous linguistic classifications of both languages.

1. Introduction

In the present work, we attempt to ground linguistic notions to sensorimotor data, in order to specify both linguistic and psychological terms based on embodied characteristics. Sensorimotor data are collected by measuring the performance of Modern Greek and American English native speakers. Based on these results, we attempt to create monolingual semantic classifications of verbs drawing on sensorimotor data and classify verbal semantic and syntactic properties. Moreover, the desired outcome of this work is a perceptual grounding of abstract linguistic notions, such as directionality and transitivity, which have traditionally been used in the linguistic studies of motion verbs. Robust classification serves both as an effective monolingual and crosslingual study of motion verbs of American English and Modern Greek. In the remainder of the paper, we describe the classifications of Modern Greek (Antonopoulou, 1987) and American English (Levin, 1993; FrameNet) motion verbs. We then discuss the verb collection criteria, the method for

Syntax - Morphology - Semantics 4 9 5 Major Trends in Theoretical and Applied Linguistics

the identification of the English equivalents of selected Greek motion verbs, the sensorimotor experimental procedure and its results. Implications for methodological considerations are also addressed. We conclude with our future work.

2. Motivation

2.1. Why Verb Classes?

Verb classes are well organized sets of semantically-related verbs sharing a range of linguistic properties (Levin), which frequently reflect corpus based properties. These classifications facilitate the investigation of the relations between the ‘representative element’ (eg. walk/ meander/…) of a verb class and its members, while crosslingual study of classes of verbs that are grounded on the same set of sensorimotor data are likely to reflect typological differences across languages.

2.2. Why Sensorimotor Data?

Cognitive studies have shown the relation between embodied experience and language, since image schemas are learnt as a sequence of interrelated sensorimotor patterns (Lakoff & Johnson, 1999; Arbib, 2008; Kemmerer, 2006; Fadiga et al, 2006).

2.3. Why do Theoretical Linguists Care?

Motor data of an event tend to be independent of existing divergences in its cross- linguistic representation. Consequently, a motor data representation suggests itself as a concrete and objective quantitative method for depicting cross-language lexical semantic relations. Furthermore, the distinction between “argument” and “adjunct” whose binary nature has been questioned across theories (Galen, Grenager & Manning, 2004) can probably be better understood by delineating conceptual representations of motion events in this way (both linguistic and motor).

2.4. Why do Computational Linguists Care?

NLP tasks, such as machine translation, construction of knowledge bases from text (Fahlman, 2006) and word sense disambiguation do profit from sound

4 9 6 Syntax - Morphology - Semantics Marietta Sionti, Leonardo Claudino, Yiannis Aloimonos, Carolyn P. Rose, Stella Markantonatou

semantic classifications of predicates and a solid knowledge of their syntactic properties. Moreover, the amalgam of linguistic and non linguistic characteristics, such as gestures and eye movement tracking in automatic discourse processing and other NLP applications (Eisenstein, 2008), promises increased efficiency.

3. Classification of Modern Greek Verbs

The classification of Modern Greek Motion Verbs (MGMV) (Antonopoulou, 1987) exhibits an overall semantic structure found with motion verbs of several Indo-European languages and, at the same time, present certain aspectual idiosyncrasies. In contrast, Antonopoulou (1987) adopted prototype theory as the most suitable method for the investigation for MGMV; prototype theory is by default closer to the cognitive approach adopted here. Antonopoulou’s (1987) taxonomic sets were defined with the use of two groups of criteria. The first group comprises criteria such as transitivity, causativity, agentivity, intentionality and aspect that are non-measurable with sensorimotor methods at the moment. The second group comprises change-of-location, directionality, path, dependent motion, as well as change of orientation, manner, medium and instrumentality, whose allocation on the sensorimotor data is feasible.

4. Classification of American English Verbs

Though the difference in perspective of Levin’s (1993) English Verb Classes and Framenet’s categorization is well attested (Baker & Ruppenhofer, 2002), both these classifications are important for this work. Levin’s classes are based on semantic grouping and valence alternations. Very much like Antonopoulou, Levin offers a rich anthology of verbs enriched with syntactic information that is crucial for our long term goal, namely distinguishing between argument vs. adjunct. On the other hand, Framenet’s grouping of words-according to conceptual structures can easily be matched to Antonopoulou’s second group of criteria and, finally, to sensorimotor data.

5. Sensorimotor Experiment

5.1. Verb Collection Criteria

Twenty five predicates (Table 1) were selected from Antonopoulou’s long list of MG verbs of motion and position (1987). Both intransitive and transitive verbs

Syntax - Morphology - Semantics 4 9 7 Major Trends in Theoretical and Applied Linguistics

were included because we assumed zero knowledge of verbal valence. The selected verbs fulfilled requirements imposed by the nature of sensorimotor experiments and lab limitations. Each action should be performed by one human; for instance, kalpazo (gallop) is left out because it applies only to horses, while akolutho (follow) was excluded since it violated the one participant prerequisite. We tried to identify the equivalent English verbs that would be performed in the same manner as the Greek verbs. However, we realized that the dictionary- based translation would not be the preferred one, in terms of everyday communication. English translational equivalents of Greek verbs traditionally provided by Greek-English dictionaries were either obsolete or did not align one to one with their suggested Greek equivalents. Therefore, at an early stage, we adopted the following method for identifying the English equivalents of selected Greek motion verbs: First, we videotaped the Greek participant who performed the selected set of actions. Then, we split the video into segments,

Figure 1. Answers to the multiple choice questionnaire, same color corresponds to same verb

Figure 2. Answers to the multiple choice questionnaire, same color corresponds to same verb

4 9 8 Syntax - Morphology - Semantics Marietta Sionti, Leonardo Claudino, Yiannis Aloimonos, Carolyn P. Rose, Stella Markantonatou

(as verb of

Lift/raise Pick up Pick and put on (lower)/ lower onto changing direction) March Stagger (e.g. Circle chair) Roll Step backStep Jump/hop Jump down Turn Turn (as Turn verb that cause to move in place) Pick up Pick and put on (higher)/ lift onto Stride/step over Walk Go down (ramp) Crouch Rotate Go (ramp) up Jump Jump over Turn overTurn Crawl Run Go (step) up

Go down (step) Lift high

English (FrameNet)

Motion Motion Posture Self motionSelf Self motionSelf Cause motion Change direction Cause to moveplace in Body movement not(raise included in the same group)

14 16 17

and put on (ramp) (ramp) 9

over 24 Lift/raise 18 Pick up Pick (lower)/ lower onto Walk 1 March 2 Go down (ramp) 11 Crouch 12 Rotate 23 Roll 25 Step backStep 3 Stagger 4 Go up (e.g. Circle chair) 21 Jump Jump over Jump down 15 Jump/hop 13a/13b Crawl Crawl 6 Run 7 up Pick and put on (higher)/ lift onto Go up (step) Go (step) up 8 Turn aroundTurn 20 Turn (e.g.Turn page) 22

Stride/step over5a/5b Go down (step) 10 Lift high 19 Turn Turn

English (Levin, 1993)

Go verbs Lift verbs Run verbs Run turn, turn, rotate, circle) same with the Greek Run verbsRun (+meander) two verbs expressesthe Roll verbs (around an axis Verb of Verb assuming a position Only the combination of the

5 6 7 19 24 21 22

jirizo

(epi topou) 13 (pano apo) (apo kapou) 15 (antithetic katefthinsi) (e.g. selida) (e.g.giro apo karekla)

Katevazo 16

Kateveno 10 Perpato Vimatizo 2 Pisopato 3 Triklizo 4 Aniforizo 9 Katiforizo 11 Hamilono (only with the body) 12 Anevazo 17 Sikono 18 Ipsono . Peristrefo 23 Anapodo Kilo 25 Dhraskelizo Busulo Trexo Aneveno 8 Pidho Pidho Pidho Jirizo Jirizo Jirizo

Antonopoulou, 1987) (

Greek verbs

Pidho Perpato verbs verbs (cause toturn) Anevazo verbs verbs (rotary motion) 2 1 form form its own class can thebe head of its own class)

Jirizo Jirizo Aneveno verbs (upward motion) Katevazo verbs (downward motion) Kateveno motion)verbs (downward trexo (upward motion). Though sikono can (

Verbs according to classes, same color belongs to same class same color to classes, according 1. Verbs Table

Syntax - Morphology - Semantics 4 9 9 Major Trends in Theoretical and Applied Linguistics

each of which corresponded to a single action according to Greek semantics. Next, we presented the segments to ten American English native speakers each one of which was asked to name the performed action according to his/ her intuition. Interestingly enough, the responses to seventeen videos were not homogeneous. In order to overcome this impasse, we designed a multiple choice questionnaire. Each question had three choices: two of them were the most frequent answers from the previous phase and the third one was selected from WordNet’s lemma. A new group of ten native speakers were shown each video segment before filling in the new questionnaire. Finally, we collected the frequent verbs for each action from both phases and we formed our English verb group (Table 1). These verbs were performed by eight American English native speakers-according to the following described experimental procedure- but still in case of ambiguity the participants of the sensorimotor experiment were asked to act according to their flair.

5.2. Experimental Procedure

5.2.1. Method

The detailed sensor data were analyzed to identify latent factors that represent stable patterns across the many dimensions of low level data. These factors appear as discrete sets (synergies) of joint angles and orientations associated with each action.

5.2.2. Equipment

A full body Moven system contains 16 inertial motion trackers. Each sensor module comprises 3D gyroscopes, 3D accelerometers and 3D magnetometers. Using advanced sensor fusion algorithms (Moven Fusion Engine) the inertial motion trackers give absolute orientation values which are used to transform the 3D linear accelerations to global coordinates which in turn give the translation of the body segments. The advanced articulated body model (23 segments and 22 joints biomechanical model) implements joint constraints to eliminate any integration drift or foot-sliding (X Moven Suit’s manual).

5.2.3. Phases

The sensorimotor experiment is divided in two phases; capturing of the main dataset for Greek –described in this presentation- and for English (8 subjects each) –to be done later.

5 0 0 Syntax - Morphology - Semantics Marietta Sionti, Leonardo Claudino, Yiannis Aloimonos, Carolyn P. Rose, Stella Markantonatou

Figure 3. Subject while performing an action

5.2.4. Subjects

The age range is 25-30 years old, 5 men and 3 women. All subjects were native Greek and American English speakers, so they were encouraged to implement each meaning according to their intuition.

5.2.5. Action Performance

Each verb was uttered by the experimenter and then the subject performed the corresponding action. When the verb was performed only with the body of the subject, the action was limited to a floor area restricted by a quadrangle. In order to normalize the distance, subjects were encouraged to start acting at a specific corner of the quadrangle (Fig.1). Effort was made to involve as few objects as possible. Still, the subjects asked for items that could be found in the lab: • a step (verbs 8, 10, 15), • a ramp (9, 11), • one or several balls (5, 14) • table, book, cylinder, chair (22-25) • chair (20, 21)

Syntax - Morphology - Semantics 5 0 1

Major Trends in Theoretical and Applied Linguistics

In order to standardize the procedure, the same objects were used throughout the experiments (whenever an object was required).

5.2.6. Analysis

In the raw data matrix, each verb is related with one row only, while each row is a set of uniformly sampled poses. For one subject, individual poses are sets of joint angles for all J joints measured by the Xsens Moven suit, with the angles being described in terms of directions x, y and z: data_matrix= { verb_1; verb_2; ... verb_M} where each row is: verb_v = [ pose_1, pose_2, ..., pose_T ] and each pose is horizontally expanded as: pose_t: [angle_joint_1_x(t), angle_joint_1_y(t), angle_joint_1_z(t), ... , angle_ joint_J_x(t), angle_ joint_J_y(t), angle_joint_J_z(t)] considering that there are J joints (e. g. head, elbow, knee, ...) and 0<=t<=T-1. When more subjects are considered, each verb (row) is computed as the average performance across all subjects, after length normalization. Principal Component Analysis (PCA) is computed on that matrix as in (Santello, 1998), so the two eigenvectors used for projection reflect 3-D and time information, allowing at the same time a two-dimensional visualization of the action scatter (Fig.3, 4). For Fig. 5-10 the matrix is (considering N performing subjects): verb_v_data_matrix= { subj_1;subj_2; ... subj_N}, each subject’s performance is subj_s = [ pose_1, pose_2, ..., pose_T΄] and each pose is horizontally expanded as in the overall data matrix (described

x Blue Y Green Z Red

Sagittal plane Blue Frontal / coronal plane Green Transverse plane Red

Figure 4. X-Moven Suit’s axes and body planes (from manual)

5 0 2 Syntax - Morphology - Semantics

Marietta Sionti, Leonardo Claudino, Yiannis Aloimonos, Carolyn P. Rose, Stella Markantonatou

above), but with 0<=t<=T΄-1. Here, T΄ is the number of time samples uniformly selected from each participant’s performance of verb v. T΄ is not necessarily the same as the T defined in the previous matrix, which was related to average verb performances instead.

5.3. Methodological Considerations

We preferred the PCA statistical approach on the grounds of objectivity and familiarity with the linguistic methodology, as opposed to mathematical models and optimization techniques used for similar research. Moreover, the correct use of necessary objects received much consideration. We avoided the usage of several different objects because the posture of the hand, the arm and the body would change the outcome significantly, e.g. when grasping. Therefore we used the most unmarked objects and the same objects were used by all subjects. In this way we standardized the procedure without focusing to the objects used. Ideally, we would like to capture every verb with a relatively large set of the objects that are denoted by the verb’s (syntactic) dependents; this is left to future research.

6. Results

Concerning the two following types of plots for all verbs: Fig 5, 7. displays the result of projecting each average verb in the previously defined data_matrix onto the first two eigenvectors computed by PCA on that matrix. In other words, each numbered dot corresponds to a mapping of a verb’s time sequence of poses into a single 2-D point. Fig.6, 8 illustrate the first 10 eigenvectors computed by PCA (rows) in terms of the weights it assigns to each joint-direction-time. Each small rectangle corresponds to a specific joint-direction (e. g., head-x), each column in that rectangle refers to a t-time instant, 0<=t<=T-1. If a column is bright at time t within the head-x rectangle, it means that there is something that the testing subjects do with the head at that direction at time t, which varies significantly among the testers and thus is picked up by PCA. Following that reasoning, by inspecting sets of joint-directions highlighted by the main principal components, one can draw a better understanding of why the scatter at Fig. 5 and 7 appear in this form.. The two main visual groups of actions are: • the rectangular: it includes walking like actions (leg-related) • the blue ellipsis: it includes manipulation of an object (arm-related) (Fig.5, 7). It is projected on both PC1 (distinct use of hips) and PC2 (emphases on knees and shoulders) (Fig. 4). Therefore, all these actions have approximately the same profile in terms of joints-angles.

Syntax - Morphology - Semantics 5 0 3 Major Trends in Theoretical and Applied Linguistics

Actions 6 (busulo-crawl) and 7 (trexo-run) are relatively apart from the leg- related group. However, they appear to be virtually projected on the same coordinate on Principal Component 1 (PC1), the one that relates to hips x (displacement on the sagittal plane). At the same time they are distant when they are projected on PC2, which involves arms as well. This is only natural because when a subject crawls, it moves its arms, hips and legs more than in actions of the green ellipsis. Action 7 (trexo-run), on the other hand, seems to be closer to the rectangular group because the displacement of the above mentioned parts encloses more energy than the rest leg-related actions but still less than action 6 (busulo-crawl). Furthermore, the fact that run-trexo and crawl-busulo are projected on different PC2 narrows down to (i) the height of the knees (upward- downward and forward-backward respectively) and (ii) the frequent movement

Figure 5. Average Actions Greek Verbs ( the numbers are explained in Table1)

Figure 6. Weights of each joint-direction-time feature according to the 10-first principal components, including time, for the Greek language)

5 0 4 Syntax - Morphology - Semantics Marietta Sionti, Leonardo Claudino, Yiannis Aloimonos, Carolyn P. Rose, Stella Markantonatou

of shoulders in all directions. Action 7 differs from the rest in terms of velocity and this is incorporated in the depiction of Time in the plot. The visualization presented here supports the results of brain imaging studies; the schematic of the distributed semantic representation in the brain of action verb processing is based on the body parts performing them (Wermter et al., 2005), e.g. arm-related and leg-related. Comparing this visualization with linguistic classifications (Antonopulou, 1987; Levin,1993; FrameNet) we see similarities. Antonopoulou’s classes are more fine-grained than the two big categories in Fig.5; here two ofthe Antonopoulou’s classes appear in the same scheme –rectangular or ellipsis. In particular, the members of the walk-group, pidho-group and kateveno-group are coiled together (green ellipsis). The same occurs for jirizo2-group, katevazo and

Figure 7. Average Actions English Verbs (the numbers are explained in Table1)

Figure 8. Weights of each joint-direction-time feature according to the 10-first principal components, including time, for the English language

Syntax - Morphology - Semantics 5 0 5 Major Trends in Theoretical and Applied Linguistics

anevazo group (blue ellipsis). Levin’s and FrameNet’s classes are similar to the groups at Fig.7. Of particular interest is the fact that the actions 13, 14, and 15 (pidho- jump) share the same morphological representation pidho but not the same performance; eventually they form one tight ‘group’. Actions 20, 21 (jirizo1) and 22, 23, 24, 25 (jirizo2) also share the same morphological representation jirizo. These actions present themselves close to each other on the PC1 projection but are separated on the PC2 projection. The first group is mainly about leg motion and the second one about object manipulation that necessarily involves arm displacement as is clearly indicated on PC2. In English all roll verbs belong to the same class. Though the plots were based on similar or even identical actions, certain divergences occurred. For instance, the verb dhraskelizo has traditionally been translated as stride, since both share longer steps. The Greek subjects always needed a small obstacle, such as a ball or a hole on the ground, to perform longer steps. On the contrary, the English subjects clearly distinguished between stride and step over (although WordNet assigns this meaning to stride as well). Similarly, we would expect vimatizo to be closer to pace rather than march, but it should be noted that the majority of the English participants were unsure for the exact representation of pace. When we compare the signals of vimatizo and march, we see significant similarity in the manner of stepping, while English subjects also emphasize the movement of the arms. Of particular interest in the Greek plot is the distance of trexo-run and busulo- crawl from the rest leg-related actions (especially for English, march and crawl show the same behavior). Although we would expect run to resemble trexo, differences occurred due to two reasons: (i) the English subjects tented to use their arms less than the Greek ones, and (ii) each group performed march in a different way; this time, the English subjects used their arms more than the Greek ones (as opposed to the performance of run). The blue and green ellipsis are projected both on the same PC1 (highlighted hips) and 2 (highlighted knees and shoulders). Furthermore, the reason why march, trexo and crawl-busulo are projected on different PC2 narrows down to (i) the height of the knees (upward-downward and forward-backward respectively) and, (ii) the frequent movement of shoulders in all directions. But still, these verbs are projected on the same PC1. PC1 is about the forward and backward displacement of the hips and emphasizes on the leg related actions, namely the walking like actions, which, in turn, is considered to be the actions’ common linguistic characteristic. The above findings are still consistent with the aforementioned linguistic analysis. The Greek trexo can form its own class in terms of velocity according to (Antonopoulou, 1983). At the same time, in Fig.5, 7, action 7 differs from all the other actions performed because of velocity. This fact is incorporated in the

5 0 6 Syntax - Morphology - Semantics Marietta Sionti, Leonardo Claudino, Yiannis Aloimonos, Carolyn P. Rose, Stella Markantonatou

depiction of time in Fig. 6. Probably Levin (1993) gives us a hint that these verbs need special treatment, since she enlists them under both the meander verbs and the run class (it must be kept in mind that Levin’s classification has taken into account the criterion of intentionality; however, intentionality is still not measurable with sensorimotor techniques).

7. Conclusions and Current Directions

In this paper we have presented work to date on corresponding linguistics terms -extracted from classifications- to sensorimotor clusters. Specifically, we explored their differences and resemblances based on totally unsupervised methods. We demonstrated the difficulties in aligning Greek and English motion verbs, while our future work aims to extend this approach. The analysis of each motion verb and the combination of detailed linguistic and motor entries that would appear as the verb’s “identification card” is feasible and helpful for sound semantic classifications.

Acknowledgement

We are grateful to reviewers and audience for their feedback at the 20th International Symposium on Theoretical and Applied Linguistics in Thessaloniki (2011), the Embodied and Situated Language Processing Workshop in San Diego (2010), and the Interdisciplinary Workshop on Verbs, the Identification and Representation of Verb Features in Pisa (2010). Moreover, we would like to thank Nathan Schneider, Lori Levin and Ben Lambert for their encouragement and feedback. This research was partially supported by the Greek Scholarships Foundation (IKY).

Syntax - Morphology - Semantics 5 0 7 Major Trends in Theoretical and Applied Linguistics

References

Antonopoulou, E. 1987. Prototype theory and the meaning of verbs, with special reference to Modern Greek verbs of motion. Unpublished PhD thesis. University of London

Arbib, M.A. 2008. From Grasp to Language: Embodied Concepts and the Challenge of Abstraction. Journal of Physiology Paris 102: pp. 4-20

Baker, C.F and Ruppenhofer, J. 2002. FrameNet’s frames, vs, Levin’s verbs classes. 20-09-2010

Clark, R.A., Miller, J.M. and Demer, J.L. 2000. Three-dimensional location of human rectus pulleys by path inflections in secondary gaze positions. Invest. Ophthalmol &Visual Sci. 41

Fadiga, L., et al. 2006. Language in shadow. Social Neuroscience, 1(2): pp.77-89.

Feldman, J. 2006. From Molecule to Metaphor. Cambridge, MA: MIT Press.

Framenet http://framenet.icsi.berkeley.edu/

Galen, A., Grenager. T. and Manning C. 2004. Verb Sense and Subcategorization: Using Joint Inference to Improve Performance on Complementary Tasks. EMNLP 2004: pp. 150-157.

Jackendoff, R. 1990. Semantics Structures. Cambridge, MA: MIT Press.

Kemmerer, D. 2006. Action verbs, argument structure constructions, and the mirror neuron system. In M. Arbib (Ed.), Action to language via the mirror neuron system. Cambridge University Press.

5 0 8 Syntax - Morphology - Semantics Marietta Sionti, Leonardo Claudino, Yiannis Aloimonos, Carolyn P. Rose, Stella Markantonatou

Lakoff, G. and Johnson, M. 1999. Philosophy in the Flesh: The Embodied Mind and Its Challenge to Western Thought. Basic Books, New York

Levin, B. 1993. English Verb Classes and Alternations: A Preliminary Investigation, University of Chicago Press, Chicago, IL.

Papafragou, A., Massey, C. and Gleitman, L. 2006. When English proposes what Greek presupposes: The cross-linguistic encoding of motion events. Cognition 98: B75-87

Santello, M., Flanders, M. and Soechting, J.F. 1998. Postural hand synergies for tool use. Journal of Neuroscience, 18.

Troje, N.F. 2002. Decomposing biological motion: A framework for analysis and synthesis of human gait patterns. Journal of Vision, 2: pp. 371–387

Wermter, S., Weber, C., Elshaw, M., Gallese, V. and Pulvermüller, F. 2005. Grounding neural robot language in action. Biomimetic Neural Learning for Intelligent Robots. Springer

Syntax - Morphology - Semantics 5 0 9 Major Trends in Theoretical and Applied Linguistics

5 1 0 Dativus Ethicus in the Balkan Languages

Ekaterina Tarpomanova

Institute for Bulgarian Language, Bulgarian Academy of Sciences [email protected]

Abstract

The paper presents a contrastive study of the ethic dative case of the personal pronouns in the languages of the Balkan Sprachbund outlining the similarities between them, but also discussing peculiarities of each language. It aims at analyzing the functioning and the meaning of different pronominal forms in dativus ethicus (1st person singular or plural, 2nd person, 1st + 2nd person of the personal pronouns, reflexive pronoun) found in authentic examples in order to indicate that dativus ethicus is one of the common features of the pronominal systems of the Balkan languages.

1. Introduction

The common features of the languages of the Balkan Sprachbund (Bulgarian, Greek, Albanian and Romanian) have been widely discussed in the linguistic literature. Since the first generalizing work on Balkan linguistics of the danish linguist Kristian Sandfeld (1930) was published, a number of authors have analyzed linguistic phenomena at all the language levels shared by the Balkan languages, such as unification processes of the phonological systems, development of analytic noun systems, syncretism of genitive and dative, future tense formation, lack of infinitive, postposition of the definite article, clitic doubling, etc. Despite some critics that Balkan linguistics has already exhausted its research field or that there is a danger to “break the structural-typological matrix” of a language if it’s only seen as being a part of a Sprachbund (as in Reiter 1994), the interest toward Balkan linguistic peculiarities doesn’t seem to have decreased. Furthermore, a number of monographs on Balkan linguistics appeared only in the last 20 years, analyzing and systematizing the common features of the Balkan languages (Steinke & Vraciu 1999, Asenova 2002, Demiraj 2004, Feuillet 2012, among others). Traditionally the focal point of research in Balkan linguistics are the common morpho-syntactic phenomena of the Balkan languages. Pronominal systems

Syntax - Morphology - Semantics 5 1 1 Major Trends in Theoretical and Applied Linguistics

haven’t been a center of interest, although they show multiple similarities. This paper presents a synchronic contrastive study of a function of the pronouns’ dative case called dativus ethicus in the modern Balkan languages. It aims at studying the usage of the ethic dative in all the forms found in the Balkan languages on the basis of authentic examples extracted from fiction and internet. The paper will show that dativus ethicus is not only preserved in those languages, but also has a high frequency and identical usages in colloquial speech.

2. Dativus Ethicus in Indo-European Languages

Dativus ethicus is a function of the personal pronouns in dative with an ancient origin in the Indo-European languages. Forms of the ethic dative are encountered in Old Greek and Latin, indicating that the person in dative has a special interest towards the person or the thing spoken (Schwyzer 1950). It is alive in many of the present-day Indo-European languages: in French – Prends-moi le bon parti. ‘You should do the right thing.’; Regarde-moi ça! ‘Look at this!’; in English the construction with the preposition on + a personal pronoun in any person and number is encountered – They’ve gone and bought that bike on us/ on you/ on me/ on him/ on them.; in Dutch – Wat doe je me daar nou? ‘What’s going on with you now?’, Waar kom je me nou mee aanzelten? ‘What are you going to do now?’; in German – Comm du mir bloss nachhause. ‘Just come into my house.’ (DuFeu 1998). The ethic dative is also found in some of the Slavic languages (mainly West and South Slavic) – Czech, Polish, Slovak, Serbo-Croatian, Bulgarian, and in the languages of the Balkan Sprachbund – Bulgarian, Greek, Romanian and Albanian. It occurs with different frequency and distribution in different languages, for instance in French, although quite current, it is restricted to the 1st person only, especially after a verb in imperative. Fischer (1997:79) mentions that the ethic dative causes some particular problems in German-to-English translation, not having a direct counterpart in English. In this inhomogeneous situation, the Balkan languages manifest impressive similarities which can be considered to be due to the convergence processes in their development.

3. Dativus Ethicus in the Balkan Folklore

A thorough analysis of the origin and functions of dativus ethicus in light of the Balkan model of the world is made by Tsivyan (Tsivyan 1999:91-103). The author sees dativus ethicus as a result of the fusion of the genitive and the dative case and the capacity of the pronominal clitics to express both a direction of an action and possession. Therefore, dativus ethicus in the Balkan languages is bound up

5 1 2 Syntax - Morphology - Semantics Ekaterina Tarpomanova

with the category of possessivity. However, this is a possession of a special kind – the speaker is situated inside the statement declaring his presence in the text. Tsivyan calls this act of appropriation “a linguistic gesture in a self-direction”. Her research is based exclusively on folklore texts. The author underlines that in the folklore, dativus ethicus does not have so much an emotional value, but it emphasizes the presence of the speaker in the world of the text by presenting it in the context of the verbal category of evidentiality.

4. Dativus Ethicus in the Modern Balkan Languages

Besides the Balkan folklore, dativus ethicus is also widely spread in the modern languages of the Balkan Schprachbund. However, in contemporary speech, in contrast to folklore texts, it is a means of the speaker to express his subjective evaluation of the situation. In Bulgarian, Romanian and Albanian, the forms of 1st person singular and 1st person plural are mostly used, while the 2nd person plural is rarely used in Bulgarian and Romanian. A frequent usage in modern Bulgarian has the short form of the reflexive pronoun. In Greek, the ethic dative has a restricted usage as compared with the other Balkan languages, which is compensated by the frequent usage of the possessive pronominal clitics, as shown by Tsivyan (Tsivyan 1999:96-98).

4.1. 1st Person of Dativus Ethicus

The 1st person expresses most distinctly “the intervention” of the speaker in the utterance by giving emotional nuances of the speech and by expressing the subjective evaluation of the speaker. In all the Balkan languages it is made by the 1st person singular, in Albanian in addition the 1st person plural is used by which the speaker presents himself as a representative of a multitude.

(1a) Bulg. Какъв си ми хубав! What are me:DE beautiful ‘You are so beautiful to me!’ (1b) Bulg. Марче, вземи се в ръце, бе момиче. Я какво си ми младо и слабичко, а аз какво да кажа с моите 85 килограма. Marche take yourself in hands part. girl part. what are me:DE young and thin and I what to say with my 85 kilograms ‘Marche, take it easy, girl. Look how young and thin you are, what should I say with my 85 kilos.’

Syntax - Morphology - Semantics 5 1 3 Major Trends in Theoretical and Applied Linguistics

(1c) Gr. Μήπως η καρδούλα σου δεν μου είναι ευτυχισμένη. Maybe the heart your not me:DE is pleased ‘Maybe your heart is not pleased?’ (1d) Gr. Θέλω να μου είσαι καλά και να περνάς όσο πιο καλά γίνεται. Want-I to me:DE be-you well and to spend-you as more well happens ‘I want you to be well and to spend your time as well as possible.’ (1e) Rom. Aici m-ai fost? (I. Creangă, GLR: 150) Here me:DE have-you been ‘So, there you are.’ (1f) Alb. Lulushen, të ma keni kujdes. (D. Agolli) Lulusha:Acc to me:DE take care ‘Take care of Lulusha.’; (1g) Alb. O bo bo, kush më qenka në këtë forum, moreee... Well, well who me:DE is in this forum part. ‘Well, well, look who was in this forum.’

In the examples above the speaker expresses his positive attitude towards the object of the speech by using dativus ethicus. In such cases it can be easily transformed in a possessive pronoun – my, my dear: Take care of Lulusha. > Take care of my (dear) Lulusha. This is a typical way of expression of a positive evaluation by appropriation or by affiliation, a means of the speaker to admit the object in his own world. Dativus ethicus can also express a negative attitude – neglect, irony or mockery:

(2a) Bulg. Ооо, тя ще ми се труди... Oh she will me:DE work ‘So, she will work.’ (2b) Bulg. Какво си ми се разциврила! What me:DE are crying ‘What are you crying!’ (2c) Rom. Ăsta îmi eşti tu! This me:DE are you ‘So, you are that kind of man!’ (2d) Alb. Na nxjerrka ai libra për bujqit! (D. Agoll) Us:DE takes out he books for the farmers ‘So, he quotes books for farmers!’

In (2a, b, c, d) a false gesture of appropriation takes place in order to express the speaker’s negative attitude towards the situation or the object spoken. This is a transpositive usage of dativus ethicus which transforms the speaker’s evaluation from positive into negative thus increasing the expressiveness of the utterance.

5 1 4 Syntax - Morphology - Semantics Ekaterina Tarpomanova

In Bulgarian and Albanian, where the verb category of evidentiality is grammaticalized, often a parallel usage of dativus ethicus and admirative verb forms is encountered. The admirative in Bulgarian and Albanian is defined as a sudden ascertainment of facts which is unexpected and surprising for the speaker and provokes a particular emotional condition (Gerdzhikov 1984:101). The expressiveness of the verbs in admirative provides a possibility for the speaker to give voice to his subjective evaluation of the situation ascertained. This is a secondary and contextually dependent function of the admirative, but quite frequent in both languages (see Tarpomanova 2010). Adding a pronoun in ethic dative the speaker gives emphasis on the subjectivity of his own appraisal.

(3a) Bulg. И това ми било мъж! And this me:DE is:adm. man ‘So, that would be a real man!’ (3b) Bulg. Изтъпанчил ми се на онзи големия баир в Минск и гради първото иглу за сезона. Stands:adm. me:DE on that big hill in Minsk and builds the first igloo for the seazon ‘He stands on the big hill in Minsk and builds the first igloo for the season.’ (3c) Alb. E, domethënë Sherif Abeceja na qenka edhe besimtar! (D. Agolli) Eh so Sherif Abeceja us:DE is:adm. also believer ‘So, Sherif Abeceja is also a believer!’

In (3a, b, c) the pronominal and the verbal form mutually intensify their meanings, expressing the emotional attitude of the speaker and his pragmatic evaluation – positive or negative, of the object of the utterance.

4.2. 2nd Person of Dativus Ethicus

The 2nd person of dativus ethicus aims to integrate the collocutor into the situation described by the utterance, as if he was an eyewitness of the events. The opposition mine/not mine takes place between the 1st and the 2nd person, on one hand, and the 3rd person, on the other. A translocation occurs in the internal correlation me/you – the speaker gives the possibility of the collocutor to express the emotional attitude of the utterance. According to Bokshi, the pronominal clitics in 2nd person should not be considered as dativus ethicus, being a neutral form for expressing the person, or an “all-personal” form. He claims that the emotional value of ethic dative can only be expressed by the 1st person, because the speaker is a concrete person, while the neutral you may be either me, you or him/her (Bokshi 2004:79-81). Bokshi’s opinion is not sustained here and a distinction is made between the 2nd person of the personal pronouns expressing impersonality (all-personality, according to Bokshi) and the 2nd person of the

Syntax - Morphology - Semantics 5 1 5 Major Trends in Theoretical and Applied Linguistics

pronominal clitics in dativus ethicus. The differences between the two meanings can be observed in the examples below. In (4a) the neutral function of the 2nd person pronoun is notable. The elimination of the pronoun would damage the sentence structure and meaning. In (4b) the pronoun expresses the speaker’s surprise and admiration and if it’s removed the speaker’ attitude will be lost, but the sentence grammaticality and meaning will be preserved.

(4a) Alb. Ti i bën mirë, ai ta kthen me të keq. You him do good he you returns bad ‘You do him good, he returns you evil.’ (4b) Alb. Pa t’u bë ai fiku sa një shtëpi. And you:DE became this fig-tree like a house ‘And the fig-tree became as big as a house.’

4.3. Parallel Usage of 1st and 2nd Person

In Bulgarian, Romanian and Albanian a parallel usage of the 1st and 2nd person pronouns in dativus ethicus is possible. In Bulgarian this usage is restricted – the two pronouns can be in enclitic position only after demonstrative (5a), interrogative (5b) and indefinite (5c) pronouns, the latter expressing pragmatic evaluation, positive or negative, intensified by the dative pronominal clitics.

(5a) Bulg. Надолу очи да метнеш, кипнала оная ми ти земя, напращели ония младоци, надули се ония ми ти пъпки по фиданките... (Chudomir) Down eyes to throw-you boiled that me:DE you:DE land burst those young people grew those me:DE you:DE buds on the trees ‘If you look down, you will see this land boiling, these young people bursting, these buds growing.’ (5b) Bulg. Каква Европа, какви ми ти там хелзинкски комитети и други подобни плутократски организации. What Europe what me:DE you:DE there Helsinki Comittees and other similar plutocratic organizations ‘What Europe, what Helsinki Committees and similar plutocratic organizations.’ (5c) Bulg. Тъкмо се настаних удобно, като се накачиха във влака едни ми ти цигани, пардон, роми. Just sat-I down comfortably when got on the trains some me:DE you:DE gypsies pardon roma ‘I just sat down comfortably when some gypsies, ops… Roma people, got on the train.’

5 1 6 Syntax - Morphology - Semantics Ekaterina Tarpomanova

In Romanian and Albanian the parallel usage of the 1st and the 2nd person of the pronouns in dativus ethicus is not submitted to usage restrictions as in Bulgarian.

(5d) Rom. Mi-au lunecat ciubotele şi am căzut în Ozana, cât mi ţi-i băietul! To me slipped the boots and fell-I on Ozana what me:DE you:DE man ‘My boots slipped and I fell on Ozana, what a man!’ (5e) Rom. Ce credeţi că mi ţi văzu? (GLR:150) What think-you that me:DE you:DE saw-I ‘And what do you think I saw?’ (5f) Alb. E të na u bë baltë e madhe, e të na u bënë, or të keqen këmbët gjak. (Bokshi 2004:80) And you:DE us:DE became mud big and you:DE us:DE became badly blood ‘And everything became mud and our feet became all in blood.’

As shown in the exemples (5a, b, c, d, e, f), the parallel usage of the 1st and the 2nd person of the pronouns acts as intensifier of the semantic property of the word modified which can be explicit, as in (5f) “really big mud”, or implicit, as in (5a) “really (beautiful) land”, in (5c) “really (awful) gypsies”, and in (5d) “what a (handsome) man”. Besides, the combination of the pronouns can modify the whole meaning of the utterance by increasing its emotional value, as in (5e) “And what do you think I saw? (It’s really incredible!)”

4.4. Combination of Dativus Ethicus with Deictic Particles

For Bulgarian Nitsolova mentions the usage of pronouns in dativus ethicus with deictic particles with predicative function (Nitsolova 1986:57). The case in point is the combination of the 2nd person of the pronominal clitics with the particles here, there. This usage is characteristic also of Greek and Romanian and may be considered as an abstract gesture of giving on the part of the speaker. The meaning of the whole construction is a negative issue of some situation that is previewed from the speaker.

(6a) Bulg. На ти сега социализъм, на ти митинг! There you:DE now socialism there you:DE demonstration ‘There you have your socialism, there you have your demonstration!’ (6b) Gr. Δεν είχαν περάσει ούτε 3 μήνες από τη μέρα που είχε βάλει την Not had passed even 3 months from the day that had-he put the υπογραφή του στο παχυλό συμβόλαιό του και να σου τώρα καμάρωνε signature his on the pompous council and there you:DE now boasted- he εμπρός από το νέο δημιούργημα του. in front of the new creation his ‘Less than 3 months have passed since he put his signature under the pompous council and there you go, he’s boasting with his creation.’

Syntax - Morphology - Semantics 5 1 7 Major Trends in Theoretical and Applied Linguistics

(6c) Rom. Iată-ţi gloria de veacuri! There you:DE glory for centuries ‘There you have your century-old glory!’

4.5. The Reflexive Pronoun in Dativus Ethicus

In Bulgarian a frequent usage in dativus ethicus has the dative reflexive pronounsi , which is a Slavic characteristic in this language. Among the other Balkan languages only in Romanian a dative reflexive pronoun exists, but its usage in dativus ethicus is marginal and it is encountered only in the folklore (Câmpeanu 1997):

(7) Rom. Toţi de Corbea că mi vorbea şi din gură mi-şi zicea. All from Corbea that to me spoke and from mouth to me refl.pron.:DE told ‘All the people from Corbea spoke to me and with their mouths told me.’

In Bulgarian the reflexive pronoun in dativus ethicus means that the action is performed for pleasure or for whim or expresses intimacy (Nitsolova 1986:68):

(8a) Bulg. Гледам си в тавана, по който мухите пълзят, и хоп, ето ти тема: „Защо мухите не падат, когато са с главата надолу?” Look-I refl.pron.:DE at the ceiling on which flies creep and up there is theme why the flies don’t fall when are upside-down ‘I was looking at the ceiling, where the flies were creeping, and suddenly a theme submerged: “Why the flies don’t fall down being upside-down?” (8b) Bulg. Кучетата си лаят, керванът си върви. The dogs refl.pron.:DE bark the caravan refl.pron.:DE goes ‘The dogs bark but the caravan goes on.”

5. Conclusion

Dativus ethicus is a common feature of the Balkan languages, an Indo- European inheritance preserved and developed in the modern languages, which contributes to the expressiveness of speech and is one of the means to articulate a pragmatic evaluation of a given situation, event or the object of the utterance. Its usage, as shown in the examples in this paper, is characteristic of colloquial speech, making it bright and vivid.

5 1 8 Syntax - Morphology - Semantics Ekaterina Tarpomanova

References

Asenova, P. 2002. Balkan Linguistics. Sofia. (In Bulgarian)

Bokshi, B. 2004. On the Personal Pronouns of Albanian. Prishtinë. (In Albanian)

Câmpeanu, E. 1997. Stylistics of the Romanian Language. Cluj-Napoca. (In Romanian)

Demiraj, S. 2004. Balkan Linguistics. Tirana. (In Albanian)

DuFeu, V. 1998. The Dativus Ethicus (DE) in the Slavonic Languages. – XII MKS KRAKÓW 1998. Retrieved from: http://www.arts.gla.ac.uk/Slavonic/VDuFeu.htm

Feuillet, J. 2012. Linguistique comparée des langues balkaniques. Paris.

Fischer, K. 1997. German-English Verb Valency: A Contrastive Analysys. Tübingen.

Gerdzhikov, G. 1984. Renarration of the verb action in Bulgarian. Sofia. (In Bulgarian)

GLR: Grammar of the Romanian Language, vol. 1. Bucarest, 1966. (In Romanian)

Nitsolova, R. 1986. Bulgarian Pronouns. Sofia. (In Bulgarian)

Reiter, N. 1994. Grundzüge der Balkanologie. Ein Schritt in die Eurolinguistik. Berlin – Wiesbadden.

Sandfeld, K. 1930. Linguistique balkanique. Problèmes et résultats. Paris.

Schwyzer, E. 1950. Griechische Grammatik. München.

Syntax - Morphology - Semantics 5 1 9 Major Trends in Theoretical and Applied Linguistics

Steinke, C. & A. Vraciu. 1999. Introduction in Balkan Linguistics. Iasi. (In Romanian)

Tarpomanova, E. 2010. The Admirative in Bulgarian and Albanian. In: Balgarski ezik, 57 (2010), 1: 45-66. (In Bulgarian)

Tsivyan, T.V. 1999. Motion and Road in the Balkan Model of the World. Moscow. (In Russian)

5 2 0 Syntax - Morphology - Semantics