Edited by: Nikolaos Lavidas Thomaï Alexiou Areti-Maria Sougari

Major Trends in Theoretical and Applied Linguistics Volume 2 Versita Discipline: Language, Literature

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.

Copyright © 2013 Selection and editorial matter: Nikolaos Lavidas, ThomaÏ Alexiou, Areti-Maria Sougari; individual contributors, their contributions.

ISBN (paperback): 978-83-7656-086-1

ISBN (hardcover): 978-83-7656-087-8

ISBN (for electronic copy): 978-83-7656-088-5

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 second volume is organized into three main sections: Discourse Analysis – Gender – Lexicography; Language Acquisition; Language Disorders. Contents

Part II Selected Conference Papers

Section 3: Discourse Analysis - Gender - Lexicography

Maria Christodoulidou The Meaning of Discourse Markers in Classroom Interaction...... 11

Angeliki Efthymiou, Zoe Gavriilidou and Eleni Papadopoulou Labeling of Derogatory Words in Modern Greek Dictionaries ...... 27

Zoe Gavriilidou Users’ Abilities and Performance in Dictionary Look Up...... 41

Μαριάνθη Μακρή-Τσιλιπάκου Σεξουαλικός προσανατολισμός και γλώσσα...... 53

Δημήτρης Ντανόπουλος Αναλύοντας τις διαδρομές των κειμένων: Εθνογραφική μελέτη παραγωγής ειδήσεων σε μια τοπική εφημερίδα...... 69

Andreas Papapavlou and Melanie Satraki Dialect and Foreign Language Features in Greek-Cypriot Advertising: Investigating Factors Influencing Recipients’ Attitudes...... 85

Theodossia-Soula Pavlidou and Angeliki Alvanoudi Grammatical Gender and Cognition...... 109 Angelina Sophiadi The Song Remains the Same… or Not? A Pragmatic Approach to the Lyrics of Rock Music ...... 125

Alexandros Tantos Verbs and Events in Discourse: A Synergy ...... 143

Efthymia Tsaroucha Do We Always Speak of What We Mean? (Non)Referring Expressions in Discourse...... 153

Section 4: Language Acquisition

Eleni Agathopoulou “Automatically Arises the Question Whether…”: A Corpus Study of Postverbal Subjects in L2 English ...... 169

Fotini Anastassiou and Georgia Andreou Multilingualism in Greece: A Study of Speech Production of Trilingual Children...... 187

Evangelia K. Asproudi Non-Target Long-Distance Wh-Questions: Crosslinguistic Typological Distinctions in Early L1 Production...... 199

Styliani S. Kalligiannaki and Marina Tzakosta Common Characteristics in Compound Formation: Evidence from Bilingual Acquisition and L2 Language Learning...... 219

Maria Kaltsa The Production of Aspect by L1 Learners of Greek...... 233

Νικόλαος Καρατσιώλης και Πηνελόπη Καμπάκη-Βουγιουκλή Η ασαφής περιοχή της επίγνωσης της διαγλωσσικής παρεμβολής μεταξύ ιταλικής και ισπανικής από Έλληνες...... 247

Sviatlana Karpava and Kleanthes K. Grohmann Greek Embedded Aspect in Child and Adult L2/L3 Acquisition...... 267

Maria Kotroni The Acquisition of Motion Verbs in Greek: Evidence from Comprehension .....289 Monique Monville-Burston and Maria Kounouni Advanced Learner Competence and Relative Clauses in Oral Narratives...... 313

Theoni Neokleous and Teresa Parodi Clitic Placement in Early Grammars: The Case of Cypriot Greek...... 331

Elena Papadopoulou and Natalia Pavlou Specificities of Syntactic Movement in Early Speech...... 349

Alexandra Prentza Pronominal Subjects in English L2 Acquisition and in L1 Greek: Issues of Interpretation, Use and L1 Transfer...... 369

Ianthi Maria Tsimpli, Maria Papakonstantinou and Ageliki Nicolopoulou Linguistic Structure and Narrative Quality in Child L1 Production...... 387

Section 5: Language Disorders

Maria Kambanaros Modality-Specific Noun–Verb Dissociations in L1 but No Effect in L2: Evidence from Bilingual Aphasia ...... 407

Maria Kambanaros, Kleanthes K. Grohmann, Eleni Theodorou and Michalis Michaelides Can Vocabulary Size Predict Narrative Abilities in Children with SLI?...... 423

Michaela Nerantzini, Despina Papadopoulou and Spyridoula Varlokosta Testing the Relativized Minimality Approach. Evidence from Wh-Question Production and Comprehension in Greek Aphasia ...... 437

Eleni Peristeri and Ianthi Maria Tsimpli Linguistic Processing and Executive Control: Evidence for Inhibition in Broca’s Aphasia...... 455 8 Section 3: Discourse Analysis - Gender - Lexicography

9 Major Trends in Theoretical and Applied Linguistics

1 0 The Meaning of Discourse Markers in Classroom Interaction

Maria Christodoulidou

Frederick University [email protected]

Abstract

There hasn’t been much research carried out to date relating the spoken classroom discourse and the correlation between meaning and interaction. In this study, I will explore the relationship between meaning and interaction as enacted by the Greek particles lipon (“so”) and endaksi (“ok”), and, in particular, I will investigate their use within classroom interaction using the tools of the conversation analytic tradition. This study investigates these lexical particles within classroom interaction in pre-primary and primary classes in schools of Nicosia. For the investigation I recorded and transcribed 20 hours of classroom interaction. The lexical items under study fit into the category of what has traditionally been framed as discourse markers. In this study I centre my attention on the classroom genre and analyze the communicative purpose of the aforementioned discourse markers in the speech of the teachers.

1. Introduction

As the main medium of instruction, classroom genres have aroused the interest of researchers. To date, most studies of classroom language have shown that classroom conversations led by the teacher and involving the whole class typically have large structural junctures that delimit lessons and tasks, and phases within them (e.g., Bellack et al., 1966; Erickson 1978; Mehan 1979; Sinclair & Coulthard 1975). Recent studies have explored a variety of discourse markers in a number of settings, but pre-primary and primary classrooms have been largely overlooked. The main purpose of this study is to analyse the occurrence of two lexical markers, that is, lipon (“so”, “well”) and endaksi? (ok?) in the speech of teachers in pre-primary and primary classrooms in Cyprus. The method that is adopted in the analysis of the data is Conversation Analysis (CA), which has its origins in the pioneering work of Sacks (1992a, 1992b).

Discourse Analysis - Gender - Lexicography 1 1 Major Trends in Theoretical and Applied Linguistics

The lexical items under study could fit into the category of what has traditionally been framed as discourse markers (DMs). According to Schiffrin (1987: 31), DMs are “textual coordinates of talk that bracket units of it”. Examples of discourse markers include the particles “oh”, “well”, “now”, “then”, “you know”, and “I mean”, and the connectives “so”, “because”, “and”, “but”, and “or”. DMs have attracted a lot of research, both in papers and in book-length studies. Some studies deal with a whole range of discourse markers (Schourup 1982, Schiffrin 1987, Watts 1989), while others concentrate on individual ones (Lakoff 1973, Svartvik 1980, Schiffrin 1985 etc.). Most available classifications of DMs are based on their core meaning as separate lexical items and/or their functions in discourse coherence. For instance, Schiffrin (1985) examines the functions of DMs in terms of conversational coherence. Blackmore, who works within the framework of relevance theory suggests mainly four ways of classifying discourse connectives: (1) Discourse connectives introducing contextual implications (e.g., so, too, also); (2) Discourse connective concerned with strengthening (e.g., after all, moreover, furthermore); (3) Discourse connective introducing denial (e.g., however, still, nevertheless, but); (4) Discourse connective indicating the role of the utterance in the discourse in which it occurs (e.g., anyway, incidentally, by the way, finally). Brinton (1996) shows that discourse markers also serve pragmatic functions and can be termed pragmatic markers, used by a speaker to comment on the state of understanding of information about to be expressed (with phrases such as ‘like’, ‘you know’); they may also be used to express a change of state (oh; Heritage 1984) or for subtle commentary by the speaker suggesting that ‘‘what seems to be the most relevant context is not appropriate’’ (well; Jucker 1993:438). These lexical items used in this manner are devoid of semantic content in and of themselves (Östman 1982; Schiffrin 1986; Vincent & Sankoff 1992) and are dependent on the local context and sequence of talk for their interpretation. To this direction an important study is Clift’s (2001) investigation of the particle ‘actually’ in interaction, because it shifts the focus from studies on the functions of DMs to investigation of the meaning of the specific particle in a range of interactional contexts. Scarce research relating the spoken classroom discourse and the use of discourse markers using Greek data has been carried out to date. For instance, Georgakopoulou and Goutsos (1998) and Archakis (2001) investigate conjunctions versus discourse markers based on Greek data. Archakis (2002) examines the discourse marker diladi (‘that is’) in classroom interaction. In Christodoulidou (2011) we centre our attention on the lecture genre and we analyze the communicative purpose of lipon (‘well’), ara (‘so’), orea (‘fine’) within spoken academic discourse. The term ‘discourse markers’ is used only provisionally in this study, because the aim is to uncover the meaning and use of these lexical items based on the

1 2 Discourse Analysis - Gender - Lexicography Maria Christodoulidou

position of the specific lexical items within the sequence and the composition of the turn in which they occur, in order to unravel their use in context. Drew and Holt’s (1998: 497) observation below sheds light on the study of language use in proposing that the study of the components of a turn’s should be analyzed by taking into consideration their place in the sequence and the construction of the turn.

The components of a turn’s construction—at whatever level of linguistic production—are connected with the activity which the turn is being designed to perform in the unfolding interactional sequence of which it is a part, and to the further development of which it contributes.

2. Data and Methodology

The extracts included in this article comprise transcriptions of approximately 20 hours of video-recorded classroom interaction in pre-primary and first and second year primary classes in schools at Nicosia, Cyprus. The DMs under study are examined in the speech of the teachers during the Greek language course in the first and second year primary classes and during the fairytale hour in the pre-primary classes. The language spoken during the recorded classroom interactions involves a combination of Cypriot Dialect and Modern Greek.1 All names of participants are replaced by pseudonyms in order to protect their privacy. For the transcription and analysis of the data we adopted the analytical tools of Conversation Analysis. The transcription symbols used in this study —cited in Appendix I— are based on the transcription conventions developed by Jefferson for the analysis of conversational turns in Anglo-American conversation (cf. Sacks, Schegloff and

1 The Cypriot Greek dialect spoken in Cyprus is an indigenous variety of Greek. Cypriot Greeks are bilingual in the Cypriot dialect and Modern Greek. The Cypriot Greek dialect is acquired naturally while Modern Greek is taught as the standard language (cf. Ferguson, 1959: 30). In recent years researchers like Davy, Panyiotou and Ioannou, (1996), Papapavlou and Pavlou (1998) and Karyolemou and Pavlou (2001) have classified village Cypriot as the basilect and town Cypriot as the acrolect of the low variety. In addition, researchers like Karyolemou (1997; 2000), Moschonas (2002: 917), Terkourafi (2004) and Arvaniti (2002) have supported the creation of a Cypriot koine in urban centres, that is, a middle variety which is different from the local vernaculars incorporating important influences from Modern Greek.

Discourse Analysis - Gender - Lexicography 1 3 Major Trends in Theoretical and Applied Linguistics

Jefferson 1974). The transcription system is intended to capture in detail the characteristics “of the sequencing of turns, including gaps, pauses and overlaps; and the element of speech delivery such as audible breath and laughter, stress, enunciation, intonation and pitch” (Hutchby and Drew 1995: 182). Significant turns for the analysis are marked with arrows. A powerful agenda for the analysis of talk-in-interaction is Schegloff, Ochs and Thompson’s (1996) proposal that the study of linguistic structures could be richly informed by consideration of their place in the wider context of social interaction:

The meaning of any single grammatical construction is interactionally contingent, built over interactional time in accordance with interactional actualities. Meaning lies not with the speaker nor the addressee nor the utterance alone . . . but rather with the interactional past, current and projected next moment. (Schegloff et al. 1996: 40)

Sacks et al. identified components of the turn—the turn-constructional units (henceforth TCUs; Sacks et al. 1974: 702–4)—as sentential, clausal, phrasal, and lexical units, which can constitute complete turns. On their potential completion, transition to another speaker turns out to be relevant. The turn is seen as “the habitat in which turn constructional units –henceforth TCUs– are housed” (Schegloff 1996: 56) and this reframing deepens our understanding of turns-at- talk. Schegloff et al. claim that “an important dimension of linguistic structures is their moment-by-moment evolving interactional production” (Schegloff et al. 1996:39). The shift of focus from sentences to turn constructional units proposed by Schegloff et al. (1996) proves to be essential for this study. In what follows I explore the theoretical and methodological implications of this claim by taking the turn and its component TCUs as the frame of reference in examining the two lexical items under study (lipon, endaksi) in classroom interaction. The DMs under study will be left untranslated within the extracts in order to uncover their interactional meaning based on their sequential order within the conversational extracts.

3. Lipon and Endaksi

According to the Greek-English dictionary of Stavropoulos (1988: 119) lipon can be (a) a deductive conjunction translated as “so”, “then”, “therefore”, “consequently”, “hence” or (b) an interjection which is translated as “so”, “well”, “then”, “now” for the expression of surprise, relief, query, decision etc. The

1 4 Discourse Analysis - Gender - Lexicography Maria Christodoulidou

questioning endaksi is an adverb used to confirm agreement and is translated as “ok?” or “all right?” (Babiniotis 1998: 622; Stavropoulos 1988: 295). As will be shown in sections 3.1, 3.2, 3.3 in the extracts under study lipon: 1. occurs in the speech of the teacher to provide students with directions. Within the same turn, endaksi? occurs in final position to invite students’ agreement (cf. Pomerantz 1984). 2. occurs to signal return to a cut-off narration or discussion and 3. can occur to initiate inductive reasoning.

3.1. Lipon - Direction Giving - Endaksi?

In the extracts that follow the discourse markers lipon and endaksi? frame (cf. Goffman, 1974; 1979) a segment of the teacher’s speech in which she is instructing the students to follow specific directions. Generally lipon signals initiation of a new topic. As was shown in Christodoulidou (2011) in the sequential environment of lecture, lipon-prefaced turns are deployed by the lecturer in order to capture the students’ attention, signal the initiation of the lesson, and the termination of any other interaction among the students. In the data from pre-primary and primary classes, analyzed here, lipon introduces direction giving by the teacher to the students. The giving of directions is followed by questioning endaksi? with which the teacher seeks to confirm agreement by the students.

Extract 1 (T: Teacher; M: Marilena; N: Nicolas; S: Stefani; G: Giorgos. The conversation takes place after the teacher has finished reading a text-poem about weekdays.)

1. M En polla astio::. It’s very fu::nny. 2. T I::ne astio. Pco su fanice astio? It i::s funny. What sounds funny to you? 3. M I Ðef – Tri::ti. Mond- Tue::sday. 4. T I TRITI. Jati:: su fanice astia I Triti? TUESDAY. Why:: does Tuesday sounds funny? 5. M E:::mm epiðι::, E::mm because::, 6. T epiði? ti θa kani::? because? what will he do::? 7. M θa psaksi, he will search,

Discourse Analysis - Gender - Lexicography 1 5 Major Trends in Theoretical and Applied Linguistics

8. T → θa psaksi mes to spiti::. Lipo::n (.) akuste me ti θelo na he will search in the hou::se. Lipo::n (.) listen what I want you to 9. kanete. θa sas ðoso ðio tria lepta cero:: na to ðjavasete po::li:: do. I’ll give you two three minutes to read ve::ry:: 10. prosektika, ºo kaθenasº (.) monos tu ºsiγa º siγa, siγa::, mesa sto carefully,ºeach oneº (.) alone ºslowlyº slowly, slowly:: in your 11. mɲaluðaci tu CE:: meta:: ean kati ðen katalavenete::, head A::ND the::n if there is something you don’t understa::nd, 12. θa:: tο º simioseteº, θa ºprospaθisete º na to ðjavasete prosextika ce you’ll º take a noteº, you’ll ºtryº to read it carefully and 13. → meta θa to sizitisume eðo oli mazi::. Endaksi::? then we’ll discuss it here all together. Endaksi::? 14. M [NE::. [YE::S. 15. N [NE::. [YE::S. 16. S [NE. [YES.

In extract 1 after the teacher has finished reading the text of the day, one of the students, M (1:1) proffers an appreciation, which triggers a discussion (1:1- 7). The teacher (1:8-9) with lipon shifts the topic by initiating a request of the students attention: Lipo::n (.) listen what I want you to do. The request of their attention is followed in turns 9-13 by directions given to the student on how to proceed to second reading of the text. With the questioning endaksi that follows the teacher seeks to secure students’ agreement.

Extract 2 (T: Teacher; M: Maria. Eleni is a university student, observer of the class.)

1. T I Eleni ine fiti::tria sto panepistimio >ce irθe eðo na parakoluθisi pu θa Eleni is a stu::dent at university >and she came to watch 2. → kanume to maθima mas, to paramiθi mas<. Lipon kaθiste anapaftika::, our course, our fairytale<. Lipon sit comfortably::, 3. fronima:: na akusete to paramiθι mas ce na to akusi ce mazi mas ce I quietly:: to listen our fairytale and with us will listen 4. → ciria Eleni, endaksi::? mrs Eleni, endaksi::? 5. M enna mas to pi i ciria Eleni? Is mrs Eleni going to read it to us? 6. T oçi eγo enna sa to po to paramiθi no I’ll read the fairytale.

1 6 Discourse Analysis - Gender - Lexicography Maria Christodoulidou

In extract 2, after the information the teacher provides the students with, with lipon she shifts the topic by commanding the students to sit comfortably::, quietly:: to listen our fairytale and with us will listen mrs Eleni, endaksi::? The questioning endaksi is deployed to secure the students’ agreement.

3.2. Lipon + Return to the Cut-Off Narration

In the extracts presented in this section the discourse marker lipon is deployed after an interruption or parenthetical discussion to the ongoing reading or narration of a fairytale. The use of lipon in turn-initial position is recurrently deployed to mark return to the narration that was in progress before the interruption.

Extract 3 (T= Teacher; A= Assistant; C= Christos; S= Stavros. The teacher is reading a fairytale.)

1. T I meres pernou::san ce I mama γlari::na ekleje epiði The days were pa::ssing and mom gu::ll was crying because 2. to micro γlaraci tis ðen ine san ta alla. Kati sinevene. her little gull was different from the others. Something was wrong. 3. ((two children are pushing each other)) 4. A ta çerja ðen ta exume ja na ðernume alla ja na aŋgaʎazume, a::::: ciria hands are not for hitting but for hugging, o:::h mrs 5. prepi na to pis ce si sta peðaca su oti ta çerja ta exume ja na xaiðevume you have to say this to your kids too, that hands are for caressing 6. oçi ja na ðernume. not for hitting. 7. C eγο piti mu epezame me to aðerfo mu alla en ton eðera. at home we were playing with my brother but I didn’t bit him. 8. T → LIPO::N, jenniθιce to mikro::= LIPO::N, was born the little::= 9. S =γlaraci =gull 10. T I mama γlarina ce o mpampas γlaros pos eɲoθan? how did mom gull and dad gull feel?

In extract 3 while the teacher is reading a fairytale (3: 1-2) she is interrupted by two children pushing each other (3: 3). The assistant (3: 4-6) addresses them with a reprehension. The teacher (3: 8) with the deployment of a loud LIPO::N shifts the footing from the interruption to the activity of reading the fairytale that was in progress as shown by the continuation of the narration in 3:8 LIPO::N, was born the little::.

Discourse Analysis - Gender - Lexicography 1 7 Major Trends in Theoretical and Applied Linguistics

Extract 4 (T= Teacher; M= Michalis. The teacher is reading the fairytale)

1. T ðen mporou::se na peta::ksi to mikro:: γlara::ci ce kaθotane panta pano the little gull couldn’t fly:: and was always sitting on 2. se mia ksilini apovaθra::. Kserete ti ine I apova::θra:: mora? a wooden do::ck. Do you know what a do::ck is kids? 3. M Ne Yes 4. T ore::a. mono o Mixalis kseri ti ine I apova::θra::? Goo::d. only Michalis knows what a do::ck is? ((the teacher shows a picture of a dock)). 5. → lipon ekaθotan panta eki pano stin apovaθra ci evlepe ta alla lipon it was sitting always on the dock and was watching the other 6. γlaraca na maθenun na petane. little gulls learning how to fly.

In extract 4: 2 the teacher cuts off reading to make a clarifying question: Do you know what a do::ck is kids?. The second pair part to the question comes with a student’s response in turn 3. After the teacher shows a picture of the dock (4: 4) she shifts the footing with the deployment of lipon which frames what follows as a continuation of the reading that was in progress before the parenthetical clarifying question.

3.3. Lipon + Inductive Reasoning

In this section lipon occurs late in the speech of the teacher in non turn-initial position and in non TCU initial position. Specifically, lipon is deployed by the teacher as a deductive conjunction to the inductive reasoning expressed by the teacher and leads the sequence to closure.

Extract 5

1. T lei ce me sinefça:: ekso I kakocefça::. It says even with cloudiness kakocefça2 goes out.

2 Kakocefça does not have an exact translation in English. It could be translated as ‘sadness’. It will be left untranslated because the conversation is about the etymology and meaning of the word.

1 8 Discourse Analysis - Gender - Lexicography Maria Christodoulidou

2. N ti ennoi ðame? what does it mean here? 3. T ti ine i kakokefça araje::? Ja sceftite li::γo:: ti leksula ka:ko:cefça::. what is kakocefça I wonde::r? Thi::nk of the word ka:ko:cefça::. 4. ((she writes the word on the board.)) 5. N kaka [ce cefça bad [and cefça 6. E [kako:: [ba::d 7. T INE- >ðe milame oli mazi< sosta to skeftomaste, ine IT’S- >we don’t talk all together< we’re thinking correctly, it 8. apo to kako:: to cefi. Kserete ti simeni exo comes from ba::d cefi3. Do you know what’s the meaning of I have 9. → cefi::? (.) Ime (.) xaru::menos. Eðο lipon kakocefça vjeni apo cefi::? (.) I’m huppy. Here lipon kakocefça comes from 10. to kako CEFI. Ðilaði:: ðen i::me xarumenos.. bad CEFI. That i::s I’m no::t huppy.

In extract the word kakocefça (5:1) triggers a discussion around its meaning in 5: 3. The teacher first explains the meaning of exo cefi as I’m huppy in 5:9. Lipon occurs in non turn-initial position and in non TCU initial position as a deductive conjunction to extract the meaning of kako cefi through inductive reasoning: If exo cefimeans I’m huppy, kako cefimeans I’m not happy.

Extract 6 (Before this extract the teacher gives to students words which include diphthongs and they try to categorize them based on orthography.)

1. T PEÐJA EA::N o ka::θenas-, akuste me liγο:: oli::, afiste CHILDREN I::F each of you- listen to me a little:: a::ll, let 2. ta molivja kato c’ akuste me, EAN o kaθenas apo esa::s (.) kani fasaria your pencils down and listen to me, IF each of you:: (.) makes noise 3. I me tin tsantula tu I pano stin karekla I milai ce either with his bag or on the chair or if he/she’s talking and 4. mu::rmu::ri::zi::, peta::jete opote nane, ÐΕN mpo::run ta πeðja na mu::rmu::ri::ng, po::ps up anytime, children caNNO::T 5. ksexorisune ce n’ akusune. Ine po::li:: li- poli:: lepti ðiafora. hear the difference. It’s a ve::ry:: sl- very:: slight difference.

3 Cefi: mirth, good mood

Discourse Analysis - Gender - Lexicography 1 9 Major Trends in Theoretical and Applied Linguistics

6. Prepi:: na kanume apoliti isiçia ja na katalavenun ta alla peðaca ce I We nee::d to be absolutely quiet for the other kids and our 7. simmaθites mas an ine a::f i a::v. EA::N o kaθenas schoolmates to hear the difference between a::f or a::v. I::F each 8. kani ena fillo etsi:: I to moli::vi tu I to svisti::ri tu::, one makes a page like thi::s or his pensi::l or hi::s ru::bbe::r, 9. ta poðara::ca tu I tin karekla tu, jinete polli fasaria o::::: mes tin his fee::t or his chair, it’s a lot of noise o:::: in the 10. → taksi o::::. Θa sas parakale::so:: lipon na iste apO::lita omos isiçi, classroom o::::. I will lipon a::sk you to be absO::lutely quiet, 11. to::so:: pu otan θa rikso tin karfitsa mu n’ akusti::. Ja na so:: mu::ch that if I throw my pin to hea::r it. Because we need to 12. → katalavenume an tixon kanun ta peðja laθος. Endaksi? Mu ðinete to understand if the children make a mistake. Endaksi? Do you give 13. loγο sas oti θa prospaθisete OLI::? me your word that you’ll A::LL try? 14. P NE:: YE::S

In extract 6 lipon occurs after a long reprehension of the teacher to the students. It occurs as a part of the final request of the teacher to the students to be quiet. The request is presented as warranted based on the reasons exposed by the teacher about what happens when students make noise in the classroom. Lipon (6: 10) occurs in non turn-initial position and also in non TCU initial position as a deductive conjunction of the teacher’s inductive reasoning. With endaksi? (6: 12) expressing request for acceptance, followed with a second request for commitment: Do you give me your word that you’ll A::LL try? (6: 12-13), the teacher invites for agreement and leads the sequence to closure.

4. Conclusion

This article presents the use of the discourse markers lipon and endaksi in classroom interaction. In the data lipon and endaksi are used in specific sequential positions within the wider interactional context and they occupy specific positions within the construction of the turn where they occur. Hence their position within the sequence as well as their position in the turn and composition of the turn –that is their relationship with the other elements of the turn, whether they proceed or follow them, whether they are freestanding, or parts of a TCU– enabled us to unfold their meaning within lecture interaction. A simple schematic representation of the position that the lexical items under study take in the turn and the actions they accomplish as revealed in the data might look like Table 1.

2 0 Discourse Analysis - Gender - Lexicography Maria Christodoulidou

Table 1. The position and composition of the DMs and the activities accomplished.

DMs Composition of the turn Activity

Non turn-initial Giving directions at the beginning of the TCU initial lesson Turn-initial Signal return to a cut-off narration or Lipon TCU initial discussion Non turn initial Inductive reasoning Non TCU-initial Turn final position Endaksi Asking for agreement Freestanding TCU

As was shown in the data lipon signals direction giving by the teacher to the students at the beginning of a new activity. It is followed by questioning endaksi? in turn final position seeking to secure student’s agreement. Lipon can also occur to signal return to the narration that was in progress before an interruption has taken place. A third use of lipon is it occurrence in the middle of the turn as a deductive conjunction as part of an inductive reasoning. Methodologically, the findings presented here underscore the importance of examining discourse markers within interactional sequences as well as within institutional discourse such as classroom interaction as examined here. Analyzing some of the uses of these DMs within the classroom discourse, the current study has revealed that the placement of each lexical particle in the turn and its component TCUs is highly consequential for the activities being undertaken in the sequence to which its turn belongs. Its placement not only characterizes as a particular type of activity—topic introducing, say, or implicative—the turn which contains it but also the turn to which it is responsive. To conclude with, the use of discourse marker has a significant role to play in the structure and coherence of classroom discourse and thus it should be further examined by focusing on the relationship between meaning and interaction.

Discourse Analysis - Gender - Lexicography 2 1 Major Trends in Theoretical and Applied Linguistics

References

Archakis, A. 2001. On discourse markers: Evidence from Modern Greek. Journal of Pragmatics, 33(8), 1235-1261.

Archakis, A. 2002. Information management in discourse: Evidence from the analysis of the discourse marker diladi in classroom interaction”. Pedagogikos Logos, 2, 201-209 [in Greek].

Arvaniti, A. 2002. The maintenance of diglossia in Cyprus and the emergence of Cypriot Standard Greek. MS.

Babiniotis, G. 1998. Leksiko tis neas Ellinikis glossas (Dictionary of the Modern Greek language). Athens: Kentro Lexikologias.

Bellack, A., H. Kliebard, R. Hyman and F. Smith. 1966. The Language of the Classroom. New York: Teachers’ College Press.

Blackmore, D. 1992. Understanding utterances. An introduction to pragmatics. Oxford: Blackwell.

Brinton, L.J. 1996. Grammaticalization and Discourse Functions. Berlin: Mouton de Gruyter.

Christodoulidou, M. 2011. Lexical markers within the University lecture. Novitas- ROYAL (Research on Youth and Language). Special issue on conversation analysis in educational and applied linguistics, 5.1, 143-160. Retrieved April 10, 2011 from http://www.novitasroyal.org/Vol_5_1/ Christodoulidou.pdf

Clift, R. 2001. Meaning in interaction: The case of actually. Language, 77(2), 245- 291.

2 2 Discourse Analysis - Gender - Lexicography Maria Christodoulidou

Davy, J., Y. Ioannou and A. Panayiotou. 1996. French and English loans in Cypriot diglossia. Travaux de la maison del’ Orient méditerranéen 25: Chypre hier et aujourd’hui entre Orient et Occident, 127-136. Lyon: Maison del’ Orient méditerranéen.

Drew, P. and E. Holt. 1998. Figures of speech: Idiomatic expressions and the management of topic transition in conversation. Language in Society, 27, 495- 522.

Erickson, F. 1978. Set ups/wrap ups. Paper Presented at the Annual Meeting of the American Educational Research Association.

Ferguson, C. 1959. Diglossia. In P.P. Giglioli. 1982. (Ed.), Language and Social Context, 232-251. London: Penguin Books.

Georgakopoulou, A. and D. Goutsos. 1998. Conjunctions versus discourse markers in Greek: the interaction of frequency, position, and functions in context. Linguistics, 36, 887-917.

Goffman, E. 1974.Frame analysis: An essay on the organization of experience. New York: Harper and Row.

Goffman, E. 1979. Footing. In Semiotica 25, 1- 29. Reprinted in E. Goffman and D. Hymes, 1981, (eds.), Forms of talk, 124-161. Philadelphia: University of Pennsylvania Press.

Heritage, J. 1984. A change of state token and aspects of its sequential placement. In J.M. Atkinson and J. Heritage (eds.), Structures of Social Action: Studies in Conversation Analysis. Cambridge: Cambridge University Press, 299- 345.

Hutchby, I. and P. Drew. 1995. Conversation analysis. In J. Verschueren, J.-O. Östman, J. Blommaert and C. Bulcaen (eds.), Handbook of pragmatics, 182-189. Amsterdam, Philadelphia: John Benjamins.

Jucker, A.H., 1993. The discourse marker ‘well’: a relevance theoretical account. Journal of Pragmatics 19, 435-452.

Karyolemou, M. 1997. Accommodation theory and the use of the aorist in the Cypriot variety. In G. Drachman, A. Malikouti-Drachman, J. Fykias and C. Klidi (eds.), Greek Linguistics: Proceedings of the Second International Conference on Greek Linguistics 2, 707-726. Salzburg: W. Neugebauer Verlag GmbH.

Discourse Analysis - Gender - Lexicography 2 3 Major Trends in Theoretical and Applied Linguistics

Karyolemou, M. 2000. Le chypriote: dialecte ou idiome. In A.-P. Christides (ed.), La langue grecque et ses dialects, 111-115. Athens: Greek Republic, Direction of International Relations and Centre for the Greek Language (bilingual edition).

Karyolemou, M. and P. Pavlou. 2001. Language attitudes and assessment of salient variables in a bi-dialectal speech community. In J. Fontana et al. (eds.), Proceedings of the first International Conference on Language Variation in Europe, 110-120. Barcelona: Universitat Pompeu Fabra.

Lakoff, R. 1973. Questionable answers and answerable questions. In B.B. Kachru, R.B. Lees, Y. Malkiel, A. Pietrangeli and S. Saporta (eds.), Issues in linguistics. Papers in honor of Henry and Rente Kahane, 453-467. Urbana, IL: University of Illinois Press.

Mehan, H. 1979. Learning lessons: Social organization in the classroom. Cambridge, MA: Harvard University Press.

Moschonas, S. 2002. Koini glossa kai dialektos: To zitima tis glossikis dimorfias stin Kipro (Common language and dialect: The issue of the linguistic diglossia in Cyprus). Nea Estia, 151(1745): 898-928.

Östman, J.-O. 1982. The symbiotic relationship between pragmatic particles and impromptu speech. In N.E. Enkvist (ed.), Impromptu Speech: A Symposium. Publications of the Research Institute of the Åbo Akademi Foundation. Åbo:Åbo Akademi, 147-177.

Papapavlou, A. and P. Pavlou. 1998. A review of the sociolinguistic aspects of the Greek Cypriot dialect. Journal of Multilingual and Multicultural Development, 19(3): 212-220.

Pomerantz, A. 1984. Agreeing and disagreeing with assessments: some features of preferred/ dispreferred turn shapes. In J.M. Atkinson and J. Heritage (eds.), Structures of social action: Studies in conversation analysis, 57-101. Cambridge: Cambridge University Press.

Sacks, H. 1992a. Lectures on conversation, [1964-1968], volume I. Edited by G. Jefferson. Oxford: Basil Blackwell.

Sacks, H. 1992b. Lectures on conversation, [1968-1972], volume II. Edited by Gail Jefferson. Oxford: Basil Blackwell.

2 4 Discourse Analysis - Gender - Lexicography Maria Christodoulidou

Sacks, H., E. Schegloff and G. Jefferson. 1974. A simplest systematics forthe organization of turn-taking for conversation. Language, 50, 696-735.

Schegloff, E. 1988. On an actual virtual servo-mechanism for guessing bad news: A single case conjecture. Social Problems, 32, 442–57.

Schegloff, E. 2007. Sequence Organization in Interaction: A Primer in Conversation Analysis, Volume 1, Cambridge: Cambridge University Press.

Schegloff, E., E. Ochs and S. Thompson. 1996. Introduction. In E. Ochs, E.A. Schegloff and S.A. Thomson (eds.), Interaction and grammar, 1-52. Cambridge: Cambridge University Press.

Schiffrin, D. 1985. Conversational coherence: The role of ‘well’. Language, 61(3): 640-667.

Schiffrin, D. 1986. Functions of and in discourse.Journal of Pragmatics 10, 41–66.

Schiffrin, D. 1987. Discourse Markers. Cambridge: Cambridge University Press.

Schourup, L. 1982. Common discourse particles in English conversation. New York: Garland.

Sinclair, J.McH. and R.M. Couthard. 1975. Towards an Analysis of Discourse: The English Used by Teachers and Pupils. London: Oxford University Press.

Stavropoulos, D.N. 1988. Oxford Greek-English learner’s dictionary. Oxford: Oxford University Press.

Svartvik, J. 1980. ‘Well’ in conversation. In S. Johansson, G. Leech, J. Svartvik (eds.), Studies in English linguistics for Randolph Quirk, 167-177. London: Longman.

Terkourafi, M. 2004. Testing Brown and Levinson’s theory in corpus of conversational data from Cypriot Greek. International Journal of the Sociology of Language, 168, 119-134.

Vincent, D. and D. Sankoff. 1992. Punctors: A pragmatic variable. Language Variation and Change 4, 205-216.

Watts, R. 1989. Taking the pitcher to the ‘well’. Native speakers’ perception of their use of discourse markers in conversation. Journal of Pragmatics, 13, 203- 237.

Discourse Analysis - Gender - Lexicography 2 5 Major Trends in Theoretical and Applied Linguistics

Appendix

Transcription Conventions

[ Separate left square brackets, one above the other on two successive lines with utterances by different speakers, [ indicates a point of overlap onset, whether at the start of an utterance or later. [[ Double sepάrate left square brackets, distinguish pairs of [[ overlapped utterances. = Equal signs ordinarily come in pairs – one at the end of a line and another at the start of a next line. If the two lines connected by the equal signs are by the same speaker, then there was a single, continuous utterance with no break or pause, which was broken up in order to accommodate the placement of overlapping talk. If the lines connected by two equal signs are by different speakers, then the second followed the first with no discernible silence between them. (2) Numbers in parenthesis indicate silence. (.) A dot in parentheses indicates a micropause. . The period indicates a falling or final, intonation contour, not necessarily the end of a sentence. ? A question mark indicates rising intonation, not necessarily a question. , A comma indicates continuing intonation, not necessarily a clause boundary. :: Colons are used to indicate the prolongation or stretching of the sound just preceding them. The more colons the longer the stretching. - A hyphen after a word or part of a word indicates a cut-off or self- interruption, often done with a glottal or dental stop. word Underlining is used to indicate stress or emphasis. WOrd Capital letters indicate louder than the rest talk. ˚ ˚ Two degree signs indicate that the talk between them is markedly softer than the talk around it. ↑ The up arrow indicate a segment starting on sharper rise. > < The combination of “more than” and “less than” symbols indicates that the talk between them is compressed or rushed. .hhh The dot followed by “h’s” indicates inbreath

2 6 Discourse Analysis - Gender - Lexicography Labeling of Derogatory Words in Modern Greek Dictionaries1

Angeliki Efthymiou, Zoe Gavriilidou and Eleni Papadopoulou

Democritus University of Thrace

[email protected], [email protected], [email protected]

Abstract

The purpose of the present paper is to examine the problems of usage labeling in two Modern Greek dictionaries (the LNEG2 and the LKN) through a survey of their treatment of ‘negative’ terms. To do this the terms studied are drawn from various semantic groups that denote nationality, racial or cultural group, lack of intelligence, age, sexual orientation, terms addressing women, bodily effluvia, etc. The investigation of the treatment of these terms in the two Modern Greek dictionaries reveals that it is difficult to support the usage labeling distinctions they make in their prefaces for derogatory, offensive, slang or taboo words.

1. Introduction

This paper surveys the treatment of ‘negative’ terms in two Modern Greek dictionaries (the LNEG2 and the LKN) and reveals a series of problems relating to the usage labels attached to many of the entries. The terms studied come from various semantic groups such as those denoting nationality, racial or cultural group (γκέκας ‘resident of Northern Epirus’, τουρκομερίτης ‘sb. who comes from a Turkish region’, σκυλάραπας ‘bloody nigger’), lack of intelligence (βλαμμένος ‘idiotic’), age (σκατόγερος ‘fucking old man’), sexual orientation (κουνιστός ‘sissy’, ντιγκιντάγκας ‘faggy’), terms for women (καραπουτάνα ‘harlot’, γύναιο ‘slut’), terms for the organs and acts of sex (ψωλή ‘dick’, μαλακίζομαι ‘jerk off’), bodily effluvia (σκατό ‘shit’, χέσιμο ‘dump’, etc. (cf. Allan & Burridge 2006; Kechagia 1997;

1 We would like to thank Jason Merchant, Anastasia Giannakidou, Lydia Mitits and Anna Sarafianou, for the help in the translation of data entries in English. We are also grateful to the anonymous reviewers for their useful suggestions.

Discourse Analysis - Gender - Lexicography 2 7 Major Trends in Theoretical and Applied Linguistics

Ξυδόπουλος 2008). The paper is structured as follows. In section 2, we offer a brief overview of the relevant literature on taboo and insulting words. We also provide a brief discussion of the literature on dictionary policies in what concerns the labeling of these words. In section 3 we explain how the data of this study were selected, and we also present and discuss our results. In the last section 4, concluding remarks are given.

2. Research Background

In this section, we give a brief overview of the previous literature on our subject. In 2.1 we present the various categories of the offensive words, whereas in 2.2 we discuss dictionary labeling of these words.

2.1. Taboo and Insulting Words

Modern Greek, like other languages, contains words, which people avoid using in most contexts, because they feel them extremely embarrassing or offensive. Words of this type vary from taboo words to insults or swearing (cf. Crystal 1995; Mercury 1995; Allan & Burridge 2006). According to Crystal (1995: 173), these three categories may overlap or coincide, but they are not identical: to call someone κώλος ‘an ass’ is to use a taboo word as insult, but if used with enough emotional force could be considered an act of swearing. On the other hand, στουρνάρι ‘blockhead ’, is a term of insult, but it is neither a taboo word nor a swear word. Finally, the swear word κατάρα ‘curse’ is neither a taboo word nor an insult (cf. Ξυδόπουλος 2008). Taboo language contains the so-called ‘dirty words’, i.e. mainly terms for bodily organs associated with sex, excretion and the act of sexual intercourse (αρχίδι ‘prick’, μουνί ‘cunt’, κώλος ‘ass’), terms for activities involving these organs (καυλώνω ‘get horny’), terms for bodily effluvia issuing from these organs (κουράδα ‘turd’, σκατό ‘shit’), terms for disease, death and the supernatural (καρκίνος ‘cancer’, πεθαίνω ‘die’). The term ‘dirty words’ denotes people’s attitudes towards the denotations and connotations of these words, which are the most emotionally evocative of all language expressions (Allan & Burridge 2006). People not only avoid using them in polite society, but also tend to replace them by a more technical term (e.g. πέος ‘penis’, κόπρανα ‘stool’, πρωκτός ‘rectum’) or a euphemism, which refers to the taboo topic in a vague or indirect way (e.g. έφυγε ‘be gone’ instead of πέθανε ‘died’, πουλάκι ‘cock’ instead of πούτσος ‘prick’) (Crystal 1995). According to Crystal (1995: 173), swearing refers to the strongly emotive use of a taboo word or phrase, and its function is to express a wide range of emotions,

2 8 Discourse Analysis - Gender - Lexicography Angeliki Efthymiou, Zoe Gavriilidou and Eleni Papadopoulou

like annoyance, frustration or anger. Swearing can mark also social distance, as for example when swearing in public (το Χριστό! ‘God damn!’), or act as an in-group solidarity marker, as when a group shares identical swearing norms (Mercury 1995; Allan & Burridge 2006; Crystal 1995). According to the same author, swearing can be further divided into smaller categories like blasphemies, which show contempt towards God, profanities, which show contempt to holy things or people, and obscenities, which involve the expression of indecent sexuality (cf. also Mercury 1995). Dysphemistic terms of insult (or abuse) are usually used in order to debase someone’s physical appearance (αλόγα ‘cow’, καράφλας ‘bald-headed’, πατάτας ‘fathead’), mental ability (στουρνάρι ‘blockhead’), character (καριόλης ‘fucker’), behaviour (καλοπερασάκιας ‘shirker’), beliefs (αγροτοπατέρας ‘trade unionist who takes advantage of farmers; lit. father of farmers), and familial or social relations (μπούλης ‘roly-poly’, μαμόθρεφτος ‘mama’s boy’) (Allan & Burridge 2006; Mercury 1995). Furthermore, terms of abuse (or vulgarisms) can be used to devalue a thing described (αραμπάς ‘ox-cart’) (Mercury 1995) In particular, the vast majority of the insulting words comment on someone’s ugliness (μαλλιαρός ‘hairy’, μπακατέλα ‘pot-bellied’), skin color or complexion (ασπρουλιάρης ‘whitey, honky’, κιτρινιάρης ‘chink, sick-looking, yellowish’, κοκκινοτρίχης ‘red-haired’), clothes (κοντοβράκι ‘knee-breeches’, μαυροφορούσα ‘dressed in black’, over- or undersize (κοιλαράς ‘pot-bellied’, κοκαλιάρης ‘skinny’, μαούνα ‘battle ship’, σαμιαμίδι ‘little worm’, στούμπος ‘shorty’), age (αρχαίος ‘ancient’, μουστόγρια ‘shriveled up old woman’, μπαμπόγερος ‘dirty old man’), physical defects (κουλός ‘armless’, καλαμοπόδαρος ‘spindle-legged’, καρπουζοκέφαλος ‘large-headed’, σακάτης ‘cripple’, στραβούλιακας ‘blind as a bat’), slovenliness or incontinence (αρχιτεμπέλης ‘lazybones’, προκομμένος ‘good for nothing’, καφενόβιος ‘a café habitué’, μεθύστακας ‘drunkard’, μπεκροκανάτα ‘boozer’), smelliness and dirtiness (κατουρλιάρης ‘pissing on one’s pants’, κλανιάρης ‘gassy’), stupidity (καρπαζοεισπράχτορας ‘punching bag’), unreliability and untruthfulness (αρχιψεύτης ‘big liar’, κατσικοκλέφτης ‘scoundrel; lit. goat thief’) or incompetence (αστοιχείωτος ‘ignorant’, αχυράνθρωπος ‘puppet man’, απήδηχτος ‘not screwed’, κουραμπιές ‘desk soldier; lit. sugared bun’, μάπας ‘soft touch’, μπάμιας ‘dupe’), greediness and flattery (πειναλέος ‘ravenous’, κωλογλείφτης ‘licking sb’s ass’), meanness (αρχίδι ‘prick’, καριόλης ‘fucker’, κουμάσι ‘sly’), tartiness, sexual laxness or perversion (καραπουτάνα ‘slut’, κνώδαλο ‘scally wag’, καμπαρετζού ‘slut’, καυλιάρης ‘horny’), sexual orientation or practice (κολομπαράς ‘pansy’, κουνιστός ‘sissy’), familial relationships (μαμόθρεφτο ‘mamma’s boy’), violent behavior (κανίβαλος ‘cannibal’), social or economic status (αριστοκράτης ‘aristocrat’, μπουρζουάς ‘bourgeois’, μπασκλάς ‘low class’), profession (καθηγητιλίκι ‘professorship (iron.)’, καραβανάς ‘ranker’, μπάτσος ‘cop’, πεθαμενατζής ‘grave digger’, προφέσορας ‘professor (iron.)’, πολιτικάντης ‘tricky politician’), religious or ideological beliefs (αλλόπιστος ‘heathen’, κομματόσκυλο ‘henchman’, κομμούνι

Discourse Analysis - Gender - Lexicography 2 9 Major Trends in Theoretical and Applied Linguistics

‘commie’), or racial or ethnic group (αράπης ‘nigger’, μπαστουνόβλαχος ‘hillbillies’). Additionally, insults may comment on supposed inadequacies of someone’s family or friends (κερατάς ‘cuckold’). It is worth pointing out that a large number of the terms of abuse derive on nouns denoting 1. animals that are conventionally associated with unpleasant characteristics (αλόγα ‘big ungainly woman’, αγελάδα ‘cow’, καρακάξα ‘magpie’, σκυλί ‘dog’, μουλάρα ‘she-mule’, μουλάρι ‘mule’, μαντρόσκυλο ‘sheepdog’), 2. tabooed bodily organs, effluvia and sexual behaviours (αρχίδι ‘prick’, μουνί ‘cunt’, κουράδα ‘turd’, κωλογλείφτης ‘licking sb’s ass’) or 3. proper names (Κατίνα ‘Katina, woman who likes gossiping’, Σταχτοπούτα ‘princess, lit. Cinderella’) (cf. for example, Allan & Burridge 2006; Kechagia 1997; Labov 1978). Finally, it should be noticed that a taboo word or a term of abuse may have several uses, which vary from insult to intimacy and solidarity. In everyday conversations one can find examples where apparent terms of abuse (ex. μαλάκας ‘asshole’) are used in a humorous way to display friendship or affection to someone close to the speaker (Allan & Burridge 2006; Crystal 1995; Kechagia 1997; Ξυδόπουλος 2008).

2.2. Labeling of Insulting Words in Dictionaries

Although taboo and insulting words are extremely frequent in everyday conversation, particularly of young people, it took dictionary writers some time to decide to include these words in their headwords (entries) (Bejoint 2000; Landau 2001). According to Landau (2001), dictionary labelling of insult is essentially political and moral. If a general purpose dictionary needs to be commercially successful, it must reflect the ideological values of its public (Béjoint 2000). On the other hand, according to Béjoint (2000: 129), the more recent a dictionary is, the more liberal it is. According to him, all contemporary general purpose dictionaries in all countries move towards greater liberalism in the inclusion of taboo or offensive words (Béjoint 2000: 127)2. Most modern dictionaries try to warn the potential user against the offensive character of these words by using usage labels, like derogatory, offensive, disparaging, etc. Nevertheless, as already mentioned in the literature (Abecassis 2008; Landau 2001; Norri 2000; Wachal 2002; Fedorova 2004; Ptaszynski 2010; Αναστασιάδη- Συμεωνίδη 2007; Κατσούδα & Tράπαλης 2007), dictionaries inconsistently label

2 It seems also that the decision whether to include or not to include offensive words in a dictionary relates also to the purpose of the dictionary, i.e. whether the dictionary is descriptive or has normative or pedagogical aims (Landau 2001: 230).

3 0 Discourse Analysis - Gender - Lexicography Angeliki Efthymiou, Zoe Gavriilidou and Eleni Papadopoulou

taboo and insulting words such as ‘colloquial’, ‘informal’, ‘derogatory’, sexist, etc. On the other hand, as Landau (2001: 233) observes, there are no agreed- upon criteria for characterizing some usages as offensive or abusive. Labels cannot tell us everything about the degree of offensiveness of specified terms under specific conditions (cf. Landau 2001). Furthermore, Norri’s study on English dictionaries revealed that with certain semantic categories (terms for nationality or race, terms for stupidity and terms for deceitfulness) the likelihood of uniform treatment is greater than with others (ex. terms for sexual orientation, physical appearance, arrogant or aggressive behavior) (cf. Norri 2000: 91). In addition, one of Norri’s findings was that terms for nationality are consistently accompanied by a ‘negative’ in at least sixty per cent of the entries and that the percentage of labeling is radically smaller in the semantic categories that refer to unintelligent or deceitful people. Finally, it is worth noting that usage labels have recently received renewed attention in Modern Greek literature (Αναστασιάδη-Συμεωνίδη 2007; Τράπαλης 2005; Τράπαλης & Κατσούδα 2007). These studies compare the labelling systems employed in Modern Greek dictionaries, point out the problems emerging from their use and proceed to suggestions for the creation of labelling systems which would be more functional.

3. Labeling of Insulting Words in the LKN and the LNEG2

As already mentioned in the introduction, the purpose of the present paper is to survey the treatment of ‘negative’ terms in two Modern Greek dictionaries, namely the LNEG2 and the LKN. The main hypothesis is that negative label usage in dictionaries varies according to the semantic class of the entry. We also expect variation in the labeling of the same negative words between the LNEG2 and the LKN.

3.1. Data Description

For the corpus elaboration of the present study first we automatically extracted all the entries of the letters Α, Κ, Μ, Π, Σ marked with the labels ειρωνικό (ειρ.) ‘ironic’, μειωτικό (μειωτ.) ‘pejorative /derogatory’, σκωπτικό (σκωπτ.) ‘satirical/jocular’, υβριστικό (υβρ.) ‘offensive’, χλευαστικό (χλευ.) ‘derisory’, χυδαίο (χυδ.) or ! ‘vulgar’ and κακόσημο (κακοσ.) ‘disparaging’ from the online version of the LKN (http://www. komvos.edu.gr/dictionaries/ dictonline/DictOnLineTri.htm) and from the LNEG2. The extraction provided

Discourse Analysis - Gender - Lexicography 3 1 Major Trends in Theoretical and Applied Linguistics

543 entries. Then an ACCESS data base was created including each entry followed by the labels used in the two dictionaries for each entry. Since very often the negative sense of a word is conveyed by the definition of that word without the use of special labels, we undertook a second extraction of the words starting with Α, Κ, Μ, Π, Σ marked as προφορικό (προφ.) ‘spoken’, μεταφορικό (μτφ.) ‘metaphorical’ or οικείο (οικ.) ‘colloquial’, and including the terms ανόητος ‘stupid’ or the expressions αρνητικός χαρακτηρισμός ‘negative characterization’, υβριστικός χαρακτηρισμός ‘offensive characterization’ or μειωτικός χαρακτηρισμός ‘derogatory characterization’ within their definition. The new extraction provided another 162 entries. In a second phase the entries were classified according to their meaning in the following categories:

• words evaluating nationality / racial or cultural group • words evaluating mental abilities • words evaluating appearance (ugliness, weight) • words evaluating behavior • words evaluating political beliefs • words evaluating sexual orientation • derogatory words for women or men • words relating to sex (terms for the organs and acts of sex) • words denoting bodily effluvia • words evaluating the age of a person or an object • words relating to religion • words relating to diseases or disabilities

From the total 705 entries only 331 were classified in the above mentioned categories. These words constituted our final corpus.

3.2. Results

3.2.1. Labeling Comparison between the LKN and the LNEG2

From the initial 543 LKN entries including the labels ειρωνικό (ειρ.) ‘ironic’, μειωτικό (μειωτ.) ‘pejoratif/derogatory’, σκωπτικό (σκωπτ.) ‘satirical/jocular’, υβριστικό (υβρ.) ‘offensive’, χλευαστικό (χλευ.) ‘derisory’, χυδαίο (χυδ.) ‘vulgar’, κακόσημο (κακοσ.) ‘disparaging’ and ! the LNEG2 had separate labels only for 242 words. 124 of the 543 shared the same label with the equivalent entry in the LKN. Table 1 presents the frequency of labels used in the LKN and the LNEG2 entries of our sample.

3 2 Discourse Analysis - Gender - Lexicography Angeliki Efthymiou, Zoe Gavriilidou and Eleni Papadopoulou

Table 1. Frequency of labels in the LKN and the LNEG2

LABELS LKN PERCENTAGE LNEG2 PERCENTAGE

Iron. 185 34% 43 17,8%

Pej. 256 47,1% 91 37,6%

Satir. 10 1,8% 24 9,9 %

Off. 51 9,3% 26 10,7%

Der. 21 3,8% - 0%

Vulg. 31 5,7% - 0%

Disp. - 0% 33 13,6%

! - 0% 25 10,3%

TOTAL 543 100% 242 100%

As was expected differences were found in the labeling systems for derogatory words in the LKN and the LNEG2. This finding is consistent with previous research (Αναστασιάδη-Συμεωνίδη 2007; Τράπαλης & Κατσούδα 2007) which has shown that labelling systems in Modern Greek dictionaries differ in respect to the number of lexicographic labels and the way they are used. In many entries of our sample no labels are used in the LNEG2 for words that are labelled, the LKN as for example in κοκκινοτρίχης ‘red-haired’ (labeled σκωπτικό ‘satirical/jocular’ in the LKN), αλλόπιστος ‘heathen’ (labeled μειωτικό ‘pejorative/derogatory’ in the LKN), κουνιστός ‘sissy’ (labeled μειωτικό ‘pejorative/derogatory’ in the LKN). In some other cases, different labels described the same entries, as for example πατσαβούρα ‘slut’ (labeled μειωτικό ‘pejorative/derogatory’ in the LKN and υβριστικό ‘offensive’ in the LNEG2), ποντικομαμή ‘insidious; lit. mouse midwife’ (labeled σκωπτικό ‘satirical/jocular’ in the LKN and υβριστικό ‘offensive’ in the LNEG2),κιτρινιάρης ‘chink, sick-looking, yellowish’ (labeled μειωτικό ‘pejorative/derogatory’ in the LKN and κακόσημο ‘disparaging’ in the LNEG2). Finally from the 124 entries which shared the same labels in the LKN and in the LNEG2 81 concerned the use of the label μειωτικό ‘pejorative/derogatory’. On the other hand, as we can see in Table 1, there is unanimity as far as the frequency of use of various labels in the two dictionaries is concerned. More precisely, the most frequent negative label both in the LKN and in the LNEG2 is the label μειωτικό ‘pejorative/derogatory’ followed by the label ειρωνικό ‘ironic’, υβριστικό ‘offensive’ and finallyσκωπτικό ‘satirical/jocular’.

3.2.2. Distribution of Labels According to Semantic Categories

Tables 2 and 3 present the distribution of labels into semantic categories in the LKN and the LNEG2, respectively.

Discourse Analysis - Gender - Lexicography 3 3 Major Trends in Theoretical and Applied Linguistics

Table 2. Distribution of labels into semantic categories in the LKN.

CATEGORIES Iron. Pej. Sat. Off. Deris. Vulg Spok. Met. Coll. Def. total

Nationality 7 1 1 9

Mental abilities 4 3 2 3 17 6 35

Appearance 9 18 2 1 4 3 10 47

Behavior 21 40 10 3 5 4 3 29 115

Political beliefs 1 15 2 4 22 Sexual 3 1 1 5 orientation Derogatory words 2 4 1 10 1 3 16 37 women/ men Sex 2 8 2 1 13

Bodily effluvia 2 2 1 9 3 1 18

Age 3 8 1 2 1 1 16

Disease 4 1 1 1 2 9

Religion 2 3 1 6

Table 3. Distribution of labels into semantic categories in the LNEG2.

CATEGORIES Iron. Pej. Sat. Off. Disp. ! Spok. Met. Coll. 0 Def. total

Nationality 2 1 1 1 2 2 9

Mental abilities 1 5 1 4 7 1 16 35

Appearance 1 23 5 1 2 2 13 47

Behavior 5 17 5 10 8 10 13 12 35 115

Political beliefs 1 11 2 3 3 20* Sexual 1 1 1 2 5 orientation Derogatory words 3 10 6 4 14 37 women/ men Sex 1 8 4 13

Bodily effluvia 1 1 9 3 1 1 16*

Age 1 3 1 4 1 1 2 3 16

Disease 3 5 1 9

Religion 1 5 6

3 4 Discourse Analysis - Gender - Lexicography Angeliki Efthymiou, Zoe Gavriilidou and Eleni Papadopoulou

A. Words Evaluating Nationality / Racial or Cultural Group

Three out of the seven sample words for nationalities share the same label in both the LKN and the LNEG2: μπουρτζόβλαχος ‘red-neck’ and αράπης ‘nigger’ (labeled μειωτικό ‘pejorative/derogatory’) and μούλος ‘bustard/‘bastard’ (labeled υβριστικό ‘offensive’). The LKN systematically labels such words as μειωτικό ‘pejorative/ derogatory’ except for two cases, μούλος labelled as υβριστικό ‘offensive’ and μακαρονάς ‘spaghetti eater’ (literal translation), which refers to Italians, labeled as χλευαστικό ‘derisory’. The labelling policy of the LNEG2 is not that systematic. In some cases there is neither a label nor a description of the negativity within the definition (e.g. αρναούτης ‘bungler’). In some others (e.g. αραπιά ‘nigger land’, αρβανίτης ‘arvanites’) the negativity of the term is expressed in the definition. Finally in the LNEG2 nationality terms are occasionally labelled as σκωπτικό ‘satirical/jocular’, υβριστικό ‘offensive’ orκακόσημο ‘disparaging’.

B. Words Evaluating Mental Abilities

The labeling tendency in the LKN for words evaluating mental abilities is to characterize them as οικείο ‘colloquial’, whereas in the LNEG2 there is a systematic description of the negativity of such terms within the word definition. Some other labels used marginally in the LKN are μειωτικό ‘pejorative/derogatory’ (e.g. στουρνάρι, στούρνος, κούτσουρο ‘blockhead’). For these words the LNEG2 adopts the label υβριστικό ‘offensive’.

C. Words Evaluating Appearance (Ugliness, Weight)

As can be noticed in Tables 2 and 3, both the LKN and the LNEG2 have the tendency to mark such terms mainly by the use of the label μειωτικό ‘pejorative/ derogatory’ or by describing the negativity in the entry definition. The LKN also quite frequently uses the label ειρωνικό ‘ironic’ for cases like στούμπος ‘shorty’, καρπουζοκέφαλος ‘largeheaded’, καράφλας ‘baldheaded’. Other marginal labels used to characterize appearance in the LKN are σκωπτικό ‘satirical/jocular’ (e.g. κοκκινοτρίχης ‘redhaired’) and χλευαστικό ‘derisory’ (e.g. μαυροτσούκαλο ‘man/ woman with dark complexion; lit. black pot’) whereas in the LNEG2 ειρωνικό ‘ironic’ (e.g. στούμπος ‘shorty’) and ! (e.g. κωλαρού ‘fat ass’).

D. Words Evaluating Behavior

The majority of our sample words evaluating behavior receive in the LKN the label μειωτικό ‘pejorative/derogatory’. A high number of cases are also labeled as ειρωνικό ‘ironic’ or their negativity is described in the definition. Finally, ten out of 115 cases were labeled υβριστικό ‘offensive’. On the contrary, studying Table 3,

Discourse Analysis - Gender - Lexicography 3 5 Major Trends in Theoretical and Applied Linguistics

it could be claimed that no clear tendency is adopted in the LNEG2: while the majority of such entries are marked for their negativity with information provided in the definition, however there is use of a wide range of labels, such as ειρωνικό ‘ironic’ (e.g. πολύξερος ‘smart ass’, κυράτσα ‘aunty’), μειωτικό ‘pejorative/ derogatory’ (e.g. κουραμπιές ‘desk soldier’, μούτρο ‘rascal’), σκωπτικό ‘satirical/ jocular’ (e.g. μπεκρούλιακας ‘drunkard’, μπεκροκανάτα ‘boozer’), υβριστικό ‘offensive’ (e.g. μπάμιας ‘bootless, coy; lit. okra’, μουλάρι ‘mule’), κακόσημο ‘disparaging’ (e.g. κοπρόσκυλο ‘bum’, καρεκλοκένταυρος ‘an executive who cannot be moved from his position’), ! (e.g. κωλοβαράω ‘lazy around’, καριόλης ‘fucker’) or μεταφορικό ‘metaphorical’ (e.g. μαϊντανός ‘someone who appears everywhere; lit. parsley’, σπάρος ‘lazybones’). Finally, the LNEG2 provides neither labels nor negative description in the definition for 12 out of 115 entries.

E. Words Evaluating Political Beliefs

Both the LKN and the LNEG2 mainly adopt the label μειωτικό ‘pejorative/ derogatory’ for words evaluating someone’s political beliefs (e.g κομμουνιστοσυμμορίτης ‘communist ganger’, μοναρχοφασίστας ‘monarcho- fascist’). This unanimity could be attributed to the fact that use conditions of such words are unambiguous; the intention of a speaker who uses such words is to express in a strongly contemptuous way his negative opinion of the other in order to offend him.

F. Words Evaluating Sexual Orientation

From the five sample words the three are described through the label μειωτικό ‘pejorative/derogatory’ in the LKN (πούστης ‘faggot’, κουνιστός ‘sissy’, κίναιδος ‘poof’). One is characterized χυδαίο ‘vulgar’ (κολομπαράς ‘pansy’) and one προφορικό ‘spoken’ (μπινές ‘sod’). In the LNEG2 the negativity of one word for sexual orientation is described in the definition as slangαργκό ( ) (κολομπαράς ‘pansy’), one is labeled μειωτικό ‘pejorative/derogatory’ (κίναιδος ‘poof’), two are labeled ! (πούστης ‘faggot’, μπινές ‘fag’) and for one there is no special indication of their pejorative character (κουνιστός ‘sissy’).

G. Derogatory Words for Women or Men

Our data show that derogatory words for women or men are usually labeled either as υβριστικό ‘offensive’ or they receive a description of their negativity in their definition both in the LKN and in the LNEG2. Quite frequent in the LNEG2 is the use of the label ! (καραπουτάνα ‘harlot’, μαλάκας ‘asshole’, μαλακισμένος ‘dickhead’, παλιοσκρόφα ‘old bitch’). Other labels marginally used in the LKN are ειρωνικό ‘ironic’ (Σταχτοπούτα ‘princess; lit. Cinderella’), μειωτικό ‘pejorative/

3 6 Discourse Analysis - Gender - Lexicography Angeliki Efthymiou, Zoe Gavriilidou and Eleni Papadopoulou

derogatory’ (καμπαρετζού ‘slut’), σκωπτικό ‘satirical/jocular’ (ποντικομαμή ‘little worm’), χυδαίο ‘vulgar’ (αρχιδάτος ‘lit. with balls’).

H. Words Relating to Sex (Terms for the Organs and Acts of Sex) / Bodily Effluvia

Both dictionaries agree in labeling words relating to sex or bodily effluvia as χυδαίο ‘vulgar’ (the LKN) or ! (the LNEG2).

I. Words Evaluating the Age of a Person or an Object

The LKN mainly labels words evaluating the age of someone as μειωτικό ‘pejorative/derogatory’. Other labels are also used marginally (ειρωνικό ‘ironic’, σκωπτικό ‘satirical/jocular’, υβριστικό ‘offensive’, χλευαστικό ‘derisory, χυδαίο (χυδ.) ‘vulgar’). The LNEG2 makes no systematic use of a wider range of labels μειωτικό ‘pejorative/derogatory’, ειρωνικό ‘ironic’, σκωπτικό ‘satirical/jocular’, υβριστικό ‘offensive’, χλευαστικό ‘derisory, !, κακόσημο ‘disparaging’, μεταφορικό ‘metaphorical’, etc.

J. Words Relating to Religion

For such words the LKN uses the labels ειρωνικό ‘ironic’, μειωτικό ‘pejorative/ derogatory’ and υβριστικό ‘offensive’. On the other hand, the LNEG2 has a clear tendency to provide negative description in the lemmas’ definitions.

K. Words Relating to Diseases or Disabilities

Finally derogatory terms for diseases or disabilities are described in the LKN with the labels μειωτικό ‘pejorative/derogatory’ (σακατιλίκι ‘the characteristic of being cripple, incapable/incompetent’), κουτσαμάρα ‘cripple’), σκωπτικό ‘satirical/ jocular’ (μύωψ ‘myopic, short-sighted’), χλευαστικό ‘derisory’ (στραβοκάνης ‘bow- legged’), προφορικό ‘spoken’ (κουφάλογο ‘deaf horse; lit. for someone who can’t hear well’). In the LNEG2 the labels used are μειωτικό ‘pejorative/derogatory’ (στραβούλιακας ‘blind as a bat’) and ειρωνικό ‘ironic’.

4. Conclusions

In this paper we surveyed the treatment of ‘negative’ terms in two Modern Greek dictionaries, the LNEG2 and the LKN. The investigation of the treatment of these terms has revealed that 1. the distinction made in the prefaces between derogatory, offensive, slang or taboo words is hard to maintain, 2. dictionaries

Discourse Analysis - Gender - Lexicography 3 7 Major Trends in Theoretical and Applied Linguistics

do not always agree in the labeling of the same negative words, 3. there is a wide range of possible descriptions of negative usage, and 4. the likelihood of a ‘negative’ indication varies from one semantic group to another. Finally, we notice that these findings are consistent with Norri’s (2000) findings about usage labeling in English dictionaries.

3 8 Discourse Analysis - Gender - Lexicography Angeliki Efthymiou, Zoe Gavriilidou and Eleni Papadopoulou

References

Abecassis, M. 2008. The Ideology of the Perfect Dictionary: How Efficient can a Dictionary be? Lexikos 18: 1-14.

Allan, K. and K. Burridge. 2006. Forbidden Words: Taboo and the Censoring of Language. Cambridge: Cambridge University Press.

Béjoint, H. 2000. Modern Lexicography. An Introduction. Oxford: Oxford University Press.

Crystal, D. 1995. The Cambridge Encyclopedia of the English Language. Cambridge: Cambridge University Press.

Fedorova, I. 2004. Style and Usage labels in Learner’s Dictionaries: Ways of Optimization. Euralex 2004 Proceedings, 265-272. Available at: http://www.euralex.org/elx_proceedings/Euralex2004/027_2004_V1_ Irina%20V.%20FEDOROVA_Style%20and%20usage%20labels%20in%20 learners%20dictionaries_Ways%20of%20optimisati.pdf

Kechagia, A. 1997. Insulting words in Modern Greek. In Proceedings of the 2nd International Conference on Greek Linguistics. Graz: Neugebauer Verlag, 297-304.

Labov, W. 1978. Le parler ordinaire: la langue dans les gettos noirs des Etats-Unis. (Language in the Inner City) transl. by A. Kihm. Paris: Minuit.

Landau, S. 2001. 2nd edition. Dictionaries. The Art and Craft of Lexicography. Cambridge: Cambridge University Press.

Mercury, R.-E. 1995. Swearning: a ‘bad’ part of language; A ‘good’ part of language learning. TESL Canada Journal 13(1): 28-36.

Discourse Analysis - Gender - Lexicography 3 9 Major Trends in Theoretical and Applied Linguistics

Norri, J. 2000. Labelling of Derogatory Words in Some British and American Dictionaries. International Journal of Lexicography 13(2):71-106.

Ptaszynski, M.O. 2010. Theoretical Considerations for the Improvement of Usage Labelling in Dictionaries: A Combined Formal-Functional Approach. International Journal of Lexicography 23(4): 411-442.

Wachal, R. 2002. Taboo or not taboo? That is the question. American Speech 77(2):195-206.

Αναστασιάδη-Συμεωνίδη, Ά. 2007. Λεξικογραφία και χρηστικά σημάδια. Στα Πρακτικά του 8ου Διεθνούς Συνεδρίου Ελληνικής Γλωσσολογίας. Ιωάννινα. 30/8-2/9 2007.

(LΚΝ) Ινστιτούτου Νεοελληνικών Σπουδών. Ίδρυμα Μ. Τριανταφυλλίδη 2007 (α΄ έκδ. 1998). Λεξικό της Κοινής Νεοελληνικής. Θεσσαλονίκη: ΑΠΘ.

(LNEG2) Μπαμπινιώτης, Γ. 2006. Λεξικό της Νέας Ελληνικής Γλώσσας. Κέντρο Λεξικολογίας.

Ξυδόπουλος, Γ. 2008. Λεξικολογία: Εισαγωγή στην ανάλυση της λέξης και του λεξικού. Αθήνα: Πατάκης.

Τράπαλης, Γ. 2005. Λεξικογραφικοί χαρακτηρισμοί στα νεοελληνικά λεξικά: Τυπολογία και συγκριτική εξέταση. Στο Γλώσσης Χάριν. Τόμος αφιερωμένος από τον Τομέα Γλωσσολογίας του ΕΚΠΑ στον καθηγητή Γεώργιο Μπαμπινιώτη. Ελληνικά Γράμματα. 2008. 664-678.

Τράπαλης, Γ. και Γ. Κατσούδα 2007. Συγκριτική μελέτη του συστήματος λεξικογραφικών χαρακτηρισμών σε ελληνικά και ξένα μονόγλωσσα λεξικά. Στα Πρακτικά του 8ου Διεθνούς Συνεδρίου Ελληνικής Γλωσσολογίας. Ιωάννινα. 30/8-2/9 2007.

4 0 Discourse Analysis - Gender - Lexicography Users’ Abilities and Performance in Dictionary Look Up1

Zoe Gavriilidou

Democritus University of Thrace [email protected]

Abstract

The purpose of the present study was to focus on dictionary users’ abilities and performance during receptive dictionary use. Forty eight students of the Democritus University of Thrace were given a worksheet containing fifteen phrases, all including a polysemous word typed in bold characters, and the Dictionary of Standard Greek of Manolis Triandafyllidis. The subjects were told to look up the bold words in the dictionary and write down their meaning as well as the exact position of the entry where the specific meaning was found. The results revealed that users located more easily in dictionaries noun and verb meanings than adjective meanings. It was also found that it was difficult for the students to locate the meaning of phraseologies. These findings support the idea of adopting specific training programs for raising dictionary use awareness.

1. Introduction

Dictionaries are important reference materials which can be used in various circumstances (e.g during reading comprehension, text production, grammar activities, oral discussions, etc) and which can become, under conditions, a valuable learning aid. However, dictionary consultation is a complex process that requires specific skills and strategies. Recent research has studied not only the strategies required for dictionary consultation (Cowie 1999, Fan 2000, Nesi & Haill 2002, Wingate 2004) but also the variables which affect strategy use during dictionary searches such as proficiency level, the type of dictionaries, the

1 I would like to thank the anonymous reviewer for his/her comments which improved the initial text.

Discourse Analysis - Gender - Lexicography 4 1 Major Trends in Theoretical and Applied Linguistics

task performed, etc (Wingate 2004, Nesi & Hua Tan 2011). The present paper focuses on the specific strategies employed during receptive dictionary use.

2. Literature Review

Receptive dictionary use, in other words dictionary use mainly while reading a text and secondarily while listening to an oral text appears to be the most popular use of dictionaries (Béjoint 1981, Cowie 1999, Gavriilidou 2002, Scholfield 2002). According to Scholfield (1999) in receptive situations, an unfamiliar word or phrase is sought in monolingual or bilingual dictionaries, consequently the sole piece of information targeted is meaning. Receptive dictionary use involves five main steps (Bogaards 1993, Scholfield 1999):

• Identification in the text of an unknown word or phrase • Decision to use a dictionary in order to resolve that problem and selection of an appropriate dictionary type • Lemmatization, that is finding the citation form of inflected forms found in the text • Localization of the correct part of the entry where different meanings of the same wordform are included • Integration of the found meaning back in the initial text or other task where the problem arose.

However, a number of problems related with each one of the above mentioned steps may arise during dictionary look up. For instance, very often during receptive dictionary use the users fail to identify the unknown words of a text either because these words have a deceptive morphological structure or they are polysemous words whose one meaning is known to the user but not the one in the current context or they are false friends between two languages or parts of phrasal idioms (Laufer 1997). In other cases, in receptive situations users avoid using dictionaries because they consider them too difficult to deal with. Instead, they simply skip the unknown words or ask the teachers for clarifications or in some cases they try to infer word meaning from context (Hosenfeld 1977). And even when they choose to use a dictionary, they often select an inappropriate type of dictionary or they face problems with lemmatization or localization of the appropriate entry or part of the entry. For instance, users select the first definition in a polysemous entry (Nesi & Haill 2002, Nesi & Hua Tan 2011) or they tend to select familiar segments from the entry (Müllich 1990 cited in Wingate 2004). These errors in receptive use “result in the learner not finding the information needed though in fact it is there, or the learner may end up with some misunderstandings” (Scholfield 1999: 19).

4 2 Discourse Analysis - Gender - Lexicography Zoe Gavriilidou

In order to resolve these problems during receptive use, the dictionary user has to apply in each step different strategies which will help him/her achieve an effective look up.

2.1. Strategies for Unknown Word Identification

Effective readers do not simply classify words into known and unknown; Instead, they bear in mind that some words may look familiar but are in reality false friends, parts of idioms or polysemous words which need to be looked up, otherwise there is a danger of text misunderstandings. Consequently, while processing a text, the reader should first draw a list with the unknown words to be looked up in the dictionary. Once the words are looked up in the dictionary, the reader should return to the text and try a second reading of it by exploiting the information found in dictionary as well the information (s) he considers to be known. If comprehension problems still persist, then (s)he should re-examine the text elements (s)he considers familiar by verifying their meaning in the dictionary.

2.2. Strategies for Deciding When to Use a Dictionary and What Type of Dictionary

Recent research (Gu & Johnson 1996, Nation 2001, Oxford 1990, Schmitt 1997) has shown that dictionary use is an important vocabulary strategy that a) occurs successfully in conjunction with guessing (or inferencing) and note-taking, b) provides information about a specific item, and c) has a positive influence on the learner’s acquisition process (Hulstjin 1993, Luppescu & Day 1993, Knight 1994, Laufer & Hadar 1997, Laufer & Hill 2000, Bruton 2007). However, excessive dictionary use may, on the other, hand inhibit users from developing other important strategies such as guessing, or asking for clarifications. Therefore, users should develop strategies for deciding when dictionary use is the optimal choice. Scholfield (1999), for instance, claims that the importance of an item is a reliable criterion if someone is about to choose whether to look up a word or not. More precisely, unknown words in titles or at the beginning of a text, content words frequently used in a text should be candidates for dictionary searches. As far as the selection of the appropriate dictionary type is concerned, research has mainly focused on the eventual advantages and disadvantages of bilingual vs. monolingual dictionaries. For instance, a lot of researchers consider that bilingual dictionaries can cause errors or problems during text understanding (Ard 1982, Hartmann 1987, Nation 2001, Summers 1995), while others claim that the effectiveness of the bilingual or monolingual dictionaries

Discourse Analysis - Gender - Lexicography 4 3 Major Trends in Theoretical and Applied Linguistics

depends on the nature of task to be solved or the level of the learner (Laufer & Hadar 1997). While bilingual dictionaries are more appropriate for beginners and during translation tasks from mother tongue to the second/ foreign language, monolingual dictionaries are more useful for the advanced level since they provide more detailed information. In any case, users should be aware of the different dictionary types and the specific categories of information found in them through special instruction programs.

2.3. Lemmatization Strategies

Users should be able identify morphological indices (stems, prefixes, suffixes, inflectional morphemes) of the unknown word in order to make hypotheses about the look-up form of that word. To do so, they should first have acquired morphological awareness. In case of oral texts, users have to make hypotheses about possible spellings of the unknown words based on their knowledge of writing conventions of their mother tongue or their second/foreign language. Additionally, dictionary users should acquire basic abilities of alphabetical sequencing through teaching and practice, otherwise lemmatization is not possible.

2.4. Strategies for Appropriate Entry or Subentry Selection and Integration of the Correct Meaning Back in the Initial Text

Nesi & Haill (2002) have shown that dictionary users often accept the first definition given for a polysemous word, even when this is not appropriate in the context. This happens because “the first definition is the first users read, so choosing it shortens their dictionary consultation time, and it is also because the first definition usually represents the most familiar meaning, and is thus most likely to confirm any knowledge they already have about the meaning of the word” (Nesi & Haill 2002: 79). However, skilful users, when they realize, while looking up a word, that the specific word has more than one meaning, they check one by one all the meanings to eliminate the unsuitable one and chose the appropriate one for a given context. Lexical phrases, idioms and complex words, on the other hand, are also difficult to locate in a dictionary, because it is difficult for a user to decide which headword to look them under. Users should be able to make inferences about the dictionary entries in which to look up such items and shouldn’t give up their searches when they cannot find a word in the place they thought it would be. According to Scholfield (1999) dictionary users should be prepared to scan to the end of an entry to find a subentry for compound words or idioms.

4 4 Discourse Analysis - Gender - Lexicography Zoe Gavriilidou

3. Purpose and Rationale

The purpose of the present study was to investigate receptive dictionary use and more precisely dictionary user’s ability to select the correct meaning of a given word as well as the parameters such as speech part, polysemy, and frozenness which affect users’ selection of the appropriate or inappropriate entry.

4. Method

4.1. Participants

Forty eight first year students of the preschool education department ofthe Democritus University of Thrace participated in the study. As students were mainly females and shared the same age, the influence of sex and age were not studied in the present paper.

4.2. Instrumentation and Procedure

Participants were given a worksheet (see appendix I) containing fifteen phrases, all including a polysemous word typed in bold characters, and the Dictionary of Standard Greek (Λεξικό της Κοινής Νεοελληνικής) of Manolis Triandafyllidis. They were told to look up the bold words in the dictionary and then select and write down the meaning of the dictionary which corresponded to the meaning of the word in each phrase of the questionnaire. They were also required to note the exact position of the dictionary entry where the specific meaning was found. There was no time restriction for the completion of the task. The distribution of task items across speech parts is presented in Table 1.

Table 1. Distribution of items across speech parts

SPEECH PARTS N NOUNS 4 VERBS 7 ADJECTIVES 4 TOTAL 15

The items included in the task were the following: two different meanings of the verb επαγγέλομαι, four different meanings of the verb κάνω, αποσπώμαι, εξωτερικός, άγιος, μαύρος in the lexical phrase μαύρη αγορά, αρσενικός, μαϊμού, βήμα, δρόμος, ζήτημα.

Discourse Analysis - Gender - Lexicography 4 5 Major Trends in Theoretical and Applied Linguistics

4.3. Scoring

Each correct search was accredited one point. Correct searches were considered the ones in which the users noted in the worksheet the correct meaning of the item word as well as the right part of the dictionary entry where that meaning was found. Erroneous searches received zero points. Erroneous answers were the ones in which the user either provided an inappropriate meaning of the entry word or did not locate the target word or meaning in the dictionary.

4.4. Data Analysis

Descriptive statistics, such as frequencies, were used to check the percentage of correct or erroneous answers for each item of the experimental task. Comparisons between the correct and erroneous answers for each item were made by using chi-square test. Comparisons of the mean scores between nouns and verbs, nouns and adjectives and verbs and adjectives were made using a paired t-Test with alpha set at .001.

5. Results

Table 2. Frequency of errors in each item.

ITEMS CORRECT WRONG CORRECT % WRONG % X2 Επαγγέλομαι1 47 1 97.9 2.1 44.08* Επαγγέλομαι2 40 8 83.3 16.7 21.33* Κάνω1 30 18 62.5 37.5 3.00 Κάνω2 39 9 81.2 18.8 18.75* Κάνω3 34 14 70.8 29.2 8.33 Κάνω4 46 2 95.8 4.2 40.33* Αποσπώμαι 39 9 81.2 18.8 18.75* Εξωτερικός 17 31 35.4 65.1 4.08 Άγια 35 13 72.9 27.1 10.08 Μαύρη 2 46 4.2 95.8 40.33* Αρσενικός 41 7 85.4 14.6 24.08* Βήμα 43 5 89.6 10.4 30.08* Δρόμος 35 13 72.9 27.1 10.08* Μαϊμού 39 9 81.2 18.8 18.75* Ζήτημα 41 7 85.4 14.6 24.08*

*p<.001

4 6 Discourse Analysis - Gender - Lexicography Zoe Gavriilidou

Table 3. Mean scores for noun, verb and adjective dictionary searches

Part of Speech Mean SD

Noun searches .82 .15

Verb searches .82 .19

Adjective searches .50 .25

The mean score for noun and verb searches was higher than for adjective searches. The paired t-test analysis showed significant differences between noun and adjective searches (t=7.04, p<.001) and verb and adjective searches (t=7.3, p<.001). In other words, users located easier noun and verb meanings than adjective meanings.

6. Discussion

The purpose of the present study was to investigate receptive dictionary use. Results showed that users looked up effectively 10 out of 15 sample words (66%). More precisely the successful word searches concerned the task-items επαγγέλλομαι1, επαγγέλλομαι2, κάνω2, κάνω4, αποσπώμαι, αρσενικός, βήμα, δρόμος, ζήτημα, μαϊμού. All of these words were polysemous but none of them was part of a lexical phrase or idiom. This finding might suggest that users of our sample do not tend to choose the first in the list meaning of a polysemous word. On the contrary they make use of semantic, syntactic or other cues to be guided in the appropriate meaning selection of a polysemous word. Thus, in our study, polysemy did not seem to affect users’ abilities in receptive dictionary use. This result is not consistent with previous research (Nesi & Haill 2002, Neubach & Cohen 1988) who found that subjects tend to take into consideration only the first meaning in dictionary entries of polysemous words. This difference might be explained however by the fact that the sample of our study consisted of Education Department students who had the opportunity to practice dictionary use during their studies. No statistical differences were found between the correct and erroneous searches of the four (26%) following task-items: Κάνω1, κάνω3, εξωτερικός, άγια which means that users provided an almost equal number of successful and unsuccessful dictionary consultations for these words. These task-items either present in context a metaphoric meaning (άγια) or they form lexical phrases (εξωτερικές υποθέσεις) or idioms (κάνω πάταγο). This finding suggests that for our sample the major source of errors in dictionary look-ups were metaphorical meanings, lexical phrases and idioms. This result is consistent with previous

Discourse Analysis - Gender - Lexicography 4 7 Major Trends in Theoretical and Applied Linguistics

research carried out by Wingate (2004) who also found that lexical phrases and idioms caused problems in her sample. The difficulty in locating an idiom or lexical phrase, clearly indicates that the participants in that study had poor knowledge or dictionary conventions. Users failed to locate only one word (6%), the word μαύρη as part of the lexical phrase μαύρη αγορά. The majority of the sample looked up this task-item under the headword μαύρος which is the first element of the lexical phrase and did not manage to find relevant information. Only two subjects searched also under the second element αγορά which is the semantico-syntactic head of the lexical phrase. Bogaards (1990) found differences between Dutch and French learners in their spontaneous approach to the look up strategies of lexical phrases or multiword compounds. Béjoint (1981) reported that learners of English show a preference for searching under the last content word. In any case, the above mentioned result indicates that the users of our sample gave up when they did not manage to find the phrase under the headword they look it up. Finally, the results showed that users located easier noun and verb meanings than adjective meanings. This happened probably because three of our four task adjectives had a metaphorical meaning or belonged to lexical phrases.

7. Conclusions and Further Investigation

The purpose of the present paper was to study receptive dictionary use. Results showed that users encountered problems mainly during look ups of lexical phrases or metaphorical meanings and revealed that the subjects lack basic strategies for dealing with such problems. This finding lends support to the idea of designing special dictionary training programs which would raise dictionary user’s strategy awareness. The present study focused on receptive dictionary use. However, more research is needed into the strategies required not only for receptive but for productive dictionary use as well. Furthermore, the effect of the part of speech on successful dictionary search has to be further investigated with the help of better proportioned test items. Finally, dictionary strategies of Greek users should be studied with a larger sample including subjects with different professional orientation and educational background.

4 8 Discourse Analysis - Gender - Lexicography Zoe Gavriilidou

References

Ard, J. 1982. The use of bilingual dictionaries by ESL students while writing. Review of Applied Linguistics 58: 1-27.

Béjoint, H. 1981. The Foreign Student’s Use of Monolingual English Dictionaries. Applied Linguistics 2(3): 207-22.

Bogaards, P. 1990. Où cherche-t-on dans le dictionnaire? International Journal of Lexicography 3(2): 79-102.

Bogaards, P. 1993. Models of Dictionary Use. Dutch Contributions to AILA ‘93. Amsterdam: Free University: 17-28.

Bruton, A. 2007. Vocabulary learning from dictionary referencing and language feedback in EFL Translational Writing. Language Teaching Research 11: 413-431.

Cowie, B. 1999. English dictionaries for foreign learners. Oxford: Clarendon Press.

Fan, M. 2000. The dictionary look-up behavior of Hong-Kong students: a large scale study. Education Journal 28 (1): 123-138.

Gavriilidou, Z. 2002. Η διερεύνηση των λόγων χρήσης λεξικού ως προϋπόθεση για τη διδασκαλία στρατηγικής χρήσης του λεξικού στην τάξη. In P. Kambaki (ed.), Η διδασκαλία της νέας ελληνικής ως μητρικής γλώσσας (Teaching Greek as L1), Komotini: 45-60.

Gu, P.Y. and Johnson, R.K. 1996. Vocabulary learning strategies and language learning outcomes. Language Learning 46: 643-679.

Hartman, R.R.K. 1987. Four perspectives on dictionary use: A critical review of research methods. In A. Cowie (ed.), The dictionary and the language learner, Tubingen, Max Niemeyer Verlag: 11-28.

Discourse Analysis - Gender - Lexicography 4 9 Major Trends in Theoretical and Applied Linguistics

Hosenfeld, C. 1977. A Preliminary Investigation of the Reading Strategies of Successful and Nonsuccessful Second Language Learners. System 5: 110-23.

Hulstjin, J.H. 1993. When do foreign language readers look up the meaning of unfamiliar words? The influence of task and learner variables. The Modern Language Journal 77(2): 139-147.

Knight, S. 1994. Dictionary use while reading: The effects on comprehension and vocabulary acquisition for students of different verbal abilities. The Modern Language Journal 78: 285-299.

Laufer, B. 1997. The Lexical Plight in Second Language Reading: Words you don’t Know, Words you Think you Know, and Words you can’t Guess. In J. Coady and T. Huckin (eds.), Second Language Vocabulary Acquisition. Cambridge: Cambridge University Press: 20-34.

Laufer, B. and Hadar, L. 1997. Assessing the effectiveness of monolingual, bilingual and ‘bilingualized’ dictionaries in the comprehension and production of new words. The Modern Language Journal, 81: 189-196.

Laufer, B. and Hill, M. 2000. What lexical information do L2 learners select in a CALL dictionary and how does it affect word retention? Language Learning and Technology 3: 58-76.

Luppescu, S. and Day, R.R. 1993. Reading, dictionaries and vocabulary learning. Language Learning, 43: 263-287.

Nation, P. 2001. Learning vocabulary in another language. Cambridge: Cambridge University Press.

Nesi, H. and Haill, R. 2002. A study of dictionary use by international students at a British university. International Journal of Lexicography 15(4): 277-305.

Nesi, H. and Hua Tan, K. 2011. The effect of menus and signposting on the speed and accuracy of sense selection. International Journal of Lexicography 24 (1): 79-96. In\

Neubach, A. and Cohen, A.-D. 1988. Processing strategies and problems in the use of dictionaries. Journal of the dictionary society of North America 10: 1-19.

Oxford, R. 1990. Language Learning Strategies: What every teacher should know. New York: Newbury House Publishers.

5 0 Discourse Analysis - Gender - Lexicography Zoe Gavriilidou

Schmitt, N. 1997. Vocabulary learning strategies. In N. Schmitt and M. McCarthy (eds.), Vocabulary: Description, Acquisition and Pedagogy. Cambridge: Cambridge University Press: 199-227.

Scholfield, P. 1999. Dictionary Use in Reception, International Journal of Lexicography 12(1): 13-34.

Scholfield, P. 2002. Why Shouldn’t Monolingual Dictionaries be as Easy to Use as Bilingual Ones? [Retrieved June 10, 2010, from http://www.longman.com/ dictionaries/teachers/articles/p-scholfield-02.html].

Summers, D. 1995. The Role of Dictionaries in Language Learning. In R. Carter and M. McCarthy (eds.), Vocabulary and Vocabulary Teaching. Harlow and London: Longman: 111-125.

Wingate, U. 2004. Dictionary use - the need to teach strategies. Language Learning Journal, 29: 5-11.

Discourse Analysis - Gender - Lexicography 5 1 Major Trends in Theoretical and Applied Linguistics

Appendix

Φύλο: Ποιο λεξικό έχεις στο σπίτι; Ποιο λεξικό χρησιμοποίησες

ΧΡΗΣΗ ΛΕΞΙΚΟΥ

Αναζητήστε τις μαυρισμένες λέξεις στο λεξικό και καταγράψτε σε ποια σημασία του λεξικού αντιστοιχεί η σημασία που έχει η λέξη στα παρακάτω παραδείγματα

1. Ο αντιπρόσωπος της επιχείρησης επαγγέλλεται τη βελτίωση των συνθηκών εργασίας. 2. Εδώ και αρκετά χρόνια επαγγέλλεται δραστηριότητες που δεν τον ικανοποιούν. 3. Δουλεύει ως εξωτερικός κριτής σε ευρωπαϊκά προγράμματα 4. Η επιδίωξή μας είναι άγια 5. Το αγόρασα 5 ευρώ στη μαύρη αγορά. 6. Είναι μαϊμού δεν είναι ΡΟΛΕΞ. 7. Η εμφάνισή μου έκανε πάταγο. 8. Ήρθε ο ηλεκτρολόγος να κάνει τη βλάβη. 9. Έκανα ένα ωραίο ταξίδι. 10. Κάνει σαν να μην ξέρει τίποτα. 11. Αποσπάστηκε από τους άλλους. 12. Ο θηλυκός παπαγάλος είναι γένους αρσενικού. 13. Δεν κάνει βήμα χωρίς να ρωτήσει. 14. Ακολούθησε το δρόμο του εύκολου κέρδους. 15. Δεν μπόρεσα να απαντήσω στο 3ο ζήτημα.

5 2 Discourse Analysis - Gender - Lexicography Σεξουαλικός προσανατολισμός και γλώσσα

Μαριάνθη Μακρή-Τσιλιπάκου

Αριστοτέλειο Πανεπιστήμιο Θεσσαλονίκης [email protected]

Abstract

This article explores the ways in which sexual orientation is articulated with language (if at all). It foregrounds the taken-for-granted construction of heterosexual identities, as opposed to the much more noticeable homosexual ones —due to the dominance of heteronormativity. A variety of data are utilized to support the claims that i) both heterosexuality and homosexuality need to be interactionally performed, ii) there is considerable overlapping of the relevant linguistic resources utilized in the construction, iii) sex/gender does not necessarily coincide with (the prescribed) sexual orientation and/or sexual practice.

1. Γλώσσα, φύλο και σεξουαλικότητα

Μέχρι πρόσφατα, η μελέτη της γλώσσας και του φύλου, δηλαδή της γυναικείας κοινωνιολέκτου σε αντιδιαστολή με την αντίστοιχη ανδρική (η οποία αποτελεί και την κυρίαρχη νόρμα), αφορούσε υπόρρητα στη γλωσσική συμπεριφορά και συγκρότηση ετεροφυλόφιλων υποκειμένων, στο πλαίσιο της ετεροκανονικότητας,1 η οποία μέσω κοινωνικών και θεσμικών δομών στηρίζει την ετεροφυλοφιλία ως την κυρίαρχη, και επομένως ‘φυσική’ και εντέλει ‘ηθική’, σεξουαλική πρακτική. Όμως, όπως το βιολογικό φύλο (sex) δεν αντιστοιχίζεται πάντοτε με το κοινωνικό (gender) —αφού οι γυναίκες και οι άνδρες αποδεδειγμένα καταλαμβάνουν ποικίλες θέσεις στο συνεχές ανάμεσα στη θηλυκότητα και την αρσενικότητα, με αποτέλεσμα να πρέπει να μιλάμε ακριβέστερα για φύλ-α (βλ. Μακρή-Τσιλιπάκου 2010)—, με τον ίδιο τρόπο, ούτε η έκφραση της σεξουαλικότητας ταυτίζεται στην πράξη με το βιολογικό/ κοινωνικό φύλο. Εκτός από την ‘πολιτικά ορθή’ ετεροφυλοφιλία, υπάρχει και η ομοφυλοφιλία και η αμφιφυλοφιλία, όπως όλες/οι γνωρίζουμε (αν και συχνά επιλέγουμε να το αγνοούμε).

1 “Those structures, institutions, relations and actions that promote and produce heterosexuality as natural, self-evident, desirable, privileged and necessary” (Cameron & Kulick 2003: 149).

Discourse Analysis - Gender - Lexicography 5 3 Major Trends in Theoretical and Applied Linguistics

Επιπλέον, όπως το φύλο δεν αποτελεί ουσία των υποκειμένων —δεν είναι, δηλαδή, μία εγγενής ιδιότητα που υπαγορεύει τις πράξεις όσων φέρουν τη μία ή την άλλη ή την παρ’ άλλη εκδοχή της, αλλά συνιστά κοινωνική/πολιτισμική κατασκευή που απαιτεί καθημερινή και αδιάλειπτη επιτέλεση—, με τον ίδιο τρόπο και ο σεξουαλικός προσανατολισμός χρειάζεται να επιτελεσθεί ώστε να αποδοθεί στα υποκείμενα η αντίστοιχη σεξουαλική ταυτότητα,2 έστω και προσωρινά. Εξάλλου, η σεξουαλικότητα3 δεν είναι απλώς η ερωτική επιθυμία, αλλά η πολιτισμικά διαμεσολαβημένη επιθυμία. Η αντίληψη αυτή άνοιξε τον δρόμο για μια σειρά από μελέτες που διερευνούν τη συμμετοχή της γλώσσας στην κατασκευή και επίτευξη και της σεξουαλικής ταυτότητας, πρωτίστως της ομοφυλόφιλης —λόγω και της γενικότερης πρωτοποριακής θέσης των σχετικών σπουδών ΛΟΑΤ(Δ)4— και δευτερευόντως της ετεροφυλόφιλης ταυτότητας.

2. Η (αδιόρατη) επιτέλεση της ετεροφυλοφιλίας

Αρχίζοντας από την ετεροφυλοφιλία, να επισημάνω και εγώ ότι η καθημερινή επιτέλεσή της περνά απαρατήρητη, ακριβώς λόγω της φυσικοποίησής της. Για να καταδείξει αυτό το γεγονός, η Κitzinger (2005) επανανέλυσε γνωστά δεδομένα από τον χώρο της Ανάλυσης της Συνομιλίας, τα οποία έχουν χρησιμοποιηθεί για τη διατύπωση θεωρητικών εννοιών. Για παράδειγμα, σε απόσπασμα από τις κλήσεις σε Κέντρο Πρόληψης Αυτοκτονιών που ανέλυσε ο Sacks (1972) —διερευνώντας τον Μηχανισμό Κατηγοριοποίησης Μελών ο οποίος, στη βάση των κοινών παραδοχών κάθε κοινότητας, αποδίδει ιδιότητες, χαρακτηριστικά και δραστηριότητες στα πρόσωπα που κατηγοριοποιούνται ανάλογα με την εκάστοτε περίσταση5— η επίκληση του θανάτου της συζύγου του ομιλητή (σειρά 2), και επομένως η συναγόμενη ετεροφυλοφιλία του, αποτελεί συνομιλιακό πόρο για την αιτιολόγηση της κατάθλιψης που βιώνει:

2 “A descriptor for the social framings through which individuals and groups are socially categorized (by themselves or others) based on their sexual orientation, beliefs about their sexuality, and/or their sexual practices” (Queen 2007: 315). 3 “The systems of mutually constituted ideologies, practices, and identities that give sociopolitical meaning to the body as an eroticized and/or reproductive site” (Bucholtz & Hall 2004: 470). 4 Λεσβιακά, Ομοφυλόφιλα Αμφισεξουαλικά, Τρανσεξουαλικά ή Διαφυλικά (άτομα), ελληνική εκδοχή του (L)esbian, (G)ay, (B)isexual, (T)ransgender, (Q)ueer/Questioning. 5 M(embership) C(ategorization) D(evice): “that collection of membership categories, containing at least a category, that may be applied to some population, containing at least a Member, so as to provide, by the use of some rules of application, for the pairing of at least a population Member and a categorization device member. A device is then a collection plus rules of application” (Sacks 1972: 32).

5 4 Discourse Analysis - Gender - Lexicography Μαριάνθη Μακρή-Τσιλιπάκου

(1) 1 Α: Έχετε βιώσει κάποιο προσωπικό πρόβλημα ή δυσκολία; 2 Β: Ναι. Μόλις έχασα τη σύζυγό μου και έχω φοβερή κατάθλιψη. (Sacks 1972: 50)6

Με τον ίδιο τρόπο, η ετεροφυλοφιλία της καταθλιπτικής ομιλήτριας (Β) —η οποία αδυνατεί να βιώσει επιτυχώς μια (ετερόφυλη) ερωτική συνάντηση (σειρές 1, 3, 7)— απορρέει από την ενεργοποίηση συγκεκριμένων κατηγοριών και συναφών δραστηριοτήτων (ώριμη γυναίκα, φανταστικό ραντεβού, ωραίο άνδρα, ποιος άνδρας θέλει…):

(2)7 1 Β: Είμαι μία ώριμη γυναίκα και έχω ένα [φανταστικό ραντεβού με έναν = 2 Α: [Έχεις κανένα 3 Β: = ωραίο άνδρα απόψε και και κάπου αισθάνομαι ότι δεν είμαι τίποτε. ((αναστεναγμός με μειδίαμα)) 4 Α: Μμ 5 Β: Και το ξέρω ότι κανείς δεν είναι τίποτε. Αλλά εγώ είμαι. Σαν να είναι όλοι οι άλλοι κάτι και κάπου εγώ τα σκάτωσα. 6 ((αργότερα)) 7 Β: Και σκέφτομαι ποιος άνδρας θέλει μια νευρωτική άκληρη σαραντάρα. Κανένας. (Sacks 1992a: 66, 68)

Ενώ σε απόσπασμα από τον Schegloff, ο ομιλητής, ταυτόχρονα με την αυτοδιόρθωση (η οποία εξετάζεται εν προκειμένω), αποδίδει και ετεροφυλοφιλία στα πρόσωπα αναφοράς (Μίτι, Γκόρντον) μέσω της συζυγικής σχέσης τους:

(3) 1 Τζο: Ξέρεις ε, η Μίτι- ο Γκόρντον, ε:- ο Γκόρντον, ο άντρας της Μίτι πέθανε. (Schegloff 1997: 519)

Ανάλογα και αναδρομικά, θα μπορούσα εύκολα να παραθέσω πολλά αποσπάσματα και από τη δική μου έρευνα η οποία εξετάζει άλλα φαινόμενα, όπου η ετεροφυλοφιλία των μελών συνάγεται λόγω των δραστηριοτήτων στις οποίες εμπλέκονται (απόσπασμα 4) —τα ζευγάρια πάνε μαζί θέατρο— και των ταξινομικών όρων (απόσπασμα 5) που χρησιμοποιούνται αξιολογικά (παλικάρι, γκόμενα), χωρίς και πάλι να δηλώνεται άμεσα:

6 Η μετάφραση όλων των αγγλικών αποσπασμάτων είναι δική μου. 7 Στην απομαγνητοφώνηση των αποσπασμάτων χρησιμοποιούνται τα εξής σύμβολα: [xxx επικάλυψη, xxx: επιμήκυνση φθόγγου, xxx= συγκόλληση εκφωνημάτων, xxx- αυτοδιόρθωση, >xxx< επιτάχυνση εκφοράς, (.) (1.0) μικρή παύση & παύση δευτερολέπτου, xxx έμφαση, XXX ιδιαίτερη έμφαση, °xxx° χαμηλόφωνη εκφορά, ↑ ανοδική επιτόνιση, !;., επιτονική καμπύλη, ( ) ακατανόητο εκφώνημα, (xxx) δυσνόητο εκφώνημα, ((xxx)) βοηθητικά σχόλια, (…) παράλειψη κειμένου.

Discourse Analysis - Gender - Lexicography 5 5 Major Trends in Theoretical and Applied Linguistics

(4) 1 Γ: Tα παιδιά χτες πήγαν θέατρο Πετράκη, [δεν κοιμόταν. 2 Π: [↑Ποια:; 3 Γ: >Tα παιδιά λέω πήγαν ΘEAτρο< δεν κοιμόταν απ΄ τις έντεκα η ώρα: 4 Π: Ποιος πήγε θέατρο:; 5 Γ: E: η Mαργαρίτα και ο Tάκης. (Μακρή-Τσιλιπάκου 2001: 591)

(5) 1 Κ: Παλικά:ρι τον απεκάλεσε ναι = 2 Π: = Tον είπε παλικάρι γιατί είχε το θάρρος τέλος πάντων να τη: να μας τη πετάξει στα μού:τρα τη τη γκό:μενα! (Μακρή-Τσιλιπάκου 1991: 321)

Ακόμη, η ετεροφυλοφιλία των προσώπων αναδύεται δια της έκφρασης της ερωτικής επιθυμίας (Archakis & Sofia Lampropoulou 2010) —τον βλέπει χωρίς μπλούζα/παντελόνι και πω πω, τη φιλά και χαρά αυτή:

(6) 1 Μ: (…) Έτσι και τον δείτε τον έχω δει και χωρίς μπλούζα πάντως και χωρίς σορτάκι όχι και χωρίς τέτοιο παντελόνι. ((γέλια)) 2 Α: Χωρίς σορτάκι δε ντρέπεσαι; ((γέλια)) 3 Μ: Αχ παιδιά δε κάναμε τίποτα και στη θάλασσα τον είδα και φορούσε τα: (βερμούδα) ( ) βασικά είχε βγάλει τη μπλούζα του στο καρναβάλι ναι ναι ι:::: 4 Φ: Θερμόαιμος. 5 Μ: Πω (πω παιδιά). Και παιδιά στο καρναβάλι μετά από κάνα τρεις βδομάδες με φίλησε χαρά εγώ. ((γέλια)) (Archakis & Lampropoulou 2010: 77, σύντμηση δική μου)

Ενώ, υπάρχουν και περιπτώσεις που δηλώνεται δια της αντιδιαστολής της με την ομοφυλοφιλία, όπως στο παρακάτω απόσπασμα από το τηλεοπτικό Greek Ιdol (Alpha TV, 5/3/2011), με τον υποψήφιο (Υ) να δηλώνει ότι όποιος δεν κοιτάει μια όμορφη γυναίκα (δηλαδή την Έλλη Κοκκίνου) είναι κουνιστός, για να συμφωνήσουν λεκτικά και τα υπόλοιπα μέλη της επιτροπής Κώστας Κα(πετανίδης) και Πέτρος Κω(στόπουλος) —η Κοκκίνου απλώς χαμογελά:

(7) 1 Kα: Φλερτ βλέπω. Με την Κοκκίνου έτσι; (.) Είσαι καρφωμένος στην Έλλη βλέπω. 2 Υ: Ε αφού όμορφη γυναίκα είναι! Να μην την κοιτάς; 3 Kα: Έλα πες τα, έλα πες τα! 4 Υ: Δηλαδή όποιος δεν κοιτάζει την Έλλη, κουνιστός θα είναι. 5 ((επιφωνήματα, γέλια)) 6 Κω: Έχεις δίκιο. Έχεις δίκιο. (http://www.youtube.com/watch?v=XSA4ZCBVhbM&feature=player_embedded)

5 6 Discourse Analysis - Gender - Lexicography Μαριάνθη Μακρή-Τσιλιπάκου

Εξάλλου, ο ομοφοβικός λόγος είναι στοιχείο που οικοδομεί όχι μόνον την ετεροφυλοφιλία αλλά και την αρσενικότητα. Έτσι, πολύ συχνά μάλιστα, οι ενδείκτες του θηλυκού/αρσενικού κοινωνικού φύλου είναι και ενδείκτες αντίστοιχης σεξουαλικότητας. Άλλωστε, πολύ λίγα γλωσσικά στοιχεία σηματοδοτούν άμεσα και αποκλειστικά το φύλο (Ochs 1994: 340), και κατ’ επέκταση και τη σεξουαλικότητα. Και αν και καμία ομιλήτρια και κανείς ομιλητής δεν εισάγουν τη συνεισφορά τους στον λόγο με την προμετωπίδα «Μιλώντας ως ετεροφυλόφιλη/ ετεροφυλόφιλος…», όπως παρατηρεί και η Kitzinger (2005: 222), η ετεροφυλοφιλία τους αναδίδεται μέσω των δημοσίων ταυτοτήτων που ενεργοποιούν, και οι οποίες συνδέονται με πολιτισμικά αναγνωρίσιμες ενασχολήσεις και τρόπους αναφοράς/ κατηγοριοποίησης —ανεξάρτητα από τις μύχιες ερωτικές επιθυμίες τους και τον τυχόν διαφορετικό σεξουαλικό προσανατολισμό που είναι δυνατόν να εκδηλώνουν ιδιωτικά (ό.π.).

3. Ομοφυλοφιλία και γλώσσα: Στερεότυπα και επιτέλεση

Σε σχέση με τα ομοφυλόφιλα υποκείμενα τώρα, και επειδή συνήθως θεωρείται ότι οι ομοφυλόφιλοι και ομοφυλόφιλες ταυτίζονται με το αντίθετο φύλο, θεωρείται επίσης λογικό ότι υιοθετούν και τις αντίστοιχες γλωσσικές συμπεριφορές. Έτσι, κατά τη στερεοτυπική αντίληψη, οι άνδρες ομοφυλόφιλοι τείνουν να πραγματώνουν την κατά Lakoff (1973/1975) γυναικεία γλώσσα: ειδικοί χρωματικοί όροι, ‘κενά’ επίθετα, επισχετικά, πρότυποι γλωσσικοί τύποι, προσεκτική εκφορά, επιμήκυνση φθόγγων (δηλαδή μακρόσυρτη προφορά), δυναμική επιτόνιση (δηλαδή συχνές και έντονες διακυμάνσεις της φωνής)· μαζί με το ειδικό λεξιλόγιο της κοινότητάς τους, που συχνά συμπεριλαμβάνει θηλυπρεπείς ή θηλυκούς μορφολογικούς τύπους για την αναφορά των ομοφυλόφιλων ανδρών στον εαυτό τους ή σε άλλα μέλη, ή για επιδοκιμασία —π.χ. «(Είσαι μεγάλος σταρ) αγάπη μου», «Θεά (τον έκανες τον μουσακά)», «(Τι ήθελα) η τρελή;», «(Είσαι τζασλή) μωρή;», «(Ήμουν πάντα) κουκλίτσα». Ενώ για τις λεσβίες, υπάρχει η πεποίθηση ότι γενικώς αντροφέρνουν, π.χ. βρίζουν, φωνάζουν, έχουν χοντρή φωνή και επίπεδη επιτόνιση, χρησιμοποιούν αρσενικούς μορφολογικούς τύπους και ονόματα, τολμηρή γλώσσα κ.λπ. (πρβ. γυναίκα ανδρούτσος). Τα σχετικά ευρήματα, δεν επιβεβαιώνουν πάντα τις στερεοτυπικές προσδοκίες. Για παράδειγμα, η φωνή των Αμερικανίδων ομοφυλόφιλων γυναικών δεν κινείται αναγκαστικά σε χαμηλότερες συχνότητες (Moonwomon [1985] 1997) —και επομένως δεν αποτελεί κριτήριο διαχωρισμού τους από τις ετεροφυλόφιλες συμπατριώτισσές τους— ούτε των ομοφυλόφιλων ανδρών σε υψηλότερες απ’ ό,τι των αντίστοιχων ετεροφυλόφιλων (Gaudio 1994), ή τουλάχιστον οι διαφορές δεν είναι στατιστικά σημαντικές.

Discourse Analysis - Gender - Lexicography 5 7 Major Trends in Theoretical and Applied Linguistics

Με το ίδιο μέτρο, και στην βάση της ακουστικής πρόσληψης μόνον, θα μπορούσαν κάποιοι ετεροφυλόφιλοι να προσληφθούν ως ομοφυλόφιλοι. Για παράδειγμα, και με τον κίνδυνο να θεωρηθώ ιερόσυλη, θα μπορούσα να επισημάνω ότι ο †Στέλιος Καζαντζίδης, όταν δεν τραγουδούσε, είχε μια καθαρά ‘θηλυκή’ φωνή, διότι και λεπτή ήταν και παρουσίαζε σημαντική επιμήκυνση των φωνηέντων και εξαιρετικά δυναμική επιτόνιση (βλ. Μακρή-Τσιλιπάκου 2001), όπως διαπιστώνεται από τις τηλεοπτικές συνεντεύξεις του. Όμως, είναι γνωστό ότι ο Καζαντζίδης ήταν όχι απλά ετεροφυλόφιλος, αλλά και τύπος ‘πολλά βαρύ’ άνδρα. Αλλά πάλι, η παρατήρηση της πραγματικότητας δείχνει ότι πολλοί άλλοι τέτοιοι ‘πολλά βαρείς’ άνδρες με αρρενωπή φωνή (όπως ο δολοφονημένος ηθοποιός †Νίκος Σεργιανόπουλος) συμβαίνει να είναι και ομοφυλόφιλοι ή έστω αμφίφυλοι. Επομένως, υπάρχει μια γκάμα επιτέλεσης της ομοφυλοφιλίας, που επίσης οικοδομείται και με πραγματολογικές όψεις, όπως η συνεργατική ανάπτυξη θέματος (Leap 1993) —η οποία, όμως, είναι ενδημική και για τις ετεροφυλόφιλες γυναίκες (Makri-Tsilipakou 1991) και, επίσης, απαντάται και ως λεσβιακή στρατηγική (Moonwomon 1994), αλλά καμιά φορά πραγματοποιείται και μεταξύ νεαρών ετεροφυλόφιλων ανδρών (Cameron 1997, Αρχάκης 2008). Επίσης, επιτελείται και θεματικά με την έμμεση ή άμεση παραπομπή στο στίγμα της ομοφυλοφιλίας και την αντίσταση σ’ αυτό (Moonwomon 1994: 522), όπως στην παρακάτω συζήτηση για θεσμικά προβλήματα ιδιοκτησίας και γάμου, τα οποία χαρακτηρίζονται απαξιωτικά:

(8) 1 Μ: Επειδή είμαστε ιδιοκτήτριες κατά το ήμισυ και κάτι τέτοιες μπούρδες. Έχω ιδιοκτησία. Και τα δύο αυτοκίνητα είναι στο όνομα μου για λόγους ασφάλισης. Τέτοιες μαλακίες. Ξες, [ 2 Τζ: [Ναι. Άμα δεν μπορείς να παντρευτείς και θέλεις, ξες [ 3 Μ:  [Σκέτες μπούρδες. Και πού να δεις της Τζην. Της Τζην το όνομα υπάρχει τώρα στον τίτλο ιδιοκτησίας. Φαίνεται κάποιος (.) ηλίθιος στο υποθηκοφυλακείο αποφάσισε ότι είμαστε αδελφές. Και της Τζην το όνομα το κάνανε Τζην Κόεν Μπέθστρονγκ. 4 ((γέλια από όλες)) (Moonwomon 1994: 519-520)

Ακόμη, μπορεί να συναρτάται και με το είδος της δραστηριότητας εντός της οποίας δρουν τα υποκείμενα και το ανάλογο ύφος που απαιτείται. Έτσι, στην πειραματική έρευνα των Smyth, Jacobs & Rogers (2003), το επίσημο ύφος κατά την ανάγνωση επιστημονικού κειμένου από ετεροφυλόφιλους άνδρες συνδέθηκε με την ομοφυλοφιλία —περισσότερο από ό,τι το δραματικό κείμενο και ο αυθόρμητος λόγος—, ίσως επειδή οι ομοφυλόφιλοι άνδρες της έρευνας έτειναν σε γενικευμένη χρήση του για όλα τα είδη λόγου που εξετάστηκαν.

5 8 Discourse Analysis - Gender - Lexicography Μαριάνθη Μακρή-Τσιλιπάκου

3.1. Σε αναζήτηση κοινωνιολέκτου;

Το ερώτημα που τίθεται είναι αν, τελικά, οι ομοφυλόφιλοι/ες με τη γλωσσική τους συμπεριφορά σηματοδοτούν το αντίθετο φύλο και έτσι θα ήταν, ίσως, δυνατόν οι ομοφυλόφιλοι άνδρες να ομαδοποιηθούν με τις ετεροφυλόφιλες γυναίκες και οι ομοφυλόφιλες γυναίκες με τους ετεροφυλόφιλους άνδρες. Ή μήπως συμβαίνει να σηματοδοτούν άμεσα (δηλ. όχι εμμέσως, διαμέσου της αντίθετης αντρικής ή γυναικείας γλώσσας) την ομοφυλόφιλη ταυτότητά τους; Στην περίπτωση αυτή, θα έπρεπε να έχουμε τουλάχιστον τέσσερις διαφορετικές κοινωνιολέκτους. Μια τέτοια κοινωνιόλεκτος ομοφυλόφιλων ανδρών έχει περιγραφεί από τον †Ηλία Πετρόπουλο στην αστική λαογραφική πραγματεία του, Καλιαρντά, που κυκλοφόρησε το (1971) και απαγορεύτηκε από τη χούντα και έστειλε και στη φυλακή τον συγγραφέα της. Τα καλιαρντά βασίζονται στο λαϊκό νεοελληνικό λεξιλόγιο, με κυρίαρχη τη λειτουργία της μεταφοράς, και με δάνεια από την τουρκική, ιταλική, γαλλική, αγγλική και την τσιγγάνικη Ρομ, που συχνά συνδυάζονται και αναπλάθονται με αναγραμματισμούς και αποκοπές, έτσι ώστε το ιδίωμα να καθίσταται ακατανόητο για τους/τις αμύητους/τες. Κάποιες λέξεις αυτής της κοινωνιολέκτου έχουν περάσει σε κοινή χρήση π.χ. τεκνό = αγόρι/νεαρός, τεκνατζού = γυναίκα κάποιας ηλικίας που της αρέσουν οι νεαροί άνδρες, τζους = φύγε, δρόμο!, πουρό = γέρος, μπερντέ = λεφτά, τσουρνέβω = κλέβω κ.ά. Αντίστοιχη είναι και η βρετανική polari/palare (Baker 2002), η οποία χρησιμοποιούνταν συνθηματικά μεταξύ των ομοφυλόφιλων ανδρών ιδίως στην περιοχή του Λονδίνου, πριν την αποποινικοποίηση της ομοφυλοφιλίας προς το τέλος της δεκαετίας του 1960. Η κρυπτική αυτή λειτουργία είναι χαρακτηριστική μιας αντι-γλώσσας, η οποία χρησιμοποιείται εντός μιας κλειστής κοινότητας, π.χ. όπως είναι η φυλακή, και κωδικοποιεί την εναλλακτική κοινωνική οπτική της καθώς και την αντίσταση στην κυρίαρχη κουλτούρα —πρβ. τα ντόρτικα της συντεχνίας των Ρομ χαλκιάδων της Ευρυτανίας (Τριανταφυλλίδης [1923] 1963). Σύμφωνα με τον Halliday (1976), η αντιγλώσσα λειτουργεί ως μέσο επανακοινωνικοποίησης. Ωστόσο, οι γλώσσες αυτές όταν χρησιμοποιούνται έχουν περιορισμένο και συγκεκριμένο είδος χρηστών και δεν αφορούν όλους τους (σύγχρονους) ομοφυλόφιλους και σίγουρα όχι παντού και πάντοτε. Ακόμη, αν σκεφτούμε τις εσωτερικές διαφοροποιήσεις εντός εκάστης κατηγορίας, αλλά και τις επικαλύψεις στη χρήση των στρατηγικών που έχουν διαπιστωθεί σε πολλές έρευνες —κάποιες επισημάνθηκαν νωρίτερα—, τότε αντιλαμβανόμαστε το μάταιο της προσπάθειας συνολικής οριοθέτησης διακριτών κοινωνιολέκτων τόσο για τα φύλα όσο και για τον σεξουαλικό προσανατολισμό, και καταλήγουμε στην κατά περίσταση και περίπτωση διερεύνηση —χωρίς όμως να παραγνωρίζουμε τη δύναμη των σχετικών συμβολικών κατηγοριών, γυναικεία/αντρική/γκέι/λεσβιακή γλώσσα, που αποτελούν το απαραίτητο ιδεολογικό υπόβαθρο και σημείο αναφοράς (βλ. Μακρή-Τσιλιπάκου 2010).

Discourse Analysis - Gender - Lexicography 5 9 Major Trends in Theoretical and Applied Linguistics

3.2. Επαναπλαισίωση και υπονόμευση

Εξάλλου, η έρευνα τείνει προς το συμπέρασμα ότι οι ομοφυλόφιλοι άνδρες και γυναίκες οικειοποιούνται συγκεκριμένα στοιχεία από τις στερεότυπες έμφυλες συμπεριφορές και κατασκευάζουν την ταυτότητά τους, δια της δυναμικής και μοναδικής διεπίδρασης συγκεκριμένων υφολογικών στοιχείων (Queen 1997: 241). Με αυτή την επαναπλαισίωση (recontextualization) των γλωσσικών φαινομένων —τη μεταφορά τους, δηλαδή, από ένα ειδικό περιβάλλον σε ένα άλλο— το κοινωνικό τους νόημα επανασυμβατικοποιείται ως δείκτης ομοφυλοφιλίας (ό.π.: 254). Έτσι, για παράδειγμα, ο ετεροφυλόφιλος μισογυνισμός ορισμένων τρόπων προσφώνησης και περιγραφής των γυναικών μετατρέπεται σε ομοφυλοφιλικό λόγο (Leap 1996: 8) όταν ανταλλάσσεται —συχνά με τη μορφή τελετουργικών αλληλοπροσβολών (ritual insults)— μεταξύ ομοφυλόφιλων ανδρών (“Work bitch!”, ή «Σκάσε, ηλίθια!», που εκσφενδόνισε και ο μουσικός παραγωγός/συγγραφέας Νίκος Μουρατίδης στον ενδυματολόγο Μάρκελλο Νύκτα κατά τη διάρκεια τηλεοπτικής μετάδοσης του Fame Story 3 (ΑΝΤ1 TV, 1/2009, http://www.youtube. com/watch?v=JXstqKIssUQ). Μια τέτοια χρήση επιτρέπει και μια δεύτερη ανάγνωση, όχι ως έκφραση μισογυνισμού, αλλά ως παρωδία του στερεοτυπικού φύλου, το οποίο αποφυσικοποιείται (Kulick 2000: 254, πβλ. Butler 1990/2009). Το ίδιο συμβαίνει και με την οικειοποίηση από τους ομοφυλόφιλους άνδρες των ‘κενών’ επιθέτων της γυναικείας γλώσσας (Lakoff), π.χ. γλυκούλης, χάρμα αλλά και των αντρικών προσφωνήσεων από τις λεσβίες, π.χ. ρε φίλε. Οι (Αμερικανίδες) λεσβίες, επίσης, μπορεί να επιστρατεύσουν ως στρατηγική τη διαστρέβλωση στην εκφορά τυπικά γυναικείων εκφράσεων, π.χ. μετατρέποντας τις αγγλικές διφθόγγους σε απλά φωνήεντα, Oh [όου] my God = O [ό] my God, (Queen 1997: 249), όπως και οι άνδρες μπορεί να κατασκευάσουν την περίσταση ως αυθεντικά ομοφυλοφιλική, παρωδώντας την υποτιθέμενη γνώση που τους αποδίδεται γύρω από τις χρωματικές αποχρώσεις (Leap 1994, 1996) —το χρώμα της κανάτας είναι καφέ, ανοικτό ταμπά, ή σκούρο σοκολατί;:

(9) 1 Α: Να πιω λίγο νερό; ((πλησιάζει τον Μπ. που πλένει πιάτα)) 2 Μπ: Έχει παγωμένο στο ψυγείο. 3 Α: Οκέι. Ευχαριστώ. ((ανοίγει το ψυγείο και κοιτάζει μέσα)) 4 Μπ: ((παρατηρεί την απραξία του Α)) Στην καφέ κανάτα. 5 Α: ((συνεχίζει να ψάχνει∙ κοιτάζει τον Μπ.)) Δεν έχει καμιά καφέ κανάτα εδώ μέσα. 6 Μπ: Έχει. Καφέ, στρογγυλή και στο πρώτο ράφι. 7 Α: ((ξανακοιτάζει)) Δεν έχει. 8 ((ο Μπ. σταματάει το πλύσιμο, σκουπίζει τα χέρια του, πάει στο ψυγείο, βγάζει την κανάτα και βάζει νερό)) 9 Α: Αυτή η κανάτα δεν είναι καφέ, είναι ταμπά (.) Ανοιχτό ταμπά. 10 Μπ: Εγώ για καφέ τη βλέπω. ((μειδίαμα)) 11 Α: Κάτσε, είπες καφέ. Και έψαχνα για κάτι σκούρο σοκολατί. (Leap 1996: 7)

6 0 Discourse Analysis - Gender - Lexicography Μαριάνθη Μακρή-Τσιλιπάκου

Έτσι, η κατάδειξη της μη ηγεμονικής ή εναλλακτικής σεξουαλικότητας μπορεί να γίνει και με τη γελοιοποίηση των ετεροφυλόφιλων ή ομοφυλόφιλων γλωσσικών στερεότυπων —που μπορεί, επίσης, να λειτουργεί ως κριτική στάση απέναντι στην ομο(φυλο)φοβία. Τέτοιο παράδειγμα, αποτελεί η οικειοποίηση της ‘λευκής’ γυναικείας γλώσσας από τους μαύρους Αμερικανούς παρενδυτικούς (drag queens) (Barret 1994), καθώς και η αυτοπαρουσίαση των Ελλήνων χρηστών του διαδυκτιακού χώρου gay.gr, που φιλοξενεί αναζητήσεις ερωτικών συντρόφων, η οποία ενώ μοιάζει να υπακούει στο ετεροφυλόφιλο σεξουαλικό πρότυπο αρσενικό-θηλυκό —με απαξίωση της θηλυπρέπειας, αφού οι ομοφυλόφιλοι αυτοπαρουσιάζονται ως αρρενωποί, αθλητικοί, γυμνασμένοι, γεροδεμένοι κ.λπ. και αναζητούν και αντίστοιχους συντρόφους (απόσπασμα 10)— στην πραγματικότητα, η αυτοπαρουσίαση αυτή οικειοποιείται την αρσενικότητα, αποσυνδέοντάς την από την ετεροφυλοφιλία, όπως διαπιστώνει ο Κανάκης (2008).

(10) Arrenopo, na goustarei ta sport kai ti peripeteia … (thiliprepeis kai gematoi please min mpainete se kopo) (ό.π.: 163)

Για όλους αυτούς τους λόγους, άλλωστε, παρατηρείται ένα συνεχές γλωσσικής συμπεριφοράς ανάμεσα στις τάξεις των ομοφυλοφίλων, όπως τεκμηριώνεται, τουλάχιστον, από τα πρόσωπα που έχουν εκδηλωθεί δημοσίως ή δεν έχουν αρνηθεί τον προσανατολισμό τους όταν δημοσιοποιήθηκε: π.χ. ο Γρηγόρης Βαλλιανάτος (σύμβουλος επικοινωνίας/δημοσιογράφος) —ο οποίος, μάλιστα σε σχετική συνέντευξή του δηλώνει «το θηλυκό γένος εμένα δεν με εκφράζει, δεν χρησιμοποιώ γυναικεία ονόματα. Αυτά τα χρησιμοποιούν οι μητροπολίτες και οι παλιοί ομοφυλόφιλοι» (Έψιλον της Κυριακάτικης Ελευθεροτυπίας, Νο 732, σελ. 44, 24/4/05)· o Χρήστος Σιμαρδάνης (ηθοποιός), ο Φώτης Σεργουλόπουλος (παρουσιαστής), ο Κωνσταντίνος Γιάνναρης (σκηνοθέτης), ο Δημήτρης Παπαϊωάννου (ζωγράφος-χορογράφος) —που εξηγεί ότι «το gay δεν είναι ιδιότητα, προτέρημα ή παράσημο. Είναι η ερωτική σου επιλογή. Δεν σημαίνει κάτι αξιόλογο για τον άνθρωπο» (http://www.luvu.gr/index.php?option=com_content& view=article&id=410&Itemid=30)· ο Γιώργος Μαρίνος (ηθοποιός), που ‘αντρίκια’ δήλωσε τον κύριο σεξουαλικό προσανατολισμό του σε δύσκολους καιρούς, ο †Γιάννης Τσαρούχης (ζωγράφος), ο †Μάνος Χατζηδάκις (συνθέτης), ο †Κώστας Ταχτσής (λογοτέχνης), ο Ντίνος Χριστιανόπουλος (ποιητής), ο Παναγιώτης Χατζηστεφάνου (εικαστικός καλλιτέχνης), η Τζένη Χειλουδάκη (μανεκέν-ιερόδουλος), η Χριστίνα Μπαλτζή (Καρλότα/Δήμος, τρανσέξουαλ), η Μίνα Ορφανού (πρωταγωνίστρια της ταινίας Στρέλλας του Κούτρα) κ.ά.

3.3. Η αθέατη γυναικεία (ομο)σεξουαλικότητα

Για τη γλώσσα των Ελληνίδων λεσβιών, τώρα, δεν έχουμε ακόμη κανένα επιστημονικό τεκμήριο —αν και έχουμε κάποιες ανθρωπολογικές μελέτες σε ένα πρόσφατο βιβλίο (Canakis, Kantsa & Yannakopoulos 2010)— και ούτε μπορούμε να αναφερθούμε σε

Discourse Analysis - Gender - Lexicography 6 1 Major Trends in Theoretical and Applied Linguistics

γνωστές σαπφίδες (όπως επίσης αυτοαποκαλούνται, http://sapphogr.net/sl/home.html), για τον απλούστατο λόγο ότι ελάχιστες έχουν εκδηλώσει δημόσια την προτίμησή τους: όπως η Ευαγγελία Βλάμη, μέλος της Ομοφυλόφιλης και Λεσβιακής Κοινότητας Ελλάδας που έγινε γνωστή από τις τηλεοπτικές παρεμβάσεις της για το ζήτημα του γάμου μεταξύ ομοφύλων· η ραδιοφωνική παραγωγός/ακτιβίστρια Μαρία Cyber (Κατσικαδάκου), η κοινωνιολόγος Σωτηρία Θεοχάρη, η σκηνοθέτιδα Χριστίνα Καλλινίκου (ονόματα που ανακάλυψα στην ιστοσελίδα http://sapphogr.net/les/afierwma/index.html). Είναι προφανές ότι το ταμπού απέναντι στην γυναικεία ομοφυλοφιλία —για να μην πούμε στη σεξουαλικότητα των γυναικών που βρίσκεται σε διαρκή υποτίμηση8— δεν έχει ακόμη υποστεί ρωγμές ανάλογες με εκείνες της ανδρικής. Ένας επιπλέον λόγος είναι το γεγονός ότι οι λεσβίες που έχουν κοινωνική προβολή αποκρύπτουν επιμελώς την προτίμησή τους, υποκύπτοντας στη συνομωσία της σιωπής που προκαλεί ο φόβος του λεσβιακού στίγματος —ίσως επειδή έχουν, επίσης, υπόψη την πορνογραφική αντιμετώπιση της λεσβιακής ερωτικής δραστηριότητας, η οποία λειτουργεί ως αφροδισιακό για τα ετεροφυλόφιλα αρσενικά. Η στάση αυτή προωθείται και εκ μέρους των ΜΜΕ που δεν χάνουν την ευκαιρία να προβάλλουν κάθε φορά ό,τι σχετικό (π.χ. λεσβιακό φιλί της Μαντόνα με την Μπρίτνεϊ Σπίαρς, της Μπρίτνεϊ Σπίαρς με τη Ριάνα κ.λπ.). Κάτι ανάλογο, με μία εφήμερη ριαλιτοδιασημότητα σε ‘αρπαχτή’ στην επαρχία να επιδίδεται σε σκηνοθετημένες ερωτικές περιπτύξεις με μια άλλη γυναίκα προς τέρψη των παρισταμένων, διαφημίστηκε σε ιντερνετικό blog (http://www.tempo.gr/ston-kosmo-mas/gossip-celebrities/19470- elina-mpekakou), το οποίο, επίσης, μας πληροφορεί ότι «οι καλεσμένοι είχαν την ευκαιρία να απολαύσουν τα παιχνίδια των 2 φιλενάδων» —αν επρόκειτο για παιχνίδια 2 φίλων μάλλον δεν θα το απολάμβαναν οι καλεσμένοι! Εν πάση περιπτώσει, είναι σίγουρο ότι η καθεμία και ο καθένας μας γνωρίζει κάποιες ομοφυλόφιλες γυναίκες, είτε άσημες είτε διάσημες (τραγουδίστριες, ηθοποιούς, στιχουργούς, μουσικούς, βουλευτίνες κ.λπ.), και μπορεί να βγάλει τα ανάλογα συμπεράσματα, για την γκάμα των γλωσσικών (και άλλων) συμπεριφορών τους.

4. Φύλα, σεξουαλικότητες και πρακτικές

Συνοψίζοντας, να επισημάνω και πάλι το συνεχές των φύλων (που δεν είναι μόνο δύο) καθώς και της σεξουαλικότητας, τα οποία είναι δυνατόν να συναρθρώνονται με ποικίλους τρόπους παρά την κυρίαρχη άποψη, γεγονός που μας προτρέπει εντέλει να αποσυνδέσουμε το βιολογικό από το κοινωνικό φύλο και τα δύο από τον σεξουαλικό προσανατολισμό. Και σε τελευταία ανάλυση, και από τη σεξουαλική πρακτική

8 Σημείο στο οποίο συμφωνήσαμε απολύτως με τη Σταυρούλα Τσιπλάκου κατά τη συνεδριακή παρουσίαση.

6 2 Discourse Analysis - Gender - Lexicography Μαριάνθη Μακρή-Τσιλιπάκου

αφού οι ομοφυλοφιλικές σωματικές πράξεις δεν αξιολογούνται πάντα ως τεκμήριο ομοφυλοφιλίας/ομοφυλόφιλης ταυτότητας —άλλωστε, η έννοια της σταθερής ‘δια βίου ομοφυλοφιλίας’ αποτελεί ιστορικά πρόσφατη δυτική επινόηση (Greenberg 1988). Για παράδειγμα ο εικοσιπεντάχρονος (Α)πόστολος Α. που εμφανίστηκε στη Στιγμή της Αλήθειας με παρουσιάστρια την Ευγενία (Μ)ανωλίδου (ΑΝΤ1, 1/5/2009) —παρά το γεγονός ότι έχει ήδη παραδεχτεί ότι η πρώτη του επαφή ήταν με τραβεστί, δεν έχει κάνει σχέση με γυναίκα, και έχει υποκύψει σε ερωτική πρόταση από άνδρα για να του βρει δουλειά— δηλώνει ότι «τρελάθηκε όταν τον είπανε γκέι (…) και τα ‘σπασε»:

(11) 1 Μ: Σε έχουν χλευάσει ποτέ στον εργασιακό σου χώρο για τη σεξουαλική σου ταυτότητα; 2 ((μεσολαβεί μουσική που επιτείνει την αγωνία)) 3 Α: Ναι. 4 ((φωνή εκτός οθόνης)) Αυτό που απάντησες, είναι (.) αλήθεια. 5 ((χειροκροτήματα)) 6 Μ: (Τι έγινε;) 7 (1.0) 8 Α: Τς. Απλά με: απλά με είχανε πει γκέι (.) και τρελάθηκα. 9 (1.0) 10 Μ: Υπερασπί [στηκες τον εαυτό σου = 11 Α: [Κι έγινε: 12 Α: = Ναι. Τρελάθηκα και τα ‘σπασα °που λεει ο λόγος:°. (http://www.youtube.com/watch?v=OyU-ywWq8co)

Αναλόγως, κατά τα άλλα ετεροφυλόφιλοι Νεοέλληνες ‘ψωνίζουν’ στη Συγγρού της κάθε πόλης. Και αν αυτά τα περιστατικά είναι δυνατόν να ερμηνευθούν και ως απώθηση μιας κοινωνικά οδυνηρής παραδοχής για τα πρόσωπα που τα βιώνουν (πβλ. «υπερασπίστηκες τον εαυτό σου» της Μανωλίδου στο προηγούμενο απόσπασμα), ας θυμηθούμε και τους αρχαίους Έλληνες που ασκούσαν ‘παιδαγωγικό έρωτα’, με τον μέντορα/εραστή να διαμορφώνει πολιτισμικά τον νεότερο ερωμένο, ενώ διήγαν ετεροφυλόφιλο βίο (βλ. Neill 2009).9 Και αν πάμε και μέχρι τη Μελανησία θα εντοπίσουμε αυτό που μερικοί ανθρωπολόγοι περιγράφουν ως ‘τελετουργική ομοφυλοφιλία’ (Herdt 1984), με τους ώριμους άνδρες να μυούν τους νεότερους μέσω της πεολειχίας.

9 Για μια περιγραφή της ομοφυλοφιλίας στην αρχαία Ελλάδα, βλ. Αριστοφάνη Νεφέλες (949-1113) —όπου οι παίδες (μαζί με τις γυναίκες, τα παιχνίδια, το φαγητό, το ποτό, το γέλιο) συγκαταλέγονται στις ηδονές (ἡδονῶν θ᾿ ὅσων μέλλεις ἀποστερεῖσθαι· παίδων, γυναικῶν, κοττάβων, ὄψων, πότων, καχασμῶν) και οι περισσότεροι συνήγοροι, τραγικοί, πολιτικοί προέρχονται Ἐξ εὐρυπρώκτων, διότι αυτοί αποτελούν την πλειοψηφία: Πολὺ πλείονας, νὴ τοὺς θεούς, τοὺς εὐρυπρώκτους (http://www.mikrosapoplous.gr/texts1.htm).

Discourse Analysis - Gender - Lexicography 6 3 Major Trends in Theoretical and Applied Linguistics

Για τις πρακτικές αυτού του είδους υποστηρίζεται ότι, σε τελευταία ανάλυση, μπορεί και να μη συνδέονται με τη σεξουαλικότητα —όπως, τουλάχιστον την εννοούμε στον σύγχρονο ή/και δυτικό κόσμο— και επομένως με την ετεροφυλοφιλία/ομοφυλοφιλία, αλλά νοηματοδοτούνται διαφορετικά εντός των συγκεκριμένων κοινωνιών. Για παράδειγμα, μια μερίδα ανθρωπολόγων συμπεραίνει ότι το σπέρμα των Μελανήσιων ανδρών μεταβιβάζεται στα νέα αγόρια ως θρεπτική ουσία, αντίστοιχη —και σε αντιπερισπασμό— με το γάλα των γυναικών (Elliston 1995/2006: 196), από τις οποίες πρέπει να διαφυλαχτεί η εξουσία και ανωτερότητα των ανδρών (: 194). Με άλλα λόγια, κάποιες φαινομενικά σεξουαλικές πρακτικές αφορούν, στην ουσία, ηλικιακές και έμφυλες ιεραρχίες και συμμετέχουν στη συγκρότηση κοινωνικών ταυτοτήτων (: 171). Και για να προσγειωθούμε στα καθ’ ημάς, ο βιασμός πολύ λίγη σχέση έχει με τη σεξουαλικότητα/επιθυμία, αλλά αποτελεί μια ωμή πράξη βίας/κυριαρχίας. Εντέλει, οι συνδυασμοί φύλων/σεξουαλικοτήτων/πρακτικών που προκύπτουν είναι καμία φορά εντελώς απρόβλεπτοι. Για παράδειγμα, φανταστείτε, αν μπορείτε, την πραγματική περίπτωση ενός προσώπου (Αμερικανού) που βιολογικά γεννιέται άνδρας, μεγαλώνοντας αποφασίζει ότι είναι διαφυλικός/ή (τρανσέξουαλ) και μάλιστα υποβάλλεται και σε εγχείρηση αφαίρεσης του ανδρικού του μορίου, έτσι ώστε να είναι ελεύθερη να ερωτευθεί —όχι όμως άνδρες, όπως θα υποθέταμε, αλλά γυναίκες και μάλιστα λεσβίες.10 Ξεκινά δηλαδή ως βιολογικός άνδρας για να καταλήξει σεξουαλικά ομοφυλόφιλος/η και κοινωνικά γυναικείου φύλου. Διότι, όπως πολύ εύγλωττα το θέτει και ο ποιητής Ευγένιος Αρανίτσης, τελικά φαίνεται ότι: «Αντίθετα από απ’ ό,τι οι σκύλοι [= τα ζώα] (…) το σώμα του ανθρώπου είναι το Έχειν του όχι το Είναι του, ένα Έχειν που αμφισβητείται διαρκώς» (Κυριακάτικη Ελευθεροτυπία, 21/11/2004, σελ. 7).

10 Την ιστορία μου αφηγήθηκε άτομο το οποίο συνέζησε μαζί του/της αγνοώντας το πλήρες βιογραφικό του/της.

6 4 Discourse Analysis - Gender - Lexicography Μαριάνθη Μακρή-Τσιλιπάκου

Βιβλιογραφία

Αρχάκης, Α. 2008. Διερευνώντας την κατασκευή ταυτότητας (φύλου) σε νεανικές συνομιλίες: Παρατηρήσεις και προεκτάσεις. Στο Ε. Λεονταρίδη και Κ. Σπανοπούλου (επιμ.), Η γλώσσα σε έναν κόσμο που αλλάζει. Πρακτικά Διεθνούς Συνεδρίου. Διδασκαλείο Ξένων Γλωσσών του Εθνικού και Καποδιστριακού Πανεπιστημίου Αθηνών, 9-11 Δεκεμβρίου 2005. Αθήνα: Διδασκαλείο Ξένων Γλωσσών Πανεπιστημίου Αθηνών, 94-100.

Archakis, A. and S. Lampropoulou. 2010. Narrativising the body as a stimulus of desire: Evidence from Greek youth storytelling. In C. Canakis, V. Kantsa and K. Yannakopoulos (eds), Language and Sexuality (through and) beyond Gender. Cambridge Scholars Publishing, 67-85.

Baker, P. 2002. Polari: The Lost Language of Gay Men. London: Routledge.

Barret, R. 1994. Appropriation of white women’s language by drag queens. In M. Bucholtz, A.C. Liang, L.A. Sutton and C. Hines (eds), Cultural Performances: Proceedings of the Third Berkeley Women and Language Conference, April 8-10, 1994. Berkeley, CA: BWLG, 1-14.

Bucholtz, M. and K. Hall. 2004. Theorizing identity in language and sexuality research. Language in Society 33: 469-515.

Butler, J. 1990. Gender Trouble: Feminism and the Subversion of Identity. New York/ London: Routledge.

Butler, J. 2009. Αναταραχή φύλου: Ο φεμινισμός και η ανατροπή της ταυτότητας, μετάφραση Γ. Καράμπελας, εισαγωγή και επιστημονική επιμέλεια Β. Καντσά. Αθήνα: Αλεξάνδρεια.

Cameron, D. 1997. Performing gender identity: Young men’s talk and the construction of heterosexual masculinity. In S. Johnson and U.H. Meinhof (eds), Language and Masculinity. Oxford: Blackwell, 47-64.

Discourse Analysis - Gender - Lexicography 6 5 Major Trends in Theoretical and Applied Linguistics

Cameron, D. and D. Kulick. 2003. Language and Sexuality. Cambridge, England: Cambridge University Press.

Canakis, C., V. Kantsa and K. Yannakopoulos (eds). 2010. Language and Sexuality (through and) beyond Gender. Newcastle upon Tyne: Cambridge Scholars Publishing.

Elliston, D. 1995. Erotic anthropology: “Ritualized homosexuality” in Melanesia and beyond. American Ethnologist 22(4): 848-867.

Elliston, D. 2006. Ανθρωπολογία του ερωτικού. Στο Κ. Γιαννακόπουλος (επιμ.), Σεξουαλικότητα: Θεωρίες και πολιτικές της σεξουαλικότητας. Αθήνα: Αλεξάνδρεια, 169- 217.

Gaudio, R.R. 1994. Sounding gay: Pitch properties in the speech of gay and straight men. American Speech 69(1): 30-57.

Greenberg, D.F. 1988. The Construction of Homosexuality. Chicago: University of Chicago Press.

Halliday, M.A.K. 1976. Anti-languages. American Anthropologist 78(3): 570-584.

Herdt, G.H. 1984. Ritualized Homosexuality in Melanesia. Berkley: University of California Press.

Κανάκης, Κ. 2008. Γλώσσα, αρσενικότητα και σεξουαλικότητα στο διαδίκτυο. Μελέτες για την Ελληνική Γλώσσα 28. Θεσσαλονίκη: Ινστιτούτο Νεοελληνικών Σπουδών (Ίδρυμα Μανόλη Τριανταφυλλίδη), 159-170.

Kitzinger, C. 2005. “Speaking as a heterosexual”: (How) does sexuality matter for talk- in-interaction? Research on Language and Social Interaction 38(3): 221-265.

Kulick, D. 2000. Gay and lesbian language. Annual Review of Anthropology 29: 243-285.

Lakoff, R. 1973. Language and woman’s place. Language in Society 2: 45-79.

Lakoff, R.T. 1975. Language and Woman’s Place. New York: Harper & Row.

Leap, W. 1993. Gay men’s English: Cooperative discourse in a language of risk. New York Folklore 19(1-2): 45-70. Special issue: Prejudice and Pride.

Leap, W. 1994. Can there be gay discourse without gay language? In M. Bucholtz, A.C. Liang, L.A. Sutton and C. Hines (eds), Cultural Performances: Proceedings of the Third

6 6 Discourse Analysis - Gender - Lexicography Μαριάνθη Μακρή-Τσιλιπάκου

Berkeley Women and Language Conference, April 8-10, 1994. Berkeley, CA: BWLG, 399-408.

Leap, W.M. 1996. Word’s Out: Gay Men’s English. Minneapolis: University of Minnesota Press.

Makri-Tsilipakou, M. 1991. Agreement/Disagreement: Affiliative vs. Disaffiliative Display in Cross-Sex Conversations. Διδακτορική διατριβή. Tομέας Θεωρητικής και Eφαρμοσμένης Γλωσσολογίας. Τμήμα Αγγλικής Γλώσσας και Φιλολογίας, Φιλοσοφική Σχολή, ΑΠΘ.

Mακρή-Tσιλιπάκου, M. 2001. O μελοδραματικός σχεδιασμός στον καθημερινό λόγο. Στο Σ. Πατσαλίδης και A. Nικολοπούλου (επιμ.), Mελόδραμα. Eιδολογικοί και ιδεολογικοί μετασχηματισμοί. Θεσσαλονίκη: University Studio Press, 579-619.

Μακρή-Τσιλιπάκου, Μ. 2010. Η «γυναικεία γλώσσα» και η γλώσσα των γυναικών. Στο Β. Καντσά, Β. Μουτάφη και Ε. Παπαταξιάρχης (επιμ.), Φύλο και κοινωνικές επιστήμες στη σύγχρονη Ελλάδα. Αθήνα: Αλεξάνδρεια, 119-146.

Moonwomon, B. 1994. Lesbian identity, lesbian text. In M. Bucholtz, A.C. Liang, L.A. Sutton and C. Hines (eds), Cultural Performances: Proceedings of the Third Berkeley Women and Language Conference, April 8-10, 1994. Berkeley, CA: BWLG, 509-524.

Moonwomon, B. [1985] 1997. Towards a study of lesbian speech. In A. Livia and K. Hall (eds), Queerly Phrased: Language, Gender and Sexuality. NewYork/Oxford: Oxford University Press, 202-213.

Neill, J. 2009. The Origins and Role of Same-Sex Relations in Human Societies. Jefferson, NC/ London: McFarland & Company, Inc., Publishers.

Ochs, E. 1992. Indexing gender. In A. Duranti and C. Goodwin (eds), Rethinking Context: Language as an Interactive Phenomenon. Cambridge: Cambridge University Press, 335- 358.

Πετρόπουλος, Η. [1971] 1993. Καλιαρντά. Αθήνα: Νεφέλη.

Queen, R. 1997. “I don’t speak spritch”: Locating lesbian language. In A. Livia and K. Hall (eds), Queerly Phrased: Language, Gender, and Sexuality. New York: Oxford University Press, 233-256.

Queen, R. 2007. Sociolinguistic horizons: Language and sexuality. Language and Linguistics Compass 1(4): 314-330.

Discourse Analysis - Gender - Lexicography 6 7 Major Trends in Theoretical and Applied Linguistics

Sacks, H. 1972. An initial investigation of the usability of conversational data for doing Sociology. In D. Sudnow (ed.), Studies in Social Interaction. New York/London: The Free Press/Collier-Macmillan, 31-74.

Sacks, H. 1992. Lectures on Conversation, 2 vols, ed. G. Jefferson, with introductions by E.A. Schegloff. Oxford: Blackwell.

Schegloff, E.A. 1997. Practices and actions: Boundary cases of other-initiated repair. Discourse Processes 23: 499-545.

Smyth, R., G. Jacobs and H. Rogers. 2003. Male voices and perceived sexual orientation: An experimental and theoretical approach. Language in Society 32: 329-350.

Τριανταφυλλίδης, Μ. [1923] 1963. Tα ντόρτικα της Eυρυτανίας (Συμβολή στα ελληνικά «μαστόρικα»). Άπαντα Mανόλη Tριανταφυλλίδη, 2ος τόμ.. Θεσσαλονίκη: Ινστιτούτο Νεοελληνικών Σπουδών [Ίδρυμα Μανόλη Τριανταφυλλίδη], ΑΠΘ, 33-45.

6 8 Discourse Analysis - Gender - Lexicography Αναλύοντας τις διαδρομές των κειμένων: Εθνογραφική μελέτη παραγωγής ειδήσεων σε μια τοπική εφημερίδα

Δημήτρης Ντανόπουλος

Αριστοτέλειο Πανεπιστήμιο Θεσσαλονίκης [email protected]

Abstract

The present study examines the practices employed in a local newspaper during the production of news. A reporter developed a dialogue with texts deriving from the local community. The interaction relationships among people were recorded in the form of an Activity System and shaped the “routes” of texts. These “routes” and the final texts echoed the ideology of the local media community and were materialized through specific writing strategies. The reporter attained these strategies throughout an interactive process between his own identity and the community identity.

1. Εισαγωγή

Γράφοντας ο ίδιος κείμενα για εφημερίδες και περιοδικά, πολλές φορές διαπίστωνα πως το τελικά δημοσιευμένο κείμενο δεν είχε ακριβώς τη μορφή που είχα επιλέξει. Αρχισυντάκτες, γραφίστες, διευθυντές, εκδότες είχαν μεσολαβήσει. Τα ποικίλα μέσα ενημέρωσης παράγουν προϊόντα που δεν είναι απλές μεταβιβάσεις πληροφοριών. Η περιγραφή του κόσμου γίνεται μέσω ενός σύμπαντος διαμεσολαβητών, των οποίων η ουδετερότητα αμφισβητείται και η πρόθεση είναι συστατικό στοιχείο της ύπαρξής τους. Η παραγωγή του γραπτού λόγου των δημοσιογράφων αποτελεί πεδίο ανταγωνισμού, συγκρούσεων, αλληλεπίδρασης φωνών και διαπραγμάτευσης ταυτοτήτων.

1.1. Βιβλιογραφική επισκόπηση

Από νωρίς, στις αρχές του 20ου αιώνα άρχισαν να καταγράφονται έρευνες σχετικές με τα έντυπα μέσα (Jankowski 2002). Στον 21ο αιώνα, σε μια εποχή που ο Τύπος αναζητά τη θέση του σε ένα συνεχώς διαφορετικά διαμορφούμενο κόσμο ενημέρωσης, το ερευνητικό ενδιαφέρον μετατοπίζεται σταδιακά.

Discourse Analysis - Gender - Lexicography 6 9 Major Trends in Theoretical and Applied Linguistics

Ο λόγος των μέσων απασχολεί ευρύτερα πεδία επιστημών από αυτό των Media Studies, πεδία που ενδιαφέρονται για ζητήματα, όπως η γλώσσα κι η επικοινωνία, η προσφορά έτοιμων δεδομένων για έρευνα και διδασκαλία, η χρήση της επιρροής των μέσων στη στάση των αναγνωστών έναντι της γλώσσας και τη χρήση του λόγου της κοινότητας, καθώς και η επιρροή των μέσων στην αναπαραγωγή αλλά και διαμόρφωση της κουλτούρας, της πολιτικής και κοινωνικής ζωής της κοινότητας (Garrett & Bell 1998: 3-5). Ιδιαίτερα σημαντική ήταν η επίδραση της Κριτικής Γλωσσολογίας (Critical Linguistics) που αναπτύχθηκε τη δεκαετία του 1970 και στηρίχθηκε στην ιδεολογική ανάλυση των γραμματικών και λεξιλογικών χαρακτηριστικών κυρίως των γραπτών κειμένων (Fairclough 1989: 246 ; 1995: 25). Παρόλο που τα κείμενα των εφημερίδων έχουν γίνει αντικείμενο μελέτης, αυτή περιορίστηκε στη στατική διερεύνησή τους ως τελικά προϊόντα. Στη διεθνή βιβλιογραφία καταγράφεται ένα επιστημονικό κενό στην έρευνα παραγωγής δημοσιογραφικών κειμένων (media texts) (van Dijk 1988: 22; Garrett & Bell 1998: 19), γι’ αυτό κι οι μελέτες πλέον επικεντρώνουν το ενδιαφέρον τους στην παρακολούθηση της παραγωγής του δημοσιογραφικού λόγου: “newsroom ethnography” (Van Hout 2011).

1.1.1.Το ελληνικό παράδειγμα Όσον αφορά στο λόγο των ελληνικών μέσων, αυτός χαρακτηρίζεται ως «αχαρτογράφητη επικράτεια» (Πολίτης 2008: 9). Τα τελευταία χρόνια δημοσιοποιήθηκαν γλωσσολογικές αναλύσεις που είχαν κύρια περιγραφικό χαρακτήρα (π.χ. τυπολογία δημοσιογραφικού λόγου) ή, έχοντας κοινωνιογλωσσολογική οπτική, ασχολήθηκαν με μεμονωμένα γλωσσολογικά στοιχεία κειμένων των εφημερίδων (π.χ. ο λόγος των κοινωνικών θεσμών ή των δημοσιογράφων και η προσαρμογή του λόγου τους με βάση το αναγνωστικό κοινό). Δεν εστίαζαν, ωστόσο, στον τρόπο που δουλεύουν οι δημοσιογράφοι (Χατζησαββίδης 1999; Πολίτης 2008).

1.2. Ο στόχος της παρούσας έρευνας

Με την παρούσα εργασία θελήσαμε να χαρτογραφήσουμε τη θέση ενός δημοσιογράφου- συντάκτη μιας τοπικής εφημερίδας μέσα σε ένα πλέγμα ιδεολογικών αλληλεπιδράσεων, που προσδιοριζόταν από το εργασιακό – κοινωνικό του περιβάλλον κατά τη διαδικασία παραγωγής των κειμένων. Στη σημερινή μας παρουσίαση θα εστιάσουμε στη διαπραγμάτευση κατασκευής εικόνων της ταυτότητας μιας πόλης.

2. Μεθοδολογία

Με δεδομένη την κοινωνική πλαισίωση του γραπτού λόγου (τη θέση δηλαδή ότι ο γραπτός λόγος διαμορφώνεται μέσα από ποικίλα πλαίσια επικοινωνίας), χρησιμοποιήσαμε

7 0 Discourse Analysis - Gender - Lexicography Δημήτρης Ντανόπουλος

ένα μοντέλο κριτικής ανάλυσης πρακτικών παραγωγής γραπτού λόγου σε μια δημοσιογραφική κοινότητα, συνενώνοντας έννοιες και τρόπους μελέτης της γλώσσας που παρουσιάζονται σε τρεις διαφορετικές προσεγγίσεις: -την Κριτική Ανάλυση Λόγου (ΚΑΛ), -την Κοινωνιογνωστική έρευνα για την παραγωγή γραπτού λόγου (Στρατηγικές Γραφής) και -την Εθνογραφική προσέγγιση του γραπτού λόγου.

2.1. Κριτική Ανάλυση Λόγου (Critical Discourse Analysis)

Πρόθεσή μας ήταν να μετακινήσουμε την εστίαση της έρευνας από τον αποσπασματικό λόγο στην κοινότητα λόγου. Χρησιμοποιήσαμε ως βάση γι’ αυτό την Κριτική Ανάλυση Λόγου (ΚΑΛ).

2.1.1. Το αρχικό μοντέλο του Fairclough Ο θεμελιωτής της ΚΑΛ, Fairclough, εκλαμβάνει το κείμενο ως “discourse” (λόγο), δηλαδή ως τρόπο αναπαράστασης ενός ξεχωριστού μέρους του κόσμου ή μιας περιοχής γνώσης και/ή εμπειρίας που παρουσιάζεται ως μέρος της κοινωνικής δραστηριότητας μιας κοινωνικής πρακτικής (Fairclough 1995: 18, 54, 41, 66 ; 2003: 129, 205- 7). Ενδιαφερθήκαμε για τη θεωρητική εξέλιξη εννοιών του αρχικού μοντέλου του Fairclough:

2.1.2. discourse – order of discourse – Discourse Βγαίνοντας από τα στενά όρια της γλωσσολογικής ανάλυσης του κειμένου, ο Fairclough υιοθέτησε τον μεταστρουκτουραλιστικό όρο “order of discourse” του Foucault (Scollon, Yung, Tsang, Li & Jones 2004: 176). Σύμφωνα μ’ αυτόν, οι λόγοι είναι προσδιορισμένοι από κοινωνικά κατασκευασμένους νόμους κι εμπεριέχουν ιδεολογίες που ελέγχονται από δυνάμεις εξουσίας (Fairclough 1989: 17, 28, 30). Στη διεθνή βιβλιογραφία καταγράφεται κι η έννοια του «Λόγου» / “Discourse” (με “D” κεφαλαίο). Πρόκειται για τον τρόπο με τον οποίο βλέπει κανείς τον κόσμο, το εν δυνάμει σύστημα κατασκευής νοημάτων, χωρίς να είναι εντελώς φανερό και χωρίς προσδιορισμένα όρια (Sunderland 2006: 48-52; Unger & Sunderland 2007: 460). Χαρακτηριστικό της ρευστότητας είναι πως το ίδιο “Discourse” μπορεί να αναπαραχθεί με διαφορετικά γλωσσολογικά «προσωπεία» (Lewis & Ketter 2004: 120-4).

2.1.3. Φωνή (voice) – (re)envoicing Η ΚΑΛ αξιοποιεί έννοιες της Μπαχτινικής θεωρίας, όπως είναι κι η «φωνή» / “voice”. Ο λόγος δεν είναι ποτέ ουδέτερος. Πάντα ενσωματώνει ιδεολογίες, που ο Bakhtin αναφέρει ως «φωνές» (Bakhtin 1981; Βολόσινοφ 1998). Όλοι οι άνθρωποι δανείζονται και μετασχηματίζουν τις φωνές των άλλων με σκοπό να δομήσουν τις δικές τους διατυπώσεις (Kamberelis & Scott 2004: 206). Οι διατυπώσεις των ανθρώπων μοιάζουν έτσι με

Discourse Analysis - Gender - Lexicography 7 1 Major Trends in Theoretical and Applied Linguistics

μωσαϊκό, αφού ενσωματώνουν λέξεις, φράσεις, στιλ, δομές, ίχνη από διατυπώσεις άλλων ανθρώπων (Kamberelis & Scott 2004: 212-3). H Kristeva γι’ αυτό το λόγο εισήγαγε τον όρο «διακειμενικότητα» (“intertextuality”) (Moi 1986: 39) –φανερή ή μη. Στη σύγχρονη έρευνα γίνεται πλέον συζήτηση για μια επαναδιατύπωση της Μπαχτινικής θεωρίας, όπου προτείνεται η παρακολούθηση της ιστορικότητας των φωνών που υπεισέρχονται στο κείμενο (Kostouli 2005: 20). Στο πλαίσιο αυτής της συζήτησης μιλούμε για “re-envoicing” και για μια αλληλεπίδραση των ετερογενών φωνών, με σκοπό τη σύνθεση από κοινού του τελικού κειμένου (Prior 2001). Θεωρήσαμε ενδιαφέρον να παρακολουθήσουμε και να μελετήσουμε τη δυναμική διαδικασία ενσωμάτωσης των διάφορων φωνών / λόγων κατά τη διάρκεια παραγωγής ενός κειμένου και όχι απλά να αναγνωρίσουμε τις διαφορετικές φωνές / λόγους σε ένα τελικό, δημοσιευμένο κείμενο.

2.2. Κοινωνιογνωστική έρευνα για την παραγωγή γραπτού λόγου (Στρατηγικές Γραφής / Writing Strategies)

Μια άλλη περιοχή που συνεισφέρει στην ανάλυση πρακτικών παραγωγής γραπτού λόγου είναι αυτή των Σπουδών παραγωγής γραπτού λόγου. Τα τελευταία χρόνια, σε αντίθεση με τη βιβλιογραφία για τη γνωστική προσέγγιση που αναλώθηκε στην αναθεώρηση (revision) ενός κειμένου και λιγότερο στις στρατηγικές γραφής, διαμορφώθηκε μια κοινωνικοπολιτισμική στροφή, που στηρίζεται στην ιδέα της διαμεσολάβησης (mediation) του Vygotsky (Lei 2008). Η παραγωγή των κειμένων θεωρείται μια «διαμεσολαβημένη πράξη». Δε μελετά κανείς στα κείμενα μόνο τα συστατικά των δομών τους, αλλά ερευνά τι έκανε ο συντάκτης (utterer) κατά τη διάρκεια της παραγωγής τους (Scollon et al. 2004: 174). Ως εξέλιξη των παραπάνω, καταγράφονται έρευνες στη διεθνή βιβλιογραφία που συνδέουν τις στρατηγικές γραφής με τη δόμηση ταυτοτήτων των υποκειμένων, πεδίο που απασχόλησε και την παρούσα εργασία. Για παράδειγμα, η Ivanic, στο έργο της Writing and Identity (1998: 4, 11-118, 182, 345) αναζήτησε τον τρόπο με τον οποίο δομείται η ταυτότητα κάποιου/κάποιας την ώρα που γράφει ποια στοιχεία της κοινότητάς του/της διατηρεί και ποια αποκρύπτει συνειδητά ή ασυνείδητα.

2.3. Κριτική Εθνογραφία (Critical Ethnography)

Κι αφού επιθυμούσαμε να ερευνήσουμε τα ίχνη των κειμένων κατά τη διάρκεια της παραγωγής τους (tracking news production), επιλέξαμε την Εθνογραφία για μέθοδο συλλογής δεδομένων. Η Εθνογραφία, το 19ο αιώνα, ορίστηκε ως η περιγραφή μιας κοινότητας ή κουλτούρας και, προκειμένου να επιτευχθεί, ο ερευνητής ή η ερευνήτρια πολλές φορές έπρεπε να ζήσει μαζί με την ομάδα των υπό έρευνα ανθρώπων για μεγάλα χρονικά διαστήματα (Atkinson & Hammersley 2007: 1).

7 2 Discourse Analysis - Gender - Lexicography Δημήτρης Ντανόπουλος

Συγκεκριμένα, χρησιμοποιήσαμε την Κριτική Εθνογραφία, που στην ουσία είναι το άλλο όνομα της Κριτικής Κοινωνικής / Ποιοτικής Έρευνας, αναζητώντας κάτι παραπάνω από την επαναδόμηση της κουλτούρας (Georgiou & Carspecken 2002: 689, 691). Η Κριτική Εθνογραφία δίνει έμφαση στη διαδικασία παραγωγής ενός πολιτισμικού προϊόντος (Carspecken 1996: 190-1). Με τη συμμετοχική παρατήρηση, συναντηθήκαμε πρόσωπο με πρόσωπο με τους ανθρώπους της υπό μελέτη κοινότητας, αποφεύγοντας να αλλάξουμε ή να διακόψουμε τις διαδικασίες παραγωγής των κειμένων και αναζητώντας πολλαπλές θέσεις οπτικής της πραγματικότητας (Philips 1998: 201-2).

2.4. Νέο μεθοδολογικό πλαίσιο: κριτική ανάλυση πρακτικών παραγωγής λόγου σε μια εφημερίδα

Στην προσπάθειά μας να διευρύνουμε την ΚΑΛ, ακολουθήσαμε το παράδειγμα του Graham Smart (2006, Writing the Economy: Activity, Genre, and Technology in the World of Banking, London, Oakville: Equinox), που ασχολήθηκε με τη μελέτη πρακτικών λόγου των οικονομολόγων της Κεντρικής Τράπεζας του Καναδά, αξιοποιώντας την “Activity Theory”. Η “Activity Theory” αποτελεί προέκταση της ιστορικο-πολιτισμικής προσέγγισης του Vygotsky (1978, Mind in society). Κεντρική θέση στη θεωρία κατέχει η έννοια “Activity System”, μια ιστορικά και πολιτισμικά προσδιορισμένη σφαίρα, στην οποία η πνευματική / διανοητική εργασία διαμεσολαβείται από πολιτισμικά εργαλεία, κοινωνικές αλληλεπιδράσεις και από κείμενα. (Smart 2006: 13-14, 28, 204) Δομώντας το δικό μας μεθοδολογικό μοντέλο, αξιοποιήσαμε τις διευρύνσεις της Κριτικής Ανάλυσης Λόγου, στοιχεία από τις Σπουδές για το γραπτό λόγο, με έμφαση στις πρακτικές παραγωγής κειμένων, σε συνδυασμό με την Κριτική Εθνογραφία. Στόχος μας ήταν να μελετήσουμε τις διαδρομές των κειμένων (routes) που παράγει/επεξεργάζεται ένας δημοσιογράφος, όπως αυτές διαμορφώνονται κατά την αλληλεπίδραση με θεσμούς και φορείς μιας πόλης (sociocultural context), προκειμένου να αναδείξουμε τις κοινωνικές πρακτικές που απηχούν. Πιο συγκεκριμένα, ενδιαφερθήκαμε για την κριτική ανάλυση των κοινωνικών πρακτικών παραγωγής δημοσιογραφικού λόγου (social practices) εντός της κοινότητας που παράγεται (discourse community). Γι’ αυτό, καταγράψαμε και μελετήσαμε τις δυναμικές σχέσεις αλληλεπίδρασης (Activity System) που συνδιαμορφώνονται σε σχέση με τις σύγχρονες διαδρομές παραγωγής διαφορετικών κειμένων αλλά και τους ανθρώπους που ζούσαν και εργάζονταν εντός της κοινότητας. Η διαδικασία αυτή παραγωγής κειμένων (texts) απηχεί ένα συγκεκριμένο ιδεολογικό “Discourse” της κοινότητας, αλλά και πραγματώνεται με συγκεκριμένες στρατηγικές γραφής (Writing Strategies) που τα μέλη της κοινότητας έχουν κατακτήσει και χρησιμοποιούν σε μια διαδικασία αλληλεπίδρασης των ταυτοτήτων τους με την ταυτότητα της κοινότητας και της πόλης από την οποία επηρεάζονται αλλά και συνδομούν ταυτόχρονα.

Discourse Analysis - Gender - Lexicography 7 3 Major Trends in Theoretical and Applied Linguistics

3. Ερευνητικά δεδομένα

Το πεδίο της έρευνάς μας ήταν ο χώρος εργασίας μιας τοπικής εφημερίδας (ΕΦ) στην Καρδίτσα, μιας κοινότητας ανθρώπων αλλά και των πρακτικών τους, καθώς και οι χώροι όπου οι δημοσιογράφοι μετακινούνταν για να συγκεντρώσουν υλικό για την παραγωγή των κειμένων τους. Εστιάσαμε την έρευνά μας στο δημοσιογράφο Δ και στο πλέγμα αλληλεπιδράσεων που αυτός ανέπτυσσε με τους εργαζόμενους / τις εργαζόμενες ή μη στο χώρο εργασίας. Ο δημοσιογράφος Δ λειτούργησε ως «διευκολυντής» μας / “facilitator” (Fetterman 1998: 33). Η προσωπική μας διασύνδεση μάς έδωσε το εισιτήριο για την πρόσβαση και την είσοδό μας στο πεδίο στην πιλοτική φάση (1 βδομάδα, Μάρτιος 2009) αλλά και στην κύρια (2 βδομάδες, Απρίλιος 2009). Συγκεντρώθηκαν στοιχεία και δεδομένα, κρατώντας ημερολόγιο και γραπτές σημειώσεις, ζώντας μέσα στο πεδίο, μεταγράφτηκαν συνομιλίες και οπτικοακουστικά ντοκουμέντα που καταγράφηκαν με ψηφιακό ηχητικό καταγραφέα και βιντεοκάμερα. Επίσης, αρχειοθετήθηκαν καθημερινά φύλλα της εφημερίδας ΕΦ, αντίγραφα από κείμενα φαξ, ηλεκτρονικά μηνύματα (e-mails), καθώς και άλλα γραπτά κείμενα εντός του πεδίου.

4. Ανάλυση δεδομένων

4.1. Η περίπτωση της εφημερίδας ΕΦ

Η εφημερίδα ΕΦ ήταν μία από τις οχτώ καθημερινές –συνδρομητικές- εφημερίδες που κυκλοφορούσαν στην Καρδίτσα. Είχε περίπου 1000 συνδρομητές. Κυκλοφορούσε για χρόνια ως τοπική μόνο εφημερίδα. Τα δύο τελευταία χρόνια, σε μια συνεταιρική μορφή, κυκλοφορούσε πανθεσσαλικά. Τα κεντρικά γραφεία βρίσκονταν στη Λάρισα. Στις θεσσαλικές πρωτεύουσες νομών λειτουργούσαν τοπικά παραρτήματα. Η σελιδοποίηση γινόταν στη Λάρισα, αλλά κάποιες σελίδες «στήνονταν» στην Καρδίτσα. Οι εργαζόμενοι κι οι εργαζόμενες επικοινωνούσαν κάθετα και οριζόντια μέσω τηλεφώνων και ηλεκτρονικών μηνυμάτων, ενώ υπήρχε on line επικοινωνία μέσω εσωτερικού ηλεκτρονικού δικτύου. Στο τοπικό παράρτημα της Καρδίτσας συνυπάρξαμε με τον εκδότη, που προερχόταν από το χώρο των οικοδομικών επιχειρήσεων, κι ήταν απογοητευμένος από την πορεία της εφημερίδας. Η εφημερίδα αντιμετώπιζε οικονομικά προβλήματα που επέφεραν μείωση προσωπικού. Οι εργαζόμενοι κι εργαζόμενες ανησυχούσαν για το μέλλον τους. Ένα χρόνο μετά την έρευνα, η εφημερίδα διέκοψε την κυκλοφορία της. Συναντηθήκαμε με τη γραμματέα, τον οικονομικό διαχειριστή και διανομέα της εφημερίδας, με τον αθλητικογράφο, τη σελιδοποιό, αλλά επικεντρωθήκαμε στο δημοσιογράφο Δ, που εργαζόταν ήδη 12 χρόνια και η ΕΦ δεν ήταν η πρώτη του δουλειά.

7 4 Discourse Analysis - Gender - Lexicography Δημήτρης Ντανόπουλος

Ο Δ είχε εργαστεί, στην αρχή της πανθεσσαλικής κυκλοφορίας της ΕΦ, στα κεντρικά γραφεία της, με συνέπεια να γνωρίζει πρόσωπα και διαδικασίες. Είχε γνώση του “habitus”1 αυτής αλλά και της συγκεκριμένης τυπολογίας που έπρεπε ν’ ακολουθούν τα κείμενα για το στήσιμο της ΕΦ (υπότιτλοι, λεζάντες κλπ). Κατά τη διάρκεια της έρευνας, ο Δ τα πρωινά απασχολούνταν στο γραφείο τύπου του δήμου της πόλης. Όταν ο Δ δεν είχε φυσική παρουσία σε χώρους ρεπορτάζ, τον αντικαθιστούσε η βοηθός του (Ε), φέρνοντάς του υλικό και πληροφορίες.

4.2. Διαδρομές κειμένων

Μελετήσαμε τις διαδικασίες παραγωγής αρκετών κειμένων και αποφασίσαμε να χρησιμοποιήσουμε το μοντέλο οργάνωσης των δρομολογίων ενός φανταστικού υπόγειου σιδηρόδρομου, προκειμένου να τις αναπαραστήσουμε. Ο αναγνώστης, διαβάζοντας καθημερινά την εφημερίδα, παρακολουθούσε την εξέλιξη κάποιων θεμάτων, αναγνωρίζοντας τις μεταξύ τους σχέσεις. Ονομάσαμε «θεματική γραμμή», λοιπόν, την ομαδοποίηση των δημοσιευμένων στην εφημερίδα κειμένων με βάση το θέμα τους, κατά τη χρονική διάρκεια της έρευνας. Πρόκειται για το φανερό / εξωτερικό δίκτυο, τους υπέργειους σταθμούς του φανταστικού σιδηρόδρομου. Ο αναγνώστης δεν καταφέρνει ποτέ να κατεβεί τα σκαλιά των υπέργειων σταθμών. Προκειμένου να ερευνήσουμε την «υπόγεια» παραγωγή των κειμένων από την αρχική ιδέα ως τη στιγμή που θα γίνονταν προϊόν της εφημερίδας, που θα επιτρεπόταν η ανάδυσή τους στον υπέργειο σταθμό, χρησιμοποιήσαμε την έννοια «διαδρομές των κειμένων», μια έννοια που δίνει την αίσθηση της πορείας, της κίνησης στο χώρο και στο χρόνο. Πρόκειται για ένα κρυφό / εσωτερικό δίκτυο διαδρομών. Οι διαδρομές αυτές δεν ήταν αυστηρά προδιαγεγραμμένες. Δε διακρίνονταν για την ευθεία τους κίνηση και τη σταθερότητα στην ταχύτητα. Οι πορείες τους συνεχώς επαναπροσδιορίζονταν, άλλοτε τέμνονταν μεταξύ τους κι άλλοτε κινούνταν παράλληλα ή ασύμπτωτα. Κάποιες διακόπτονταν, δίνοντας προτεραιότητα σε άλλες κι έμεναν ανοιχτές κι ανολοκλήρωτες, περιμένοντας τη σειρά τους. Όλες, όμως, είχαν ως τελικό σκοπό να περάσουν, ως δημοσιευμένα κείμενα, στο εξωτερικό / φανερό δίκτυο, να γίνουν μέρη μιας ακόμα θεματικής γραμμής, αφού πρώτα περνούσαν με επιτυχία από παρακάτω διαδοχικά στάδια (σταθμούς) ελέγχου και επιλογής: Ως πρώτο σταθμό ή αφετηρία της διαδρομής των παραγόμενων κειμένων θεωρήσαμε την υποδοχή των ηλεκτρονικών μηνυμάτων, Δελτίων Τύπου, προσκλήσεων για συνεντεύξεις τύπου, ρεπορτάζ, κειμένων φαξ ή τηλεφωνημάτων σε ένα κουτί εισερχόμενων μηνυμάτων (Inbox).

1 Ο όρος “habitus” χρησιμοποιήθηκε από τον Bourdieu (1972) και αναφέρεται στην ενσωμάτωση των ατομικών πράξεων σε ένα σύστημα κοινωνικών κανόνων, αντίληψης και προτύπων συμπεριφοράς, που χωρίς να προκαθορίζει πράξεις εγγυάται ότι τα άτομα έχουν μεγαλύτερη προδιάθεση να δράσουν με ένα συγκεκριμένο τρόπο απ’ ότι με διαφορετικούς (Hillier & Rooksby 2005: 21).

Discourse Analysis - Gender - Lexicography 7 5 Major Trends in Theoretical and Applied Linguistics

Τα «εισερχόμενα μηνύματα» αξιολογούνταν από το δημοσιογράφο για το αν θα γίνουν κείμενα προς δημοσίευση. Γι’ αυτό και είχαμε και περιπτώσεις μη υλοποιημένων διαδρομών (Χ). Στη συνέχεια, ο δημοσιογράφος Δ τοποθετούσε στο κουτί των «εξερχόμενων μηνυμάτων» τα έτοιμα κείμενα προς δημοσίευση. Το “outbox” έμοιαζε με σταθμό μετεπιβίβασης. Τα κείμενα λάμβαναν οι αρχισυντάκτες, μέσω ηλεκτρονικού ταχυδρομείου, στα κεντρικά γραφεία της ΕΦ, και έπαιρναν την τελική απόφαση για το αν θα τα οδηγήσουν στον τελικό σταθμό, τη δημοσίευση, ή όχι. Από το σημείο αυτό και μετά, αναλάμβαναν δραστηριότητα οι φύλακες της ταυτότητας της εφημερίδας. Συγκεκριμένα, πρόσθεταν στοιχεία, συνέθεταν κείμενα, τα συνόψιζαν ή τα τροποποιούσαν, αφαιρούσαν υπότιτλους, φράσεις, φωτογραφίες ή λεζάντες, προωθούσαν κείμενα στην πρώτη σελίδα ή τα υποβάθμιζαν δημοσιεύοντάς τα στις τελευταίες σελίδες.

4.3. Διαχείριση φωνών

Ζώντας στο πεδίο και μελετώντας, στη συνέχεια, την εθνογραφική περιγραφή των διαδρομών πολλών κειμένων σε ένα κρυφό και φανερό αυτού δίκτυο, διαπιστώσαμε πως ο δημοσιογράφος Δ βρισκόταν εντός ενός πλέγματος ιδεολογικών επιδράσεων και σχέσεων εξουσίας, στην ουσία ενός Activity System (2Α.4.).

4.3.1. Παράγοντες / διαστάσεις επίδρασης

(1) Η ταυτότητα της πόλης Ο Δ ήταν φορέας του Λόγου της τοπικής κοινότητας. Γνώριζε καλά τις αξίες και την ιστορία της πόλης. Οι πρακτικές του απηχούσαν το ιδεολογικό της Discourse. Τα κείμενά του ήθελε να διακρίνονται για την τοπο-κεντρικότητά τους. Για παράδειγμα, απέκρυψε επιστολή της οργάνωσης WWF κατά της εκτροπής του Αχελώου. Η εκτροπή του Αχελώου θεωρείται σημαντικό έργο για την περιοχή. Ο Δ στα κείμενά του το χαρακτήριζε ως «εθνικό έργο» και σημείωνε πάντα τη φράση «έργο μερικής εκτροπής του άνω ρου του Αχελώου», φράση που χρησιμοποιούταν από τους τοπικούς φορείς. Ο ίδιος είχε κατά νου μια συγκεκριμένη ταυτότητα της πόλης, ως μικρής πόλης της επαρχίας. Μιλώντας με τη βοηθό του Ε, σχολίασε: «Εσύ νομίζεις πως αυτά που γίνονται στην Καρδίτσα επηρεάζουν τη ζωή σου;» Τα δομικά στοιχεία της ταυτότητας της πόλης, όπως αυτά καταγράφτηκαν στη διάρκεια της έρευνας, εντασσόμενα άλλοτε στο φανερό κι άλλοτε στο κρυφό δίκτυο των διαδρομών, συνέθεταν φωνές, για παράδειγμα, πολιτικών: Ο αγροτικός χαρακτήρας της περιοχής κυριαρχούσε ως δομικό στοιχείο της ταυτότητας της κοινότητας. Ο δήμαρχος δώρισε πίνακα ζωγραφικής με θέμα «αγροτικές εργασίες» στον πρόεδρο της Βουλής. Ωστόσο, καταγράψαμε και την προσπάθεια ενσωμάτωσης καινούριων στοιχείων, όπως ο «αγροτουρισμός» αλλά και τα «φωτοβολταϊκά πάρκα».

7 6 Discourse Analysis - Gender - Lexicography Δημήτρης Ντανόπουλος

Ένα κυρίαρχο σημείο ενδιαφέροντος της δημόσιας (αλλά πολλές φορές αδημοσίευτης) συζήτησης ήταν ο τρόπος ανάπτυξης της τεχνητής Λίμνης Πλαστήρα, «της θετικής παρέμβασης του ανθρώπινου παράγοντα», όπως διαπίστωσε ο νομάρχης. Καταγράψαμε και αντιρρήσεις ως προς τον προσανατολισμό, για παράδειγμα, τη διατύπωση περιφερειακού δημάρχου πως «τα συνέδρια δε βοηθούν τις τουριστικές επιχειρήσεις». Στοιχείο της ταυτότητας της πόλης αποτελούν οι πανεπιστημιακές σχολές, όπως η Κτηνιατρική, που για τον ντόπιο υφυπουργό Παιδείας ήταν η «ναυαρχίδα της εκπαίδευσης» στην πόλη, ενώ, από την άλλη, υπάρχει σχετική υποβάθμιση των τμημάτων ΤΕΙ (ο Δ απέρριψε δύο σχετικά δελτία τύπου). Παρούσα στη διαμόρφωση της ταυτότητας της πόλης είναι η τοπική Εκκλησία. Διαπιστώσαμε την ύπαρξη χριστιανικών εικόνων, συμβόλων στο γραφείο του δημάρχου, την άριστη σχέση μεταξύ δημοσιογράφων και κληρικών, καταγράψαμε την αγωνία των επισήμων να προσφωνήσουν πρώτους τους εκπροσώπους της εκκλησίας σε κοινωνικές εκδηλώσεις. Παρ’ όλα αυτά ο Δ προσπάθησε να επιλέξει φωτογραφίες που να μην εικονίζουν κληρικούς («Να φάω τους παππάδες», σχολίασε). Εκτός από τον τοπικό άγιο, κάθε πόλη, δομώντας την ταυτότητά της, χρειάζεται τους ήρωές της. Ιδεολογία ασκούν μορφές όπως αυτή του πολεμιστή του ’21 Γεώργιου Καραϊσκάκη, καθώς και του στρατιωτικού-πολιτικού, Νικόλαου Πλαστήρα. Ως εμβληματική μορφή, από το χώρο της τέχνης, καταγράφτηκε ο ζωγράφος Δημήτρης Γιολδάσης. Αναζητώντας ιδανική γυναικεία μορφή, η μόνη περίπτωση που καταγράψαμε ήταν αυτή της ‘ηρωίδας «Μάνας» αγρότισσας’, μιας και παραβρεθήκαμε σε αποκαλυπτήρια ομώνυμου αγάλματος. Στο κείμενο τού χαιρετισμού τού δημάρχου συνέβαλε κι ο Δ, γράφοντας: [Η μάνα] «…να ποτίζει το βασιλικό της, … να φροντίζει, …να προσφέρει τα πάντα, …». Η ταυτότητα της πόλης χρειάζεται «φύλακες». Τα σώματα ασφαλείας και οι τοπικές στρατιωτικές αρχές είναι παρούσες. Το εθνικό στοιχείο το αντιλαμβάνεται κανείς ακόμα και σε εκδηλώσεις, όπως τα αποκαλυπτήρια αγάλματος που γίνονται αφορμή για ανάκρουση του εθνικού ύμνου. Κοιτώντας προς το μέλλον, η πόλη αναζητά τη θέση της στην ψηφιακή εποχή. Κατά τη διάρκεια της έρευνάς μας, δήμος και νομαρχία έριζαν για το ποιος από τους δυο θα την οδηγήσει προς τα κει. Στην περίπτωση της ΕΦ, ο Δ ενίσχυε τη φωνή του δήμου. Αποσιωπήθηκε “off the record” συζήτηση που αναδείκνυε προβλήματα οικονομικής διαχείρισης προγράμματος τηλε-ιατρικής στο δήμο. Ο Δ έγραψε κείμενο με τίτλο «Ο Δήμος …εισέρχεται δυναμικά στην ‘Ψηφιακή Εποχή’…» κι από την άλλη αποσιωπήθηκαν δύο περιπτώσεις δραστηριοποίησης της νομαρχίας. Στην πόλη αυτή υπάρχουν κι οι «άλλοι», για τους οποίους, όταν αναφέρονται τα μέσα ή οι φορείς, προσπαθούν να μη μιλήσουν «ρατσιστικά». Για παράδειγμα, ο Δ για να μεταφέρει την είδηση επίσκεψης σε παιδική βιβλιοθήκη τάξης σχολείου τσιγγάνων, χρησιμοποίησε τη λέξη «τσιγγανόπουλα» αντί για «τσιγγανόπαιδα» του αρχικού κειμένου και σκέφτηκε: «Λίγο ρατσιστικό, ε; Αλλά πώς να το βάλεις για να το καταλάβουν;». Στη λεζάντα της φωτογραφίας προτίμησε τη φράση: «μικροί μαθητές».

Discourse Analysis - Gender - Lexicography 7 7 Major Trends in Theoretical and Applied Linguistics

Ένα μόνο κείμενο που διαχειρίστηκε ο Δ αναφερόταν σε άλλη εθνότητα. Προερχόταν από τον τοπικό Σύλλογο Αλβανών Μεταναστών. Το κείμενο απορρίφθηκε από τους αρχισυντάκτες.

(2) Η τοπική δημοσιογραφική κοινότητα Ιδεολογική επίδραση στο δημοσιογράφο Δ ασκούσε η τοπική δημοσιογραφική κοινότητα και η κουλτούρα που είχε αναπτύξει. Ο Δ χρησιμοποιούσε κοινωνικές πρακτικές (Journalism as social practice) παραγωγής των γραπτών του κειμένων που έμαθε εντός αυτής της επαγγελματικής κοινότητας και αξιοποιούσε το δίκτυο επικοινωνιών του με τις/τους δημοσιογράφους των υπόλοιπων τοπικών μέσων. Τη δύναμη που νιώθει η τοπική δημοσιογραφική κοινότητα τη διαπιστώσαμε στη φράση εκδότη άλλης τοπικής εφημερίδας: «Καλά, διαβάστε αύριο στην εφημερίδα!». Επίσης, οι δημοσιογράφοι χρησιμοποιούνταν ως μάρτυρες σε προφορικές συνομιλίες («εδώ είναι οι δημοσιογράφοι»), ενώ δημοσιεύματα γίνονταν τεκμήρια στα οποία αναφέρονταν πολιτικά πρόσωπα. Στην περιοχή της Θεσσαλίας υπάρχει μεγάλη παράδοση τοπικού τύπου, που ξεκινά από τα τέλη του 19ου αι. Το γεγονός αυτό έχει δημιουργήσει μια ανάλογη τοπική κουλτούρα στο αναγνωστικό κοινό, που αναγνωρίζει τις κοινωνικές πρακτικές των δημοσιογράφων και τους παρέχει δικαίωμα λόγου στη δημόσια σφαίρα. Ο Δ το γνωρίζει αυτό κι έτσι χρησιμοποιεί πολλές συντομογραφίες και γράφει με τρόπο που προϋποθέτει έναν έμπειρο ή τουλάχιστο έναν ενημερωμένο αναγνώστη. Ωστόσο, τον ακούσαμε να λέει: «Πολλοί θα το δουν λίγοι θα το προσέξουν».

(3) Η εργασιακή σχέση με την ΕΦ Ο Δ, όμως, δεν έπρεπε να ξεχνά την εργασιακή του σχέση με την εφημερίδα ΕΦ. Ο εκδότης ήταν ιδιοκτήτης ξενοδοχειακής μονάδας. Έτσι, ο Δ, σε περιπτώσεις εκδηλώσεων στο ξενοδοχείο του εκδότη, έλεγε: «Να βάλουμε το ξενοδοχείο […] να φαίνεται το αφεντικό». Ο Δ είχε εργαστεί, στην αρχή της πανθεσσαλικής κυκλοφορίας της ΕΦ, στα κεντρικά γραφεία της, με συνέπεια να γνωρίζει πρόσωπα και διαδικασίες που συνέθεταν την ταυτότητα της ΕΦ. Ο Δ, ως φορέας της, είχε την προδιάθεση να δρα με τον τρόπο που αυτή προσδιόριζε. Ο ίδιος ο εκδότης σε συζήτηση που είχαμε μάς εξέφρασε τη δυσαρέσκειά του για το κόμμα που κυβερνούσε (ΝΔ), επειδή τον καθυστέρησε σε εκδοτικό του πρότζεκτ. «Μας έχουν πρήξει με το ΠΑΣΟΚ», σχολίαζε κι η βοηθός Ε.

(4) Η εργασιακή σχέση με το δήμο Ο Δ ήταν φορέας και της φωνής του δήμου. Η ταυτόχρονη εργασιακή του σχέση με το δήμο τον έκανε πολλές φορές να ταυτίζει τον εαυτό του με αυτόν, χρησιμοποιώντας α’ πληθυντικό πρόσωπο στις μεταξύ μας συζητήσεις: «…Εμείς έχουμε το άγαλμα, τη Μάνα».

(5) Η ιδεολογική επίδραση του ερευνητή Τέλος, υπήρχε κι η ιδεολογική επίδραση του ερευνητή. Για παράδειγμα, ο Δ δεν επέλεξε να βάλει φωτογραφία του δημάρχου, λέγοντας: «Έχω γίνει αντικείμενο μελέτης…», «πειραματόζωο, πειραματόζωο…».

7 8 Discourse Analysis - Gender - Lexicography Δημήτρης Ντανόπουλος

4.3.2. Είδη πλαισιώσεων Ο δημοσιογράφος, αναλαμβάνοντας τη διαχείριση των φωνών της τοπικής κοινότητας, διαφοροποιούσε τη στάση του, ανάλογα με τη φωνή που διαχειριζόταν. Για παράδειγμα, ο Δ δεν ασχολούνταν με κείμενα που προέρχονταν από ιδιώτες και δε δήλωναν κάποια θεσμική θέση ή συμμετοχή σε συλλογικό φορέα. Ή ανάλογα με το θέμα του κειμένου που έγραφε, για παράδειγμα η περίπτωση του διαφορετικού χειρισμού της φωνής του WWF, που στην πιλοτική φάση τη χρησιμοποίησε για να προσδώσει κύρος σε εκδήλωση του δήμου και σε δεύτερη περίπτωση την κατέπνιξε, αφού η οργάνωση χαρακτήριζε «… περιβαλλοντικό και κοινωνικό έγκλημα αξίας εκατοντάδων εκατομμυρίων ευρώ» την εκτροπή του Αχελώου. Ο Δ ερχόταν σε έναν άτυπο ανταγωνισμό με τους αρχισυντάκτες, που θα ήταν οι τελικοί κριτές των παραγόμενων κειμένων του. Διεκδικούσε μέσω των κειμένων του θέση και λόγο στη διαμόρφωση της ταυτότητας της ΕΦ αλλά και της πόλης. Έτσι, συνοπτικά, διαχειρίστηκε τα «εισερχόμενα μηνύματα», χρησιμοποιώντας διαφορετικά είδη πλαισίωσης, με αποτέλεσμα: - Να αναδείξει φωνές, όπως του δήμου. - Να αφήνει κάποιες φωνές να ακουστούν ίσα-ίσα. Όπως του κομμουνιστικού κόμματος, που το θεωρούσε ξεπερασμένο, ακόμα και για τον τρόπο που στέλνει τα κείμενά του, με φαξ. Ο Δ σχολίασε: «Αυτή η εποχή για μας πέρασε…». Έτσι, διαμόρφωνε την πρακτική του με βάση το τεχνικό υλικό. Ένας άλλος παράγοντας διαμεσολάβησης, που καμιά φορά υποτιμάται. -Να καταπνίγει φωνές. Συγκεκριμένα, για τον ΠΟΛΙΤΕΥΤΗ_ΝΔ_Π, σχολίασε εισερχόμενο μήνυμά του: «Α, ρε … Μ’ αρέσει που είσαι αισιόδοξος. Έχει δει ότι δε βάζουμε κανένα και στέλνει…» Κι αν ο Δ παρουσιαζόταν ως θεματοφύλακας της ταυτότητας της πόλης, οι αρχισυντάκτες ήταν οι φύλακες της ταυτότητας της ΕΦ.

4.4. Η ταυτότητα της εφημερίδας / η ταυτότητα της πόλης

Η ΕΦ, ως εμπορικό προϊόν, στόχευε στη μέση και μεγαλύτερη ηλικία συνδρομητών. Στοιχεία της ταυτότητάς της που ενδιαφέρουν τη συζήτησή μας, αποτελούσαν τα παρακάτω: - Ο λόγος της ήταν απλός, λαϊκός, συντηρητικός, χωρίς εναλλακτική χρήση του θηλυκού γένους. - Οι προφορικές συνομιλίες της τοπικής διαλέκτου μεταγράφονταν και δημοσιεύονταν ως συνομιλίες / δηλώσεις στην καθιερωμένη επίσημη ελληνική γλώσσα. - Γινόταν λόγος κύρια για επίσημους φορείς και μεγάλα κόμματα. - Κυριαρχούσαν οι φωνές των κυρίαρχων ομάδων και αποσιωπούνταν φωνές ιδιωτών. Μέσα από την ΕΦ αναδυόταν μια ταυτότητα ομοιογένειας του πληθυσμού της κοινότητας που μιλά την καθιερωμένη (επίσημη) ελληνική γλώσσα. Η κοινότητα είχε συγκεκριμένη θρησκευτική (χριστιανική) παράδοση, παρόλη την τάση ακόμα κι από τους αρχισυντάκτες να μειώσουν τη φωνή της Εκκλησίας: αφαιρώντας λέξεις από τίτλους / υπότιτλους ή και φωτογραφίες.

Discourse Analysis - Gender - Lexicography 7 9 Major Trends in Theoretical and Applied Linguistics

Μπορούμε να πούμε πως η εφημερίδα αναπαρήγαγε και ενίσχυε μια ταυτότητα που δεν ανταποκρινόταν στη σύγχρονη πραγματικότητα, αφού δεν είχαν επικαιροποιηθεί τα δομικά της στοιχεία.

5. Συμπεράσματα

Συνοψίζοντας τις σκέψεις μας, θα λέγαμε ότι: Τα κείμενα που δημοσιεύονται στις εφημερίδες αποτελούν την εξωτερική επιφάνεια ενός ιδεολογικού προϊόντος, αφού, εντός αυτού δημιουργούνται σχέσεις ανταγωνισμού, σύνθεσης, αλληλεπίδρασης, ρήξης φωνών και ταυτοτήτων που συνεχώς διεκδικούν, προσδιορίζουν και επαναπροσδιορίζουν την πορεία των κειμένων μέσα στο χώρο και στο χρόνο. Η παρούσα έρευνα κατέδειξε ότι η μελέτη των διαδικασιών που οδηγούν στη διαμόρφωση ενός κειμένου φέρνει στην επιφάνεια τους τρόπους με τους οποίους ο δημοσιογράφος διαχειρίζεται ποικίλες φωνές (ενισχύει, καταπνίγει, υποβαθμίζει), αλλά και πώς αυτές επηρεάζουν τη διαδικασία σύνθεσης ενός κειμένου. Η παραγωγή των κειμένων μοιάζει με διαδρομή που έχει ως τελικό σκοπό τη δημοσίευση. Την τελική τους μορφή αποφασίζουν οι αρχισυντάκτες.

5.1. Προεκτάσεις

Προέκυψαν ανάγκες προέκτασης της έρευνας, υπό τα εξής ερωτήματα: - Πώς προσλαμβάνουν οι αναγνώστες όλο αυτό το παιχνίδι διαπραγμάτευσης των ταυτοτήτων και τι θέση λαμβάνουν εντός αυτού του πεδίου; - Πώς διαμορφώθηκε και παρουσιάστηκε η ταυτότητα της συγκεκριμένης κοινότητας στις τοπικές εφημερίδες από το 19ο αιώνα και ποια στοιχεία προστέθηκαν ή αφαιρέθηκαν ως σήμερα;

5.1.1. Παιδαγωγικές προεκτάσεις Τα τελευταία χρόνια, θεωρητικοί της Κριτικής Παιδαγωγικής αναζητούν τρόπους συνάντησης της «εκπαίδευσης για τα μέσα» με την «εκπαίδευση στα μέσα» (media education), καθώς και ένταξης αυτών ως γνωστικό αντικείμενο στο αναλυτικό πρόγραμμα σπουδών των σχολείων (Buckingham 1996). Έχοντας στο νου μας τα συμπεράσματα της παρουσιαζόμενης έρευνας, θεωρούμε πως μια προσπάθεια κριτικού γραμματισμού στο δημοσιογραφικό λόγο πρέπει να έχει ως αφετηρία την έξοδο των παιδιών από το σχολείο και τη σύνδεσή τους με την κοινωνία. Η επαφή τους με τη δημοσιογραφική κουλτούρα θα ελαχιστοποιήσει το διαμεσολαβητικό ρόλο τού/τής εκπαιδευτικού, ώστε τα παιδιά να έρθουν σε επαφή με τα κείμενα, έχοντας πρώτα αντίληψη της κουλτούρας που τα δημιούργησε.

8 0 Discourse Analysis - Gender - Lexicography Δημήτρης Ντανόπουλος

Βιβλιογραφία

Atkinson, P. and Hammersley, M. 1995/2007. Ethnography: Principles in Practice. London & New York: Routlendge. 2nd/3rd edition.

Bakhtin, M. 1981. In M. Holquist (ed.), The Dialogic Imagination. Four Essays. Translated by C. Emerson and M. Holquist. Austin: University of Texas Press.

Bourdieu, P. 1977. Outlines of a Theory of Practice. Cambridge: University Press. 1st published 1972.

Buckingham, D. 1996. Critical pedagogy and media education: a theory in search of a practice. Journal of Curriculum Studies 28, 6: 627-650.

Βολόσινοφ, Β. 1998. Μαρξισμός και Φιλοσοφία της Γλώσσας. Μετάφραση από το Β. Αλεξίου. Αθήνα: Παπαζήση. 1η έκδοση: 1929.

Carspecken, P.F. 1996. Critical Ethnography in Educational Research. New York & London: Routledge.

Fairclough, N. 1989. Language and Power. London & New York: Longman.

Fairclough, N. 1995. Media Discourse. London, New York, Sydney, Auckland: Arnold.

Fairclough, N. 2003. Analysing Discourse. Textual analysis for social research. London & New York: Routledge.

Fetterman, D. 1998. Ethnography. Step by Step. Thousand Oaks, London, New Delhi: SAGE.

Garret, P. and Bell, A. 1998. Media and Discourse: A Critical Review. In A. Bell & P. Garrett (ed.), Approaches to Media Discourse. Oxford: Blackwell, 1-20.

Discourse Analysis - Gender - Lexicography 8 1 Major Trends in Theoretical and Applied Linguistics

Georgiou, D. and Carspecken, P.F. 2002. Critical Ethnography and Ecological Psychology: Conceptual and Empirical Explorations of a Synthesis. Qualitative Inquiry, 8, 6: 688-706.

Hillier, J. and Rooksby, E. 2005. Introduction to first Edition. In J. Hillier and E. Rooksby (ed.), Habitus: A sense of place (2nd edition). Hants & Burlington: Ashgate. 1st published 2002.

Ivanic, R. 1998. Writing and Identity: The discoursal construction of identity in academic writing. Amsterdam, Philadelphia: John Benjamins Publishing Company.

Jankowski, N. 2002. Creating Community with Media: History, Theories and Scientific Investigations. In L. Lievrouw and S. Livingstone (ed.), Handbook of New Media. Social Shaping and Consequences of ICTs. London, Thousand Oaks, New Delhi: SAGE Publications, 34-49.

Kamberelis, G. and Scott, K.D. 2004. Other people’s voice. The Coarticulation of Texts and Subjectivities. In N. Shuart-Faris and D. Bloome (ed.), Uses of Intertextuality in Classroom and Education Research. Connenticut: IAP, 201-250.

Kostouli, T. 2005. Introduction: Making Social Meanings in Contexts. In T. Kostouli (ed.), Writing in Context(s). USA: Springer, 1-26.

Lei, X. 2008. Exploring a sociocultural approach to writing strategy research: Mediated actions in writing activities. Journal of Second Language Writing, 17(4): 217–236.

Lewis, C. and Ketter, J. 2004. Learning as social interaction: Interdiscursivity in a teacher-researcher book group. In R. Rogers (ed.), An introduction to critical discourse analysis in education. Mahwah, NJ: Lawrence Erlbaum Associates, 117-146.

Moi, T. 1986. The Kristeva Reader. New York: Columbia University Press.

Philips, S.U. 1988. The Language Socialization of Lawyers: Acquiring the “Cant”. In G.D. Spindler (ed.), Doing the Ethnography of Schooling. Illinois: Waveland Press, Inc, 176-209.

Prior, P. 2001. Voices in text, mind, and Society. Sociohistoric accounts of discourse acquisition and use. Journal of Second Language Writing, 10: 55-81.

Πολίτης, Π. (επιμ.) 2008. Ο λόγος της μαζικής επικοινωνίας. Το ελληνικό παράδειγμα. Ινστιτούτο Νεοελληνικών Σπουδών (Ίδρυμα Μανόλη Τριανταφυλλίδη).

8 2 Discourse Analysis - Gender - Lexicography Δημήτρης Ντανόπουλος

Scollon, R., Yung, V., Tsang, W.K., Li, D. and Jones, R. 2004. Voice, Appropriation, and Discourse representation in a student writing task. In N. Shuart-Faris and D. Bloome (ed.), Uses of Intertextuality in Classroom and Education Research. Connenticut: IAP, 173-199.

Smart, G. 2006. Writing the Economy: Activity, Genre, and Technology in the World of Banking. London, Oakville: Equinox.

Sunderland, J. 2006. Language and Gender. London & New York: Routledge.

Unger, J. and Sunderland, J. 2007. Gendered Discourses in a Contemporary Animated Film: Subversion and Confirmation of Gender Stereotypes in Shrek. In N. Fairclough, G. Cortese and P. Ardizzone (ed.), Discourse and Contemporary Social Change. Bern: Peter Lang, 459-486.

Van Dijk, T.A. 1988. News as Discourse. Hillsdale, New Jersey, Hove and London: Lawrence Erlbaum Associates, Publishers.

Van Hout, T. 2011. Writing from news sources: The case of Apple TV. Journal of Pragmatics, 43: 1876-1889.

Vygotsky, L. 1978. Mind in society: The development of higher psychological processes. Cambridge, MA: Harvard University Press.

Χατζησαββίδης, Σ. 1999. Ελληνική γλώσσα και δημοσιογραφικός λόγος. Θεωρητικές και ερευνητικές προσεγγίσεις. Αθήνα: Gutenberg.

Discourse Analysis - Gender - Lexicography 8 3 Major Trends in Theoretical and Applied Linguistics

8 4 Discourse Analysis - Gender - Lexicography Dialect and Foreign Language Features in Greek-Cypriot Advertising: Investigating Factors Influencing Recipients’ Attitudes

Andreas Papapavlou and Melanie Satraki

University of Cyprus [email protected], [email protected]

Abstract

Although advertising aims at promoting goods, by incorporating phenomena like code-mixing, it becomes an interesting area for linguists to study. In this paper, we aimed at investigating Greek-Cypriots’ view of the appearance of dialect and foreign features in local advertising. In doing so, an effort was made to test influence by the medium used to present the advertisement or the stereotype reflected. The results support studies that showed preference towards the standard variety (Papapavlou 1998). However, advertisers’ insistence on associating different types of products with different codes (Pavlou 1992, 2004) does not bear support.

1. Introduction

Since the establishment of the Republic of Cyprus, the linguistic situation of the island has been linked to that of identity. Although official language policy recognises Standard Modern Greek (SMG) as the official language of the Greek-Cypriot community, people never stopped using the Greek Cypriot Dialect (GCD) in their everyday life (Papapavlou & Pavlou 1998; Karyolemou 2001). GCD has borrowed heavily from many different languages – since its history has been interlinked with that of over thirty-two countries (Varella 2006). English, as the lingua franca, has influenced many languages worldwide and in Cyprus, this does not appear only in the form of loanwords, but, also, in code-switching and code-mixing. Extended research has been conducted within the area of education (Papapavlou & Pavlou 2004, 2007; Yiakoumetti et al. 2006), on the use of the dialect (Papapavlou & Kouridou 2007) and English (Papapavlou 1994, 1997; Goutsos 2005). Among these studies there have been more focusing

Discourse Analysis - Gender - Lexicography 8 5 Major Trends in Theoretical and Applied Linguistics

on the issue of language attitudes. Although the most frequent conclusion is that people are more in favour of the standard variety rather than the dialect (Papapavlou 1998), what has been recently observed is that negative feelings towards GCD are declining (Tsiplakou 2003; Themistocleous 2007; Papapavlou & Sophocleous 2009). On the other hand, although English obtains an important position in people’s speech, the native codes are still not threatened since they will always be an inseparable part of the community’s identity (McEntee- Atalianis & Pouloukas 2001). Given that language attitudes constitute the topic of the present paper, it needs to be noted that the present paper focuses mainly on advertising. The most triggering issue about the language of advertising is that while “with other genres we usually know where our loyalties lie; with ads we are just confused” (Cook 2006: 3). Going a step further, it becomes obvious that attitudes towards the language of advertising become an even more fascinating topic.

2. Attitudes to Language

Attitudes have raised much discussion as to what they are and how they can be measured. Baker (1992) described attitudes as “construct[s] used to explain the direction and persistence of human behavior” (p. 10). Going one step further, Baker refers to the questionable relationship between attitude and action. Attitudes have been claimed to include three components: cognitive, affective and readiness for action. The first one refers to the beliefs related to a linguistic variety, the second one concerns feelings about it and the third one is associated with plans for action. However, in some cases people supported the maintenance of a language, even if they do not hold favourable feelings towards it. But, studies showed that actions may not always work as an indicator of people’s attitudes (Choi 2003). Garrett, Coupland and Williams (2003) discussed the main distinction in approaching language attitudes: ‘societal treatment’, ‘direct measures’ and ‘indirect measures’. The first one, which has been the least used, concerns observation of people’s reaction to a linguistic variety within society. Direct measurement refers to direct elicitation of people’s attitudes and it can be done through questionnaires - with open-ended questions, likert scale etc - or interviews by asking people about their opinion on a linguistic code. Finally, indirect approaches can be observations or deceptive measurement where the participants are not fully aware of the purpose of the study. The most famous indirect approach is the matched-guise technique. It was initially used by Lambert, Hodgson, Gardner and Fillebaum (1960) and it “involves playing recordings of [the same speaker] reading aloud the same passage of prose but using different accents, dialects or languages [as guises]” (Trudgill 1992: 51). After listening to each recording, the judges are asked to evaluate the

8 6 Discourse Analysis - Gender - Lexicography Andreas Papapavlou and Melanie Satraki

speaker based on a questionnaire which usually includes a semantic differential scale of opposite adjectives.

3. Advertising and the Field of Linguistics

Although at first sight advertising seems to be solely a component ofthe marketing process, nobody can doubt that advertising is a kind of communication. Advertising is not as straightforward as other genres and aims at promoting products or promoting identities (Martin 2006; Piller 2001). Advertising language differs from ordinary language and it is so powerful that it manages to enter people’s everyday language use (Kelly-Holmes 2005). At a purely linguistic level, features appearing in advertisements concern phonology, morphology, syntax, semantics and pragmatics and have been extensively studied by linguists (Bruthiaux 1996; Cook 2006; Dyer 1982; Harris et al. 2002, Tanaka 1994; Vestergaard & Schrøder 1985; Wyer 2002). Nevertheless, linguistics goes even deeper in advertising since phenomena of bi- / multilingualism and bi- / multidialectism are observed. Although existing research on these issues is not as rich as on the ones mentioned above, researchers have already come up with some interesting conclusions.

3.1. The Use of Foreign Languages in Advertising

As Kelly-Holmes argues, the use of foreign languages in advertising is ‘symbolic’ since it is recognised as a way of “constructing identities for the product […] rather than communicating facts” (2005: 65). A general conclusion of the studies drawn on multilingual advertising is that certain languages are associated with certain products cross-culturally. Hence, French attributes fashion and elegance, Spanish is used as a sign of freedom and adventure, German signifies “reliability, precision, and superior technology” and Italian is linked to “good food and a positive attitude toward life” (Piller 2003: 172-173). As far as English is concerned, it is the most frequent language appearing in multilingual advertising and, although it may be associated with the cowboy spirit, youth culture and hip hop rebellion, “internationally, English has become a general symbol of modernity, progress, and globalization” (175). Nevertheless, the role of the lingua franca in multilingual advertisements may be different from that of any other language. Piller argues that even if the recipients do not speak the foreign language used in an advertisement, they will be able to activate the stereotypes about that culture and “transfer them to the product” (Piller 2001: 163). Besides, this unique role of English in advertising has been supported by research within different linguistic settings (Rash 1996; Chen 2006; Krishnasamy 2007; Hornikx et al. 2010; Planken et al. 2010).

Discourse Analysis - Gender - Lexicography 8 7 Major Trends in Theoretical and Applied Linguistics

3.2. The Use of Nonstandard Varieties in Advertising

It is claimed that although at the beginning advertising adopted a formal register, gradually it has come to adopt the characteristics of everyday communication (Kelly-Holmes 2005). This change prompted advertisers to use indigenous varieties to “create a feeling of community and to associate this with a particular product” (109). For instance, Mutonya (2008) argues that advertisers in Nairobi use nonstandard forms to appeal to a greater number of audiences and portray the contemporary identity of the community and address taboo topics. On the other hand, Atkinson and Kelly-Holmes find the standard Castilian to be used asa ‘globalising’ language and to show modernity, whereas nonstandard Catalan stands as a means of establishing either solidarity or “local and often ‘traditional’ associations” (2006: 255).

4. Advertising in the Greek-Cypriot Community

When the British took over the administration of Cyprus from the Turks in 1878, the colonizers permitted the publication of the first Cypriot newspaper CYPRUS which included the first commercial advertisements. At that time, these ‘announcements’ were written in puristic Greek (“katharevousa”). However, at the beginning of the 20th century, the archaic forms started gradually being simplified and, by the 1960s, Demotic Greek became the language of literary texts. Nonetheless, during this linguistic transition, Cypriot dialect features have been observed in advertisements of as early as 1912, in order to create a satirical effect (Sophocleous 2004). The satirical use of the dialect still exists nowadays. But, research on the language of media and advertising has been limited (Karyolemou 1994, Pavlou 1992, 2001, 2004). Pavlou (1992) aimed at investigating dialect use in radio commercials. Pavlou claims that GCD is – successfully – used not only to create humourous effects, but also, to promote traditional food, products competing similar imported ones and products associated with the maintenance of traditional values. The reason why the dialect is preferred rather than the standard in advertising such products lies in the belief that recipients need to identify the lifestyle they employ with the product, and validate it through the depiction of it in other domains of their life. Moreover, Pavlou (2001) further investigates the appearance of foreign language elements and how language choice in advertising functions as a ‘carrier’ of a desired Cypriot identity. Based on the view that advertising serves as a powerful sociopolitical tool used to promote values and identities, Pavlou concluded that advertising promotes a kind of bourgeois society, since products

8 8 Discourse Analysis - Gender - Lexicography Andreas Papapavlou and Melanie Satraki

reflecting such a lifestyle are advertised through the use of SMG and foreign language elements. By contrast, the dialect is used to address lower-class people and, thus, promoting a traditional way of life. In summing up, from the above-mentioned research on the language used in the Cypriot advertising, one realises that further investigation is essential. While previous efforts have investigated the use of the local and foreign varieties in advertising, none of the studies focused on people’s view of the appearance of GCD, SMG and English features. Specifically, the aim of the current study is to provide answers to the following questions:

a) Do Greek-Cypriots have positive attitudes towards the appearance of GCD, SMG and English in advertising? b) Does the stereotypical nature of a product (e.g. traditional or foreign cul- ture-specific value) affect Greek-Cypriots’ attitudes? c) Does the medium (oral or written) used for the advertisements influence Greek-Cypriots’ attitudes?

5. The Study

5.1. Participants

A total of 38 adult Greek Cypriots (20 males and 18 females) participated in the main study. The participants were randomly selected from middle-class families, from both urban and rural areas of Cyprus.

5.2. Design

For the purpose of the study, a two-part questionnaire was designed. The questionnaire consisted of four closed questions. After providing some personal information, the participants were asked to answer questions based on self- made advertisements designed by the experimenters along with the assistance of a professional Cypriot advertiser. The language of the advertisement was decided to be closer to what is considered ‘correct Cypriot’ (i.e. Tsiplakou et al. 2005), since a more basilectal level might have triggered negative feelings towards a variety not used anymore in contemporary Cyprus (Papapavlou & Sophocleous 2009). For Part A, the participants were asked to study three categories of advertisements and then answer three questions. Further details about the advertisements are found in Table 1.

Discourse Analysis - Gender - Lexicography 8 9 Major Trends in Theoretical and Applied Linguistics

Table 1. Features of self-made advertisement (Part A)

CATEGORY I STEAKHOUSE CATEGORY II ZIVANA CATEGORY III CLOTHING

Ib Ic IIa IIc IIIa IIIc Ia SMG with SMG SMG with IIb SMG SMG IIIb SMG with SMG English with GCD English SMG with GCD with GCD SMG English features features features features features features

The selection of the above products was based on how these are usually perceived by the members of the community under study: the steakhouse signifies luxury and foreignness, zivana–a local drinking spirit–is associated with the local tradition, and casual clothing is normally considered as neutral since it is part of everyday life. For the first two questions, participants were instructed to evaluate the three advertisements of each category by comparing them and ranking them (1, 2, 3 from high to low) with the use of a table of several items (see Tables 3 and 4). Whereas in Question 1 they were asked to evaluate the advertising text in terms of aesthetics, vocabulary and target audience, for Question 2, they should evaluate the advertised product in terms of quality, price and target audience. The way of perception of the items is left to the judges (McGroarty 1996). These questions studied attitudes towards the three codes, while, at the same time, they would reveal any influence by the nature of the product. Then, Question 3 should be answered through categorization of the advertisements as whether they should be presented in oral media (radio, TV etc) or written media (magazine, newspaper, leaflet etc). This was an effort to find out whether people’s views of the codes are affected by the mode of communication – oral or written. Finally, Part B included a question asking participants to rank (1, 2, 3 from high to low) three advertising texts of each of the four products given. Although the format of the question differed from that of Question 1, the purpose was similar. Further details are found in Table 2.

Table 2. Features of self-made advertisement (Part B)

ENGLISH CYPRIOT TAVERN CHAMPAGNE FORMAL CLOTHING BOOKSHOP SMG SMG SMG SMG features features features features features features features features SMG with GCD SMG with GCD SMG with GCD SMG with GCD SMG with English SMG with English SMG with English SMG with English

9 0 Discourse Analysis - Gender - Lexicography Andreas Papapavlou and Melanie Satraki

Before distribution, the questionnaire was piloted for clarity of the questions used in the questionnaire and authenticity of language of the advertisements.

6. The results

The data collected from the questionnaires were analysed and the findings are presented in four different subsections with the use of tables and figures. Each of the subsection corresponds to each of the four questions included in the questionnaire. It must be noted that the higher the score, the lower the preference, since participants indicated the order of the most preferred advertisement (numbered as 1) to the least preferred one (numbered as 3).

6.1. Evaluation of Advertising Texts

The first question of the questionnaire asked participants to express their preference of the advertising texts (a, b, c) in each category (I, II, III) comparatively, and rank them (1, 2, 3 from high to low) based on criteria such as aesthetics, vocabulary and target audience. The results are presented in Table 3, in the order given to the participants—random ordering prevents mechanical answering.

Table 3. Mean evaluation of advertising texts (Question 1)

STEAKHOUSE ZIVANA CLOTHING

SMG ENG GCD ENG SMG GCD GCD SMG ENG

AESTHETICS

Pleasant 1,68 1,95 2,37 2,39 1,71 1,89 2,08 1,82 2,11

Modern 1,76 1,63 2,61 1,39 1,95 2,66 2,50 1,76 1,68

Attractive 1,47 1,89 2,66 2,29 1,55 2,16 2,39 1,76 1,84

VOCABULARY

Suitable 1,32 1,92 2,76 2,42 1,32 2,26 2,37 1,58 2,05

Functional 1,29 1,89 2,82 2,24 1,53 2,24 2,42 1,58 2,00

Familiar 1,84 2,08 2,05 2,53 1,97 1,68 2,03 1,79 2,18

TARGET AUDIENCE

High economic status 1,58 1,58 2,84 1,82 1,61 2,58 2,45 1,68 1,87

High educational status 1,42 1,76 2,82 1,79 1,68 2,53 2,47 1,63 1,89

Discourse Analysis - Gender - Lexicography 9 1 Major Trends in Theoretical and Applied Linguistics

The mean values of each code for each product are presented in Figure 1.

3

2 M 1

0 Steakhouse Zivana Clothing SMG 1,55 1,67 1,70 GCD 2,62 2,25 2,34 ENG 1,84 2,11 1,95

Figure 1. Mean evaluation of GCD, SMG and English (Question 1)

A general observation is that for all three products, SMG is the most preferred code, whereas GCD is the one with the least preference. However, differences are detected in the deviation between the codes’ values from one product to the other. SMG and English are mostly preferred in steakhouse advertising, whereas GCD is mostly preferred for zivana (Figure 1). Moreover, although results of SMG seem to be homogeneous from one product to the other, interesting points are raised from the results of GCD and English. Although GCD is generally the least preferred code, it is found to be the most pleasant and familiar in the advertisement of zivana. Also, English is reported to be the least familiar code, but, the most modern one in advertising zivana and clothing, signaling a high economic and educational status. Last, for these two advertisements English is considered as the least attractive, suitable and functional, too.

6.2. Evaluation of Advertised Products

The same analysis was followed for Question 2, where the participants show their preference of the advertised products of the same advertisements of each category and rank them in the same way, but, based on criteria of quality, price and target audience. The results are recorded in Table 4.

9 2 Discourse Analysis - Gender - Lexicography Andreas Papapavlou and Melanie Satraki

Table 4. Mean evaluation of advertised products (Question 2).

STEAKHOUSE ZIVANA CLOTHING

SMG ENG GCD ENG SMG GCD GCD SMG ENG

QUALITY

High 1,21 1,87 2,92 2,24 1,53 2,24 2,63 1,50 1,87

Reliable 1,21 2,03 2,76 2,21 1,63 2,16 2,47 1,50 2,03

PRICE

Expensive product 1,66 1,58 2,76 1,61 1,74 2,66 2,61 1,79 1,61

TARGET AUDIENCE

High economic status 1,55 1,68 2,76 1,76 1,63 2,61 2,55 1,95 1,68

High educational status 1,45 1,74 2,82 1,89 1,66 2,45 2,58 1,66 1,76

The mean values of each code for each product are given in Figure 2.

3

2 M 1

0 Steakhouse Zivana Clothing SMG 1,42 1,64 1,68 GCD 2,60 2,42 2,57 ENG 1,78 1,94 1,79

Figure 2. Mean evaluation of GCD, SMG and English (Question 2)

What is mainly observed is that again SMG occupies the first place in people’s preference, English is the one that follows and GCD comes last. Nevertheless, it is worth pointing out that English corresponds to the most expensive product in all three categories and adds higher economic status to clothing advertisement. On the contrary, this code is associated with least reliability and equally low quality to GCD in advertising zivana, although its quality is quite unreliable for the other products too.

Discourse Analysis - Gender - Lexicography 9 3 Major Trends in Theoretical and Applied Linguistics

6.3. The infl uence of the Medium

Further on, Question 3 instructed participants to categorise the advertisements according to which kind of media is most suitable for each of them to appear in. The fi rst category included oral media (eg. radio or TV) and the second category concerned written media (eg. magazine, newspaper or leafl et). Participants could write the number of the advertisements in both categories, if they regarded them suitable for both kinds (see Figures 3 and 4).

ORAL GCD 70,18% MEDIA ENG 55,26% SMG 44,74% ORAL GCD 70,18% MEDIA ENG 55,26% SMG 44,74%

Figure 3. Advertisements/Codes suitable for oral media (Question 3)

WRITTEN

MEDIA SMG 68,42%

ENG 53,51% WRITTEN GCD 34,21% MEDIA SMG 68,42%

ENG 53,51%

GCD 34,21%

Figure 4. Advertisements/Codes suitable for written media (Question 3)

9 4 Discourse Analysis - Gender - Lexicography

Andreas Papapavlou and Melanie Satraki

The initial conclusion drawn from the fi gures is that GCD is perceived to be more suitable for oral media while SMG is the least suitable variety for such kind of mass communication. On the other hand, the opposite happens with written media. SMG occupies the fi rst position whereas GCD comes at the bottom of the fi gure. However, SMG in oral media is accepted by more people than GCD in written media. At last, English comes in the middle position of both fi gures since half of the participants prefer it in oral presentation and half of them in writing.

6.4. Suitability of the Code Used

Finally, the participants were asked to rank (1, 2, 3 from high to low) three advertising texts of four products (Cypriot tavern, champagne, formal clothes and English bookshop), according to which they believe to be the most suitable for each product. The results are presented in Figure 5.

3 2,5 2

M 1,5 1 0,5 0 Cypriot Champagne Formal English tavern clothes bookshop SMG 1,13 1,29 1,32 1,18 GCD 2,47 2,74 2,87 2,76 ENG 2,39 2,00 1,82 2,05

Figure 5. Mean value of GCD, SMG and English (Question 4)

The results show that GCD is regarded as the least suitable code for all products with the highest suitability being observed in the advertisement of Cypriot tavern and the least one in that of formal clothing. Once again, SMG is the code mostly preferred, especially for promoting the Cypriot tavern and then the English bookshop. Similarly, English still occupies the middle place with the least suitability being found in the advertisement of the Cypriot tavern. The English bookshop, champagne and formal clothing follow from less suitable to more suitable.

Discourse Analysis - Gender - Lexicography 9 5

Major Trends in Theoretical and Applied Linguistics

7. Discussion and Conclusions

Bi-/multilingual and bi-/multidialectal settings have always been a challenge to linguists due to their palpable linguistic wealth that can bring about remarkable conclusions. Such settings, in combination with equally intriguing domains of language use, fascinate experts in language studies. This has provided the stimulation for the present study which focuses on advertising in the Greek-Cypriot community. Based on limited available literature, one can hypothesize that Greek-Cypriots have recently begun to hold a more favourable attitude towards dialect use in advertisements, especially of products reflecting local tradition. Similarly, English should be preferred for products of foreign origin or luxury. Hence, the following questions can be addressed:

Do Greek-Cypriots have positive attitudes towards the appearance of GCD, SMG and English in advertising?

From the results obtained from Questions 1, 2 and 4, it can be concluded that Greek-Cypriots have more positive attitudes towards SMG. English is the second preferred code in advertising, whereas the local dialect comes last (Figures 1, 2 and 3). Deviations from this pattern were found in some products of Questions 1 and 2, for some traits. However, none of them affected the ordering of the codes. People’s preference of SMG supports earlier studies that showed greater preference towards the standard variety rather than the dialect (Karyolemou 1994; Papapavlou 1998, 2007; Papapavlou & Sophocleous 2009). This is rather a universal phenomenon which shows that in many linguistic settings around the world people have more favourable views of the standard variety (Kalmar et al. 1987; Bettoni & Gibbons 1988; Luhman 1990; Adegbija 1994; Beckford Wassink 1999; de V. Cluver 2000; Broermann 2007). The finding showing that Greek-Cypriots prefer a foreign code rather than their mother tongue raises several concerns. A possible explanation is that the dialect does not have a writing system and it is used only in oral communication. The high position of English in professional domains has also been observed by McEntee-Atalianis and Pouloukas (2001). This study additionally found that English does not constitute a threat for the two native codes. If the written version of the advertisements did not play a role in the participants’ judgment, people’s preference for English may have grown from 2001 until today. Such a situation may be due to such factors as globalization and technological development where access is achieved through the lingua franca – like in the case of Africa (Adegbija 1994).

9 6 Discourse Analysis - Gender - Lexicography Andreas Papapavlou and Melanie Satraki

Does the stereotypical nature of a product (eg. traditional or foreign culture-specific value) affect Greek-Cypriots’ attitudes?

In studying Greek-Cypriots’ preference of the language of advertising, it is worth stating that the stereotype, or nature, or value reflected in the advertised product does not seem to change people’s views (Questions 1, 2 and 4). Nevertheless, minor stereotype influence is observed, in cases people judged with the use of traits (Questions 1 and 2), even though it does not affect the final result. In the case of the foreign and luxury product (steakhouse), people consider English intervention as the most modern, but, the least familiar text, advertising the most expensive product. Moreover, in the case of a product which is of Cypriot origin and tradition (zivana), GCD is thought to be the most familiar kind of language. Similarly, in Question 4, GCD is more acceptable for the Cypriot tavern than any other product. On the other hand, English carries the label of the most modern in advertising zivana, but, the least pleasant, attractive, suitable and familiar code, adding unreliability and cost to the product. This can serve as a counterbalance to the previous concern since it can be a sign that English remains a foreign element of no trust within the community.

Does the medium (oral or written) used to present the advertisements influence Greek-Cypriots’ attitudes?

From the answers to Question 3 of the questionnaire, we detect influence of the medium of presentation on people’s view of the advertising language. The written mode is more preferable for SMG and, then, English. GCD is the least favoured code for written media, but, it is the first appearing in oral presentation. What has most likely driven people to such a categorization is the fact that GCD is the only investigated code that has not been codified yet; therefore, it remains unacceptable or ´weird` for it to appear in written form. An alternative explanation is that writing is a more formal kind of communication – where people are expected to use SMG – whereas orality is more associated with casualty and everyday life – where people use GCD. Additionally, English can be found in written advertisements, but, not as often as a more native and comprehensible code (SMG). People’s support for English more than SMG in oral media is rather interesting. This can stand as an indication of an already existing situation; the inadequate usage of SMG. Although SMG is the official language of the state, in trying to speak SMG, Greek-Cypriots end up with a mixture called Cypriot Standard Greek (Arvaniti 2006).Conversely, English language’s dominance over SMG in oral communication shows that people are more used to or in favour of listening and speaking English rather than SMG. Previous research has recognized the

Discourse Analysis - Gender - Lexicography 9 7 Major Trends in Theoretical and Applied Linguistics

important role of English in Cyprus, but, it has been claimed that it is related to certain social contexts (Goutsos 2005) and people “share a common understanding of their national linguistic identity, cultural heritage and language preferences” that do not let foreign elements to weaken it (McEntee-Atalianis & Pouloukas 2001:35). However, anglicisms are constantly increasing, changing people’s linguistic preferences. In conclusion, as Kelly-Holmes (2005) argues, while advertisers appear to support that the nature of the product requires a certain linguistic code that is best for promoting it, recipients themselves do not seem to share this view . The present study demonstrated that while people are not absolutely negative towards the use of GCD in advertising local products, they still prefer SMG as in other domains of their life (Karyolemou 1994; Papapavlou 1998, 2007; Papapavlou & Sophocleous 2009 etc). This view is more firm in the case of written media, where SMG is the most acceptable code, by contrast to GCD which is preferred for oral presentation. Additional studies are needed to further explore the intricacies of this phenomenon.

9 8 Discourse Analysis - Gender - Lexicography Andreas Papapavlou and Melanie Satraki

References

Adegbija, E. 1994. Language Attitudes in Sub-Saharan Africa: A Sociolinguistic Overview. Clevedon: Multilingual Matters.

Arvaniti, A. 2006. Linguistic practices in Cyprus and the emergence of Cypriot Standard Greek. In San Diego Linguistic Papers, 2: 1-24.

Atkinson, D. and H. Kelly-Holmes. 2006. Linguistic normalisation and the market: Advertisingand linguistic choice in El Periódico de Catalunya. In Language Problems and Language Planning, 30 (3). John Benjamins, 239-260.

Baker, C. 1992. Attitudes and Language. Clevedon: Multilingual Matters Ltd.

Beckford Wassink, A. 1999. Historic low prestige and seeds of change: Attitudes toward Jamaican Creole.In Language in Society, 28. Cambridge University Press, 57-92.

Bettoni, C. and J. Gibbons. 1988. Linguistic purism and language shift; a guise- voice study of the Italian community in Sydney. In International Journal of the Sociology of Language, 72. Amsterdam: Mouton de Gruyter, 15-35.

Broermann, M. 2007. Language attitudes among minority youth in Finland and Germany. In International Journal of the Sociology of Language, 187/188: 129- 160.

Bruthiaux, P. 1996. The Discourse of Classified Advertising. Oxford: Oxford University Press.

Chen, W.-Y.C. 2006. The mixing of English in magazine advertisements in Taiwan. In World Englishes, 25 (3/4). Oxford: Blackwell, 467-478.

Discourse Analysis - Gender - Lexicography 9 9 Major Trends in Theoretical and Applied Linguistics

Choi, J. 2003. Language attitudes and the future of bilingualism: the case of Paraguay. In International Journal of Bilingual Education and Bilingualism, 6 (2): 81-94.

Cook, G. 2006. The Discourse of Advertising (Second Edition). London, New York: Routledge.

de V. Cluver, A. 2000. Changing language attitudes: The stigmatisation of Khoekhoegowap in Namibia. In Language Problems and Language Planning, 24 (1). Amsterdam: John Benjamins Publishing Company, 77-100.

Dyer, G. 1982. Advertising as Communication. London, New York: Methuen & Co. Ltd.

Garrett, P., N. Coupland and A. Williams. 2003. Investigating Language Attitudes. Cardiff: University of Wales Press.

Goutsos, D. 2005. A discourse-analytic approach to the use o English in Cypriot Greek conversations. In A. Papapavlou (ed.), Contemporary Sociolinguistic Issues in Cyprus. Thessaloniki: University Studio Press, 181-212.

Harris, R., R. Sturm, M. Klassen and J. Bechtold. 2002. Language in advertising: a psycholinguistic approach. In Original Research and Theoretical Contributions. EBSCO Publishing, 1-26.

Hornikx, J., F. van Meurs and A. Boer. 2010. English or a local language in advertising? The appreciation of easy and difficult English slogans in the Netherlands. In Journal of Business Communication, 47 (2). Nijmegen: Association for Business Communication, 169-188.

Kalmar, I., Z. Yong and X. Hong. 1987. Language attitudes in Guangzhou, China. In Language in Society, 16. Cambridge University Press, 499-508.

Karyolemou, M. 1994. Linguistic attitudes and metalinguistic discourse: An investigation in the Cypriot press. In I. Philippaki-Warburton, K. Nicolaidisand M. Sifianou (eds.),Themes in Greek Linguistics. Amsterdam: John Benjamins, 253-259.

Karyolemou, M. 2001. From linguistic liberalism to legal regulation: The Greek language in Cyprus. In Language Problems and Language Planning, 25 (1): 25-51.

Kelly-Holmes, H. 2005. Advertising as Multilingual Communication. New York: Palgrave Macmillan.

1 0 0 Discourse Analysis - Gender - Lexicography Andreas Papapavlou and Melanie Satraki

Krishnasamy, K. 2007. English in Tamil: The language of advertising. In English Today, 23 (3/4). Cambridge University Press, 40-49.

Lambert, W., R. Hodgson, R. Gardner and S. Fillenbaum. 1960. Evaluational reactions to spoken language. In Journal of Abnormal and Social Psychology, 60: 44-51.

Luhman, R. 1990. Appalachian English stereotypes: Language attitudes in Kentucky. In Language in Society, 19. Cambridge University Press, 331-348.

Martin, E. 2006. Marketing Identities through Language: English and Global Imagery in French Advertising. New York: Palgrave Macmillan.

McEntee-Atalianis, L. and S. Pouloukas. 2001. Issues of identity and power in a Greek-Cypriot community. In Journal of Multilingual and Multicultural Development, 22 (1): 19-38.

Mutonya, M. 2008. Swahili advertising in Nairobi: Innovation and language shift. In Journal of African Cultural Studies, 20 (1). Routledge, 3-14.

Papapavlou, A. 1994. Language Contact and Lexical Borrowing in the Greek Cypriot Dialect: Sociolinguistic and Cultural Implications. Athens: N.C. Grivas Publications.

Papapavlou, A. 1997. The influence of English and its dominance in Cyprus: Reality or unfounded fears? In Journal of Mediterranean Studies, 7: 1-32.

Papapavlou, A. 1998. Attitudes towards the Greek Cypriot dialect: Sociocultural implications. In International Journal of the Sociology of Language, 134: 15-28.

Papapavlou, A. 2007. Language policy and planning: The role of language attitudes in decision making. In A. Papapavlou and P. Pavlou (eds.), Sociolinguistic and Pedagogical Dimensions of Dialects in Education. Newcastle: Cambridge Scholars Publishing, 194-215.

Papapavlou, A. and T. Kouridou. 2007. Bidialectism and metalinguistic awareness. In A. Papapavlou and P. Pavlou (eds.), Sociolinguistic and Pedagogical Dimensions of Dialects in Education. Newcastle: Cambridge Scholars Publishing, 216-235.

Papapavlou, A. and P. Pavlou. 1998. A review of the sociolinguistic aspects of the Greek Cypriot Dialect. In Journal of Multilingual and Multicultural Development, 19 (3): 212-220.

Discourse Analysis - Gender - Lexicography 1 0 1 Major Trends in Theoretical and Applied Linguistics

Papapavlou, A. and P. Pavlou. 2004. Issues of dialect use in education from the Greek Cypriot perspective. In International Journal of Applied Linguistics, 14 (2). Oxford: Blackwell Publishing, 243-259.

Papapavlou, A. and P. Pavlou. 2007. The interplay of bidialectalism, literacy and educational policy. In A. Papapavlou and P. Pavlou (eds.), Sociolinguistic and Pedagogical Dimensions of Dialects in Education. Newcastle: Cambridge Scholars Publishing, 101-121.

Papapavlou, A. and A. Sophocleous. 2009. Relational social deixis and the linguistic construction of identity. In International Journal of Multilingualism:1-16.

Pavlou, P. 1992. The use of the Cypriot-Greek Dialect in the commercials of the Cyprus Broadcasting Corporation. Paper presented at the 37th Conference on the International Linguistics Association, Washington DC.

Pavlou, P. 2001. The use of dialectal and foreign language elements in radio commercials and their function in the construction of contemporary Cypriot identity. Paper given at the international conference Mèditerrannée: Ruptures et Continuités, Nicosia, Cyprus.

Pavlou, P. 2004. Greek dialect use in the mass media in Cyprus. In International Journal of the Sociology of Language, 168. Walter de Gruyter, 101-118.

Piller, I. 2001. Identity constructions in multilingual advertising. In Language in Society 30.Cambridge: Cambridge University Press, 153-185.

Piller, I. 2003. Advertising as a site of language contact. In Annual Review of Applied Linguistics 23. Cambridge: Cambridge University Press, 170-183.

Planken, B., F. van Meurs and A. Radlinska. 2010. The effects of the use of English in Polish product advertisements: Implications for English for business purposes. In English for Specific Purposes, 29. Nijmegen: Elsevier, 225-242.

Rash, F. 1996. Attitudes to the use of English in Swiss German advertising language. In The Web Journal of Modern Language Linguistics, 1. University of Newcastle upon Tyne: School of Modern Languages.

Sophocleous, A. 2004. The History of Advertising in Cyprus. Nicosia: Laiki Group Cultural Centre.

1 0 2 Discourse Analysis - Gender - Lexicography Andreas Papapavlou and Melanie Satraki

Tanaka, K. 1994. Advertising Language: A Pragmatic Approach to Advertisements in Britain and Japan. London, New York: Routledge.

Themistocleous, C. 2007. Written Cypriot Greek in online chat: Usage and attitudes. In Proceedings of the 17th International Postgraduate Linguistics Conference. Manchester: Department of Linguistics and English language, 473- 488.

Trudgill, P. 1992. Introducing Language and Society. London: Penguin English.

Tsiplakou, S. 2003. Attitudes towards language and language change: A two-way relationship? [in Greek]. In Proceedings of the 6th International Conference on Greek Linguistics.

Tsiplakou, S., A. Papapavlou, P. Pavlou and M. Katsoyannou. 2005. Levelling, koineization and their implications for bidialectism. In Language Variation— European Perspectives: Selected Papers from the 3rd International Conference on Language Variation in Europe. John Benjamins, 265-276.

Varella, S. 2006. Language Contact and the Lexicon in the History of Cypriot Greek. Bern: Peter Lang.

Vestergaard, T. and K. Schrøder. 1985. The Language of Advertising. Oxford: Blackwell.

Wyer, R. 2002. Language and advertising effectiveness: mediating influences of comprehension and cognitive elaboration. In Psychology and Marketing 19 (7-8). Wiley InterScience, 693-712.

Yiakoumetti, A., M. Evans and E. Esch. 2006. Language awareness in a bidialectalsetting: The oral performance and language attitudes of urban and rural students in Cyprus. In Language Awareness, 14 (4). London: Routledge, 254- 260.

Discourse Analysis - Gender - Lexicography 1 0 3 Major Trends in Theoretical and Applied Linguistics

Appendix A Advertisements

CATEGORY I

a) JACK’S steakhouse Για ‘σας που το φαγητό είναι απόλαυση και που πάντα ζητάτε τις πιο ξεχωριστές γεύσεις. Εκλεπτισμένα πιάτα, φτιαγμένα με υλικά άριστης ποιότητας και επιμελημένα από κορυφαίους μάγειρες.

b) STEVEN’S steakhouse Αν δεν αρκείστε στο fast food και είστε λάτρης της gourmet κουζίνας, το Steven’s steakhouse είναι εδώ για ‘σας. Γεύματα από τα χέρια καταξιωμένων chef που μπορείτε να απολαύσετε σε μια chic ατμόσφαιρα.

c) KEN’S steakhouse Για πολυτέλεια τζαι καλό φαϊν ελάτε τζαι ‘σεις στο Ken’s. Σερβίρουμε φαγητά από την διεθνή κουζίνα που ικανοποιούν όλα τα γούστα. Ken’s steakhouse για γεμάτα στομάχια που δεν σκέφτονται...την πούγκαν τους. CATEGORY II

a) Ζιβάνα ΧΟΛΕΤΡΙΑ Cypriot spirit only by ΧΟΛΕΤΡΙΑ. Η προτίμηση των προγόνων μας είναι και δική μας. Αφού πέρασε τόσα tests μπορούμε να την χρησιμοποιούμε άφοβα και για τα δικά μας parties.

b) Ζιβάνα ΤΡΕΙΣ ΕΛΙΕΣ Ντόπια ζιβάνα μόνο από τις ΤΡΕΙΣ ΕΛΙΕΣ. Εμπνευσμένη από την παράδοση της Κύπρου και σαν κληρονομιά από τους προγόνους μας.

c) Ζιβάνα ΠΑΛΙΟΜΥΛΟΣ Το ποτόν των παππούων μας που επέρασεν που γενιά σε γενιάν. ΠΑΛΙΟΜΥΛΟΣ γιατί τούτον μας εμάθαν οι παππούες μας τζαι τούτον ακολουθούμεν! CATEGORY III

a) Κατάστημα ΟΛΑ ΣΕ ΕΝΑ Για τη μάμμα, τον παπάν, την κόρην τζαι το γιόν! Στο κατάστημα μας μπορείτε να ντύσετε όλη την οικογένεια. Τζαι μεν έσιετε έννοιαν που θα πάτε: Δουλειά; Σπίτι; Έξοδο; Εμείς αναλαμβάνουμε για όλα!

1 0 4 Discourse Analysis - Gender - Lexicography Andreas Papapavlou and Melanie Satraki

b) Κατάστημα ΓΙΑ ΣΕΝΑ Ρούχα για τον άνδρα, τη γυναίκα, το παιδί. Ρούχα για να φορέσεις οποιαδήποτε στιγμή της ζωής σου: στο γραφείο, στο πάρκο, για καφέ. Ρούχα που μπορείς να αγοράσεις ακόμα και σε περίοδο οικονομικής κρίσης.

c) Κατάστημα ΦΟΥΣΤΑ-ΠΑΝΤΕΛΟΝΙ Φούστες, jeans, υποκάμισα, T-shirts σε όλα τα sizes. Γεμίστε το trolley σας άφοβα. Enjoy your shopping! Ελάτε στο κατάστημα μας για να φροντίσουμε όλη την οικογένεια σας ντύνοντας τη με τα πιο cool ρούχα.

Discourse Analysis - Gender - Lexicography 1 0 5 Major Trends in Theoretical and Applied Linguistics

Appendix B Questions

ΜΕΡΟΣ A ΠΡΟΣΩΠΙΚΑ ΣΤΟΙΧΕΙΑ Φύλο: ...... Ηλικία: ...... Τόπος διαμονής: ......

ΜΕΡΟΣ Β Τα παραπάνω διαφημιστικά έχουν χωριστεί σε τρεις κατηγορίες (Ι, ΙΙ και ΙΙΙ). Η κάθε κατηγορία αποτελείται από τρία διαφημιστικά (α, β και γ).

1. Αξιολογείστε τα τρία διαφημιστικά κείμενα (α, β, γ) της κάθε κατηγορίας (Ι, ΙΙ, ΙΙΙ) μεταξύ τους, με βάση τα πιο κάτω κριτήρια. Βάλτε σε σειρά προτίμησης, χρησιμοποιώντας την κλίμακα 1, 2, 3 (1 αντιπροσωπεύει το πρώτο της κατηγορίας για το αντίστοιχο κριτήριο, ενώ 3 αντιπροσωπεύει το τελευταίο).

ΚΑΤΗΓΟΡΙΑ Ι ΚΑΤΗΓΟΡΙΑ ΙΙ ΚΑΤΗΓΟΡΙΑ ΙΙΙ

α β γ α β γ α β γ

ΑΙΣΘΗΤΙΚΗ

Ευχάριστο

Μοντέρνο

Ελκυστικό

ΛΕΞΙΛΟΓΙΟ

Κατάλληλο

Λειτουργικό

Οικείο

ΚΟΙΝΟ ΤΟ ΟΠΟΙΟ ΣΤΟΧΕΥΕΙ

Υψηλό οικονομικό επίπεδο

Υψηλό μορφωτικό επίπεδο

2. Αξιολογείστε τα τρία διαφημιζόμενα προϊόντα (α, β, γ) της κάθε κατηγορίας (Ι, ΙΙ, ΙΙΙ) μεταξύ τους, με βάση τα πιο κάτω κριτήρια. Βάλτε σε σειρά προτίμησης, χρησιμοποιώντας την κλίμακα 1, 2, 3 (1 αντιπροσωπεύει το πρώτο της κατηγορίας για το αντίστοιχο κριτήριο, ενώ 3 αντιπροσωπεύει το τελευταίο).

1 0 6 Discourse Analysis - Gender - Lexicography Andreas Papapavlou and Melanie Satraki

ΚΑΤΗΓΟΡΙΑ Ι ΚΑΤΗΓΟΡΙΑ ΙΙ ΚΑΤΗΓΟΡΙΑ ΙΙΙ

α β γ α β γ α β γ

ΠΟΙΟΤΗΤΑ

Υψηλή

Αξιόπιστη

ΤΙΜΗ

Ακριβό προϊόν

ΚΟΙΝΟ ΤΟ ΟΠΟΙΟ ΣΤΟΧΕΥΕΙ

Υψηλό οικονομικό επίπεδο

Υψηλό μορφωτικό επίπεδο

3. Σε ποιο μέσο μαζικής επικοινωνίας θεωρείτε ότι πρέπει να προβάλλονται κάθε ένα από τα παραπάνω διαφημιστικά μηνύματα; Γράψε τον αριθμό του κάθε διαφημιστικού (Ια, Ιβ, Ιγ, ΙΙα, ΙΙβ, ΙΙγ, ΙΙΙα, ΙΙΙβ, ΙΙΙγ) σε όποια από τις παρακάτω κατηγορίες θεωρείται ότι είναι πιο κατάλληλη.

Λεκτικά Γραπτά (π.χ. ραδιόφωνο, τηλεόραση) (π.χ. περιοδικό, εφημερίδα, φυλλάδα)

ΜΕΡΟΣ Γ 4. Ποιό από τα παρακάτω κείμενα θεωρείτε πως χρησιμοποιεί την πιο κατάλληλη γλώσσα για διαφήμιση των παρακάτω προϊόντων; Βάλτε σε σειρά προτεραιότητας 1, 2, 3, με τον αριθμό 1 να αντιπροσωπεύει την πιο κατάλληλη.

α. Κυπριακή ταβέρνα __ Προσφέρονται όλων των ειδών οι κυπριακοί μεζέδες. Για σκάπουλλους τζαι κορούες, για νέους τζαι παππούες. __ Για απολαυστικές βραδυές με νόστιμους μεζέδες εμπνευσμένους από την κυπριακή παράδοση ελάτε σε εμάς. __ Αν είστε πελάτες που ζητούν super γεύματα, extra service και special προσφορές, ελάτε στην ταβέρνα μας.

Discourse Analysis - Gender - Lexicography 1 0 7 Major Trends in Theoretical and Applied Linguistics

β. Σαμπάνια __ Αν γιορτάζετε κάτι ξεχωριστό, γιορτάστε το με την πολυτέλεια που του αρμόζει. __ Η πολυτέλεια στο ποτήρι σας αλλά τζαι στη ζωή σας. Ποτό για πρίγκιπες τζαι βασιλιάες. __ Γιατί τίποτα δε θα μπορούσε να αποδώσει καλύτερα το glamour ενός special δείπνου.

γ. Ρούχα επίσημα __ Σε ‘μας μπορείτε να βρείτε elegant ρούχα που δείχνουν glamour και prestige. __ Ψάχνετε ρούχα μοναδικά που θα σας αναδείξουν; Μόνο σε εμάς μπορείτε να τα βρείτε. __ Παντρεύκεστε; Χαρτώνεστε; Έσιετε κάποιο σημαντικό γεγονός; Για ρούχα επίσημα, ελάτε κοντά μας.

δ. Αγγλικό βιβλιοπωλείο __ Για bestsellers στα Αγγλικά που θα είναι το passport σας για ένα ταξίδι του μυαλού στο Buckingham Palace. __ Τζαι Εγγλέζος αν μεν είσαι, φτάνει να ξέρεις Εγγλέζικα. Αγγλικά βιβλία για τζείνους που ενδιαφέρουνται για τον αγγλικό πολιτισμό. __ Είστε αγγλομαθής ή σας αρέσει να διαβάζετε αγγλικά βιβλία; Βιβλία για να βελτιώσετε τις γνώσεις σας.

1 0 8 Discourse Analysis - Gender - Lexicography Grammatical Gender and Cognition

Theodossia-Soula Pavlidou and Angeliki Alvanoudi

Aristotle University Thessaloniki [email protected], [email protected]

Abstract

Empirical research on “structural relativity” (Lucy 1996) has indicated that the grammatical structuring of languages impacts on speakers’ thinking. In particular, the experimental investigation of the relationship between grammatical gender and cognition has repeatedly brought evidence to correlations between the two. Some researchers argue, however, that grammatical gender effects appear only in languages with a two-gender system. Others suggest that the language of instructions for the experiments affects speakers’ performance of tasks. The present paper examines these two issues on the basis of experimental data from a three-gender language, Modern Greek, and suggests that the overall picture is not coherent on several levels.

1. Introduction

The aim of this paper is to present the first findings of an on-going project within the framework of linguistic relativity. Research on linguistic relativity, more specifically on what has been called “structural relativity”,1 has shown that different encodings of color, time, space and number, across languages affect speakers’ thinking about these domains e.g. when speakers codify a scene or an event (experiencing for speaking, Levinson 2003: 302-303) or when they talk about it (thinking for speaking, Slobin 1996: 76).2 Along these lines a number of studies have sought to examine the influence of grammatical gender on speakers’ thinking about the world as ‘female’ or ‘male’ (e.g., Sera et al. 1994,

1 Lucy (1996: 41, 52) uses the term ‘structural relativity’ to refer to the (possible) impact of the formal characteristics of a language on the thought or behavior of its speakers. He differentiates this kind of relativity from ‘discursive relativity’, i.e. the (possible) influence of patterns of use of a language on its users’ thought. 2 Cf. e.g. Gumperz & Levinson 1996; Niemeier & Dirven 2000; Pütz & Verspoor 2000; Bowerman & Levinson 2001; Gentner & Meadow 2003.

Discourse Analysis - Gender - Lexicography 1 0 9 Major Trends in Theoretical and Applied Linguistics

Bowers et al. 1999, Sera et al. 2002, Boroditsky et al. 2003, Vigliocco et al. 2005, and partly Mills 1986 as well as Flaherty 2001).3 Some of this research indicates that grammatical gender effects appear systematically in languages with a two-gender system and that they are confined to specific semantic categories. Another finding suggests that the language in which instructions are given may affect the speakers’ behavior in performing the task. Drawing on this tradition of research, our project aims at the examination of two languages with a three-gender system (Greek and German), the impact of different semantic categories, and the influence of the language of instructions. At this stage, only results with respect to the Greek language are available. In what follows, we will first discuss the relevant studies (section 2), on the basis of which our research design (section 3) developed. We will then present some of our results (section 4). Finally, we will conclude with a more general discussion (section 5), embedding our findings in the overall picture of research on grammatical gender and cognition.

2. Previous Research on Grammatical Gender and Cognition

Studies on grammatical gender and cognition have employed various experimental tasks in order to illuminate possible effects of the gender system on the way we think about the world as ‘female’ or ‘male’. One of the methods used is the sex attribution task, in which speakers are asked to assign a female/ male proper name or voice to animals and objects (denoted by words, depicted, or actually presented). For example, Mills (1986) asked young adults and children to give proper names to toys (actual toys in the case of children, words denoting toys in the case of adult subjects). While the majority of adult English subjects attributed male sex to the toys, German adults consistently assigned male sex to animals/objects denoted by masculine nouns and female sex to animals/objects

3 There is another type of studies (e.g. Ervin 1962; Clarke et al. 1981; Konishi 1993; and, partly, Mills 1986 and Flaherty 2001), which investigate whether masculine and feminine nouns carry connotations of masculinity and femininity, respectively, using semantic differential tests. Such studies have been criticized on several grounds. For one, semantic differential tests reveal at best something about speakers’ knowledge of grammatical gender rather than about their thinking of the world as ‘male’ or ‘female’ since participants were asked to judge words (cf. Sera el al. 2002). For another, these studies yield an unclear pattern of results (cf. Vigliocco et al. 2005). A further problem, in our opinion, lies in the association of the adjectives employed in the scales with properties of femininity and masculinity: as long as the criteria for this association are not explicated, such scales remain pretty arbitrary, simply reflecting gender stereotypes.

1 1 0 Discourse Analysis - Gender - Lexicography Theodossia-Soula Pavlidou and Angeliki Alvanoudi

denoted by feminine nouns. However, given that sex attribution to toys tended to align with the younger subjects’ own sex (a tendency that diminished with older age groups), Mills also suggested that grammatical gender cannot be the sole factor influencing sex attribution. Similar research designs were adopted by Topsakal (1995) for Greek and Flaherty (2001) for Spanish and English. Their conclusions are compatible with those of Mills (1986). Later studies, however, argue that the cognitive effects of grammatical gender vary across languages depending on the properties of the gender system itself. Sera et al. (2002), for example, conducted a series of comparative studies of sex attribution in Spanish, French, German and English. Among other things, speakers of these languages were asked to assign a male or a female voice to different items (humans, animals, and artificial or natural objects) presented in pictures.4 Results yielded that speakers of Spanish and French made grammatically consistent gender assignments, while German speakers, aged 5-9 years,5 did not. The effect of grammatical gender was more pronounced when the pictures were accompanied by labels (for subjects older than 7 years of age). Based on these findings, Sera et al. (2002) concluded that the effects of grammatical gender are limited to languages with a two-gender system, in which the high correlation between grammatical and natural gender leads to an overgeneralization of masculine and feminine traits to inanimate objects. With respect to English, which lacks the grammatical category of gender, Sera et al. (2002), but also Sera et al. (1994), found that speakers of this language tend to classify natural objects as ‘female’ and artificial objects as ‘male’. Similar conclusions were reached by Vigliocco et al. (2005), who examined the influence of grammatical gender on speakers’ semantic representations of animals and objects in Italian and German. In a series of meaning-similarity judgment and semantic substitution error tasks, these researchers tested two hypotheses associated with language-learning mechanisms. According to the first (the ‘similarity and gender’ hypothesis), nouns of the same grammatical gender tend to have similar meanings. According to the second (the ‘sex and gender’ hypothesis), the association between grammatical gender (of nouns denoting humans) and sex (of these humans) is generalized to nouns which refer to animals or to nouns for whose denotation sex is not a relevant conceptual dimension. Words and pictures denoting/depicting animals and

4 Half of the pictures were labeled with the noun denoting the depicted item, while the rest had no such labels. 5 No adult German speakers were included in the sample because, according to Sera et al. (2002), it was not possible to find a sufficient number of native monolingual German- speaking adults.

Discourse Analysis - Gender - Lexicography 1 1 1 Major Trends in Theoretical and Applied Linguistics

artifacts were deployed as stimuli in order to check whether the grammatical gender effect is limited to the lexico-semantic representation of words or extends to the corresponding concept of the word. The study yielded such effects in Italian for the semantic category of animals (but not for that of artifacts); in other words, the sex and gender hypothesis was confirmed in this case. However, no comparable effects were observed in German. Vigliocco et al. (2005) thus concluded that grammatical gender effects are limited to languages with a two-gender system because of the high degree of transparent correspondence between the gender of nouns denoting humans and their sex in these languages, and that they are limited to certain semantic categories. Moreover, they argued that grammatical gender effects are limited to tasks that require verbalization, since no effects were detected when only pictures (without labels) were used as stimuli. While the above-mentioned studies explore the influence of grammatical gender on speakers’ thinking in their native language, a series of other studies (Boroditsky & Schmidt 2000; Phillips & Boroditsky 2003; Boroditsky et al. 2003) looks at the influence of grammatical gender on German and Spanish speakers’ mental representations of objects when speakers think in a different, non-native, genderless language, such as English. According to these scholars, conducting experiments in the same language ensures that speakers of different languages understand the stimuli and instructions in the same way. Their experiments (testing the effect of grammatical gender on speakers’ ability to remember word-name pairs, word description, and similarity rating of object-person pairs) provide evidence for the influence of grammatical gender on cognition –for both Spanish and German! More specifically in the memory task, Phillips & Boroditsky (2000) first ‘taught’ the subjects that 24 nouns with different grammatical genders in Spanish and German have specific proper names. They then tested the subjects’ ability to remember these word-name pairs and found that both Spanish and German speakers’ memory for word-name pairs was better “when the gender of the proper name was consistent with the grammatical gender of the word denoting the object than for pairs where the two genders were inconsistent” (Boroditsky et al. 2003: 68). The studies discussed above offer useful insights, but they are not without flaws. For example, it has been argued (Boroditsky et al. 2003) that insex attribution tasks the subjects may be using grammatical gender as a strategy for performing the task itself. Moreover, the fact that explicit mention of sex is made, e.g. when asking the subjects to assign a ‘male’/‘female’ voice to objects, as in Sera et al. (2002), can reveal the purpose of the study and, thus, influence the results (cf. also Vigliocco et al. 2005). On the other hand, alternative methods such as the similarity rating employed by Boroditsky et al. (2003), are open to similar skepticism. For example, in the picture similarity experiment the subjects may also be using grammatical gender as a conscious strategy for rating

1 1 2 Discourse Analysis - Gender - Lexicography Theodossia-Soula Pavlidou and Angeliki Alvanoudi

an object as similar to a male or female person. Besides this, instructions in different languages may lead to different understandings of the task involved, eventually triggering different behaviors, which are not necessarily related to “differences in thought” (Boroditsky et al. 2003: 67). Finally, it is notself- evident that ‘words’ can be equated with ‘objects’ as it is implicitly done in the experiments conducted by Boroditsky et al. (2003).

3. Current Research Design

Taking into account the findings of previous research on grammatical gender and cognition, our study aims at examining a) the possible effects of grammatical gender in two languages, Greek and German, which both have a three-gender system, b) whether effects of grammatical gender are confined to certain semantic categories (animals) but do not affect others (inanimate objects), and c) whether the language in which instructions for the experimental tasks are given affects the results. Given, though, that none of the relevant studies is without weaknesses, we chose to adopt two experimental tasks that have been prominent in previous research, adapting them for our purposes in order to allow for comparability of the findings. For both tasks two different conditions were tested: in the first case (A) both instructions and stimuli were given in Greek, while in the second (B) stimuli were given in Greek but instructions in English. All participants (70 university students for the first and another 38 for the second task) were native speakers of Greek, equally divided into women and men. Additionally, those participants who conducted the task under condition (B) had a very good knowledge of the English language.6 Task one involved sex attribution to objects/animals/persons shown on pictures through the use of proper names distinct for men-women.7 A set of 40 colored pictures or drawings depicting persons, animals and inanimate objects (both natural and artificial) together with a list of 20 Greek proper names for women and men in equal numbers was presented to all participants. The choice of pictures was made on the basis of two criteria: a) following Sera et al. (2002) and Boroditsky et al. (2003), the depicted items were matched for

6 We would like to express once again our sincere thanks to all students who participated in our experiments and to the staff of the School of Philology and the School of English Language and Literature who enabled our access to their audience during classes. Special thanks go to Despina Papadopoulou for providing us with the E-Prime software. For his assistance with the SPSS we would also like to thank Apostolis Batsidis. 7 This is basically a modified version of the sex attribution tasks reported in Mills (1986), Topsakal (1995), Flaherty (2001), Sera et al. (2002).

Discourse Analysis - Gender - Lexicography 1 1 3 Major Trends in Theoretical and Applied Linguistics

same/different grammatical gender of the nouns denoting them in Greek and German (i.e. 18 items with the same grammatical gender in both languages, 22 items with different gender); b) following Flaherty (2001), care was taken to avoid objects stereotypically connected with women or men. Finally, we did not make use of labels, i.e. words denoting the depicted items, in order to avoid employment of grammatical gender as a strategy on the subjects’ part (cf. section 2). Participants were asked to ‘name’ the objects by choosing a proper name from the list. Task two replicated the memory experiment as conducted by Boroditsky et al. (2003), i.e. it involved memorizing pairs of words/objects and proper names (distinct for men/women). Participants were then tested, using the E-Prime software, for how well they could remember a given name for a word/object. The stimuli consisted of 28 words denoting animals and inanimate objects (whereby attempt was made to deploy as many of the items used in the previous task as possible) each appearing on the computer screen with a different proper name for five seconds, and was automatically followed by the next pair in random order. Each pair was presented only once. Male and female proper names were chosen to be similar (Vasilis/Vasiliki, Alekos/Aleka, etc.) in order to increase the difficulty of the task. After the ‘learning’ stage, participants completed a1-2 minute distraction task unrelated to the study. In the final stage, each word/ object from the learning set was presented on screen, this time followed by two similar proper names (e.g. Alekos/Aleka); participants were asked to indicate the proper name (feminine or masculine) that had been associated with that word by pressing one of the two keys on the keyboard. For task one, our hypothesis was (given Topsakal’s work for Modern Greek) that the subjects would assign male or female sex to objects/animals/persons according to the grammatical gender of the noun denoting those items. For task two, we hypothesized (given the work of Boroditsky et al.) that the subjects’ memory would be better for word/object–name pairs when there is a fit between grammatical gender and sex than for those pairs in which there is no such fit. Finally, for both tasks the hypothesis was that the language of the instructions would impact on the results.

4. Results

In the following we present only some of the findings of our study due to limitations of space. Examination of the relationship between the grammatical gender of the noun denoting a depicted item and the sex attributed to it (via choice of proper name) yielded the following two tables for each of the two experimental conditions (A: Instructions + Stimuli in Greek, B: Instructions in English, Stimuli in Greek).

1 1 4 Discourse Analysis - Gender - Lexicography Theodossia-Soula Pavlidou and Angeliki Alvanoudi

Table 1A. Grammatical Gender by Attributed Sex (Instructions + Stimuli in Greek)

attributed sex male female TOTAL adj. residual gram. gender masculine 435 70.3% 184 29.7% 619 100% 10.8

feminine 256 38.2% 415 61.8% 671 100% 9.2

neuter 313 50.3% 309 49.7% 622 100% 1.3

TOTAL 1004 100% 908 100% 1912 100% p ≤ .000

Table 1B. Grammatical Gender by Attributed Sex (Instructions in English, Stimuli in Greek)

attributed sex male female TOTAL adj. residual gram. gender masculine 213 75.8% 68 24.2% 281 100% 9.9

feminine 96 31.8% 206 68.2% 302 100% 8.5

neuter 137 48.2% 147 51.8% 284 100% 1.3

TOTAL 446 100% 421 100% 867 100% p ≤ .000

Tables 1A and 1B show that there is a high correlation between grammatical gender and the sex attributed to persons/animals/objects. Under condition A, 70.3% of the masculine nouns correlate with male sex and 61.8% of the feminine nouns correlate with female sex. Moreover, under condition B, 75.8% of the masculine nouns correlate with male sex and 68.2% of the feminine nouns correlate with female sex. In other words, the correlation is more pronounced when instructions are given in English. The neuter gender is almost evenly distributed between male and female under both conditions. The above results were broken down by semantic category (persons, animals, inanimate objects), for each of the two experimental conditions.

Table 2A. Grammatical Gender by Attributed Sex – Persons (Instructions + Stimuli in Greek)

attributed sex male female TOTAL adj. residual gram. gender masculine 129 91.5% 12 8.5% 141 100% 14.6

feminine 6 6.3% 90 93.8% 96 100% 10.0

neuter 1 2.1% 47 97.9% 48 100% 6.9

TOTAL 136 100% 149 100% 285 100% p ≤ .000

Discourse Analysis - Gender - Lexicography 1 1 5 Major Trends in Theoretical and Applied Linguistics

Table 2B. Grammatical Gender by Attributed Sex – Persons (Instructions in English, Stimuli in Greek)

attributed sex male female TOTAL adj. residual gram. gender masculine 62 95.4% 3 4.6% 65 100% 10.6

feminine 0 0.0% 44 100% 44 100% 8.0

neuter 2 9.1% 20 90.9% 22 100% 4.1

TOTAL 64 100% 67 100% 131 100% p ≤ .000 Tables 2A and 2B show that for persons there is an almost perfect fit between masculine nouns and male sex (91.5% under condition A and 95.4% under condition B) and feminine nouns and female sex (93.8% under condition A and 100.0% under condition B).8 Tables 3A and 3B below show that for animals the correlation between gender (masculine, feminine) and sex is at about the same level as the overall results (cf. Tables 1A and 1B): 71.3% (condition A) and 73.8% (condition B) of the masculine nouns were given male sex, while 67.8% (condition A) and 73.8% (condition B) of the feminine nouns were given female sex.

Table 3A. Grammatical Gender by Attributed Sex – Animals (Instructions + Stimuli in Greek)

attributed sex male female TOTAL adj. residual gram. gender masculine 102 71.3% 41 28.7% 143 100% 4.5

feminine 46 32.2% 97 67.8% 143 100% 7.0

neuter 92 64.3% 51 35.7% 143 100% 2.5

TOTAL 240 100% 189 100% 429 100%

p ≤ .000

Table 3B. Grammatical Gender by Attributed Sex – Animals (Instructions in English, Stimuli in Greek)

attributed sex male female TOTAL adj. residual gram. gender masculine 48 73.8% 17 26.2% 65 100% 4.1

feminine 17 26.2% 48 73.8% 65 100% 5.3

neuter 39 59.1% 27 40.9% 66 100% 1.2

TOTAL 104 100% 92 100% 196 100%

p ≤ .000

8 This is a rather expected result if one takes into account the findings reported in Pavlidou et al. (2004) with respect to the grammatical gender-sex fit in the Greek vocabulary.

1 1 6 Discourse Analysis - Gender - Lexicography Theodossia-Soula Pavlidou and Angeliki Alvanoudi

Finally, tables 4A and 4B below show that for inanimate objects there is a correlation between gender (masculine, feminine) and sex, but this correlation is less pronounced than the one found in the case of animals (and, of course, persons): 60.9% (condition A) and 68.2% (condition B) of the masculine nouns were assigned male sex respectively, while 52.8% (condition A) and 59.1% (condition B) of the feminine nouns were given female sex respectively. Once again, these correlations are higher under condition B.

Table 4A. Grammatical Gender by Attributed Sex – Inanimate Objects (Instructions + Stimuli in Greek)

attributed sex male female TOTAL adj. residual gram. gender masculine 204 60.9% 131 39.1% 335 100% 3.7

feminine 204 47.2% 228 52.8% 432 100% 2.7

neuter 220 51.0% 211 49.0% 431 100% .7

TOTAL 628 100% 570 100% 1198 100%

p < .001 Table 4B. Grammatical Gender by Attributed Sex – Inanimate Objects (Instructions in English, Stimuli in Greek)

attributed sex male female TOTAL adj. residual gram. gender masculine 103 68.2% 48 31.8% 151 100% 4.8

feminine 79 40.9% 114 59.1% 193 100% 3.7

neuter 96 49.0% 100 51.0% 196 100% .9

TOTAL 278 100% 262 100% 540 100%

p < .000

Examination of the relationship between attributed sex and type of object (natural vs. artifact) for each grammatical gender did not yield consistently significant results for the masculine and feminine genders.9 Only in the case of neuter, attribution of maleness-femaleness seems to be dependent on the type of inanimate object, i.e. it being an object of nature or an artifact, as tables 5A and 5B below indicate.

9 Results are statistically significant only under condition B for the masculine gender, in which case artificial objects get associated with maleness much more frequently than with females.

Discourse Analysis - Gender - Lexicography 1 1 7 Major Trends in Theoretical and Applied Linguistics

Table 5A. Attributed Sex by Type of Inanimate Object – Neuter (Instructions + Stimuli in Greek)

attributed sex male female TOTAL adj. residual gram. gender natural 57 39.9% 86 60.1% 143 100% 3.3

artificial 163 56.6% 125 43.4% 288 100% 3.3

TOTAL 220 100% 211 100% 431 100%

p < .001

Table 5B. Attributed Sex by Type of Inanimate Object – Neuter (Instructions in English, Stimuli in Greek)

attributed sex male female TOTAL adj. residual gram. gender natural 20 30.3% 46 69.7% 66 100% 3.7

artificial 76 58.5% 54 41.5% 130 100% 3.7

TOTAL 96 100% 100 100% 196 100%

p < .000

As seen from the two tables above, natural objects are associated more frequently with the female rather than with the male sex; artifacts, on the other hand, are associated slightly more frequently with the male rather than the female sex. This is the case under both experimental conditions, whereby the association of ‘naturalness’ with the female sex and of ‘artificiality’ with the male sex is more pronounced when instructions are given in English. In other words, there seems indeed to be a kind of cultural bias with respect to the conception of inanimate objects as male, if ‘artificial’, or female, if ‘natural’, which is overridden, however, by grammatical gender and its association with ‘maleness’ if masculine, and ‘femaleness’ if feminine. The memory task yielded no statistically significant results under either condition A or condition B. In other words, no connection between the grammatical gender- attributed sex fit, on the one hand, and memory, on the other, could be detected –a fact that runs counter to the results reported in Boroditsky et al. (2003).10 In sum, then, the first hypothesis was confirmed, the second hypothesis was not confirmed, while the third hypothesis was confirmed for task one but not for task two.

10 It is important to note here that we originally intended to use only pictures of objects (without labels) in the memory task in order to avoid the implicit equation of ‘word’ and ‘object’ (cf. section 2) in the Boroditsky et al. (2003) design. However, a first pre-test we ran with 14 pictures yielded no mistakes at all in the subjects’ recalling of names for depicted objects. Assuming that the number of items used was too small, and hence the object-name pairs too easy to memorize, we increased the number of items to 26, and ran a second pre-test, but the pattern of mistakes detected this time was not systematic with respect to grammatical gender.

1 1 8 Discourse Analysis - Gender - Lexicography Theodossia-Soula Pavlidou and Angeliki Alvanoudi

5. Discussion

As we have seen in the previous section, the sex attribution task yielded that grammatical gender correlates with the sex attributed to depicted items even in the case of a three-gender language. It also yielded that this correlation is not restricted to the semantic category of animals, but also holds for inanimate objects. Moreover, it showed that such correlations are stronger when the language of instructions is English. We could thus claim that our work provides evidence for grammatical gender effects on cognition in Modern Greek. On the other hand, the memory task showed no effects of the congruence between the gender of nouns for animals/objects and the gender of names given to these animals/objects on the subjects’ ability to remember these names, regardless of the language of instructions. Therefore, on these grounds, we could conclude that no evidence for correlations between grammatical gender and cognition can be provided. This is a rather peculiar situation that needs to be accounted for. A first question to ask is: how do these results fit in with the landscape of research on grammatical gender and cognition? The answer is not a simple one, given that there is no unilateral alignment between our findings and those of other scholars. Due to space limitations, we can only hint here at some convergences and divergences between what we have found and what others report, without providing a detailed account. For one, as it has already been mentioned, our memory task results contradict what is reported in Boroditsky et al. (2003); on the other hand, our work aligns with theirs in so far as the language of instructions was found to impact on the sex attribution task. Moreover, our results are compatible with all those reporting gender effects in languages with a three-gender system (e.g., Mills 1986; Topsakal 1995; Boroditsky et al. 2003), though not necessarily with regard to all semantic categories, but diverge from those in Sera et al. (2002) and Vigliocco et al. (2005) who found no such effects. Finally, with respect to semantic categories our findings for animals and inanimate objects diverge from Vigliocco et al. (2005) and Sera et al. (2002), respectively, but match those in Mills (1986) and Topsakal (1995). Taking into account the fact that some of these scholars criticize the work of one another11 (cf. section 2), one could invoke one or several methodological divides among the various approaches in order to account for this unclear pattern of results. However, the partial match of our findings across different experimental paradigms may well point to inherent methodological problems in

11 Curiously, though, among scholars working on gender and cognition there is a kind of selective reading of the available literature. For example, Sera et al. (1994, 2002), Boroditsky et al. (2003), Phillips & Boroditsky (2003) do not refer to Mills (1986) or Flaherty (2001).

Discourse Analysis - Gender - Lexicography 1 1 9 Major Trends in Theoretical and Applied Linguistics

each one of them. Additionally, it may also raise the issue of the comparability of the findings. Presumably, all studies understand grammatical gender in the same way; but can we say the same for e.g. ‘cognition’ or ‘speakers’ thinking for speaking’ and so on? In our opinion, it is such matters that have to be adequately resolved before an intervention of culture in the relationship between grammatical gender and cognition is appealed to.12 But it is also these issues that make the field of grammatical gender and cognition even more complex than language and cognition already is, and an area that one should fear to tread.13

12 Researchers working within the framework of linguistic relativity commonly argue that it is difficult to separate cultural effects from linguistic effects in experiments testing the relationship between language and cognition (cf. e.g. Lucy 1992). 13 It may be accidental, but it is certainly interesting that Gleitman & Papafragou (2005) in their excellent article on language and thought, do not include grammatical gender in their survey of the “semantic arenas of the present day language–thought investigation”.

1 2 0 Discourse Analysis - Gender - Lexicography Theodossia-Soula Pavlidou and Angeliki Alvanoudi

References

Boroditsky, L. and L.A. Schmidt. 2000. Sex, syntax, and semantics. Proceedings of the 22nd Annual Meeting of the Cognitive Science Society, 42-47.

Boroditsky, L., L.A. Schmidt and W. Phillips. 2003. Sex, syntax, and semantics. In D. Gentner and S. Goldin-Meadow (eds), Language in the Mind: Advances in the Study of Language and Thought. Cambridge, MA: The MIT Press, 61-79.

Bowerman, M. and S.C. Levinson (eds). 2001. Language Acquisition and Conceptual Development. Cambridge: Cambridge University Press.

Bowers, J.S., G. Vigliocco, H. Stadthagen-Gonzalez and D.P. Vinson. 1999. Distinguishing language from thought: experimental evidence that syntax is lexically rather than conceptually represented. Psychological Science 10(4): 310- 315.

Clarke, M.A., A. Losoff, M. Dickenson McCracken and J. Still. 1981. Gender perception in Arabic and English. Language Learning 31: 159-169.

Ervin, S.M. 1962. The connotations of gender. WORD 18(3): 249-261.

Flaherty, M. 2001. How a language gender system creeps into perception. Journal of Cross-Cultural Psychology 32(1): 18-31.

Gentner, D. and S. Goldin-Meadow (eds). 2003. Language in the Mind: Advances in the Study of Language and Thought. Cambridge, MA: The MIT Press.

Gleitman, L. and A. Papafragou. 2005. Language and thought. In K.J. Holyoak and R.G. Morrison (eds), The Cambridge Handbook of Thinking and Reasoning. Cambridge: Cambridge University Press, 633-661.

Discourse Analysis - Gender - Lexicography 1 2 1 Major Trends in Theoretical and Applied Linguistics

Gumperz, J. and S.C. Levinson (eds). 1996. Rethinking Linguistic Relativity. New York: Cambridge University Press.

Konishi, T. 1993. The semantics of grammatical gender: a cross-cultural study. Journal of Psycholinguistic Research 22(5): 519-534.

Levinson, S.C. 2003. Space in Language and Cognition: Explorations in Cognitive Diversity. Cambridge: Cambridge University Press.

Lucy, J. 1992. Grammatical Categories and Cognition: A Case Study of the Linguistic Relativity Hypothesis. Cambridge: Cambridge University Press.

Lucy, J. 1996. The scope of linguistic relativity: an analysis and review of empirical research. In J. Gumperz and S.C. Levinson (eds), Rethinking Linguistic Relativity. New York: Cambridge University Press, 37-69.

Mills, A.E. 1986. The Acquisition of Gender: A Study of English and German. Berlin: Springer-Verlag.

Niemeier, S. and R. Dirven (eds). 2000. Evidence for Linguistic Relativity. Amsterdam/Philadelphia: John Benjamins.

Pavlidou, T.-S., A. Alvanoudi and E. Karafoti. 2004. [in Greek] Grammatical gender and semantic content: preliminary remarks on the lexical representation of social gender. Studies in Greek Linguistics 24: 543-553.

Phillips, W. and L. Boroditsky. 2003. Can quirks of grammar affect the way you think? Grammatical gender and object concepts. Proceedings of the 25th Annual Meeting of the Cognitive Science Society, 928-933.

Pütz, M. and M.H. Verspoor (eds). 2000. Explorations in Linguistic Relativity. Amsterdam: John Benjamins.

Sera, M.D., C.A. Berge and J. del Castillio Pintado. 1994. Grammatical and conceptual forces in the attribution of gender by English and Spanish speakers. Cognitive Development 9: 261-292.

Sera, M.D., C. Elieff, J. Forbes, M. Clark Burch, W. Rodriguez and D. Poulin Dubois. 2002. When language affects cognition and when it does not: an analysis of grammatical gender and classification. Journal of Experimental Psychology, General 131: 377-397.

1 2 2 Discourse Analysis - Gender - Lexicography Theodossia-Soula Pavlidou and Angeliki Alvanoudi

Slobin, D.I. 1996. From ‘thought and language’ to ‘thinking for speaking’. In J. Gumperz and S.C. Levinson (eds), Rethinking Linguistic Relativity. New York: Cambridge University Press, 70-96.

Topsakal, M. 1995. [in Greek] The Influence of the Grammatical Gender on the Perception of Gender in Modern Greek: Comparison with German and English. Unpublished MA Thesis, Department of Linguistics, School of Philology, Aristotle University Thessaloniki.

Vigliocco, G., D.P. Vinson, F. Paganelli and K. Dworzynski. 2005. Grammatical gender effects on cognition: implications for language learning and language use. Journal of Experimental Psychology: General 134: 501-520.

Discourse Analysis - Gender - Lexicography 1 2 3 Major Trends in Theoretical and Applied Linguistics

1 2 4 Discourse Analysis - Gender - Lexicography The Song Remains the Same…or Not? A Pragmatic Approach to the Lyrics of Rock Music1

Angelina Sophiadi

Aristotle University of Thessaloniki [email protected]

Abstract

This paper explores how the lyrics of popular rock music associate with change as it is imprinted in language. More specifically, it aims to investigate the lyrics from a pragmatic perspective and to discern a possible fluctuation of certain pragmatic features (deixis, politeness strategies and, to some extent, speech acts) through time. Subsequently, it examines whether there is a connection between this fluctuation and ordinary language use. The research involves a preliminary questionnaire, and the use of a concordances program, which is employed for the processing of the lyrics of 225 songs.

1. Introduction

Popular music constitutes an integral part of the wider realm of popular culture and, since it operates within the social context, one can assume that the language employed in it, constitutes a “form of social practice” (Fairclough 2001: 18). Rock music, in particular, has been the dominant music genre2 in consumer popularity

1 This paper constitutes a condensed version of my MA thesis submitted in the Department of Theoretical and Applied Linguistics in the School of English of the Aristotle University of Thessaloniki. The thesis was supervised by Dr. Eliza Koutoupi-Kitis, to whom I am thankful and grateful for her constant support and guidance. Of course, I am responsible for all mistakes and obscurities. 2 See http://76.74.24.142/8EF388DA-8FD3-7A4E-C208-CDF1ADE8B179.pdf for the RIIA 2008 ten year demographic survey on consumer profiles, http://www.riaa.com/newsitem. php?id=370FD0E5-4FCF-0724-8AEC-277A1D52A260 for the RIIA 1996 ten year trend data demographic survey on consumer preferences and http://www.bpi.co.uk/music-business/article/ the-market.aspx for information on the British market from the British Recorded Music Industry.

Discourse Analysis - Gender - Lexicography 1 2 5 Major Trends in Theoretical and Applied Linguistics

for many decades and it “has been a crucial element in several youth culture movements” (Rutten 1993: 38) from the 1960s to the present. Having a life-span of more than half a century, rock music has operated as a versatile force and has evolved into a resilient genre of popular music, incorporating all the good, and being the scapegoat3 for all the bad aspects of it.

2. Anybody Listening? Lyrics as the Subject of Analysis

In spite of their standardized nature4, there is a diversity of foci one can take on when studying lyrics. These range from content analysis and the connection between shifts in lyrics and changes in the socio-economic conditions (Mooney 1954; Pettijohn & Sacco 2009), politics (Grossberg 1993; Frith 1993), how lyrics lead to the deviant behavior of youth (Knobloch-Westerwick et al. 2008; Robinson et al. 1996) to even language teaching (Kimbel & Protivnak, 2010; Paquette & Rieg, 2008). Moreover, a considerable amount of research has been conducted within the domain of sociolinguistics regarding code switching between the African American Vernacular English and the respective standard language in the lyrics of rap music around the world5. Kuhn in 1999 attempts a pragmatic6 approach to the analysis of persuasive strategies in blues lyrics, employing speech act theory and politeness theory. She observes that there is common ground in songwriters in their attempt to mitigate the face-threatening act of a request, by means of redressive action in the form of positive–and negative politeness (Kuhn 1999: 528-533).

3 See http://www.slate.com/id/1008314/, http://www.snopes.com/rumors/radio.asp and http://en.wikipedia.org/wiki/List_of_songs_deemed_inappropriate_by_Clear_Channel_ following_the_September_11,_2001_attacks and for further information on the controversial 2001 Clear Channel memorandum regarding lyrically questionable –mainly rock- songs which radio stations had to avoid playing on air, following the September 11, 2001 attacks. 4 The notion of standardization, which “aims at standard reactions” (Adorno [1941] 2002: 442) by the listeners, and that of pseudo-individualism, which offers the illusion of free choice (ibid, 445), are central to Adorno’s overview on popular music, concluding that this industry “provides trademarks of identification for differentiating between the actually undifferentiated” (ibid, 446). 5 Just to name a few, Androutsopoulos & Scholz (2002), who deal with European rap music, and Pennycook (2003), who discusses code switching in the rap music of Malaysia, Japan, Korea and Pacific Rim countries, New Zealand, Fiji and Samoa. 6 Kuhn’s research is quite well structured, although a bit small-scale (she only examines five songs). Her references however, reveal the scarcity of literature combining linguistics and popular music.

1 2 6 Discourse Analysis - Gender - Lexicography Angelina Sophiadi

3. Methodology and Research – The Questionnaire

The research was initiated with a preliminary questionnaire7 to investigate the hypothesis that lyrics are noticed by listeners. The questionnaire was created with the use of an online survey software called Survey Monkey. Nine closed questions and one open-ended question8 were constructed. The goal was to make it quick and easy to fill in, without getting into complex details. The processed questionnaires were those filled in online during a two-month period (April – June 2010). The total number of participants was 185 and they represented different age groups and nationalities. The results regarding the music genre preference place rock music in the first position [71,9%], which becomes even higher when taking into account the fact that 14,5% of the participants who actually selected the other option, mentioned subgenres of rock music like hard or progressive rock and heavy metal. The overwhelming majority of the participants [88%] listens to music very often or always. In the question concerning whether they pay attention to the lyrics of songs, a safe percentage of 78% validates the presupposition. Almost 60% of the participants answered that the lyrics of a song can affect their preference towards it, which could be an indication that lyrics play a role in the popularity of a given song. However, this percentage displays a decrease [37%] when the subjects were asked whether they relate to the lyrics of the songs they listen to; popularity may not be a matter of self-expression, but rather a matter of taste or preference.

4. Methodology and Research – Interrogating the Corpus

After securing support for the assumption that listeners do pay attention to the lyrics, the next stage was to compile the corpus using a concordances program. AntConc 3.2.1 was selected, especially because of its simple and “well-structured […] user interface” (Wiehman & Fuhs 2006: 118). The aim was to collect the same number of popular rock songs for each decade from the 1960s up to the 2000s. Whatever information was available was employed

7 See http://www.surveymonkey.com/s/5K78WDY for the online version of the questionnaire. 8 In this question, participants were asked to quote an example of lyrics that they relate to. It was optional and included for future research.

Discourse Analysis - Gender - Lexicography 1 2 7 Major Trends in Theoretical and Applied Linguistics

and crosschecked with other sources regarding sales and overall reception of music production, as well as official and unofficial chart collections9 for most European countries. Other parts of the music industry such as the Grammy awards, music television networks and magazines had to be taken into account as well. Bearing all this in mind,10 five lists were created, one for each decade consisting of 25 rock songs each. The same procedure was followed for pop music, collecting, however, a slightly smaller sample of 20 songs for each decade. This discrimination was allowed for two reasons; first, the differentiation between rock and pop was a secondary hypothesis and, given the time, it was considered reasonable to have a smaller sample. The second reason had to do with the general genre of pop music and the range of subgenres it includes, or has come to include, over the years. Questions like whether dance music or rap music actually comprise pop music inevitably came up. Half a century ago, there was no such thing as rap music, which can be thought of as a separate genre, but in some cases it may be considered to be pop, as official charts show.

5. Interrogating the Corpus - Results

There were some revealing results in the samples regarding a specific pair of personal pronouns, namely I and you; the emergence of this pair, along with the constant presence of the third person, seems not to be arbitrary, considering Singer’s analysis of Peirce’s first, second and third person as “the underlying structures of communication and signification which form and express the semiotic self” (1989: 242). This tripartition is conspicuous in the lyrics throughout the decades with the three parts –first, second and third person- always dominating the first three positions in the frequency order. Their omnipresence is important since it indicates that, quantitatively speaking, the main focus on lyrics has to do with the communication and the signification of the self. Even without fluctuations in the frequency of occurrences, their mere dominance would still verify that lyrics follow the same structure of human interaction and communication since, their use constitutes “a condition for the formation of selves and person” (Singer 1989: 284).

9 www.billboard.com and www.billboard.biz for the official charts for all over the world. www.billboard.com/specials/hot100/index.shtml for various official top 100 lists during the years 1958-2008. http://www.theofficialcharts.com/music-charts/ the official UK charts website and www.vinylsurrender.com , an unofficial site containing many charts based both on visitors’ vote and on official data and charts 10 One last thing one had to consider when compiling the corpus of song lyrics was to try to find artists or groups who were of diverse nationalities. For a list of the songs employed in all corpora, please contact me via e-mail.

1 2 8 Discourse Analysis - Gender - Lexicography Angelina Sophiadi

8,00% 7,00% 6,00% 5,00% 4,00% 3,00% 2,00% 1,00% 0,00% 1960s 1970s 1980s 1990s 2000s I 4,49% 3,53% 6,16% 6,08% 7,01% you 5,02% 3,31% 4,25% 4,10% 3,13% we 0,54% 0,93% 0,83% 0,88% 0,51%

Figure 111. Percentages of the appearances of ‘I’, ‘you’, and ‘we’ in Rock music.

But apart from this, a fluctuation was observed between the occurrences of the first and the second person singular pronouns. The observed pairing of these two is also not random, since they are entirely indexical in terms of their reference, which makes them more easily relatable by the audience. When there is no referential restriction, one can interpret these pronouns in any way one wishes and this can be considered to be a popularity factor as far as song lyrics are concerned. Observing the occurrences of the two pronouns diachronically, the shift from you to I is prominent: the first person pronoun doubles the percentage of its appearance, whereas the second person singular pronoun ends up in the 2000s having almost half the appearances it had in the 1960s. It is also of importance, however, to notice the difference in the order of their occurrences. During the 1960s the most frequently occurring word in the lyrics

11 The first person plural pronoun is also mentioned in the tables above butnot commented on. This is because, before even embarking on this research, one had the firm assumption that there would be expressions of solidarity or absence of it fluctuating through time in rock music, mainly by means of employing the inclusive and exclusive we. The results however, showed very little fluctuation. Also,we holds a very low position in the order of frequency in both corpora, leading one to assume that expressing solidarity –as its main function- is not of high significance for either genre. Considering however specific subgenres of rock or any other music for that matter, the results would have probably been different and one might anticipate a more prominent expression of inclusiveness and exclusiveness in the lyrics.

Discourse Analysis - Gender - Lexicography 1 2 9 Major Trends in Theoretical and Applied Linguistics

was the second person pronoun you and the second –by a slight difference- was the first person I. This tuistical12 dominance quickly changes and in the 1970s the order is reversed, displaying an egotistical trend. Gradually, the gap between the frequency of occurrences of the two pronouns becomes more prominent, as in the 1980s the first person has 384 occurrences, whereas the second person has 265 occurrences, and in the 1990s 418 and 282 respectively. In the sample of lyrics collected for the 2000s the second person pronoun drops even more in terms of frequency with 201 occurrences, whereas the first person pronoun remains dominant with 449 occurrences. Since the selection of the songs was based on criteria that had to do with their appeal, we cannot assume that this variation is random. If we consider music to be a type of media discourse relating to the discourse of those to whom it appeals, it is expected and well justified that these two pronouns were the most prominent ones in terms of frequency, exactly because their reference is indeterminate and, therefore, adaptable to the hearer’s/interpreter’s subjectivity or circumstances. This is a clear case of what one might call ‘reference adaptability’ of indexicality, a prevalent phenomenon in the genre of poetry. The tendency, however, towards the first person combined with the decline in the occurrences of the second person, opens the ground for additional inferences. Lyons states that the deictic context of an utterance “is centered upon the speaker’s here and now: it is, in this respect, egocentric” (1981: 305). In the case of the lyrics however, the egocentricity of the deictic context is even more prominent due to the gradual dominance of the first person singular pronoun. This is enhanced when taking into account that you has a generic use as well, replacing the indefinite one in more informal contexts (Quirk et al. 1972: 222). Also, the first person singular pronoun I can only be a subject whereas the second person you can operate either as a subject or an object. Fowler points out that it is characteristic of media discourse, to position as subjects people of authority, whereas “those with less power occur as objects” (1991: 98). Based on this suggestion, we could infer that the dominance of the first person singular may reflect a gradual ascension of the speaker/singer to a dominant position in relation to his ‘addressee’. In other words, we witness a gradual shift of emphasis on the performer. As the you of the lyrics denotes the addressee of the singer’s narrative or a secondary character of the discourse enacted, that is, the narrative of the song, but not the actual physical recipient of the song, the latter is expected to identify with the performer as the main agent of the narrative rather than with secondary characters of the narrative (you).

12 This term, along with egotistical and idistical as used to refer to the second, first and third person respectively, that will be used further on, is taken from Singer (1989: 239).

1 3 0 Discourse Analysis - Gender - Lexicography Angelina Sophiadi

Therefore, the focus is undoubtedly on the self, that is, the I. And this discourse of the self is quite reminiscent of recent work of psychologists, claiming that the current generation is more egocentric than the previous ones, “a generation unapologetically focused on the individual” (Twenge 2007: 2). This can be explained on the basis of the globalization experienced by every society, along with the commodification of almost everything: “In a society awash with flashy commodities […] our language for representing […] our image of self-identity is more and more fixed into a syntax of possession [and] ownership” (Elliot and Lemert, 2006: 41). This, of course, does not necessarily imply that people are becoming more selfish, for this would be a naïve assumption to make. The language employed in this genre in general, being the most popular one, could be indicative of a relative trend in the ordinary use of language; but this is something that has to be verified by means of historical linguistics, to see whether there is indeed a shift in the frequency of use of personal pronouns. In pop music on the other hand, the results were slightly different. It looks like pop music, by and large, follows the pronominal patterns of rock music, only in more subtle ways. This may be attributed to genre differences between rock and pop music, or to the fact that –without any intention of standardizing pop music- the latter is more or less, thematically restricted to love and sexuality, and probably this restriction is translated into language that is not fully representative of its time of production.

7,00%

6,00%

5,00%

4,00%

3,00%

2,00%

1,00%

0,00% 1960s 1970s 1980s 1990s 2000s I 4,74% 4,64% 4,40% 4,69% 5,17% you 5,02% 4,22% 4,14% 6,54% 4,55% we 0,56% 1,51% 1,61% 0,44% 0,59%

Figure 2. Percentages in the appearances of ‘I’, ‘you’, and ‘we’ in Pop music.

Discourse Analysis - Gender - Lexicography 1 3 1 Major Trends in Theoretical and Applied Linguistics

6. Speech Acts – The Problem

Recognizing speech acts within the scope of lyrics was perhaps the most difficult task to tackle. Before attempting to see song lyrics in terms of speech acts, the first thing that should be done is investigate their context. In song lyrics, the elements comprising the context are blurred. After all, we are talking about a mass communication text which “is likely to be consumed in various sorts of places and at various times” (Fairclough, 1995: 36). It is also characterized by “one-sidedness” as “there is a sharp divide between the producers and the interpreters” (Fairclough: 2001: 41), since it is designed to be received by mass audiences. Moreover, when talking about speech acts, one should initially consider the tripartite distinction of sentence types –the declarative, the interrogative and the imperative. This may not cover Searle’s distinction of speech acts, but still “provide[s] us with an empirical base for constructing the common-sense or cognitive model of speech acts” (Croft 1994: 471). Lyons also seems to agree by arguing that these sentence-types constitute the three basic speech acts (1981: 251). Of course, this should not light-heartedly be extended to the respective speech acts. These conventional markers may be successful at pointing towards a specific illocutionary force, the actual speech act, however, may not necessarily correspond to a grammatical indication or form. At this point it should be mentioned that, even if one considers literary texts to be a development of ordinary language (Fabb 2010: 1230), one cannot ignore the fact that the language employed by the former is not actually doing things to the maximum of its illocutionary force, but rather it pretends to be doing things (acts as if). The speaker of fictional discourse “pretends to perform a series of illocutionary acts, normally of the representative type […] without undertaking the commitments that are normally required” (Searle 1975: 325, 326). Fish also seems to agree that there is “a kind of discourse that is characterized by the suspension of the rules to which speech acts are normally held accountable” (1976: 1017). This can be the case for song lyrics as well, and this actually is what gives the audience the illusion of serious13 discourse, as Searle would put it. Under these circumstances, it might seem meaningless to undertake the task of interpreting speech acts or investigating the use of politeness strategies in song lyrics, since the force of utterances is somehow lost in ‘as-if’ illocutionary performances, as the utterer is safe in terms of politeness in the sanctuary of fictional or literary discourse.

13 Searle makes a distinction between serious and nonserious literature not in a derogatory sense, but rather in the sense that if an author of a novel or a poem tells us “that it is raining outside, he isn’t seriously committed to the view that it is at the time of writing actually raining outside” (1975: 321)

1 3 2 Discourse Analysis - Gender - Lexicography Angelina Sophiadi

7. The Alternative – Sentence Types and Politeness Strategies

Considering all the above, an alternative approach to the research presented itself; it is evident and undeniable that most of the sentences in the lyrics collected are declaratives; However, one cannot discern beyond the declaratives in the lyrics to interpret them as indirect speech acts, and in this sense, the two remaining sentence-types, namely imperative and interrogative sentences, as well as their frequency of appearance, took on the primary role in this research. Compiling the interrogatives and imperatives for each corpus14 proved to be a tedious task, since the concordances software did not offer a relevant feature; therefore, it had to be done manually. The benefit to this however, was the option to exclude repetitions of the same sentence, given the recurrent nature of some lyrics.

80

70

60

50

40

30

20

10

0 1960's 1970's 1980's 1990's 2000's questions 19 29 28 37 16 39 32 52 73 66 imperatives Figure 3. Distribution of interrogative and imperative sentences in Rock music.

The results were indicative of an increasing tendency in using more imperatives in relation to an opposite trend in interrogatives. More specifically, in the rock music corpora, during the 1960s the questions employed are half the number

14 The pop music corpora were not included in this part of the research, due to the sample’s diversity consisting of new musical subgenres such as rap and hip hop, as mentioned earlier.

Discourse Analysis - Gender - Lexicography 1 3 3 Major Trends in Theoretical and Applied Linguistics

of imperatives -19 and 39 respectively, whereas in 1970s this difference becomes quite small -29 over 32. In the 1980s however, the gap between the two becomes visible once again with 28 questions over 52 imperatives, and it gradually increases in the 1990s -37 over 73 respectively – when, upon investigating the 2000’s the imperatives are more than four times the number of questions -66 over 16. The use of imperatives and questions in song lyrics offers a “sense of personal interaction” (Fowler, 1991: 65) in itself. The evident, however, preference towards employing imperatives may have implications regarding politeness strategies. Using the imperative generally gives the impression of imposing on the addressee, which may threaten their negative face15. Brown and Levinson assert that “direct imperatives stand out as clear examples of bald-on-record usage” (1987: 95). This usage is employed for maximum efficiency in a communicative urgency, which can be also metaphorical when one is begging; this seems to agree also with Grice’s Maxim of Manner regarding brevity of the contribution. (1989: 27). Imperatives can also be used in “pre-emptive invitations” including welcoming, farewells, and offers. (Brown and Levinson, 1987: 95-101) These two explanations, however, entail the assumptions that the addressee will translate this use as a politeness strategy, and that the speaker does not intend to threaten the face of the addressee. But consider the following examples taken from the corpora:

1. Save me from the nothing I’ve become. Bring me to life16 2. Give me your heart, make it real or else forget about it17 3. Welcome to this place, I’ll show you everything18 4. Go drill your deserts - Go dig your graves. Then fill your mouth with all the money you will save19 5. Here we are now, entertain us20 6. Don’t try to push your luck, just get out of my way21

15 Of course, imperatives employed in the addressee’s interest should be excluded, since they operate as an “attempt to overcome H’s reluctance to impose on S, by pretending that S has the power to force H to act” (Brown and Levinson, 1987: 229), and therefore contain no threat against the face of the addressee. 16 Bring me to life by Evanescence 17 Smooth performed by Santana and Rob Thomas 18 With arms wide open by The Creed 19 Psychosocial performed by Slipknot 20 Smells like teen spirit by Nirvana 21 Back in black performed by AC/DC

1 3 4 Discourse Analysis - Gender - Lexicography Angelina Sophiadi

Even without the help of contextual information, it can be inferred that examples (1) and (2) fall into the communicative urgency category, be it literal or metaphorical. The potentially face threatening act expressed with the imperative is performed without the use of the particle please and without any identity markers or certain address forms that may mitigate the threat (Brown and Levinson, 1987: 107-110). This intensifies the urgency expressed, but still, the lack of further redressive action weighs the scale towards a weightier FTA. The third example is obviously a welcoming, covering the second polite option Brown and Levinson present. The three remaining examples are a bit problematic, since they cannot be explained in terms of politeness strategies. Consider the calculus Brown and Levinson (1987) have suggested:

Wx = D (S, H) + P (H, S) + Rx

where Wx is the value that measures the weightiness of the given act. It includes three variables, the social distance (D) between the speaker and the hearer, the relative power (P) between the two and the absolute ranking (R) of the imposition in the given situation. Brown and Levinson acknowledge that the variables at work here may be connected in more complex ways, they state however, that “all three dimensions P, D and R contribute to the seriousness of an FTA” (1987: 76). By considering that two of the variables, D and R are constant, then the weight of the FTA would depend on the third one. Given that examples (4) to (6) constitute quite addressee-adversarial acts, it can be assumed that the speaker deems the relative power (P) between him/her and the addressee to be quite high. In other words, the speaker does the FTA because of the power he/she may feel to have over the hearer. This, however, is only a subjective estimation on behalf of the hearer and may not actually correspond to reality; Still, the speaker goes on and performs the FTA apparently on the assumed power –be it physical, psychological or emotional- he/she has over the addressee. Although this claim lacks sturdy grounding in a thoroughly systematic observation and analysis of the imperatives employed in lyrics that would unambiguously delineate their true illocutions, there is still clear evidence for positing the impolite use of the imperative; this pragmatic feature, too, seems to squarely fall in place with the previous results of pronominal occurrence concerning the ego-centricity projected in rock music. This, of course, does not constitute a generalised assumption that all the imperatives employed in the corpora are of this kind, but since their frequency increases dramatically over time, it is sensible to assume that this use is also extended. Finally, both pop and rock corpora were processed in order to see the frequency of use of potential address terms that may convey in-group membership and

Discourse Analysis - Gender - Lexicography 1 3 5 Major Trends in Theoretical and Applied Linguistics

therefore be employed as redressive action22 in politeness strategies. The results reveal once again that there is a decreasing trend in the use of such forms, which is clearly prominent in the 2000’s, where there are only 4 instances of address forms in a corpus of 6404 word tokens -3 instances of baby and 1 of darling-. Overall, the only address form used consistently in both music genres and throughout the decades, is baby and babe. Apart from that, there are other terms employed for addressing the hearer in the rock music corpora, their occurrences, however, are represented by one-digit numbers scattered in each corpus.

35

30

25

20

15

10

5

0 1960s 1970s 1980s 1990s 2000s baby/babe 32 29 19 19 3 darling/darlin 2 4 2 1 girl 9 9 1 1 laddy 1 lady 1 man 4 1 woman 3 6 honey 2 1 1 brother 2

Figure 4. Distribution of address forms – and ‘please’ in the Rock music corpora.

During the 1960s 48 such terms are estimated, 32 of which are of the word baby, whereas the 1970s follow the same pattern with 50 occurrences. In the 1980s and the 1990s the decrease in the use of terms is obvious, since there are only 26 and 25 –respectively- overall occurrences, 19 of which are of the term baby, whereas in the 2000s there are only 4 occurrences, 3 of the term baby and 1 of the term darling.

22 The data were also studied manually in order to verify that it was indeed a case of addressing a speaker.

1 3 6 Discourse Analysis - Gender - Lexicography Angelina Sophiadi

80

70

60

50

40

30

20

10

0 1960s 1970s 1980s 1990s 2000s

1960s 1970s 1980s 1990s 2000s baby/babe 71 10 25 48 66 darling 19 1 girl 9 9 2 7 love 1 12 man 4 3 woman 19 honey 14 12 sugar 10 please 4 2 7 8 7

Figure 5. Distribution of address forms – and ‘please’- in the Pop music corpora.

The pattern is different in the pop music corpora. The term baby is again systematically present in all decades; however, there doesn’t seem to be a decrease in the term’s occurrences. A sharp fall is observed during the 1970s (10 appearances) compared to 1960s (71 appearances), but since then the occurrences have been gradually increasing. This may be attributed once again to the nature of pop songs, focusing mainly on love and affection, which is expressed even in ordinary language with such endearment terms. This gradual decline in the occurrences of such terms in rock music and their rarity in general, indicates that either employing such terms is not a common strategy in song lyrics or it is considered that there is neither need nor space for such terms. This second assumption seems to agree with what was suggested earlier considering the findings so far; lyrics in popular music are more assertive, the speaker is –or considers himself/herself to be- more powerful, and subsequently language becomes rather bald, without really worrying about civility, courtesy or politeness in the lyrics discourse.

Discourse Analysis - Gender - Lexicography 1 3 7 Major Trends in Theoretical and Applied Linguistics

The mitigating particle please also seems to be absent in the corpora of both genres, which shows that the conventional politeness norms are not actually followed in song lyrics. This however, seems to be a general phenomenon as Lakoff observes:

One way in which conventional politeness is communicating is by taking a bit of extra time, saying that additional syllable or two that conveys that […] attending to the human needs of the interlocutor comes first. But in America, we say, “Time is money.” The removal of “please” merely instantiates that belief in a newly explicit way. (Lakoff, 2005: 33-34)

In this sense, it can be ascertained that the pragmatic demeanour of song lyrics may in fact represent a general tendency towards being more terse and direct, and subsequently avoid the use of politeness strategies that might take up space and time while taking their toll on substance.

8. Conclusion – Does the Song Remain the Same?

The results regarding the rock music corpora suggest that there is indeed a pragmatic shift, implying that assertiveness is more prominent in song lyrics and that the projected power of the speaker over the assumed addressee is increasing leaving less room for politeness strategies. Concerning the relation between rock and pop music, the findings seem to be less systematic. Pop music does not always follow the rock tendency discussed, mainly because of its diachronically constant content. This difference may be explained by means of an oxymoron. Pop music probably does not represent the pop culture of a time, at least to the same extent as rock music does. In this sense, rock is more pop than pop is. To answer the original question posed as the title to this essay, no, the song does not remain the same. Not only does it develop in terms of musical styles and techniques, but also in terms of lyrics. There has to be an interchange between the lyrics and the people who listen to the lyrics and this interchange may reflect or even affect both sides in terms of language use. However, there remains much work to be done regarding the joint areas of popular music –lyrics- and language use. The need for further research should be stressed; although there is a vast amount of literature on popular music, the pragmatic aspect of this genre has been somehow overlooked mainly due to the impediments discussed.

1 3 8 Discourse Analysis - Gender - Lexicography Angelina Sophiadi

References

Adorno, T. 2002. On Popular Music [with the assistance of George Simpson] (1941). In Essays on Music: Theodore W. Adorno; Selected with Introduction, Commentary and notes by Richard Leppert; New Translations by Susan H. Gillespie, 437-469. Berkeley and Los Angeles, California: University of California Press.

Androutsopoulos, J. and Scholz, A. 2002. On the recontextualization of Hip-Hop in European speech communities: A contrastive analysis of rap lyrics. Philologie im Netz (PhiN) (19), 1-42.

Brown, P. and Levinson, S.C. 1987 (1978). Politeness: Some Universals in Language Usage. Cambridge: Cambridge University Press.

Croft, W. 1994. Speech Act Classification, Language Typology and Cognition. In S.L. Tsohatzidis (Ed.), Foundations of Speech Act Theory. Philosophical and Linguistic Perspectives, 460-477. London and New York: Routlege.

Elliott, A. and Lemert, C.C. 2006. The new individualism: the emotional costs of globalization. Oxon: Routlege.

Fabb, N. 2010. Is literary language a development of ordinary language? Lingua 120 (5), 1219–1232.

Fairclough, N. 1995. Media Discourse. London: Edward Arnold.

Fairclough, N. 2001. Language and Power (2nd ed.). London: Longman.

Fish, S. E. 1976. How to do things with Austin and Searle: Speech Act Theory and Literary Criticism. MLN (Modern Language Notes) , 91 (5), 983-1025.

Discourse Analysis - Gender - Lexicography 1 3 9 Major Trends in Theoretical and Applied Linguistics

Fowler, R. 1991. Language in the News: Discourse and Ideology in the Press. New York: Routlege.

Frith, S. 1993. Popular Music and the Local State. In T. Bennett, S. Frith, L. Grossberg, J. Shepherd and G. Turner (Eds.), Rock and Popular Music: Politics, Policies, Institutions, 14-24. Routlege.

Grice, P.H. 1989. Meaning. In P.H. Grice, Studies in the ways of words, 213-223. Cambridge, Mass.: Harvard University Press.

Grossberg, L. 1993. The Framing of Rock: Rock and the New Conservatism. In T. Bennett, S. Frith, L. Grossberg, J. Shepherd and G. Turner (Eds.), Rock and Popular Music: Politics, Policies, Institutions, 193-209. Routlege.

Kimbel, T.M. and Protivnak, J.J. 2010. For Those About to Rock (With Your High School Students), We Salute You: School Counselors Using Music Interventions. Journal ofCreativity in Mental Health, 5 (1), 25 - 38.

Knobloch-Westerwick, S., Musto, P. and Shaw, K. 2008. Rebellion in the TopMusic Charts: Defiant Messages in Rap/Hip-Hop and Rock Music 1993 and 2003. Journal of Media Psychology, 20 (1), 15-23.

Kuhn, E.D. 1999. ‘I just want to make love to you’ - Seductive strategies in blues lyrics. Journal of Pragmatics (31), 525-534.

Lakoff, R.T. 2005. Civility and its Discontents: or, Getting in your Face. In R.T. Lakoff and S. Ide (Eds.), Broadening the Horizon of Linguistic Politeness, 23-44. Amsterdam and Philadelphia: John Benjamins Publishing Company.

Lyons, J. 1981 (1995). Linguistic Semantics: An Introduction. Cambridge University Press.

Mooney, H.F. 1954. Songs, Singers and Society, 1890-1954. American Quarterly, 6 (3), 221-232.

Paquette, K.R. and Rieg, S.A. 2008. Using Music to Support the Literacy Development of Young English Language Learners. Early Childhood Educ Journal (36), 227-232.

Pennycook, A. 2003. Global Englishes, Rip Slyme, and performativity. Journal of Sociolinguistics, 7 (4), 513-533.

1 4 0 Discourse Analysis - Gender - Lexicography Angelina Sophiadi

Pettijohn, T.F. and Sacco, D.F. 2009. The Language of Lyrics : An Analysis of Popular Billboard Songs Across Conditions of Social and Economic Threat. Journal of Language and Social Psychology, 297-311.

Quirk, R., Greenbaum, S., Leech, G. and Svartnik, J. 1972. A Grammar of Contemporary English. London: Longman.

Robinson, T.O., Weaver, J.B. and Zillmann, D. 1996. Exploring the Relation between Personality and the Appreciation of Rock Music. Psychological Reports (78), 259–269.

Rutten, P. 1993. Popular Music Policy: a Contested Area - The Dutch Experience. In T. Bennett, S. Frith, L. Grossberg, J. Shepherd and G. Turner (Eds.), Rock and Popular Music: Politics, Policies, Institutions, 37-51. London and New York: Routlege.

Searle, J.R. 1975. The Logical Status of Fictional Discourse. NLH (New Literary History), IV, 319-332.

Singer, M. 1989. Pronouns, Persons and the Semiotic Self. In B. Lee and G. Urban (Eds.), Semiotics, Self and Society, 229-298. Berlin: Mouton de Gruyter.

Twenge, J.M. 2007. Generation Me: Why Today’s Young Americans Are More Confident, Assertive, Entitled--and More Miserable Than Ever Before. New York: Free Press.

Wiehman, D. and Fuhs, S. 2006. Corpus Linguistics Resources: Concordancing software. Corpus Linguistics and Linguistic Theory (2-1), 109-130.

Discourse Analysis - Gender - Lexicography 1 4 1 Major Trends in Theoretical and Applied Linguistics

1 4 2 Discourse Analysis - Gender - Lexicography Verbs and Events in Discourse: A Synergy

Alexandros Tantos

Aristotle University of Thessaloniki [email protected]

Abstract

Verbal semantics has been shown a useful resource for reconstructing not only the semantics of whole sentences, but also of inferring relations between larger pieces of text. As Asher and Lascarides (1995), Danlos (2001a,b) and Tantos (2008, 2009) point out, lexical items have the power to simplify discourse inference and explain the ease with which we resolve ambiguities and construct the meaning of discourse. Based on the well-studied case of causative verbs in syntax-semantics interface, this paper aims to prove that it is not only a desired strategy to include lexical semantic knowledge in modeling discourse inference and interpretation, but also essential to include such lexically triggered information within the analyses of discourses.

1. Introduction

Since the first dynamic semantic theories (File Change Semantics by Heim 1981 and Discourse Representation Theory (DRT) by Kamp and Reyle 1993), it was made clear that content and structure in discourse go hand in hand within a hierarchical model of semantic representation and interpretation. The establishment of structural constraints in the update of discourse information paved the way for providing descriptive adequateness for a number of semantic phenomena, such as plurals, negation and pronoun resolution. One of the most successful tools and diagnostics for verifying or falsifying the reality of hierarchical models of discourse structure is anaphora resolution.

(1) a. John’s roommatei came to the party yesterday. Hei brought his own beer.

b. * If his roommate went to a partyi yesterday he must have been drunk. Iti was crazy.

c. Johni left yesterday and if hei went to the opera he must have been very satisfied.

Discourse Analysis - Gender - Lexicography 1 4 3 Major Trends in Theoretical and Applied Linguistics

In (1a) the two-sentence discourse allows John’s roommate to be the antecedent of the pronoun in the second sentence.1 (1a) is the easy case of resolving anaphoric elements, since the semantics of the discourse can be seen as a simple conjunction of the contents of the two sentences rendering all its content available for building a coreference relationship. Therefore, if the pronoun it denotes the variable x, then the assignment function g has no other option (with respect to its gender assignment) than to assign the value of the constant s brought by his roommate to the variable x, i.e. g(x)=s. However, in (1b) things are different. The pronoun it and the indefinite a party do not corefer with no apparent reason prima facie. The reason why this semantic misfire happens lies in the semantic constraint brought by the fact that the potential antecedent a party occurs within the conditional of the first sentence. Entities referred or discourse referents within conditionals have been shown not to be available for a coreference relationship with discourse referents outside the conditional. on the contrary, discourse referents within conditionals have access to referents outside of it as in (1c). This asymmetry as well as a number of other grammatical constraints relevant to anaphora resolution, such as quantificational operators, plurals and temporal phenomena, show that grammar affects and is affected by the hierarchical structure of discourse. In fact, the violation of the constraints leads to discourse incoherence in a strict interpretation parallel to that of ungrammaticality. Semantic structure is then the first and most obvious factor for imposing a hierarchical structure to discourse for expressing connections between bits of information within discourse. Soon after entering formal semantics’ machinery for interpreting discourse, influential theories of discourse semantics, such as DRT, acknowledge the presence of pragmatic factors that structure and shape partially but substantially the notion of discourse coherence. For instance, the function of using the second sentence in (2) is to explain the event described by the verb in the first sentence. Following textual order, the pushing enters discourse after the falling. The occurrence of falling sets a temporal anchor or reference point after which the next-mentioned event initiates. However, the default interpretation is that the pushing occurred before the falling and, thus, the semantic content of the discourse is not simply a conjunction of the contents of its parts in a compositional way similar to that of the semantic composition of the majority of single sentences.2 (2) John fell. Max pushed him.

1 I assume here that coindexing expresses the coreference relationship for the sake of the examples in this paper without implying any one-to-one connection between the syntactic mechanism of binding and the semantic coreference. 2 Asher and Lascarides (2003) offer a rich and well-established inventory of substantial reasons why dynamic semantics is not enough for interpreting discourse.

1 4 4 Discourse Analysis - Gender - Lexicography Alexandros Tantos

Therefore, the rhetorical structure of the discourse is an additional parameter that provides discourse its hierarchical flavor. In (2),explanation is the rhetorical function that imposes the reverse temporal order of the events in the discourse. Among the existing computational and pragmatic theories of discourse structure, only Rhetorical Structure Theory (RST, Mann and Thomason 1987) and Segmented Discourse Discourse Representation Theory (SDRT, Asher 1993, Asher and Lascarides 2003) take on board the idea that the representation of discourse information should include a well-defined inventory of rhetorical relations. Particularly, RST and SDRT implement discourse coherence as directed graphs with edges connecting Elementary Discourse Units (EDUs) or else discourse segments with rhetorical relations as representational devices. Moreover, only SDRT offers a formally precise framework that allows intrasentential information to interact with intersentential ones. Specifically, SDRT’s discourse structure, SDRS, is a triple (A, F, Last), where: • A is a set of labels that represent the EDUS, which represent the primitives of discourse or else the discourse terms of our formal apparatus and func- tion as arguments to discourse relations. • Last is a label in A (the EDU label that represents the content of the last clause inserted to the discourse logical form). • F is a function that assigns each member of A a member of a formula of the SDRS language, which includes formulas of some version of dynamic semantics (DRT, DPL, Update Semantics, MLTT, etc.).

Due to the formal semantic machinery that SDRT provides, which is quite illuminating for our purposes of associating lexical with discourse semantics, in the rest of the paper, I will first provide a crash course to the main factor that regulates discourse hierarchy, the principle called Right Frontier Constraint (RFC) as implemented within SDRT, in order to assess the validity of the examples and the conclusions of the paper. Then, I will show why it is not only more effective, but also essential to allow rhetorical relations to interact between EDUs as well as lexically triggered discourse units. The last part of the paper is devoted to a new view of how segmentation of discourse information works and provides an insight as to what criteria are necessary to define EDUs.

2. RFC and Discourse Relations

The Right Frontier Constraint is originally proposed by Polanyi (1988) and is one of the most well-documented constraints in literature of formal discourse semantic and pragmatic theories. RFC is deemed responsible for the way discourse update is built and discourse anaphora is resolved. In order to understand the importance of RFC for explaining the hierarchical flavor of discourse, it is essential to grasp the distinction between two types of discourse

Discourse Analysis - Gender - Lexicography 1 4 5 Major Trends in Theoretical and Applied Linguistics

relation, subordinating and coordinating ones. Briefly, the idea behind this distinction is that by uttering the current segment the author-speaker either provides information in the same direction with respect to a prior segment within a range of different utterance modes, such as explaining, elaborating or commenting, (subordinating relations) or entirely shifting the topic of discourse to new directions with various ways, such as narrating, providing a background or a result of the prior segment (coordinating relations). The type of rhetorical relation is the most important part of RFC. Schematically, the difference between the two types with respect to RFC is reflected within graphical representations. The nodes of a SDRS graph represent its labeled constituents and its edges represent the discourse relations between these constituents. Thereby, downward edges express subordinating relations and horizontal edges express coordinating ones. The downward direction in the case of subordinating relations signals the degree of granularity of the information provided. An important restriction is that two nodes in a graphical representation cannot be connected using both a subordinating and a coordinating relation. For instance, in (3), a simple three-sentence discourse, contains five EDUs and the SDRS graph is depicted in figure 1. π2 and π3 elaborate on the visit to the ophthalmologist mentioned in π1.3 On the other hand, π4 and π5 do not provide any more detail about the event described in π1. These EDUs denote a shift on the topic and they continue the discourse in a different direction, in the sense that they describe new events that happened one after the other, the driving and the drinking. π4 builds a narration relation with π1 and not with π3 or π2 and, hence, since narration is a coordinating relation, it is connected with a horizontal edge with π1.4 Accordingly, π5, being related narratively with π4, is connected with a horizontal edge with it.

(3) Yesterday evening I visited an ophthalmologist (π1). He diagnosed a small eye problem (π2) and wrote me a prescription (π3). Then, I drove back home (π4) and drank a beer (π5).

π1 Nar π4 Nar π5

Elab

π2 Nar π3

The SDRS graph of (3).

3 EDUs are usually represented by the Greek letter π. Throughout the paper this convention is used. 4 According to Asher and Vieau (2005), narration and elaboration are the prototypical coordinating and subordinating discourse relations respectively.

1 4 6 Discourse Analysis - Gender - Lexicography Alexandros Tantos

Asher (1993) implements RFC by inserting a rule defining the right frontier. The EDUs that lie on the right frontier of the discourse graph built so far are available for attachment; in other words the update of information in the discourse allows new information to either attach to the last entered EDU β in a discourse structure or to some other constituent γ such that (β,γ) is in the transitive closure of the subordination relation. More formally, the right frontier is defined in SDRT as the set of available nodes for attachment falling under the following possibilities: 1. The label α = LAST; 2. Any label γ ≥D* α, where ≥D* is defined recursively: a)  R(γ,α) is a conjunct in F(l) for some label l, where R is a subordinating discourse relation; b)  R(γ,δ) is a conjunct in F(l) for some label l, where R is a subordinating discourse relation and  F(δ) contains as conjunct R’(δ’,α) or R’(α,δ’), for some R’ and δ’; or c)  R(γ,δ) is a conjunct in F(l) for some label l, where R is a subordinating discourse relation and  δ ≥D* α.

Attachment of new information is directly influenced by the number and placement of available EDUs in the right frontier. At the same time, availability of attachment points on the right frontier should allow no more and no less than the EDUs that correspond to the context, since otherwise its definition would lead to over- and/or undergeneration in terms of discourse structures.

3. Discourse Segmentation and Anaphora Resolution

The majority of formal discourse semantic theories, such as SDRT, acknowledge the essence of lexical semantic knowledge for inferring and interpreting discourse. However, the need for including lexical semantics should be established within and by a solid theory based on real language examples on a case-by-case study. For our purposes, anaphora resolution is perhaps the most reliable test and criterion by which one can enter the discussion of how natural language speakers segment utterances for encoding semantic relations between bits of information in the discourse. In other words, following Asher (1993), I intend to cash out the distinction between real and natural language metaphysics by assuming that pronouns refer to real entities of natural language metaphysics with no commitment as to the model interpretation of those entities. Keeping that in mind one could factor out possible antecedents to anaphors as real entities independently of their grammatical form.

Discourse Analysis - Gender - Lexicography 1 4 7 Major Trends in Theoretical and Applied Linguistics

The immediate repercussion of this idea is that it is possible to focus on semantic entities that would do not necessarily correspond to syntactic forms. The appeal to semantic entities that are part of our daily discussions is not only desired but also necessary in order to provide simple analyses close to linguistic reality, since otherwise the coherence of smaller or larger discourses cannot be explained. Let us consider an illuminating example below:

(4) Yesterday John broke the carafe (π1). Its pieces were lying all over the floor (π2) and it was such an elegant carafe (π3). He was deep in his thoughts (π4) while he was walking up and down the room (π5) and hit it with his knee (π6).

In (4) it seems we are able to pick up and talk about parts of the event of breaking.5 Moreover, these parts are well-defined and they are related to the cause and result of the breaking event, they are not arbitrary spatiotemporal bits of information and they influence discourse coherence. The EDU π1 refers to an event of breaking with an agent and a theme. π2 elaborates on the state of the broken carafe and π3 comments on a property of the theme in π1, before the breaking took place. Furthermore, both verbs of π2 and π3 are in the same past tense and there is no obvious temporal marking that differentiates between the before and after the attainment of the broken state. π4 and π5 provide background information and π6 elaborates on the cause of the breaking state. Furthermore, π6 is related to π1, elaborates on the cause and not on the result of the breaking. If one does not take for granted the presence of EDUs triggered by lexical items, the verb ‘break’ in (4), there is no obvious way of explaining: a) why we take a reading in which π2 refers to a state initiated well after the initiation of the state of John being-deep-in-thoughts in π4 and the state of walking-up-and-down-the-room in π5, b) why we interpret π2 elaborating on π1 and at the same time π4 and π5 provide background information on π1 while π2, π4 and π5 being states related to π1, and c) why we can resolve the anaphor ‘it’ in π6 with the carafe in π1, when RFC is violated.6

An immediate answer to these issues and the ease with which we infer and interpret events in various discourses would be to include some Neo-

5 We are also able to talk about properties of the participants of an event. However, this has always been considered part of discourse processing of salient entities. 6 It violates RFC, since EDU 6 cannot be related to 1 due to the fact that it is not in an accessible attachment site. For more on RFC, cf. Asher and Lascarides (2003).

1 4 8 Discourse Analysis - Gender - Lexicography Alexandros Tantos

Davidsonian event-based theory following Parsons (1990) that separate the causing part of an event, as a subevent that precedes the event of a result state. However, this approach does not take us much further for the following reasons: a) one should appeal again to subevents as part of the analysis, and since subevents are ultimately events in their turn , this leads us to problems related to their identity conditions, as Tantos (2011) notes; b) I see no apparent way of formally encoding relations between events that would cover rhetorical relations. Questions relating to issues of elaborating on an event, providing background to it, or explaining it on the basis of another event, are totally unclear in terms of model interpretation. Discourse relations connect bits of related information in a specific context of usage. The solution I suggest is to include lexical labels in the repository of discourse theories and assign them a special status on account of their implicit presence. By implicit presence I mean that their presence is not marked by any sort of explicit marker and is explicated only when discourse inference and interpretation is pursued. Following work by Danlos (2001a, b) and mainly by Tantos (2008, 2009), I claim that lexical EDUs are present in discourse inference and carry typed information, but these EDUs should not be constrained in the same way that other EDUs are. In (4) the verb ‘break’ inserts two lexical EDUs, 1a and 1b, which are not subject to the RFC and, thus, they are accessible to EDUs π2, π4 π5 and π6. Moreover, π1 is still present and accessible to other EDUs following the standard rules of discourse inference. That is why π3 has access to it and comments on the carafe independently of the cause and result of breaking. If π1 were not present, and all there was were the lexical EDUs, it would be incoherent to talk about the carafe with no mention to the event of breaking (cause and result). In other words, verbs insert lexical EDUs without disturbing the segmentation of the discourse. Discourse segmentation is a difficult task for any formal semantic theory and any system of computational semantics. The main reason lies on the fact that until now the majority of attempts supports a one-to-one correspondence between surface forms such as sentences and propositions. Only recently, Schlangen (2003) and Schlangen, D. and A. Lascarides (2003a, b) consider a formal approach in which surface utterance forms may correspond to complex EDUs. Discourse segmentation, then, has the main burden for a whole class of inferences when causative verbs are mentioned in the discourse. Lexical EDUs function as default templates for inferring connections with other utterances in the discourse and have a conceptual status clearly intermediating between linguistic and extralinguistic knowledge. It is this level of description that one needs to investigate more in order to decode inference patterns on the one hand and the limits of linguistic theorizing in discourse.

Discourse Analysis - Gender - Lexicography 1 4 9 Major Trends in Theoretical and Applied Linguistics

4. Conclusion

By acknowledging the existence of lexically triggered information, we are in the position to explain in a natural way why discourses are interpreted easily by human interpreters without resorting to heavy inference machinery about possible event connections. In this paper I showed that a) the current approach to discourse segmentation along with its definition of attachment availability is not able to describe the complexity of discourse connections within a single example that includes a causative verb and that b) one should not be based on event-based theories for inferring and interpreting discourse connections, since these are semantic entities for sentential interpretation and do not directly affect discourse ineference and interpretation. Moreover, one needs to assume lexical EDUs that have clearly a conceptual status and serve default inferences in discourse.

1 5 0 Discourse Analysis - Gender - Lexicography Alexandros Tantos

References

Asher, N. 1993. Reference to Abstract Objects in Discourse. Dordrecht: Kluwer.

Asher, N. and A. Lascarides. 1995. Lexical disambiguation in a discourse context. Journal of Semantics 12(1):69–108.

Asher, N. and A. Lascarides. 2003. Logics of Conversation. Cambridge: Cambridge University Press.

Asher, N. and L. Vieu. 2005. Subordinating and coordinating discourse relations. Lingua, 115(4), 591–610.

Danlos, L. 2001a. Event coreference between two sentences. In H. Bunt and R. Muskens, eds., Computing Meaning 2. Kluwer Academic Publishers.

Danlos, L. 2001b. Event coreference in causal discourses. In P. Bouillon and F. Busa, eds., The Language of Word Meaning. Cambridge University Press.

Heim, I. 1982. The Semantics of Definite and Indefinite Noun Phrases, University of Massachusetts: Ph.D. dissertation; published 1989, New York: Garland.

Kamp, H. and U. Reyle. 1993. From Discourse to Logic. Dordrecht: Kluwer Academic Publishers.

Parsons, T. 1990. Events in the Semantics of English. MIT Press.

Polanyi, L. 1988 A formal model of the structure of discourse. Journal of Pragmatics, 12, 601–639.

Discourse Analysis - Gender - Lexicography 1 5 1 Major Trends in Theoretical and Applied Linguistics

Schlangen, D. 2003. A Coherence-Based Approach to the Interpretation of Non- Sentential Utterances in Dialogue. PhD Thesis, School of Informatics. University of Edinburgh.

Schlangen, D. and A. Lascarides. 2003a. The Interpretation of Non-Sentential Utterances in Dialogue, in Proceedings of the 4th SIGdial Workshop on Discourse and Dialogue, Sapporo, Japan, July 2003.

Schlangen, D. and A. Lascarides. 2003b. A Compositional and Constraint-Based Approach to Non-Sentential Utterances, in S. Müller (ed.) Proceedings of the 10th International Conference on Head-Driven Phrase Structure Grammar, Michigan, USA, July 2003.

Tantos, A. 2008. Computing Events in Discourse: A Case Study Involving Light “have”. PhD Dissertation, Konstanzer Online-Publikations- System (KOPS), Konstanz.

Tantos, A. 2009. Lexically Triggered Discourse Relations. In Selected Papers from the 18th International Symposium on Theoretical and Applied Linguistics, A. Tsangalidis (eds.), Dept of English Language and Literature, Aristotle University of Thessaloniki.

Tantos, A. 2011. Κειμενογλωσσολογικοί Παράγοντες στη Διδασκαλία του Κλιτικού Αναδιπλασιασμού της Ελληνικής σε Φυσικούς Ομιλητές Γερμανικών Γλωσσών. Στο Μελέτες για την Ελληνική Γλώσσα 31, 485–495, Αριστοτέλειο Πανεπιστήμιο Θεσσαλονίκης: Θεσσαλονίκη.

1 5 2 Discourse Analysis - Gender - Lexicography Do We Always Speak of What We Mean? (Non)Referring Expressions in Discourse

Efthymia Tsaroucha

Aristotle University of Thessaloniki [email protected]

Abstract

This paper investigates the use of (non)referring expressions by assuming that speakers do not always use reference to speak of what they mean or of what they actually intend to mean. (Non)reference is shaped both by the intention of the speaker and the way the audience makes the identification with what is being referred. Examples are drawn from newspaper articles, headlines, political speeches and advertisements. (Non)referring expressions will not be solely restricted to attributive or equatative functions since their meaning is shaped by the interrelation of the situation derived from a particular context, on the basis of who, to whom, why, when, where and how norms of interaction, interpretation and genres are used.

1. Introduction

This paper aims to investigate persuasive language in political and media discourse, particularly advertising language. Persuasion as a form of communication is ubiquitous not only in formal texts, but also in everyday speech, since its basis lies on the transmission of feelings, moods and attitudes towards audiences. Persuasion, however, is composed of two versions. On the one hand, persuasion can be termed as a kind of “language-like representations that capture the ideational content of the mind” (Eysenck and Keane 2000: 246), or to put it better as an information processing activity, in which thoughts are actively manipulated to create new beliefs and attitudes. On the other hand, persuasion is interrelated with the notions of conviction and manipulation. In particular, Woolbert (1923: 19) claimed that the fixing of a line between what is called conviction and what is called persuasion could not be successfully done because both notions seem to seek the same end. When it comes to

Discourse Analysis - Gender - Lexicography 1 5 3 Major Trends in Theoretical and Applied Linguistics

the interrelation of persuasion with manipulation Campell ([1989] 2000: 57) proposed that manipulation refers to the degree according to which the language user is a willing or a naïve participant in the process of being compelled towards products, or ideas. In this paper, I have decided to investigate the first version of persuasion, without however, ignoring the fact that language itself can be misleading in many cases since the hidden meaning behind utterances and texts cannot be conceived at a superficial reading; on the contrary, it is my belief that persuasive language should be analyzed on the basis of both linguistic knowledge and of social, political, economical and cultural factors as well. After having presented the version of persuasion I am going to analyze, I should also note that approaching linguistic meaning under discourse analysis and critical discourse analysis concerns at a starting point the relationship between the discourse and the speakers and hearers, by and for whom it is produced. To be more precise, meaning is explained by means of analyzing the way speakers take and relinquish the role of the speaker, social roles and relationships between the participants (Coulthard 1981: 11). According to Fairclough and Wodak (1997: 258) critical discourse analysis sees discourse- language use in speech and writing as a form of social practice due to the dialectical relationship between a particular discursive event and the situations, institutions and social structures which frame it. As it was mentioned in the very beginning of the introduction, persuasive language exists not only in texts but also in every day speech. In this respect, critical linguistics and critical discourse analysis focus not only on texts, spoken or written as objects of inquiry. On the contrary, individuals or groups should account on a theorization and description of both social processes and structures in order to create meanings in their interaction with text (Wodak 2001a: 3). As Jager (2001: 35) points out discourse creates the conditions for the formation of subjects and the structuring and shaping of societies. With regard to the aforementioned remarks, I would like to add that the notion of persuasion is strongly related to the concept of power. Power is not simply dependent on grammatical forms of a text, especially, if we consider speaker’s control over a social affair regarding the genre of a text; on the contrary power is reflected in the way people make use of grammatical structures and with regard to this, it shapes meaning and creates ideology. Power, in my opinion, constitutes a fundamental component of discourse and critical discourse analysis since it encompasses a wide range of discourses, such as news, political speeches and advertisements. In this paper, I will attempt to analyze the concept of power and its relation with persuasion on the grounds of referring expressions (either if they serve as counterparts of reference, i.e., deictics, proper names, generics, indexicals etc., or as non-referring expressions). In addition, I view that a critical approach to persuasive language indicates that the concept of power can be interpreted on the basis of certain features. To

1 5 4 Discourse Analysis - Gender - Lexicography Efthymia Tsaroucha

be more precise, desirability, importance-significance, warrantability, probability, comprehensibility, obviousness, utility, expectability and appropriateness contribute to the interpretation of persuasive language simply because they are subject to certain notions like respect, freedom, efficiency, morality, trust (worthiness), legality, virtuousness, beauty, wealth, quantity, magnitude, scope and dimensionality, which by definition aim at communicating ideas or products (Graham 2003: 114). Summarizing the introductory part of the paper, I should note that the investigation of persuasive language on the grounds of (non)referring expressions and under the discipline of discourse and critical discourse analysis indicates that the combination of both linguistic and extra-linguistic knowledge is necessary. In the following sections I will attempt to propose that the usage of (non)referring expressions in political speech and in advertising serves as persuasive tools because the function of (non)reference is strongly related with the concept of power. As it will be shown, (non)referring expressions, when used in particular discourses, such as politics reflect a certain kind of ideology and point of view, which affects the cognition of the given audience. In conclusion, I should note the aim of this paper is to emphasize the “hidden” meaning of the overall of context, in which (non)referring expressions serve as an instance of persuasive language. My claim is based on Scollon’s (2003: 169) argument stating that language (in our case, the use (non)referring expressions) affects the cognition of its speakers, but at the same time they should be aware of the circumstances in which they come to use the language in question. Lastly, a strong emphasis will be given on context, since as it has been proposed contextual analysis is a combination of not only immediate language or internal co-text, but also of extra-linguistic sociological variables reflected in the broader sociopolitical and historical contexts (Wodak 2001b: 67).

2. The Special Status of News, Political Speeches and Advertisements as Discourse

This section aims to investigate the discursive dimensions of news, politics and advertisements. In particular, it will be argued that these discourses have a strong and dynamic impact on the presupposition of knowledge as they affect judgment and appreciation. Judgment and appreciation should constitute a fundamental component of critical discourse analysis, as they point at deep- seated processes of marketization (Billig 2003: 36) and institutional positions (Chouliaraki and Fairclough 1999: 9), which shape the special status of news, politics and advertisements.

Discourse Analysis - Gender - Lexicography 1 5 5 Major Trends in Theoretical and Applied Linguistics

To begin with the special status of news as discourse, it should be noted that their status lies on cultural determination, report of the event, register in terms of generic structure and information addressed either to a general public or to an objective point of view or even to journalist’s self image (Caldas-Coulthard 2003: 274-5). Moreover, news selection is regulated by general news values such as frequency, size of the event, (un)ambiguity or clarity of the event, (un) predictability, (un)expectedness, mixture of different kinds of events, reference to elite persons and nations, personalization and negativity (ibid: 277). According to Meyer (2001: 27) there are certain discursive strategies the researcher should follow, while analyzing news; for instance, nomination, predication, argumentation, perspectivation and intensification are necessary steps for the analysis of the particular genre. In my understanding the application of this mechanism is grounded by some further steps the researcher should follow. I think that focus on semantic macrostructures, an extensive analysis of local meanings and formal structures (by taking into account specific linguistic realizations) and context result in a complete and critical processing of news. Moreover, politics as a discourse type concerns the overall institutionalization of public sphere. According to Calhoum (1992: 12-13) politics is characterized by a kind of social intercourse (status), mutual willingness to accept given roles, notion of common interest, or right policy and problematization of areas that until now may not have been questioned. Public space is generated by public discourse due to the link between power, legitimacy and discourse. Benhabib (1992: 85-95) proposed a discursive model of public space according to which the issue of social differentiation is reflected in the aspects of modernity, personality formation and bifurcation, and public space should keep a reflexive and critical attitude towards the individuals. At this part, I would like to propose that the concept of public sphere could be studied as a feature of public discourse, or even as an instance of political discourse. More precisely, if we consider that the language of politics is composed of a body of private persons, which has to discuss social matters in public, we could further consider that the issues of discursive interaction and public opinion need to be dominant. Lastly, advertisements as a discourse type are composed of text, which in its formal definition is used to mean linguistic forms, related with context. According to Cook (1992: 1-2) the context of advertisements includes substance (physical material which carries text, music and pictures), paralanguage (meaningful behavior accompanying language such as voice quality, facial expressions and choice of type face), situation (properties and relations of objects and people in the vicinity of the text, as perceived by the participants), co-text (text that precedes or follows that under analysis), inter-text (text which the participants perceive as belonging to other discourse), participants (senders, addressers,

1 5 6 Discourse Analysis - Gender - Lexicography Efthymia Tsaroucha

addressees and receivers) and function (what text is intended to do by the senders and addressers, or perceived to do by the addressees). It is further worth mentioning that the substance of advertisements lies on communication. According to Cook (1992: 23-25) advertisements use secondary substances of language and should be analyzed by virtue of their purposes such as, attention-getting, talking micro-chips, novel substance or fixing the product more firmly in the memory by actively involving the reader. Despite the usage of secondary substances of language, advertisements are also rich in signs, which in their turn aim to serve as persuasive tools. According to Danesi (2002: 17) signs mediate reality and never really denote or represent the whole truth, because they “constitute convenient selections from the infinite realm of the knowable”. In the following section, it will be claimed that the “hidden” meaning of both news along with politics and advertisements is governed not only by the written text, but also by the current signs. As it will be shown, (non)referring expressions in relation to signs construct associations and shape meaning. Signs along with (non)referring expressions constitute a fundamental part of visual modality, which in the case of political speeches, news and advertisements contributes to the perception of a hidden quality and latency of the text. Lastly, the methodological analysis I have decided to follow is based on Blanchard’s (1980: 22) observations on the semiological reading of texts. According to Blanchard (1980: 22) a semiological reading of a text generally entails the perception and the internalization of a stylistic structure, which consists of a self-referential value of the text. Thus, the description of any discourse type holds our attention because something is partial and something else may, or has to be missing.

3. On (Non) Referring Expressions

The theorization of referring expressions primarily deals with the notions of deixis, definiteness, (in)direct reference and indexicality. Among the years of theoretical descriptions on meaning various rules, exceptions, regulations and even newer counterparts of referring expressions emerged. For reasons of economy I will not discuss all these theoretical accounts; however, I will present a definition on the counterparts of reference by investigating their function within genres of persuasive language, which constructs the subject of this paper. I will attempt to connect each category of referring expressions with the “hidden” meaning lying behind political speeches, advertisements and newspaper headlines. Firstly, deixis and deictic expressions constitute a fundamental part of referentiality. According to Huang (2007: 132), the term deixis derives from the Greek verb /dihno/, which means to show or to point out, and it is interrelated both with the linguistic structure and the context in which language is used.

Discourse Analysis - Gender - Lexicography 1 5 7 Major Trends in Theoretical and Applied Linguistics

The process of pointing or showing encodes universal features, no matter how strong cultural and linguistic diff erences may be, since a language without deictics cannot serve communication. Lastly, the organization of deixis follows an egocentric way because it represents not only the speaker of an utterance (“deictic origo”), but also the audience (“ground zero”) (Huang 2007: 135). Furthermore, deictic expressions are subcategorized into person, spatial and temporal aspects. To be more precise, person-social deixis concerns addressor- addressee relations and possibly addressor-general (public) audience relations. Social-person deixis is ruled by “[…] the codifi cation of the social status of the speaker, the addressee, or a third person or entity referred to, as well as the social relationships holding between them” (Huang 2007:163). The sub-category of spatial deixis concerns the place of the utterance and is introduced with locative adverbs, demonstratives, and determiners; it is also possible to show a similar sort of dominance relation to the personal pronouns (Cruse 2004: 333). Finally, temporal deixis refers to the so-called “time axis”, which stands for the moment the utterance is produced and at the same time it is grounded as the reference point. After having described the behavior of deictic expressions, I would like to point out that deixis is widely used in advertisements since its combination with visual modality draws our attention eff ortlessly towards the current product. The given advertisement is composed of a text, weaved together with the logo of the national bank of Greece (the trading product), in order to promote its creditworthiness over other banking loan agreements.

Figure 1. Advertisement

1 5 8 Discourse Analysis - Gender - Lexicography Efthymia Tsaroucha

As I have mentioned in the previous section (where I have argued for the special status of adverisements as a discourse type), one of the basic purposes of the particular genre is to present the product more firmly by actively involving the reader. In the particular example, the reader seems to seek for the most convenient financial transaction, if we consider the fact that other possible mortgage packages are crossed out (figure ).1 It seems to me that in the given advertisement, we face a complex instance of deixis since the indefinite description [mia] (“one”) functions deictically (by embracing both spatio-temporal and social readings), because it refers to the national bank of Greece, which happens to be pointed out (the trading product, as illustrated in Figure 1, is circled). The national bank of Greece is launched as the right place for such a transaction and at the same time the advertisement urges its readers to choose the trading product by constructing the implication that their consumer dilemma came to an end. My assumption derives from the fact that opponent products being crossed out, whereas the bank’s logo is circled. Lastly, with regard to the social function I believe that the deictic expression [mia] (“one”) is associated with the socio-economic situation of Greece, whereby the overall credit system is under devaluation; thus, the particular bank may ensure stability and credibility. Along the same lines, it is also worth mentioning that deixis and reference are interrelated, and reference is further expanded into three subcategories: definite, indefinite and generic. Firstly, definite reference is defined asthe identification of referents (of definite descriptions), which helps the hearer to reconstruct what the speaker has expressed (Cruse 2004:320). Definiteness is known as the familiarity hypothesis since the denoted entities are familiar to both speakers and hearers so that general knowledge in relation to cross- reference or associative uses can arise (Lyons 1999: 4). Secondly, identifiability in combination with the notions of inclusiveness and uniqueness is in parallel relations with definiteness because the hearer is in a position to exploit clues in the linguistic and extra-linguistic context so that to establish the identity of the referent (Lyons 1995: 5-6). Lastly, generic reference concerns the general or overall reference towards a class of referents whereby something can be predicated of the whole class referred to, or of each member of the class (Cruse 2004: 320, 323). Is it actually possible for a referring expression not to be purely referential if it brings together all the above functions? I think that in persuasive language and particularly in media discourse the perception and the internalization of a stylistic structure is generated from a hidden quality or latency; something may be partial or something else has to be missing.

(1) Chronology of Terror, 11 September 2001 (USA Today)

Discourse Analysis - Gender - Lexicography 1 5 9 Major Trends in Theoretical and Applied Linguistics

This example comes from a newspaper headline and equates a particular social situation with a specific date. In my understanding this headline can be treated as an elliptical equatative clause (comma links the two expressions and equates them). Ellipsis is expected in newspapers’ headlines either for reasons of economy or for stylistic factors regarding news selection such as personalization, negativity and (un)predictability (as analyzed in Section 2). As a consequence, the equation of the expression “chronology of terror” with “11 September 2001” refers to a definite and inclusive event, which is familiar to the readers because they are able to extract clues (either from the linguistic or extra-linguistic context) and identify the referent. Despite the fact that the event is definite and specific (a particular temporal and spatial location), the public sphere in which it refers to is not. It concerns a body of participants, either passive (the relatives of the victims of the terrorist attack, the overall body of US citizens, and citizens of foreign countries as well) or active (the victims of the terrorist attack). In this respect, I claim that the persuasive function of the given headline, despite the ellipsis, aims to introduce the terrorist attack of 9/11, as a chronology of terror, which every man who is opposed to violence has to refute. In the particular case, it is observed that a definite referent may also function as the associative referent of a mass public group. As a consequence, definite reference can co-exist with generic reference if we think of media discourse as a more flexible genre (news) by virtue of structure and style (ellipsis). What is more, I view that the distinction between constant-variable reference on the one hand and objectual-descriptive from the other is needful for the analysis of (non)referring expressions in the field of persuasive language. According to Saeed (1997: 26-27) constant reference concerns expressions that have the same referent across a range of different variables whereas variable reference denotes expressions, which are totally dependent on each context (1997:26-7). Extension is a part of variable reference because referents can be either picked out by uttering the expression in a particular context or extracted from a set of things, which could possibly be the referent of that expression (1997:27). Lastly, as far as the distinction between objective and descriptive reference is concerned, Bach (1987: 66) supports that the first type occurs when the addressee identifies the referent despite of what the addressor has already produced, while the second type emerges in case that the speaker intends to talk about whatever uniquely satisfies the individual concept (and the hearer is able to recognize the determining referent). The examples below are drawn from newspaper articles and concern political statements. In (2) the “US president” functions as constant reference because it expresses the same referent across the world. However, the expression “an apparent terrorist attack” is used attributively not only because of the presence of the propositional attitude verb “call”, which denotes personal perspective

1 6 0 Discourse Analysis - Gender - Lexicography Efthymia Tsaroucha

and intention, but also because hearers are in a position to make the exact identification with what the speaker intends to denote by relying either on the particular context, or on general and shared knowledge about the world. In (3) the expression “a White House spokesman” is used indefinitely, while the expression “a threat” is an instance of definite reference, because it illustrates a particular political situation. If we assume that the utterance concerns a political statement (since the speaker a White House’s spokesman), we could also claim that the audience expects him/her to refer to the current political agenda, as the special status of politics foresees (Section 2) and they can further perceive spokesman’s words as the intended referent.

(2) US president […] called the destruction in New York “an apparent terrorist attack on our country” and ordered a full-scale investigation to hunt down the folks who committed this act. (The New York Times)

(3) Yesterday a White House spokesman stressed Iraq was a threat under Saddam before the 2003 invasion. (The Daily Mirror)

4. A Step beyond Reference: Indexicality, Politics and the Public Sphere

As far as indexicality is concerned, Koutoupi-Kitis (2009-10: 262) claims that there are certain elements of speech, geographically or socially located, which are considered to be indexes that not only put the speaker in a particular situation, but also witness details concerning speaker’s identity. Under this proposal, indexicality is another type of reference, which carries, apart from purely contextual or structural information, some further details deriving from the general world context, speakers’ inferences and idiosyncratic beliefs. This type of reference surpasses the strict linguistic environment of the utterance and it is associated with semantic features. More specifically, Cappelen and Lepore (2002: 276) claim that the nature of indexicals is constrained both by their syntactic and epistemological side; the former refers to anaphoric relations, whereas the latter generates certain kinds of a priori truths. From my point of view, if the linguistic meaning of an expression is self-referential with its structure, it grounds universal truths; thus, any referring expression of this type must have a universal character, which undergoes denotational changes at any given context. I also believe that in the case of persuasive language, both the syntactic and epistemological side of indexicals is in a cline. To be more precise, indexicals that participate in overt anaphoric relations, like the personal pronoun “I”, are deep in the sense that they make precise what is universally true.

Discourse Analysis - Gender - Lexicography 1 6 1 Major Trends in Theoretical and Applied Linguistics

The observation of the character of the indexical (“I”) at any given context, witnesses its function as the agent of that context. In (4) the indexical “I” refers to the speaker of the particular utterance, the agent of the particular context; The political statement is drawn from Obama’s campaign speech at the Democratic National Committee (fall meeting 2007), where he was called to support his predominance over his political opponent for the presidential elections.

(4) The only reason I am standing here today is because somebody somewhere stood up when it was risked. (DNC)

However, I view that its indexical function is expanded into descriptive reference because the speaker intends to talk about whatever “[…] uniquely satisfies the individual concept that the hearer is able to recognize as determining the referent” (Bach 1987:66). I suppose that the personal pronoun “I” designates multiple identities because the speaker is presented both as a political figure and as the expectant president and even as the most right choice for the elections, because the referring expression is extended to general addressees like “somewhere” and “somebody”. Consequently, I propose that in politics, when the personal pronoun “I” functions as an indexical is neither a pure indexical nor a pure referring expression. I think that it should be approached as an in-between reference and indexicality element. It functions as a collective a body since it expresses a set of referents. It embraces the political ideology of the speaker by invoking audience’s political acceptance.

5. Concluding Remarks

This paper investigated the function of (non)referring expressions under the field of media and political discourse. Instances of these discourses (such as advertisements, political speeches, newspaper articles and headlines) showed a strong flexibility by virtue of structure and style. In these genres language can be used in a more expressive manner on the basis of speaker’s intentions and beliefs. The factor of public sphere with reference to social, political and economical features shapes the way language is used. The initial question of the paper, if we always speak of what we mean or if we always mean of what we speak, was answered multiply. Firstly, deictic expressions can encompass spatial, temporal and social readings simultaneously. Secondly, definite reference equates different expressions, if hearers are in a position to make the exact identification with what the speaker intends to denote. Moreover, referring expressions cannot be approached as purely referential, if their

1 6 2 Discourse Analysis - Gender - Lexicography Efthymia Tsaroucha

readings are a combination of a range of different variables, either extended or extracted from a set of things, which hearers may be in a position to recognize as the determined referent(s). In conclusion, I believe that reference without reference is ubiquitous in language, in spite of genres and discourses. However, cases of “misleading” language can be more easily examined in more formal discursive structures because of effortless access (either with respect to advertisements or news). The theorization of rules and regulations of (non)referring expressions is shaped by addressor-addressee, addressor-audience and addressee-public sphere relations. As it was shown in this paper, the use of (non)referring expressions in certain discourses, such as news, political speeches and advertisements has a persuasive function since it reflects a certain kind of ideology and social behavior appealing to the target audience.

Discourse Analysis - Gender - Lexicography 1 6 3 Major Trends in Theoretical and Applied Linguistics

References

Bach, K. 1987. Thought and Reference. Oxford: Oxford UP.

Benhabib, S. 1992. Models of Public Sphere: Hannah Arendt, the Liberal Tradition and Jürgen Habermas. In C. Calhoun (ed.), Habermas and the Public Sphere. Cambridge, Massachusetts: MIT Press, 73-98.

Billig, M. 2003. Critical Discourse Analysis and the Rhetoric of Critique. In R. Wodak and G. Weiss (eds.), Critical Discourse Analysis: Theory and Interdisciplinarity. New York: Palgrave Macmillan, 35-46.

Blanchard, M.E. 1980. Description: Sign, Self, Desire: Critical Theory in the Wake of Semiotics. The Hague and New York: Mouton Publishers.

Caldas-Coulthard, C.R. 2003. Cross-Cultural Representation of ‘Otherness’ in Media Discourse. In R. Wodak and G. Weiss (eds.), Critical Discourse Analysis: Theory and Interdisciplinarity. New York: Palgrave Macmillan, 272-296.

Campell, C. [1989] 2000. The Puzzle of Modern Consumerism. In M. Lee (ed.), The Consumer Society Reader. Malden, Massachusetts: Blackwell, 48-72.

Capellen, H. and Lepore, E. 2002. Indexicality, Binding Anaphora and a Priori Truth. Analysis, 62 (4): 271-281.

Chouliaraki, L. and Fairclough, N. 1999. Discourse in Late Modernity: Rethinking Critical Discourse Analysis. Edinburgh: Edinburg UP.

Cook, G. 1992. The Discourse of Advertising. London and New York: Routledge.

Coulthard, M. 1981. An Introduction to Discourse Analysis. London and New York: Longman.

1 6 4 Discourse Analysis - Gender - Lexicography Efthymia Tsaroucha

Cruse, A. 2004. Meaning in Language. Oxford: Oxford UP.

Danesi, M. 2002. Understanding Media Semiotics. London, UK and New York: Arnold, Hodder Headline Group and Oxford UP.

Eysenck, M. and Keane, M. 2000. Cognitive Psychology, 4th edition. Hove: Psychology Press Ltd.

Fairclough, N. and Wodak, R. 1997. Critical Discourse Analysis. In T. van Dijk (ed.), Discourse Studies: A Multidisciplinary Introduction. London: Sage, 258-284.

Graham, P. 2003. CDA and Evaluative Meaning: Interdisciplinarity as a Critical Turn. In R. Wodak and G. Weiss (eds.), Critical Discourse Analysis: Theory and Interdisciplinarity. New York: Palgrave Macmillan, 110-129.

Huang, Y. 2007. Pragmatics. Oxford: Oxford UP.

Jager, S. 2001. Discourse and Knowledge: Theoretical and Methodological Aspects of A Critical Discourse and Dispositive Analysis. In R. Wodak and M. Meyer (eds.), Methods of Critical Discourse Analysis (1st edition). London: Sage, 32-62.

Kitis-Koutoupi, E. (2009-10), (MS). Semantics: The Fundamentals of Meaning in Language. Thessaloniki: Aristotle University.

Lyons, C. 1999. Definiteness. Cambridge: Cambridge UP.

Lyons, J. 1995. Linguistic Semantics, An Introduction. Cambridge: Cambridge UP.

Meyer, M. 2001. Between Theory, Method and Politics: Positioning of the Approaches to CDA. In R. Wodak and M. Meyer (eds.), Methods of Critical Discourse Analysis (1st edition). London: Sage, 14-31.

Saeed, J. I. 1997. Semantics (1st edition). Oxford: Blackwell.

Scollon, S. 2003. Political and Somatic Alignment: Habitus, Ideology and Social Practice. In R. Wodak and G. Weiss (eds.), Critical Discourse Analysis: Theory and Interdisciplinarity. New York: Palgrave Macmillan, 167-198.

Wodak, R. 2001a. What Critical Discourse Analysis Is About- A Summary of Its History, Important Concepts and Its Developments. In R. Wodak and M. Meyer (eds.), Methods of Critical Discourse Analysis (1st edition). London: Sage, 1-13.

Discourse Analysis - Gender - Lexicography 1 6 5 Major Trends in Theoretical and Applied Linguistics

Wodak, R. 2001b. The Discourse-Historical Approach. In R. Wodak and M. Meyer (eds.), Methods of Critical Discourse Analysis (1st edition). London: Sage, 63-94.

Woolbert, C.H. 1923. The place of logic in a system of persuasion. The Quarterly Journal of Speech Education, 9(4): 19-39.

1 6 6 Discourse Analysis - Gender - Lexicography Section 4: Language Acquisition

1 6 7 Major Trends in Theoretical and Applied Linguistics

1 6 8 Language Acquisition “Automatically Arises the Question Whether…”: A Corpus Study of Postverbal Subjects in L2 English1

Eleni Agathopoulou

Aristotle University of Thessaloniki [email protected]

Abstract

The current study offers a novel set of data on phenomena regarding SV/VS in L2 English. We explored structures with unergative and unaccusative predicates in the corpus of Greek learners of English (GRICLE) and in comparable native English corpora. Results showed that for both the learners and the natives word order is conditioned by properties (a) of the lexicon-syntax interface, i.e. VS appears only with unaccusatives and never with unergatives and (b) of the syntax-discourse interface, i.e. postverbal S in unaccusatives expresses focus rather than topic. Moreover word order is conditioned by properties of the syntax-phonology interface, in that postverbal subjects tend to be phonologically heavy, albeit only in the learner data. Our results almost replicate results in previous corpus-based research.

1. Introduction

In generative second language (L2) research, subject (S) - verb (V) order has been considered mainly as one of the properties of the NullS(ubject) Parameter (e.g. White 1985, 1986; Liceras 1989; Tsimpli & Roussou 1991). This parameter, formulated by Rizzi (1982, 1986) in the general spirit of Chomsky’s (1981) theory, suggested that one of the differences between non-NullS languages like English and NullS languages like Italian is that the former generally disallow postverbal S, while the latter allow both SV and VS order. Results from the above mentioned L2 studies have shown that L1 speakers of NullS languages generally

1 I would like to thank Maria Dimitrakopoulou for her insightful comments on an earlier version of this paper.

Language Acquisition 1 6 9 Major Trends in Theoretical and Applied Linguistics

disprefer VS order in English. Moreover, corpus-based studies (Rutherford 1989; Zobl 1989; Oshita 2000, 2004; Lozano & Mendikoetxea 2010) have revealed that L1 speakers of NullS languages such as Italian, Spanish, Arabic and Japanese use VS order in English almost exclusively with certain intransitive verbs like the ones in the following examples.

(1) In the town lived a small Indian… (L1 Spanish, Rutherford 1989: 178) (2) … . . . there exist two kinds of jobs. (L1 Italian, Oshita 2000: 315)

These verbs belong to a class of intransitive verbs called ‘unaccusative’ and differ from the other class of intransitive verbs, called ‘unergative’, on the basis of semantics and syntax. In unergatives, e.g. walk, laugh, S has the semantic role Agent and originates in the Specifier of VP (3); in unaccusatives, e.g. arrive, occur, S has the semantic role Theme (or Patient) and originates inside the VP in a V-complement position (4) (Perlmutter 1978, Levin & Rappaport-Hovav 1995, among others).

(3) [TP [DP Mary]i [VP ti [V΄ [V walked]]]]]

(4) [TP [DP Mary]i [VP [V΄ [V arrived] [ti]]]]]

In English the default SV order that surfaces both with unaccusative and unergative verbs is due to that S has to move to the specifier of Tense Phrase (TP) to satisfy the language-specific requirement for an overt element in this position and to get case. In null S languages the overt postverbal S originates within the VP and gets case through an Agree relation with a null expletive pronoun (pro) assumed to occupy the specifier of TP (see, e.g., Roussou & Tsimpli 2006: 319 and references therein; also see Section 2). Importantly, VS order appears in English only with some unaccusatives which express existence or appearance, e.g. come, appear, exist, arrive and live. In this case the specifier of TP is filled by elements such as an expletive there (2), a Prepositional Phrase (PP) in locative inversion constructions (5), an adverb or another XP (see Biber et al. 1999: 912-3; also see section 4.2 for examples). Moreover, word order with unaccusatives is conditioned by discourse factors, such as the principle of ‘end focus’ (e.g. Leech & Short 2007, Chapter 7). Postverbal subjects tend to be focus, that is, they express new information (5) while preverbal subjects tend to be topic, that is, they refer to old information (6).

(5) In it, lives a family. (6) This family lives in our neighbourhood.

Another factor conditioning speakers’ choice of word order has been accounted for by the ‘end weight’ principle (Quirk & Greenbaum 1977: 410-411), which states that that structurally complex constituents tend to appear at the end of

1 7 0 Language Acquisition Eleni Agathopoulou

an utterance for processing reasons. This principle can be illustrated by what is called ‘heavy Noun Phrase (NP) shift’. As shown by the following examples (from Hawkins 2004: 26), shifting a complex direct object NP before its simple complement PP (7) renders the sentence easier to parse, rather than keeping this complex object in its original position (8).

(7) Mary VP[gave PP[to Bill] NP[the book she had been searching for since last Christmas]] (8) Mary VP[gave NP[the book she had been searching for since last Christmas] PP[to Bill]]

Note that structural complexity implies phonological heaviness: complex constituents generally consist of more syllables or words than simple constituents and, therefore, the former can be considered ‘heavy’ and the latter ‘light’. (See Section 4.1 about the criteria for such a distinction.) The ‘end weight’ principle may affect word order also with unaccusatives: postverbal subjects tend to be more complex than preverbal ones (see Lozano & Mendikoetxea 2010 and references therein), as can be shown by (9) and (10) respectively.

(9) Then there arrived a man who knew of a proven remedy for the illness of the prince. (10) A tall man arrived before noon.

Given the above, SV/VS order in English pertains to three linguistic interfaces: the lexicon-syntax interface (unergative vs unaccusative verbs), the syntax- discourse interface (focus vs. topic S) and the syntax-phonology interface (heavy vs light S). The aim of the present paper is to investigate whether these three interface conditions restrict word order in the Greek-English interlanguage by partially replicating Lozano and Mendikoetxea’s (2008, 2010) research, detailed in Section 3. We also attempt to trace L1 effect comparing our data with those in the latter mentioned study.

2. Subject Position in Greek

Greek is a NullS language allowing VS order both with unaccusative and unergative verbs regardless of whether S is topic or focus as illustrated by (11) and (12).

(11) Eftasan kapji fitites (focus) / i fitites (topic) unaccusative arrived some students / the students (12) Edho dulevun kapji fitites (focus) / i fitites (topic) unergative here work some students / the students

Language Acquisition 1 7 1 Major Trends in Theoretical and Applied Linguistics

However, VS seems to be favoured with unaccusatives and SV with unergatives. This can be shown by comparing word order between and within the two verb classes in answers to a wide-focus question such as “What happened?” (13 vs. 14, cf. Lozano 2006: 373-74), where # indicates a pragmatically less felicitous word order (for references see, e.g. Lozano ibid.).

(13) A: Ti ejine? What happened? B: (i) #Ena pedhi irthe (#SV) unaccusative A boy arrived. (ii) Irthe ena pedhi (VS) (14) A: Ti ejine? What happened? B: (i) Ena pedhi eklapse (SV) unergative A boy shouted. (ii) #Eklapse ena pedhi (#VS)

The underlying difference in subject position between unaccusatives and unergatives in Greek is illustrated by (15) and (16) respectively.

(15) [TP proi [VP [V’] [V irthe]] [ena pedhii]]]]]

(16) [TP [DP Ena pedhi] i [V eklapse] k] [VP [V’ ti] [V tk]]]]]

In the unaccusative VS structure the specifier of TP is assumed to be occupied by an expletive null pro that licenses the postverbal S and shares with it agreement and case features. This is the structure assumed also for NullS languages such as Spanish and Italian (Rizzi 1986). Moreover word order in unaccusatives may be affected by discourse factors in that a topic S increases the probability for SV order, as shown by the comparison between (17) and (18) (from Tsimpli et al. 2004: 88).

(17) Htes vradhi to moro ekleje2 Last night the baby was crying. (18) Htes vradhi ekleje ena moro Last night a baby was crying.

2 Tsimpli et al. (2004: 88) state that “indefinite DPs rarely make good topics”. It could then be assumed that the topic/focus distinction roughly corresponds with the definite/ indefinite distinction. Others, however, support that VS orders in Greek do not display any definiteness restrictions unlike their counterparts in English, Icelandic, Dutch or French” (Alexiadou and Anagnostopoulou 1999: 96).

1 7 2 Language Acquisition Eleni Agathopoulou

Last, research has shown that the principle of ‘end weight’ applies in Greek too since phonologically heavy S usually occurs postverbally (Laskaratou 1989; Spyropoulos & Revythiadou 2009).

3. Interfaces of the Unergative/unaccusative Split in L2 English

Given the scarcity of research in the area described by the section title, this review concerns only three studies, of which the first two were carried by Lozano and Mendikoetxea (hereafter L&M) in 2008 and 2010. L&M’s studies resemble each other with respect to aims, method and results (detailed below) but differ regarding the following that concern us here. Only the 2008 study employed L2 English data from learners of two different L1s and only the 2010 study included data from native English speakers too. For purposes of brevity the following description will not distinguish between the two studies, unless necessary. As mentioned in Section 1, L&M investigated whether SV/VS order in L2 English is restricted by all three linguistic interfaces that restrict L1 English SV/VS order: the lexicon-syntax interface (unergative vs. unaccusative verbs), the syntax- discourse interface (focus vs. topic S) and the syntax-phonology interface (heavy vs. light S). To this aim they examined written corpora produced by Spanish and Italian university students with upper intermediate to advanced English language proficiency. These corpora consisted of argumentative essays written according to the guidelines provided by the International Corpus of Learner English (ICLE, Granger et al. 2002; also see http://www.uclouvain.be/en-317607.htm) and totalled more than 200,000 words each. L&M also analysed comparable native English corpora. Results showed that learners, like English natives, produced VS only with unaccusatives and mainly so when S was focus (new information) and heavy (long). However learners produced VS with unaccusatives significantly more than the natives and made syntactic errors. In both learner corpora these errors were due mostly to insertion of it expletive S (19) (38 occurrences in the Spanish corpus and 27 occurrences in the Italian corpus (L&M 2008: 107)), while the rest of the errors consisted in wrongful omission of there expletive (20).

(19) *In the name of religion it had occurred many important events. (20) *It is difficult that Ø exist volunteers with such a feeling against it.

Based on their results, L&M concluded that for both natives and learners SV/ VS order in English is constrained by properties of all three of the discussed interfaces but that learners differed from the natives with respect to “the (un) grammaticality of the outputs of syntactic encoding” (L&M 2010: 494).

Language Acquisition 1 7 3 Major Trends in Theoretical and Applied Linguistics

In addition, the two learner groups differed from each other in that the Spanish produced VS significantly more than the Italians (8.1% and 2.6% respectively). L&M attributed these results to two factors. First, the verbs exist and appear were more frequent in the Spanish than in the Italian corpus and there was a “significant correlation between the frequency of concordances and the frequency of inversions in both corpora” (L&M 2008: 115). Second, the Spanish learners’ English proficiency level was assumingly lower than that of the Italian’s, as attested also by the higher percentage of ungrammatical VS sentences in the Spanish than in the Italian corpus (65 and 53.3 respectively) 3. The above seem to offer a plausible explanation for the differences between the two learner groups. Note that L1 effect is rejected by L&M (ibid) “given the similarities between Italian and Spanish” with respect to word order. However, other research has revealed that there are fewer restrictions on VS(O) order in Spanish than in Italian (see Roussou & Tsimpli 2006: 321 and references there). Therefore, L1 effect could be one of the reasons for the discrepancies observed between the two learner corpora in L&M’s study. Danavassi (2009) investigated the lexicon-syntax and the syntax-phonology interface in unergatives and unaccusatives via a grammaticality judgement task administered to adult Greek learners of English and native English controls. The learners were divided into an intermediate and an advanced group according to their performance at the Oxford Quick Placement test (UCLES 2001). Results showed that the learners resembled the English natives in that (a) they accepted VS more with unaccusatives than with unergatives and (b) they accepted VS more with unaccusatives when S was heavy rather than light. However, while the within-group difference in the acceptance between heavy and light postverbal S was significant for the natives and the advanced group, it was only near- significant (p=0.07) for the intermediate group. Additionally, the within-group difference in the acceptance between VS with unaccusatives and unergatives was significant for the natives and the intermediate learners but not forthe advanced ones. Danavassi attempted to account for her results based on Sorace (2000) who has classified intransitive verbs into a continuum from core unaccusatives to core unergatives, depending on whether they take ‘be’ or ‘have’ as their auxiliaries (in e.g. Italian). Thus verbs denoting change of location (e.g. arrive, come) are categorized as core unaccusatives while verbs meaning existence

3 For the basis of this assumption see reference in (L&M 2008: 115). However, as pointed out by the researchers the specific learner corpora do “not provide an independent measure of proficiency for each participant” (L&M: ibid).

1 7 4 Language Acquisition Eleni Agathopoulou

or state (e.g. exist, lie) are considered peripheral unaccusatives. Verbs related with controlled process (e.g. work, speak) are core unergatives and those implying uncontrolled process (e.g. sleep, cry) are peripheral unergatives. Danavassi suggested that her results may be due to that the acceptability judgement task included mostly ‘core’ unergatives but both ‘core’ and ‘peripheral’ unaccusatives and that the advanced learners were more able than the intermediate learners to “treat the most peripheral unaccusative verbs as being less unaccusative and more unergatives”. However, if that were the case, one wonders why the same factor, i.e. coreness or peripherality of verbs in either the unaccusative or the unergatives category did not affect the native speakers’ use of word order. Also, Danavassi speculated that lack of significant within-group difference between heavy and light postverbal S in the intermediate group is due to that these learners had not had enough exposure to L2 input. Recall, however, that the “end-weight” principle is assumed to be a universal one (see references in Callies 2009: 17) and has been shown to apply in Greek too (see Section 2). Therefore Danavassi’s speculation seems unwarranted. Perhaps, it could be the case that syntax (i.e. word order) overrides universal phonetic principles at less advanced stages of learning L2.

4. The Present Study

Our research question was whether in the Greek-English interlanguage SV/ VS word order with intransitive verbs is conditioned by the three interfaces mentioned above. In view of results from previous L2 research discussed in section 3, we hypothesized that the Greek learners will resemble the English natives in that they will use

• VS with unaccusatives but not with unergatives (Hypothesis 1, regarding the lexicon-syntax interface) • VS with unaccusatives when S is focus rather than topic (Hypothesis 2, regarding the syntax-discourse interface) and • VS with unaccusatives when S is heavy rather than light (Hypothesis 3, regarding the syntax-phonology interface).

4.1. Data

The non-native English data used here came from The Greek Corpus of Learner English (GRICLE, Hatzitheodorou & Mattheoudakis 2009) and totaled 233,570 words. GRICLE was compiled according to the ICLE guidelines (see Section 3) at

Language Acquisition 1 7 5 Major Trends in Theoretical and Applied Linguistics

the Department of English Language and Literature of the Aristotle University in Greece. The essay authors of GRICLE (230 students, 2 essays each) were assumed to be advanced learners of English by virtue of the fact that they were third or fourth year students of the above mentioned Department. The native English data were from essays written by American students in two corpora, LOCNESS (Louvain Corpus of Native English Essays) and PELCRA ((Polish and English Language Corpora for Research and Applications) and amounted to 175,047 words altogether (for more details on the native English data see e.g. Hatzitheodorou & Mattheoudakis 2011). The target verbs were sixteen in total, eight unaccusative and eight unergatives (Table X). These verbs are shown in Table 1.

Table 1. Target verbs in the present study.

Unaccusative verbs Unergative verbs

Core Peripheral Core Peripheral

appear, come, arise, begin, exist, laugh, run, speak, cry, breathe emerge lie, occur talk, walk, work

Using the concordance software MonoConc Pro2 (Barlow 2003), first we identified the target verbs and then we carried out a manual search to examine the contexts in which these verbs appeared. This search included also possible misspellings (e.g. ‘occured’, ‘ocur(s) ‘ocurs’) or wrong verb forms (e.g. ‘arouse’ or ‘aroused’). We counted out all cases of the target verbs that were transitive (e.g. They began the revolution), passive (e.g. *There have been appeared many problems), non-finite (e.g. *It is essential to exist room for hope), followed by a non-finite clause (e.g. … begin to improve) or in interrogative clauses. We also excluded all of the target verbs that occurred in fixed or highly frequent expressions than may have been memorized by the learners as chunks (e.g. the time has come / dreams come true / come home / come in contact / come to a standstill). After this filtering, of the target tokens, the native English corpora yielded 163 unergatives and 289 unaccusatives and the learner corpus 139 unergatives and 372 unaccusatives. To identify the information status of S used with the target verbs, that is, whether S was topic or focus, we read through each of the essays in which these verbs appeared. Last, we considered as phonologically light subjects that consisted of a pronoun, a noun, a determiner+noun or a determiner + adjective + noun. The rest or the subjects that had a larger number of words and more complex structure were categorized as phonologically heavy (cf. Lozano & Mendikoetkea 2010: 485). For examples, see the next section.

1 7 6 Language Acquisition Eleni Agathopoulou

4.2. Results

First we present results from SV vs. VS order in the unergative and unaccusative predicates dealt with here. As Table 2 reveals, in both the native speaker (NS) and the non-native speaker (NNS) corpora, (a) the predominant order was SV and (b) VS appeared only with unaccusatives. As expected, the differences in word order between the two verb types was statistically significant for both the NS and the NNS data (χ2 (1,451) = 6.953, p <.01 and χ2 (1,510) = 13,610, p <.01respectively). However, the NNS employed VS with unaccusatives more than twice as much as the NS (9.13% and 4.15%, respectively), which proved to be a statistically significant between-group differenceχ ( 2 (1,660) = 6,249, p=.012).

Table 2. VS at the lexicon-syntax interface

SV VS

Unergatives Unaccusatives Unergatives Unaccusatives 100% 95.85% NS 0% (0/163) 4.15% (12/289) (163/163) (277/289) 100% 90.87% NNS 0% (0/139) 9.13% (34/372) (139/139) (338/372)

Regarding types of VS structure, in the NNS data the most common one (19/32) was that of There-insertion (21). The next more common structures included AdvP-insertion (23) or another XP-insertion (24), while locative inversion occurred only once (26). Ø-insertion structure (27) occurred twice and the ungrammatical It-insertion once (28) In the NS data, on the other hand, there was only one instance of There-insertion (22) with the rest of the structures including XP-insertion (25)4.

There-insertion5 (21) There appeared the first feminists who… NNS (22) … there existed social ills… NS

AdvP-insertion (23) …and thus begun the phenomenon…

4 All examples are presented as they appeared in the corpora. 5 Examples (21) and (22) illustrate the distinction between ‘presentational’ there and ‘existential’ there (respectively). We return to this distinction in Section 5.

Language Acquisition 1 7 7 Major Trends in Theoretical and Applied Linguistics

XP-insertion (24) … due to this alienation came to the surface all … NNS (25) With the rioting came a surge of articles… NS

Locative inversion (26) …in certain channels appear fashion designers… NNS

Ø-insertion (27) In contrast to the advantages that a university student has, Ø comes unemployment. NNS

It-insertion (28) It could exist serious problems in everyday life… NNS

With respect to errors in VS structures, perhaps the only clearly ungrammatical construction was the one with It-insertion (28). To verify this, we asked six highly educated native English speakers to judge the acceptability of all sentences with VS in our data. However, these judgements did not enlighten us much. Besides the sentence with It-insertion, which was unanimously disliked, regarding the rest of the sentences there was great variation between the judges and their remarks revealed that judgements were affected mainly by matters of style. Many of these remarks also showed indecision and were of the kind “I don’t like it much but people say it” or “May be it would be OK in a different context”. Let it be pointed out that, probably for the same reasons, the judges also disliked some VS structures produced by the NS. Table 3 presents token numbers as well as percentage frequencies of VS for each of the target unaccusative verbs in the corpora. The shadowed rows include data about the core unaccusatives.

Table 3. Token numbers and frequency of VS with unaccusatives

NS NNS

appear 0% (0/23) 3,63% (2/55)

arise 0% (0/13) 9.09% (1/11)

begin 0% (0/69) 2.98% (2/67)

come 8.03% (11/137) 9.56% (11/115)

emerge - 16.6% (2/12)

exist 4.34% (1/23) 14. 6% (13/89)

lie 0% (0/8) 18.75% (3/16)

occur 0% (0/43) -

1 7 8 Language Acquisition Eleni Agathopoulou

In all corpora come was the most frequent unaccusative. In the NS data, (a) VS appeared almost exclusively with come – except once with exist too, (b) the next most frequent unaccusatives were begin, occur, appear and exist in this order and (c) there were no tokens of emerge. In the NNS data, (a) VS appeared with all of the target verbs used but mostly with lie, (b) the next most frequent unaccusatives after come were exist, begin and appear in this order and (c) there were no tokens of emerge. Last, word order does not seem to be affected by whether a verb is a ‘core’ or a ‘peripheral’ unaccusative in the NNS data, while this may not hold for the NNS data, where VS was used almost only with the core unaccusative verb come. Table 4 displays SV vs. VS order with unaccusatives at the syntax-discourse interface. In all corpora, preverbal S is slightly more often topic rather than focus but it is predominantly focus when in postverbal position (100% and 93.75% for the NS and the NNS respectively). The statistical analysis showed that for both the NS and the NNS the within group difference between S Focus and S Topic in VS was statistically significant (NS: χ2 (1,288) = 14,085, p <.01; NNS: χ2 (1,371) = 32,887, p <.01), and that there were no significant between-group differences in this respect. Also no significant within- or between-group differences were found regarding S Focus and S Topic in SV structures.

Table 4. SV vs. VS with unaccusatives at the syntax-discourse interface

SV VS

S Focus S Topic S Focus S Topic English 44.76% 55.24% 100% 0% Natives (124/277) (153/277) (12/12) (0/12) 46.6% 57.4% 93,75% 6.25% Greeks (144/338) (194/338) (32/34) (2/34)

Here are some examples from VS structures.

S topic (29) So it is true that there lies the battle of the two sexes… NNS

S focus (30) With a good football team comes free publicity. NS (31) There exists some prejudice in the workplace… NNS

Table 5 provides results regarding the syntax-phonology interface. As shown, in SV order light S was about twice as much as heavy S in both the NS and the NNS corpora. On the other hand, in VS order the NS used light and heavy S exactly to

Language Acquisition 1 7 9 Major Trends in Theoretical and Applied Linguistics

the same degree (50% in each case) while the NNS used predominantly heavy S (82.35%). The statistical analysis showed that the heaviness of the subject had main effects for the NNS χ( 2 (1,371) = 31,882, p <.01) and that in the VS condition the between-group differences were statistically significantχ ( 2 (1,45) = 4,815, p <.05).

Table 5. SV vs. VS with unaccusatives at the syntax-phonology interface

SV VS

S Light S Heavy S Light S Heavy 65.34% 34.66% 50% 50% English Natives (181/277) (96/277) (6/12) (6/12) 66.86%/ 33.14% 16,65% 82,35% Greeks (226/338) (112/338) (6/34) (28/34)

Here are some examples from VS structures with light and heavy S.

S light (32) Along with the traffic congestion comes pollution. NS (33) There may exist different religions but… NNS

S heavy (34) Along with respected sources come well thought studies that have been completed. NS (35) …. there emerged a necessity to sustain and perpetuate the fear of the individual. NNS

5. Discussion

First, results showed that in all corpora (a) VS occurred only with unaccusatives and never with unergatives and that (b) in VS with unaccusatives, S was predominantly focus, not topic. These results seem to verify our first two hypotheses, namely that in both the NS and the NNS data SV/VS order would be conditioned by properties of the lexicon-syntax interface (Hypothesis 1), and by properties of the syntax-discourse interface (Hypothesis 2). On the other hand, effect of phonological heaviness of S on word order was found only in the NNS corpora, where in VS structures S was predominantly heavy. Thus, our Hypothesis 3, according to which SV/VS order would also be conditioned by properties of the syntax-phonology interface, is not fully confirmed or even disconfirmed. Our results probably indicate that the “end-weight” principle, linked in this case with

1 8 0 Language Acquisition Eleni Agathopoulou

phonological properties of S, may not override (a) the clear bias for SV order in English, hence the NS results and (b) the clear bias for VS in Greek, which is manifested as L1 effect in the NNS data. This interpretation is of course a tentative one, given also that our findings regarding (lack of) phonological effect in the NS data disagree with those in the studies by Lozano and Mendikoetxea (2008, 2010) and by Danavassi (2009), discussed in Section 3. The difference in findings between these studies and ours may be attributed to difference in methodology, given that Danavassi had employed grammaticality judgements and that L&M’s corpus search included 31 types of unaccusatives while our search included 8 types only. Although L&M’s findings seem to have a more solid base than ours, let us mention that when we further searched our NS corpora for all of the 31 types of unaccusatives dealt with in L&M, we did not discover any more VS constructions than those we had already found with the 8 types of unaccusatives in our initial search. Thus, all in all, whether word order in intransitives is conditioned by properties of the syntax-phonology interface seems an issue worthy of further investigation. Our results also showed that the NNS used VS significantly more than the NS, which points to L1 effect. Recall that a similar effect was attested in L&M’s L2 English corpora with essays from L1 speakers of NullS languages (Italian and Spanish). However, a comparison between our data and L&M’s revealed that VS with English unaccusatives was used the most by the Spanish (52/640=8.1%) and the least by the Italians (15/574=2.6%), while the Greeks were somewhere in-between (34/372=4.23%). Moreover, we found a statistically significant difference between the Spanish and the Italian dataχ ( 2 (1,1213) = 17.630, p <.01) as well as between the Italian and the Greek data (χ2 (1,945) 19.576, p <.01) but not between the Spanish and the Greek data. To account for the differences between the learner data, we could comply with L&M’s assumption that the Italians were advanced while the Spanish were intermediate learners of English and also assume that the Greek learners’ average level of English proficiency was in-between, that is, upper intermediate (contra Hatzitheodorou and Mattheoudakis 2009, 2011). Nevertheless, this is merely a speculation since there was no independent measure for the learners’ level of proficiency in English in any of the corpora. A final point worthy of discussion is why (ungrammatical) It-insertion in VS constructions prevailed in the Spanish and the Italian corpora (see Section 3) while it occurred only once in the Greek learner corpus. Omitting details, Cardinaletti (1997) has proposed that languages have two subject positions, one hosted by a higher Agreement Phrase and the other by a lower one (Agr1P and Agr2P respectively), depending on the referentiality of the subject. In NullS languages like Italian, referential subjects, such as full DPs and overt strong pronouns (e.g. lui ‘he’) as well as overt weak subject pronouns (e.g. egli ‘he’) occupy the Specifier of Agr1P, while the Specifier of Agr2 hosts pro and a weak

Language Acquisition 1 8 1 Major Trends in Theoretical and Applied Linguistics

tu pronoun used in subjunctives. In non-NullS languages like English only strong subjects occur in the highest specifier position whereas non-referential (i.e. expletive) it occurs at the Specifier of Agr2P. Based on Cardinaletti’s proposal we may assume that the error of It-insertion in the Italian corpus is due to that Italian has more cases of phonological realisation of a pronoun in subject position than Greek (which has no weak subject pronouns) and it is this option that is used in the English-Italian interlanguage, in the form of the expletive it. Regarding the same error in the Spanish corpus, on the assumption that the Spanish learners had an intermediate level of English language proficiency, L&M (2010: 493) offer the following explanation. While learners of English are exposed to many instances of existential there (e.g. There are two chairs …) from early on, presentational there (e.g. There appeared…) is not as frequent in the input. As a result, learners learn ‘there + be’ constructions as formulaic units, which may be attested by that learners also overuse such constructions (see references therein). Consequently these learners may not have yet acquired the use of there as an independent expletive and thus use it interchangeably with the expletive it. We agree with L&M (ibid.) that the latter account they offer is “highly speculative”. However, as we are yet unable to provide a less speculative hypothesis, we may concede that the wider occurrence of It-insertion errors in the Spanish than in the Greek corpus could be due to that the Spanish learners had had less exposure to English than the Greek learners (but see relevant remark above in this section).

6. Conclusion

Our results indicate that adult L2 acquisition cannot be driven only by input, given that in English VS is rare, even with unaccusatives. Thus an input-based approach to L2 could not explain the strong discrimination between unaccusatives and unergatives in English attested in data from L1 speakers of a NullS language like Greek where VS is the unmarked option. Rather, adult L2 acquisition seems to be guided by knowledge concerning differences in syntactic and semantic properties between the two classes of intransitive predicates. Given that these differences are not generally involved in L2 instruction (see also L&M),the nature of L2 knowledge discussed here is more plausible driven by Universal Grammar principles. Our study largely replicated the Italian and Spanish data in L&M, offering further support to L&M’s assumption that VS in interlanguage is conditioned not only by properties of the syntax-lexicon interface, but also of the syntax-discourse and the syntax-phonology interface. With respect to the two latter interfaces, however, it is possible that our learner data may reflect L1 effect and not inner knowledge of universal principles related with discourse or phonology.

1 8 2 Language Acquisition Eleni Agathopoulou

References

Alexiadou, A. and E. Anagnostopoulou. 1999. EPP without Spec,IP. In D. Adger, S. Pintzuk, B. Plunkette and G. Tsoulas (eds), Specifiers: Minimalist Approaches. Oxford: Oxford University Press, 93-109.

Barlow, M. 2003.Concordancing and Corpus Analysis Using MP2.2. Houston: Athelstan.

Biber, D., S. Johansson, G. Leech, S. Conrad and E. Finnegan. 1999. Longman Grammar of Spoken and Written English. Harlow: Pearson.

Callies, M. 2009. Information Highlighting in Advanced Learner English. Amsterdam/ Philadelphia: John Benjamins.

Cardinaletti, A. 1997. Subjects and Clause Structure. In L. Haegeman (ed.), The New Comparative Syntax. London, N.Y.: Longman.

Chomsky, N. 1981. Lectures on Government and Binding. Dordrecht: Foris, 33-63.

Danavassi, T. 2009. The Status of the Unaccusative-Unergative Split in the Greek- English Interlanguage. Upublished MA thesis, Aristotle University of Thessaloniki, Greece. Retrieved 2 July 2012 from http://invenio.lib.auth.gr/record/114660/ files/danavasi.pdf?version=1

Granger, S., E. Dagneaux and F. Meunier. 2002. The International Corpus of Learner English. Louvain-la-Neuve: Presses Universitaires de Louvain.

Hatzitheodorou, A.-M. and M. Mattheoudakis. 2009. “It is more than true that television reproduces life”: The effect of Greek rhetorical conventions on Greek learners’ academic writing in English. In A. Tsagalidis (ed.), Selected Papers from the 18th International Symposium on Theoretical and Applied Linguistics. Thessaloniki, Greece, 167-176.

Language Acquisition 1 8 3 Major Trends in Theoretical and Applied Linguistics

Hatzitheodorou, A.-M. and M. Mattheoudakis. 2011. The impact of culture on the use of stance exponents as persuasive devices: the case of GRICLE and English native speaker corpora. In A. Frankenberg-Garcia, L. Flowerdew and G. Aston (eds), New Trends in Corpora and Language Learning. Continuum Press, 229-246.

Hawkins, J.A. 2004. Efficiency and Complexity in Grammars. New York: Oxford University Press.

Laskaratou, C. 1989. A Functional Approach to Constituent Order with Particular Reference to Modern Greek. Implications for Language Learning and Language Teaching. Athens: Parousia Monograph Series 5.

Leech, G. and M. Short. 2007. Style in Fiction: A Linguistic Introduction to English Fictional Prose, 2nd Edition. Harlow: Pearson/Longman.

Levin, B. and M. Rappaport Hovav. 1995. Unaccusativity at the Syntax–Lexical Semantics Interface. Cambridge, MA: MIT Press.

Liceras, J.M. 1989. On some properties of the pro-drop parameter: looking for missing subjects in non-native Spanish. In S.M. Gass and J. Schachter (eds), Linguistic Perspectives on Second Language Acquisition. Cambridge: Cambridge University Press.

Lozano, C. 2006. The development of the syntax-discourse interface: Greek learners of Spanish. In V. Torrens and L. Escobar (eds), The Acquisition of Syntax in Romance Languages. Amsterdam; Philadelphia, PA: John Benjamins, 371-399.

Lozano, C. and A. Mendikoetxea. 2008. Postverbal subjects at the interfaces in Spanish and Italian learners of L2 English: a corpus analysis. In G. Gilquin, S. Papp and M.B. Díez-Bedmar (eds), Linking up contrastive and corpus learner research. Amsterdam: Rodopi, 85-125.

Lozano, C. and A. Mendikoetxea. 2010. Interface conditions on postverbal subjects: a corpus study of L2 English. Bilingualism: Language and Cognition 13(4): 475-497.

Oshita, H. 2000. What is happened may not be what appears to be happening: A corpus study of ‘passive’ unaccusatives in L2 English. Second Language Research 16(4): 293-324.

Oshita, H. 2004. Is there anything there when there is not there? Null expletives and second language data. Second Language Research 20(2): 95-130.

1 8 4 Language Acquisition Eleni Agathopoulou

Perlmutter, D.M. 1978. Impersonal passives and the Unaccusative Hypothesis. Berkeley Linguistics Society 4: 157-189.

Quirk, R. and S. Greenbaum. 1977. A University Grammar of English. London: Longman.

Rizzi, L. 1982. Issues in Italian Syntax. Dordrecht: Foris.

Rizzi, L. 1986. Null objects in Italian and the theory of pro. Linguistic Inquiry 17(3): 501-58.

Roussou, A. and I.-M. Tsimpli. 2006. On Greek VSO again! Journal of Linguistics 42(2): 317-354.

Rutherford, W. 1989. Interlanguage and pragmatic word order. In S.M. Gass and J. Schachter (eds), Linguistic Perspectives on Second Language Acquisition. Cambridge: Cambridge University Press, 163-182.

Sorace, A. 2000. Gradients in auxiliary selection with intransitive verbs. Language 76(4): 859-890.

Spyropoulos, V. and A. Revithiadou. 2009. Subject chains in Greek and PF processing. In C.D. Halpert, J. Hartman and D. Hill (eds), Proceedings of the 2007 Workshop in Greek Syntax and Semantics at MIT. Cambridge, MA, 293-309.

Tsimpli I.-M. and A. Roussou. 1991. Parameter resetting in L2? University College London Working Papers in Linguistics 3: 148-69.

Tsimpli I.-M., A. Sorace, C. Heycock and F. Filiaci. 2004. First language attrition and syntactic subjects: a study of Greek and Italian near-native speakers of English. International Journal of Bilingualism 8: 257-277.

UCLES (2001). Quick placement test (Paper and pencil version). Oxford: Oxford University Press.

White, L. 1985. The pro-drop parameter in adult second language acquisition. Language Learning 35: 47-62.

White, L. 1986. Implications of parametric variation for adult second language acquisition: An investigation of the pro-drop parameter. In V. Cook (ed.), Experimental approaches to second language acquisition. Oxford: Pergamon Press, 55-72.

Language Acquisition 1 8 5 Major Trends in Theoretical and Applied Linguistics

Zobl, H. 1989. Canonical typological structures and ergativity in English L2 acquisition. In S.M. Gass and J. Schachter (eds), Linguistic Perspectives on Second Language Acquisition. Cambridge: Cambridge University Press, 203-221.

1 8 6 Language Acquisition Multilingualism in Greece: A Study of Speech Production of Trilingual Children

Fotini Anastassiou and Georgia Andreou

University of Thessaly [email protected], [email protected]

Abstract

During the last twenty years Albanian immigrants have come to Greece and they have formed a large community in different parts of our country. Their children now are of the second generation and they go to Greek state schools, being taught according to the official syllabus in the same classes as Greek students. We will present a study on the vocabulary interaction amongst the three languages of fifty trilingual children. Our hypothesis is that trilingual children will seek help from their L1 (either Greek or Albanian) while speaking their L3 (English). Furthermore, we will investigate cross- linguistic influence: a) Interactional strategies and b) Transfer lapses as previously described by Cenoz (2001). Since multilinguals have the ability to activate several languages at the same time, cross-linguistic influence in speech production has been a popular field within the studies on multilingual lexicon, since according to Cenoz (2001) the way a person’s languages interact with each other can help us comprehend the multilingual and the mental lexicon.

1. Introduction

Many attempts have been made so far to approach a definition of trilingualism. Most researchers have attempted to define trilingualism through bilingualism and trilingualism has sometimes been regarded as a branch of bilingualism or even as an extension of bilingualism. Haugen refers to multilingualism as “a kind of multiple bilingualism” (1956: 9). Oksaar defines bilingualism as “the ability of a person to use here and now two or more languages as a means of communication in most situations and to switch from one language to the other if necessary” (1983: 19), whereas Skutnabb-Kangas (1984) suggests that more than two languages can be present in the person she defines as bilingual.

Language Acquisition 1 8 7 Major Trends in Theoretical and Applied Linguistics

Cenoz and Genesee (1998) suggest that a student should be defined as trilingual if s/he can use her/his three languages to communicate in both oral and written speech. Furthermore, they describe multilingualism as the final result of the process of acquisition of several non- native languages. Moreover, since many scholars have proposed that learning an L3 is a different process than learning an L2, Hufeisen (1998) suggested that multilingualism should be used to refer to the learning of more than two languages. According to Herdina and Jessner (2000), multilingualism is to be considered as a varied phenomenon involving bilingualism and monolingualism as possible forms, but addressing mainly those languages learned after a second one, so, therefore, bilingualism cannot be identified with multilingualism.

2. Speech Production and Trilinguals

Lambert (1990) gives us his working hypothesis on the speech production of trilinguals. “Bilingualism provides a person with a comparative, three-dimensional insight into language, a type of stereolinguistic optic on communication that the monolingual rarely experiences” (1990: 212). We could assume that if a bilingual possesses the qualities that Lambert (1990) hypothesizes then a trilingual would possess the same and even more. Lambert (1990) goes on to propose that bilinguals have a certain ability to switch codes when referring to speakers of either language (L1 or L2) or even when communicating with bilingual speakers that speak the same languages, especially in diglossic speech communities. If that is the case with bilinguals, we could then hypothesize that a consecutive trilingual who went through a bilingual period and then proceeded to the trilingual period has built up his/her language facilities and can switch codes in a more rapid and effective way. Jessner (1999) also points to the advantage of bilinguals over monolinguals regarding their interactional competence, that is, their ability to communicate with others, to perform and interpret communicative actions on the basis of the sociocultural and sociolinguistic norms of a particular speech community. In fact, as argued by Jessner (1999), bilinguals might show a higher degree of pragmatic development than monolingual speakers. We suggest that, by extension, Jessner’s approach may apply to trilinguals too. According to Cenoz (2001), language production and perception are more complex when more than two languages are involved. Speech production in the different languages a multilingual uses may share most of the general characteristics of speech production in monolinguals and bilinguals but is bound to me more complex and may be linked with characteristics which are the outcome of the interaction between different linguistic systems (Cenoz, 2001). According to Grosjean (2001), when bilinguals are in monolingual mode their other language is deactivated to a great extent and their language production

1 8 8 Language Acquisition Fotini Anastassiou and Georgia Andreou

resembles that of monolinguals. Both languages are activated when they are in bilingual mode, which is when their speech production indicates characteristics of bilingual speech, such as borrowing and code switching. Hoffmann & Stavans (2007) propose that Grosjean’s model could be used in trilingual speech production too; hence, trilinguals would then be able to operate in monolingual, bilingual and multilingual modes according to the degree of activation or deactivation of their three linguistic systems.

“The study of child trilingualism can be frustrating as it is often hampered by practical problems relating to the collection and interpretation of data on the one hand, and the absence of theoretical models that might be used for comparison on the other. The rewarding feature of such endeavours, however, is that they do allow fascinating glimpses into the human capacity of processing language and the linguistic resourcefulness of multilinguals” (Hoffmann & Stavans, 2007: 55).

Moreover, Hoffmann and Stavans (2007) point out that most of the research on trilingualism focuses on individuals who acquire or learn a third language in a school context (Cenoz & Genesee, 1998; Cenoz, Hufeisen, & Jessner, 2001) or immigrant minorities who learn a third language in social contexts (Baetens-Beardsmore, 1993). Also, the majority of these studies look into the development of a third language consecutively to the development of one or two other languages, just like our study presented below. The aim of the present study was to investigate the vocabulary interaction of the three languages of bilingual primary school children who are learning their L3. Specifically, this study aimed to explore cross- linguistic influence as far as interactional strategies and transfer lapses are concerned, as previously described by Cenoz (2001). Our hypothesis was that trilingual children would use their L1 (either Greek or Albanian) as the source language or default supplier of cross-linguistic influence, while speaking their L3 English.

3. Methodology

3.1. Participants

Our participants were 50 bilingual children, whose L1 and L2 were either Greek or Albanian and who were learning their L3. All of the children came from an immigrant background and were selected according to their bilingualism and their level in their L3, English. More particularly, participants were all “A1 level” learners

Language Acquisition 1 8 9 Major Trends in Theoretical and Applied Linguistics

of English that their L1 and L2 were either Greek or Albanian. In order to define the participants’ level in English (L3) they were asked to narrate two picture stories. According to their produced “text” their level in English was then estimated “A1” according to the Common European Framework of Reference for Languages. The participants’ ages ranged between 9 and 12 years old and their mean age was 10 years and 7 months. Their three languages were Albanian and Greek (L1 or L2) and English (L3). Our sample consisted of 26 girls and 24 boys. All of them were children of Albanian immigrants and they were born in Greece. First of all, we collected data regarding our learners through questionnaires. The aim of this questionnaire was to elicit demographic data about the participants (age, sex, class they attended), as well as information on their three languages (which one were their L1 and their L2 and the years they had been using their languages, how many years they had been learning their L3). On the basis of the data, two groups of participants were formed:

Group 1: this included 23 children (14 girls and 9 boys) who had as their L1 Albanian, L2 Greek and L3 English. Group 2: this included 27 children (12 girls and 15 boys) who had as their L1 Greek, L2 Albanian and L3 English.

Regarding Group 2 the children’s parents explained that they had decided to speak to their children in Greek, even though their level of Greek might not have been so high at the time, because they believed, this might help their children adapt more easily to Greek society. This is probably a common strategy immigrants follow with their children.

3.2. Instruments

We followed Cenoz’s (2001) method which aimed to investigate trilingual children’s speech production and specifically cross- linguistic influence. Cenoz provided the bilingual participants with two wordless picture stories (one taken from their English language school text book that they had already been taught and another one they had not seen before). She asked the children to narrate the two stories in their L3 (English) that they were learning at school. Her participants were also divided in two groups according to their L1 and finally their interactional strategies and transfer lapses were identified for each one of the two groups. We too used the same wordless story book with Cenoz: “The boy, the dog and the frog” by Mercer Mayer (1969). This is a commonly used series of story books for language studies (“Story 1” from now on). We also asked children to narrate a wordless picture story taken from their English school book “Sail away 2” (“Story 2” from now on). Before the children started narrating they were told that they

1 9 0 Language Acquisition Fotini Anastassiou and Georgia Andreou

could draw on and use whichever language they wanted while narrating in their L3. Specifically, they were told:“the languages you speak are all yours, you can use them if you feel you need to”. Our aim was to identify the language (L1 or L2) that the participants would use every time they needed help while narrating in their L3. We considered two types of cross- linguistic influence, following Cenoz (2001): “Interactional strategies which are intentional switches into languages other than the target language and their presence will depend on language mode so that their frequency is related to the bilingual and monolingual mode adopted by the speaker. Transfer lapses that are non-intentional switches and are not preceded by a pause or false start and can be regarded as automatic” (2001: 107). All interviews were audiotaped and transcribed and all cases of cross-linguistic influence were identified. We also used a bilingual interpreter (Greek, Albanian) to help us with the context of the Albanian transfer lapses.

4. Results

The narrations of the participants were transcribed and analyzed and the cases of interaction with the interlocutor as well as the cases of transfer lapses were identified. For each one of the two groups that were formed according to their L1, interactional strategies and transfer lapses were identified and analyzed for both stories that were narrated.

Group 1 Interactional strategies

Story 1: The main language that the children of Group 1 used to narrate Story 1 was their target language, L3, English. However, when seeking for help from their interlocutor all of the learners of Group 1 would use Greek, their L2. Examples:

(1a) «Το ξυπνάω πώς είναι;» (1b) «Τον κουβά πώς τον λέμε στα αγγλικά;» (1c) «πώς το λέμε κοιτάζει;» (1d) «Αυτό δεν μπορώ να το περιγράψω καλά.»

Story 2: Children of Group 1 were found to use the same interactional strategy that they used while narrating Story 1 when they narrated Story 2. So, during their narrations in their L3, children of Group 1 would ask for help from their interlocutor in Greek again. Examples:

Language Acquisition 1 9 1 Major Trends in Theoretical and Applied Linguistics

(1e) «Δεν κατάλαβα καλά». (1f) Το κασετόφωνο είναι “radio”, ε; (1g) «Χορεύει και είναι χαρούμενο αλλά δεν θυμάμαι όλες τις λέξεις στα αγγλικά». (1h) Πειράζει που δεν θυμάμαι την αγγλική «σκούπα»;

Transfer lapses While children were narrating in their target language L3, as they were requested to do, transfer lapses were found. Story 1 was the one that children had not seen before and transfer lapses in their L1, Albanian were a small percentage out of the total utterances produced. However, there was a higher percentage of intrasentential code switching in their L2.

Story 1: In a total of 638 utterances, 124 (19,5%) were transfer lapses from their L1 Albanian. Examples:

(2a) They go to the tekpem (lake) to catch the frog. (2b) They try to catch a bretkoca (frog).

However, when these children code switched to L1 while narrating in L3, in 80,5% of the utterances they automatically continued their sentence in L2. Examples:

(2c) Bretkoca (frog) πήγε να κάνει μια βόλτα. (2d) Tsuni (child) πήγε βόλτα.

Story 2: While children narrated Story 2 which they had previously worked with in class, transfer lapses were found in their L1, Albanian. These lapses were more in comparison with Story 1. However, the intrasentential code switching to L2 that was found in a high rate in Story 1 was not found in Story 2 at all. In a total of 285 utterances, 114 (40%) were transfer lapses from their L1 Albanian while speaking their L3. Examples:

(3a) Then he piu (drinks) milk. (3b) He pastron (cleans up). (3c) His mom wants him to go heret (early) to school. (3d) He picks up his canta e shkolles (school bag) and leaves.

With respect to the second group, that their L1 was Greek, the types of cross- linguistic influence used are as follows:

1 9 2 Language Acquisition Fotini Anastassiou and Georgia Andreou

Group 2 Interactional strategies During their narrations children of Group 2 used their L1, Greek, every time they would ask for help from their interlocutor in both Story 1 and Story 2. Examples:

Story 1:

(4a) Ο βάτραχος είναι the frog αλλά το πηδάω είναι jump; Δεν είμαι σίγουρος.. (4b) Θα πω και ελληνικά εδώ, εντάξει; (4c) Ωραία ιστορία. Πού το βρήκατε το βιβλίο; (4d) Το παιδί και ο βάτραχος κάνουν παρέα. Πώς είναι όμως «η παρέα» στα αγγλικά;

Story 2:

(4e) Αν πω άλλη λέξη και δεν είναι σωστή; (4f) Μπορώ να ρωτήσω μερικές λέξεις; (4g) Το ρομπότ βλέπω ότι καθαρίζει. Στα αγγλικά πώς να είναι; (4h) Η μαμά του θέλει να ξυπνήσει αμέσως. Θα θυμηθώ τώρα. Το «ξυπνήσει» ψάχνω…

Transfer lapses Children of Group 2 that their L1 was Greek, were found to be transferring from Greek. Differences were observed in the rate of transfer lapses in Story 1 and Story 2. Specifically, there were more transfer lapses in Story 1 (that children had not seen before) than in Story 2 (that they had already been taught in class).

Story 1: In a total of 459 utterances 319 (70%) were transfer lapses from their L1 Greek. Examples:

(5a) The frog looking με θυμό. (5b) The boy wants να πιάσει the frog. (5c) The frog ακολούθησε τα αποτυπώματα του and he went to his house. (5d) He saw the dog and the frog να πλενόντουσαν.

Story 2: In a total of 158 utterances, 97 (61,40%) were transfer lapses from their L1 Greek. Examples:

(5e) When he ξύπνησε he drank milk. (5f) Mom τον ξυπνάει. (5g) One robot πάει να ξυπνήσει το παιδί. (5h) Reads and plays κιθάρα.

Language Acquisition 1 9 3 Major Trends in Theoretical and Applied Linguistics

5. Discussion

The data presented above indicated that children transferred from their L1, either Greek or Albanian, while they were narrating in their L3, English. This was in line with our hypothesis that participants would use their L1. We found out that in both groups there was more cross- linguistic transfer while the children narrated the first story that had not been taught at school before. Previous studies (Griessler, 2001; Kellerman, 2001) indicated that L3 learners used both L1 and L2 while speaking their L3. Our study showed that most of our learners used their L1 while transfer lapses occurred. However, it is difficult to explain the prevalence of L1 and not of L2 in the case of transfer lapses, since most research studies point out that L2 is the main source of transfer lapses while one is speaking in his/her L3. (Hammarberg, 2001; Dewaele, 1998; Ringbom, 1987). Code switching serves as a tool that facilitates interaction. It helps individuals to overcome communication obstacles in one or several of their languages. The choice of L1 as the main source of transfer in L3 production might have to do with individual factors such as character of the interlocutor, age of the subjects, anxiety or characteristics of the languages involved. During interaction with the interlocutor our learners used Greek which was either their L1 or their L2. Bilinguals and multilinguals have been reported to be sensitive to the linguistic context of language use and appear to be “intuitively responsive to the linguistic needs of their interlocutors. We see this heightened sensitivity as part of the trilingual’s overall metalinguistic awareness and language competence” (Hoffmann & Stavans, 2007, p. 58). Also, Andreou (2007) points out that the experience of three languages is likely to result in enhanced awareness of the analysis and control components of language processing on the part of trilinguals. The preference for using two rather than three languages at a time can be partially explained with reference to context, preference, and experience. However, it cannot be fully explained in cognitive terms. According to Hoffmann & Stavans (2007), more data from more diverse sociolinguistic contexts might be able to explain if trilinguals tend to switch in a bidirectional manner and if this is something that indicates their trilingual competence. The most interesting part of our findings is the fact that our first group whose L1 was Albanian had this language sequence when they code switched while using their L3 in “story 1” (the one that was new to them):

► L3 → L1 (Albanian) → L2 (Greek).

According to Jessner (2008), it has been observed that L2 in a trilingual system takes up a specific role: L3 learners or users do not rely on their L1 as one would expect, but mainly on their L2, which serves as the “bridge language”.

1 9 4 Language Acquisition Fotini Anastassiou and Georgia Andreou

In our study the children were speaking in their L3 and when they felt that they needed help they used their L1 and then they continued speaking in their L2. There is not a safe explanation for this finding on our behalf that can be given just yet. We believe that when we reach a larger sample we will be able to confirm this finding and hopefully propose some reasoning. However, factors that could possibly affect cross-linguistic influence have been suggested by researchers. One of them may be “recency” according to Hammaberg (2001). He proposes that it could be hypothesized that learners are more likely to borrow from a language they actively use than from other languages they may know but they do not use as often. Another factor is the “foreign language effect” or “L2 status” (Meisel, 1983; Cenoz, 2001; Hammaberg, 2001), which is the tendency of L3 learners or users to activate their first foreign language they have acquired or learnt. Also, several studies have indicated that learners appear to have a pattern of using the L2 or languages other than their L1 as the source language of cross-linguistic influence (Clyne, 1997; Williams & Hammaberg, 1997; De Angelis & Selinker, 2001). We hope that our research on trilingualism/ multilingualism regarding this combination of languages (Greek, Albanian and English) will give more data on the way these specific languages interact. More specifically, we hope to provide some insight on these speakers’ language choice and how they automatically choose how they code switch each time they need to.

Table 1. Number of transfer lapses from L1 per group in the two stories

Group 1 (L1 Albanian) Group 2 (L1 Greek) Total Number Total Number Lapses Lapses of Utterances of Utterances Story 1 638 124 285 114

Story 2 459 319 158 97

Language Acquisition 1 9 5 Major Trends in Theoretical and Applied Linguistics

References

Andreou, G. 2007. Phonological awareness in bilingual and trilingual schoolchildren. The Linguistics Journal, 3 (3), 8-15.

Baetens-Beardsmore, H. (ed.) 1993. European models of bilingual education. Clevedon: Multilingual Matters.

Cenoz, J. 2001. The effect of linguistic distance, L2 status and age on cross- linguistic influence in L3 acquisition. In J. Cenoz, B. Hufeisen and U. Jessner (eds.), Cross-linguistic Influence in Third Language Acquisition: Psycholinguistic Perspectives, 8-20. Clevedon: Multilingual Matters.

Cenoz, J. and Genesee, F. 1998. Psycholinguistic perspectives on multilingualism and multilingual education. In J. Cenoz and F. Genesee (eds), Beyond Bilingualism: Multilingualism and Multilingual Education, 16-32. Clevedon: Multilingual Matters.

Cenoz, J., Hufeisen, B. and Jessner, U. 2001. Looking beyond second language acquisition: Studies in tri- and multilingualism. Tubingen: Stauffenberg Verlag.

Clyne, M. 1997. Some of the things trilinguals do. The international Journal of Bilingualism 1, 95-116.

De Angelis, G. and Selinker, L. 2001. Interlanguage Transfer and competing linguistic systems in the multilingual mind”. In J. Cenoz, B. Hufeisen and U. Jessner (eds.), Cross-linguistic Influence in Third Language Acquisition: Psycholinguistic Perspectives, 8-20. Clevedon: Multilingual Matters.

Dewaele, J.-M. 1998. Lexical inventions: L2 versus L3. Applied Linguistics, 19, 471-490.

1 9 6 Language Acquisition Fotini Anastassiou and Georgia Andreou

Griessler, M. 2001. The effects of third language learning on second language proficiency: An Austrian example. International Journal of Bilingual Education and Bilingualism, 4 (1), 50-60.

Grosjean, F. 2001. The bilingual’s language modes. In J. Nichol (ed.), One mind, two languages: Bilingual language processing, 1-22. Oxford: Blackwell.

Hammarberg, B. 2001. Roles of L1 and L2 in L3 production and acquisition. In J. Cenoz, B. Hufeisen and U. Jessner (eds.), Cross-linguistic influence in third language acquisition: Psycholinguistic perspectives, 21-41. Clevedon: Multilingual Matters.

Haugen, E. 1956. Bilingualism in the Americas: A Bibliography and Research Guide. (Publication of the American Dialect Society, No. 26). University of Alabama Press.

Herdina, P. and Jessner, U. 2000. The dynamics of third language acquisition. In J. Cenoz and U. Jessner (eds.), English in Europe: The acquisition of a third language, 84-98. Clevedon: Multilingual Matters.

Hoffmann, C. and Stavans, A. 2007. The evolution of trilingual codeswitching from infancy to school age: The shaping of trilingual competence through dynamic language dominance International Journal of Bilingualism 11, 55-72.

Hufeisen, B. 1998. L3-Stand Der Forschung – Was Bleibt zu tun? In B. Hufeisen and B. Lindemann (eds.), Tertiarsprachen: Theorien, Modelle, Methoden, 169-185. Tubingen: Stauffenburg.

Jessner, U. 1999. Metalinguistic awareness in multilinguals: Cognitive aspects of third language learning. Language Awareness 8, 201-209.

Jessner, U. 2008. Teaching third languages: Findings, trends and challenges. Language Teaching 41:1, 15-56. Cambridge University Press.

Kellerman, E. 2001. New uses for old language: Cross-linguistic influence in the depiction of motion and emotion. In J. Cenoz, B. Hufeisen and U. Jessner (eds.), Cross-linguistic Influence in Third Language Acquisition: Psycholinguistic Perspectives, 170-191. Clevedon: Multilingual Matters.

Lambert, W.E. 1990. Persistent issues in bilingualism. In B. Harley, A. Patrick, J. Cummins and M. Swain (eds.), The development of second language proficiency, 201-218. Cambridge, U.K.: Cambridge University Press.

Language Acquisition 1 9 7 Major Trends in Theoretical and Applied Linguistics

Meisel, J. 1983. Transfer as a second language strategy. Language and Communication 3, 11-46.

Oksaar, E. 1983. Multilingualism and multiculturalism from the linguist’s point of view. In T. Husén and S. Opper (eds.), Multicultural and multilingual education in immigrant countries, 17-36. Wenner Gren Symposium Series, Vol. 38. Oxford: Pergamon Press.

Ringbom, H. 1987. The role of first language in foreign language acquisition. Clevedon: Multilingual Matters.

Skutnabb-Kangas, T. 1984. Bilingualism or not. Clevedon: Multilingual Matters.

Williams, S. and Hammaberg, B. 1997. L1 and L2 influence in L3 production. Stockholm: Stockholm University, Centre for Research on Bilingualism.

1 9 8 Language Acquisition Non-Target Long-Distance Wh-Questions: Crosslinguistic Typological Distinctions in Early L1 Production

Evangelia K. Asproudi

Aristotle University of Thessaloniki [email protected]

Abstract

This paper investigates the production of non-target long-distance wh-questions by preschool-age children across languages, including L1 Greek. Previous studies show that children frequently resort to partial wh-movement, wh-copying and wh- subextraction. To examine whether Greek children follow a similar developmental route, ninety 4-to-7-year-old children participated in production tasks designed along the principles of McDaniel et al. (1996) and Crain & Thornton (1998). Certain differences aside, Greek children seem to behave similarly to other children: partial wh-movement, wh-copying and wh-subextraction are interpreted in light of children’s preference for the more economical LF- rather than PF- convergent representations of long-distance wh-chains.

1. Introduction

The aim of this paper is to investigate the production of non-target long-distance (LD) wh-questions by preschool-age children across languages, including L1 Greek. Specifically, children’s production of LD wh-interrogatives is examined with respect to two questions:

A. Is there evidence for children’s sensitivity to the successive cyclic nature of LD movement that operates in the adult language from early on? B. Do children’s deviant patterns during the development of LD questions reveal a strategy aiming at one-to-one mappings between semantic (LF) and phonetic (PF) output representations?

In order to provide answers to the above questions, reference will be made to certain parametric variations that wh-movement across languages is subject to. These are partial wh-movement (PM), wh-copying and wh-subextraction.

Language Acquisition 1 9 9 Major Trends in Theoretical and Applied Linguistics

1.1. Long-Distance Wh-Movement Across Languages

PM refers to structures for question formation that serve to express LD wh- dependencies along with or instead of the common structure that involves LD extraction of the wh-phrase. In these structures the wh-phrase appears in the left edge of the subordinate clause, while another wh-word appears in the matrix scope-taking position (e.g. McDaniel 1989; Dayal 1994). Apart from PM, in a number of languages another strategy is employed as an alternative way of expressing LD wh-dependencies in interrogative structures. This strategy is known as wh-copying and refers to constructions that involve repetition of the wh-item in all the cycles of the derivation (e.g. Felser 2004; Bruening 2006). PM and copying are allowed in languages like German and Dutch but not in ones like English, and they are respectively exemplified in (1)-(3) and (4)-(6) below:

German (McDaniel 1989) (1) Was glaubt Hans wen Jakob anruft? what think-3SG Hans who Jakob call-3SG ‘Who does Hans think Jakob is calling?’

Dutch (Jakubowicz & Strik 2008) (2) Wat zeg je (zeg) waar Marie (waar) heengaat? what say-2SG you where Marie go-3SG ‘Where do you say Maria goes?’

English (McDaniel 1989) (3) * What does Hans think whom Jakob is calling?

German (Felser 2004) (4) Wen glaubst du wen sie liebt? who think-2SG you who she love-3SG ‘Who do you think she loves?’

Dutch (Jakubowicz & Strik 2008) (5) Wie denk je wie verhalen leest? who think-2SG you who histories read-3SG ‘Who do you think reads histories?’

English (Felser 2004) (6) * Who do you think who she loves?

Turning to wh-subextraction, this is the phenomenon where, in case of a D-linked wh-phrase (Pesetsky 1987), only the wh-determiner raises to the

2 0 0 Language Acquisition Evangelia K. Asproudi

edge of the left periphery. The accompanying nominal element is not pied- piped but remains stranded to a position lower in the clause, this often being its base-generated position. The possibility for wh-subextraction does not apply in languages like English ((7)) and Dutch ((8)) but is valid in other languages, as its application is largely determined by morphological ‘richness’. Morphologically rich languages like the Slavic ones (e.g. Polish, (9)) freely allow for subextraction of the nominal part of the wh-phrase (e.g. Chen et al. 1998; Gavarró & Solà 2004; van Kampen 2004).

English (Chen et al. 1998) (7) * How much do you think I have money?

Dutch (Chen et al. 1998) (8) *Hoeveel denk je dat ik gelg heb? howmuch think-2G you that I money have-1SG ‘How much money do you think I have?’

Polish (van Kampen 2004) (9) Jaki wykreciles numer? which dial-2SG number ‘Which number did you dial?’

1.2. Long-Distance Wh-Movement in Greek

PM and wh-copying do not constitute licit options of expressing LD wh- dependencies in Greek. LD extraction is formed in the more usual way of fronting the wh-phrase to the matrix left periphery; it is allowed in the presence of ‘oti’ and ‘na’, as illustrated in the following examples:

(10) Ti ipes oti tha dhjavasun ta pedhja? what said-2SG that will read-3PL the children ‘What did you say that the children will read?’

(11) Ti ipes na dhjavasun ta pedhja? what said-2SG to read-3PL the children ‘What did you tell the children to read?’

Furthermore, it is interesting to note that despite being a morphologically rich language, Greek does not allow for wh-subextraction:

Language Acquisition 2 0 1 Major Trends in Theoretical and Applied Linguistics

(12a) * Posa perilamvani piimata i siloji? how many contain-3SG poems the collection ‘How many poems does the collection contain?’ (12b) Posa piimata perilamvani i siloji? how many poems contain-3SG the collection ‘How many poems does the collection contain?’

As shown in (12), the raised wh-phrase must necessarily be accompanied by the nominal element it specifies.

2. Child Studies on Long-distance Question Production

As stated from start, this paper focuses on the investigation of the patterns employed by children during LD question production. Across languages, extensive research has been conducted on child LD questions; yet, this area has so far remained largely uninvestigated with regard to child Greek. Studies in other early languages have shown that children often produce well-formed LD questions in which the wh-phrase surfaces in the matrix left periphery. (13) and (14) below exemplify child well-formed LD questions:

English (Thornton 1990) (13) What do you think is in the white box?

French (Jakubowicz & Strik 2008) (14) Où a dit Lala que le poisson nage? where have-3SG said Lala that the fish swim-3SG ‘Where has Lala said that the fish swims?’

As will be explained in the end of this section, well-formed LD questions provide evidence in favour of child sensitivity to successive cyclicity. Further evidence in this direction comes from deviant PM and wh-copying constructions, which are frequently used by children in target LD contexts. Let us first look at examples of PM:

English (Thornton 1990) (15) What do you think who jumped over the can?

French (Jakubowicz & Strik 2008) (16) Qu’ est- ce que Lala a dit où le poisson nage? what be-3SG it that Lala have-3SG said where the fish swim-3SG ‘Where did Lala say that the fish is swimming?’

2 0 2 Language Acquisition Evangelia K. Asproudi

Dutch (Jakubowicz & Strik 2008) (17) Wat zei Lala waarom Kikker weggaat? what said-3SG Lala why Frog leave-3SG ‘Why did Lala say that Frog leaves?’

And here are examples of child wh-copying:

English (Thornton 1990) (18) What do you think what the baby drinks?

French (Oiry 2004) (19) Tu crois quoi que je bois quoi? you believe-2SG what that I drink-1SG what ‘What do you believe I am drinking?’

Dutch (Jakubowicz & Strik 2008) (20) Wat zei Billy wat Kikker eet? what said-3SG Billy what Frog eat-3SG ‘What did Billy say that Frog eats?’

Apart from PM and copying, another non-target strategy employed by children during LD question production is wh-subextraction; this is found in early languages like English (e.g. Chen et al. 1998; Gavruseva & Thornton 2001) and Dutch (e.g. van Kampen 1994, 1997).

English (Chen et al. 1998) (21) How many do you think marbles are in there?

Dutch (van Kampen 1994) (22) Welke wil jij liedje zingen? which want-2SG you song sing-INF ‘Which song do you want to sing?’

In addition, feature copying is attested in child language. Feature copying refers to instances of wh-subextraction, where the stranded nominal is accompanied by some type of resumptive element (e.g. Chen et al. 1998). Let us look at examples:

(23) Who do you think this bottle is? (24) How many do you think how many marbles are in there? (Chen et al. 1998)

Language Acquisition 2 0 3 Major Trends in Theoretical and Applied Linguistics

The stranded nominal element may be accompanied by a resumptive determiner ((23)) or wh-word ((24)) which doubles the phi- and case features of the fronted wh. On the whole, the occurrences of PM, wh-copying and wh-subextraction/ feature copying in child grammar are assumed to be triggered by the UG constraint on successive cyclic movement. In PM and wh-copying, the medial wh-element is viewed as the overt expression of LD movement through the intermediate CP (Thornton 1990); in other words, the medial wh is a spelled-out copy of the raised wh-phrase. As noted in Stromswold (1995), medial-wh questions may be the result of children’s effort to produce UG-consistent variants of LD questions that their grammars cannot yet generate productively in a well-formed way. A copy-theoretical account has also been put forward for the explanation of feature copying constructions. The material that surfaces in the base-generated position of the wh-element and accompanies the stranded nominal head doubles features of the fronted wh-determiner, and it is hence attributed a resumptive function (cf. Chen et al. 1998). On the basis of this, feature copying is assumed to represent the spelling-out of a copy during the formation of the wh-chain by the incomplete child grammar (Yamane et al. 1999). In other words, this doubling phenomenon seems to reflect intermediate stages of successive cyclic movement (cf. Felser 2004). To sum up, it can be concluded that apart from well-formed LD questions, even deviant structures like PM, wh-copying and wh-subextraction/feature copying attest for the presence of successive cyclic wh-movement in child grammar from an early age. Within Chomsky’s minimalist theory (1995), this is interpreted, in turn, in light of children’s preference for LF-like representations (van Kampen 1997): the child computational system gives priority to those options that involve movement of the minimum for LF/PF convergence, even if these options override language-specific PF conditions on the Spell-out (van Kampen 1996, Gavruseva & Thornton 2001).

3. The Present Study

3.1. Predictions

In view of previous research, the predictions formulated with regard to Greek children’s production of LD wh-questions were the following:

Prediction 1: In their LD questions, the Greek children are not expected to converge fully with adult LD movement. Apart from well-formed questions, they should also employ - on a par with their peers crosslinguistically - alternative

2 0 4 Language Acquisition Evangelia K. Asproudi

strategies: PM and wh-copying are expected as variants to target LD questions (e.g. Thornton 1990; Oiry 2004; Jakubowicz & Strik 2008).

Prediction 2: In cases of D-linking, the crosslinguistic picture is not very clear. Early English studies (e.g. Chen et al. 1998; Gavruseva & Thornton 2001) have shown that children consistently pied-pipe wh-phrases, with wh-subextraction attested scarcely. On the contrary, Dutch children have been found to produce subextracted wh-phrases quite often (van Kampen 1994, 1997). With regard to Greek children, it is predicted that wh-subextraction should be quite frequent, especially in the younger ages. This is closely related to the rich morphological specification of Greek wh-phrases. On one assumption, the rich morphological load might impede the raising of the full wh-phrase to the matrix SpecCP; on another assumption, the rich morphology might render the link between the wh- and the nominal element so strong that there is no need for the child to also raise the nominal element in matrix SpecCP. Either way, the result should be the stranding of the nominal, which is in line with both children’s assumed early preference for economical LF-like representations and the processing limitations of child grammar (e.g. van Kampen 1996; Chen et al. 1998; Gavruseva & Thornton 2001). Consistent pied-piping is expected to be the preferred choice for the older children.

3.2. Participants

The study group consisted of ninety typically developing children aged 4;0 to 7;0. For the analysis of the data, these children were divided into three equivalent subgroups A, B and C. Group A included thirty children between four and five (mean age range: 4;6), group B thirty children between five and six (mean age range: 5;5) and group C thirty children between six and seven years old (mean age range: 6;7). Group A and B children were in their first and second year in kindergarten respectively, while group C children attended the first grade in primary school.

3.3. Materials and Procedure

The children participated in a production task designed along the principles of McDaniel et al. (1996) and Crain & Thornton (1998). The aim was to see whether children produce adult-like LD questions. A puppet called ‘Astrulis’ that had come from another planet was introduced to the child, and both of them participated in two games. The first game included prompts of the type Rotise ton Astruli X mandevi aftos (‘Ask Astrulis X he guesses’), where X stood for the

Language Acquisition 2 0 5 Major Trends in Theoretical and Applied Linguistics

respective wh-element. As for the second game, it consisted of prompts which were of the form Rotise ton Astruli X protimai aftos (‘Ask Astrulis X he prefers’), where X stood again for the respective wh-element. Here is a sample of the first and second game protocol as well as ofthe stimulus sentences used.

1st game: The experimenter presents five toys: one car, one pink and one blue comb, and one red and one yellow ball. She asks both the child and the puppet to cover their eyes while she is hiding each of these toys in different numbered boxes. Then she asks the child and the puppet to uncover their eyes, and the game proceeds. After the experimenter has elicited the child’s guess, she prompts the child to elicit the puppet’s guess.

Stimulus sentences: Experimenter: Sto kuti 3 ekripsa mia htena. Ja mandepse pja. ‘In box 3 I hid one of the combs. Guess which one.’ (the child says his/her guess…) Experimenter: Esi mandevis oti sto kuti 3 ekripsa tin … htena. Rotise ton Astruli pja mandevi aftos. ‘You are guessing that in box 3 I hid the … comb. Ask Astrulis which one he is guessing.’

Target question: Pja htena mandevis oti ekripse sto kuti 3? ‘Which comb are you guessing that she (i.e. the experimenter) hid in box 3?’

In sum, 9 LD questions were elicited in this game from each child.

2nd game: The experimenter presents four toy characters: one rabbit, one dog and two horses. She explains to the child that three of these toy characters have to be matched with certain actions.

Stimulus sentences: Experimenter: Kapjo alogaki θa pai volta. Rotise ton Astruli pjo protimai aftos. ‘One of the two horses is going for a walk. Ask Astrulis which one he prefers.’

Target question: Pjo alogaki protimas na pai volta? ‘Which horse do you prefer that it go for a walk?’

2 0 6 Language Acquisition Evangelia K. Asproudi

In this second game, 6 LD questions were elicited from each child. The duration of this task was about 20 minutes, 15 minutes for the first and 5 minutes for the second game, with the elicited LD questions on the part of each child being 15 in total. Each child was tested separately in a room next to their classroom, while each session was tape recorded and transcribed at a later stage. In case the child did not react to a prompt, this prompt (and the relevant part of the game) was repeated twice; if still no question was elicited, the procedure continued with the presentation of the next part of the game and the corresponding prompt.

4. Results and Discussion

Overall, the results obtained were in line with the predictions outlined in section 3.1. A detailed presentation of the relevant results is provided below.

4.1. Prediction 1

In Prediction 1 the interest lies in LD question production, and specifically in whether children gave PM and copying structures. Overall, LD questions were not the most preferred choice on the part of the children.

Table 1. Target: LD Questions. Elicited Patterns

GROUP A GROUP B GROUP C

Pattern No Mean No Mean No Mean

LD (target) 74/450 16.45% 126/450 28% 220/450 48.89%

SD (non-target) 370/450 82.22% 322/450 71.56% 227/450 50.44% Isolated wh 1/450 0.22% 0/450 0% 0/450 0% (non-target) No response 5/450 1.11% 2/450 0.44% 3/450 0.67% (non-target)

As shown in table 1, in place of target LD questions, non-target short-distance (SD) question production was the pattern to which children resorted mostly in the two younger groups. Examples of all error types attested are presented below.

SD (25) Esi pja xtena mandevis? you which comb guess-2SG target: Pja xtena mandevis oti ekripse sto kuti 3? (= ‘Which comb are you guessing that she hid in box 3?’)

Language Acquisition 2 0 7 Major Trends in Theoretical and Applied Linguistics

(26) Pja ine sto kuti 5? which be-3SG in the box 5 target: Pja bala mandevis oti ine sto kuti 5? (= ‘Which ball are you guessing to be in box 5?’) (27) Ti na kani to skilaki Astruli? what to do-3SG the dog Astruli target: Ti protimas na kani to skilaki? (= ‘What do you prefer that the dog do?’) isolated wh (28) Pjo? which target: Pjo alogaki mandevis oti ekripse sto kuti 2? (= ‘Which horse are you guessing that she hid in box 2?’) no response (29) … target: Pja bala mandevis oti ine mesa sto kuti 5? (= ‘Which ball are you guessing to be in box 5?’)

With a closer focus now on LD production, the majority of the produced questions were well-formed. Table 2 presents the rates of grammatical and ungrammatical LD questions, with an emphasis on the PM and copying rates of children of all groups.

Table 2. Long-Distance Question Production1

GROUP A GROUP B GROUP C

Pattern No Mean StdD* No Mean StdD No Mean StdD Grammatical LD 50/74 67.57% 11.64 86/126 68.25% 13.65 170/220 77.27% 15.57 questions LD questions involving subextraction 18/74 24.32% 10.31 22/126 17.46% 6.20 13/220 5.91% 3.43 (ungrammatical) LD questions involving feature copying 6/74 8.11% 3.14 9/126 7.14% 7.36 21/220 9.55% 11.77 (ungrammatical) Partial wh- movement questions 0/74 0% 0.00 2/126 1.59% 2.25 7/220 3.18% 4.88 (ungrammatical) Wh-Copying questions 0/74 0% 0.00 7/126 5.56% 9.88 8/220 3.64% 5.09 (ungrammatical) Wh in-situ LD 0/74 0% 0.00 0/126 0% 0.00 1/220 0.45% 1.08 questions *StdD = Standard Deviation

1 The number of contexts in each group reflects the number of LD questions produced by each group in total (see table 1).

2 0 8 Language Acquisition Evangelia K. Asproudi

As shown in the above table, children’s LD questions were highly grammatical throughout. PM and copying were extremely rare and appeared only in the two older groups of children, as illustrated in the following figure.

6% 5.56%

5%

4% 3.64% 3.18% 3%

2% 1.59%

1% 0% 0% 0% group A group B group C

partial wh-movement wh-copying

Figure 1. Partial Wh-Movement and Wh-Copying in Long-Distance Question Production

A two-way mixed ANOVA analysis was performed on the results: PM/copying (2) ¥ group (3). Overall, the main within-subject effects of PM/copying [F(1,87)=1.929, p=.168] and PM/copying by group interaction [F(2,87)=6.672, p=.175] were found to be non-significant. Between groups, no significant main effect was found either [F(2,87)=1.021, p=.365]. Given that no significant main effects or interactions were found, no post-hoc tests were run. PM and wh-copying questions are respectively exemplified in (30)-(31) and (32)-(33) below.

(30) Pu mandevis esi pja ine i kafe bala? where guess-2SG you which be-3SG the brown ball target: Pu mandevis oti ine i kokini bala? (= ‘Where are you guessing the red ball to be?’)

(31) Esi pjo mandevis sto 3 ti ine, to mikro aftokinitaki i to megalo? you which guess-2SG in the 3 what be-3SG the little car or the big target: Pjo aftokinitaki mandevis oti ine sto kuti 3? (= ‘Which car are you guessing to be in box 3?’)

Language Acquisition 2 0 9 Major Trends in Theoretical and Applied Linguistics

(32) Astruli esi ti mandevis sto 3 ti exi? Astruli you what guess-2SG in the 3 what have-3SG target: Pjo aftokinitaki mandevis oti ine sto kuti 3? (= ‘Which car are you guessing to be in box 3?’)

(33) Esi pu mandevis i kokini bala pu ine? you where guess-2SG the red ball where be-3SG target: Pu mandevis oti ine i kokini bala? (= ‘Where are you guessing the red ball to be?’)

To sum up with regard to Prediction 1, the overwhelming majority of the produced LD questions were well-formed, with PM and wh-copying attested very rarely only in the older children’s productions2. Across languages, the production of LD wh-interrogatives has also been found to involve use of PM and wh-copying. Certain studies are in line with the Greek data in the sense that PM and copying were rather rare compared to well-formed full wh-fronted LD questions; such studies come from English (e.g. Thornton & Crain 1994) and French (e.g. Jakubowicz & Strik 2008). Other studies, however, have revealed high frequency in use of PM/copying compared to full-movement questions. Such evidence is found in Thornton (1990) for English, in Strik (2007) and Oiry (2004, 2009) for French, in van Kampen (1996, 1997, 2009a, b) for Dutch and in Gutierrez (2004, 2006) for Basque and Spanish. On the whole, then, the Greek data differs from the majority of other early language data in that PM and copying were produced sporadically rather than systematically. A further difference between Greek and other early language data with respect to PM is that the structures produced by the Greek children were actually reverse to the PM structures produced by other children in other languages. The Greek children inserted a wh-element in the embedded SpecCP, while the true wh raised to the matrix SpecCP, as exemplified in (30)-(31) above. As a matter of fact, examples like (31) are reminiscent of a similar pattern attested in cases of D-linking in some German dialects (Fanselow & Ćavar 2001) as well as in early English (Thornton 1990; Thornton & Crain 1994) and early Dutch (van Kampen 1996, 1997, 2009a, b); there, the true D-linked phrase raises to the matrix periphery and another wh-word is introduced in the embedded SpecCP. Yet, with non-D-linked wh-phrases, PM in all child languages resembles the more usual adult PM in raising the true wh to the embedded periphery and inserting a wh-

2 It is noteworthy that no PM and wh-copying instances were produced by the youngest children. Given the fact that such instances were scarce even in the older groups, their total lack in the youngest group may be attributable to chance performance and not to a certain underlying theoretical reason.

2 1 0 Language Acquisition Evangelia K. Asproudi

scope marker in the matrix periphery. It is precisely in this respect that the Greek data diverges, since the instances of ‘PM’ concerned not only D-linked but also non-D-linked wh-phrases. Irrespective of these differences, however, PM and copying questions constitute an overt manifestation of successive cyclicity in the sense that they involve copying without deletion at PF. As a consequence, both well-formed and medial-wh questions reflect children’s sensitivity to the successive cyclic formation of LD wh-chains.

4.2. Prediction 2

Prediction 2 states that in cases of D-linking, Greek children should often resort to wh-subextraction (and feature copying), especially in the younger ages; consistent pied-piping is expected in the oldest group. In order to test this prediction, the focus here will be only on the instances of D-linked LD questions that were produced by children; specifically, it is interesting to see whether children consistently pied-piped the extracted wh-phrase or produced alternative patterns instead. Table 3 presents the relevant patterns and their respective rates.

Table 3. Patterns of D-Linking in Long-Distance Question Production3

GROUP A GROUP B GROUP C

Pattern No Mean StdD No Mean StdD No Mean StdD 40.38 74.81 Pied-piping (target) 6/30 20 % 37.08 21/52 41.19 101/135 35.99 % % Wh-subextraction 42.31 9.63 18/30 60 % 42.24 22/52 40.88 13/135 14.98 (non-target) % % Feature copying 17.31 15.56 6/30 20 % 36.65 9/52 31.97 21/135 33.32 (non-target) % %

As displayed in the above table, pied-piping, wh-subextraction and feature copying were the D-linking patterns to which children resorted, with only the first one being the target (and grammatical) choice. Schematically, the above results look as follows.

3 The number of contexts in each group corresponds to the total number of D-linked wh- phrases produced on the part of the children.

Language Acquisition 2 1 1 Major Trends in Theoretical and Applied Linguistics

100% 20% 17.31% 15.56% 9.63% 80%

42.31% 60% 60%

40% 74.81%

40.38% 20% 20%

0% group A group B group C

pied-piping (target) wh-subextraction (non-target) feature copying (non-target) Figure 2. Patterns of D-Linking in Long-Distance Question Production

A two-way mixed ANOVA analysis was performed: pattern of D-linking (3) ¥ group (3). Overall, the main within-subject effects of D-linking pattern [F(2,144)=5.751, p=.004] and D-linking pattern by group interaction [F(4,144)=8.184, p=.000] were found to be significant. Between subjects, however, no main effect of group was observed [F(2,72)=1.713, p=.188]. This implies that there is heterogeneity of performance within groups. Focusing more narrowly on pairwise within- group comparisons, pied-piping prevailed and differed significantly from wh-subextraction (p=.000) and feature copying (p=.000) only in C. In A and B wh-subextraction was the prevailing pattern, with feature copying rates being significantly lower (A: p< .02, B: p< .05). Examples of these three patterns are presented below.

pied-piping: (34) Pja bala mandevis oti ine sto 5? which ball guess-2SG that be-3SG in the 5 ‘Which ball are you guessing to be in 5?’ (35) Pjo alogaki protimas na pai volta? which horse prefer-2SG to go-3SG walk ‘Which horse do you prefer that it go for a walk?’

wh-subextraction: (36) Astruli pjo mandevis alogaki na ine sto numero 2? Astruli which guess-2SG horse to be-3SG in the number 2 ‘Which horse are you guessing to be in number 2?’

2 1 2 Language Acquisition

Evangelia K. Asproudi

(37) Pjo protimas Astruli na pai me to kilo alogaki? which prefer-2SG Astruli to go-3SG with the dog horse ‘Which horse do you prefer that the dog take with him?’

feature copying: (38) Pjo de bori na pari mazi tu o Petros kapjo pragma? which not can-3SG to take-3SG with him the Peter some thing ‘Which thing can’t Peter take with him?’ (39) Pjo protimas to alogaki na pari to skilaki? which prefer-2SG the horse to take-3SG the dog ‘Which horse do you prefer that the dog take?’

As illustrated in the above examples, in cases of wh-subextraction the stranded nominal element occurred either in situ ((37)) or in medial COMP ((36)), with the former position being the most preferred one (base position preference rates: A: 68.42%, B: 83.33%, C: 56.25%). In feature copying the nominal element was marked by a definite ((39)) or an indefinite ((38)) determiner which doubled the phi- and case features of the fronted wh. Similarly to LD questions involving subextraction, in the ones involving feature copying the nominal element, along with the resumptive determiner, surfaced either in situ ((38)) or in medial COMP ((39)). Again, the former position was most preferred as the landing site for the DP constituent (base position preference rates: A: 80%, B: 88.89%, C: 91.67%). The overt intermediate copies in feature copying may be taken as proof for the successive cyclic derivation of D-linked questions (Cinque 1990; Rizzi 1990). On the whole with respect to Prediction 2, pied-piping predominated only in the oldest group, with the younger children resorting most often to subextraction and feature copying. Crosslinguistically, similar studies have revealed two main tendencies. In languages like English, on the one hand, subextraction and feature copying are found more rarely than pied-piping in children’s productions (Gavruseva 1997; Chen et al. 1998; Gavruseva & Thornton 2001). In languages like early Dutch, on the other hand, subextraction and feature copying are employed much more frequently (van Kampen 1997, 2004). On the basis of these comparisons, Greek children performed more in line with their Dutch rather than with their English peers, since subextraction and feature copying instances were rather frequent, at least in groups A and B. This performance may be due to that the rich case and agreement morphology in Greek (similarly to Dutch but unlike English) on both the wh-determiner and its accompanying noun facilitates the treatment of the two as independent morphological units (cf. Gavarró & Solà 2004). This might lead, in turn, to an incorrect licensing of a stranded nominal. Besides, in terms of processing, the rich morphological load that wh-determiner and noun bear in combination might discourage children from fronting the full wh-phrase to the matrix left periphery.

Language Acquisition 2 1 3 Major Trends in Theoretical and Applied Linguistics

The result will be nominal stranding to a lower position, which is in line with the assumption that, under the dictations of natural economy, children prefer the overriding of PF requirements and the construction of LF-like representations with movement of the minimal wh-material to the matrix SpecCP (Chomsky 1995; van Kampen 1997; Gavruseva & Thornton 2001). Finally, it is interesting to note that feature copying pertained to a partial set of features, with interrogative force being absent from the lower copy; the lower copy encoded feature doubling under quantificational rather than under interrogative force. On this ground, it is plausible to argue that the phenomenon of feature copying casts doubts on the full identity of all copies advanced in Chomsky’s copy theory of movement (Chomsky 1995). A more satisfactory proposal seems to be the one put forward by Felser (2004), who claimed that all copies in a wh-chain are not identical in terms of feature specification: the highest copy carries the wh-feature, whereas the lower copies carry features other than the wh-feature. Similarly, Radford (2009) argued for the possibility of discontinuous spell-out of a moved constituent, with a partial copy of it that lacks the wh-feature specification appearing in a lower position.

5. Conclusions

In conclusion, this paper has dealt with the L1 acquisition of LD wh-movement. Apart from full-fronted wh-questions, children across languages, including Greek, often resort to alternative constructions, such as PM, wh-copying and wh- subextraction/feature copying. On the basis of such findings, more generalised conclusions can be drawn about the fundamental questions shaped in the introductory section. Question A was whether there is evidence for the availability of successive cyclic movement in early child grammar. To begin with, the production of well- formed LD questions points to this direction. Further support for successive cyclicity comes from deviant question production. On the one hand, the rare occurrences of PM and wh-copying constructions manifest overtly the intermediate positions of a wh-dependency chain (e.g. Thornton 1990; van Kampen 1997; Gutierrez 2006); on the other hand, instances of feature copying with an overt copy in medial-COMP position also serve the same function. As a matter of fact, the reverse-to-usual PM as well as the feature copying pattern observed in the Greek data challenges the full identity of copies claimed by Chomsky (1995) in favour of the claim that lower copies are partial copies of the fronted wh deprived of wh-feature specification (Felser 2004; Radford 2009). Turning to question B, it concerned child preference for early PF/LF convergence. The major source of evidence suggestive of this convergence is children’s deviant production of D-linked questions. Given that movement at LF

2 1 4 Language Acquisition Evangelia K. Asproudi

pertains to the minimum, namely to movement of the wh-feature for checking purposes (Chomsky 1995), children’s wh-subextraction and feature copying are closer to this LF representation than full pied-piping which carries excessive non-wh-material to the matrix left periphery. Instances of subextraction and feature copying in the Greek data were quite frequent, and this shows that children’s overriding of PF wh-and-noun adjacency requirements in favour of LF-like representations was systematic. Furthermore, indirect evidence for the reduction of PF/LF discrepancies in child grammar comes from medial- wh questions. The occurrences of PM and wh-copying structures constitute, despite their rareness, manifestation of the fact that children’s wh-chains might involve copying without deletion at PF. In other words, the successive cyclic LF construction of the wh-chain becomes equivalent to its PF realization for children, who thus override the PF requirement of overtly spelling out only one copy of the wh-chain (Chomsky 1995). As a final note, all this evidence for early availability of successive cyclic movement and for early convergence of PF/LF representations can be subsumed under a general tendency to natural economy. Through the overtly manifested local steps, children’s production system seems to ‘refresh’ the wh-element involved in a LD dependency; in other words, it may be concluded that during LD question production children attempt to minimise on the processing burden in every way possible (cf. Jakubowicz & Strik 2008).

Language Acquisition 2 1 5 Major Trends in Theoretical and Applied Linguistics

References

Bruening, B. 2006. Differences between the Wh-Scope-Marking and Wh-Copy Constructions in Passamaquoddy. Linguistic Inquiry 37: 25-49.

Chen, D., M. Yamane and W. Snyder. 1998. On the nature of left-branch violations in child English. BUCLD Proceedings 22: 91-96.

Chomsky, N. 1995. The Minimalist Program. Cambridge, Massachusetts: MIT Press.

Cinque, G. 1990. Types of A-Dependencies. Cambridge, Massachusetts: MIT Press.

Crain, S. and R. Thornton. 1998. Investigations in Universal Grammar. A Guide to Experiments on the Acquisition of Syntax and Semantics. Cambridge, Massachusetts: MIT Press.

Dayal, V. 1994. Scope marking as indirect wh-dependency. Natural Language Semantics 2: 137-170.

Fanselow, G. and D. Ćavar. 2001. Remarks on the economy of pronunciation. In G. Müller and W. Sternefeld (eds.), Competition in Syntax. Berlin: de Gruyter, 107-150.

Felser, C. 2004. Wh-copying, phases, and successive cyclicity. Lingua 114: 543- 574.

Gavarró, A. and J. Solà. 2004. Wh-subextraction in child Catalan. Paper presented at Second Lisbon meeting on language acquisition, Universidade de Lisboa, June 1st, 2004.

Gavruseva, E. 1997. Syntax of Possessive Phrases and Acquisition of Whose- Questions: a Comparative Study of Child L1 and Child L2 Grammars. PhD Dissertation, Georgetown University, Washington.

2 1 6 Language Acquisition Evangelia K. Asproudi

Gavruseva, E. and R. Thornton. 2001. Getting It Right: Acquisition of Whose- Questions in Child English. Language Acquisition 9: 229-267.

Gutierrez, M.J. 2004. The acquisition of L1 Basque Long-Distance wh-questions: a longitudinal investigation. Paper presented at BIDE04, Bilbo, Spain.

Gutierrez, M.J. 2006. Acquiring long-distance wh-questions in L1 Spanish. In V. Torrens and L. Escobar (eds.), The Acquisition of Syntax in Romance Languages. Amsterdam, Philadelphia: John Benjamins, 251-287.

Jakubowicz, C. and N. Strik. 2008. Scope-marking Strategies in the Acquisition of Long Distance wh-Questions in French and Dutch. Language and Speech 51: 101-132.

van Kampen, J. 1994. The learnability of the left-branch condition. In R. Bok- Bennema and C. Cremers (eds.) Linguistics in the Netherlands 1994. John Benjamins, Amsterdam, 83-94.

van Kampen, J. 1996. PF/LF Convergence in Acquisition. In K. Kusumoto (ed.), North East Linguistics Society Conference (NELS) 26. Amherst: GLSA, 149-163.

van Kampen, J. 1997. First Steps in Wh-Movement. PhD Dissertation, University of Utrecht.

van Kampen, J. 2004. An acquisitional view on optionality. Lingua 114: 1133- 1146.

van Kampen, J. 2009a. The learnability of A-bar chains. In M. Everaert, T. Lentz, H. De Mulder, O. Nilsen and A. Zondervan (eds.), The Linguistic Enterprise. Amsterdam, Philadelphia: John Benjamins, 115-139.

van Kampen, J. 2009b. The ‘Phased’ Learnability of Long Wh-questions. Cascadilla Proceedings Project.

McDaniel, D. 1989. Partial and multiple wh-movement. Natural Language and Linguistic Theory 7: 565-604.

McDaniel, D., C. McKee and H. Smith Cairns (eds.). 1996. Methods for Assessing Children’s Syntax. Cambridge, Massachusetts: MIT Press.

Oiry, M. 2004. Evidence from L1 Acquisition for the Syntax of Wh-Scope Marking. Paper presented at 2nd WoSS, University of Barcelona, September 15-17, 2004.

Language Acquisition 2 1 7 Major Trends in Theoretical and Applied Linguistics

Oiry, M. 2009. A case of true optionality: Wh-in-situ patterns like Long Movement in French. Paper presented at ISTAL 19, School of English, Aristotle University of Thessaloniki, 3-5 April 2009.

Pesetsky, D. 1987. Wh-in-situ: movement and unselective binding. In E. Reuland and A. ter Meulen (eds.), The Representation of (In)definiteness. Cambridge, Massachusetts: MIT Press, 98-129.

Radford, A. 2009. Analyzing English Sentences: A Minimalist Approach. Cambridge: CUP.

Rizzi, L. 1990. Relativized Minimality. Cambridge, Massachusetts: MIT Press.

Strik, N. 2007. L’acquisition des phrases interrogatives chez les enfants francophones. Psychologie française 52: 27-39.

Stromswold, K. 1995. The Acquisition of Subject and Object Wh-Questions. Language Acquisition 4(1): 5-48.

Thornton, R. 1990: Adventures in long-distance moving: the acquisition of complex wh-questions. PhD Dissertation, University of Connecticut.

Thornton, R. and S. Crain. 1994. Successful cyclic movement. In T. Hoekstra and B. Schwartz (eds.), Language Acquisition Studies in Generative Grammar. Amsterdam, Philadelphia: John Benjamins, 215-252.

Yamane, M., D. Chen and W. Snyder. 1999. Subject-object asymmetries and Children’s Left Branch Extractions. BUCLD Proceedings 23(2): 732-740.

2 1 8 Language Acquisition Common Characteristics in Compound Formation: Evidence from Bilingual Acquisition and L2 Language Learning

Styliani S. Kalligiannaki and Marina Tzakosta

University of Crete [email protected], [email protected]

Abstract

The aim of the paper is to explore the mechanisms underlying compound formation in Greek L1 and L2. The focus is on Albanian, Russian and Swedish speakers of Greek. The data of these L2 groups are cross-checked with data of native speakers in order to assess the proficiency level of learners of Greek as well as the degree to which these learners may be characterized as bilinguals. The data reveal that learners of Greek employ the same mechanisms as native speakers do in the formation of compound forms and provide high production scores, though not as high as native speakers. We further assume that typological language adjacency determines language proficiency level.

1. Introduction

Compounding provides important information regarding the order of acquisition of word formation processes as well as the characteristics of a languages vocabulary. More specifically, compounding presupposes the acquisition of inflection and derivation, since, on the one hand, inflection is an obligatory process contra to derivation and compounding which are non-obligatory processes, and, on the other hand, compounding is a form of derivation, though even more complex than derivation (cf. Tzakosta 2009, 2010, 2011, in press). According to Selkirk (1982) compounds in English and cross-linguistically are word structures usually consisting of two major constituents belonging to the category of either a N(oun), an A(djective), a V(erb) or a P(reposition). It has also been argued that compound heads are located at the right edge of the newly formed form (Williams 1981). There are plenty of theoretical studies on Greek compounding focusing on, first, the internal structure of compound forms with emphasis onthe

Language Acquisition 2 1 9 Major Trends in Theoretical and Applied Linguistics

determinant role of heads, second, the existence of linking vowels and, third, the relation holding between compound elements (cf. Ράλλη 1999, Nespor & Ralli 1996, Ralli 2005). For Greek, it has been proposed that compounds belong to two major categories, namely, they may be lexical or morphosyntactic. Lexical compounds are of two types; they may either have the structure [[stem + stem] + derivational suffix] or the structure [stem + word].1 Representative data of both structures are given in (1a) and (1b), respectively:

(1a) pali-ό-filos ‘old friend-MASC.NOM.SG.’ (1b) pali-ο-fílos ‘old friend-MASC.NOM.SG.’

In (1a) none of the compound constituents retains all of its morphophonological properties. More specifically, only the stem of the first constituent is preserved, while the second constituent preserves its grammatical but not all of its prosodic properties; stress is moved to a neutral element of the newly formed word, namely, the linking vowel. Stress readjustment is a fundamental characteristic of [s + s] compounds. (1b), on the other hand, is an example of [s + w] compounds in which the second constituent, which also happens to be the head of the word, retains all of its morphophonological characteristics, especially its stress. Morphosyntactic compounds consist of two words not combined in one form. Both constituents retain their morphophonological characteristics, i.e. the number of their syllables, their stress pattern, their grammatical category and other grammatical characteristics, such as gender, case and number. The constituents of morphosyntactic compounds may be separated by a hyphen but not another word. Therefore, (2a) is a possible morphosyntactic compound but (2b) is not. In this study, the focus is on lexical compounds.

(2a) to pedί-thávma ‘the wonder child – NEUT.NOM.SG.’ (2b) to pedί to thávma ‘(the) wonder (the) child – NEUT.NOM.SG.’

The linking element – ο – is considered to be a relic from ancient Greek (cf. Anastasiadi-Simeonidi 1983, Ralli & Raftopoulou 1999, Ralli 2005, 2007). Linking elements appear in compound forms cross-linguistically, though they take different shapes in languages other than Greek (cf. Tzakosta 2009, 2010, 2011, in press). In Greek it takes the shape of a vowel and usually appears

1 From this point on [[stem + stem] + derivational suffix] compounds are referred to with the abbreviation [s + s] and [stem + word] compounds are referred to with the abbreviation [s + w].

2 2 0 Language Acquisition Styliani S. Kalligiannaki and Marina Tzakosta

when the second compound constituent starts with a consonant, as shown in (3).2

(3) pefk-ό-dasos/ pefk-ο-dásos ‘pine forest-NEUT.NOM.SG.

Finally, like in other languages, Greek lexical compounds’ heads emerge at the right edge of the word no matter which compound type they belong to (Drachman & Malikouti-Drachman 1994; Malikouti – Drachman 1997; Nespor & Ralli 1994, 1996). The aim of the present study is to assess the proficiency level of different groups of L2 learners of Greek as opposed to Greek preschool children, i.e. children who are still in the process of acquiring Greek as a mother language. It is important to note that the learners’ proficiency level is defined on the basis of the experimental data and not any tools of proficiency assessment. The remainder of the paper is organized as follows: section 2 reviews previous theoretical and experimental findings on Greek compounding. Sections 3, 4 and 5 are short introductions to the principles governing compounding in Albanian, Russian and Swedish, respectively. Section 6 discusses the experimental results, while section 7 concludes the paper.

2. Previous Research and Goals of the Present Study

Experimental data from adult native speakers of Greek (Tzakosta 2009) and Dutch, German and Turkish learners of Greek (Tzakosta 2010, 2011, in press) have highlighted that both native speakers and L2 learners of Greek - irrespective of their linguistic backgrounds - provide equivalent results regarding the strategies which govern compound formation and the statistical frequencies of the latter. This entails that native speakers and L2 learners of a language draw from the same pool of repair mechanisms empowering the role of Universal Grammar (hereafter UG) in language acquisition and language learning. In addition, good or low performance rates for L2 learners depend on the language proficiency level of the latter.

2 There are relatively sparse cases in which the linking vowel appears in Consonant _ Vowel environments, as in μεγαλ-ο-απατεώνας ‘big cheater – MASC.NOM.SG.’ However, such cases are considered to be semantically and morphologically less coherent compared to phonological environments in which the linking vowel is missing. See Ralli (2007) for relevant and detailed discussion.

Language Acquisition 2 2 1 Major Trends in Theoretical and Applied Linguistics

More specifically, native speakers and L2 learners of Greek exhibit ahigh degree of variation in the formation of [s + s] and [s + w] forms which are semantically indifferent and can be used interchangeably, as demonstrated in (4a). In addition, both groups display that head misperception leads to semantic ambiguity (4b) (see also Jarema et al. 1999; Kehayia et al. 1999; Tsapkini et al. 1999; Tzakosta 2009, 2010, 2011, in press).

(4a) pon-o-kefalos vs. kefal-ό-ponos ‘headache- MASC.NOM.SG.’ (4b) kreat-ό-rizo vs. riz-ό-kreas ‘meat & rice-NEUT.NOM.SG.’

The linking vowel appears in environments where it could be phonologically prohibited, i.e. in C/V_V contexts (5a). Finally, variation – also being the cause of head misperception - appears more systematically in novel compounds, i.e. in forms which are morphologically possible but semantically ambiguous (5b). It appears that variation in existing compounds is restricted by mnemonic mechanisms.

(5a) xion-o-ánθropos vs. xio-nánθropos ‘snowman-MASC.NOM.SG.’ (5b) aftj-ó-ponos vs. pon-aftí ‘ear-ache-MASC.NOM.SG.’

In the present study we rely on the above theoretical and experimental findings with the aims, first, to assess the proficiency of additional groups of native speakers and L2 learners and, second, to categorize the latter as bilinguals or L2 learners of Greek. More specifically, while Tzakosta (2009) tested 40 adult native speakers of Greek (age range 18-58 years) for the present study we recruited a group of five Greek preschoolers (age range 4-6 years) who served as controls, as well as six more preschoolers, four of whom are native speakers of Albanian (age range 20-30 years), one is a native speaker of Russian (30 years old) and one is a native speaker of Swedish (45 years old). We need to clarify that we are not interested in the time of exposure of L2 participants to Greek. In addition, we are not interested in whether, if bilinguals, our L2 participants are successive or simultaneous bilinguals. We rather rely on the experimental findings in order to prove whether they are bilinguals or not. Given the above, our working hypotheses are summed up in the following: first, we expect that if L2 speakers are true bilinguals, they will exhibit as high performance rates as native speakers in the experimental task, second, L2 language learners may exhibit equivalent but lower rates than native speakers regarding their performance in compound formation. Finally, we expect that true bilinguals will demonstrate higher rates in the novel compounds task compared to L2 learners. This will be evidence of the fact that bilinguals are capable of making productive use of various word formation mechanisms.

2 2 2 Language Acquisition Styliani S. Kalligiannaki and Marina Tzakosta

Although the tested groups, especially the Russian and Swedish which consist of only one participant, cannot lead us to safe findings and theoretical generalizations with respect to our research questions, we argue that they can provide us with some first hints regarding the order of compound formation in such and comparable populations. In addition, we are interested in these languages because, first, their (morphological) system has typological similarities to that of Greek (Russian and Swedish share more similarities with Greek, Albanian fewer), and, second, the tested native speakers are integrated in the Greek educational system. The testing material is an off-line task which takes the shape of two distinct questionnaires (adopted from Tzakosta 2009, 2010, 2011, in press) distributed to all groups. These two questionnaires demanded the formation of existing compounds (T1) and the formation of novel compounds which are morphologically possible but semantically vague (T2). Both questionnaires consist of 138 items. Representative examples of T1 and T2 are provided in (6a) and (6b), respectively.

(6a) péfko + dásos → pefkódasos AND pefkodásos ‘pine forest – NEUT. NOM.SG.’ (T1) (6b) velóna + dásos → velonódasos AND velonodásos ‘needle forest – NEUT.NOM.SG.’ (T2)

3. Compounding in Albanian

In sections 3 to 5 we briefly review the principles which govern compound formation in Albanian, Russian and Swedish, the languages which are related to our study. In Albanian compounding there are no length restrictions with respect to the number of syllables of the compound forms. In addition, compounds are written as two separate words with or without hyphens. There is sporadic use of the linking elements /a/ and /o/ (Orel 2000, Elsie 2006), but the conditions which determine their emergence are not clearly defined. Some representative data adopted from Newmark et al. (1982: 175) are given in (7).

(7a) gusht-o-vjeshtë < gusht, vjeshtë ‘end of August-beginning of September’ ‘August’, ‘autumn’ (7b) dash-a-mir < dash mirë ‘beloved’ ‘love’, ‘good’

The fact that Albanian compounds are written as two separate words entails that they do not undergo stress readjustment, unlike Greek (Orel 2000, Elsie 2006).

Language Acquisition 2 2 3 Major Trends in Theoretical and Applied Linguistics

4. Compounding in Russian

Russian compounding shares more properties with Greek compound formation than Albanian. More specifically, Russian has one and two-word compounds. In one-word compounds the use of the linking vowels /e/ and /o/ are reported, as illustrated in (8). There is only one main stress which is carried by the compound head. In Russian, like in Greek, compounds are right-headed (cf. Gouskova & Roon 2009).

(8) ùglj-e-do-bitʃ-a < ùg(o)lj-Ø, do-bítʃ-a ‘coal’, ‘mine’ ‘coal mine’ (adopted from Cubberley 2002: 270)

5. Compounding in Swedish

In Swedish, compounding is a very productive and recursive word formation process (Friberg 2008). In other words, two or more word roots may be involved in the newly formed word. Like Russian and Greek, the rightmost constituent is the head of the compound. In one-word compounds, there are two stress peaks, namely a primary and a secondary one. In addition, Swedish makes extensive use of linking elements. These are /s, e, a, u, o, and Ø/ (Hedlund 2002). Representative examples of compound formation in Swedish are provided in (9) below.

(9a) järnindustri < jarn + industri ‘iron industry’ (9b) skol+bok-s+hylla ‘book shelf for school books’

6. Results and Discussion

In what follows we present the statistical analyses of the data and make some theoretical proposals. Table 1 summarizes the frequency rates for answered (A) and correctly answered (√-A) questions. It becomes obvious that native speakers exhibit much higher rates for both answered and correctly answered questions compared to all L2 tested groups. However, Russian and Swedish learners perform much better than Albanian speakers of Greek. It is interesting to notice that, although native speakers score high with respect to their answered questions, they do not score as high in their correctly answered questions. Apparently, native speakers are positive in answering the questions of both T1 and T2, though they make many mistakes. On the other hand, L2 learners are more reluctant to answer the questions, though they perform well when it comes to their correctly answered rates.

2 2 4 Language Acquisition Styliani S. Kalligiannaki and Marina Tzakosta

Table 1. Rates of answered and correctly answered questions

Nom A √-A A √-A A √-A A √-A

Native Albanian Russian Swedish 51,4 T1 73% 70% 37,4% 72,3% 83% 50% 56% % T2 88% 74% 34,9% 43,1% 57% 70% 60% 71,4%

Total 80,5% 72% 36,1% 57,7% 54,3% 76,5% 55% 63,7%

Table 2 provides a first general picture of the performance of all groups regarding the major tested variables, namely, right-headedness (RH), the position and the use of the linking vowel (LV), the preferred compound types (CT), and repairs that are involved in compound formation, such as stress readjustment (Stress R) and syllable reduction (SR).

Table 2. Tested variables and compound formation repairs

R-H LV CT Stress-R SR

Greek √ √ √ √ √

Albanian √ √↓ √ x x Russian √ √ √ x x

Swedish √ √↑ √ x x

All groups satisfy the RH, LV and CT variables; however, the Albanian group scores badly when it comes to the use of the LV while the swedish group scores well. This is expected given that in Albanian the emergence of the LV is rare and in Swedish there is extensive use of various linking elements. The repairs of stress-R and SR appear only in the native speakers’ group. It seems that L2 learners prefer to consistently preserve the stress of one of the two compound constituents in the compound form rather than readjust stress in the newly formed form. Besides, stress readjustment is not operative in either Albanian, Russian or Swedish. Table 3 illustrates the rates of native and L2 speakers with respect to the position of compound heads ((R)ight or (L)eft). As already discussed, morphophonological heads tend to appear at the right edge of the word cross-linguistically (Williams 1981). The data seem to support this claim. More specifically, native speakers place heads at the right edge of existing compounds across-the-board. The rate remains very high (R, 88%) for non-existing/ novel compounds. Albanian, Russian and Swedish learners exhibit high rates of right-

Language Acquisition 2 2 5 Major Trends in Theoretical and Applied Linguistics

headedness for existing compounds, though the rate falls radically for non- existing compounds. All L2 findings are equivalent.

Table 3. Head position

Heads R L R L R L R L

Native Albanian Russian Swedish

T1 100% 0% 75% 25% 81% 19% 84% 16%

T2 88% 12% 40% 60% 44% 66% 45% 65%

Total 94% 6% 57,5% 42,5% 62,5% 42,5% 64,5% 40,5%

The linking vowel (LV) appears to be an important cue for compound formation in Greek L1 and L2. This is evident by the fact that native speakers not only prefer (Pr) to use the LV in 90% of the cases in T1, but the rate of LV use is even higher in T2 (96% of the cases). Albanian speakers of Greek employ the LV more frequently in T1 than T2. However, the Russian and Swedish speakers, like Greek native speakers, use the LV more often in T2 than T1. As already mentioned, the LV is an important indicator of compound formation in Russian and Swedish but not in Albanian. This may give an answer as to why the LV is less frequently found in the data of the Albanian speakers of Greek.

Table 4. Use of the linking vowel (LV)

LV Pr N-Pr Pr N-Pr Pr N-Pr Pr N-Pr

Native Albanian Russian Swedish

T1 90% 10% 84% 16% 86% 14% 82,3% 17,7%

T2 96% 4% 69% 31% 95% 5% 95,3% 4,7%

Total 93% 7% 76,5% 23,5 90,5% 9,5% 88,8% 11,2%

The picture is vague with respect to the preference of specific compound types over others. More specifically, native speakers preferably form [s +w] compounds in both T1 and T2. This finding goes contra the findings of Tzakosta (2009) according to which adult native speakers of Greek prefer to form [s + s] compound words. [s + s] forms presuppose the knowledge of the mechanism of stress readjustment, as shown in (1a) above, while in [s + w] forms the second constituent, which also happens to be the head of the compound, retains its morphophonological characteristics, especially its stress position. Therefore, it

2 2 6 Language Acquisition Styliani S. Kalligiannaki and Marina Tzakosta

is expected that [s + w] rather than [s + s] compounds are easier to be realized by preschool native speakers who may not have already acquired all word formation mechanisms. On the contrary, adult native speakers are in a better position to master complex word formation mechanisms; therefore they prefer to form [s + s] forms. Similarly to the preschool native speakers’ data, the Russian and Swedish speakers prefer [s + w] to [s + s] compounds, though Albanian speakers prefer [s + s] compounds for both T1 and T2.

Table 5. Preferred compound types (CT)

CT [s+ w] [s + s] [s + w] [s + s] [s + w] [s + s] [s + w] [s + s]

Native Albanian Russian Swedish

T1 51% 49% 41%% 59% 40% 60% 76,4% 23,6%

T2 61% 39% 37% 63% 60% 40% 52,3% 38%

Total 56% 44% 39%% 61% 50% 50% 64,3% 44%

In relation to the above table, table 6 presents the frequencies of correct (R) and non-correct (W) stress readjustment in compound words. Depending on the preferred compound type, native speakers readjust stress correctly (R) in 98% in both T1 and T2. Albanian speakers perform better in T1 than T2. It is interesting that, although in Swedish two stress peaks appear in compound forms, the Swedish speaker readjust stress correctly across-the-board in both T1 and T2; so does the Russian speaker.

Table 6. Stress readjustment

Stress-R R W R W R W R W

Native Albanian Russian Swedish

T1 98% 2% 82% 18% 85,7% 14,5% 100% 0%

T2 98% 2% 62,5% 37,5% 100% 0% 100% 0%

Total 98% 2% 72%% 27% 93% 7,3% 100% 0%

Finally, syllable reduction (SR) is a strategy non-frequently employed in any of the participating groups. Apparently, native speakers and L2 learners, although of preschool age, have passed off the stage of syllable reduction being used as a process which facilitates language development. Stress reduction was never activated for L2 learners either.

Language Acquisition 2 2 7 Major Trends in Theoretical and Applied Linguistics

Table 7. Syllable reduction

SR Pr N-Pr Pr N-Pr Pr N-Pr Pr N-Pr

Native Albanian Russian Swedish

T1 2% 98% 0,3% 99,7% 5,7% 94,3% 3% 97%

T2 0% 100% 0% 100% 0% 100% 0% 100%

Total 1% 99% 0,15% 98,5%% 2,85% 97,2 1,5% 98,5%

The data in (10) are representative cases of syllable reduction. The second piece of data in (10a) and the first in (10b) are reduced regarding the number of their syllables.3 Both examples in (10a) are equally acceptable, while the first example in (10b) is not acceptable. In both cases the reduced syllable belongs to the first constituent. Put differently, the head of the compound retains all of its morphophonological properties because of the firmed character of the former.

(10a) a.ste.r-o-ci.ni.gós (N6) vs. a.str-o-.ci.ni.gós (N5) ‘star hunter – MASC.NOM.SG.’ (10b) kli.d-ó-.tri.pa (N4) vs. kli.da.r-ó-.tri.pa (N5) ‘key hole – FEM.NOM.SG.’

7. Conclusions

The aim of the present study was to assess the capacity of Greek native speakers of preschool age and the proficiency level of Albanian, Russian and Swedish L2 learners of Greek in order to investigate whether L2 speakers could potentially be (or become) bilinguals. The data lead us to two major findings; first L2 groups provided equivalent results to L1 preschool acquirers of Greek, though they did not score as high as native speakers. Second, there is clear variation in the scores of different L2 groups; more specifically, the Russian and Swedish speakers performed better than Albanian learners in the formation of both existing (T1) and non-existing (T2) compounds. The discussion regarding the general characteristics of compound formation in Greek, Albanian, Russian and Swedish revealed that Greek is typologically more adjacent to Russian and Swedish than Albanian. In other words, typological similarity is responsible for the fact that the

3 The number of syllables is indicated in the parentheses. Syllable boundaries are indicated by dots.

2 2 8 Language Acquisition Styliani S. Kalligiannaki and Marina Tzakosta

Swedish and Russian speakers perform better in forming compound words than the Albanian speakers. Therefore, the closer two languages are regarding their morphological synthesis the easier and faster they are learned or acquired. This entails that the inter-typological language adjacency should be seriously taken into consideration in the design and substantiation of the relevant teaching material and teaching methods. As far as our main research question is concerned, namely whether our non native participants are bilinguals or L2 learners, we assume that the scores of non native participants cannot lead us to safe theoretical claims. In addition to that, we argue that typological language adjacency in combination with the fact that the participants of all groups are preschoolers provides them with the possibility to develop as bilinguals with time. To be more specific, although L2 learners exhibit lower scores than native speakers, they are potential bilinguals due to the fact that, as preschoolers, they are still in the process of language acquisition. As a result, adequate exposure to the relevant linguistic stimuli in combination with the benefit of language structural adjacency will help L2 groups to develop into true bilinguals and not into mere L2 learners. We recognize that our experimental groups, especially the Russian and Swedish groups, consist of the minimal possible number of participants; therefore our findings and our claims are difficult to be generalized. However, we believe that even such small groups provide us with the necessary basis in order to expand our research.

Language Acquisition 2 2 9 Major Trends in Theoretical and Applied Linguistics

References

Anastasiadi-Simeonidi, A. 1983. La Composition en Grec Moderne d’un Point de Vue Diachronique. Lalies 2: 77-90.

Cubberley, P. 2002. Russian: A Linguistic Introduction. Cambridge: Cambridge University Press.

Drachman, G. and A. Malikouti-Drachman. 1994. Stress and Greek Compounding. Phonologica 1992: 55-64.

Elsie, R. 2006. Albanian Literature: A Short History. London: I.B. Taurus.

Friberg, K. 2008. Decomposing Swedish Compounds Using Memory-Based Learning. NODALIDA 2007 Conference Proceedings: 224-230.

Gouskova, M. and K. Roon. 2008. Interface Constraints and Frequency in Russian Compound Stress. In J. Reich, M. Babyonyshev and D. Kavitskaya (eds.), Proceedings of the 17th Meeting of Formal Approaches to Slavic Linguistics. Ann Arbor: Michigan Slavic Publishers, 49-63.

Hedlund, T. 2002. Compounds in dictionary-based cross-language information retrieval. Information Research 7(2).

Jarema, G., C. Busson, R. Nikolova, K. Tsapkini, and G. Libben 1999. Processing Compounds: A Cross-Linguistic Study. Brain and Language 68: 362-369.

Kehayia, E., G. Jarema, K. Tsapkini, D. Perlak, A. Ralli and D. Kadzielawa. 1999. The Role of Morphological Structure in the Processing of Compounds: The Interface between Linguistics and Psycholinguistics. Brain and Language 68(1-2): 370-377.

2 3 0 Language Acquisition Styliani S. Kalligiannaki and Marina Tzakosta

Malikouti-Drachman. A. 1997. Prosodic Domains in Greek Compounding. In G. Drachman, A. Malikouti-Drachman, S. Klidi and J. Fykias (eds.), Proceedings of the 2nd International Conference of Greek Linguistics. Graz: Neubaurer Verlag, 87-96.

Nespor, M. and A. Ralli. 1994. Stress Domains in Greek Compounding. In I. Philippaki-Warburton, K. Nicolaidis and M. Sifianou (eds.), Themes in Greek Linguistics I. Amsterdam: John Benjamins, 201-208.

Nespor, M. and A. Ralli. 1996. Morphology-Phonology Interface: Phonological Domains in Greek Compounds. Linguistic Review 13(3-4): 357-382.

Newmark, L., P. Hubbard and P. Prifti. 1982. Standard Albanian: A Reference Grammar for Students. Stanford: Stanford University Press.

Orel, V. 2000. A Concise Historical Grammar of the Albanian Language: Reconstruction of Proto-Albanian. Boston: Brill Academic Publishers.

Ράλλη, Α. 1999. To Φαινόμενο της Σύνθεσης στη Νέα Ελληνική: Περιγραφή και Ανάλυση. Παρουσία ΙΑ-ΙΒ: 183-205.

Ralli, A. 2005. Μορφολογία. Athens: Patakis.

Ralli, A. 2007. Η Σύνθεση Λέξεων: Διαγλωσσική Μορφολογική Προσέγγιση. Athens: Patakis.

Ralli, A. and M. Raftopoulou. 1999. Η Σύνθεση ως Διαχρονικό Φαινόμενο Σχηματισμού Λέξεων. Studies in Greek Linguistics 20: 389-403.

Selkirk, E.O. 1982. The Syntax of Words. MIT Monograph Seven.

Tsapkini, K., E. Kehayia and G. Jarema. 1999. Does Phonological Change Play a Role in the Recognition of Derived Forms across Modalities? Brain and Language 68: 318-323.

Tzakosta, M. 2009. Perceptual Ambiguities in the Formation of Greek Compounds by Native Speakers. In Giannakis, G.K., M. Baltazani, G.I. Xydopoulos and A. Tsaggalidis (eds.), Electronic Proceedings of the 8th International Conference of Greek Linguistics. Department of Philology: University of Ioannina, 545-557 (ISBN: 978-960-233-195-8).

Tzakosta, M. 2010. External and Internal Factors Affecting Compound Formation in L2: the Case of Dutch Learners of Greek. Proceedings of the 30th Annual

Language Acquisition 2 3 1 Major Trends in Theoretical and Applied Linguistics

Meeting of Greek Linguistics. Department of Philology: Aristotle University of Thessaloniki.

Tzakosta, M. 2011. L1 Transfer in L2 Learning: Compound Forms in the Speech of Turkish Learners of Greek. In E. Kitis, N. Lavidas, N. Topintzi and T. Tsangalidis (eds.), Selected papers from the 19th International Symposium on Theoretical and Applied Linguistics (19th ISTAL). Department of English Studies. Aristotle University of Thessaloniki. Thessaloniki: Monochromia, 459-468.

Tzakosta, M. in press. L1 Transfer in L2 Word Formation. Electronic Proceedings of the 9th International Conference of Greek Linguistics. Department of Linguistics: University of Chicago.

Williams, E. 1981. On the notions ‘lexically related’ and ‘head of the word’. Linguistic Inquiry 12(2): 245-276.

2 3 2 Language Acquisition The Production of Aspect by L1 Learners of Greek

Maria Kaltsa

Aristotle University of Thessaloniki [email protected]

Abstract

This study examines the production of aspect and telicity by children acquiring Greek as a native language. Telicity is compositionally determined by the aspectual class of the verb, morphological aspect and the presence of a quantized object. A sentence is assumed to refer to a telic event if the event is represented as having an endpoint beyond which the event cannot continue (Depraetere 2007). The aim of the present study is to identify the nature of the relation between aspect and telicity by comparing evidence drawn from the developmental steps monolinguals take in expressing telicity in activity and motion events.

1. Introduction

Telicity lies at the syntax-discourse interface and is determined by the interaction of different factors, that is (a) the aspectual class of the verb (cf. Vendler 1967, Dowty 1979, Smith 1991, Ramchand & Svenonius 2002, Rothstein 2002), (b) morphological aspect and (c) the presence or absence of objects and particles (cf. van Hout 2003), where factors (b) and (c) are language specific. The interaction of these factors offer information about the internal temporal organisation of an event and lead to a compositional understanding of aspectual meaning. Telicity being a by-product of the aspectual marking of VPs has been controlled by a variety of language-specific criteria such as the adverbial modification test, the conjunction test and the progressive entailment test (see Borik 2006). Vendler’s classification of verbs distinguishes between states and activities being atelic from achievements and accomplishments being telic (Vendler 1967, Dowty 1979). Smith (1991) discusses the shift from one situation type (i.e. aspectual verb class) to the other as in build, an activity, to build a house, an accomplishment. This shift renders the atelic verb into a telic predicate. A

Language Acquisition 2 3 3 Major Trends in Theoretical and Applied Linguistics

sentence receives a telic interpretation if the event is represented as having an endpoint beyond which the event cannot continue (Depraetere 2007). The manifestations of an endpoint in Depraetere’s view are the lexical semantics of the verb, (e.g. find, sneeze, explode), contextual knowledge shared by discourse participants (e.g. He sang. understood as He sang a song), specific object NPs which provide the natural endpoint to the situation or via the inclusion of a cardinal number (e.g. drink a glass of wine) and the presence of a PP-GOAL which acts as a boundary of the action denoted by the verb (e.g. He drove the car into the garage). In Greek, the visibility of the endpoint may be established within the sentence and prior to pragmatic interpretation through lexical and grammatical means. For example, the Greek sentence Zografise ena portreto (“He painted a portrait”) is unambiguously telic in that the portrait is completed while the sentence Zografize ena portreto (“He was painting a portrait”) leaves open the possibility that the portrait is still not completed (Tsimpli & Papadopoulou 2006). The difference lies on the aspectual marking of the VP. Since perfectivity makes visible the endpoint of the event (Smith 1991) the presence of the object serves to identify the endpoint with the completion of the object painting event. The imperfective aspect, on the other hand, does not make visible the endpoint of the event and leaves open the possibility of identifying the completion. Thus, in Greek telicity can be unambiguously defined within the sentence with the combination of perfective aspect and a specific object DP (cf. Mozer 1994; Chila-Markopoulou & Mozer 2001). Furthermore, even when telicity is unambiguously defined within the sentence, this is the result of a compositional interpretation of the aspectual form and the quantized object or a PP-GOAL. In line with Zubizarreta & Oh (2007), we assume that motion verbs may include an inherent ‘directed motion’ feature by selecting a PP-GOAL complement. Consider the following examples:

(1) a. To agori etrehe stin kuzina. the boy ran-IMP in-the kitchen “The boy was running in/towards the kitchen. b. To agori etrekse stin kuzina. the boy ran-PERF in-the kitchen “The boy ran into/towards/inside the kitchen.”

Each of the two sentences in (1a) and (1b) is ambiguous between a telic and an atelic interpretation. Specifically, in (1a) the PP stin kuzina can be construed as the GOAL argument or as the LOCATIVE modifier of the motion event. In (1b), on the other hand, the ambiguity increases since the telic interpretation in which the PP-GOAL is interpreted as the endpoint of the

2 3 4 Language Acquisition Maria Kaltsa

event is also available. Thus, in (1a) the event is unbounded and therefore atelic while in (1b) the telic interpretation is also available. Despite the ambiguity just described, native speakers develop a strong preference for the atelic locative interpretation in (1a) and the telic in (1b) (Tsimpli & Papadopoulou 2009). These preferences are associated with the aspectual distinction on the verb form in each case. Studies in first language acquisition have shown that telicity particles in languages such as Dutch are acquired early (van Hout 1998, 2000). Van Hout (1998, 2000) tested typically-developing Dutch children on the comprehension of telicity showing that by age 3 children have acquired the telic interpretation of particle verbs in Dutch. Wagner (2010) reviews the evidence for the acquisition of telicity in various languages and suggests that initially children show a preference for mapping telicity onto aspectual morphemes, particles or PPs. This preference extends to associating goal-oriented interpretations with transitivity. Wagner (2006) included directed motion verbs with a goal PP in the acquisition study and found that even the youngest children (2-year olds) were accurate in interpreting telicity when a PP-GOAL was included in the sentence. In L1 Greek, Kaltsa’s comprehension study on telicity (2012) shows that (a) perfective activity VPs with a DP complement precede perfective motion VPs with a PP complement in development; and (b) the interpretation of telicity becomes adult-like between 7 to 8 years old. The delay in the Greek data compared to findings in other languages such as Dutch (van Hout 1998, 2000) and German (Schultz & Wittek 2003), where the age of 3 is considered a turning point in the acquisition of telicity, can be attributed to the fact that telicity is not lexicalized in Greek and maintains an interface status.

2. Predictions

The aim of the empirical study is to investigate the role of the aspectual verb form (Perfective vs Imperfective) and the nature of the complements (DPs vs PPs) in the production of activity and motion VPs by monolingual Greek children and adults. We adopt the notion that the acquisition of syntax precedes the acquisition of features that lie at the syntax-discourse interface; the acquisition of morphological aspectual distinctions is in place by the age of 3 (Katis 1984, Tsimpli 1996). Consequently, the acquisition of the morphosyntax of aspect precedes the acquisition of telicity which is determined at the interface level. The working hypotheses are that a) activity verbs will precede motion verbs in acquisition as a result of the complements they receive and b) imperfective motion verbs will be acquired earlier than perfective ones due to the underspecification of telicity in the latter.

Language Acquisition 2 3 5 Major Trends in Theoretical and Applied Linguistics

3. The Production Study

3.1. Participants

Five groups of monolingual learners of Greek participated in the study. There were 250 participants in total aged from 5 to 10 years old. For each age group there were 50 participants. At the time of testing the 5-6 yrs old group had a mean age of 5;5 (Female: 21 & Male: 29), the 6-7 yrs old group a mean age of 6;5 (Female: 23 & Male: 27), the 7-8 yrs old one a mean age of 7;5 (Female: 26 & Male: 24), the 8-9 yrs old group a mean age of 8;6 (Female: 24 & Male: 26) and the 9-10 yrs old group a mean age of 9;3 (Female: 26 & Male: 24). All participants were typically developing learners of Greek and did not suffer from any language or mental disorder. Additionally, a group of 40 adult native speakers of Greek with a mean age of 36;7 (Female: 27 & Male: 13) was included for control purposes.

3.2. Materials

The production task consisted of 64 short video stimuli. For each verb, we constructed two videos, one presenting a telic/complete event and the other depicting an atelic/ongoing event. To avoid repetition effects two versions of the task were created with 32 video stimuli each, so that the participants never saw the same verb-event/video more than one time in each session. There were 18 target verbs (6 intransitive, 6 activity and 6 motion verbs) and 14 filler verbs. The activity verbs were zografizo (‘paint’), dhiavazo (‘read’), troo (‘eat’), pino (‘drink’), katharizo (‘clean’) and ftiahno (‘make’). The motion verbs were strifogirizo (‘turn around’), sernome (‘crawl’), treho (‘run’), pidho (‘jump’), pigeno (‘go’) and horopidho (‘gambol’). The test verbs were selected on the basis of the results of two pre- tasks, an acceptability judgment task and a verb frequency task (for a detailed account see Kaltsa 2012: 114-118).

3.3. Procedure

The participants were shown one video at a time presenting either the telic or atelic condition and a test question followed. The test question was not morphologically marked with aspect (2) and it included a time adverbial to indicate past reference:

(2) Τι έκανε χθες το παιδί; what do-PAST yesterday the-NOM child “What did the child do yesterday?”

2 3 6 Language Acquisition Maria Kaltsa

The task for the participant was to describe the event. Examples of each targeted VP construction may be seen in (3) and (4):

(3) Activity VP (a) Hthes to koritsi efage ena milo. yesterday the girl eat-PERF an apple ‘Yesterday the girl ate an apple.’ (b) Hthes to koritsi etroge ena milo. yesterday the girl eat-IMP an apple ‘Yesterday the girl was eating an apple.’

(4) Motion VP (a) Hthes to agori etrekse sti kuzina. yesterday the boy run-PERF o/in the kitchen ‘Yesterday the boy ran to the kitchen.’ (b) Hthes to agori etrehe sti kuzina. yesterday the boy run-IMP inside the kitchen ‘Yesterday the boy was running in the kitchen.’

In (3a), the perfective activity VP was considered as a target utterance for a telic event, that is a completed event where the goal has been accomplished (the whole apple has been consumed). In the case of the imperfective activity VP (3b), the utterance was accepted as a description of an atelic event during which the goal has not been achieved (the apple remains uneaten). In (4a), the perfective motion VP with a PP-GOAL was considered as a target utterance for a telic motion event. In the case of the imperfective motion VP (4b), the utterance was accepted as a description of an atelic event.

3.4. Results

For the data analysis of the production experiment we filtered out utterances that were not morphological marked as past and / or were not lexically related to the verb-target. For the rest of the responses we examined the use of perfective and imperfective aspectual verb forms and the use of overt or null DPs and PPs for activity and motion verbs respectively when describing telic events.

Activity Events and Aspect For the event type comparisons (telic vs. atelic) in the aspectual marking of activity verb constructions we performed two-way group-independence chi- square tests. The comparisons show that the event type the participants were shown significantly affects the aspectual choice of the 7 yrs old and older

Language Acquisition 2 3 7 Major Trends in Theoretical and Applied Linguistics

children. Adults also show this effect. The two youngest age groups, though, do not appear to be equally sensitive to the role of telicity (5-6 yrs old: χ²=.669, df= 1, p=.413; 6-7 yrs old: χ²=.500, df= 1, p=.480; 7-8 yrs old: χ²=7.386, df= 1, p=.007; 8-9 yrs old: χ²=9.589, df= 1, p=.002; 9-10 yrs old: χ²=127.396, df= 1, p=.000; Adults: χ²=25.794, df= 1, p=.000). Figure 1 presents the percentages of the aspectual marking for activity VPs of each age group when they were provided with a telic event.

100% 90% 80% 70% 60% Imperfective 50% P erfective 40% 30% 20% 10% 0% 5-6 yrs old 6-7 yrs old 7-8 yrs old 8-9 yrs old 9-10 yrs old A dults

Figure 1. Aspectual Marking: Activity Telic Events

The data reported in Figure 1 show that the 5-6 yrs old group along with the 6-7 yrs old group regardless of the fact that they are given a telic event they significantly prefer to use imperfective aspect. The 7-8 yrs old group appears to be undetermined and performs at chance level. The 8-9 and 9-10 yrs old groups, however, show adult-like performance in selecting perfective aspect to describe telic events. These observations were statistically supported by one- way-goodness-of-fit chi-square tests (5-6 yrs old: χ²=32.508, df= 1, p=.000; 6-7 yrs old: χ²=16.078, df= 1, p=.000; 7-8 yrs old: χ²=2.032, df= 1, p=.154; 8-9 yrs old: χ²=4.829, df= 1, p=.028; 9-10 yrs old: χ²=71.201, df= 1, p=.000; Adults: χ²=14.286, df= 1, p=.000). These findings indicate that perfective aspect is used to mark telicity for activity VPs. Turning to the atelic condition, Figure 2 presents the percentages of the aspectual choice for activity VPs of each age group when they were presented with an atelic event. The data reported in Figure 2 show that all groups uniformly select imperfective aspect to describe atelic activities. This observation was statistically supported by one-way-goodness-of-fit chi-square tests (5-6 yrs old: χ²=23.403, df= 1, p=.000; 6-7 yrs old: χ²=23.432, df= 1, p=.000; 7-8 yrs old: χ²=5.568, df= 1, p=.018; 8-9 yrs old: χ²=4.761, df= 1, p=.029; 9-10 yrs old: χ²=56.427, df= 1,

2 3 8 Language Acquisition Maria Kaltsa

p=.000; Adults: χ²=11.571, df= 1, p=.001). It appears that imperfective aspect is mapped on atelicity for activity verb constructions early on in acquisition.

100%

80%

60% Imperfective P erfective 40%

20%

0% 5-6 yrs old 6-7 yrs old 7-8 yrs old 8-9 yrs old 9-10 yrs old A dults

Figure 2. Aspectual Marking: Activity Atelic Events

The between group comparisons for the telic condition show very clearly the developmental steps of telicity in the aspectual marking of activity VPs. The 5-6 and 6-7 yrs old children perform alike and the 7-8 and 8-9 yrs old ones similarly to each other. When compared to adults, children are found not to have attained the target performance with the exception of the 8-9 yrs old group. Quite importantly, the differences to the adult performance are not always in the same direction for all child groups. The youngest groups’ (5-8 yrs old) use of imperfective is high but it gradually lowers getting closer to the adult preference for perfective. The oldest group (9-10 yrs old), on the other hand, is significantly different to adults not because it selects the non-target aspect, that is imperfective, but due to the fact that it selects perfective aspectual marking more robustly. Thus, it is in the same direction with adults mapping telicity on perfective activities. In the atelic condition, the differences among the groups are more subtle and the preferences of each groups uniform. To summarize, the examination of aspectual marking for activity verbs makes evident that (a) telicity affects the aspectual choices of children and adults, (b) perfective aspect is selected for telic events adult-like by children from the age of 8 and (c) imperfective aspect is used for atelic activities by children as young as 5 yrs old.

Motion Events and Aspect For the event type comparisons (telic vs. atelic) in the aspectual marking of motion verb constructions we performed two-way group-independence chi-

Language Acquisition 2 3 9 Major Trends in Theoretical and Applied Linguistics

square tests. The comparisons show that the event type the participants were shown significantly affects the aspectual choice of all child groups and adults (5-6 yrs old: χ²=10.919, df= 1, p=.001; 6-7 yrs old: χ²=10.989, df= 1, p=.001; 7-8 yrs old: χ²=7.614, df= 1, p=.006; 8-9 yrs old: χ²=21.658, df= 1, p=.000; 9-10 yrs old: χ²=34.998, df= 1, p=.000; Adults: χ²=20.198, df= 1, p=.000). Figure 3 presents the percentages of the aspectual marking for motion VPs of each age group when they were provided with a telic event.

100% 90% 80% 70% 60% Imperfective 50% P erfective 40% 30% 20% 10% 0% 5-6 yrs old 6-7 yrs old 7-8 yrs old 8-9 yrs old 9-10 yrs old A dults

Figure 3. Aspectual Marking: Motion Telic Events

The data reported in Figure 3 show that children appear to be confused over the aspectual marking of motion telic events, while adults significantly prefer using perfective aspect. In particular, the use of imperfective is predominant for the youngest child groups. The 6-7 yrs old group significantly selects imperfective aspect, while the rest of the child groups are undetermined and perform at chance level indicating the acquisitional difficulty posed by telic motions. These observations were statistically supported by one-way- goodness-of-fit chi-square tests (5-6 yrs old: χ²=2.701, df= 1, p=.100; 6-7 yrs old: χ²=5.538, df= 1, p=.019; 7-8 yrs old: χ²=3.115, df= 1, p=.078; 8-9 yrs old: χ²=2.843, df= 1, p=.092; 9-10 yrs old: χ²=.261, df= 1, p=.610; Adults: χ²=4.840, df= 1, p=.028). Figure 4 presents the percentages of the aspectual marking for motion VPs of each age group when they were provided with an atelic event. The data reported in Figure 4 show that, unlike the findings in the telic condition, in the atelic condition children perform adult-like is selecting the imperfective to describe the atelic motion events (5-6 yrs old: χ²=35.438, df= 1, p=.000; 6-7 yrs old: χ²=39.112, df= 1, p=.000; 7-8 yrs old: χ²=27.174, df= 1, p=.000; 8-9 yrs old: χ²=23.450, df= 1, p=.000; 9-10 yrs old: χ²=57.110, df= 1, p=.000; Adults: χ²=17.280, df= 1, p=.000).

2 4 0 Language Acquisition Maria Kaltsa

100% 90% 80% 70% 60% Imperfective 50% P erfective 40% 30% 20% 10% 0% 5-6 yrs old 6-7 yrs old 7-8 yrs old 8-9 yrs old 9-10 yrs old A dults

Figure 4. Aspectual Marking: Motion Atelic Events

The between group comparisons for the telic condition show the children’s performance is quite uniform. For the 6-7 yrs old group comparisons show a significant difference to the two oldest age groups (8-9 and 9-10 yrs old groups). Also, the 7-8 yrs old children appear to differ to the 8-9 yrs old ones. The age of 8 is of importance for the acquisition of telicity for motion verbs since it is from that age on that we get adult-like aspectual preferences by the children. In the atelic condition, there is a similar performance across the groups and the preference of imperfective mapping on atelic motion events appears set early on in acquisition. The 6-7 yrs old group differs to the 8-9 yrs old group and adults. Quite interestingly, as for activity verbs, the 9-10 yrs old group overuses imperfective aspect, which is the target aspectual choice, and, thus, appears to be significantly different to the 8-9 yrs old group and adults. The between verb-type comparisons per event type appear in Table 1. In the telic condition, the two verb categories are differentiated by children with the exception of 6-7 and 8-9 yrs old children, but crucially not by adults who appear not equally sensitive. In the atelic condition, both verb types are treated similarly by all groups with the exception of 7-9 yrs old children who significantly distinguish between the two. Overall, children are more responsive to VP differences. To summarize, the examination of aspectual marking for motion verbs makes evident that (a) telicity affects the aspectual choices of children and adults, (b) perfective aspect is selected for telic events by adults but children do not attain the adult performance not even up to the age of 10, (c) imperfective aspect is used for atelic activities by all child groups and adults and (d) children are sensitive to the verb type when selecting aspect for telic events. These findings are pointing to the difficulties motion verb constructions pose in acquisition and are in accordance to our predictions.

Language Acquisition 2 4 1 Major Trends in Theoretical and Applied Linguistics

Table 1. Between Verb-Type Comparisons: Aspectual Marking

Telic Event * Aspect Atelic Event * Aspect Groups Activity vs. Motion Verbs Activity vs. Motion Verbs

5-6 yrs old χ²= 8.019, df= 1, p=.005 χ²= 2.058, df= 1, p=.151

6-7 yrs old χ²= 1.235, df= 1, p=.267 χ²= 2.932, df= 1, p=.087

7-8 yrs old χ²= 5.130, df= 1, p=.024 χ²= 5.008, df= 1, p=.025

8-9 yrs old χ²= .089, df= 1, p=.765 χ²= 4.392, df= 1, p=.036

9-10 yrs old χ²=35.133 , df= 1, p=.000 χ²=.024 , df= 1, p=.876

Adults χ²= 1.087, df= 1, p=.297 χ²= .407, df= 1, p=.523

Telic Events and Complement Use In the following set of data, we examine the use of complement in the telic condition when it is paired with perfective aspect. Figure 5 presents the percentages of DP-use in perfective VPs of each age group when they were provided with a telic activity event.

100% 90% 80% 70% 60% null DP s 50% overt DP s 40% 30% 20% 10% 0% 5-6 yrs old 6-7 yrs old 7-8 yrs old 8-9 yrs old 9-10 yrs old A dults

Figure 5. Telic Activities: Perfective Aspect & Complement Use

The data reported in Figure 5 suggest that the 5-6 year old children avoid the use of overt complement DP for telic activities, while the 6-7 year old children show optionality in the overt vs. null use of DPs. From 7 years-old and onwards, however, the children acquire the adult performance in marking telicity with the use of overt complements as endpoints. These observations were statistically supported by one-way goodness-of-fit chi-square tests (7-8 yrs old:χ ²= 31.113, df= 1, p= .000; 8-9 yrs old: χ²= 7.530, df= 1, p= .006; 9-10 yrs old: χ²= 64.286, df= 1, p= .000; adults: χ²= 27.842, df= 1, p= .000).

2 4 2 Language Acquisition Maria Kaltsa

Figure 6 presents the percentages of PP-use in perfective VPs of each age group when they were provided with a telic motion event.

100% 90% 80% 70% 60% null P P s 50% overt P P s 40% 30% 20% 10% 0% 5-6 yrs old 6-7 yrs old 7-8 yrs old 8-9 yrs old 9-10 yrs old A dults

Figure 6. Telic Activities: Perfective Aspect & Complement Use

Turning to telic motion events, the data reported in Figure 6 suggest that the use of complement PPs is differentiated to that of DP production. The 5-6 year old children show very restricted use of PPs, which becomes optional for the rest of the L1 learners, while adults use overt PPs to indicate endpoints and encode telicity in motion events. These observations were statistically supported by one-way goodness-of-fit chi-square tests (5-6 yrs old: χ²= 13.886, df= 1, p= .000; adults: χ²= 13.787, df= 1, p= .000). Turning to the between verb-type comparisons, Table 2 presents two-way group-independence chi-square tests to determine the role of telicity for each verb category in relation to the complement production when perfective aspect was provided by the participants. Children from the age of 7 and up do differentiate between the two verb-categories. Adults, on the other hand, when selecting perfective aspect do not make the same distinction. Table 2. Between Verb-Type Comparisons: Perfective VPs & Complement Use

Telic Event Groups Activity vs. Motion Verbs

5-6 yrs old χ²= 2.426, df= 1, p=.119

6-7 yrs old χ²= 2.058, df= 1, p=.151

7-8 yrs old χ²= 16.850, df= 1, p=.000

8-9 yrs old χ²= 12.263, df= 1, p=.000

9-10 yrs old χ²= 37.677, df= 1, p=.000

Adults χ²= .815, df= 1, p=.367

Language Acquisition 2 4 3 Major Trends in Theoretical and Applied Linguistics

In sum, the production results show that for L1 learners activity verbs select perfective aspect and complements more strongly than motion verbs. The use of the imperfective for both verb types is preferred by the 5-7 year old children. The 9-10 year old learners, on the other hand, distinguish between activity and motion verbs in marking telicity and show adult like performance with regard both to aspectual marking and complement use for activities but not for motion events. Overall morphological aspect is used to denote telicity earlier than complement DPs or PPs in the acquisition of telicity.

4. Conclusion

This study examined the production of telicity by L1 learners of Greek so as to identify the nature of the relation between aspect and telicity by comparing evidence drawn from activity and motion verb constructions. The production results show that aspect is found to operate in relation to telicity. Perfective aspect is used to mark telicity on both activity and motion events but poses difficulties developmentally especially in the case of the latter. Imperfective aspect, on the other hand, is related to atelic events for both verb categories and is in place earlier in acquisition than perfective aspectual marking. The results on the joint use of perfective aspect and overt complements for telic events indicate two separate developmental milestones for each verb category. Learners aged 7 and up develop an adult-like preference for perfective activity VPs with overt complements but this preference does not develop for the motion construction even at the age of 10.

Acknowledgments

This research has been co-financed by the European Union (European Social Fund – ESF) and Greek national funds through the Operational Program “Education and Lifelong Learning” of the National Strategic Reference Framework (NSRF) - Research Funding Program: Heracleitus II. Investing in knowledge society through the European Social Fund.

2 4 4 Language Acquisition Maria Kaltsa

References

Borik, O. 2006. Aspect and reference time. OUP.

Chila-Markopoulou, D. and A. Mozer. 2001. Telicity and referentiality in the VP of Modern Greek: Aspect and Determiner. In Y. Agouraki et al. (eds.), Proceedings of the 4th International Conference on Greek Lingustics. Nicosia: University Studio Press, 138-145.

Depreatere, I. 2007. (A)telicity and Intentionality. Journal of Linguistics 45.2: 243-269.

Dowty, D. 1979. Word Meaning and Montague Grammar. Dordrecht: Reidel.

Kaltsa, M. 2012. The acquisition of telicity in the native language. AUTH: Unpublished PhD Thesis.

Katis, D. 1984. The acquisition of the Modern Greek verb with special reference to the imperfective past and perfect classes. Ph.D. Dissertation. University of Reading.

Mozer, A. 1994. Ποιόν και Απόψεις του Ρήματος [Situational and View point Aspect of the Verb]. Athens: Parousia.

Ramchand, G. and P. Svenonius. 2002. The Lexical Syntax and Lexical Semantics of the Verb Particle Construction. WCCFL 21 Proceedings. University of Tromsø: Cascadilla Press, 387-400.

Rothstein, S. 2002. What is lexical aspect? Talk given at Trans Seminar of Syntax and Semantics, UiL OTS, Utrecht University.

Schultz, P. and A. Wittek. 2003. Opening doors and sweeping floors: What children with specific language impairment know about telic and atelic verbs.

Language Acquisition 2 4 5 Major Trends in Theoretical and Applied Linguistics

In B. Beachley et al. (eds.), Proceedings of 27 BUCLD. Somerville: CascadillaPress, 727-738.

Smith, C. 1991. The parameter of Aspect. Studies in Linguistics and Philosophy 43. Dordrecht, Boston, London: Kluwer Academic Publishers.

Tsimpli, I.M. 1992/1996. The Prefunctional Stage of First Language Acquisition: A crosslinguistic Study. Ph.D. Dissertation, UCL. New York and London: Garland.

Tsimpli, I.M. and D. Papadopoulou. 2006. Aspect and argument realization : A study on antecedentless null objects in Greek. Lingua 116, 1595-1615.

Tsimpli, I.M. and D. Papadopoulou. 2009. Aspect and the Interpretation of Motion Verbs in L2 Greek. In N. Snape, Y.I. Leung and M. Sharwood Smith (eds), Representational Deficits in SLA. John Benjamins, 187-227.

Van Hout, A. 1998. On the role of direct objects and particles in learning telicity in Dutch and English. In A. Greenhill et al. (eds.), Proceedings of 22 BUCLD. Somerville: CascadillaPress, 397-408.

Van Hout, A. 2000. Event semantics in the lexicon-syntax interface: Verb frame alternations in Dutch and their acquisition. In C. Tenny and J. Pustejovsky (eds.), Events as grammatical objects. Stanford: CSLI, 239-282.

Van Hout, A. 2003. Acquiring telicity crosslinguistically: On the acquisition of telicity entailments associated with transitivity. In M. Bowerman and P. Brown (eds.), Crosslinguistic perspectives on argument structure: Implications for learnability. Hillsdale: Lawrence Erlbaum.

Vendler, Z. 1976. Verbs and times. Linguistics in Philosophy. Ithaca: Cornell UP.

Wagner, L. 2006. Aspectual Bootstrapping in Language Acquisition: Telicity and Transitivity. Language Learning and Development 2(1), 51-76.

Wagner, L. 2010. Inferring Meaning from Syntactic Structures in Acquisition: The Case of Transitivity and Telicity. Language and Cognitive Processes 25, 1354- 1379.

2 4 6 Language Acquisition Η ασαφής περιοχή της επίγνωσης της διαγλωσσικής παρεμβολής μεταξύ ιταλικής και ισπανικής από Έλληνες

Νικόλαος Καρατσιώλης και Πηνελόπη Καμπάκη-Βουγιουκλή

Αριστοτέλειο Πανεπιστήμιο Θεσσαλονίκης, Δημοκρίτειο Πανεπιστήμιο Θράκης [email protected], [email protected]

Abstract

This pilot study investigates the awareness of the interference among Greek learners of Spanish and Italian and the relationship between this phenomenon and factors such as gender, age, educational level, studies orientation, first foreign language, and number/ level of other languages. An experimental questionnaire was designed in order to investigate if and to what extent the utilization of specific strategies could limit the extent of negative transfer. An innovative tool bar instead of Likert scales is used in order to exploit the fuzziness of such a procedure. The results show that the knowledge of more foreign languages seem to reduce the phenomenon of negative transfer probably due to subjects’ phychotypologies.

1. Εισαγωγή

Το πλέον εντυπωσιακό στοιχείο στην εκμάθηση ξένων γλωσσών τα τελευταία χρόνια στην Ελλάδα είναι ότι φαίνεται να αφορά πλέον γλώσσες που παραδοσιακά δεν αποτελούσαν επιλογή των μαθητών, όπως η ιταλική, η ισπανική, η ρωσική, η αραβική, ή ακόμη και η κινεζική. Κατά συνέπεια, το ενδιαφέρον των γλωσσολόγων για τις διαδικασίες εκμάθησης και διδασκαλίας σε επίπεδο περισσοτέρων της μίας γλωσσών κερδίζει διαρκώς έδαφος διεθνώς και οι θεωρίες που αφορούν στις συγκεκριμένες διαδικασίες απασχολούν όλο και περισσότερους ερευνητές. Ένα από τα γλωσσικά φαινόμενα που, πάντα ενδιέφερε τους ερευνητές που ασχολούνται με την εκμάθηση μιας οποιασδήποτε ξένης γλώσσας, είναι η γλωσσική παρεμβολή/ μεταφορά (language interference/transfer) (Selinker 1969) είτε αρνητική (negative interference/transfer), είτε θετική/ διευκολυντική (positive transfer/ facilitation) (Selinker & Gass 1983) που φαίνεται να ενισχύεται από την υπόθεση της κοινής υποκείμενης γλωσσικής ικανότητας (common underlying proficiency hypothesis) του Cummins (1984). Συγκεκριμένα η ανάπτυξη της γλωσσικής ικανότητας στην πρώτη γλώσσα μπορεί να μεταφερθεί και στη δεύτερη και σε περισσότερες γλώσσες.

Language Acquisition 2 4 7 Major Trends in Theoretical and Applied Linguistics

Ακόμη, φαίνεται φυσιολογικό ένας μαθητής που γνωρίζει,π.χ. ιταλικά, να μάθει πιο εύκολα ισπανικά σε σχέση με μια άλλη γλώσσα μικρότερου βαθμού συγγένειας, κυρίως χάρη στη διευκόλυνση (facilitation) που του προσφέρει η συγγένεια αυτή. Είναι όμως σίγουρο πως θα μπορέσει να φτάσει στο ίδιο σημείο κατάκτησης της γλώσσας και στις δυο αυτές περιπτώσεις; Η απάντηση εξαρτάται από πολλούς παράγοντες, όπως η συνειδητή γνώση της σχέσης μεταξύ των δύο ή περισσοτέρων γλωσσών, δηλαδή η ψυχοτυπολογία (psychotypology) των χρηστών (Kellerman 1979), καθώς και η ποιότητα της διδασκαλίας (Cummins 1984). Από την άλλη πλευρά, αν και θα καταβάλει πολύ περισσότερο χρόνο στην εκμάθηση μιας λιγότερο συγγενικής γλώσσας, στο τέλος φαίνεται να την κατακτά χωρίς τα προβλήματα παρεμβολής που έχει με μια γλώσσα πιο στενής συγγένειας. Ο Πετρούνιας (87:2004) χαρακτηριστικά αναφέρει:

Όταν η μητρική και η γλώσσα-στόχος μοιάζουν πολύ ή έστω μοιάζουν πολύ σε ένα τομέα, παρουσιάζεται κίνδυνος σύγχυσης ανάμεσα στις δύο γλώσσες. Ιταλός και Ισπανός δε θα ήταν δύσκολο να μάθουν ικανοποιητικά τις αντίστοιχες γλώσσες, επειδή οι δύο μοιάζουν πολύ, θα ήταν όμως κατόρθωμα να τις διακρίνουν συνεχώς και με συνέπεια.

Ο κίνδυνος σύγχυσης, όμως, μπορεί να προέρχεται και από μια τρίτη γλώσσα Γ3, που είναι πιο κοντινή στη γλώσσα-στόχο απ’ ότι η μητρική Γ1 όπως στην περίπτωση των Ελλήνων που μιλούν ιταλικά και ισπανικά ως ξένη γλώσσα χωρίς αυτό να σημαίνει πως η μια γλώσσα δεν συμβάλλει θετικά στην κατάκτηση της δομής της άλλης.

2. Προηγούμενη έρευνα

2.1. Θεωρίες Γλωσσικής Παρεμβολής

Κατά τη δεκαετία του ’50 στο πνεύμα των δομιστικών/συμπεριφοριστικών προσεγγίσεων αναπτύχθηκε η λεγόμενη Υπόθεση της Αντιπαραβολικής Ανάλυσης (Contrastive Analysis Hypothesis), συγκριτική μελέτη ανάμεσα σε μητρική και ξένη γλώσσα που αποσκοπούσε στην πρόβλεψη των πιθανών λαθών (Fries 1945; Lado 1957), δηλαδή στην αρνητική μεταφορά (negative transfer) των χαρακτηριστικών της μητρικής ή μιας άλλης ήδη κατακτημένης γλώσσας στην νέα γλώσσα που οδηγεί σε λάθη σε όλα τα επίπεδα. Στα τέλη του ’60 τίθεται υπό αμφισβήτηση ενώ εμφανίζεται ένα νέο ρεύμα αυτό της Ανάλυσης Λαθών (Error Analysis), που προσπαθεί να ταξινομήσει και να εξηγήσει τα λάθη μέσω της σύγκρισης (Corder 1967). Σχετικές έρευνες αποδίδουν μικρότερο βαθμό σημαντικότητας στο ρόλο της παρεμβολής από τη μητρική ή μια γλώσσα Γ3 καθώς δεν είναι οι μοναδικοί λόγοι παρέκκλισης από το σωστό τύπο (Dulay et al. 1982). Στις αρχές του ’70 ο Selinker (1972) εισάγει την έννοια της διαγλώσσας (interlanguage) και έρχεται

2 4 8 Language Acquisition Νικόλαος Καρατσιώλης και Πηνελόπη Καμπάκη-Βουγιουκλή

να συμπληρώσει και να ενισχύσει το ρόλο της Ανάλυσης Λαθών. To 1983 o Sharwood Smith εισάγει τον όρο διαγλωσσική επίδραση (cross-linguistic influence) αναφερόμενος στην αλληλεπίδραση μεταξύ των κεκτημένων γλωσσών ενός ομιλητή, είτε πρόκειται για επίδραση της γλώσσα Γ3 στη Γ2, είτε Γ2 στην Γ1 κ.ο.κ.. Σε έρευνα του Cenoz (2001) σε δίγλωσσους ομιλητές (ισπανικής και βασκικής) στην εκμάθηση της αγγλικής, εκτός από την συγγένεια των γλωσσών φαίνεται να υπάρχουν κι άλλοι παράγοντες που μπορούν να επηρεάσουν την διαγλωσσική επίδραση, όπως η ηλικία και συγκεκριμένα οι μεγαλύτεροι ομιλητές παρουσιάζουν μεγαλύτερο βαθμό διαγλωσσικής επίδρασης (Cenoz et al. 2001). Τέλος, η Bayona (2009) σε έρευνά της σε ομιλητές με Γ1 την αγγλική, Γ2 τη γαλλική και Γ3 την ισπανική υποστηρίζει πως το κοινωνικό και εκπαιδευτικό υπόβαθρο είναι παράγοντες που επηρεάζουν τη διαγλωσσική επίδραση. Η γλωσσική παρεμβολή ως γλωσσικό φαινόμενο έχει μελετηθεί πολύ στο παρελθόν. Οι σχετικές μελέτες επικεντρώνονται κυρίως στην ανάλυση λαθών στα λεγόμενα σώματα κειμένων μαθητικού λόγου (corpus-based error analysis) στον προφορικό και στο γραπτό λόγο και αφορούν κυρίως την παρεμβολή από τη μητρική γλώσσα και σε μικρότερο βαθμό από μια δεύτερη ξένη γλώσσα Γ3. Παρόλα αυτά, τα αποτελέσματά τους δεν μπορούν να δώσουν μια εικόνα για την επίγνωση που είχαν οι ίδιοι οι μαθητές για τα γλωσσικά φαινόμενα που οφείλονται σε γλωσσική παρεμβολή.

2.2. Χρήση στρατηγικών

Όσον αφορά στις στρατηγικές που χρησιμοποιούν οι ομιλητές, η Schachter (1974), σε μελέτη σε ομιλητές με διαφορετική μητρική γλώσσα (αραβική, περσική, ιαπωνική και κινέζικη), παρατήρησε πως αυτοί με μητρική την αραβική και περσική χρησιμοποιούσαν δύο ή και τρεις φορές περισσότερες αναφορικές προτάσεις ως μαθητές της αγγλικής, γεγονός που το απέδωσε στην κοινή λογική της δομής των αναφορικών προτάσεων ανάμεσα στις γλώσσες αυτές. Αναφορικά με τους Κινέζους και Ιάπωνες η Schachter απέδωσε την απουσία αναφορικών στη στρατηγική της αποφυγής (avoidance strategies), δηλαδή οι μαθητές/τριες πολλές φορές προκειμένου να κάνουν λάθος σε μια δομή στην οποία δεν είναι εξοικειωμένοι καταφεύγουν στην αποφυγή της ή στη χρήση μιας παραπλήσιας δομής. Με άλλα λόγια, αποφεύγει εσκεμμένα τη χρήση ορισμένων δομών προκειμένου να μην κάνει λάθος και προτιμάει να χρησιμοποιεί τις δομές για τις οποίες είναι σίγουρος-η, κάτι που αποτελεί ένα από τα αδύνατα σημεία της Ανάλυσης Λαθών (Error Analysis). Τώρα, όσον αφορά στη χρήση στρατηγικών σε σχέση με τους παράγοντες φύλο, ηλικία, μορφωτικό επίπεδο, κατεύθυνση σπουδών, πρώτη ξένη γλώσσα και αριθμός άλλων ξένων γλωσσών, μια σύντομη ματιά στη διεθνή βιβλιογραφία αποκαλύπτει ενδιαφέροντα στοιχεία όπως ότι ο αριθμός των μελετών που αναφέρονται στις πεποιθήσεις των δύο φύλων για την εκμάθηση της ξένης γλώσσας (BALLI- Beliefs about Language Learning Inventory, Horwitz 1987) είναι λίγες. Ενδεικτικά, ο Siebert (2003) βρήκε ότι οι άνδρες πίστευαν στις δυνατότητές τους πολύ περισσότερο από τις γυναίκες και ότι χρειάζονται

Language Acquisition 2 4 9 Major Trends in Theoretical and Applied Linguistics

πολύ λιγότερο χρόνο για να μάθουν μια ξένη γλώσσα από ότι οι γυναίκες. Ανάλογα αποτελέσματα είχε και η Tercanlioglu (2005), σε τουρκικό πανεπιστήμιο. Αντίθετα οι Bacon and Finnenmann (1992), δεν είχαν στατιστικά σημαντικές διαφορές. Όλες αυτές οι μελέτες βέβαια έχουν αρκετές ιδιορρυθμίες και δεν ταυτίζονται απόλυτα ερευνητικά. Αναφορικά με την ηλικία και την εκμάθηση γλωσσών υπάρχει πολύ μεγάλη βιβλιογραφία όπως στον Krashen et al. (1983), όπου υποστηρίζεται ότι οι μεγαλύτεροι προχωρούν ταχύτερα από τα παιδιά στη σύνταξη και τη μορφολογία, ότι τα μεγαλύτερα παιδιά είναι ταχύτερα από τα μικρότερα αλλά ότι μαθαίνει καλύτερα όποιος εκτίθεται γρηγορότερα και περισσότερο στην ξένη γλώσσα. Παρόμοιες απόψεις εκφράζονται και από τον Cook (1986). Στενή σχέση με την ηλικία έχει και το μορφωτικό επίπεδο των μαθητών, εφόσον οι μεγαλύτεροι έχουν ένα μορφωτικό επίπεδο υψηλότερο από των νεαρών μαθητών. Όπως χαρακτηριστικά αναφέρεται στους Walsh and Diller (1978), οι μεγαλύτεροι και πλέον μορφωμένοι ενήλικες είναι σε σημαντικούς τομείς της διαδικασίας εκμάθησης μιας γλώσσας καλύτεροι από τους νεότερους και λιγότερο μορφωμένους. Όπως έχει υποστηριχθεί από τον Cummins (1984), η ανάπτυξη της γλωσσικής ικανότητας σε μια γλώσσα μεταφέρεται στην επόμενη και μετά στην επόμενη, καθιστώντας τους πολύγλωσσους ανταγωνιστικούς όχι μόνο σε γλωσσικό αλλά και σε γενικότερα ακαδημαϊκό επίπεδο (Collier 1992; Ramirez 1992).

3. Σκοπός και δικαιολόγηση της παρούσας έρευνας

Η μελέτη της επίγνωσης που πιθανόν να έχουν οι μαθητές της ξένης γλώσσας για το φαινόμενο της παρεμβολής, δηλαδή πώς το αντιλαμβάνονται οι ίδιοι δεν φαίνεται να έχει ερευνηθεί μέχρι τώρα τουλάχιστον, ούτε στην Ελλάδα ούτε διεθνώς. Θεωρώντας δεδομένη την διευκόλυνση που μπορεί να προσφέρει μια κεκτημένη γλώσσα στην εκμάθηση μιας νέας, στην παρούσα εργασία θα μελετήσουμε πιλοτικά την επίγνωση της ύπαρξης διαγλωσσικής αρνητικής παρεμβολής μεταξύ ιταλικής και ισπανικής στους Έλληνες χρήστες των δύο αυτών γλωσσών. Συγκεκριμένα, θα προσπαθήσουμε να μελετήσουμε εάν η προσωπική εκτίμηση που έχουν οι χρήστες για το φαινόμενο της παρεμβολής συσχετίζεται με παράγοντες όπως το φύλο, την ηλικία, το μορφωτικό τους επίπεδο και υπόβαθρο, τη γνώση επιπλέον γλωσσών καθώς και ποιες στρατηγικές χρησιμοποιούν για τον περιορισμό ή την αποφυγή του φαινομένου.

3.1. Η ασάφεια του φαινομένου

Η γλώσσα είναι μια ασαφής περιοχή και κατά συνέπεια η αναζήτηση του απολύτως αντικειμενικού αποτελεί ματαιοπονία. Έτσι, εάν σε ένα ερωτηματολόγιο χρησιμοποιήσουμε μία σαφή κλίμακα, π.χ. στην ερώτηση «πόσο συχνά χρησιμοποιείτε λεξικό;», οι επιλογές είναι 0=ποτέ, 1=σπάνια, 2=κάποιες φορές, 3=συχνά, 4=πάντα, αντιλαμβανόμαστε ότι στο ενδιάμεσο χάνονται πολλές άλλες απαντήσεις που ίσως

2 5 0 Language Acquisition Νικόλαος Καρατσιώλης και Πηνελόπη Καμπάκη-Βουγιουκλή

ικανοποιούσαν περισσότερο τους ερωτώμενους. Προκειμένου να θεραπεύσουμε, εν μέρει πάντα, αυτή την αδυναμία προχωρήσαμε στη χρήση μιας εφαρμογής των ασαφών συνόλων (fuzzy sets theory), συγκεκριμένα της ράβδου (bar) όπως αυτή προτείνεται από τους Kambaki-Vougiouklis and Vougiouklis (2008) ως εναλλακτική μέθοδος αντί των χρησιμοποιούμενων κλιμάκων Likert. Συγκεκριμένα προτείνεται η αντικατάσταση της γνωστής κλίμακας με μια ράβδο της οποίας τα δύο άκρα προσδιορίζονται με τους αριθμούς «0» και «1»:

0 1

Οι συμμετέχοντες, αντί να αποφασίσουν να επιλέξουν από μια προκαθορισμένη κλίμακα, «κόβουν» τη ράβδο με μια κάθετη γραμμή στο σημείο που κατά τη γνώμη τους εκφράζει την απάντησή τους. Τα πλεονεκτήματα της επιλεγείσας ράβδου αφορούν τόσο το στάδιο της συμπλήρωσης όσο και το στάδιο της επεξεργασίας των αποτελεσμάτων. Όποιος καλείται να απαντήσει δεν χρειάζεται ιδιαίτερη εκπαίδευση και δεν μπαίνει στην διαδικασία να διακρίνει την, έτσι κι αλλιώς, δυσδιάκριτη διαφορά μεταξύ δυο διαβαθμίσεων της κλίμακας. Όσον αφορά στην επεξεργασία των ερωτηματολογίων με ράβδο επαφίεται πλήρως στον/στην ερευνητή/ τρια. Ο/Η ερευνητής/τρια έχει την ευχέρεια να ‘κλιμακώσει’ τις απαντήσεις χωρίς να είναι δεσμευμένος/η από τη αρχή αν έχει 4 ή 5 ή 10 σκαλοπάτια και να αποσαφηνίσει τις διαφορές μεταξύ των διαβαθμίσεων. Το σπουδαιότερο είναι ότι έχει την ευχέρεια να κάνει ισομερή ή όχι κλίμακα, ανάλογα με τις απαιτήσεις της συγκεκριμένης έρευνας.

4. Μέθοδος

4.1. Υποκείμενα

Στην έρευνα συμμετείχαν 27 άτομα, ηλικίας από 26 χρονών και πάνω και συγκεκριμένα 5 άντρες και 22 γυναίκες, όλοι τους πανεπιστημιακής μόρφωσης, τόσο θεωρητικής όσο και θετικής κατεύθυνσης. Οι 5 είναι κάτοχοι ενός μεταπτυχιακού ή διδακτορικού διπλώματος. Κοινό χαρακτηριστικό τους η ελληνική ως μητρική γλώσσα και η γνώση της ιταλικής και της ισπανικής σε ικανοποιητικό επίπεδο. Όσον αφορά στη γλωσσομάθεια όλα τα υποκείμενα ομιλούν από 3 έως 5 γλώσσες, συμπεριλαμβανομένης και της ιταλικής και ισπανικής. Η επιλογή των υποκειμένων έγινε με σκοπό το υπόβαθρό τους να είναι τέτοιο ώστε να γινόταν αντιληπτά τα φαινόμενα παρεμβολής μεταξύ των δύο γλωσσών, φαινόμενα τα οποία έπρεπε να αξιολογήσουν ο καθένας με βάση την προσωπική του εμπειρία στα πλαίσια της εκμάθησης ή ακόμα και διδασκαλίας των δύο εξεταζόμενων γλωσσών.

Language Acquisition 2 5 1 Major Trends in Theoretical and Applied Linguistics

4.2. Εργαλεία της έρευνας

Για την πραγματοποίηση της έρευνας συντάχθηκε πειραματικά ερωτηματολόγιο στο οποίο κωδικοποιήθηκαν τα σημαντικότερα φαινόμενα παρεμβολής μεταξύ των δύο γλωσσών στην παραγωγή και κατανόηση του προφορικού και του γραπτού λόγου. Η επιλογή των φαινομένων παρεμβολής έγινε με εμπειρικό τρόπο σε συνδυασμό με βιβλιογραφικές αναφορές σε ζητήματα που αφορούν τη μορφολογία και τις διαφορές μεταξύ των δύο γλωσσών λαμβάνοντας υπόψη όλα τα επίπεδα της γραμματικής. Η κωδικοποίηση έγινε με βάση κοινά χαρακτηριστικά ορισμένων φαινομένων, δίνοντας ένα χαρακτηρισμό στο είδος παρεμβολής π.χ. διαφορές στον τονισμό/επιτονισμό, δηλαδή φαινόμενα που αφορούν την παρεμβολή σε επίπεδο Φωνητικής-Φωνολογίας. Για την κατανόηση του φαινομένου παρεμβολής δίνονται δύο ως τέσσερα χαρακτηριστικά παραδείγματα στην ιταλική σε αντιπαραβολή με την ισπανική γλώσσα ώστε να γίνεται απόλυτα κατανοητό από τους συμμετέχοντες το φαινόμενο το οποίο εξετάζεται. Στα φαινόμενα παρεμβολής λαμβάνεται υπόψη τόσο ο προφορικός όσο και ο γραπτός λόγος και περιλαμβάνονται φαινόμενα φωνολογικού, συντακτικού, μορφολογικού και σημασιολογικού επιπέδου. Το ερωτηματολόγιο απαρτίζεται από δύο τμήματα. Στο πρώτο συμπεριλαμβάνονται τα προσωπικά στοιχεία του συμμετέχοντα που μας ενδιαφέρουν, δηλαδή το φύλο, η ηλικία το μορφωτικό επίπεδο, το επίπεδο γλωσσομάθειας με χρονολογική σειρά εκμάθησης των γλωσσών. Το δεύτερο τμήμα απαρτίζεται από 3 μέρη εκ των οποίων το πρώτο αφορά 9 γενικές ερωτήσεις περί παρεμβολής ως γλωσσικού φαινομένου στις τέσσερις δεξιότητες (παραγωγή-κατανόηση γραπτού και προφορικού λόγου), το δεύτερο μέρος περιλαμβάνει 12 ερωτήσεις για ειδικότερα φαινόμενα παρεμβολής σε συγκεκριμένη χρήση στη γλώσσα αντιπαραβολικά στις δύο υπό έρευνα γλώσσες και τέλος, και το τρίτο περιλαμβάνει 5 ερωτήσεις που αφορούν στρατηγικές μάθησης σε προσωπικό επίπεδο σε σχέση με την αποφυγή ή περιορισμό της παρεμβολής (το πλήρες ερωτηματολόγιο, στο Παράρτημα).

4.3. Διαδικασία συλλογής δεδομένων

Το ερωτηματολόγιο καταρχήν δοκιμάστηκε από τρεις πανεπιστημιακούς για να ελεγχθεί στοιχειωδώς η εγκυρότητά του. Στη συνέχεια διανεμήθηκε στους συμμετέχοντες μεταξύ Ιανουαρίου και Φεβρουαρίου του 2010. Αποκλείστηκαν τρία ερωτηματολόγια, ενός συμμετέχοντα που δεν είχε ως μητρική γλώσσα την ελληνική και δύο που δεν ήταν απόφοιτοι κάποιου ανώτερου/ανώτατου εκπαιδευτικού ιδρύματος αλλά απόφοιτοι λυκείου.

2 5 2 Language Acquisition Νικόλαος Καρατσιώλης και Πηνελόπη Καμπάκη-Βουγιουκλή

5. Στατιστική ανάλυση

Για τη στατιστική ανάλυση χρησιμοποιήσαμε το πακέτο PASW Statistics v.18 για Windows. Προκειμένου να έχουμε συγκρίσιμα και κατανοητά αποτελέσματα για την πλειοψηφία των ερευνητών που είναι εξοικειωμένοι με τις κλίμακες Likert, μετατρέψαμε εύκολα τα αποτελέσματα σε μία κλίμακα 0-4, όπως αυτή που χρησιμοποιείται στο SILL (Oxford, 1990). Μετά την καταχώριση των δεδομένων υπολογίσαμε τους μέσους όρους καθεμιάς από τις κατηγορίες ερωτήσεων. Προβήκαμε σε περιγραφική ανάλυση (descriptive analysis) του δείγματος και στη συνέχεια συγκρίναμε τους μέσους όρους των κατηγοριών ως προς τις μεταβλητές: φύλο, ηλικία, εκπαίδευση, κατεύθυνση σπουδών, 1η ξένη γλώσσα εκμάθησης και συνολικό αριθμό ξένων γλωσσών που κατέχουν. Λόγω του ικανοποιητικού σχετικά δείγματος και μετά από έλεγχο Kolmogorov-Smirnov για κανονικότητα, βρέθηκε πως το δείγμα μας έχει κανονική κατανομή και έτσι προβήκαμε σε παραμετρικές αναλύσεις του δείγματος (parametric tests). Για τις συγκρίσεις χρησιμοποιήσαμε το “Independent Sample T-Test” για τις μεταβλητές φύλο, εκπαίδευση, κατεύθυνση σπουδών, 1η ξένη γλώσσα, και ανάλυση διακύμανσης “AΝΟVA” για τις μεταβλητές ηλικία και αριθμό ξένων γλωσσών. Επίσης, ελέγχθηκε η αξιοπιστία των απαντήσεων (reliability statistics) της κάθε κατηγορίας υπολογίζοντας τον δείκτη “Cronbach’s Alpha” για κάθε κατηγορία απαντήσεων ξεχωριστά όπως και για το σύνολο. Τέλος, υπολογίσαμε τις εσωτερικές συσχετίσεις των μεταβλητών που βρέθηκε να παρουσιάζουν στατιστικά σημαντική διαφορά και ενδιαφέρον για την έρευνα.

6. Αποτελέσματα

6.1. Έλεγχος Αξιοπιστίας

Για τον έλεγχο αξιοπιστίας του συνολικού αριθμού των απαντήσεων καθώς και καθεμιάς από τις επιμέρους κατηγορίες υπολογίστηκε ο δείκτης “Cronbach’s Alpha” οποίος δείχνει πως υπάρχει αρκετά υψηλή αξιοπιστία στις ερωτήσεις (>.065), με εξαίρεση τη χαμηλή αξιοπιστία (.543) στην τρίτη ομάδα των ερωτήσεων που απαρτίζεται από 5 ερωτήσεις σχετικά με στρατηγικές αποφυγής της παρεμβολής, γεγονός που ενδεχομένως να οφείλεται στον μικρό αριθμό των ερωτήσεων ή ακόμα στον διαφορετικό τρόπο αντιμετώπισης του προβλήματος από το κάθε υποκείμενο.

Language Acquisition 2 5 3 Major Trends in Theoretical and Applied Linguistics

6.2. Περιγραφική ανάλυση του δείγματος

Πίνακας 1. Περιγραφική ανάλυση του δείγματος (%)

ΦΥΛΟ Άνδρες Γυναίκες Σύνολο

26-30 3,7% 33,4% 37,1%

31-35 14,8% 33,3% 48,1% ΗΛΙΚΙΑ 36-40 0% 7,4% 7,4%

40+ 0% 7,4& 7,4%

ΕΚΠΑΙΔΕΥΣΗ ΑΕΙ/ΤΕΙ 18,5% 63% 81,5%

ΜΠΣ/Διδακτ. 0% 18,5% 18,5%

Θεωρητική 14,8% 63% 77,8% ΚΑΤΕΥΘΥΝΣΗ Θετική 3,7% 18,5% 22,2%

Ιταλικά 7,4% 66,7% 74,1% 1η ΞΕΝΗ ΓΛΩΣΣΑ Ισπανικά 11,1% 14,8% 25,9

Τρεις 7,4% 37,04% 44,4%

Τέσσερις 7,4% 37,04% 44,4% ΣΥΝΟΛΟ ΓΛΩΣΣΩΝ Πέντε 3,7% 7,4% 11,1%

Τέσσερις

18,5% 81,5% 100%

6.3. Η παρεμβολή ως προς τους παράγοντες φύλο και ηλικία

Μελετήσαμε τη σχέση ανάμεσα στους μέσους όρους της καθεμιάς από τις 3 ομάδες ερωτήσεων με τους παράγοντες φύλο, ηλικία, εκπαίδευση, κατεύθυνση σπουδών, 1η γλώσσα εκμάθησης και γνώση επιπλέον γλωσσών. Οι 3 ομάδες ερωτήσεων συμπεριλαμβάνουν: η 1η γενικά πώς αντιλαμβάνονται την γλωσσική παρεμβολή, η 2η την αντίληψη της παρεμβολής ειδικά σε συγκεκριμένα παραδείγματα των 2 γλωσσών και η 3η την αντιμετώπιση του φαινομένου της παρεμβολής με τη χρήση στρατηγικών. Εφαρμόζοντας το T-Test με εξαρτημένη μεταβλητή το φύλο των υποκειμένων και ανεξάρτητες μεταβλητές τον Μ.Ο. των απαντήσεων για κάθε μια από τις 3 ομάδες καθώς και τον Μ.Ο. του συνολικού του τεστ δεν βρέθηκε καμία στατιστικά σημαντική διαφορά. Προβληματικό ωστόσο παραμένει το γεγονός πως το δείγμα μας απαρτίζεται κατά συντριπτική πλειοψηφία από γυναίκες και δεν έχουμε ισομερή κατανομή με τους άντρες. Για την 1η ομάδα έχουμε t(25)= .271, p= .78, για τη δεύτερη t(25)= 1.09, p=.29 και για την 3η t(25)=.15 , p= .89 και για το συνολικό τεστ t(25)= .22, p=.40 > 0,05. Για τη διερεύνηση του παράγοντα ηλικία δεδομένου ότι έχουμε 4 ηλικιακές ομάδες εφαρμόζουμε “ΑΝΟVA” με εξαρτημένη μεταβλητή την ηλικία των υποκειμένων και

2 5 4 Language Acquisition Νικόλαος Καρατσιώλης και Πηνελόπη Καμπάκη-Βουγιουκλή

ανεξάρτητες μεταβλητές τον Μ.Ο. των απαντήσεων για κάθε μια από τις 3 ομάδες και τον Μ.Ο. του συνολικού τεστ, όπου και πάλι δεν βρήκαμε καμία στατιστικά σημαντική διαφορά. Αξίζει να αναφέρουμε πως αν και έχουμε 4 ομάδες (26-30, 31-35, 36-40 και 40+άνω) είναι χαρακτηριστική η απουσία μικρότερων ηλικιών δηλ. ανάμεσα 18- 25 ενδεχομένως λόγω του ότι είναι σπάνιο κάποιος ‘Έλληνας χρήστης της ιταλικής και ταυτόχρονα της ισπανικής να βρίσκεται σε αυτή την ηλικιακή ομάδα. Επομένως, αν είχαμε ένα μεγαλύτερο εύρος ηλικιών πιθανόν να είχαμε διαφορετικά και πιο ενδιαφέροντα αποτελέσματα.

6.4. Η παρεμβολή ως προς τους παράγοντες εκπαίδευση και κατεύθυνση σπουδών

Θέσαμε ως εξαρτημένη μεταβλητή το μορφωτικό επίπεδο των υποκειμένων και ανεξάρτητες μεταβλητές τους Μ.Ο. των απαντήσεων της καθεμιάς ομάδας και του συνολικού τεστ και τα αποτελέσματα δεν έδειξαν κάποια στατιστικά σημαντική διαφορά ανάμεσά τους. Αυτό μας οδηγεί στο συμπέρασμα πως ούτε το μορφωτικό επίπεδο αλλά ούτε το Τμήμα από το οποίο προέρχονται δεν φαίνεται να επιδρά στην αντίληψη του φαινομένου της παρεμβολής. Κάτι τέτοιο ήταν μη αναμενόμενο για τους προερχόμενους από ξενόγλωσσα Τμήματα, δεδομένου ότι προσεγγίζουν τη γλώσσα όχι μόνο σε πρακτικό αλλά και σε θεωρητικό επίπεδο. Η έλλειψη ενδιάμεσων εκπαιδευτικών βαθμίδων χαμηλότερων της ανωτέρας και ανωτάτης εκπαίδευσης δεν επιτρέπει και μεγάλα περιθώρια ερμηνείας σχετικά με τον παράγοντα εκπαίδευση. Σε κάθε περίπτωση θα μπορούσαμε να πούμε πως ο Μ.Ο. της 2η ομάδας και του συνολικού τεστ παρουσιάζει κάποια τάση για διαφορά (p= .09 και p=.08 αντίστοιχα) χωρίς όμως να είναι στατιστικά σημαντική για να επιτρέψει εξαγωγή συμπερασμάτων.

6.5. Η παρεμβολή ως προς τους παράγοντες 1η γλώσσα εκμάθησης και γνώση επιπλέον γλωσσών

Για τη διερεύνηση της σχέσης μεταξύ της 1ης γλώσσας εκμάθησης ανάμεσα στις δυο υπό έρευνα γλώσσες, σε σχέση πάντα με τα φαινόμενα παρεμβολής που μας ενδιαφέρουν πραγματοποιήσαμε T-Test με εξαρτημένη μεταβλητή την 1η γλώσσα και ανεξάρτητες μεταβλητές τους Μ.Ο. όλων των ομάδων και του συνόλου του τεστ. Τα αποτελέσματα δεν δίνουν στατιστικώς σημαντική διαφορά ώστε να μπορούμε να πούμε πως αντιλαμβάνονται διαφορετικά τα φαινόμενα παρεμβολής ή χρησιμοποιούν διαφορετικές στρατηγικές τα άτομα που μάθανε πρώτα τη μία ή την άλλη γλώσσα. Όσον αφορά στο σύνολο των γλωσσών που ομιλούν τα υποκείμενα κυμαινόταν από 3 έως 5 γλώσσες και το χαρακτηριστικό ήταν πως όλοι μιλούσαν την αγγλική και φυσικά

Language Acquisition 2 5 5 Major Trends in Theoretical and Applied Linguistics

την ιταλική και την ισπανική. Η ανάλυση διακύμανσης ANOVA έδωσε ενδιαφέροντα αποτελέσματα καθώς υπήρξε στατιστικά σημαντική διαφορά μεταξύ αυτών που μιλούσαν 3 και αυτών που μιλούσαν 5 γλώσσες στο Μ.Ο. της 2ης ομάδας καθώς και του Μ.Ο. του συνόλου των απαντήσεων. Η σχέση αυτή απεικονίζεται στα σχεδιαγράμματα που ακολουθούν.

Σχεδιάγραμμα. Στον άξονα X ο συνολικός αριθμός γλωσσών και στον άξονα Υ στο 1ο σχεδιάγραμμα ο Μ.Ο. της 2ης ομάδας ερωτήσεων και στο 2ο σχεδιάγραμμα ο Μ.Ο των ερωτήσεων του συνολικού τεστ.

2 5 6 Language Acquisition Νικόλαος Καρατσιώλης και Πηνελόπη Καμπάκη-Βουγιουκλή

Τέλος, όσον αφορά τον υπολογισμό της συσχέτισης Pearson μεταξύ της 2ης ομάδας και του συνολικού τεστ με το σύνολο των γλωσσών βρέθηκε πως υπάρχει μέτρια συσχέτιση του Μ.Ο. με το σύνολο γλωσσών και μάλιστα και στις δύο περιπτώσεις η συσχέτιση αυτή είναι αρνητική, δηλαδή όσο αυξάνεται ο αριθμός των γλωσσών μειώνεται ο μέσος όρος των απαντήσεων. Συσχετισμοί ανάμεσα στις μεταβλητές που επιδιώξαμε έδειξαν να έχουν στατιστικά σημαντικές διαφορές και μέτρια συσχέτιση μεταξύ τους: (r= -.47 DF=25 p=.013) και (r= -.38 DF=25 p=.05) αντίστοιχα. Αυτό μας οδηγεί στο συμπέρασμα πως οι πολύγλωσσοι αντιλαμβάνονται τη γλωσσική παρεμβολή δύο συγγενικών γλωσσών καλύτερα από κάποιους που κατέχουν λιγότερες γλώσσες, προφανώς διότι αντιλαμβάνονται τις ομοιότητες αλλά κυρίως τις διαφορές μεταξύ των γλωσσών- έχουν ορθή ψυχοτυπολογία.

7. Συζήτηση

Το λάθος οφείλεται στην ασάφεια της γλώσσας και γι’αυτό πολλές φορές είναι δύσκολο να το προσδιορίσουμε. Σήμερα η γλωσσολογία θεωρεί το λάθος μια απαραίτητη και μάλιστα άκρως παραγωγική διαδικασία για την κατάκτηση της γλώσσας (Fernández López 1989:35). Μέσα από το λάθος κάνουμε τις συγκρίσεις μας, ελέγχουμε, διορθώνουμε και προχωράμε. Ακόμη, το λάθος που δεν διορθώνεται, παρά τις προσπάθειες, γίνεται η νόρμα της επόμενης γενιάς. Όπως χαρακτηριστικά αναφέρει ο Fernández López (1989:32) «όταν δεν κάνει κανείς λάθη είναι επειδή δεν εξασκεί τη γλώσσα και όταν κανείς δεν την εξασκεί δεν τη μαθαίνει». Γι’ αυτό θέλουμε να διευκρινίσουμε πως σαφώς δεν ενοχοποιούμε το λάθος αλλά το θεωρούμε βασικό σημείο αναφοράς στο πλαίσιο της εκμάθησης της ιταλικής και της ισπανικής. Με την συνειδητοποίηση των ομοιοτήτων ο εκπαιδευόμενος αντιλαμβάνεται την νέα γλώσσα ως κάτι το ήδη οικείο, κοντινό και ιδιαίτερα προσιτό, γεγονός που ενθαρρύνει την εκμάθηση. Ενισχύεται, δηλαδή, η δημιουργία της σωστής ψυχοτυπολογίας τού αναφορικά με τις δύο αυτές γλώσσες. Από την άλλη πλευρά, μπορεί να δοθεί έμφαση στις διαφορές δομής και λεξιλογίου με τη χρήση υλικού που επισημαίνει αντιπαραβολικά τις διαφορές αυτές μέσα στα κατάλληλα περικείμενα ώστε να γίνεται σαφής η οποιαδήποτε διαφορετικότητα που μπορεί να οδηγήσει σε φαινόμενα παρεμβολής, πολύ πριν παγιωθεί (fossilized) στη χρήση της από τον εκπαιδευόμενο. Ένα τέτοιο παράδειγμα είναι οι συγκεντρωτικοί πίνακες με τις ψευδόφιλες μονάδες (faux-amis) ανάμεσα σε δύο γλώσσες, λέξεις που μπορούν να οδηγήσουν σε αδιέξοδο ή σε παρερμηνεία τη γλωσσική επικοινωνία. Εάν υπάρχει επίγνωση των φαινομένων που ίσως οδηγούν σε παρεμβολή, η χρήση στρατηγικών θα μπορούσε να τις μετριάσει ή να τις εξαφανίσει. Η παρούσα έρευνα έδειξε πως η πολυγλωσσία ίσως μπορεί να αποτελέσει θετικό στοιχείο έναντι στη γλωσσική παρεμβολή, διότι προσφέρει τη μεταγλωσσική εμπειρία που βοηθάει στη συνειδητοποίηση των ομοιοτήτων αλλά και των διαφορών μεταξύ των γλωσσών.

Language Acquisition 2 5 7 Major Trends in Theoretical and Applied Linguistics

7.1. Αδυναμίες της παρούσας έρευνας /μελλοντική έρευνα

Το ερωτηματολόγιο συντάχθηκε εμπειρικά, χωρίς ουσιαστικό έλεγχο εγκυρότητας, και επιδέχεται τροποποιήσεις, ιδιαίτερα στο 3ο μέρος όπου η αξιοπιστία των ερωτήσεων ήταν χαμηλή. Αναφορικά με το δείγμα, είναι μειονέκτημα ότι ήταν σχετικά περιορισμένο και όχι καλά κατανεμημένο ως προς τους παράγοντες φύλο, ηλικία και κατεύθυνση σπουδών. Ταυτόχρονη έρευνα για την παρεμβολή βασισμένη στην παραγωγή λόγου από τα ίδια υποκείμενα μπορεί να έδινε τελείως διαφορετικά αποτελέσματα από αυτά του ερωτηματολογίου που αφορά στην προσωπική εκτίμηση του καθενός για την παρεμβολή. Κάτι τέτοιο οφείλεται και πάλι στην ασάφεια των ορίων που τίθενται. Θα ήταν ενδιαφέρον να δούμε παράλληλα αν η εκτίμησή τους συμπίπτει με τα πραγματικά δεδομένα παρεμβολής που εμφανίζουν στο λόγο τους, αυτό όμως παραμένει αντικείμενο νέας έρευνας.

2 5 8 Language Acquisition Νικόλαος Καρατσιώλης και Πηνελόπη Καμπάκη-Βουγιουκλή

Βιβλιογραφία

Bacon, S.M.C. and M.D. Finnemann. 1992. Sex differences in self-reported beliefs about language learning and authentic oral and written input. Language Learning 42(4), 471- 495.

Βayona, P. 2009. Crosslinguistic influence in the acquisition of Spanish as a third language. Doctoral Dissertation. The University of Western Ontario.

Cenoz, J., B. Hufeisen and U. Jessner. 2001. Crosslinguistic influence in third language acquisition: Psycholinguistics perspectives. Clevedon: Multilingual Matters.

Collier, V.P. 1992. A synthesis of studies examining long-term language minority student data on academic achievement. Bilingual Research Journal 16(1-2), 187-212. Retrieved 22 September 2010 from http://www.ncela.gwu.edu/pubs/nabe/brj/v16/16_12_collier.htm .

Cook, V.J. (ed.) 1986. Experimental Approaches to Second Language Learning, Chapter 2 Web version.

Corder, S.P. 1967. The significance of learners’ error. International Review of applied Linguistics 5(4), 161-170.

Cummins, J. 1984. Bilingualism and special education. Clevedon: Multilingual Matters.

Dulay, H., M. Burt and S. Krashen. 1982. Language two. Oxford: Oxford University Press.

Fernández López, S. 1989. Corregir y evaluar desde una perspectiva comunicativa. Cable 4, 30-35.

Fries, C.C. 1945. Teaching and learning English as a foreign language. Michigan: University of Michigan Press.

Language Acquisition 2 5 9 Major Trends in Theoretical and Applied Linguistics

Greco, S. χ.ημ. “Lenguas Afines” REDele N.6. Perugia:Università di Perugia, Retrieved 4 September 2010 από

Horwitz, E.Κ. 1987. Surveying students’ beliefs about language teaming. In A.L. Wenden and J. Rubin (eds), Learner Strategies in Language Learning. London: Prentice Hall, 119-132.

James, C. 1998. Errors in Language Learning and Use-Exploring Error Analysis. Essex: Pearson.

Κambaki-Vougioukli, P. and T. Vougiouklis. 2008. Bar instead of scale. Ratio Sociologica 3, 49-56.

Kellerman, E. 1979. Transfer and non-transfer: where we are now. Studies in Second Language Acquisition 2, 37-57.

Krashen, S.D., R.C. Scarcella and M.H. Long (eds.). 1982. Child-Adult Differences in Second Language Acquisition. Newbury House: Rawley, MA.

Lado, R. 1957. Linguistics across cultures. Applied linguistics for language teachers. Michigan: University of Michigan Press.

Λεονταρίδη, E. 2008. Επικίνδυνες συγγένειες; Γλωσσικές παρεμβολές της Ισπανικής και της Ιταλικής στη διαδικασία εκμάθησης της ξένης γλώσσας: η περίπτωση των Ψευδόφιλων Λεξικών Μονάδων. Επετηρίδα του Τμήματος Ιταλικής Γλώσσας και Φιλολογίας, Τόμος ΙΙΙ/ 2006-2007 ψηφιακό μέρος-pdf.21 (ISSN: 1792-006Χ), Θεσσαλονίκη: Αριστοτέλειο Πανεπιστήμιο Θεσσαλονίκης.

Leontaridi, E., N.S. Peramos and M.R. Morales. 2009. Errores en la interlengua escrita de estudiantes grecófonoes de español como lengua extranjera. Zona Próxima, 12-31.

López Jimeno, Α. 2000. Algunos falsos amigos del aprendiz de griego (moderno) hispano-hablante. Apuntes, invierno 2000, 15-17.

Oxford, R. 1990. Language learning strategies: What every teacher should know. New York, USA: Newbury House.

Πετρούνιας, Ε. 2004. Νεοελληνική Γραμματική και Συγκριτική («Αντιπαραθετική») Ανάλυση Τόμος Α’: Φωνητική και Εισαγωγή στη Φωνολογία. Θεσσαλονίκη: Ζήτη.

Ramirez, D.J. 1992. Executive summary. Bilingual Research Journal 16, 1-62.

2 6 0 Language Acquisition Νικόλαος Καρατσιώλης και Πηνελόπη Καμπάκη-Βουγιουκλή

Schachter, J. 1974. An error in error analysis. Language Learning 24, 205-214.

Selinker, L. 1969. Language transfer. General Linguistics 9, 67-92.

Selinker L. 1972. Interlanguage. IRAL 10(3), 209-231.

Selinker, L. and S. Gass (eds). 1983. Language transfer in Language Learning Rowley. Mass: Newburh House.

Sharwood Smith, M. 1983. Crosslinguistic aspects of second language acquisition. Applied Linguistics 4(3), 192-199.

Siebert, L.L. 2003. Student and teacher beliefs about language learning. THE ORTESOL Journal 21, 7-39.

Tercanlioglu, L. 2005. Pre-service EFL teachers’ beliefs about foreign language learning and how they relate to gender. Electronic Journal of Research in Educational Psychology, 5-3(1), 145-162. Retrieved 15 February 2011 from hhhp://investigacion- psicopedagogica.org/revista/articulos/ 5/English/Art 5 58.pdf.

Walsh, T.M. and K.C. Diller. 1978. Neurolinguistic Foundations to Methods of Teaching a Second Language. International Review Of Applied Linguistics 16, 1-14.

Language Acquisition 2 6 1 Major Trends in Theoretical and Applied Linguistics

ΠΑΡΑΡΤΗΜΑ

1° ΤΜΗΜΑ: ΠΡΟΣΩΠΙΚΑ ΣΤΟΙΧΕΙΑ

1. Φύλο: Α. Άντρας Β. Γυναίκα

2. Ηλικία: Α.20-25 Β. 26-30 Γ. 31-35 Δ. 36-40 Ε. 40 και άνω

3. Σπουδές (Συγκεκριμένα ποια Σχολή/Τμήμα: παρακαλώ σημειώστε μόνο την ΥΨΗΛΟΤΕΡΗ βαθμίδα):

A. Λύκειο: B. Πτυχίο: Γ. Μεταπτυχιακό : Δ. Διδακτορικό :

4. Ξένες γλώσσες που γνωρίζετε με τη σειρά που τις μάθατε:

Α. Ε. Β. ΣΤ. Γ. Δ.

5. Προσδιορίστε πόσο καλά ξέρετε -κατά τη γνώμη σας- την κάθε γλώσσα κόβοντας τη ράβδο που ακολουθεί στο σημείο που κρίνετε πως αντιπροσωπεύει το επίπεδό σας λαμβάνοντας υπόψη ότι 0=ΚΑΘΟΛΟΥ και 1 = ΑΡΙΣΤΑ: πχ. 0______|_1

Α. 0______1 Β. 0______1 Γ. 0______1 Δ. 0______1 Ε. 0______1 ΣΤ. 0______1

2° ΤΜΗΜΑ: ΕΡΩΤΗΜΑΤΟΛΟΓΙΟ Απαντήστε κόβοντας τη ράβδο στο σημείο που εκφράζει την προσωπική σας άποψη έχοντας υπόψη ότι ΑΥΤΗ ΤΗ ΦΟΡΑ 0 = ΚΑΘΟΛΟΥ και 1 = ΑΠΟΛΥΤΑ :

2 6 2 Language Acquisition Νικόλαος Καρατσιώλης και Πηνελόπη Καμπάκη-Βουγιουκλή

ΜΕΡΟΣ Α 1. Πιστεύω πως η γνώση άλλων ξένων γλωσσών, βοηθάει στην εκμάθηση μιας νέας γλώσσας. 0______1 (ΚΑΘΟΛΟΥ) (ΑΠΟΛΥΤΑ)

2. Όσο πιο συγγενική είναι μια νέα γλώσσα που μαθαίνω με μια που ήδη γνωρίζω, τόσο γρηγορότερα μπορώ να τη μάθω. 0______1

3. Όσο πιο συγγενική είναι μια νέα γλώσσα που μαθαίνω με μια που ήδη γνωρίζω, τόσο καλύτερα μπορώ να τη μάθω. 0______1

4. Όταν μαθαίνω μια νέα γλώσσα παρουσιάζω φαινόμενα «παρεμβολής», δηλαδή έχω αρνητική επιρροή από τις ήδη γνωστές μου γλώσσες, τόσο σε λεξιλογικό όσο και συντακτικό επίπεδο. 0______1

5. Τα φαινόμενα «παρεμβολής» είναι δυσκολότερο να ξεπεραστούν όταν πρόκειται για γλώσσες με μεγάλο βαθμό συγγένειας όπως τα ιταλικά και τα ισπανικά. 0 ______1

6. Αυτή η «παρεμβολή» αφορά στην παραγωγή γραπτού λόγου (writing) 0 ______1

7. Αυτή η «παρεμβολή» αφορά την κατανόηση γραπτού λόγου (reading) 0 ______1

8. Αυτή η «παρεμβολή» αφορά την παραγωγή προφορικού λόγου (speaking) 0 ______1

9. Αυτή η «παρεμβολή» αφορά την κατανόηση προφορικού λόγου (listening) 0 ______1

Language Acquisition 2 6 3 Major Trends in Theoretical and Applied Linguistics

ΜΕΡΟΣ Β Με βάση την προσωπική σας εμπειρία, «αξιολογήστε» το βαθμό παρεμβολής για κάθε ένα από τα επιμέρους φαινόμενα. Για διευκόλυνση, σας παραθέτουμε και τα ανάλογα παραδείγματα στην ιταλική και στην ισπανική. Απαντήστε κόβοντας τη ράβδο στο σημείο που εκφράζει την προσωπική σας κρίση έχοντας υπόψη ότι ΑΥΤΗ ΤΗ ΦΟΡΑ 0 = ΚΑΘΟΛΟΥ ΠΑΡΕΜΒΟΛΗ και 1 = ΥΨΗΛΟΣ ΒΑΘΜΟΣ ΠΑΡΕΜΒΟΛΗΣ:

1. Ψευδόφιλες Λεξικές Μονάδες : Λέξεις όμοιες ή σχεδόν όμοιες λέξεις με τελείως διαφορετικό νόημα σε κάθε γλώσσα: πχ. salire-salir (ανεβαίνω – βγαίνω) / πχ. il burro – el burro (βούτυρο – γάιδαρος)/ πχ. l’ arena - la arena (το αμφιθέτρο – η άμμος) 0 ______1 (ΚΑΘΟΛΟΥ) (ΥΨΗΛΟΣ)

2. Λέξεις όμοιες ή σχεδόν όμοιες με παρόμοιο νόημα στις δύο γλώσσες: πχ. incontrare – encontrar (συναντώ-βρίσκω)/ πχ. sperare –esperar (ελπίζω –περιμένω) / πχ .la carta / la carta (το χαρτί - το γράμμα) 0______1

3. Διαφορά στην ορθογραφία : πχ. difficile - difícil. / πχ. Spagna –España/ πχ. adattare – adaptar/ πχ. penso-pienso/ πχ. la porta – la puerta…. 0______1

4. Διαφορετική προφορά ή τονισμός: πχ. visita – visita (βίζιτα – μπισίτα) / πχ. bello – bello (μπέλλο –μπέγιο) / πχ. guerra –guerra (γκουέρρα –γέρρα) / πχ. unicamente – únicamente (ουνικαμέντε – ούνικαμέντε) … 0______1

5. Διαφορές στην παραγωγή λέξεων που έχουν την ίδια ετυμολογία, (συνήθως από τη λατινική) και ίδια σημασία πχ. pensiero – pensamiento/ πχ. conoscenza –conocimiento/ πχ. professoressa – profesora/ πχ. destinazione – destino … 0______1

6. Λειτουργική αναντιστοιχία ή μερική αντιστοιχία βασικών ρημάτων κοινής ετυμολογίας από τα λατινικά: πχ. essere-ser / πχ.stare-estar / πχ. avere-haber/ πχ. tenere- tener/ πχ. andare-andar … 0______1

7. Λειτουργική αντιστοιχία ή μερική αντιστοιχία στη χρήση των προθέσεων/συνδέσμων: πχ. Per me – para mí , por mí / πχ. sono a Roma - estoy en Roma/ πχ. vado in Grecia - voy a Grecia … 0______1

2 6 4 Language Acquisition Νικόλαος Καρατσιώλης και Πηνελόπη Καμπάκη-Βουγιουκλή

8. Διαφορές στη χρήση σημείων στίξης (ερωτηματικά, θαυμαστικά, τόνοι, απόστροφος κ.α): πχ. Ciao! Come stai?- ¡Hola! ¿Qué tal? /πχ. Caro amico, - Querido amigo: /πχ. ultimo – último / πχ. l’ amica – la amiga ... 0______1

9. Διαφορά στη χρήση οριστικού/αόριστου άρθρου : πχ. un’altro – otro /πχ. il mio amico – mi amigo/ πχ. il fiore – la flor/ πχ. alcuni amici – unos amigos ... 0______1

10. Διαφορά στη σύνταξη παρόμοιων δομών: πχ. ai giovani piace - a los jóvenes les gusta/ πχ. sognare qualcuno – soñar con alguien/ πχ. amo mia madre – amo a mi madre/ πχ. piú giovane di lui - más jóven que él ... 0______1

11. Διαφορά στη χρήση χρόνων/εγκλίσεων πχ. ieri ho mangiato - ayer comí/ πχ. credo che tu abbia ragione – creo que tú tienes razón/ πχ. non mi dire - no me digas ... 0______1

12. Εκφράσεις, όπως σύνδρομες εκφράσεις, φράσεις κλισέ, παροιμίες κτλ., που δεν παρουσιάζουν αντιστοιχία στην μετάφραση από τη μια γλώσσα στην άλλη: πχ. sempre più = cada vez más (=όλο και περισσότερο) /πχ. mi dispiace = lo siento (=λυπάμαι) / πχ. mi manchi =me haces falta(=μου λείπεις)/ πχ. forse = a lo mejor/ tal vez (=ίσως)... 0______1

Language Acquisition 2 6 5 Major Trends in Theoretical and Applied Linguistics

ΜΕΡΟΣ Γ Απαντήστε κόβοντας τη ράβδο στο σημείο που εκφράζει την προσωπική σας κρίση έχοντας υπόψη ότι αυτή τη φορά 0 = ΠΟΤΕ και 1 = ΠΑΝΤΑ: Κατά την εκμάθηση λεξιλογίου μιας νέας γλώσσας χρησιμοποιώ «στρατηγικές» για την καλύτερη εμπέδωσή του. Συγκεκριμένα:

1. Μαθαίνω το λεξιλόγιο μιας γλώσσας-στόχου πάντα συγκριτικά, συσχετίζοντάς το με μια που ξέρω καλά ή που είναι συγγενική (πχ. ιταλικά- ισπανικά) 0______1

2. Οι συσχετισμοί που κάνω αφορούν στις ομοιότητες των δύο γλωσσών, πχ. σε λεξιλογικό, συντακτικό επίπεδο. 0______1

3. Οι συσχετισμοί που κάνω αφορούν στις διαφορές των δύο γλωσσών, πχ. σε λεξιλογικό, συντακτικό επίπεδο. 0______1

4. Αποφεύγω να συσχετίζω τη γλώσσα-στόχο με κάποια συγγενική γλώσσα για να αποφεύγω τις γλωσσικές παρεμβολές. 0 ______1

5. Γενικά, νομίζω ότι οι στρατηγικές που προσωπικά χρησιμοποιώ για την εκμάθηση μιας νέας γλώσσας είναι αποτελεσματικές για τον περιορισμό των φαινομένων παρεμβολής από άλλες γλώσσες τουλάχιστον για μένα προσωπικά. 0 ______1

2 6 6 Language Acquisition Greek Embedded Aspect in Child and Adult L2/L3 Acquisition

Sviatlana Karpava and Kleanthes K. Grohmann

University of Cyprus [email protected], [email protected]

Abstract

The focus of the present work is the acquisition of Greek embedded aspect by early and late L2/L3 learners. The study was conducted in two countries, Mainland Greece and the Republic of Cyprus, with five groups of participants: Russians and Georgians living in Cyprus and acquiring Cypriot Greek (CG) as their L2/L3, Russians who live in Greece and use Standard Modern Greek (SMG) as their L2, and two control groups of native speakers of CG and SMG, respectively. Considerations of typological distance between the languages were compared to the different aspectual paradigms looking for possible transfer from L1/L2 (Russian) to L2/L3 (CG or SMG); length of residence in the country was found to influence the participants’ production most.

1. Introduction

In line with the Interpretability Hypothesis (Tsimpli 2003) aspect, being a grammaticalized, interpretable feature in Greek, is relevant to the syntax– semantics and the syntax–discourse interfaces. This study investigates influence of L1 (or possibly L2) on L2 (L3) acquisition of Greek aspect, comparing participants who have a homogeneous L1 background (Russian) in Mainland Greece (L2 Standard Modern Greek/SMG) and Cyprus (L2 Cypriot Greek/CG) with participants whose L1 is Georgian, L2 Russian, and L3 Cypriot Greek — consequently, with Russian being an overlapping language in all three non-native groups, as L1 in one and L2 in the other group. That is, this study aims to find out whether the grammar of Russian shared by all participants interacts at all with mastering embedded aspect in Greek in either or both of these groups. Both in CG and SMG verb complementation takes a finite form, while in Russian this form is infinitival. The main interest of the study will be subjunctive clauses in Greek with na, the subjunctive marker (Roussou 2009), where the

Language Acquisition 2 6 7 Major Trends in Theoretical and Applied Linguistics

main verb controls the na-complement clause. The choice of aspect in the na-clause depends on the type of matrix verb and the interaction between lexical aspect (states, activities, achievements, and accomplishments) and grammatical aspect (perfective and imperfective) (Malagardi 1993; Moser 1993). Russian, in contrast, allows two types of verb complementation: (i) infinitival complementation, in which case the subject of the main clause and the embedded clause should be the same, and (ii) finite verb complementation, in which case the complementizer čtoby ‘in order to’ is used but the verb should be only in the past tense, and matrix and embedded subject should not be identical. The paper is organized as follows. Section 2 provides an overview of aspect in Georgian, Russian, and Greek and presents the research questions to be investigated in the study. Section 3 deals with the methodology of the experiments. In Section 4 the results of the written and oral tests are presented. Section 5 concludes and offers an interpretation of the results in light of current L2/L3A theories.

2. Background

2.1. Aspect, Lexical vs. Grammatical

The internal temporal constituency of a situation or event is described by aspect through beginning, continuation, and completion (Comrie 1976: 16). Aspect can be distinguished into lexical and grammatical. The inherent lexical or semantic meaning of a verb is determined by lexical aspect, which has four types: states, activities, accomplishments, and achievements (Vendler 1957). States (e.g., ‘love’, ‘enjoy’, ‘hate’) describe atelic, stable consistent situations without change, dynamicity or successive stages (Smith 1991). Activities (e.g., ‘eat’, ‘walk’) describe atelic, but dynamic events without a natural endpoint. Accomplishments (e.g., ‘The children built a sandcastle.’) show the heterogeneous process with successive internal stages that leads to a completion or an outcome of an action or change of state. Achievements are instantaneous events that result in a change of state and lack duration (e.g., ‘Our team won the game’). States and activities share the feature of atelicity, but they differ in dynamicity: states are [–dynamic], activities are [+dynamic], while accomplishments and achievements share the feature of telicity, but differ in duration: accomplishments are [+durative] and achievements are [–durative]. Grammatical aspect is responsible for aspectual distinctions by means of morpho-syntactic marking with the help of grammatical devices such as

2 6 8 Language Acquisition Sviatlana Karpava and Kleanthes K. Grohmann

auxiliaries, inflectional or derivational morphology. Grammatical aspect is distinguished into perfective and imperfective. Perfective aspect describes a situation as a whole, without showing the internal view of an action or situation. Imperfective aspect describes the internal structure of a situation and views it as an open event (Comrie 1976; Shirai & Li 2000; Smith 1991).

2.2. Aspect, Cross-Linguistically

The category of aspect is universal (Smith 1991), but there is cross-linguistic variation in the realization and encoding of lexical and grammatical aspect across languages. The mapping between lexical and grammatical aspect varies as well and so do forms and meanings of aspect. There is morphological aspect (e.g., the perfective/imperfective distinction in Slavic languages), aspectual tenses (e.g., perfect/imperfect in Romance languages, aorist/preterit in Greek and Bulgarian), or aspectual words and affixes (e.g., English –ing or Dutch and German ge–). The concern of this study is the aspectual system of Georgian, Russian, and Greek — and how learners with L1 Russian and learners with L1 Georgian/L2 Russian acquire L2/L3 Greek embedded aspect. (When the distinction between CG and SMG does not matter, we simply refer to Modern Greek as ‘Greek’.) To begin with, Georgian is characterized by rich synthetic morphology with inflectional verbal morphology and a complex case system, and can be described with the help of screeves, a term used to refer to tense, aspect, and mood marking on Georgian verbs. There are eleven screeves which can be divided into three series: present, aorist, and perfect (Harris 1981: 47). The present series has two sub-series, present (present, imperfect, present subjunctive) and future (future, conditional, future subjunctive). The aorist series consists of two screeves, aorist and optative subjunctive, while the perfect series consists of three screeves, perfect first evidential, pluperfect (second evidential), and perfect subjunctive (third evidential). There are four verb classes in Georgian: transitive, intransitive, medial, and indirect verbs. A lot of verbs have common stems and differ only in terms of prefixes and thematic suffixes. Prefixes are realized only in the aorist and perfect series, and thematic suffixes only in the present series (e.g., root shen, infinite form asheneba). In the present series, the thematic suffix -eb is present, but without the preverb: present indicative (v-a-shen-eb), imperfect (v-a-shen-eb-d-I), present subjunctive (v-a-shen-eb-d-e). In the aorist series, there is a preverb, but no thematic suffix: aorist indicative (a-v-a-shen-e), optative (a-v-a-shen-o). In the perfective series, both preverb and the thematic suffix are present: perfect (a-m-i-shen-eb-i-a), pluperfect (a-m-e-shen-eb-in-a), perfect subjunctive (a-m- e-shen-eb-in-o-s). Only the present sub-series express the imperfective aspect, all other forms are perfective (Hillery 2006). Melikishvili (2008: 146) states

Language Acquisition 2 6 9 Major Trends in Theoretical and Applied Linguistics

that the imperfective construction is unmarked and dominant in comparison to the aorist construction. All verbs have the imperfective series, but there are some with a defective paradigm lacking aorist and perfective series. The frequency of imperfective forms is higher than perfective and aorist. In Russian, there is also a rich verbal inflectional system, where even infinitives are marked for aspect and voice. There are three tenses (present, past, future) and two aspects (perfective and imperfective) (Borik 2002; Slabakova 2005). Present tense allows only imperfective aspect (imperfective present tense stem plus personal endings, as in čitau/–esh/–et/–em/–ete/– ut ‘I/you/she/we/you/they read.IMP’). Past tense allows both perfective (perfective past tense stem/infinitive stem plus suffix –l plus gender endings, as in pročital/–a/–o/–i ‘I/she/it they read.PERF’) and imperfective aspect (imperfective past tense stem/infinitive stem plus suffix –l plus gender endings, as in čital/–a/–o/–i ‘I/she/it they read.IMP’). Future tense allows both perfective (auxiliary byt’ ‘be’ in personal form plus the imperfective infinitive, as inbudu/–esh/–et/–em/–ete/–ut čitat’ ‘I/you/she/we/you/they will read.IMP.INF’) and imperfective aspect (perfective present tense stem plus personal endings, as in pročitau/–esh/–et/–em/–ete/–ut ‘I/you/she/we/you/ they will read.PERF’). The realization of aspect is done by means of morphology: The imperfective form of the verb is default; perfective aspect is achieved either by addition of perfective prefix to the imperfective stem of the verb (e.g., pisat ‘write. IMP.INF’) plus the prefix na to yield na-pisat ‘PERF-write.INF’) or by irregular formation, verb stem alternation, and stress shift (e.g., brat ‘take.IMP’ vs. vzyat ‘take.PERF.INF’) (Filin 1979: 40-41). Perfective forms can undergo a secondary imperfectivization through aspectual suffixation (e.g., otkyt ‘open.PERF’ vs. otkryvat ‘open.IMP’). In Greek there are eight tenses, which represent certain aspects: imperfect (elina ‘I solved.IMP’), aorist (elisa ‘I solved.PERF’), present (lino ‘I solve.IMP’), present perfect (eho lisi ‘I have solved.PERF’), past perfect (iha lisi ‘I had solved.PERF’), future progressive (tha lino ‘I will solve.IMP’), future simple (tha liso ‘I will solve.PERF’), and future perfect (tha eho lisi ‘I will have solved. PERF’). Perfective aspect is derived by vowel change of the stem of the imperfective verbal aspect form (e.g., trogo ‘I eat.IMP’ vs. efaga ‘I ate.PERF’) and/or changes of the final consonant of the imperfective verbal stem (e.g., a sigmatic morpheme as above in lino vs. elina vs. elisa) (Joseph & Smirniotopoulos 1993; Warburton 1970). The type of direct object influences the choice of telic/atelic predicate: definite objects/quantized incremental theme objects yield a telic/perfective interpretation, while cumulative incremental theme objects give rise to atelic/imperfective interpretation, and in this respect Greek differs from Russian (Krifka 1998; Dowty 1991; Verkuyl 1972).

2 7 0 Language Acquisition Sviatlana Karpava and Kleanthes K. Grohmann

The difference between Russian, Greek, and Georgian with respect to verb complementation, in particular the aspectual form of the embedded verb, is going to be examined in the present study to reveal possible transfer effects from L1 Russian into L2 Greek, L2 Russian in L3 Greek, or L1 Georgian in L3 Greek. In certain sentential environments, Russian allows two aspectual options in the complement verb (either perfective or imperfective), while Greek permits only one. More specifically: when the accomplishment main verb has a continuous interpretation (prospathusa ‘try’), the state main verb has a continuous interpretation (borusa ‘be able’), and when the main verb is perfective non-ingressive (fovithika ‘be afraid’), the complement verb should be only in perfective aspect; in Russian, it can be either perfective or imperfective. In addition, there is a restriction in Greek when the matrix verb thelo ‘want’ is in past tense; it then requires a perfective aspect finite verb in the embedded clause, which is not the case for Russian. Activity verbs normally do not allow na-complements; they only do so when the clause expresses a purpose or goal. Also, with na as a short form of gia na ‘in order to’ (Moser 1993: 572), the verb of the complement clause should be perfective in Greek and either perfective or imperfective in Russian. Regarding Georgian, it is known that in ‘want’-type clauses, the complementizer is omitted and the embedded verb has an infinitival form (Haspelmath 2002). Only the verbs of the aorist and perfective series (focused either on the completed event or the result/resultative state of the event and thus being perfective) can be used for past tense (Samsel 1992). The target sentences in Greek under investigation are in past tense; this means that if Georgian learners of Greek transfer from their L1, they would use the aorist and perfective verb series, and predominantly employ perfective aspect, meaning that imperfective condition should more difficult for them than perfective. But if they transfer from their L2 Russian, then the perfective condition might be more difficult than the imperfective. There can be a third option, namely to mix between the three languages. Discrepancies between aspectual paradigms and the way how aspect is encoded in a language cause difficulties in form-to-meaning relations, which is in line with the Morphological Salience Hypothesis (van Hout 2008), according to which form-to-meaning realations are influenced by how aspect is realized in a language: the more salient the morphological paradigm, the earlier the acquisition takes place. In Russian, aspect is realized by means of a prefix, but tense and agreement with the help of a suffix, while in Greek prefixes are not used at all for this purpose, only suffixes. So, Russian learners of L2/L3 Greek might have problem with the choice of the right morphological paradigm for aspect and with the mapping of verbal form to its meaning.

Language Acquisition 2 7 1 Major Trends in Theoretical and Applied Linguistics

2.3. Research Questions

The main research questions of the present study are the following:

1. Do learners transfer from L1 Russian to L2 Greek, or from L1 Georgian to L3 Greek, or even from L2 Russian to L3 Greek in the initial stage of acquisition, with respect to embedded aspect (supporting the Full Transfer/Full Access Hypothesis)? Can L2/L3 learners reach a native-like stage of performance or do they consistently show variability even at advanced levels?

2. Does typological distance between Georgian, Russian, and Greek play a big role in non-native production? Does the CG–SMG bilectal situation in Cyprus influence production of non-native participants in Cyprus in comparison to Mainland Greece?

3. Which type of main and embedded verb is the most problematic for L2/ L3 learners (e.g., state, activity, accomplishment, achievement)?

4. Which variables influence L2/L3 production (e.g., length of residence in the country, L2 usage, age of onset to L2, education, L2 proficiency, frequency of L2 usage)?

5. Is there a difference between child and adult L2/L3 production, and is it possible to test the Domain-by-Age Model and the Morphological Salience Hypothesis with the results of the experimental study?

3. The Present Study

3.1. Participants

308 participants took part in the written task: 75 bilingual Russian–CG speakers (25 adults, 50 children), 32 multilingual Georgian–Russian–CG speakers (7 adults, 25 children), 63 bilingual Russian–SMG speakers (13 adults, 50 children), and two control groups of 75 monolingual SMG speakers (25 adults, 50 children) and 63 monolingual CG speakers (25 adults, 38 children). More detailed data are summarized in Table 1, including length of residence (LoR), age of onset (AoO) of Greek, and the language acquisition (LA) type for the different groups:

2 7 2 Language Acquisition Sviatlana Karpava and Kleanthes K. Grohmann

Table 1. Participants for the written task

Group N Gender Age LoR AoO LA Type

male female years years years

Russian–CG adults 25 0 25 (100%) 36.2 8.8 28.4 adult L2 15.5 Russian–CG children 50 19 (38%) 31 (62%) 7.6 5.7 child L2 Georgian–Russian–CG 33 7 3 (43%) 4 (57%) 10.4 22 adult L3 adults Georgian–Russian–CG 25 12 (48%) 13 (52%) 15 7.3 7.3 child L3 children Russian–SMG adults 13 3 (23%) 10 (77%) 38.5 12.5 28.8 adult L2

Russian–SMG children 50 17 (34%) 33 (66%) 16 12.8 3.2 bilingual L1/L2

CG adults 25 10 (40%) 15 (60%) 39.8 L1

CG children 50 24 (48%) 26 (52%) 14 L1

SMG adults 25 5 (20%) 20 (80%) 35.4 L1

SMG children 38 16 (42%) 22 (58%) L1

124 participants took part in the oral video stimuli task: 36 Russian–CG speakers (18 adults and 18 children), 42 Russian–SMG speakers (5 adults and 37 children), and two control groups of 30 CG speakers (9 adults and 21 children), 16 SMG speakers (10 adults and 6 children). This information is provided in Table 2:

Table 2. Participants for the oral video stimuli task

Group N Gender Age LoR AoO LA Type

male female years years years

Russian–CG adults 18 1 (6%) 17 (94%) 31.3 8.6 22.7 adult L2

Russian–CG children 18 6 (33%) 12 (37%) 6.5 5.8 0.2 bilingual L1/L2

Russian–SMG adults 5 3 (60%) 2 (40%) 34.4 12.5 21 adult L2

Russian–SMG children 37 23 (62%) 14 (38%) 10.7 9.6 1.2 bilingual L1/L2

CG adults 9 4 (44%) 5 (56%) 36.4

CG children 21 12 (57%) 9 (43%) 5.4

SMG adults 10 3 (30%) 7 (70%) 31.2

SMG children 6 3 (50%) 3 (50%) 6.1

Language Acquisition 2 7 3 Major Trends in Theoretical and Applied Linguistics

3.2. Materials

3.2.1. Written Task

For the written study, the materials included a language history questionnaire (25 questions), modeled on Li et al. (2006), and a two-task test based on Malagardi (1993) and Moser (1993). There was a forced-choice task with 36 items (6 for each type of main verb, 30 items with a perfective target and 6 items for the imperfective condition), where participants had to choose between the two sentences that differed in the aspectual form of the embedded, and a cloze task with 36 items (6 for each type of main verb, 30 items with perfective target and 6 items for the imperfective condition), for which participants had to fill the gap with the complement verb with a relevant aspectual form. For the perfective condition, embedded environments were chosen where in Greek only perfective aspect is allowed, while in Russian either perfective or imperfective can be used; for the imperfective condition, those types of sentences were chosen which allow only imperfective aspect in both Russian and Greek. For the perfective condition, five different types of main verbs were used in a certain aspectual form which require only perfective aspect of the verb in the embedded clause: the accomplishment verb prospathusa ‘try’ with a continuous interpretation, the perfective non-ingressive state verb fovithika ‘be afraid of’, the perfective ingressive state main verb borese ‘be able’, the perfective volitional verb in the past tense ithela ‘want’, and activity verbs expressing a purpose or a goal with na being a short form of gia na ‘in order to’. The first four require complement clauses and the last one an adverbial clause. The sentences were supported by pictures, which restricted certain pragmatic conditions. One example is given in (1):

(1a) Προσπαθούσε να επιλέξει το δώρο για τη μαμά της. Prospathuse na epileksi to doro gia ti mama tis. ‘She was trying to choose (PERF) the present for her mum.’ (at that certain moment, not habitual) [TARGET] (1b) Προσπαθούσε να επιλέγει το δώρο για τη μαμά της. Prospathuse na epilegi to doro gia ti mama tis. ‘She was trying to choose (IMP) the present for her mum.’

For the imperfective condition, there was one type of the main verb arhise ‘start’, which requires embedded imperfective. An example is given in (2):

2 7 4 Language Acquisition Sviatlana Karpava and Kleanthes K. Grohmann

(2a) Άρχισε να χορεύει. Arhise na horevi. ‘He started to dance (IMP).’ [TARGET] (2b) Άρχισε να χορέψει. Arhise na horepsi. ‘He started to dance (PERF).’

3.2.2. Oral Video-Stimuli Task

The oral-production video task included 36 short video stimuli with 6 types of the main verbs (6 video clips for each type) and two conditions: perfective and imperfective. The video clips showed on-going actions. After watching the video clip, the participants were told the beginning of a sentence (in an oral mode), such as in (3a), and asked to complete the sentence with their own words, using their own embedded clause in the past tense (orally), such as (3b) for the condition and (4b) for the perfective condition. The session was audio-recorded and then analyzed accordingly.

(3a) To koritsi arhise na… ‘The girl began NA…’ (3b) … pleni ta heria ‘…to wash her hands-IMP’ (4a) To koritsi pire tin petra gia na… ‘the girl took the stone for NA…’ (4b) … petaksi sti thalassa ‘…throw-PERF into-the sea’

4. Results and Discussion

4.1. Written Task

Unsurprisingly, native speakers performed better than the bilingual Russian– SMG and Russian–CG participants as well as the multilingual Georgian– Russian–CG participants. In addition, the monolingual and bilingual participants in Greece performed better than the corresponding groups in Cyprus. Russian–SMG speakers performed best among the non-native groups, while the Georgian–Russian–CG speakers had the worst production; details are provided in Table 3.

Language Acquisition 2 7 5 Major Trends in Theoretical and Applied Linguistics

Table 3. Non-target test production of the groups.

GREECE CYPRUS

Non-target 63 Russian– 63 SMG adults 75 Russian– 75 CG adults+ 32 Georgian– production SMG adults+ + children CG adults+ children Russian–CG children children adults+ children

prospathuse 1.21% 0.33% 1.94% 0.38% 2.21%

fovithike 3.30% 1.10% 3,66% 2.46% 4.64%

mborese 0.97% 0.28% 1,35% 0.16% 3.47%

gia na 1.52% 1.21% 2.48% 1.20% 3.34%

arhise 1.52% 0.33% 2.09% 0.40% 3.29%

ithele 0.90% 0.11% 1.16% 0.31% 2.9%

Total 9.43% 3.37% 12.70% 4.94% 19.85%

The perfective condition was more problematic than imperfective for Russian non-native groups, since in L1 Russian both perfective and imperfective aspect is possible, so either positive or negative transfer might take place. The most problematic types of the main verbs for all groups were fovithike ‘afraid’, activity verbs with gia na ‘in order to’ (perfective condition). According to an ANOVA (means) statistical analysis, there is a statistically significant difference between Russian–CG speakers and the control groups of CG and SMG speakers, between Russian–SMG speakers and a control group of CG speakers, and between Georgian–Russian–CG speakers and all other native as well as non-native groups. The results are given in Table 4:

Table 4. ANOVA statistics analysis

Group Group t-value DF Prob.

CG Georgian–Russian–CG –5.532 105 .0000*

Russian–CG CG 3.947 148 .0001*

Russian–CG SMG 2.058 136 .0415*

Russian–CG Georgian–Russian–CG –2.480 105 .0147*

Russian–SMG CG 3.523 136 .0006*

Russian–SMG Georgian–Russian–CG –2.608 93 .0106*

SMG Georgian–Russian–CG –4.032 93 .0001*

2 7 6 Language Acquisition Sviatlana Karpava and Kleanthes K. Grohmann

Again unsurprisingly, the test production of monolingual children and adults was better than that of bilingual children and adults in both countries. Bilingual Russian–CG adults performed better than bilingual Russian–SMG adults, however, while monolingual SMG adults performed better than CG adults. Children performed better than adults, except for the monolingual CG and the multilingual Georgian–Russian–CG groups. Russian–SMG speakers performed best among the non-natives and Georgian–Russian–CG children had the worst production, with SMG children performing better than CG children. Table 5 provides the details:

Table 5. Non-target test production: adults vs. children

GREECE CYPRUS

non-target 13 Russian–SMG 25 SMG 25 Russian–CG 25 CG 7 Georgian– production adults adults adults adults Russian–CG adults

Total 28.41% 2.33% 15.7% 4.38% 14.48%

25 Georgian– non-target 50 Russian–SMG 38 SMG 50 Russian–CG 50 CG Russian–CG production children children children children children

Total 4.5% 4.05% 11.16% 5.22% 21.38%

Embedded activity verbs were the most diffi cult as they can be associated both with perfective and imperfective aspect:

Figure 1. Non-target production: embedded verbs

As it can be seen from the following fi gures, it is obvious that variables such as length of residence in the country (which by and large goes with length of exposure to Greek) and age of onset play an important role in participants’ test production:

Language Acquisition 2 7 7 Major Trends in Theoretical and Applied Linguistics

Figure 2. Non-target production and length of residence in L2/L3 country

Figure 3. Non-target production and the age of onset to L2/L3

But there are certain diff erences among the groups that the bilingual Russian–CG and Russian–SMG participants have more in common, while the multilingual Georgian–Russian–CG speakers exhibit a diff erent pattern. It seems that the participants with 6-10 years of exposure to Greek (all three non-native groups) behave in a similar way concerning the test production, as can be seen from Figure 2. Figure 3 shows that an early age of onset makes a markedly bigger diff erence for the L1 Russian bilingual speakers than the L1 Georgian multilingual participants.

4.2. Oral Task

The results of the video task are consistent with the written task. The non- native groups of bilingual Russian–CG and Russian–SMG speakers performed worse than the native group of monolingual CG and SMG participants. Children in both native and non-native groups performed better than adults, except for the SMG group, where performance was the same. Russian–SMG speakers performed better than Russian–CG participants. The most problematic types

2 7 8 Language Acquisition Sviatlana Karpava and Kleanthes K. Grohmann

of the main verbs for all groups were fovithike ‘be afraid’ and activity verbs with gia na ‘in order to’ (perfective). The results are provided in Table 6:

Table 6. Non-target production in the oral video stimuli task

Non-target Russian–CG CG Russian–SMG SMG production

adults 18 9 5 10

children 18 21 37 6

prospathuse 6.01% 1.23% 0.26% 0% 1.11% 0.15% 0.27% 0.46%

mborese 2.93% 0.92% 0.26% 0% 0% 0.07% 0.27% 0.46%

fovithike 4.32% 1.54% 0.26% 0.30% 0.55% 0.15% 0.27% 0%

ithele 3.85% 0.61% 0.52% 0% 0.55% 0.07% 0.22% 0.46%

giana 4.01% 0.77% 0.13% 0% 3.88% 0.37% 0% 0.46%

arhise 2.77% 1.23% 0.13% 0% 0% 0% 0% 0%

total non-target 23.91% 6.32% 1.58% 0.30% 6.11% 0.75% 1.38% 1.38%

The results from the written and oral experiments are in line with the Lexical Aspect Hypothesis (Andersen 1991; Salaberry 1998), according to which learners first acquire perfective and then imperfective aspect, and the telicity of the verbal predicate influences the mapping between lexical aspect and verb morphology in the initial stage of learning. So, at the beginning of the L2 acquisition process, dynamic and durative predicates are combined with imperfective aspect, and telic predicates with perfective aspect. The pictures and videos in the experiments showed on-going events, so the L2 beginners were more likely to choose the imperfective aspect in comparison to the L2 intermediate or advanced learners (see e.g. Chin 2009 and the collection of papers in Kempchinsky & Slabakova 2005). Besides, according to Borik (2002), imperfective aspect in Russian describes not only on-going and episodic situations, but also knowledge about facts, so the participants might have chosen the imperfective aspect instead of the correct perfective simply to state the fact about the picture or a video in front of them. According to an ANOVA analysis, there is a statistically significant difference only between Russian–CG speakers and the CG and SMG control groups:

Language Acquisition 2 7 9 Major Trends in Theoretical and Applied Linguistics

Table 7. ANOVA statistics analysis

Group Group t-value DF Prob.

SMG Russian–CG –5.555 50 .0000*

CG Russian–CG –6.504 64 .0000*

Russian–CG Russian–SMG 7.38 76 .0000*

Again, embedded activity verbs were the most diffi cult, especially for the Russian– CG group, as they can be associated both with perfective and imperfective aspect:

Figure 4. Non-target production: embedded verbs

Children, more than adults, substituted embedded aspectual forms with general all purpose (GAP) or light verbs, such as kano banio ‘have a shower’ that do not have aspectual specifi cation in Greek:

Figure 5. Non-target production of GAP verbs

Variables such as the length of residence in the L2 country (only for the Russian–SMG group) and age of onset to L2 (both for the Russian–SMG and the Russian–CG groups) play a crucial role in non-native production: the younger the participants, the better the test production.

2 8 0 Language Acquisition Sviatlana Karpava and Kleanthes K. Grohmann

Figure 6. Non-target production and the length of residence in L2 country

Figure 7. Non-target production and the age of onset to L2

4.3. Written vs. Oral Task

Comparing the written and oral tasks (two modes of the test: oral and written), it can be seen that all groups, except for the Russian–CG adults, performed better in the oral task than in the written task. This means that the oral mode is easier for the participants than the written mode:

Figure 8. Written vs. oral task non-target test production

Language Acquisition 2 8 1 Major Trends in Theoretical and Applied Linguistics

Figure 9. Written perfective vs. imperfective task performance

Figure 10. Oral perfective vs. imperfective task performance

The two graphs above show the groups’ production in written (Figure 9) and oral task (Figure 10) in terms of perfective and imperfective conditions: the participants performed slightly better for the imperfective than for perfective conditions in both written and oral tasks, but this diff erence is not signifi cant. The behavior of adult Russian–SMG and adult Russian–CG participants in the written and oral tests diff er.

5. Conclusion

In general, the results of the study can be taken to support the Full Transfer/ Full Access Hypothesis (Schwartz & Sprouse 1994; Montrul & Slabakova 2003; Slabakova 2005), although there was no big diff erence between production for perfective and imperfective conditions, meaning that Russian learners of Greek use either positive or negative transfer L2 learners of Greek transfer from L1 Russian at the initial stage of acquisition and then with more exposure to L2

2 8 2 Language Acquisition Sviatlana Karpava and Kleanthes K. Grohmann

transfer decreases. The behavior of the Georgian mulitilingual group was not consistent with the bilingual Russian non-native groups. Russian is an overlapping language for the groups of Russian–CG, Russian– SMG, and Georgian–Russian–CG speakers, though for the first two groups this language is their L1, while for the third group it is L2. The difference in the error patterns between the Russian and the Georgian groups might be due to the order and direction of transfer: L3 learners (as L2 learners) tend to transfer from L1 Georgian to L3 rather than from L2 Russian; alternatively (or in addition), it may be because of the extra cognitive load: imperfective aspect is unmarked in Georgian, screeves are characterized by feature complexity, but only aorist and perfective verb series (i.e. perfective aspect) can be used with the past tense, so transferring from their L1 Georgian, they tend to use perfective aspect instead of imperfective in the past tense (imperfective condition). Errors in the perfective condition, when imperfective aspect is used instead of perfective, can be explained by the unmarked feature of the imperfective aspect in Georgian. Another issue is that the research was conducted in two countries, and the participants in Cyprus are exposed to two varieties of Greek: CG and SMG. This complex socio-linguistic, bi(dia)lectal, diglossic situation (documented in Grohmann & Leivada 2012 and Rowe & Grohmann 2013, for example) increases the difficulty of L2 acquisition of Greek and puts the L2 population in Cyprus in a more disadvantageous position than of the L2 population in Greece. Russian– CG speakers have to switch among three modes: CG, SMG, and Russian. The Georgian–Russian–CG group has a more difficult task, as they need to use CG, SMG, Russian, and Georgian on a daily basis. So, non-native residents of Cyprus require more complex patterns in the activation of languages in language production and perception, being in a multilingual setting (Ceñoz et al. 2001), than the relevant population in Greece, who are in a non-dialectal bilingual setting. Typological distance between languages can explain transfer preferences to some extent: Russian–CG and Russian–SMG speakers transfer from L1 Russian into L2 Greek, since Russian and Greek are typologically (relatively) closely related, belonging to the Indo-European language family; Georgian–Russian–CG speakers, though with L2 knowledge of Russian, prefer to transfer from their L1 Georgian, which is a non-Indo-European Caucasian language and typologically more distant from Greek. The successful test production of the native control groups, both in Greece and in Cyprus, shows that the experimental tools are reliable. Error analysis showed that the most problematic type of the main verb for L2/L3 learners was state verbs, as they might be easily associated with imperfective rather than perfective aspect, supporting the Aspect Hypothesis (Andersen 1991; Shirai & Li 2000). Embedded activity verbs caused more non-target production than the

Language Acquisition 2 8 3 Major Trends in Theoretical and Applied Linguistics

other types of verbs (achievement, accomplishment, and state verbs), as they can have either perfective or imperfective interpretation. The study also revealed some differences between children and adult L2 production: L2 children were closer to L1 children rather than L2 adults, which can partially support the Domain-by-Age Model (Schwartz 2003), according to which there is a similarity between child L2 and child L1 acquisition in the domain of inflectional morphology, but in the domain of syntax, child L2 acquisition should rather be more similar to adult L2 acquisition; age of onset should then only influence the domain of inflectional morphology and L1 transfer the domain of syntax. Since aspect belongs to the syntax–morphology interface, some overlap may be expected (which may in the future be further tested with the Interface Hypothesis of Tsimpli 2003, for example). The Morphological Salience Hypothesis can explain the difficulties of L2 learners in the correct choice of embedded aspect, with perfective aspect more difficult than imperfective. L2 learners cannot easily map form to meaning, as Greek and Russian have different aspectual paradigms. Another interesting point is that L2 children, more than adults, substituted verb forms (perfective or imperfective) with general all purpose or light verbs, which do not have aspect specification in Greek. This issue will be picked up in the future in more detail but is in line with concurrent studies on language development in Greek (see Grohmann & Leivada 2013; Kambanaros & Grohmann 2013). One important issue of this study is that a large sample of participants was used to compare the acquisition of L2 Greek embedded aspect in monolingual (Greece) and bilectal settings (Cyprus), with SMG as the overlapping variety of Greek, by learners with one prior language (L1 Russian) and learners with two languages (L1 Georgian and L2 Russian), with overlapping Russian, in written and oral as well as comprehension and production modes.

2 8 4 Language Acquisition Sviatlana Karpava and Kleanthes K. Grohmann

References

Andersen, R. 1991. Developmental sequences: the emergence of aspect marking in second language acquisition. In T. Huebner and C. Ferguson (eds.), Crosscurrents in Second Language Acquisition and Linguistic Theories. Amsterdam: John Benjamins, 305-324.

Borik, O. 2002. Aspect and Reference Time. PhD Dissertation, University of Utrecht.

Ceñoz, J., B. Hufeisen and U. Jessner. 2001. Introduction. In J. Ceñoz, B. Hufeisen and U. Jessner (eds.), Cross-Linguistic Influence in Third Language Acquisition: Psychological Perspectives. Clevedon, OH: Multilingual Matters, 1-7.

Chin, D.H.-J. 2009. Cross-linguistic Effects on the Acquisition of L2 Aspect: A Bi-directional Investigation. Saarbrücken: VDM Verlag Dr. Müller.

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

Dowty, D. 1991. Thematic proto-roles and argument selection. Language 67: 547-619.

Filin, F. 1979. Russkij jazyk: Enciklopedija. Moscow: Sovetskaja enciklopedija.

Grohmann, K.K. and E. Leivada. 2013. Lightverbhood in child language: Evidence from Cypriot Greek. In E. Blom, J. Verhagen and I. van de Craats (eds.), Dummy Auxiliaries in First and Second Language Acquisition. Berlin: De Gruyter Mouton, 141-170.

Grohmann, K.K. and E. Leivada. In press. Lightverbhood in child language: Evidence from Cypriot Greek. In E. Blom, J. Verhagen and I. van de Craats

Language Acquisition 2 8 5 Major Trends in Theoretical and Applied Linguistics

(eds.), Dummy Auxiliaries in First and Second Language Acquisition. Berlin: De Gruyter Mouton.

Harris, A. 1981. Georgian Syntax: A Study in Relational Grammar. Cambridge: Cambridge University Press.

Haspelmath, M. 2002. On the cross-linguistic distribution of same-subject and different subject ‘want’ complements: Economic vs. iconic motivation. Paper presented at the Typology Symposium at the University of Frankfurt am Main, November 2002.

Hiller, P.J. 2006. The Georgian language: An outline grammatical summary. Ms., http://www.armazi.com.

van Hout, A. 2008. Acquiring perfectivity and telicity in Dutch, Italian and Polish. Lingua 118: 1740-1765.

Joseph, B. and J. Smirniotopoulos. 1993. The morphosyntax of the Modern Greek verb as morphology and not syntax. Linguistic Inquiry 24: 388-398.

Kambanaros, M. and K.K. Grohmann. In press. More GAPs in children with SLI? Evidence from Greek for not fully lexical verbs in language development. Applied Psycholinguistics.

Kempchinsky, P. and R. Slabakova (eds.). 2005. Syntax, Semantics, and Acquisition of Aspect. New York: Springer Publishers.

Krifka, M. 1998. The origins of telicity. In S. Rothstein (ed.), Events and Grammar. Dordrecht: Kluwer, 197-235.

Li, P., S. Sepanski and X. Zhao. 2006. Language History Questionnaire: A web- based interface for bilingual research. Behavior Research Methods 38: 202- 210.

Malagardi, I. 1993. Problems of Greek aspect morphology and the identification of projection for tense and aspect. In I. Philippaki-Warburton, K. Nicolaidis and M. Sifianou (eds.), Themes in Greek Linguistics. Amsterdam: John Benjamins, 161-169.

Melikishvili, I. 2008. Georgian as an active ergative split language. Bulletin of the Georgian National Academy of Sciences 2: 138-147.

2 8 6 Language Acquisition Sviatlana Karpava and Kleanthes K. Grohmann

Montrul, S. and R. Slabakova. 2003. Competence similarities between native and near-native speakers: An investigation of the preterite/imperfect contrast in Spanish. Studies in Second Language Acquisition 25: 351-398.

Moser, A. 1993. The interaction of lexical and grammatical aspect in Modern Greek. In I. Philippaki-Warburton, K. Nicolaidis and M. Sifianou (eds.), Themes in Greek Linguistics. Amsterdam: John Benjamins, 137-145.

Roussou, A. 2009. Selecting complementizers. Lingua 119: 1811-1836.

Rowe, C. and K.K. Grohmann. 2013. Discrete bilectalism: Towards co-overt prestige and diglossic shift in Cyprus. International Journal of the Sociology of Language 224: 119–142.

Salaberry, R. 1998. The development of aspectual distinctions in academic L2 French. Canadian Modern Language Review 54: 508-542.

Samsel, J. 1992. A role and reference grammar analysis of Georgian morphosyntax. MA Thesis, University of California, Davis.

Schwartz, B.D. 2003. Child L2 acquisition: paving the way. In B. Beachley, A. Brown and F. Conlin (eds.), Proceedings of the 27th Annual BUCLD. Somerville, MA: Cascadilla Press, 26-50.

Schwartz, B.D. and R. Sprouse. 1994. Word order and nominative case in nonnative language acquisition: A longitudinal study of (L1 Turkish) German interlanguage. In T. Hoekstra and B.D. Schwartz (eds.), Language Acquisition Studies in Generative Grammar. Amsterdam: John Benjamins, 317-368.

Shirai, Y. and P. Li. 2000. The Acquisition of Lexical and Grammatical Aspect. Berlin: Mouton de Gruyter.

Slabakova, R. 2005. What is so difficult about telicity marking in L2 Russian? Bilingualism: Language and Cognition 8: 63-77.

Smith, C. 1991 (19972). The Parameter of Aspect. Dordrecht: Kluwer.

Tsimpli, I.-M. 2003. Clitics and determiners in L2 Greek. In J. Liceras, H. Zobl and H. Goodluck (eds.), Proceedings of the 6th Conference on Generative Approaches to Second Language Acquisition. Somerville, MA: Cascadilla Proceedings Project, 331-339.

Language Acquisition 2 8 7 Major Trends in Theoretical and Applied Linguistics

Vendler, Z. 1967. Verbs and times. In Z. Vendler (ed.), Linguistics in Philosophy. Ithaca, NY: Cornell University Press, 97-121.

Verkuyl, H. 1972. On the Compositional Nature of the Aspect. Dordrecht: D. Reidel.

Warburton, I. 1970. On the Verb in Modern Greek. Bloomington, IN: Indiana University.

2 8 8 Language Acquisition The Acquisition of Motion Verbs in Greek: Evidence from Comprehension1

Maria Kotroni

Aristotle University of Thessaloniki [email protected]

Abstract

This study addresses the issue of how motion verbs interact with aspect in the expression of manner of motion regarding child language acquisition. More specifically, we investigate children’s and adults’ preferences in interpreting motion events as telic, atelic locative or atelic directional when expressed by

an ambiguous manner of motion verb followed by a PP(path/goal) (real or novel

and in both aspectual forms). Within two offline comprehension sentence- video matching tasks, the results from the control group indicate that in Greek the interpretation of manner of motion verbs and the attribution of their aspectual values involves the integration of: morphological aspect, motion event and structural information. However, the results from Greek children point to inappropriate aspectual knowledge of aspect at discourse level and a developmental pattern in the acquisition of such integration, due to its compositional nature. Thus, this study concludes in favor of an account of Lexicon-Morphosyntax Interface.

1. Introduction

The present study is an account of lexicon-morphosyxntax interaction in the expression of manner of motion. More specifically, the study addresses the issue of how motion verbs interact with aspect in the expression of manner of motion and how this interaction evolves in child language acquisition of Greek. We investigate children’s and adults’ preferences in interpreting motion events as telic, atelic locative or atelic directional when expressed by an ambiguous

1 This study has been funded by the Greek State Scholarship Foundation (IKY), Leventis Foundation, Propondis Foundation, as well as from the Greek Ministry of Education.

Language Acquisition 2 8 9 Major Trends in Theoretical and Applied Linguistics

manner of motion verb followed by a PP(path/goal), (adjunct or complement). The aspectual form of the verb (perfective or imperfective) was also a dependent variable. For example, in Greek the sentence η πεταλούδα πέταξε στην κουζίνα is ambiguous: a. the butterfly was outside the kitchen and it was flying towards the kitchen but the event stops before the butterfly reaches the kitchen (atelic directional), b. the butterfly was outside the kitchen and it was flying towards and reaches the kitchen (telic-endpoint reached) and c. the butterfly was inside the kitchen and it was flying there (atelic locative). Moreover, the sentence η πεταλούδα πετούσε στην κουζίνα is also ambiguous between a locative and a directional reading (a. the butterfly was outside the kitchen and it was flying towards the kitchen but the event stops before the butterfly reaches the kitchen b. the butterfly was inside the kitchen and it was flying there. The study comprised both real and novel verbs, in order to examine the role of aspect independently from verb meaning. Previous studies on lexicon-morphosyntax interface follow two basic accounts. Reseachers such as Mac Whinney 1999, Bates and Goodman 1997, Tomasello 2000, Marchman and Bates 1994 claimed in favour of a Lexicon first account. On the other hand, researchers such as Gleitman 1990, Landau and Gleitman 1985, Naigles 1990, Naigles & Terrazas 1998, Fisher et al 1994 supported a Morphosyntax first account (also called syntactic bootsrapping) in the interpretation of such an interaction. In some previous comprehension studies, researchers conclude that 3 to 5 year old Polish and Russian speaking children know the aspectual semantics of perfective and imperfective aspect (Weist et al.1991, Vinnitskaya and Wexler 2001, Stoll 1998). However in two other comprehension studies (Kazanina and Philips 2002 and van Hout 2002 for Russian, Polish and Dutch), 3 to 5 year old children have acquired the aspectual semantics of the perfective-imperfective aspect, but they do not employ it appropriately in a pragmatic sense, i.e. they have not yet acquired the interface with discourse, knowledge intergrating aspect in discourse is not yet fully developed, which in turn inhibits the proper interpretation of the aspects. The same is the case with Italian, where aspectual meanings are conflated by the various past tenses (Hollebrandse and van Hout 2001). Even the 5-year olds did not show full, discourse-appropriate understanding of aspectual meanings. Moreover, no developmental advantage of aspect as a dedicated grammatical category is found. The present study comprises two offline comprehension sentence-video matching tasks which aim at investigating the degree native speakers of Greek use the perfective-imperfective distinction to differentiate between locative and directional atelic readings and telic-endpoint reached and atelic-locative interpretations. The dependent variable is motion and the independent variables are aspect, verb type (real vs novel) and age group.

2 9 0 Language Acquisition Maria Kotroni

Based on Horrocks and Stavrou 2003, 2007, Papadopoulou 1996, Papafragou and Selimis 2010, we predicted (see Table 1) that imperfective aspect mostly denotes atelic locative events with the PP being an adjunct. In a more restricted use, imperfective aspect can also denote atelic directional events, but in that case the verb requires a complement PP usually introduced by the preposition pros. On the other hand, Greek perfective aspect is ambiguous between a telic and an atelic reading. In the telic reading, the PP is a complement. This use is more strongly preferred with perfective aspect. In the atelic reading of the perfective aspect, which is more restricted in use, the PP can be either a complement or an adjunct. Consequently, imperfective aspect and atelic readings are predicted to prevail over perfective aspect and telic readings respectively. Therefore, the aspectual value of verb stems involves the combination of the aspectual character with the choice of grammatical aspect.

Table 1. Predictions

Context Perfective Imperfective

Atelic locative (PP adjunct) √ (restricted) √√

Atelic directional (PP complement) √ (restricted) √ (very restricted, mostly with pros)

Telic-endpoint reached (PP complement) √√

2. The Pilot Study

The pilot study, conducted on 122 adults native speakers of Greek, was an acceptability judgment task (in a 5-likert scale questionnaire) which aimed at identifying the hierarchy of acceptability regarding locative and directional readings and the role of aspect and preposition in the interpretation of the sentence. This study resulted in the recruitment of the critical items used in the main study. In Table 2, we identify the conditions tested.

Table 2. The conditions tested

+ tempP se/punctual (directional- PP complement) Perfective aspect + tempP ja/durative (locative- PP adjunct)

+ tempP se/punctual (directional- PP complement) Imperfective aspect + tempP ja/durative (locative- PP adjunct)

Language Acquisition 2 9 1 Major Trends in Theoretical and Applied Linguistics

Based on Tsimpli and Papadopoulou 2009, we assume that both perfective and imperfective can be interpreted as directional (PP-complement) or locative (PP- adjunct). However, the imperfective cannot have a telic interpretation. Two word orders were tested:

a. S V[+perf/imperf] PPpath PPtemp(se mia ora/ja mia ora)

The temporal PP does not intervene between the V and PP(path), therefore there is no bias in favour of or a restriction for an adjunct/complement interpretation

of the PPpath.

b. S V[+perf/imperf] PPtemp (se mia ora/ja mia ora) PPpath

The temporal PP intervenes between the V and PP(path), which could in a sense favour the adjunct interpretation. However, Greek has free word order which restricts such a bias. Regarding the first word order, the conditions that are predicted to be rated higher are the ones with ja mia ora (durative-atelic) with either perfective or imperfective aspectual form and the adjunct interpretation (atelic locative) is favoured over the complement one. On the other hand, se mia ora (punctual- telic) cannot combine with imperfective aspect. In Table 3, there is an overview of the means in all four conditions.

Table 3.

Preference (mean):V-PPpath-PPtemp

Conditions Mean Std.Deviation

Perfective-se 2.01 1.3

Perfective-ja 3.14 1.593

Imperfective-se 1.65 1.101

Imperfective-ja 3.61 1.567

Total se 1.83 1.217

Total ja 3.38 1.596

In both participant and item analysis, the main effect of aspect (F1,58= ,621, p=,434 and F1,23= ,191, p=,666 respectively) was not significant. However, the main effect of preposition was significant (F1,58= 228,439, p=,000 and F1,23=49,802, p=,000) as well as the interaction between aspect and preposition (F1,58= 43,527, p=,000 and F1,82= 18,260, p=,000). Moreover, the difference in the aspect using the same preposition is significant (se: t=5,601, df=58, p=,000 and ja: t=-4,232, df=58 and p=,000 for participant analysis and se: t=3,987, df=23, p=0,001 and ja: t=2,783, df=23 and p=0,011 for item analysis), which is also the case with the difference in the preposition keeping the aspectual type

2 9 2 Language Acquisition Maria Kotroni

constant (participant analysis: t=10,516, df=58, p=,000 for perfective aspect and t=14,831, df=58 and p=,000 for imperfective aspect and item analysis: t=4,694, df=23, p=,000 for perfective aspect and t=8,442, df=23 and p=,000 for imperfective aspect). The difference between the punctual preposition se and the durative preposition ja is also found significant (participant=15,066, df=58 and p=,000 and item: t=6,939, df=23 and p=,000). Therefore, in both participant and item analysis, the predictions regarding the role of aspect and preposition are verified.

Regarding the second word order, we predicted that, since the PPtemp ja mia

ora favours locative atelic readings (PPpath=adjunct), it is compatible with both perfective and imperfective aspect and thus is expected to be rated higher than se mia ora, which is only compatible with perfective aspect and available only with a telic reading. Moreover, imperfective aspect is predicted to show higher ratings with ja mia ora, whereas perfective aspect is predicted to show almost similar ratings with se mia ora and ja mia ora. An overview of the means in all four conditions appears in Table 4.

Table 4.

Preference (mean): V-PPtemp-PPpath

Conditions Mean Std.Deviation

Perfective-se 2.08 1.345

Perfective-ja 3.25 1.572

Imperfective-se 1.86 1.283

Imperfective-ja 3.46 1.556

Total se 1.97 1.318

Total ja 3.35 1.567

In the inferential analysis, the main effect of aspect was not significant (participant: F1,82=,024, p=,877 and item: F1,23= ,025, p=,876). However, the main effect of preposition was significant (participant: F1,82= 148,177, p=,000 and item: F1,23= 81,978, p=,000), as well as the interaction between aspect and preposition (participant: F1,82= 8,185, p=,005 and item: F1,82= 11,430, p=,003). The difference in the aspect using the same preposition is significant (participant: t=2,629, df=8, p=0,010 when the preposition is se and t=1,927, df=82 and p=0,57 (marginal) when the preposition is ja and item analysis: preposition se t=2,457, df=23, p=0,022 and but preposition ja t=1,706, df=23 and p=0,102-one tailed). The difference in the preposition keeping the aspectual type constant is also significant (participant analysis: t=9,852, df=82,

Language Acquisition 2 9 3 Major Trends in Theoretical and Applied Linguistics

p=,000 for perfective aspect and t=10,569, df=82 and p=,000 for imperfective aspect and item analysis: t=7,817, df=23, p=,000 for perfective aspect and t=8,962, df=23 and p=,000 for imperfective aspect). The difference between the punctual preposition se and the durative preposition ja is also found significant (participant analysis: t=12,134, df=82 and p=,000 and item analysis: t=8,982, df=23 and p= ,000). Once more, the predictions regarding the role of preposition and aspect are verified. In the pilot study, we also tested each item separately in all four conditions, in order to classify them in a hierarchy regarding the inherent role of aspect and recruit the most appropriate ones (See the appendix for the means of the responses and the standard deviations. No statistical analysis is included because of lack of space). The pilot study concluded that some verbs are ambiguous and the role of aspect is significant, whereas some other verbs are interpreted as atelic locative or directional regardless of aspect. There were also intermediate points between the two ends of the hierarchy. In Table 5, the verbs tested are classified, on the basis of the means of the responses for each item separately in each of the four conditions.

Table 5. Hierarchy of acceptance

Verbs interpreted as Ambiguous verbs- Verbs with a tension Verbs interpreted as atelic-directional (low role of aspect to be interpreted as atelic-locative effect of aspect) significant atelic-locative (low effect of aspect)

- στρίβω - πηδώ - μπουσουλώ - κυλώ

- σκαρφαλώνω - κολυμπώ - σέρνομαι

- γλιστράω - οδηγώ - τρέχω

- περπατώ

- πετώ

3. The Main Study: Method

3.1. Participants

Three age groups comprised of 272 monolingual native speakers of Greek were recruited: one group of adult monolingual native speakers of Greek (control group-76 adults) and two groups of Greek monolingual children (1st group:

2 9 4 Language Acquisition Maria Kotroni

mean age 5;7, 90 children, and 2nd group: mean age 9;6, 106 children). Children and adults with a suspicion of low IQ, language, hearing and visual deficits were excluded.

Table 6. Profile of participants

Locative-directional task Telic-atelic task Age group N of Participants Mean N of Participants Mean

1st group 50 5;7 40 5;7

2nd group 58 9;6 48 9;6

Adults-control 36 35;3 40 20;4

total 144 128

3.2. Materials

The critical sentences denote a motion event expressed by a manner of motion verb followed by a path PP. The tasks consist of 36 experimental items (12 real verbs, 12 novel verbs and 12 fillers) and 36 couples of videos. Each couple of videos illustrated each experimental item in two different events. Regarding the first task, they illustrated either a locative event (e.g. a man was in the living room and he was running there) or a directional event (e.g. a man was in the kitchen and he was running towards the living room). Regarding the second task, each couple of videos illustrated either an atelic-locative event (e.g. a man was in the living room and he was running inside it) or a telic event with a reached endpoint (e.g. a man a man was in the kitchen and he run and reached the living room). In each motion event, the verb appeared in both aspectual forms (perfective and imperfective) in different versions. Therefore each item appeared in four conditions. The sentences and the couples of videos were divided into two versions so that each participant never saw the same set of videos more than once and never heard the same item more than once. The items and the videos were pseudorandomized. The choice of the real items was based on the pilot study, whereas the formation of the novel items followed past-tense formation rules and patterns regarding Greek as they appear in Holton et al 1997, Triantafyllides 2000, Ralli 1988, Klairis and Babiniotis 2004. The conditions tested were the following as shown in Table 7.

Language Acquisition 2 9 5 Major Trends in Theoretical and Applied Linguistics

Table 7. Conditions

Locative (PP adjunct) atelic vp[vp[V…]PP]] Perfective+ PPpath Directional e.g. O antras etrekse sto saloni (PP complement) atelic the . man ran s-the . living room . NOM.S.M NOM.S.M. PERF.3S. ACC.S.N ACC.S.N vp[V PP] The man ran into/in the living room. Endpoint reached (PP complement) telic vp[V PP] Locative (PP adjunct) atelic Imperfective+ PPpath vp[vp[V…]PP]] e.g. O antras etrexe sto saloni the man ran s-the living room NOM.S.M. NOM.S.M. IMPERF.3S. ACC.S.N. ACC.S.N. Directional The man was running into/in the living room. (PP complement) atelic vp[V PP]

3.3. Procedure

The participant sits in front of a computer screen and listens to an audio recorded sentence which contains a manner of motion verb either in perfective or imperfective aspect. Then he/she watches two videos simultaneously and finally he/she listens to the same sentence again. The subject has to choose which video matches the sentence the best.

4. Results

4.1. Locative vs. Directional Task

The aim of this task was to investigate the role of the aspectual verb form in the comprehension of ambiguous manner-of-motion verbs. More specifically do native speakers of Greek use the perfective/ imperfective distinction to differentiate between locative and directional readings? The variables tested in this task appear in Table 8. As is clear from Figure 1, a basic pattern is observed. All three age groups prefer the locative interpretation with imperfective aspect. The same is the case with perfective aspect regarding children. However, adults prefer the directional interpretation with perfective aspect. Such a tendency is observed even within the second group, where the gap between a locative and a directional interpretation is narrowing. Therefore, the prediction that a locative interpretation with imperfective aspect prevails over a directional one with perfective aspect is verified.

2 9 6 Language Acquisition Maria Kotroni

Table 8. Dependent and Independent variables tested

1. directional interpretation Dependent variable a. Directed motion 2. locative interpretation

1. perfective a. Aspect 2. imperfective

1. mean age 5;7 (range 1;7)

Independent variables b. Age group 2. mean age 9;6 (range 1;8)

3. Adults-control group

1. real verbs c. Verb type 2. novel verbs

R eal v erb s 100 80 38 69 60 77 81 86 82 loc ativ e 40 62 direc tional 20 31 Preference (%) 23 19 14 18 0 perfective perfective perfective imperfective imperfective imperfective 1st group 2nd group adults Ag e g ro u p s

Figure 1. Descriptive Statistics-Real verbs

Regarding novel verbs, Figure 2 makes clear that a locative interpretation outnumbers the directional one within imperfective aspect for all age groups and within perfective aspect in the case of children. In contrast to real verbs, adults assigned a locative interpretation with perfective aspect which is however near chance level. More specifially, in an inferential analysis, a three way mixed anova pointed out that, in the participant analysis of real verbs, the main effect of aspect (F(1,141)= 31.250, p= 0.000) and the aspect by group interaction were significant

Language Acquisition 2 9 7 Major Trends in Theoretical and Applied Linguistics

Novel verbs

100

80 58 60 74 77 locative 84 91 91 40 directional

preference (%) preference 20 42 26 23 16 0 9 9 perfective perfective perfective imperfective imperfective imperfective 1st group 2nd group adults age groups

Figure 2. Descriptive statistics-Novel verbs

(F(2, 141) =24.653, p= 0.000). The main effect of verb type (real or novel) (F(1,141)= 70.086, p= 0.000), as well as the verb type by group interaction (F(2, 141) =26.583, p= 0.000) were also significant. Additionally, the aspect-by-verb type interaction was significant (F(1,141)= 7.644, p= 0.006) and the three-way interaction between aspect, verb type and group was marginally significant (F(2,

141) =2.803, p= 0.064). Finally, the main effect of age was significant (F(1,141) = 22.750, p = 0.000). Regarding the role of aspect in the participant analysis, within the 1st group, within real verbs (t= 0.431, df=49, p < 0.334, one-tailed) and within novel verbs (t= 0.586, df=49, p <0.561, one-tailed), the difference between perfective and imperfective aspect is not significant. Therefore, the 1st group children’s preference for locative interepretation over a directional one is not an effect of aspect. Regarding the second group, in the group of real verbs, there was a significant difference between the conditions (z=4.255, N-Ties= 46,p=0.000 ). In the group of novel verbs there is a marginally significant difference between the conditions (z=1.878, N-Ties= 40, p=0.060). Therefore, the 2nd group children’s preference for locative interpretation over a directional one is an effect of aspect. Finally, within the 3rd group, within real verbs, the difference between perfective and imperfective aspect is significant (t= 10.736, df=35, p < 0.000). Therefore, the adults’ preference for locative interpretation with imperfective aspect and a directional one with perfective aspect is an effect of aspect.

2 9 8 Language Acquisition Maria Kotroni

Within novel verbs, the difference between perfective and imperfective aspect is significant (t= 7.218, df=35, p < 0.000). Therefore, the adults’ preference for locative interpretation with both perfective and imperfective aspect is an effect of aspect. Regarding the role of verb-type (novel-real) in the participant analysis, within the 1st group, there was not a significant difference between the conditions (real vs novel) in the group of verbs in perfective aspect (z=1.408, N-Ties= 32, p=0.159) as well as imperfective aspect (z=1.728, N-Ties= 31, p=0.084). Therefore, the 1st group’s preference for locative interepretation over a directional interpretation is not an effect of verb type in either aspect. Regarding the 2nd age group, in the group of verbs in perfective (z=4.874, N-Ties= 41, p=0.000) as well as imperfective aspect (z=2.371, N-Ties= 31, p=0.018), there is a significant difference between the conditions (real vs novel verbs). Therefore,nd the2 group’s preference for locative interpretation over a directional interpretation is an effect of verb type in the case of both perfective and imperfective aspect. Finally, regarding adults, within both perfective (z=3.682, N-Ties= 30, p=0.000) and imperfective verbs (z=3.301, N-Ties= 24, p=0.000), there is a significant difference between the conditions. Therefore, the adults’ preference for locative interpretation with imperfective aspect and a directional one with perfective aspect, as well as their preference for a locative interpretation over a directional one in both aspects is affected by verb type. Moving to the item analysis, a three-way mixed anova revealed significance in the main effect of aspect (F(1,33)= 5.236, p= 0.029), the aspect by group interaction (F(2,33)= 3.042, p= 0.061-marginal), the main effect of verb type (F(1,33)= 77.414, p= 0.000), the verb type by group interaction (F(2,33)= 26.903, p= 0.000), the aspect by verb type interaction (F(1,33)= 3.701, p= 0.063-marginal), as well as the main effect of age group (F(2,33)= 5.812, p= 0.007). However, the three way interaction between aspect, verb type and group was not significant. (F(2,33)= 0.702, p= 0.503). More specifically, regarding the role of aspect, within the 1st group, the difference between perfective and imperfective aspect is not significant within real verbs (t=1.287, df=11, p<0.224) and within novel verbs (t=1.055, df=11, p<0.314). Therefore, the 1st group’s preference for locative interpretation over a directional interpretation in both real and novel verbs is not an effect of aspect. Within the nd2 group, the difference between perfective and imperfective aspect is significant within real verbs (t=2.598, df=11, p<0.025), as well as within the novel verbs. Therefore, the 2nd group’s preference for a locative over a directional interpretation is an effect of aspect. Within the 3rd group, within both real (t=6.515, df=11, p<0.000) and novel verbs (t=6.082, df=11, p<0.000), the difference between perfective and imperfective aspect is significant. Therefore the adults’ preference for a locative interpretation with imperfective and a directional with perfective as well as adults’ preference for a locative over a directional interpretation is an effect of

Language Acquisition 2 9 9 Major Trends in Theoretical and Applied Linguistics

aspect. Regarding the role of verb type, within the 1st group, the difference in verb type (real vs novel verbs) is not significant within either perfective aspect verbs (t=.681, df=11, p<0.510) and imperfective aspect verbs (t=0.718, df=11, p<0.488). Therefore, the 1st group’s preference for locative over directional interpretation is not an effect of verb type in both perfective and imperfective aspect. Within the 2nd group, within perfective aspect verbs, the difference of verb type is marginally significant (t=2.060, df=11, p<0.064), whereas, within imperfective aspect verbs, the difference of verb type is not significant (t=0.997, df=11, p<0.340). Within adults, within perfective aspect verbs, the difference of verb type is significant (t=2.876, df=11, p<0.015). However, within imperfective aspect verbs, the difference of verb type is not significant (t=1.235, df=11, p<0.243). Therefore, the older children’s and adults’ preference for locative interpretation over a directional interpretation is an effect of verb type in the case of perfective aspect, but not in the case of imperfective aspect. Overall, adults seem to rely on aspect alone for the interpretation of imperfective motion verbs, whereas for the interpretation of perfective motion verbs, aspect, lexical (semantic values, saliency, frequency) and syntactic cues were also considered. In any case, the effect of aspect is shown to be significant. In the case of children, a developmental pattern is observed. Older children rely on aspect for the interpretation of ambiguous manner of motion verbs and syntactic and lexical cues are slightly related to their preference regarding perfective aspect. Younger children’s data, however, revealed that aspectual, syntactic and lexical cues are not related to their preferences. They prefer the locative interpretation because the PP is an adjunct and within children there is no preference for complement. We also checked each item separately regarding the pair of perfective- imperfective aspect within real and within novel verbs and we observed a. a developmental pattern (for example, regarding real verbs like ‘στρίβω, μπουσουλώ, κυλώ, πετώ’: 1st group: Z=0.660, N-Ties=12, p=509, 2nd group: Z=1.706, N-Ties=12, p=0.88 and adults: Z=3.655, N-Ties=1, p=0.000. Similar is the case for the rest of the verbs, real and novel) and b. that some verbs seem more resisting to the effect of aspect (for example: regarding the real verbs στρίβω,‘ μπουσουλώ, κυλώ, πετώ’ Z=0.660, N-Ties=12, p=509, whereas for the rest of the real verbs: Z=0.871, N-Ties=21, p=381)) and in the hierarchy of acceptance they were classified towards the ends of the continuum. We also checked the role of the degree of ambiguity following the hierarchy of acceptance from the pilot study (Figure 3) in both a within group and a between group analysis and we observed that the relation between motion and degree of ambiguity is significant within the nd2 group in perfective aspect (participant analysis: t=2.246, df=88, p=0.025) and within the 1st group in imperfective aspect (item analysis: t=3.500, df=5, p=0.017)). However, it was non-significant within the control group in both aspects and in all other conditions. Moreover,

3 0 0 Language Acquisition Maria Kotroni

within perfective aspect, all three age groups differ from each other. Similar is the condition of highly ambiguous imperfective verbs within imperfective aspect, except for the comparison between the younger group and adults (U=7828.5,

N1=150, N2=108, p=0.637). Regarding low ambiguity imperfective verbs, all st nd three age groups do not differ significantly (1 vs 2 group: U=12384, N1=150, st N2=174, p=0.407, 1 vs adult group: U=7735.5, N1=150, N2=108, p=0.521 and nd 2 vs adult group: U=8311.5, N1=174, N2=108, p=0.088).

High vs Low ambiguity

100

80 29,6 46,3 62,6 76 73,3 60 74,7 76,9 87 80,7 82,8 85,3 88,5 40 70,4 53,7 preference (%) preference 20 37,4 24 26,7 25,3 23,1 13 19,3 17,2 14,7 0 11,5 high low high low high low high low high low high low perfective imperfective perfective imperfective perfective imperfective 1st group 2nd group adults conditions

directional locative

Figure 3. High vs. Low ambiguity

Finally, we examined the role of frequency (Figure 4) in a within and between group analysis. The analysis revealed that in the item analysis and within all three age groups, there is no effect of frequency in both aspects. A significant effect is observed in the participant analysis of imperfective aspect in all age groups (1st group: Z=2.086, N-Ties=26, p=0.037, 2nd group: Z=3.799, N-Ties=37, p=0.000 and adults: Z=2.200, N-Ties=28, p=0.028) and of perfective aspect within older children (Z=3.849, N-Ties=42, p=0.000). Regarding the between group analysis, within perfective aspect, there is no relationship between motion and frequency, except for the children groups within highly frequent perfective

aspect verb forms (U=1449.5, N1=50, N2=58, p=0.997). On the other hand, in all age groups there is no effect of frequency regarding imperfective aspect except for the older children and adults within highly frequent imperfective aspect verb

forms (U=792.5, N1=58, N2=36, p=0.008).

Language Acquisition 3 0 1 Major Trends in Theoretical and Applied Linguistics

High vs Low Frequency

50

40 16,7 21,3 28,4 38

30 37,7 38,8 39,3 39,4 40,2 42,7 42,6 46,8 20 preference 33,3 10 28,7 21,6 12 12,3 9,8 11,2 10,7 10,6 7,3 7,4

0 3,2 high low high low high low high low high low high low freq freq freq freq freq freq freq freq freq freq freq freq perfective imperfective perfective imperfective perfective imperfective 1st group 2nd group adults conditions

directional locative Figure 4. High vs. Low frequency

4.2. Telic vs. Atelic Task

The aim of this task was to investigate the role of the aspectual verb form in the comprehension of ambiguous manner-of-motion verbs, i.e do native speakers of Greek use the perfective/ imperfective distinction to differentiate between telic and atelic readings? The variables tested appear in Table 9.

Table 9. Dependent and Independent variables

1. telic (endpoint reached) Dependent variable a. Telicity 2. atelic (locative)

1. perfective a. Aspect 2. imperfective

1. 5;7 (age range 1;6)

Independent variables b. Age group 2. 9;6 (age range 1;11)

3. 20;4 (age range 11;0)

1. real verbs c. Verb type 2. novel verbs

3 0 2 Language Acquisition Maria Kotroni

Within real verbs (Figure 5), a basic pattern is observed. All three age groups prefer the atelic locative interpretation with imperfective aspect. With perfective aspect however adults prefer the telic interpretation, although children still prefer the locative one. However, within older children, the gap between telic and atelic is narrowing.

Real verbs 100 21 80 55 75 74 60 87 85 40 79 45

preference (%) preference 20 25 26 13 15 0 perfective perfective perfective imperfective imperfective imperfective 1st group 2nd group adults age groups telic (endpoint reached) atelic (locative)

Figure 5. Descriptive statistics- Real verbs

Regarding novel verbs (Figure 6), a locative interpretation outnumbers the telic one within imperfective aspect for all age groups and within perfective aspect within children. However, adults assign a telic interpretation to perfective motion verbs. In the inferential participant analysis (with the use of a three-way mixed anova) significance was observed for the main effect of verb type (F(1,125)=15,319, p=0.000), the verb type by age group interaction (F(2,125)=12,007, p=0.000), the main effect of aspect (F(1,125)= 96,524, p=0.000), the aspect by age group interaction (F(2,125)= 34,080, p=0.000), the verb type by aspect interaction (F(1,125)= 12,051, p=0.001), the main effect of age (F(2,125)= 9,651, p=0.000) as well as the three-way interaction between aspect, verb type and age group (F(2,125)= 3,324, p=0.039). Regarding the role of aspect, within the 1st group, the difference between perfective and imperfective aspect is not significant within real verbs (t=0.405,

Language Acquisition 3 0 3 Major Trends in Theoretical and Applied Linguistics

Novel verbs 100

80 38

60 71 70 70 91 90 40 62

preference (%) preference 20 29 30 30 0 9 10 perfective perfective perfective imperfective imperfective imperfective 1st group 2nd group adults age groups telic (endpoint reached) atelic (locative)

Figure 6. Descriptive statistics-Novel verbs

df=39, p<0.687) as well as within novel verbs (t=0.327, df=39, p<0.745). Therefore, the 1st age group’s preference for an atelic over a telic interpretation is not an effect of aspect. However, within the 2nd group, within the real verbs, the difference between perfective and imperfective aspect is significant within both real (t=5.661, df=47, p<0.000) and novel verbs (t=4.515, df=47, p<0.000). Similar is the case with the 3rd age group, real verbs: t=10.115, df=39, p<0.000 and novel verbs: t=6.846, df=39, p<0.000. Therefore, the 2nd and 3rd age group’s preference for an atelic over a telic interpretation is an effect of aspect. Regarding the role of verb type, within the 1st group, the difference of verb type was not significant within perfective verbs (t=1.049, df=39, p<0.301) as well as within imperfective (t=1.389, df=39, p<0.173). Therefore, the 1st group’s preference for an atelic over a telic interpretation is not an effect of verb type. The 2nd group’s preference for an atelic over a telic interpretation is an effect of verb type in the perfective aspect (t=4.103, df=47, p<0.000) but not in the imperfective (t=1.319, df=47, p<0.096, one-tailed). Similar is the case with the 3rd group: perfective verbs (t=4.594, df=39, p<0.000) and imperfective verbs (t=1.704, df=39, p<0.096). Therefore, the 3rd group’s preference for an atelic over a telic interpretation in imperfective aspect is not an effect of verb type, but their preference for a telic over an atelic interpretation within perfective aspect is an effect of verb type.

3 0 4 Language Acquisition Maria Kotroni

In an item analysis, the three way mixed anova verified a signigicance in the main effect of verb type (F(1,33)=4.162, p=0.049), the verb type by age group interaction (F(2,33)=3.076, p=0.060-marginal), the main effect of aspect (F(1,33)=208.232, p=0.000), the aspect by age group interaction (F(2,33)=78.085, p=0.000). However, the verb type by aspect interaction (F(1,33)=2.681, p=0.111), as well as the three way interaction between verb type, aspect and age group were not significant (F(2,33)=0.916, p=0.410). Regarding the role of aspect within the 1st group, the difference between perfective and imperfective aspect was not significant within real verbs (t=0.411, df=11, p<0.689) as well as within novel verbs (t=0.114, df=11, p<0.912). Therefore, the 1st group’s preference for an atelic over a telic interpretation is not an effect of aspect. Within the 2nd group, the difference between perfective and imperfective aspect was significant within real verbs (t=11.851, df=11, p<0.000) as well as within novel verbs (t=5.803, df=11, p<0.000). Similar is the case within the 3rd group: real verbs real verbs (t=11.525, df=11, p<0.000) and novel verbs (t=5.922, df=11, p<0.000). Therefore, the 2nd and 3rd group’s preference for an atelic over a telic interpretation within novel verbs and the latter’s preference for telic interpretation within real verbs is an effect of aspect. Regarding the role of verb type, within the 1st group, the difference between real and novel verb type was not significant within both perfective (t=1.157, df=11, p<0.272) and imperfective verbs (t=0.549, df=11, p<0.594), which is also the case with the 3rd group: perfective verbs, t=1.563, df=11, p<0.146 and imperfective verbs, t=1.593, df=11, p<0.139. For the 2nd group, within perfective aspect verbs, the difference between real and novel verb type was significant (t=2.602, df=11, p<0.025), whereas within imperfective verbs, the difference between real and novel verb type was not significant (t=1.093, df=11,p<0.298 ). The results show that older children rely on aspect for the interpretation of imperfective as well of perfective motion verbs. Younger children and adults do not seem to rely on aspect or semantic cues for the interpretation of motion

verbs. Children avoid the use of NPsobject, so the telic condition, where the PP is a complement, is not preferred irrespectively of aspect or verb type. In the analysis of each item separately, a developmental pattern is also verified similar to the previous task. Regarding the role of frequency and ambiguity, within all age groups, there was no relationship between telicity and degree of frequency as well as telicity and degree of ambiguity.

4.3. Between-Task Analysis

If we attempted to compare conditions from different tasks, this might cause task effects in the analysis. Therefore, in an attempt to make a between task analysis, we compared these two conditions through comparisons with the atelic locative

Language Acquisition 3 0 5 Major Trends in Theoretical and Applied Linguistics

condition. Such an analysis is only available within perfective aspect, because the telic condition is incompatible with imperfective aspect. Regarding real verbs (Table 10), within older children and adults, telic interpretations outnumber directional ones. The telic condition is more transparent than the directional one.

Table 10. Between task analysis (real verbs)

Real verbs

Participant Item anal anal Age groups Conditions p p Directional vs sig Locative>directional 1st group .000 .000 locative sig Locative>telic => Telic vs locative .000 .000 non-sig Telic>directional Directional vs Sig Locative>directional 2nd group .022 .000 locative non-sig.Locative>telic => Telic vs locative .171 .158 Sig Telic>directional Directional vs non-sig Directional>locative Control group .178 .001 locative Sig telic>locative => Telic vs locative .000 .001 Sig telic>directional

However, within novel verbs (Table 11), such an observation is not significant, since novel verbs are restricted from any syntactic or lexical cue.

Table 11. Between task analysis-novel verbs

Novel verbs

Participant Item analysis analysis Age groups Conditions p p Directional vs sig locative>directional 1st group .000 .000 locative sig locative>telic => Telic vs locative .000 .000 n.sig telic>directional Directional vs sig locative>directional 2nd group .000 .000 locative sig locative>telic => Telic vs locative .000 .000 n.sig telic>directional Directional vs n.sig.locativedirectional => Telic vs locative .155 .486 n.sig telic>directional

3 0 6 Language Acquisition Maria Kotroni

5. Discussion and Conclusions

Based on the results analysed in the previous parts, the prediction that the interpretation of imperfective as atelic is more preferred than the interpretation of perfective as telic and the prediction that the interpretation of imperfective as locative is more preferred than the interpretation of perfective as directional are verified. Moreover, a developmental pattern in the interpretation of manner of motion verbs (late acquisition) is also observed, which in turn led to the assumption that except for aspect, there are probably other factors that interact with the interpretation of real verbs: lexical cues and morphosyntactic information. These observations lead to a more general conclusion that the interpretation of verb forms and the attribution of their aspectual values involves the integration of various features in a compositional nature. This interaction involves a. morphological aspect (i.e. perfective and imperfective), b. motion (locative-directional and telic-atelic conditions) and c. structural information (PP as a complement or as an adjunct). This observation that makes morphosyntax, semantics, pragmatics and syntax integrate with a compositional effect, gives an account of Lexicon-Morphosyntax Interface (Johson-Laird 1983). The acquisition of such an interaction (not of each part separately) seems to be a prerequisite for an adult like performance and results in late acquisition and in the developmental pattern observed. Finally, we assume that late acquisition does not reveal insufficient processing resources due to processing load, but insufficient knowledge of such an interface even at the age of 10 forGreek children, which in turn fails to establish proper discourse relations. However, further research on language interfaces is required so that we arrive at more robust conclusions and a widely accepted approach.

Language Acquisition 3 0 7 Major Trends in Theoretical and Applied Linguistics

References

Bates, E. and Goodman, J. 1997. On the inseparability of grammar and the lexicon: Evidence from acquisition, aphasia, and real-time processing. Language and Cognitive Processes 12, 507-584.

Fisher, C., Hall, D.G., Rakowitz, S. and Gleitman, L.R. 1994. When it is better to receive than to give: Syntactic and conceptual constraints on vocabulary growth. Lingua 92, 333-375.

Gleitman, L. 1990. The structural sources of verb meanings. Language Acquisition 1, 3-55.

Hollebrandse, B. and van Hout, A. 2001. On the acquisition of the aspects in Italian. In J.-Y. Kim and A. Werle (eds.) The proceedings of SULA. The semantics of underrepresented Languages in the Americas. UMOP 25, 111-120. GLSA, Amherst.

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

Horrocks, G. and Stavrou, M. 2003. Actions and their results in Greek and English: The complementarity of morphologically encoded (viewpoint) aspect and syntactic resultative predication. Journal of Semantics 20, 297-327.

Horrocks, G. and Stavrou, M. 2007. Grammaticalized aspect and spatio-temporal culmination. Lingua 117, 605-644.

Johnson-Laird, P. 1983. Mental Models: towards a cognitive science of language, inference and consiousness. New York: Cambridge University Press.

Kazanina, N. and Philips, C. 2002. Russian children’s comprehension of aspectual distinctions. Talk presented at LSA. www.ling.umd.edu/ninaka

3 0 8 Language Acquisition Maria Kotroni

Klairis, C. and Babiniotis, G. 2004. Grammatiki tis neas Ellinikis, [Modern Greek Grammar]. Ellinika Grammata.

Landau, B. and Gleitman, L. 1985. Language and experience: Evidence from the blind child. Cambridge, MA: Harvard University Press.

MacWhinney, B. (ed.) 1999. The Emergence of Language. Mahwah, NJ: Erlbaum.

Marchman, V. and Bates, E. 1994. Continuity in lexical and morphological development: a test of the critical mass hypothesis. Journal of Child Language 21, 339-366.

Naigles L.R. 1990. Children use syntax to learn verb meanings. Journal of Child Language 17, 357-374.

Naigles, L.R. and Terrazas, P. 1998. Motion-verb generalizations in English and Spanish: Influences of language and syntax. Psychological Science 9, 363-369.

Papadopoulou, D. 1996. Ρήματα κίνησης στην ελληνική: σημασιοσυντακτική προσέγγιση [Motion verbs in Greek: a syntactic and semantic approach]. MA thesis, University of Athens.

Papafragou, A. and Selimis, S. 2010. Event categorisation and language: a cross- linguistic study of motion. Language and Cognitive Processes 25,2.

Ralli, A. 1988. Elements de la morphologie du Grec Moderne: La structure du verbe. Unpublished PhD Dissertation, Universite de Montreal, Montreal, Canada.

Stoll, S. 1998. Acquisition of Russian aspect. First Language 18, 351-377.

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

Triantafyllidis, M. 2000. Neoelliniki Grammatiki. Modern Greek Studies Institute. Aristotle University of Thessaloniki.

Tsimpli, I.M. & Papadopoulou, D. 2009. Aspect and the Interpretation of Motion verbs in L2 Greek. In Snape, Neal, Yan-kint Ingrid Leung and M. Sharwood Smith (eds.), Representational Deficits in SLA. 187-227.

van Hout, A. 2002. Imperfect imperfectives. Paper presented at the Workshop on Syntax, Semantics and Acquisition of Aspect. University of Iowa, May 24-26, 2002.

Language Acquisition 3 0 9 Major Trends in Theoretical and Applied Linguistics

Vinnitskaya, I. and Wexler, K. 2001. The role of pragmatics in the development of Russian aspect. First Language 21, 143-186.

Weist, R., Wysocka, H. and Lyytinen, P. 1991. A cross-linguistic perspective on the development of temporal systems. Journal of Child Language 18, 67-92.

3 1 0 Language Acquisition Maria Kotroni

Appendix

Imperf- Imperf- item Perf-se S.d. Perf-ja S.d. S.d. S.d. se ja

1 τρέχω 1,88 1,054 4,00 1,225 1,50 ,855 4,43 1,284

2 περπατώ 1,85 1,345 3,64 1,598 2,07 1,592 3,50 1,618

3 οδηγώ 1,79 1,188 3,50 1,286 1,28 ,575 4,08 1,165

4 σκαρφαλώνω 4,07 1,385 2,72 1,447 3,08 1,656 3,29 1,858

5 τρικλίζω 1,39 ,850 1,69 ,855 1,29 ,611 4,00 1,359

6 κυλιέμαι 1,50 1,000 3,07 1,639 2,00 1,177 3,06 1,349

7 γλιστράω 1,93 ,997 2,64 1,598 1,56 ,856 1,92 1,038

8 τσουλάω 1,71 1,204 1,89 1,023 1,46 1,127 2,36 1,336

9 στρίβω 2,83 1,581 1,46 1,198 1,64 1,008 1,64 1,008

10 μετακινούμαι 2,15 1,519 3,43 1,604 1,43 ,938 3,39 1,539

11 ταξιδεύω 2,79 1,424 1,64 1,216 2,38 1,668 1,38 ,870

12 πετώ 1,31 ,630 3,21 1,888 1,79 1,424 4,18 1,334

13 υπνοβατώ 1,57 1,089 4,41 ,939 1,08 ,277 4,50 1,019

14 βηματίζω 1,61 ,916 3,92 1,441 1,43 ,646 3,29 1,858

15 καλπάζω 1,31 ,855 4,00 1,359 1,50 1,160 4,17 1,383

16 προχωρώ 1,79 ,802 2,79 1,528 1,72 1,127 4,00 1,414

17 πεζοπορώ 2,21 1,424 2,94 1,305 1,46 ,967 4,14 1,231

18 κολυμπώ 2,00 1,177 3,50 1,653 1,33 ,594 3,54 1,561

19 κυλώ 1,86 1,292 3,06 1,514 1,15 ,376 4,21 ,975

20 βαδίζω 2,11 1,023 4,38 ,768 1,64 1,008 4,21 1,477

21 σέρνομαι 2,62 1,502 4,57 ,514 1,93 1,439 4,47 ,717

22 μπουσουλώ 1,79 ,975 4,14 1,460 1,94 1,056 4,08 1,441

23 πηδώ 2,14 1,610 2,39 1,539 1,46 ,776 4,21 1,251

24 πλέω 2,06 1,349 2,92 1,553 1,57 ,938 4,29 1,437

Language Acquisition 3 1 1 Major Trends in Theoretical and Applied Linguistics

3 1 2 Language Acquisition Advanced Learner Competence and Relative Clauses in Oral Narratives

Monique Monville-Burston and Maria Kounouni

University of Cyprus, Cyprus University of Technology [email protected], [email protected]

Abstract

This article examines the use of relative clauses in the oral narratives of advanced Cypriot students of French. The data were collected through story- retelling techniques. Seven native speakers of French and nine non-native Cypriot learners participated in the study. Similarities and differences in the narrative discourse of the two groups were identified. It is shown that two main features characterise narrative competence at advanced level: the use of il y a...qui (‘there is someone…who/something…which’) presentative constructions and of autonomous relatives. Possible cross-linguistic influence from Greek narrative style on the French oral productions of non-native speakers is also considered.

1. Introduction

The work presented here is part of a larger project on the acquisition of relativisation by advanced native Greek-speaking learners of French. In previous studies we dealt with the acquisition of relative clauses, of their structure and introductory terms, in students’ written productions (Monville-Burston 2008; Monville-Burston & Kounouni 2008; Monville-Burston & Kakoyianni-Doa 2009; Monville-Burston & Kounouni 2010). In this article we concentrate on the advanced learners’ use of relative clauses in discourse: what we examine is the nature and functioning of relativised sentences in oral narratives, on the basis of the bilingual ‘parallel’ corpus collected by Kounouni (Kounouni 2010). Relativisation, as a grammatical/discursive feature, has been singled out by Bartning and her students/collaborators (Bartning & Kirchmeyer 2003; Bartning & Schlyter 2004; Hancock 2002; Hancock & Kirchmeyer 2002; Kirchmeyer 2002) as a marker of advanced non-native speaker (NNS) speech vs. native speaker (NS) speech in French and can thus be regarded as one of the indices of naturalness and fluency in advanced learners-users’ speech.

Language Acquisition 3 1 3 Major Trends in Theoretical and Applied Linguistics

The following research questions were addressed: 1) What is the comparative frequency of relative clauses [RCs] in the productions of native speakers’ [NS] and non-native speakers’ [NNS] speech? 2) Which relative pronouns [RPs] (and which relativised grammatical functions) are used the most by the two groups? 3) What are the informational functions of the RCs used by both groups: identification-referential [restrictives] or predicative-attributive [non- restrictives]? 4) Where the differences are important between NS and NNS, can cross- linguistic influence be invoked as a possible reason?

First we will describe the study design, we will then give an account of the results obtained, contrasting NS and NNS speech, and in relation to the investigated questions, we will discuss the results, paying particular attention to qui (subject-RP) RCs. Finally in the light of the narrative discourse of Cypriot Greeks we will try to evaluate possible Greek>French interferences in NNS speech.

2. Methodology and Participants

The corpus data (gathered in 2010) in the present study come from sixteen subjects, seven native speakers of French (Erasmus exchange students, 3 males, 4 females) and nine Fourth Year and Masters’ Cypriot French students (9 females), all at the University of Cyprus. The oral narratives were elicited with the help of three short films shown to the participants on a computer screen, during individual interviews. The films were “silent” cartoons by Mordillo and lasted between 50 and 70 seconds1. Participants were then asked to retell the story they had just watched, giving as many details as possible and speaking directly into the computer microphone. Thus, each participant produced three oral stories which gave a total of forty-eight texts. Altogether the data for NS included 3,230 words and for NNS 3,290. In 2011 additional recordings were made of eight Cypriot subjects (3 males, 6 females), of the same age group and level of education as the first two groups. They were asked to retell the same three stories in Greek (2,877 words).

1 Some of the elicitation material can be consulted on http://www.youtube.com/watch?v= Y3p5ay1WchE (First aid). Unfortunately the other two videos (At the circus; The child’s friend) have been withdrawn from the Web. A summary of the three stories is given in the appendix.

3 1 4 Language Acquisition Monique Monville-Burston and Maria Kounouni

The recorded performances were then transcribed, with an indication of features like syllable or pause length, where relevant. These details are important for the identification of autonomous RCs. The interviewer intervened the least possible, for example to help a NNS in difficulty because of a missing vocabulary item or to ask a participant to include an episode (s)he had overlooked. The NNS group was approximately at Level B2 on the Common European Framework of Reference scale.

3. Frequency of Use of Relative Clauses

Regarding the first research question, the following results were obtained (Table 1). In this table, all categories of RCs (see infra, Section 4) are grouped together.

Table 1. Number of relative clauses used by NS and NNS

L2=French Code L1=French Code L1=Greek NS1 12 NNS1 7

NS2 4 NNS2 9

NS3 11 NNS3 9

NS4 11 NNS4 8

NS5 10 NNS5 3

NS6 16 NNS6 4

NS7 6 NNS7 6

NNS8 5

NNS9 10

Total 70 Total 61

Average 10 Average 6.7

We see that, on average, NNS use a smaller number of RCs than the NS. Actually, on the whole, NS use more subordinate clauses (including RCs) than NNS. This is not surprising in the light of research done on syntactic complexity in language acquisition. Studies on subordination in interlanguages draw attention to the “grammaticalization” of the language of learners (e.g. Klein & Perdue 1992, 1997; Chini 1998; Giacalone Ramat 2000). Based on Givon’s functional- typological (1984/2001) or on Lehmann’s typological model (1988), such studies hypothesize and show that in second language acquisition learners go through a context-bound pragmatic mode of expression which is predominantly paratactic, with minimal integration, towards modes of expression where subordination is increasingly present and is the sign of a stronger syntactic integration.

Language Acquisition 3 1 5 Major Trends in Theoretical and Applied Linguistics

4. Frequency of Use of Particular RPs Grammatical Functions

In order to answer our second research question, we considered the RCs produced by the participants from a morphosyntactic point of view: we examined the RPs used and their grammatical functions in the embedded RC: qui (‘who, which’) for subject, que (‘whom, which’) for direct object, où (‘where/ when, in which’) for locational or temporal relations, the inflected form lequel for prepositional phrases (e.g. ‘with whom/with which’) (see Tables 2 and 3). The occurrences of ce qui (subject) and ce que (object), introducing RCs without NP antecedents (equivalents of which), have been combined for simplification, although ce qui is more frequent then ce que. Dont (‘of which, whose’) is absent from the data. Table 2 (NS) and Table 3 (NNS) show the distribution of RPs with their functions. The oral narratives for both groups exhibit a very high frequency of qui pronouns and there does not seem to be much difference between the two groups. Aside from confirming the Keenan-Comrie (1977) hierarchy and the high accessibility of the subject to relativisation across languages, these results support other specific studies for spoken French, such as Gadet (2003) or Morel & Danon-Boileau (1998).

Table 2. Relative pronouns and their grammatical functions (NS)

qui que. direct où locative/ lequel in ce qui/ Code Total subject object temporal PP que

NS1 10 - 1 - 1 12

NS2 4 - - - - 4

NS3 6 1 - 1 3 11

NS4 8 - 1 - 2 11

NS5 10 - - - - 10

NS6 13 - 2 - 1 16

NS7 6 - - - - 6

Total 57 1 4 1 7 70

% 81.5% 1.4% 5.7% 1.4% 10% 100%

3 1 6 Language Acquisition Monique Monville-Burston and Maria Kounouni

Table 3. Relative pronouns and their grammatical functions (NNS)

qui que. direct où locative/ lequel in ce qui/ Code Total subject object temporal PP que

NNS1 5 - - 2 - 7

NNS2 8 - 1 - - 9

NNS3 8 - - - 1 9

NNS4 4 4 - - - 8

NNS5 3 - - - - 3

NNS6 3 1 - - - 4

NNS7 4 - 2 - - 6

NNS8 4 - 1 - - 5

NNS9 9 - - 1 - 10

Total 48 5 4 3 1 61

% 78.7% 8.2% 6.5% 5% 1.6% 100%

5. Types of qui Relative Clauses

Leaving aside the low frequency items (columns 3, 4, 5, 6), we now concentrate on the qui RCs found in the data. As it is well-known, RCs, although identical in form, may not function in the same way within a sentence. In our classification we adopt the traditional opposition between restrictive and non-restrictive (appositive) RC.2 We use de Vries (2006) and Morel & Danon-Boileau (1998) descriptions of the behavior of non-restrictive RCs3 and create within the latter category subcategories that are adapted to the type of texts at hand (retelling oral narratives). All the examples are taken from our corpus.

2 We refer the reader to Kleiber (1987) for instance, for a discussion of the difficulties raised by this two-way classification and various proposals put forward to remedy them. But we consider that this distinction has enough validity for the purposes of this article. 3 Non-restrictive RCs (appositive) according to de Vries have the following properties, among others: they specify the whole antecedent, they can be deleted without a significant shift in meaning or loss of acceptability, they can follow other complements of the antecedent, they can be stacked asyndetically, they can be extraposed from their antecedent NP, they can have antecedents of any category (De Vries 2006: 21-32). Morel & Danon-Boileau (1998: 53-58), referring only to spoken French use intonational patterns to classify relatives and their rhematic or non-rhematic function. So there are relatives that are integrated into the antecedent NP (restrictive or non-restrictive), while others are dissociated from their antecedent NP: they can be presentative, parenthetical or autonomous.

Language Acquisition 3 1 7 Major Trends in Theoretical and Applied Linguistics

5.1. Restrictive Relative Clauses

Since their property is to restrict the number of potential referents of an NP, they have an identification-referential function.

(1) l’homme qui était sur la terre [NNS2] ‘the man who was lying on the ground’ (as opposed to the others)

5.2. Non-Restrictive Relative Clauses

Non-restrictive RCs add a comment or a predication concerning the referent of the antecedent NP, without modifying its extension (see also footnote 3). In our data, there are three main types of qui non-restrictive relatives.

5.2.1. The Canonical Variety, Often Called ‘Explicatives’ (or ‘Appositives’) in French Grammars

The RC plays no role in the identification of the referent. It provides accessory information, a complementary description for instance.

(2) le magicien porte un chapeau qui est un peu bizarre. [NNS2] ‘the magician wears a hat which is a bit strange.’

5.2.2. Qui Clauses with a Predicative-Attributive Function: So- Called “Presentative” Sequences (c’est…qui, il y a…qui)

It should be noted (see examples below) that French presentative RCs are often not translated with RCs in other languages. Presentative structures are discontinuous: “Presentative word + NP + qui” where the first part serves as a support for a predication on the referent (represented by the NP), the predication being performed by a RC. For example:

(3) [Incipit] Alors, c’est un monsieur qui est dans une voiture et le pneu de la voiture éclate. [NS1] (So, it is a man who is in a car and the tire of the car blows out) ‘So a man is in a car and the tire blows out.’

(4) Il y a une ambulance qui arrive. [NS2] (There is an ambulance which/that is arriving) ‘An ambulance arrives.’

3 1 8 Language Acquisition Monique Monville-Burston and Maria Kounouni

Semantically, presentation is an operation which determines the mode of existence (or the identity or the presence) of an object or a process. This existence is always more or less tightly linked to a localisation in place or time. The referent (see Lambrecht 1988) is cognitively non-accessible, non- identifiable before being introduced into the structure. Once introduced itis promoted to a high degree of topicality, which makes it a potential topic/theme for the RC and other clauses that follow. Two types of presentative structures (illustrated in (3) and (4)) are traditionally recognized in French, the c’est…qui (‘it is…who/which’) type and the il y a…qui (‘there is…who/which’) type. In c’est…qui, ce is by nature deictic-anaphoric. Since the referent of the NP to be presented is not accessible, it cannot be textually anaphoric. It then functions as an anaphoric which presupposes the existence of the object/ person in a situation anterior to the present one. This is the case, in particular, at the very beginning of a text (as in the oral narratives considered here, see (3)). C’est can also be given a deictic reading: it is as if the referent existed under the eyes of the speaker (which is the case after viewing a video clip). Being deictic, c’est presupposes the presence of a speaker at the origin of the act of referencing. Rather than asserting existence (as c’est), il y a posits the referent as an object/person which is going to play a role in the speech situation. It presents the referent in a more abstract, objective way than c’est; the speaker seems to be less involved. While il y a introduces the new topic, its RC brings in new information which normally takes the form of an event related to the topic (see (4), the arrival of the ambulance) [for a discussion of these matters, see Lambrecht 1988].

5.2.3. Qui Clauses with a Predicative-Attributive Function, Where the Antecedent is Governed by a Verb of Perception

These RCs are always introduced by the subject RP qui.

(5) On entend les gens qui applaudissent. [NS2, NS7, NNS5] (We hear the people who applaud) ‘We hear people applauding/cheering.’

They are clearly predicative: Sentence (5) can be glossed as “I hear the people and they are cheering.” Perception RCs always involve events. Since they depict perception in action, they are considered to be deictic (see Cadiot 1976; Furukawa 2005).

Language Acquisition 3 1 9 Major Trends in Theoretical and Applied Linguistics

5.2.4 Qui Clauses with a Predicative-Attributive Function: the So-Called “Autonomous” or “Narrative” Relative Clause

Sentences (6) and (7) are examples of such relatives:

(6) il les [=les ballons] donne à l’enfant // qui cette fois-ci accepte. [NS2] ‘He gives them to the child, who this time accepts. [// indicates a pause]

(7) Là-dessus, il voit un marchand de ballons remplis de gaz là //qui volent d’hélium //qui marche. [NS3] ‘Thereupon, he sees a man selling balloons, filled with gas, there, that fly, of helium… who is walking.’

Creissels (2004) notes that, with this type of construction, one refers to an event chronologically linked to a preceding event which is talked about in the phrasal unit to which the antecedent belongs. In other words, what is denoted is the succession of two events (See (6)). The RP behaves more like a coordinating element than a subordinating one: ‘he gives them to the child… who this time accepts them (= and this time he accepts them)’. Such RCs are rather independent from their antecedents, hence the term “autonomous” used by Morel & Danon-Boileau and others to designate them. Morel & Danon-Boileau (1998) and Kirchmeyer (2002) give a list of features which characterise this type of relatives. These can be summarized as follows: autonomous/narrative relatives are similar to independent utterances; they can be paraphrased with juxtaposed or coordinated constructions; they are often separated from their antecedent by several words or phrases; they are also marked by specific intonational contours and preceded by pauses; qui can be repeated and its final vowel elongated. (See example (7) as an illustration of these features.)

6. Comparing the Use of Qui Relatives in NS and NNS Oral Narratives

With the discussion in Section 5 in mind, we can now compare the uses of qui RCs in NS and NNS speech. Table 4 (NS) and Table 5 (NNS) give detailed information on the results, while Table 6 summarizes the main findings in a comparative fashion.

3 2 0 Language Acquisition Monique Monville-Burston and Maria Kounouni

Table 4. Types of qui relatives used by NS

Code Non-restrictives Restrictives with perception Canonical C’est…qui Il y a…qui Autonomous verbs NS1 - 2 4 2 1 1

NS2 - 1 2 - 1 -

NS3 - - - 1 4 1

NS4 1 4 1 - 1 1

NS5 3 - 3 - 3 1

NS6 1 - 3 4 3 2

NS7 - 2 1 1 2 -

Total 5 9 14 8 15 6 % out of 8.8% 15.9% 24.5% 14% 26.3% 10.5% 57 items Restrictives TOTAL Non-restrictives 51=89.5% 6 = 10.5%

Table 5. Types of qui relatives used by NNS

Code Non-restrictives Restrictives with perception Canonical C’est…qui Il y a…qui Autonomous verbs NNS1 - 4 - - 1 -

NNS2 3 3 - - 1 1

NNS3 5 2 - - 1 -

NNS4 1 1 1 1 - -

NNS5 - 2 - 1 - -

NNS6 - 1 - - - 2

NNS7 2 - 1 1 - -

NNS8 - - 1 2 - 1

NNS9 - - 2 6 1 -

Total 11 13 5 11 4 4 % out of 23% 27% 10.4% 23% 8.3% 8.3% 48 items Restrictives TOTAL Non-restrictives 44=91.7% 4= 8.3%

Language Acquisition 3 2 1 Major Trends in Theoretical and Applied Linguistics

Table 6. Frequency of types of qui relatives – A comparison

NS NNS

Canonical 8.8% 23%

C’est...qui 15.9% 27%

Non-restrictives Il y a...qui 24.5% 10.4%

with perception verbs 14% 23%

Autonomous 26.3% 8.3%

Restrictives 10.5% 8.3%

7. Use of Qui Relatives by NS and NNS: Discussion

We observe first that non-restrictivequi RCs are far more frequent than restrictive RCs, in the narratives of both groups. The predicative function clearly dominates over the identificational function. This may be explained by the fact that the stories to be told by the participants did not necessitate precise identification of the protagonists, objects and places depicted, since interviewee and interviewer had both seen the cartoons and thus knew what the stories were about. The ratio might have been different with other types of oral genres, personal narratives for example. Secondly the results indicate that both groups use the two types of presentational RCs (c’est...qui and il y a...qui) and the perception RCs, though not with the same frequency. NNS seem to prefer the c’est...qui and the perception relatives, and to not favor the il y a...qui relatives. A possible explanation could be, on the one hand, that French perception RCs have RC counterparts in Greek (see below, Section 8). As for the c’est...qui sequences, they are more frequent in French than the il y a...qui sequences, since they are widely used in contexts other than the presentational ones (for example: for identification, clefting, focussing (see Clech-Darbon et al., 1999: 84-88). NNS actually sometimes use c’est…qui presentational constructions where il y a…qui would be preferred by a NS. Thirdly, the data indicate that NNS shy away from the use of narrative RCs. Kirchmeyer in her study of Swedish advanced learners (2002) obtained similar results. NNS appear to use relatives more confidently at the local level of the sentence, as in (2) for example (le magicien porte un chapeau qui est un peu bizarre, ‘the magician is wearing a hat which is a little strange’) where the canonical non- restrictive relative is better integrated into the NP. On the contrary, a sentence like (7) [produced by a NS], where the RC escapes the scope of the NP requires boldness and dexterity, which NNS, although advanced, do not seem to have yet.

3 2 2 Language Acquisition Monique Monville-Burston and Maria Kounouni

8. Greek>French Cross-Linguistic Influence?

We hypothesised that the learners’ native language and its discourse structures could have some influence on their oral narrations in the target language. To ascertain this, the narratives (in Greek) of the eight Cypriots who participated in the second part of the experiment were examined with respect to relativisation.4 The most significant results are the following: i) Each Cypriot speaker [CS] used an average of 7,9 RCs, as compared to the French NS group (10) and the NNS group (6.7). The learners’ speech in French is therefore less complex (see section 3) than it would probably be in their own language. ii) The relativiser5 was overwhelmingly found in subject function (73% of occurrences), a figure comparable to those for the NS and NNS groups (81.5% and 78.7%, respectively - See Table 1). iii) However, examination of the discursive functions of the RCs with subject relativisers revealed important differences between French (NS), interlanguage French (NNS) and Greek (CS). Table 7 presents the results which will be commented upon. Within the limits of this article, we will concentrate on three points: the presentative structures, the perception verbs and the autonomous RCs.

Table 7. Types of RCs with a subject relativiser, used by Cypriots (in Greek)

Code Non- restrictives Restrictives Iparxi… Ine… pou with with pou (‘there Canonical (‘it is… who appearing percept. Autonomous is… who / /which’) verbs verbs which’) CS1 2 - - 1 - - 2 CS2 - 2 - 1 - - 1 CS3 5 - - 1 1 - - CS4 - - - 1 - 1 2 CS5 1 - 1 - 1 - 2 CS6 2 - - 1 3 4 6 CS7 - - - - 2 - 1 CS8 1 - - - 1 - - Total: 11 2 1 5 8 5 14 46 % out 24% of 46 4.5% 2% 10.75% 17.5% 10.75% 30.5% Restrictives Non-restrictives 32 = 69.5% 14 = 30.5%

4 The recordings revealed a number of interesting differences in French NS and Greek NS narratives, which would need to be studied further. 5 We choose the term ‘relativiser’ here to cover both pou (relativiser-complementizer) and o opios (relative pronoun) found in participants’ narratives.

Language Acquisition 3 2 3 Major Trends in Theoretical and Applied Linguistics

1) There are equivalents in Greek for c’est...qui and il y a...qui (‘ine…pou/o opios’ and ‘iparxi…pou/o opios’ respectively), but they are underused by CS as Table 7 shows, in comparison with another type of presentational structure, illustrated in (8).

(8) Emfanizete sto parko kapios o opios poula balonia. [CS3] ‘Appears in the park someone who sells balloons.’

These verb-first sentences which feature a small set of intransitive “appearing” verbs (emfanizome, ‘appear’; erxome ‘come’; parousiazome, ‘present oneself’) often announce unpredictable information (Holton et al. 1997: 428). They are used to avoid having in initial position of the sentence (as theme) an indefinite NP, i.e. an NP whose referent is unidentifiable by the hearer (see Lambrecht 1988: 147-149). Such constructions are rare in French and typical of the formal language. For these reasons, the Greek “appearing” RCs do not interfere in the oral French of advanced learners, who have integrated in their interlanguage the French presentational structures.

2) We saw above (Tables 4 and 5) that, compared to NS, NNS used more perception verb RCs than NS (23% vs 14%). This seemingly important difference may be due to cross-linguistic influence. Table 8 shows that perception verb RCs represent 17.5 % of the RCs with a “subject” relativiser in CS data. In CS’ narratives, vlepo (‘see’) in the first person, singular or plural [and more marginally akouo ‘to hear’] is frequent at the incipit of a narrative, or within the body of the retold story, to introduce a new protagonist. Actually, the presence of presentational vlepo in the CS narratives is heightened by the fact it can be accompanied by an “infinitive” clause, in addition to a RC.6 Example (9) illustrates this construction.

(9) Vlepi kapion kirio na perna me balonia. [CS9] ‘he sees a man passing by with ballons.’ [na perna = ‘infinitive’ markerna + verb 3rd pers. present].

The two structures –perception RCs and perception na-clauses– appear therefore to be favorite constructions in Greek to convey the simultaneity of two events, with a mediating verb which expresses a visual mode of contact. There may therefore be a propensity for NNS to overuse the parallel French perception RC.

6 For a complete list of perception verbs, see Klairis & Babiniotis 2010: 570.

3 2 4 Language Acquisition Monique Monville-Burston and Maria Kounouni

3) Autonomous relatives are almost absent in CS narratives. In our data, RCs in the Greek texts are much more closely integrated into their antecedent NPs than in French. This may help to understand the low number of such RCs in NNS narratives.

9. Conclusion

The analysis of our data revealed a number of similarities between NS and advanced NNS narratives. a) For the two groups, the ratio of non-restrictive RCs to that of restrictive RCs is high, as is the ratio of RCs introduced by qui as compared to other RPs. b) NNS seem to control rather well the use of the French presentative RCs, which are considered to be ‘gallicisms’, i.e. very specific to French, and may not be readily acquired by learners. Like NS, they make use of these theme-rheme structures in the introduction of their narratives (Alors, c’est un monsieur qui est dans une voiture [Example (3)]), to set up the scene and present the protagonists, or they use them to introduce new episodes as “continuative” RCs (Il y a une ambulance qui arrive [Example (4)]). They have thus mastered rather well these text cohesion devices, with the reservation that they do not favour the il y a…qui presentative sequences. There are however differences between the two groups. a) If subordination is taken as an index of complexity, NNS’ texts are less complex than NS’ texts. b) Possibly under the influence of Greek narrative discourse characteristics, NNS use more predicative perception RCs than NS. c) Within the qui clauses group, NNS –more than NS– have recourse to the regular, canonical non-restrictive RCs that serve to give complementary information about properties of the antecedent/referent (descriptions, for example). Such descriptive RCs (like the restrictive RCs) tend to be integrated into, dependent upon, the antecedent NP, at the local micro-syntactic sentential level. This result is consistent with Hankock & Kirchmeyer’s study (2002) of advanced Swedish learners’ narrative competence (see also Bartning & Kirchmeyer 2003; Bartning & Schlyter 2004). In the case of Cypriot learners, this tendency may also be explained by syntactic-discursive features in their native language. Autonomous RCs are significantly more frequent in NS narratives. It seems therefore that NS, as compared to NNS, focus less on the local organization of sentences, and more on the overall discursive organization of the oral texts they produce. The use of less canonical RCs, dissociated from their antecedents, and of il y a…qui presentative constructions can thus be considered to be, in French, a characteristic of NS and very advanced learner/ user narrative competence.

Language Acquisition 3 2 5 Major Trends in Theoretical and Applied Linguistics

References

Bartning, I. and N. Kirchmeyer. 2003. Le développement de la compétence textuelle à travers les stades acquisitionnels en français L2. Acquisition et Interaction en Langue Etrangère 19: 9-39.

Bartning, I. & S. Schlyter. 2004. Itinéraires acquisitionnels et stades de développement en français L2. French Language Studies 14: 281-299.

Cadiot, P. 1976. Relatives et infinitives «déictiques» en français.DRLAV 13: 1-64.

Chini, M. 1998. La subordinazione in testi narrativi di apprendenti tedescofoni: forma e funzione. Linguistica e Filologia 7: 121-157.

Clech-Darbon, A., G. Rebuschi and A. Rialland. 1999. Are there cleft sentences in French? In G. Rebuschi and L. Tuller, The Grammar of Focus. Amsterdam: John Benjamins, 83-118.

Creissels, D. 2004. Cours de syntaxe de sciences du langage, Université Lyon 2. http://www.infotheque.info/cache/8875/lesla.univ-lyon2.fr/article. php3%3fid_article=562.html Retrieved 29 September 2011.

Furukawa, N. 2005. Pour une sémantique des constructions grammaticales. Thème et thématicité. Bruxelles: De Boeck-Duculot.

Gadet, F. 2003. La relative française, difficile et complexe. In S. Kriegel (ed.), Grammaticalisation et réanalyse – Approches de la variation créole et française. Paris: CNRS Éditions, 251–268.

Giacalone Ramat, A. 2000. Typological considerations on second language acquisition. Studia Linguistica 54(2): 123-135.

3 2 6 Language Acquisition Monique Monville-Burston and Maria Kounouni

Givón, T. 2001. Syntax – A Functional-typological Introduction. Amsterdam: John Benjamins. [first ed. 1984]

Hancock, V. 2002. L’emploi des constructions en c’est x qui/que en français parlé: une comparaison entre apprenants de français et locuteurs natifs. Proceedings of the XV Skandinaviske romanitkongress, Romansk Forum 16: 379-388.

Hancock, V. and N. Kirchmeyer. 2002. À la recherche des traits d’une organisation discursive avancée en français L2. La relative aux micro- et macro-niveaux dans un corpus d’apprenants. L’Information grammaticale 93: 3-9.

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

Keenan, E. and B. Comrie. 1977. Noun phrase accessibility and universal grammar. Linguistic Inquiry 8(1): 63-99.

Kirchmeyer, N. 2002. Étude de la compétence textuelle des lectes d’apprenants avancés. Aspects structurels, fonctionnels et informationnels. Doctoral thesis, Forskningsrapporter 17, Institutionen för franska italienska och klassiska språk, Stockholms Universitet.

Kλαίρης, Χ. και Γ. Μπαμπινιώτηs. [Klairis, K. and G. Babiniotis]. 2010. Γραμματική της Νέας Ελληνικής. Αθήνα: Ελληνικά γράμματα. 13rd ed.

Kleiber, G. 1987. Relatives restrictives/relatives appositives: dépassement(s) autorisé(s). Langages 88: 41-63.

Klein, W. & C. Perdue. 1992. Utterance Structure. Developing Grammars Again. Amsterdam: John Benjamins.

Klein, W. & C. Perdue. 1997. The basic variety (or: Couldn’t natural languages be much simpler ?). Second Language Research 13(4) : 310-347.

Kounouni, M. 2010. Le récit oral en L1 et L2: Étude particulière de la connectivité de ‘donc’. MA. thesis, University of Cyprus.

Lambrecht, K. 1988. Presentational cleft constructions in spoken French. In J. Haiman and S.A. Thompson (eds.), Clause Combining in Grammar and Discourse. Amsterdam: John Benjamins, 135-180.

Language Acquisition 3 2 7 Major Trends in Theoretical and Applied Linguistics

Lehmann, C. 1988. Toward a typology of clause linkage. In J. Haiman and S.A. Thompson (eds), Clause Combining in Grammar and Discourse. Amsterdam: Benjamins, 181-225.

Monville-Burston, M. 2008. Évaluer la compétence par un test de jugement de grammaticalité: le cas de la relativisation chez des apprenants chypriotes hellénophones avancés en FLE. In J. Durand, B. Habert & B. Laks (eds.) Congrès Mondial de Linguistique Française (Paris 9-12 juillet), Paris: Institut de Linguistique Française. CD-Rom, 1779-1791.

Monville-Burston, M. & F. Kakoyianni-Doa. 2009. Aspects of the interlanguage of advanced Greek-speaking Cypriot learners of French: Relative clauses. In E. Labeau & F. Myles (eds.), The Advanced Learner Variety: The Case of French. Bern: Peter Lang, 142-170.

Monville-Burston, M. & M. Kounouni. 2008. Introspection et pédagogie : Que disent les apprenants sur la relativisation? In J. Burston, E. Gabriel, M. Monville- Burston & P. Pavlou (eds.), Languages for Intercultural Dialogue. Nicosia: The European Parliament Office in Cyprus and the Ministry of Education and Culture of the Republic of Cyprus, 103-118.

Monville-Burston, M. & M. Kounouni. 2010. Deux procédures expérimentales pour tester la relativisation dans l’interlangue d’apprenants chypriotes hellénophones en FLE. In A. Psaltou-Joycey and M. Mattheoudakis (eds.), Advances in research on language acquisition and teaching. Selected papers. Actes du 14ème Colloque International de l’Association Grecque de Linguistique Appliquée (14-16 déc. 2007).

Morel, M.-A. and L. Danon-Boileau. 1998. Grammaire de l’intonation. L’exemple du français oral. Paris: Ophrys.

de Vries, M. 2006. The syntax of appositive relativisation: On specifying coordination, false free relatives, and promotion. Linguistic Inquiry 37: 229-270.

3 2 8 Language Acquisition Monique Monville-Burston and Maria Kounouni

Appendix

Summary of the three Mordillo cartoons to be narrated

1. First aid Along a country road, a man is driving a car that does not seem to be running very well. He finally crashes into a tree. An ambulance arrives at once. Three male nurses come out. Two of them load the wrecked car onto a stretcher while the third one puts bandages on the tree. The ambulance departs, leaving the stunned driver on the road.

2. At the circus The scene shows a circus performer throwing knives around his partner, a woman, who is backed by a wooden panel. You can hear the audience applaud. Then the woman ties a headband around the artist’s eyes: it is the second part of the performance. Then we can hear the woman’s footsteps and we assume she is leaving the stage. But we see her hidden behind the panel. After having thrown all his knives, the man takes off his headband among the whistles and applause of the audience. We discover then that his partner had resumed her initial position between the blades.

3. The child’s friend The scene takes place in a park. A woman is sitting on a bench watching her child playing in a sandbox. A man arrives, greets the woman and sits beside her. The child does not like this man. When he addresses his mother the child responds by giving him a big kick in the leg. The man does not lose his cool and tries to coax the child by giving him a candy. The child then bites his finger and throws sand into his eyes. Then along comes a balloon seller. The man has an idea and rushes towards him. He returns with all the balloons and promptly offers them to the child, who accepts them. The child, who is lighter than the balloons, leaves the ground and slowly disappears into the air. The man having gotten rid of the child is finally alone with the woman whom he kisses on the cheek.

Language Acquisition 3 2 9 Major Trends in Theoretical and Applied Linguistics

3 3 0 Language Acquisition Clitic Placement in Early Grammars: the Case of Cypriot Greek1

Theoni Neokleous and Teresa Parodi

University of Cambridge [email protected], [email protected]

Abstract

In this paper, we investigate the acquisition of clitic placement in Cypriot Greek, on the basis of spontaneous speech and elicited production data, both cross– sectional and longitudinal. Data analysis reveals ceiling percentages of misplaced clitics in proclisis contexts for a subset of children below the age of 3, while enclisis contexts are adult–like from the onset. On the basis of this result, we propose that enclisis is the default pattern for clitic placement in Cypriot Greek and we attribute the overgeneralization of enclisis to the overgeneralization of verb movement (Petinou & Terzi 2002). Clitics, first language acquisition, Cypriot Greek, syntax

1. Introduction

The first language acquisition of clitic constructions has been widely studied cross–linguistically. Clitics are pronominal elements morpho–phonologically deficient, in Cardinaletti & Starke’s (1999) sense, that historically derive from strong pronouns and have variable placement. Clitics are lexical elements highly dependent on the functional domain. Thus, they constitute a good tool for assessing the development of this domain. Over the last twenty years, many studies have been conducted on the first language acquisition (henceforth L1A)

1 This research was conducted within the project «L1 Acquisition of Cypriot Greek Pronominal Clitics» coordinated by Dr. Kleanthes Grohmann (University of Cyprus). This project was funded by the Cyprus Research Promotion Foundation within the program «Young Researchers of Cyprus» (project protocol no. ΠΕΝΕΚ/0609/42), under the «Framework Program for Research, Technological Development and Innovation 2009–2010», that is co–funded by the Republic of Cyprus and the European Regional Development Funds.

Language Acquisition 3 3 1 Major Trends in Theoretical and Applied Linguistics

of clitics in a number of languages including French (Hamann et al. 1996), Dutch (Schaeffer 2000), European Portuguese (Costa & Lobo 2007 et seq.), Standard Modern Greek (Marinis 2000, Stephany 1997), Italian (Guasti 1993/94), Romanian (Babyonyshev & Marin 2005), Serbo/Croatian (Ilic & Ud Deen 2003), Spanish and Catalan (Wexler et al. 2004). The focus of the aforementioned studies regarding the syntax of clitics has been the phenomenon of clitic realization and omission at the onset of L1A. An interesting discrepancy has been observed among clitic languages: children acquiring languages like Standard Modern Greek (henceforth SMG) have adult–like clitic production by age 3 (Marinis 2000), whereas for language like European Portuguese, high rates of clitic omission have been reported in the speech of much older children (Costa & Lobo 2007 et seq.). There is, however, another intriguing phenomenon regarding the L1A of clitics that is so far understudied: clitic misplacement. This phenomenon has been observed in early European Portuguese (Costa & Lobo 2007 et seq., Duarte & Matos 2000) and Cypriot Greek (Petinou & Terzi 2002). Clitic misplacement in early European Portuguese (henceforth EP) has not been systematically investigated so far, whereas for early Cypriot Greek (henceforth CG), Petinou and Terzi (2002) studied the phenomenon on the basis of data from 5 typically developing children and 5 children diagnosed with SLI. Both EP and CG are languages with a mixed pattern of clitic placement. Yet, unlike Romance languages, in EP and CG clitics appear post–verbally in finite contexts, unless the clause is headed by a proclisis–trigger. Notably, in languages where clitics usually precede the finite verb, no instances of clitic misplacement have been reported even at the earliest stages of L1A. This observation leads to another interesting discrepancy among clitic languages: clitic languages with and clitic languages without clitic misplacement at the onset of L1A. Clitic misplacement is an indication of an impoverished clausal structure in child grammar, whereas for clitic omission this may not be necessarily the case. Following Marinis’ (2000) argumentation, we speculate that within the minimalist program and assuming the syntax–morphology interface, as in the framework for Distributed Morphology (Halle & Marantz 1993), the computational system operates prior to spell out with formal features, and lexical items (clitics as well) are inserted after spell out. Thus, if clitics are absent from child speech, this may well be the result of an incomplete lexicon, while clitic misplacement is sufficient evidence for an impoverished computational system. Early CG constitutes a good test for the phenomenon of clitic misplacement, because young CG–speaking children have good clitic production (Grohmann 2010) even at the earliest stages of L1A, whereas for early EP (Costa et al. 2007 et seq.) both clitic misplacement and high rates of clitic omission have been reported.

3 3 2 Language Acquisition Theoni Neokleous and Teresa Parodi

2. CG Pronominal Clitics

CG, like SMG, has only pronominal clitics, as it lacks the reflexive, locative and auxiliary clitics found in Romance and Slavic languages (Terzi 1999a:86, footnote 2); henceforth the terms clitics / object clitics / pronominal clitics will be used inter–changeably to refer to pronominal object clitics in CG. CG clitics are inflected for number, case, person and gender (see table 1) and are verb adjacent. They are banned from clause initial position and have variable placement appearing either pre– or post–verbally depending on the syntactic context.

Table 1. The morphological paradigm of CG clitics

Number / 3rd person 1st person 2nd person Case Masculine Feminine Neuter Singular Genitive mu su tu tis tu Accusative me se to(n) ti(n) to Plural Genitive mas sas tus tus tus Accusative mas sas tus tes ta

CG clitics immediately follow the finite verb (1), unless the clause is headed by a proclisis–trigger. Negative particles, modal particles, wh–elements, XPs bearing contrastive focus and the factive complementizer pu trigger proclisis, with clitics immediately preceding the finite verb (2).

(1) Efera to. Brought–1S it–CL.ACC ‘I brought it’

(2) En to efera. NEG it–CL.ACC brought–1S ‘I didn’t bring it’

To date, three syntactic analyses have been put forward to account for clitic placement in CG and they all assume that enclisis derives from proclisis as a result of the verb raising past the clitic (Agouraki 2001, Terzi 1999a; 1999b). However, they attribute verb movement to different operations and they propose different landing sites for the verb. Terzi (1999a, 1999b) adopts Kayne’s (1994) antisymmetry proposals and offers two distinct but similar analyses. In both, Greek clitics in general and

Language Acquisition 3 3 3 Major Trends in Theoretical and Applied Linguistics

CG clitics in specific are considered X°s2. Terzi’s first proposal (1999a) departs from the idea that there is a designated functional head, where clitics adjoin in all syntactic environments. She advocates a bipartition with respect to the type of functional heads that serve as adjunction sites for clitics: clitics adjoin to T° when tense features are weak and to mere placeholders in all remaining contexts. Thus, clitics adjoin to T° in gerunds and imperatives3 and to an F° head4 in all other contexts. A T° with weak features (defective or specified as [–T]) becomes a legitimate adjunction site for clitics, since the verb does not adjoin to it overtly for feature checking and a configuration of multiple adjunction is not created5. Terzi proposes this bipartition based on the relative order of the dative and accusative clitic in double clitic constructions. She argues that in CG finite contexts, there is no free ordering of the dative and accusative clitic within clitic clusters, with the former always preceding the latter, while the imperative verb (bearing [–tense] Infl in her analysis) allows free ordering6 (Terzi 1999a:115). According to Terzi, CG clitics originally adjoin to a functional head higher than IP in all syntactic environments, patterning with SMG clitics (3). In CG clitic constructions, the finite verb undergoes movement, with M°, the head of Mood Phrase and the functional head that encodes modality, as its landing site, giving rise to the order V–clitic (1), unless the clause is headed by a functional head with operator–like properties7 (2).

2 Terzi bases her claim on clitic doubling facts such as Ton ida ton Yianni ‘him–CL saw–1S the John–ACC’, where the doubled phrase is considered to be the Specifier of a determiner head, following Uriagereka (1995), hence, clitics are heads (X°s) (1999a: 86, footnote 2). This type of construction is found in both SMG and CG, so provided Terzi’s argument is verified, the status of clitics is the same in both varieties. 3 Terzi assumes that the imperative tense involves a [–tense] Infl (Terzi 1999a: 94). 4 Terzi argues that the F head “is a functional head partially reminiscent of the head of Clitic Voice of Sportiche (1996) (but significantly different from Uriagereka’s (1995) F)” (Terzi 1999a: 93) and does not check features of the verb overtly. 5 After spell out, the verb raises to T° in order to have its Tense features checked (Terzi 1999a: 96). 6 If this piece of evidence were valid, it would have offered direct support for her claim that clitics do not always adjoin to the same functional head but their adjunction site depends on the Tense properties of their verbal host. However, in CG neither the finite nor the imperative verb allow free ordering of the dative and accusative clitic within clitic

clusters: the relative order is CLDAT–CLACC in all syntactic contexts. 7 The fact that V–to–M movement is manifested in CG alone, but not in SMG, is attributed by Terzi to the Infl make–up of CG, specifically, to the feature composition of M°. She bases her claim on two pieces of evidence: (1) the fact that unlike SMG, CG lacks [–present] compound tenses, and (2) the fact that CG lacks the future particle of SMG (Terzi 1999a: 110, footnote 24).

3 3 4 Language Acquisition Theoni Neokleous and Teresa Parodi

(3) [C…[F to [F 0 [T diavasai [Agr ei [V ei]]]]]] ((14) in Terzi 1999a:93) it read–1S

In her second proposal, Terzi (1999b) suggests that clitics in both CG and SMG adjoin to the same featureless functional head F°, which takes IP as its complement (4). Her claim regarding the post–verbal position of CG clitics remains the same as her original proposal: namely, she assumes that the finite verb in CG undergoes movement with M° as its landing site. Yet, in her second proposal, verb movement is attributed to the licensing requirements of CG clitics: they have strong features that must be licensed in the internal domain of a functional head with operator– like properties8, i.e. the Neg head, the Mood head, WH–operators or XPs bearing contrastive focus. In the absence of such a functional head, the verb undergoes V–to–M movement and clitics surface post–verbally.

(4) [CP [C’ COMP [MP [M’ [FP [F CL [F’ [IP [VP [V’….]]]]]]]]]] ((19) in Terzi 1999b: 231).

Agouraki (2001) offers a different analysis. She suggests that CG clitics head clitic phrases located between IP and CP (5), and clitic placement does not involve cliticization; it only depends on verb placement. She postulates that CG has a filled C° requirement, for clause–typing purposes; “the item which fills C° informs us on the type of the clause” (2001:14). In Agouraki (2001) the C° position can host overt or null (i.e. WH and Focus, when the [Spec,CP] position is filled by wh–phrases, or XP–foci, respectively) complementizers, negation or modality markers. If the C° position is filled, V–to–C raising is blocked, otherwise V–to–C is manifested.

(5) [CP WH/F/Top [C’ C ([WH/F]) [NegP [Neg’ Neg [ClP [Cl’ Cl [IP [I’ I VP]]]]]]]]] (based on the clause structure in (7) in Agouraki 2001:5)

Early CG constitutes a good test for the aforementioned analyses. If the phenomenon of clitic misplacement is generalized across young CG–speaking children, the existence of a fully–fledged clausal structure in child Greek–Cypriot grammar will be challenged. Moreover, child data will test the availability of the two functional heads suggested so far, M and C, by Terzi (1999a, 1999b) and Agouraki (2000) respectively, to be the adjunction site for the finite verb.

8 “This is a type of licensing that involves a head–complement rather than Spec–head relation and must be satisfied before spell out” (Terzi 1999b: 233).

Language Acquisition 3 3 5 Major Trends in Theoretical and Applied Linguistics

3. Research Questions

This paper examines the L1A of object clitics in CG. The aim is twofold. Firstly, we want to investigate clitic misplacement in early grammars, focusing on early CG. This means that we will try to replicate Petinou and Terzi’s (2002) results on the basis of a larger database of child language and test their generalizability across typically developing children. The first set of research questions is related to the existence of the phenomenon and its nature: a. Do young Greek–Cypriot children misplace clitics? b. Do they manifest true optionality in their clitic placement or are their choices systematic? In the second place, we want to test the syntactic accounts so far proposed for clitic placement in CG, and try to answer the following question: c. How can this non–adult–like pattern be accommodated within a syntactic account?

The first question can be re–phrased as follows: do children use post–verbal clitics in proclisis contexts and / or pre–verbal clitics in enclisis contexts? Root clauses and imperatives are enclisis contexts, thus adult–like production would involve a post–verbal clitic (6), while pre–verbal clitic placement would be considered non–adult–like (7). Reversely, in proclisis contexts, i.e. in negatives, clauses headed by wh–elements, modality markers or XPs bearing contrastive focus, pre–verbal clitics are grammatical (8) and post–verbal clitics are ungrammatical9 (9).

(6) Efera to. Brought–1S it–CL.ACC ‘I brought it’ (7) *To efera. (8) Na to fero. M it–CL.ACC bring–1S ‘To bring it’ (9) *Na fero to.

9 The use of the terms non–adult–like for the former possible instantiation (7) and ungrammatical for the latter (9) is not arbitrary. Pre–verbal clitic placement in root clauses (7) is adult–like in SMG, therefore such an instantiation could constitute evidence for influence from the Standard variety. Yet, the use of post–verbal clitics in negative clauses and clauses headed by wh–elements, modality markers or XPs bearing contrastive focus is not available in either adult CG or adult SMG; it is, thus, ungrammatical in both varieties.

3 3 6 Language Acquisition Theoni Neokleous and Teresa Parodi

The second question has theoretical implications for the nature of child grammar. Random use of pre– and post–verbal clitics indicates true optionality regarding clitic placement, showing lack of a systematic grammar. In contrast, the over–generalization of either the proclisis or the enclisis pattern, and the over–production of pre–, or post–verbal clitics, respectively, across syntactic contexts, is an indication of a systematic, yet non–adult–like grammar.

4. Methodology

The L1 acquisition of clitic placement in CG was studied on the basis of spontaneous speech and elicited production data, both cross–sectional and longitudinal. All the participants of our studies were monolingual CG–speaking children, with a monolingual CG–speaking background and no history of cognitive deficits or language impairments. As for the socio–economic status of their families, they come from middle–class families residing in Limassol. The participants of the spontaneous speech study are different from the participants of the elicited production study, thus our database consists of data from 58 children overall.

4.1. Spontaneous Data

Samples of spontaneous speech were collected from 8 children, aged 2;4 to 3;4. Children were audio–recorded at home, while they interacted with the experimenter. Picture books, wooden puzzles and stickers were used as prompts for the elicitation of constructions involving object clitics. Each recording session lasted approximately an hour and the data were orthographically transcribed in accordance with the conventions of the CHAT transcription format (MacWhinney 2000; 2011), used by CHILDES. These transcripts constituted the corpus of utterances for each child. Data analysis included only (a) fully intelligible, (b) multi–word, and (c) spontaneously used utterances (imitations and immediate self–repetitions were discarded) and involved the following stages: A. Pin–pointing of all the clitic constructions produced. B. Calculation of the proclisis and enclisis contexts: the number of contexts where an adult would produce pre–, and post–verbal clitics, respectively. C. Calculation of the raw numbers and the respective percentages of correct / incorrect clitic placement per condition. Table 2 shows the overall clitic production and the number of clitics produced per condition. 912 clitic constructions were produced, with a mean number of 114 clitics produced per child: 443 clitics were produced in enclisis contexts and 469 in proclisis contexts.

Language Acquisition 3 3 7 Major Trends in Theoretical and Applied Linguistics

Table 2. Clitic production (Spontaneous data)

Enclisis Contexts Proclisis Contexts

912

443/912 469/912

100%

49% 51%

In enclisis contexts, all the children performed adult–like and placed the clitics post–verbally. Yet, this was not the case for proclisis contexts: a subset of children performed adult–like, producing pre–verbal clitics, while another subset had ceiling percentages of incorrect clitic placement, producing post– verbal clitics. Figure 1 shows the proportion of correct and incorrect clitic placement per child in proclisis contexts; the relevant fi gures for incorrect clitic placement are reported on the graph.

Figure 1. Clitic placement in proclisis contexts (Spontaneous data)

S1, the younger participant, was followed longitudinally for a period of 6 months. Table 3 shows his clitic production per context and per construction from age 2;4 to 2;10. All the clitic constructions he produced in enclisis contexts were adult–like with regard to clitic placement, while all the clitic constructions he produced in proclisis contexts were non–adult–like, apart from a single negative clause at age 2;8. His consistency in incorrect clitic placement in proclisis contexts from age 2;4 to 2;10 is visualised in fi gure 2.

3 3 8 Language Acquisition

100

80

60 Percentage of S1 Misplaced Clitics 40 (%) 20

0 2 2.2 2.4 2.6 2.8 3 3.2 Age (years.months) Theoni Neokleous and Teresa Parodi

Table 3. Clitic production (Longitudinal data)

Enclisis Proclisis Contexts Contexts Overall Age Clitic Production IMPER ROOT M NEG WH PU

15 20 30 2 0 2 2;4 69 (100%) 35 (51%) 34 (49%)

25 39 33 2 0 0 2;5 99 (100%) 64 (65%) 35 (35%)

6 33 21 9 5 2 2;8 76 (100%) 39 (51%) 37 (49%)

8 32 53 3 2 1 2;10 99 (100%) 40 (40%) 59 (60%)

100

80

60 Percentage of S1 Misplaced Clitics 40 (%) 20

0 2 2.2 2.4 2.6 2.8 3 3.2 Age (years.months)

Figure 2. Clitic misplacement in proclisis contexts (Longitudinal data)

4.2. Elicited Production

50 Greek–Cypriot children from three age groups: A:2;6–3;0 (n=18), B:3;0–3;6 (n=22), C:3;6–4;0 (n=10) participated in this study. They were randomly recruited from 5 nursery schools, after approval from the directors and upon written parental consent. Each experimental session involved the implementation of a picture–based task and was preceded by a warm–up session. It lasted approximately 20 minutes and was audio–recorded.

Language Acquisition 3 3 9 Major Trends in Theoretical and Applied Linguistics

The picture–based task is an elicited–production task for 3rd person singular object clitics. The experimental material consists of 12 pictures taken from the book First Hundred Words in English (edited by Amery & Cartwright 2009) matched with 12 questions. The task was designed for the elicitation of 3 types of constructions: (1) root clauses, (2) negatives and (3) na/enna–clauses. Root clauses constitute an enclisis context, while negatives and na/enna–clauses constitute proclisis contexts. Only 3rd person singular object clitics were elicited and we controlled for the genitive and accusative case and all the 3 genders to be equally represented in the task. The children were shown the pictures one at a time and the experimenter would point at a picture and ask a question related to the situation depicted (10), in order to elicit a clitic construction (11). 4 clauses per condition were elicited.

(10) Ti θeli na kami to koritsaki to kaδro? [Experimenter] What want–3S M do–3S the girl–NOM the frame–ACC ‘What does the girl want to do the frame?’ (11) Na to kremasi. [Elicited Clause] M it–CL.ACC hang–3S ‘To hang it’

Table 4 shows the overall number of clitic constructions produced per condition and the mean number of clitic constructions produced per child per condition. While the production of root and na/enna–clauses was good, the number of negative clauses elicited was extremely low. Thus, no safe conclusions can be drawn on the basis of the negative clauses.

Table 4. Clitic production (Experimental data)

Adult–like clitic Structure Overall production Mean production placement Root Clauses 138/200 (69%) 2.76/4 137/138 (99%)

Na/enna–clauses 123/200 (61%) 2.46/4 107/123 (87%)

Negatives 18/200 (9%) 0.36/4 16/18 (89%)

The last column of table 4 reports the proportion of adult–like clitic placement per experimental condition. If we compare root and na/enna–clauses, we observe that children perform adult–like in root clauses (only 1% of non–adult–like productions) but not in na/enna–clauses (13% of non–adult–like productions). Since table 4 presents group results, it is worth noting that these 16 na/enna– clauses involving misplaced clitics were produced by 7 children below the age of 3;3, who exhibited consistent post–verbal clitic placement in all syntactic contexts.

3 4 0 Language Acquisition Theoni Neokleous and Teresa Parodi

Figures 3–5 show the proportion of children who consistently manifest correct (grey color) or incorrect (black color) clitic placement in na/enna–clauses per age group. The fi gures reveal that the number of children who consistently misplace clitics decreases with age, with no children older than 3;3 producing any misplaced clitics. A factorial ANOVA was used for between groups analysis and has shown that children’s incorrect clitic placement in na/enna–clauses diff ers signifi cantly among age groups,F (2) = 3.64, p = .034, while Scheff e post– hoc comparisons indicated that the performance of age group A (M = .72) diff ers signifi cantly from the performance of age group B (M = .14) as well as from age group C (M = .00).

Figure 3. Age Group A: 2;6–3;0 Adult–like: 72% (n=13) Non–adult–like: 28% (n=5)

Figure 4. Age Group B: 3;0–3;6 Adult–like: 91% (n=20) Non–adult–like: 9% (n=2)

Figure 5. Age Group C: 3;6–4;0 Adult–like: 100% (n=10) Non–adult–like: 0% (n=0)

Language Acquisition 3 4 1 Major Trends in Theoretical and Applied Linguistics

Based on the results obtained from the spontaneous speech data (figure 1) and the elicited production experiment (table 4), we report ceiling percentages of adult–like clitic placement in enclisis contexts from the onset of L1A in CG: all the children produced exclusively post–verbal clitics. Notably, the youngest child who participated in this study has adult–like clitic placement in enclisis contexts from age 2;4. Yet, this does not apply to proclisis–contexts. Children’s clitic placement shows a bimodal distribution (figure 1): most children perform adult–like, whereas a subset of children younger than 3 produces post–verbal instead of pre–verbal clitics at ceiling percentages, regardless of the presence of proclisis triggers. Noteworthy, only enclisis–pro–proclisis was found, not the reverse. The longitudinal study (figure 2) shows that children included in this subset may be consistent regarding their incorrect clitic placement for months, while figures 3–5 show that clitic misplacement decreases with age.

5. Discussion

Returning to the research questions originally posed, we report that young CG– speaking children misplace clitics and do not manifest true optionality in their clitic placement: their choices are systematic. The outcome of our study partly confirms Petinou and Terzi (2002), who argue that the initial stage of clitic L1A in CG is characterized by the overgeneralization of enclisis. Yet, our results differ from theirs in three crucial aspects. First, we have shown that this pattern is not generalized across participants, since a subset of children below the age of 3 performs adult–like in both enclisis and proclisis contexts. In particular, only 28% of the children between 2;6 and 3;0 overgeneralize the enclisis pattern (figure 3). In the second place, Petinou and Terzi report percentages for clitic misplacement ranging between 7%–66% even for the youngest participants of their study (2002:13, table 2), whereas the analysis of our data revealed that those children who misplace clitics do so in 98%–100% of the proclisis contexts (figure 1). Thirdly, Petinou and Terzi (2002) report gradual decrease of clitic misplacement over a period of 2 months (i.e. O.X. 10% → 2% → 0%, N.A. 66% → 28% → 12%, A.I. 62% → 14% → 0%). However, based on our longitudinal study, children with ceiling percentages of misplaced clitics may be consistent in their incorrect clitic placement for as long as 6 months. Given that our study is based on data from 58 typically developing children, a larger database than Petinou and Terzi’s 5 children, it, arguably, constitutes a safer basis for conclusions. Turning, now, to the last research question posed, we will try to accommodate this pattern of clitic placement within a syntactic account. We argue in favor of a designated adjunction site for clitics in the spirit of Uriagereka (1995). Proclisis–enclisis alternation in CG does not correlate with finiteness, as in

3 4 2 Language Acquisition Theoni Neokleous and Teresa Parodi

SMG10 (Mavrogiorgos 2009), thus if CG clitics left–adjoin to the verbal host and ex–corporation is not a possibility (Kayne 1994), enclisis cannot be derived. CG clitics head their own projections, resembling both Agouraki’s (2001) Clitic phrase and Terzi’s (1999a, 1999b) F phrase. Enclisis derives from proclisis through the manifestation of verb movement past the clitic, again in accordance with both Agouraki (2001) and Terzi (1999a, 1999b). If we try to account for clitic misplacement found in early CG based on Agouraki’s (2001) analysis, we face an important obstacle: there is no functional head located between the clitic phrase and the NegP / C° for the verb to adjoin in order to get the linear order modal/negative particle + verb + clitic (see structure in (5)). An alternative would be to assume that modal and negative particles, as well as elements that appear in Spec CP (i.e. wh–elements), are adjuncts in child language, in the spirit of Tsimpli (1996), and that the verb moves and adjoins to C°. Yet, postulating that elements in C° and Spec CP are adjuncts entails that V– to–C is not legitimate. Terzi’s analyses (1999a, 1999b), on the other hand, seem more promising for accommodating the pattern attested. Following the argumentation in Petinou and Terzi (2002), we attribute clitic misplacement to the overgeneralization of V– to–M movement in all syntactic contexts. Of course, on the basis of the structure given in (4) the M° head in negatives and na/enna–clauses is filled by negative, and modal particles, respectively. Thus, there is no available head for the verb to adjoin. What Petinou and Terzi (2002) suggest is that children perceive the negative and modal particles as phrasal and locate them in the Spec MP instead of the M° head. In Spec MP, these particles cannot check the verbal features of M° and V–to–M is needed for the V–features of M° to be checked. As soon as children acquire the properties of the modal and negative particles, they will locate them in M°, the verbal features of M° will be checked, no verb movement will be needed, and clitic placement will be pre–verbal, thus, adult–like. Petinou and Terzi’s (2002) proposal seems promising for accommodating early CG data, yet the authors offer no strong evidence to support their claims. They take the following facts to indicate children’s misconception of the properties of the inflectional particles: the omission of na and a (single) illicit use of o(h) i, the negator used for constituent negation, instead of (dh)en. As for the first point, according to Stephany (1997), na is either omitted or reduced to schwa until the age of 28 to 34 months in early Greek, yet, clitic placement in SMG– speaking children is adult–like from the onset. Moreover, both phenomena used by Petinou and Terzi to justify their claim are found in SLI data alone, and, as

10 “Proclisis correlates with non–restricted/full person agreement on T, while enclisis correlates with restricted person agreement on T” (Mavrogiorgos 2009: 292).

Language Acquisition 3 4 3 Major Trends in Theoretical and Applied Linguistics

the authors themselves observe, “none of the study children with NLD (normal language development) omitted any subjunctive particle” or “made the wrong choice of negative marker” (2002:24). Therefore, Petinou and Terzi’s (2002) analysis can account for early CG data provided that their suggestion for the misanalysis of the negative and modal particles at the onset of L1A in CG is verified by ample evidence, ideally, on the basis of phenomena unrelated to clitic placement. On the basis of our study, this parameter cannot be examined and remains open for future research. Leaving aside the specifics of the syntactic analyses, the main finding of our investigation is that the enclisis pattern is over–generalized at the onset of L1A of clitic constructions in CG by a subset of children below the age of 3, while no child older than 3;3 misplaces clitics. This confirms that age 3 is a milestone for language development with regard to clitic placement as well. The fact that only enclisis–pro–proclisis was found and not the reverse order can be taken as an indication that the default pattern for clitic placement in CG is the enclisis pattern. It is worth exploring if enclisis is the default pattern for languages exhibiting similar positioning restrictions for clitic placement with CG, like EP and Galician.

3 4 4 Language Acquisition Theoni Neokleous and Teresa Parodi

References

Agouraki, Y. 2001. The position of clitics in Cypriot Greek. Proceedings of the First International Conference of Modern Greek Dialects and Linguistic Theory. University of Patras, 1–18.

Babyonyshev, M. and S. Marin. 2005. The acquisition of object clitic constructions in Romanian. In R.S. Gess and E.J. Rubin (eds.), Theoretical and experimental approaches to Romance linguistics. Amsterdam: John Benjamins, 21–40.

Cardinaletti, A. and M. Starke. 1999. The Typology of Structural Deficiency: A case study of the three classes of pronouns. In H. van Riemsdijk (ed.), Clitics in the languages of Europe. Berlin: Mouton de Gruyter, 145–233.

Costa, J. and M. Lobo. 2007. Clitic omission, null objects or both in the acquisition of European Portuguese? In S. Baauw, F. Drijkoningen and M. Pinto (eds.), Romance Languages and Linguistic Theory 2005. Amsterdam: John Benjamins, 59–72.

Duarte, I. and G. Matos. 2000. Romance Clitics and the Minimalist Program. In J. Costa (ed.), Portuguese Syntax: New Comparative Studies. Oxford University Press, 116–142.

Grohmann, K. 2011. Some directions for the systematic investigation of the acquisition of Cypriot Greek: A new perspective on production abilities from object clitic placement. In E. Rinke and T. Kupisch (eds.), The Development of Grammar: Language acquisition and diachronic change – Volume in honour of Jürgen M. Meisel. (Hamburg Studies on Multilingualism 11). Amsterdam: John Benjamins, 179–203.

Guasti, M.T. 1993/94. Verb syntax in Italian child grammar: finite and nonfinite verbs. Language Acquisition 3(1): 1–40.

Language Acquisition 3 4 5 Major Trends in Theoretical and Applied Linguistics

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 linguistics in honor of Sylvain Bromberger. Cambridge, MA: MIT Press, 111–176.

Hamann, C., L. Rizzi and U.H. Frauenfelder. 1996. On the acquisition of subject and object clitics in French. In H. Clahsen (ed.), Generative perspectives on language acquisition. Amsterdam/ Philadelphia: John Benjamins, 309–334.

Ilic, T. and K.U. Deen. 2004. Object raising and cliticization in Serbo–Croatian child language. In J. van Kampen and S. Baauw (eds.), Proceedings of GALA 2003 (Generative Approaches to Language Acquisition) Volume 1. Utrecht: LOT, 235– 243.

Kayne, R.S. 1994. The antisymmetry of syntax. Cambridge, MA: MIT Press.

MacWhinney, B. 2000. The CHILDES project: tools for analyzing talk (3rd edition). Mahwah, NJ: Lawrence Erlbaum Associates.

MacWhinney, B. 2012. The CHILDES project: tools for analyzing talk (electronic edition). Downloadable at http://childes.psy.cmu.edu/manuals/chat.pdf

Marinis, T. 2000. The acquisition of clitic objects in Modern Greek: Single clitics, clitic doubling, clitic left dislocation. ZAS Papers in Linguistics 15. Berlin: ZAS, 259–281.

Mavrogiorgos, M. 2009. Proclisis and enclisis in Greek. Unpublished PhD dissertation, University of Cambridge.

Petinou, K. and A. Terzi. 2002. Clitic misplacement among normally developing children and children with specific language impairment and the status of Infl heads. Language Acquisition 10(1): 1–28.

Schaeffer, J.C. 2000. The acquisition of direct object scrambling and clitic placement: syntax and pragmatics. Amsterdam: John Benjamins.

Sportiche, D. 1996. Clitic Constructions. In J. Rooryck and L. Zaring (eds). Phrase Structure and the Lexicon. Dordrecht: Kluwer Academic Publishers, 213–276.

Stephany, U. 1997. The acquisition of Greek. In D.I. Slobin (ed.), The crosslinguistic study of language acquisition Vol.4. Mahwah, NJ: Lawrence Erlbaum Associates, 183–334.

3 4 6 Language Acquisition Theoni Neokleous and Teresa Parodi

Terzi, A. 1999a. Clitic combinations, their hosts and their ordering. Natural Language and Linguistic Theory 17: 85–121.

Terzi, A. 1999b. Cypriot Greek clitics and their positioning restrictions. In A. Alexiadou, G. Horrocks and M. Stavrou (eds.), Studies in Greek Syntax. Natural Language and Linguistic Theory Series. Dordrecht: Kluwer Academic Publishers, 227–240.

Tsimpli, I.M. 1992/1996. The Prefunctional Stage of First Language Acquisition: A crosslinguistic Study. PhD Dissertation, University College London. New York / London: Garland.

Uriagereka, J. 1995. Aspects of the syntax of clitic placement in western Romance. Linguistic Inquiry 26(1): 79–123.

Wexler, K., A. Gavarró and V. Torrens. 2004. Feature checking and object clitic omission in child Catalan and Spanish. In R. Bok–Bennema, B. Hollebrandse, B. Kampers–Manhe and P. Sleeman (eds.), Romance Languages and Linguistic Theory 2002. Amsterdam/Philadelphia: John Benjamins, 253–268.

Language Acquisition 3 4 7 Major Trends in Theoretical and Applied Linguistics

3 4 8 Language Acquisition Specificities of Syntactic Movement in Early Speech

Elena Papadopoulou1,2 and Natalia Pavlou3

1University of Essex, 2University of Cyprus, 3University of Chicago [email protected], [email protected]

Abstract

This paper discusses syntactic errors and strategies found in child speech in an attempt to outline the linguistic development in early speech. The focus lies on children’s ungrammatical utterances appearing to violate the Pied-piping condition in D-linked wh-questions as these are found in two different elicitation methodologies. The violation of grammatical constraints, such as the one discussed, supports the role of Economy in languages with a relevant discussion on the syntactic dependencies involved. The study targets the understanding of Movement specificities in Greek Cypriot children in order to identify the errors in these structures and the specificities of the variety.

1. Introduction

This paper aims to investigate patterns of syntactic errors in the production of D-linked wh-questions in Cypriot Greek (henceforth, CG) based on two different elicitation experiments. The study of errors or non-target responses in child speech can reveal the development of language and more specifically the order of the acquisition of syntactic phenomena. In first language acquisition (L1), it is assumed that certain structures are late acquired either because of their syntactic or semantic complexity. The complexity involved in a structure is surely a determining factor for the toddler, but the errors produced in the process of acquiring that complexity can indicate children’s simple understanding of syntax in their mastering of language. With focus on a specific variety, Cypriot Greek, data were drawn from two different experiments (Papadopoulou 2013; Pavlou 2010b) at different times of testing, where children aged 2;8 – 6;11. participated. The specificities observed involved a series of similar patterns found in the acquisition of D-linked questions, which are syntactically assumed to show pied-piping (Ross 1967), as indicated in (1) below:

Language Acquisition 3 4 9 Major Trends in Theoretical and Applied Linguistics

(1) [Inda milon] kofki ɪ kopela? which apple.ACC cutting.3SG the.NOM girl.NOM ‘Which apple is the girl cutting?’

Following the different errors and non-target responses observed in the two experiments, we argue that these are syntactically-motivated patterns, which are driven by syntactic reasons and thus appear even at 6 years old (henceforth, yr) children. In this paper, we aim to briefly discuss the background literature with regard to errors found in the acquisition of D-linked1 questions in several languages in order to list the different error patterns involved. In Section 2, we will give an overview of wh-questions and their formation in Cypriot Greek. Data were drawn from a syntactic priming experiment, presented in Section 3 and an elicitation task, presented in Section 4. Section 5 focuses on decomposing the errors observed in the experiments and providing an explanation that supports Economy in syntax and illustrates a clearer picture for specificities in early speech. By illustrating the different examples of errors appearing in the data, we will provide a theoretical analysis by adopting the Immediate Move Hypothesis (Pavlou 2012).

1.1. Background Literature

To start with, pied-piping in D-linked questions is generally assumed to involve a wh-word, which is a determiner that moves to the target position and pied- pipes the NP along with it, see (1) above. The study of pied-piping (Ross 1967) in D-linked questions and other structures has been a matter of many studies (Butler & Mathieu 2005; Cable 2008; Fanselow & Cavar 2002; Heck 2008; Mathieu 2002 among others) for decades with focus on the optionality or not of pied-piping in certain languages such as French, as in (2) below.

(2a) Combien as-tu lu de livres? how-many have-you read.2SG of books ‘How many books have you read?’ (2b) Combien de livres as-tu lu? how-many of books have-you read.2SG ‘How many books have you read?’

1 These questions are also called Referential, because they refer to the binding chain of the referred NP with the wh-word.

3 5 0 Language Acquisition Elena Papadopoulou and Natalia Pavlou

A recent acquisition study of ninety pre-school children (4;0-7;0) in Standard Modern Greek (SMG) reported error findings related to sub-extraction of wh- phrases (Asproudi 2011). Asproudi reports that sub-extraction of wh-phrases was the most frequent technique and argues that these are in line with Dutch data (van Kampen 1997). In her conclusions, she argues that the morphological richness of SMG is a key factor to the possibility of sub-extraction of wh-phrases in child speech. Stavrakaki (2006) after testing 8 SLI children with 2 control children for each one tested, argues that even though children acquired the formation of wh- questions by age 4, they still produced errors. First, this study indicates that there was frequent omission of the NP in D-linked subject and object questions. According to Stavrakaki, these errors may have been the consequence of the phonological similarity between pjos ‘which’ and pjos ‘who’. Second, SLI children showed a tendency of converting a non-D-linked question into D-linked question or a non-D-linked who-object question into a D-linked which-object question. Most importantly, this study reports gap-filling errors2, which are characterized by splitting of the wh-phrase and the NP (3). Stavrakaki concludes that the interpretation of D-linked questions requires the discourse linking with the NP and the costly simultaneous participation of syntactic and discourse-relevant operations (Avrutin 2000).

(3) (SLI response) *O andras pion htipise ton piθiko? the.NOM man.NOM which hit.3SG the.ACC monkey.ACC ‘Which monkey did the man hit? (Stavrakaki 2006 p. 390)

Van Kampen (1994, 1996 and subsequent work) argues for a PF/LF discrepancy in child language (see van Kampen 1996 for a detailed discussion) when children produce this kind of wh-subextraction errors in (3). In her analysis, X’ raising is triggered by morphological greed or by a PF adjacency condition. This kind of movement is proposed to have a direct link with the satisfying of any PF needs as in (4).

(4) *welke wil jij [twh liedje] zingen? (Sarah 3;7) which want you song sing?

‘[Welk liedje]DP wil jij tDP zingen?’ (Adult/later child Dutch) ‘Which song want you sing?’ (van Kampen 1996)

2 For consistency matters, we refer to this term as wh-subextraction.

Language Acquisition 3 5 1 Major Trends in Theoretical and Applied Linguistics

It should be noted that Nomura and Himoru (2005) showed in their study with 15 Japanese-speaking children (4;4-5;2) that, unlike Dutch (van Kampen 1997) and English (Chen et al. 1998), their participants did not violate the Pied- piping condition (Ross 1967). Catalan (Gavarró & Solà 2004a; Gavarró & Solà 2004b) is another language showing wh-subextraction errors in the acquisition of pied-piping in D-linked questions. Gavarró & Solà (2004) argue that errors, such as sub-extractions in child speech are explained by Kayne’s (2002) remnant movement, which is determined by Case requirements. Last, Roeper and Perez-Leroux (1997) discuss the interpretation of questions by children (Schaeffer 1991) expressing lack of movement of the NP in D-linked questions. For example, wh-subextraction errors appear in wh-possessor questions (also see Gavruseva & Thornton 2001; Thornton & Gavruseva 1996) as in (5) below.

(5)* Whose did you see book ‘Whose book did you see?’ Wh [‘s book] (Roeper & Perez-Leroux 1997 p.16)

As they explain, “the morphological constituency is altered by the phonological creation of a single word whose” (p.16) and this causes the need for pied-piping. It appears that errors in the acquisition of pied-piping are met often in several child languages. To sum up, consider Table 1 below:

Table 1. Studies that report errors in the acquisition of D-linked questions

Asproudi 2011 Sub-extraction of whs SMG Stavrakaki 2006 Sub-extraction of whs, NP omission

Dutch van Kampen 1997 Sub-extraction of whs

Japanese Nomura & Himoru 2005 No errors

Catalan Gavarró & Solà 2004 Sub-extraction of whs

English Gavruseva & Thornton 2001 Sub-extraction of possessors

The specificities found in the acquisition of D-linked questions appear in different languages that follow different grammatical rules and different structures. We will not elaborate further on the differences between the languages presented above, as our focus is on Cypriot Greek. We turn in the next section to examine the syntactic distribution of questions in CG.

3 5 2 Language Acquisition Elena Papadopoulou and Natalia Pavlou

2. Wh-questions in CG

In this paper, we explore only one type of question, namely, the D-linked (or Referential) pco/inda/ti ‘which’ question ((6) below).

(6) Pco /Inda vivlio Θkyavazi o andras? which book.ACC reading.3SG the.NOM man.NOM ‘Which book is the man reading?’

With regard to CG question formation, this shows morphological resemblance to SMG with minor pragmatic-semantic and morpho-phonological differences (Newton 1972) as well as substantial formation differences with respect to the embu ‘is-it-that’ strategy and inda mbu3 ‘what/why’ wh-phrases. Inda4 ‘what/which’ is invariant in gender, number, and case and it is used either prenominally (‘what/which NP’) or pronominally (simple ‘what’). For a summary of the SMG and CG wh-phrases, consider the Table below:

Table 2. Wh-phrases in CG and SMG (Pavlou 2012)

SMG CG Meaning pios/pjos pcos ‚who’ ti ti/ inda mbu5 ‚what’ pu pu ‚where’ pote pote ‚when’ poso poso(n) ‚how much’ jati jati/ inda/ inda mbu ‚why’ pos pos/ indalo(i)s ‚how’ (apo pu) pothen ‚from where’

An example of inda ‘what’ in D-linked questions is given below. Pied-piping in D-linked questions is not optional and it is characterized by movement of the noun along with the operator (7).

3 Pavlou (2010a), contra to Papadopoulou (2013) argues that inda mbu has a complex syntax with inda being the wh-phrase and mbu being on C, following the discussion in Grohmann et al. (2006). 4 When inda ‘what’ is adjoined to embu ‘is-it-that’, resulting in indambu ‘what is-it-that’ (Papadopoulou 2013) or inda mbu (Pavlou 2010a) four other allomorphs are identified, namely innambu, nambu, tambu and ambu (Pavlou 2010a). 5 Inda mbu is sometimes treated as a single element, depending on the analysis assumed (see Pavlou 2010a; Papadopoulou 2013).

Language Acquisition 3 5 3 Major Trends in Theoretical and Applied Linguistics

(7) Inda milon troi ɪ kopela? which apple.ACC eating.3SG the.NOM woman.NOM ‘Which apple is the woman eating?’

Grohmann and Papadopoulou (2010) note that inda ‘why’ cannot remain in- situ and always need to be fronted (8a), but inda ‘what’ in a complex wh-phrase can be found in-situ (8b).

(8a) Inda vɪvlion θkiavazi o Nikos? which book.ACC reading.3SG the.NOM Nick.NOM ‘Which book is Nick reading?’ (8b) O Nikos θkiavazi inda vɪvlion? the.NOM Nick.NOM reading.3SG which book.ACC ‘Nick is reading which book?’

Split-DPs in complex wh-phrases, or better split wh-constructions, were allowed in Classical Greek as presented in Mathieu and Sitaridou (2005). At that time, wh-elements did not need to raise together with the relevant nominal. These structures appear with the use of tis ‘his/hers’, which was inflected for φ-features. Today, the possibility for split-constructions as shown in SMG appears only in wh-constructions that involve a possessor marked with genitive Case (Table 3) (also examples (9a) & (9b) below).

Table 3 Wh-phrases that allow(ed) Split-DPs (Pavlou 2012)

Classical Greek SMG CG Meaning

tis-tina-tis ti ti ‚what’

tinosPOSS tinosPOSS ‚whose’

Pianu/PjanuPOSS PcuPOSS ‘whose’

CG does not allow any split-DPs in wh-constructions except in the case that a possessor element is involved. Tinos ‘whose’ appears in Split wh-constructions today in both SMG and CG, but as Pavlou (2012) argues Greek Cypriot speakers disallow the possession reading, when there is a possibility of a second reading. The second reading is the one which specifies the possessor wh-phrase as the indirect object of a ditransitive verb. The availability of sub-extraction from a tinos-phrase, with no change in meaning can be seen in (9b) for SMG. When tinos is separated from to vivlio, as in (9b), the same interpretation is possible. Even though CG also employs tinos-phrases, when the reading in (10) is available, the reading corresponding to (9b) becomes unavailable.

3 5 4 Language Acquisition Elena Papadopoulou and Natalia Pavlou

(9a) Tinos to vɪvlio eferes? whose.GEN the.ACC book.ACC brought.2SG (9b) Tinos eferes to vɪvlio? whose.GEN brought.2SG the.ACC book.ACC ‘Whose book did you bring?’ (Horrocks and Stavrou 1987 p.89) (10a) Tinos to vɪvlio eferes? whose.GEN the.ACC book.ACC brought.2SG ‘Whose book did you bring” (10b) Tinos eferes to vɪvlio? to whom.ACC brought.2SG the.ACC book.ACC ‘To whom did you bring the book?’ (Pavlou 2012)

Cypriot Greek does not allow any Split-DPs in questions in adult speech, so children are expected not to show any patterns of split wh-constructions in their acquisition of questions. We will consider relevant cases in the first experiment discussed in the next section.

3. Syntactic Priming in CG

3.1. Syntactic Priming Experiment in Cypriot Greek

A Syntactic Priming Experiment in Cypriot Greek (SPE-CG) (Papadopoulou 2013) was conducted with a hundred three monolingual native speakers-children of CG, aged 2;8 - 6;5. All children participating in the experiment attended kindergartens around the area of Larnaca and Limassol district. After being randomly selected, they were distributed in three age groups (Table 4 below).

Table 4. SPE-CG participants

Age group Age range Number of participants Mean age Standard deviation

AG1 2;8–3;11 22 3;4 3 months

AG2 4;0–4;11 26 4;3 2 months

AG3 5;0–6;5 45 5;7 3 months

3.2. Material and Procedure

The experiment involved thirty sentences of a prime (P) and a target (T), with a different verb, agent and patient for each P and T. Three wh-words were tested,

Language Acquisition 3 5 5 Major Trends in Theoretical and Applied Linguistics

namely, pco ‘which’ D-linked, ti ‘what’ non D-linked and the dialectal element inda ‘which’ D-linked, each in 10 sentences. Test sentences were distributed across the topicalization or not of the subject and appearance of embu ‘is-it- that’ following (11) and (12) below respectively; which in accordance with the

appearance or not of embu ‘is-it-that’ result in four main conditions which were distributed between groups.

(11) O andras pco vivlio (embu) Өkyavazi? the.NOM man.NOM which book.ACC (is-it-that) reading.3SG ‘Which book is the man reading?’ (12) Pco vivlio (embu) Өkyavazi o andras? which book.ACC (is-it-that) reading.3SG the.NOM man.NOM ‘Which book is the man reading?’

3.3. Results

Due to full pushing for priming children did not deviate much from the expected priming questions, even at a very young age. Raw numbers are found in Table 5 below.

Table 5. SPE-CG raw numbers

Prime Wh + V + Subj Subj + Wh + V Wh + V + Subj + Wh Wh + V + Subj + Wh Target Ungrammatical Ungrammatical Subj + V Subj + V 135 170 AG1 2 1 133 - 88 81

155 200 AG2 4 6 151 - 28 168

345 330 AG3 1 - 334 - 10 320

As depicted in Figure (1) below children perform at almost 100% when they are primed with the word order ‘Wh + V + Subj’. AG3 performs the same irrespective of the word order condition provided (Figures 1 and 2). In contrast AG2 seems to perform slightly less (84%) when they are given the topicalized word order as prime ‘Subj + Wh + V’. The younger group (AG1) seems to have great difficulty with the topicalized word order condition. They (AG1) performed at 47% following the target word order and reversed to the non-topicalized word order at 52% of the cases.

3 5 6 Language Acquisition Elena Papadopoulou and Natalia Pavlou

Figure 1. SPE-CG Wh+V+Subj results

Figure 2. SPE-CG Subj+Wh+V results

Concentrating on specifi cities of syntactic movement in early speech failing to be primed by the topicalized word order condition suggests possible diffi culties to move the subject at a topic position for AG1 (see inter allies Papadopoulou 2013), Papadopoulou & Pavlou 2012). Results above do not refer to the priming

Language Acquisition 3 5 7 Major Trends in Theoretical and Applied Linguistics

effects or not ofembu ‘is-it-that’, since it was not highly primed and its priming –or not- was not related to syntactic movement difficulties, given that it is generated at Spec-CP hence, not moved at that position (see again Papadopoulou 2013 for an in depth investigation). Overall, children across all age groups made a few ungrammatical errors, failing to obey the primed word order (5.1% across age groups). The majority of the errors made referred to the inability to obey pied-piping conditions resulting in a split DP with the wh-word correctly moved to Spec-CP but, with the noun remaining VP-internally resulting in examples like (13) instead of providing the supposed target (14) below.6

(13) *Pco kaӨarizi ɪ kopela pcato? which cleaning.3SG the.NOM girl.NOM plate.ACC ‘Which plate is the girl cleaning?’ (14) Pco pcato kaӨarizi ɪ kopela? which plate.ACC cleaning.3SG the.NOM girl.NOM ‘Which plate is the girl cleaning?’

Children have also produced sentences with both the wh-object phrases and the overt object by moving on the one hand the wh-word to Spec, CP but at the same time pronouncing the NP in the VP as in (15)7 below rather than (16).

(15) *O andras ti aniγi ðoro? the.NOM man.NOM what opening.3SG present.ACC ‘Which present is the man opening?’ (16) O andras ti aniγi? the.NOM man.NOM what opening.3SG ‘What is the man opening?’

Even though these utterances are limited in number, a greater number of these errors are found in the Guess What Game discussed in the next section.

The Guess What Game Similarly to SPE-CG, data were collected from Greek Cypriot children who were attending either public or private kindergartens in Limassol, the southern town in Cyprus. Participants were distributed in four age groups, as summarized in Table 6.

6 2 children aged 3;1 failed to obey pied piping condition twice and 4 more, aged 5;05, 4;11, 4;0 and 3;7, failed to obey the condition but also changed the wh-word in 8 instances 7 2 children aged 4;11 and 4;2 at 3 instances have produced wh-questions similar to the example given where wh/object doubling is observed.

3 5 8 Language Acquisition Elena Papadopoulou and Natalia Pavlou

Table 6. GWG participants

Age group Age range Number of participants Mean age Standard deviation

AG1 3;0–3;11 19 3;7 3 months

AG2 4;0–4;11 22 4;7 3 months

AG3 5;0–5;11 22 5;5 3 months

AG4 6;0–6;11 18 6;2 1 month

3.4. Material and Procedure

The materials used were two puppets, a baby frog and a baby lion, so as to provide enthusiasm to the children. The procedure followed was the same for all children tested and each child was introduced to it individually. The researcher would introduce the child to the idea that they were going to play a game and the game was to collect chickens with baby lion and baby frog and see who can collect the most chicks. The one who would collect the most chicks would be the winner and would win a prize at the end. What the child needed to do was to ask a question about each picture shown. The child believes that s/he is competing with the puppets, but by manipulating the way that the puppets respond, the child always wins the game. The test had 24 items in six sections with each one investigating a different syntactic structure. In this paper, we will discuss the findings for block 4, which involved D-linked questions. Each set of test items was preceded by two warm up items. In the warm up items, the child simply copied the adult’s questions but then s/he was told that s/he must go ahead to ask the questions directly. The same scenario was repeated for each set of items. An example, as used in Block 4, is provided below:

Warm-up 2: Inda aftokinitaki krata ɪ korua? (Researcher) which car.ACC holding.3SG the girl.NOM ‘Which car is the girl holding?’

Puppet: En su milo esena. (To research.) not you.GEN talk.1SG you.ACC Pezo mono me mora. play.1SG only with children.ACC ‘I am not talking to you. I only play with children’.

Language Acquisition 3 5 9 Major Trends in Theoretical and Applied Linguistics

Researcher: ΘelIs na rotɪsɪs esi ton vatrahulin? (To child) want.2SG to ask.2SG you.NOM the.DET baby-frog.ACC ‘Do you want to ask the baby frog?’

Child: Inda aftokinitaki krata ɪ korua? (To puppet) which car.ACC holding.3SG the.NOM girl.NOM ‘Which car is the girl holding?’

Puppet: To kotʒino. (To child) ‘The red one’.

Researcher: Ate, rota ton ya. (To child) come on ask.2SG him.ACC for.PRE tutɪn tɪn fotoγrafi an. this.DEM the.DET photograph.ACC ‘Now, ask baby frog about this picture’.

Target 1: Inda ðoro aniɪ o andras? (Child) which present.ACC opening.3SG the.NOM man.NOM ‘Which present is the man opening?’ (Pavlou 2012)

The child then produced other 3 more questions and the researcher repeated the same procedure for the next blocks. 3.5. Results

A control group with 10 adults also participated in the experiment and as expected, provided the target responses with a high percentage.

Figure 3. Production of Object D-linked questions in CG by Adults

Adults mostly produced inda-questions, following the target responses, but a relatively low percentage (12%) responded with a non-target ti-question. Overall, the successful production of ‘which’ questions was relatively poor with raw numbers found in Table 7 below and summarized in Figure 4.

3 6 0 Language Acquisition Elena Papadopoulou and Natalia Pavlou

Table 7. Production of Object D-linked questions in CG

inda ‘what/which’ ti ‘what/which’ Declarative No Response Other

39 3 yr 16 4 17 4 35

50 4 yr 25 0 13 0 50

60 5 yr 8 0 20 2 58

47 6 yr 0 0 25 7 40

D-linked questions had the lowest percentages in comparison with the elicitation of the other wh-questions in the experiment.

Figure 4. Production of D-linked questions: Overall results

Children performed very poorly in the successful production of target questions and showed a substantial preference for the CG-like wh-phrase ti ‘what’. Very low percentages were observed for the production of questions with inda ‘which’ and this appears in the youngest group (3 yr) and the older groups (5 yr & 6 yr). The percentages shown in Figure 4 can be sub-divided into further categories as other sub-types were observed. These percentages do not just show the successful production of the pied-piped structure with a wh-phrase, but also production of a question with omission of the noun and ungrammatical questions characterized by lack of movement of the noun.

Language Acquisition 3 6 1 Major Trends in Theoretical and Applied Linguistics

Figure 5. Sub-types of responses to D-linked questions

Figure 5 corresponds to the overall question production with ti ‘which’. Note that the successful pied-piping in wh-questions decreases by age. Very high percentages were observed with regard to the omission of NP from the D-linked question. In addition, errors were also observed showing the predicted lack of movement of the noun phrase and sole movement of the operator. This kind of error appeared with both a stranded NP and a determiner + NP sequence, but most importantly it also appeared in complex wh-phrases of the type ‘what colour car’, as in (17).

(17) *Ti aniγɪ kutin o andras? which opening.3SG box.ACC the.NOM man.NOM ‘Which box is the man opening?’

(18) *Ti hroma krata aftokinitaki o andras? which colour.ACC holding.3SG car.ACC the.NOM man.NOM ‘What is the colour of the car that the man is holding?’

Figure 6 below shows responses from children when attempting to produce a question with inda ‘which’. The breakdown of response patterns can be seen to diff er substantially from the pattern seen in Figure 5.

3 6 2 Language Acquisition Elena Papadopoulou and Natalia Pavlou

Figure 6. Successful pied-piping and errors with inda ‘which’ in Block 4

Even though there was no successful production of questions with the use of the Cypriot-specifi c inda ‘which’ in the 3 yr and 4 yr groups, the limited utterances of inda in 5 yr olds and 6 yr olds (see (19) and (20) below) show that the children performed at ceiling in any attempt made. The children exhibited target pied-piping of an NP with inda ‘which’ and formation of a D-linked wh- question.

(19) *Inda fori o andras kaphelo? which wearing.3SG the.NOM man.NOM hat.ACC ‘Which hat is the man wearing?’

(20) *Inda hroma krata aftokinitakin o andras? which colour.ACC holding.3SG car.ACC the.NOM man.NOM ‘What is the colour of the car that the man is holding?’

Errors in this case appear only in the 3 yr old group, which is the youngest group and would expectedly show the greatest frequency of errors for a late-acquired structure.

4. Discussion

The types of errors produced by children and explored here involve a logical explanation under which fundamental notions, such as Economy, are expressed through diff erent structures. Based on the data taken from the two experiments discussed above, children’s errors in D-linked questions appear

Language Acquisition 3 6 3 Major Trends in Theoretical and Applied Linguistics

in similar ages and are not affected by any specific methodology. It is evident that full pushing for priming minimized errors but it did not prevent them. Our analysis supports that the errors are not speech errors, but innately-motivated patterns that follow a theoretical reasoning in syntax. Two experiments were discussed in order to outline similar patterns appearing in children production of D-linked questions. Children participating in a syntactic priming experiment produced a specific number of utterances showing sub-extraction of the wh- phrase from a complex wh-phrase. A similar type of errors was also found in an elicitation game targeting production of wh-questions in the same variety. In the same experiment, wh-subextraction was also found in strings of the type ‘which colour car’ by having the wh-phrase and the following NP moving the beginning of the sentence and stranding the second following NP. This contrastive study was aiming to show that the actually methodology does not play a very significant role, when a specific path of acquisition of a structure is involved. We have seen that errors appeared even when the participants were fully primed and should not have produced these. We should note, however, that there is a substantial difference in the number of errors appearing between the two experiments. This type of errors is found cross-linguistically without following a strict path of acquisition. Several accounts have been proposed in order to account for these errors, but there was no agreement in the literature upon the phenomenon. As presented in Section 2, van Kampen (1994 and subsequent work) argues that there is a PF/LF discrepancy, whereas Gavarró & Solà (2004) argue that there is involvement of remnant movement. For Greek, Stavrakaki (2006) has argued that the errors are a result of a phonological similarity of the wh-word with the wh-object question, but as these errors have appeared in other languages as well, this explanation cannot be valid. Asproudi (2011) refers to the morphological richness of the language, which can be a possibility since both Greek and Cypriot Greek are morphologically rich languages. This still however does not explain why these error patterns should occur. These particular errors in child speech provide strong arguments for the Economy in language as well as its different applications on a theory of grammar. Simplicity, in other words, in language acquisition lies at the core of generative theory, as we try to represent a system which requires the minimum effort or procedure in all aspects of language. We adopt the Immediate Move Hypothesis (Pavlou 2012), as outlined in (21), based on two crucial characteristics of the errors presented. Children move as little as they can in their production of D-linked questions, but as much as they need. This analysis predicts that when children produce these errors, C attracts the goal as follows:

3 6 4 Language Acquisition Elena Papadopoulou and Natalia Pavlou

(21) Immediate Move Hypothesis Move α iff: a) α carries the target feature b) α is immediately contained within the nearest to the probe maximal projection containing the target feature c) No β is contained in α such that β immediately contains the target feature d) If α forms an XP, then it must immediately contain the target feature (Pavlou 2012)

In languages that do not allow split-DPs, C attracts as little as it can, while at the same time it is satisfying with it any of its needs, which are the uninterpretable features in C. In children’s syntax, both conditions above apply separately. While they could move the first DPMAX that contains the relevant features to satisfy the condition ‘Move as little as you can’, they also apply ‘Move as much as you need’ and therefore ignore the presence of the shortest (in distance) outer DPMAX and move only the internal DPMAX as in (22) below where only the wh-word is moved.

(22) *Inda fori o andras kapelo? which wearing.3SG the man.NOM hat.ACC ‘Which hat is the man wearing?’

To sum up, Immediate Move Hypothesis was proposed to account for sub- extraction phenomena in D-linked questions based on data from Cypriot Greek and other environments of similar type. To conclude, any generalizations defining these errors as speech errors, and not innately-motivated patterns, as suggested by Nomura and Hirotsu (2005), are not validated.

Language Acquisition 3 6 5 Major Trends in Theoretical and Applied Linguistics

References

Asproudi, E. 2011. Non-Target Long-Distance Wh-Questions. Crosslinguistic Typological Distinctions in Early L1 Production. Proceedings of the 20th International Symposium on Theoretical and Applied Linguistics (ISTAL 20). Aristotle University of Thessaloniki, Thessaloniki 1-3 April.

Avrutin, S. 2000. Comprehension of Wh-questions by children and Broca’s aphasics. In Y. Grodzinsky, L.P. Shapiro and D.A. Swinney, eds. Language and the brain: Representation and processing, San Diego: Academic Press, 295-312.

Butler, A. and E. Mathieu. 2005. Split-DPs, generalized EPP, and visibility. MIT Working Papers in Linguistics, 49, 49-57.

Cable, S. 2008. There is no such thing as Pied-Piping. Ms., MIT.

Chen, D., M. Yamane and W. Snyder. 1998. Children’s left-branch violations: Evidence for a non-parametric account. 22nd Annual Boston University Conference on Language Development, Somerville, MA: Cascadilla Press.

Fanselow, G. and D. Cavar. 2002. Distributed Deletion. In A. Alexiadou, ed. Theoretical Approaches to Universals. Amsterdam, John Benjamins, 65-107.

Gavarró, A. and J. Solà. 2004a. Wh-subextraction in child Catalan. Second Lisbon Meeting on Language Acquisition, Universidade de Lisboa, 1 June.

Gavarró, A. and J. Solà. 2004b. Subextraction in Romance interrogatives. Going Romance 2004, Leiden University, 12 April.

Gavruseva, E. and R. Thornton. 2001. Getting it right: acquisition of whose- questions in child English. Language Acquisition, 9(3), 229-267.

3 6 6 Language Acquisition Elena Papadopoulou and Natalia Pavlou

Grohman, K.K. 2011. Some Directions for the Systematic Investigation of the Acquisition of Cypriot Greek: A New Perspective on Production Abilities from Object Clitic Placement. In E. Rinke and T. Kupisch, eds. The Development of Grammar: Language Acquisition and Diachronic Change — Volume in Honor of Jürgen M. Meisel. (Hamburg Series on Multilingualism 11.) Amsterdam: John Benjamins, 179-203.

Grohmann, K.K., P. Panagiotidis and S. Tsiplakou . 2006. Properties of Wh-Question Formation in Cypriot Greek. 2nd International Conference on Modern Greek Dialects and Linguistic Theory, University of Patras, Mytilene, 30 September-3 October 2004.

Grohmann, K.K. and E. Papadopoulou. to appear. Question(able) Issues in Cypriot Greek. Linguistic Analysis, 37, 1-31.

Heck, F. 2008. On Pied-piping. Wh-movement and Beyond. Berlin: Mouton de Gruyter.

Horrocks, G. and M. Stavrou. 1987. Bounding Theory and Greek syntax: Evidence for wh-movement in NP. Journal of Linguistics. 23, 79-108.

Kayne, R. 2002. On some preposition that look like DP internal: English ‘of’ and French ‘de’, Ms., NYU.

Mathieu, E. 2002. The Syntax of Non-Canonical Quantification: A comparative Study. University College London: Doctoral Dissertation.

Mathieu, E. and I. Sitaridou. 2005. Split WH-constructions in Classical and Modern Greek: A diachronic perspective. In M. Batllori, M-L. Hermanz, C. Picallo and F. Roca, eds. Grammaticalization and parametric change. Oxford Scholarship Online, Oxford, 236-250.

Newton, B. 1972. Cypriot Greek: Its Phonology and Inflections. The Hague: Mouton.

Nomura, M. and K. Hirotsu. 2005. The Left Branch Condition in the Acquisition of Japanese. In M. Nomura, F. Niinuma and L. Reglero, eds. University of Connecticut working papers in linguistics, 13, 119-144.

Papadopoulou, E. 2013. The acquisition of wh-questions: Evidence from Cypriot Greek. Phd Dissertation, University of Essex.

Papadopoulou, E. and N. Pavlou. 2012. What I say, you say! Illustration of syntactic priming in Cypriot Greek. In K.K. Grohmann, A. Shelkovaya, and D. Zoumbalides

Language Acquisition 3 6 7 Major Trends in Theoretical and Applied Linguistics

(eds.). Linguists of Tomorrow: Selected Papers from the 1st Cyprus Postgraduate Conference in Theoretical and Applied Linguistics. Newcastle upon Tyne: Cambridge Scholars Publishing. [expected publication: late 2011/early 2012].

Pavlou, N. 2010a. Mbu! On wh-objects and true adjuncts of Cypriot Greek. 4th International Conference on Modern Greek Dialects and Linguistic Theory, University of Patras, Patra, 11-14 June 2009.

Pavlou, N. 2010b. Inda mbu, nambu or ine ti pu? Acquiring Complex Structures in Cypriot Greek. The ISCA Tutorial and Research Workshop on Experimental Linguistics, ISCA & University of Athens, Athens, 25-27 August.

Pavlou, N. 2012. Pied-piping in wh-questions. What do children say about it? MA dissertation, University of York.

Roeper, T. and A. Pérez-Leroux. 1997. The interpretation of bare nouns in semantics and syntax: inherent possessive, pied piping, and root infinitives. MIT Occasional Papers in Linguistics 12. MA: MIT Press.

Ross, J. 1967. Constraints on variables in syntax. Doctoral dissertation, MIT

Schaeffer, J. 1991. The Italian child C-system. Ms. University of Venezia.

Stavrakaki S. 2006. Developmental perspectives on Specific Language Impairment: Evidence from the production of wh-questions by Greek SLI children over time. Advances in Speech-Language Pathology, 8, 384-396.

Thornton, R. and E. Gavruseva. 1996. Children’s split “Whose-questions” and the structure of possessive NPs. Unpublished paper presented at the 21st Annual Boston University Conference on Language Development, Boston, United States.

van Kampen, J. 1994. The learnability of the Left Branch Condition. In R. Bok- Bennema and C. Cremers, eds. Linguistics in the Netherlands 1994. 83-94. Amsterdam: John Benjamins.

van Kampen, J. 1996. PF/LF convergence in acquisition. In K. Kusumoto, ed. Proceedings of the NELS 26, 149-163.

van Kampen, J. 1997. First Steps in Wh-movement. Delft: Eburon.

3 6 8 Language Acquisition Pronominal Subjects in English L2 Acquisition and in L1 Greek: Issues of Interpretation, Use and L1 Transfer

Alexandra Prentza

Aristotle University of Thessaloniki [email protected]

Abstract

This study examines the L2 acquisition of English pronominal subjects by Greek learners and its possible causes. It also considers the factors that affect pronoun resolution in L1 Greek. Null pronominal subjects were found to be used even by advanced learners, especially in topic-continuity contexts. This is attributed to the transfer of the L1 property of null subjects and to the interpretive effects thereby achieved. Regarding pronoun resolution in Greek, we found that the syntactic constraint related to the null/overt realization of pronominal subjects is such a decisive factor that its effect surfaces even when antecedents are pragmatically inferred.

1. Theoretical Background

1.1. Introduction

Optionality in phenomena at the syntax-discourse interface has been the focus of much recent research. L2 inconsistent performance has been accounted for in two ways: narrow syntax vs. the interfaces. Interface Vulnerability Hypotheses (Hulk & Muller 2000; Sorace & Filiaci 2006; Belletti et al. 2007; Sorace & Serratrice 2009) maintain that formal features are acquirable in L2 acquisition but the interfaces and the related interpretable features are not. This implies that cross-linguistic influence may not be relevant to any learnability problems. On the other hand, Formal Features Deficit Accounts (Smith & Tsimpli 1995; Hawkins & Hattori 2006) like the Interpretability Hypothesis (Tsimpli & Dimitrakopoulou 2007) attribute L2 variability to a deficit in uninterpretable features which are

Language Acquisition 3 6 9 Major Trends in Theoretical and Applied Linguistics

assumed to remain inaccessible, unlike interpretable features that are more readily available. Here, the implication is that L1 transfer may be associated with L2 instability. Previous L2 findings on pronoun interpretation have reported residual optionality manifested in two ways: first, learners were found to overuse overt pronominal subjects in topic-maintenance contexts (Bini 1993; Pérez-Leroux & Glass 1999; Margaza & Bel 2006; Sorace & Filiaci 2006; Belletti et al. 2007; Lozano 2008) and, second, to a much lesser extent, they inaccurately used null pronominal subjects in topic-shift contexts (Monrtul & Louro 2006; Lozano 2008, 2009; Jegerski, et al. 2011). By contrast, concerning production, some studies have reported that speakers of non null-subject languages (NNSLs) learning a null-subject language (NSL) produce null pronominal subjects at rates comparable to those of native speakers (Liceras Diaz & Maxwell 1999; Belletti et al. 2007). Studies on pronoun resolution in Greek have shown that speakers interpret null subject pronouns as anaphoric to the first-clause subject and overt subject pronouns as anaphoric to the first-clause object (Dimitriadis 1996; Tsimpli et al. 2003, 2004; Miltsakaki 2007; Papadopoulou et al. 2007). Similar findings have also been reported for other NSLs (Carminati 2002, 2005 for Italian, Mayol 2008; Jegerski, et al. 2011 for Spanish)1. For this study, a replication of such approaches is important for two reasons: first, it can show that syntactic issues are fundamental to the resolution of pronominal ambiguity in Greek and second, it will provide strong evidence that null pronoun use in L2 English reflects the transfer of the L1 property of subject drop in topic-continuity contexts.

1.2. Pronominal Subjects: Licensing and Distribution

It is well known that NSLs like Greek feature null lexical and pronominal subjects in matrix and subordinate clauses, a possibility not available for English, as (1) demonstrates:

(1) pro ipe oti pro tha mas tilefonisi Said.3SG that will us call.3SG *Said that will call us ‘He/She said that he/she will call us’

1 However, other studies have shown that overt subject pronouns have a more flexible procedure of antecedent assignment (Alonso-Ovalle et al. 2002; Geber 2006; Meridor, 2006). The discussion of this issue is beyond the scope of this paper.

3 7 0 Language Acquisition Alexandra Prentza

Within minimalism these differences have been accounted for as differences in the way the uninterpretable phi-features of T is valued. In English the phi- features of T are valued by Move/Merge of an overt subject/expletive in the Spec,TP. Conversely, in Greek, the valuation proceeds through verbal agreement morphology, which is assumed to be nominal in nature. Therefore, the SpecTP/ IP position need not be occupied by an overt subject (Barbosa 1995, 2009; Alexiadou & Anagnostopoulou 1998; Spyropoulos & Philippaki-Warburton 2001; Roussou & Tsimpli 2006). According to minimalist assumptions on cross- linguistic variation (Chomsky 1995, 2001), this difference between Greek and English amounts to a divergence in the feature specification of the functional head(s) associated with subject-verb agreement or, in other words, to a difference with uninterpretable features. The distribution of null and overt subjects in NSLs is assumed to be regulated by the interpretable [topic-shift] feature: null pronouns denote topic-maintenance while overt pronouns topic-shift (Cardinaletti & Starke 2001; Carminati 2002; Tsimpli et al. 2003, 2004). In English overt pronouns are obligatory and thus discursively unmarked. See (2-4):

(2) -Milise o Janis stin Anna? Talked.3SG the.NOM Janis.NOM to.ACC Anna.ACC ? ‘Did Janis talk to Anna?’ (3a) -Ne, pro tis milise (topic-continuity) yes, pro her talked.3SG (3b) -??Ne, aftos tis milise (topic-shift) yes, he.NOM her talked.3SG (4) -Did John talk to Anna? -Yes, he did (a&b)

In NSLs, subject pronoun interpretation with competing antecedents has been argued to depend on the null/overt realization of the pronoun to be referenced: whereas for null pronouns the preferred antecedent is in subject position, for overt pronouns it is in object position (Carminati 2002 for Italian, Tsimpli et al. 2004; Miltsakaki 2007; Papadopoulou et al., 2007 for Greek). See (5):

(5a) Ενώ η Κατερίνα μιλούσε στην κυρία pro σκόνταψε σε μια πέτρα

(5b) Ενώ η Κατερίνα μιλούσε στην κυρία αυτή σκόνταψε σε μια πέτρα

(5c) While Katerina-i was talking to the lady-k, pro-i/she-k tripped over a stone

Besides the factor of Pronoun Type (PT) (null/overt realization of pronominal subject), which is related to the exploitation of a parametric option available in

Language Acquisition 3 7 1 Major Trends in Theoretical and Applied Linguistics

Greek, our study considers the variable of World Knowledge Inference (WKI), i.e. whether the choice of an antecedent is inferred based on world knowledge or not. Our aim was to explore how Greek speakers resolve the ambiguity in contexts where both variables are incorporated; whether they proceed based on syntax and the interpretive distinctions encoded by parametric choices or based on the pragmatics of a sentence and the reference assignment thereby dictated. Endorsing the view that uninterpretable features are related to prolonged acquisitional problems, we predict that in the English tests i) Greek learners will accept/produce null pronominal subjects even in advanced stages of proficiency and ii) Greek learners but not the English natives will accept/produce more null pronominal subjects when these refer to the first-clause2 subject than when they are referentially disjoint from it transferring an L1 property. In the Greek pronoun resolution test we expect speakers to assign null pronouns to subject antecedents and overt pronouns to object antecedents.

2. The Experiment

2.1. Subjects and Materials

Seventy-two Greek learners and a control group of twenty-five English native speakers (NS) took the English tests while sixty-seven Greek speakers took the Greek test. The Greek learners of the first experiment were tested by the Oxford Quick Placement Test (2001) and divided into thirty-five intermediate (INT) and thirty-seven advanced (ADV) learners. They were administered one judgement and two production tasks: a Paced Grammaticality Judgement Task (PGJT), a Sentence Completion Task (SCT) and a Cloze Test (CT). The PGJT aimed to test the acceptability of null and overt pronominal subjects in English subordinate clauses. It consisted of ten items in which the subject pronoun referred to the first-clause subject (Coreferential condition, henceforth, CoR) and of ten items in which the subject pronoun was referentially disjoint from it (Disjoint Reference condition, henceforth, DjR). This categorization is related to the Pronoun Reference Type (PRT) variable. In each type, half of the items were ungrammatical, while the other half were grammatical. Sentences in

2 We use this term because the linearly first clause is either matrix or subordinate in the tasks used.

3 7 2 Language Acquisition Alexandra Prentza

(6) provide examples:

(6a) *We won’t finish on time if pro don’t start now (CoR) Mary was angry because she had lost her job. (6b) *It is so hot in here that pro sweat all the time (DjR) The movie was so boring that Mary fell asleep.

Participants had to indicate their judgements using a -2 to +2 scale as follows: -2 would be given to an ungrammatical sentence, +2 to a grammatical sentence and 0 if they thought that a sentence had equal chances of being grammatical or ungrammatical. Zero responses were not eliminated in the statistical analysis since it was decided that in this study uncertainty in judgement would be exploited. Participants were also given choices -1 and +1. They ought to use them when they were not as sure about the (un)grammaticality of a sentence as they were when they used the extremes of the scale. For reasons of presentation, during data inputting this scale was matched to a 1-5 scale. In this way, for both grammatical and ungrammatical items the scale categories encoded the same choices regarding accuracy and certainty in judgement. See (7):

(7) 5: accurate-categorical (-2 for ungram., +2 gram.) 4: accurate-non-categorical (-1 for ungram., +1 for gram.) 3: inaccurate (0 for ungram./ gram.) 2: inaccurate-non-categorical (+1 for ungram., -1 for gram.) 1: inaccurate-categorical (+2 for ungram., -2 for gram.)

The subjects read on a screen and at the same time listened to the test items. Each item had to be judged during a 5-second gap. The SCT tested the production of null and overt subject pronouns. It consisted of sixteen items in which the targeted pronoun referred to the first-clause subject and of sixteen items in which it did not. Participants had to conjoin the sets of clauses presented to them. Only the predicate of the second clause was given. See (8) (anticipated responses in italics):

(8a) When she heard the news [tell/ all friends] (CoR) When she heard the news she told all her friends (8b) We can’t pay you because [you/ not complete/ the work] (DjR) We can’t pay you because you haven’t completed the work yet

Performance was analyzed in terms of target and non-target responses which were related to pronominal subject use and omission respectively. The CT also tested the production of null and overt pronominal subjects, yet in a less restricted context. Participants were given ten short passages including

Language Acquisition 3 7 3 Major Trends in Theoretical and Applied Linguistics

fourteen test items and sixty-two fillers. In eight of the test items the targeted response was a pronoun coreferential with the first-clause subject while in the remaining six the pronoun was disjoint in reference. See (9) (anticipated response in italics):

(9a) They hold one’s interest…., because …… they ……… are carefully written. (CoR) (9b) Hunters sell …..their….skin because … it …. is very expensive’ (DjR)

Target and non-target responses corresponded to overt pronominal subject production vs. omission respectively. In the Greek test participants had to indicate the antecedent of the pronominal subject of the second clause choosing from a set of three options: the first- clause3 subject, the first-clause object or either. The test consisted of fifteen items in which the response was not pragmatically biased and of sixteen items in which it was. This categorization is associated with the variable of WKI (-/+ WKI). In half of the +WKI items the felicitous response was the subject, whereas in the other half it was the object. Moreover, in 8 items of the +WKI type and in eight items of the -WKI type the pronoun to be referenced was overt, while in the remaining items in both types the pronoun was null. This distinction is related to the variable of Pronoun Type (PT). Sentences in (10) exemplify –WKI items. Sentences in (11) and (12) illustrate +WKI items:

(10a) Καθώς η Κατερίνα μιλούσε στην κυρία, αυτή σκόνταψε σε μια πέτρα. While Katerina was talking to the lady, she tripped over a stone (10b) Η Ζωή μίλησε με την αδερφή της, αλλά pro δε δέχτηκε να συμφωνήσει. *Zoe talked to her sister but pro didn’t agree to consent

Subject Biased Items (11a) Καθώς ο γιατρός μιλούσε με τον ασθενή, αυτός συνταγογραφούσε. While the doctor was talking to the patient, he was making up the prescription (11b) O ένοικος παραπονέθηκε στον αστυνομικό για τη φασαρία, αλλά pro δεν ήθελε να κάνει μήνυση. *The tenant complained to the policeman about the noise but pro didn’t want to press any charges

3 All items were instances of forward anaphora, but the linearly first clause could be either matrix or subordinate.

3 7 4 Language Acquisition Alexandra Prentza

Object Biased Items (12a) H κοπέλα απολογήθηκε στη μητέρα της και αυτή έδειξε κατανόηση. The girl apologized to her mother and she was understanding (12b) Η Λίνα έφτιαχνε σούπα στην αδερφή της, επειδή pro ήταν άρρωστη *Lina was cooking soup for her sister because pro was sick

The statistical analysis was conducted as follows: for the PGJT, a random effects factorial GLM (ANOVA) approach was used to assess factors with a significant effect. This approach contributed to accounting for repeated responses. Tukey HSD post-hoc tests were conducted to test for between and within-group differences. For the SCT and the CT, analyses of main effects and interactions were performed through a logistic regression model. In the Greek test, chi- square tests were applied to test for independence between the variables involved. Multinomial logistic regression analyses were applied in order to detect significant differences on the response

2.2. Results 2.2.1. PGJT

Multiple 3 x 2 x 2 analyses yielded a highly significant main effect of Group (F2,376

= 114.98, p < 0.001) and Grammaticality (F1,376 = 143.01, p < 0.001), as well as

a main effect of PRT (F1,376= 13.32, p < 0.05). Group was found to interact with

Grammaticality in a highly significant way (F1,376 = 47.44, p < 0.001) and with

PRT in a significant way (F2,376 = 4.17, p < 0.05). The interaction of Group x PRT

x Grammaticality was not significant (F2,376 = 1.47, p > 0.05). Figure 1 illustrates group overall accuracy rates:

5 4,92 4,6 4,46 4,73 4 4,16

3 2,92 GRAM 2 UNGRAM Accuracy Accuracy 1 0 INT ADV NS

Figure 1. Overall Accuracy Rates

Language Acquisition 3 7 5 Major Trends in Theoretical and Applied Linguistics

Within-group comparisons showed a grammaticality effect but only for L2 groups (INT, ADV: p < 0.001, NS: p > 0.05). Between-group comparisons in the ungrammatical sentences revealed that all groups differ significantly from one another (INT vs. NS, INT vs. ADV, ADV vs. NS: p < 0.001). In the grammatical sentences the INT and the ADV were significantly less accurate than the NS (INT vs. NS: p < 0.001, ADV vs. NS: p < 0.05) but no differences were found between the two L2 groups (INT vs. ADV: p > 0.05). The scores in the ungrammatical items were further analyzed by the variable of PRT4 and are presented in Figure 2.

5 4,72 4,74 4,33 4 3,97 INT 3 3,26 2,58 ADV 2 Accuracy NS 1 0 CoR DjR

Figure 2. Rejection Scores of null CoR vs. null DjR items

Within-group comparisons revealed that the INT were significantly more accurate in rejecting null pronoun items of the disjoint than of the coreferential type (p < 0.001). The ADV rejected more readily ungrammatical disjoint than ungrammatical coreferential items but this difference did not reach significance (p > 0.05). Native performance did not differ in the two conditions. As for between-group comparisons, all groups differed from one another in the coreferential condition, and, crucially the ADV from the NS, as predicted (INT vs. NS, INT vs. ADV, ADV vs. NS: all ps < 0.001). In the disjoint condition, as anticipated, there is an improvement since, although the INT are significantly less accurate than the controls (p < 0.001), the ADV perform native-like (p > 0.05).

4 We are allowed to do so, since the interactions between Group and Grammaticality and Group and PRT were significant.

3 7 6 Language Acquisition Alexandra Prentza

2.2.2. SCT

A logistic regression analysis returned a main effect of Group in both reference conditions (CoR: Wald test = 53.333, p < 0.001, DjR: Wald test = 21.230, p < 0.001). This was further explored through between-group comparisons on pronominal subject use rates organized by reference type (see Figure 3):

100 99,49 100 100 80 92,5 88,5 CoR 60 69,29 40 DjR 20 0 INT ADV NS

Figure 3. Pronominal Subject Use: CoR vs. DjR

In the coreferential condition there were across group differences. As predicted, the ADV produced significantly fewer overt subject pronouns than the NS group (all ps < 0.01). By contrast, in the disjoint condition, the ADV were as accurate as the controls exhibiting improved performance (INT vs. ADV: p < 0.05, INT vs. NS: p < 0.01, ADV vs. NS: p > 0.05). Additional comparisons between coreferential and disjoint items in each group revealed that the INT and, crucially, the ADV produce more null pronominal subjects when these refer to the first-clause subject than when they are referentially disjoint from it, as anticipated (both ps < 0.0001). No similar difference was detected in the NS group (p > 0.05).

2.2.3. CT

The logistic regression analysis indicated a highly significant main effect of Group (Wald test = 77.200, p < 0.001) and a main effect of PRT (Wald test = 19.152, p < 0.01). However, no significant interaction was detected between Group and PRT (Wald test = 0.323, p > 0.05). Figure 4 demonstrates overall pronominal subject use rates. The INT group produced null subject pronouns at a significantly higher rate than the ADV (p < 0.001) and the NS group (p < 0.001). As predicted, the ADV also performed significantly less accurately than the NS group (p < 0.05). Next, group performance was organized by PRT. See Figure 5:

Language Acquisition 3 7 7 Major Trends in Theoretical and Applied Linguistics

100 99 80 90,97 60 79,57 40 20 0 INT ADV NS

Figure 4. Pronominal Subject Use Overall Scores

100 80 93,29 99,5 98,67 84,29 87,5 CoR 60 72,5 40 DjR 20 0 INT ADV NS

Figure 5. Pronominal Subject Use: CoR vs. DjR

These scores reveal that the L2 groups produce more null subject pronouns in cases of coreference than of disjoint reference. However, this tendency did not reach significance (both ps > 0.05). In contrast, the controls produce overt subject pronouns irrespective of reference type (p > 0.05).

2.2.4. Greek Test

First, the results of the –WKI items of the test will be presented. See Figure 6. The statistical analysis showed that there is a main effect of the PT variable 2 (i.e. overt /null realization of pronoun) on the response (χ = 206.481, df = 2, p < 0.0001). When the subject pronoun is null, participants significantly favor the first-clause subject over the ‘object’ and ‘either’ options. By contrast, when the subject pronoun is overt, the object antecedent is preferred over the ‘subject’ and ‘either’ options at a significant rate. These findings show that in the absence of other factors Greek speakers resolve pronominal ambiguity based on the semantic properties related to the parametric option of null subjects in Greek; null pronouns are assigned to the most prominent antecedent, the first-clause subject, while overt pronouns are assigned to the first-clause object, as expected.

3 7 8 Language Acquisition Alexandra Prentza

100% 16,04 7,46 80% 22,76 EITHER 60% 67,8 OBJECT 40% SUBJECT 61,19 20% 24,73

PreferredAntecedent 0% NULL PRONOUN OVERT PRONOUN

Figure 6. Responses in –WKI items

Next we will present the results from the +WKI items. Figures 7 and 8 provide responses in the subject and object biased items of the test respectively:

98,13 100 84,33 80 SUBJECT 60 OBJECT 40 EITHER 20 12,69 0,37 1,99 2,99

PreferredAntecedent 0 NULL PRONOUN OVERT PRONOUN

Figure 7. Responses in Subject Biased Items

98,88 100 79,85 80 SUBJECT 60 OBJECT 40 EITHER 17,54 20 2,61 1,12 0

PreferredAntecedent 0 NULL PRONOUN OVERT PRONOUN

Figure 8. Responses in Object Biased Items

Language Acquisition 3 7 9 Major Trends in Theoretical and Applied Linguistics

The analysis showed a main effect of the variable of WKI on the response χ( 2 = 742.402, df = 2, p < 0.001). Greek speakers conform to the pragmatic constraints incorporated in the items and respond accordingly: in the subject biased items they overwhelmingly give a subject response (see Figure 7) while in the object biased items they clearly opt for the object antecedent (see Figure 8). Interestingly though, the analysis also indicated a main effect of the variable of PT on participant preferences (χ2 =26.374, df = 2, p < 0.001), as well as a significant interaction of the variables involved (Wald test = 231. 012, p < 0.001). Multinomial regression tests applied to further examine all the possible differences resulting from this interaction showed that the responses given in the subject biased items and in the object biased items differed significantly with respect to the ratio of overt/null items (Wald test = 5.716, p < 0.05). This means that despite the preponderance of the subject responses in the subject biased type and of the object responses in the object biased type, the type of pronominal subject (overt/null) inflicted statistically significant distinct preference rates: in the subject biased type the subject response was favored significantly more when the pronoun to be referenced was null than when it was overt (p < 0.05). Accordingly, in the object biased items the object response was opted for at a significantly higher rate when the pronoun to be referenced was overt than null (p < 0.05). These results are remarkable because they show that although the variable of PT may not override that of WKI, the syntactic constraint related to the occurrence of null and overt subject pronoun variants in Greek is such a decisive factor in pronoun resolution that its effect surfaces even when pragmatic bias is involved.

3. Discussion

Starting with the acquisition of English pronominal subjects, it was predicted that due to the involvement of parameterized formal features, the L1 syntactic option of null pronominal subjects will be active in L2 grammars even in advanced stages of proficiency. Overall scores from all tasks have verified this prediction: not only INT but ADV learners as well accept/produce significantly more null pronominal subjects than the controls. However, studies by Liceras Diaz & Maxwell (1999) and Belletti et al. (2007) report that the production of null pronominal subjects in the L2 acquisition of NSLs by speakers of NNSLs was not distinguishable from that of the controls. Two points can be made on that: First, the direction of acquisition (i.e. acquisition of a null form when the L1 has only overt forms or acquisition of an overt form when L1 has null and overt forms) may involve different levels of difficulty (see Parodi & Tsimpli, 2005). Second, these studies used only production tests, while our experiment used both judgement and production tasks, which may have affected results.

3 8 0 Language Acquisition Alexandra Prentza

Returning to our study, the obtained data allows us to assume that parameterized uninterpretable features related to the obligatorily overt manifestation of English subjects are inaccessible in adult L2 acquisition. For this reason, L2 learners transfer the L1 option of null pronominal subjects in developing and advanced L2 grammars. Based on that, it can be argued that accounts like the Interpretability Hypothesis which propose that uninterpretable features are not acquirable in adult L2 acquisition are supported by our data. Moving on to interpretation, it was hypothesized that L2 learners but not the NS will accept/produce more null pronominal subjects when these refer to the first-clause subject than when they are referentially disjoint from it, transferring an L1 property. Data from the PGJT and the SCT support this hypothesis. Findings from the CT attest only a tendency towards that direction. This could stem from the fact that the CT is a more demanding task as test items were presented in short texts the meaning of which participants had to grasp in order to complete the task. The case might be that participants were more reluctant to omit subject pronouns. Given the above, we can maintain that the parametric choice of null subjects which, as previously discussed, appears to be active in L2 grammars, seems to be exploited by Greek learners to encode interpretative effects in the interlanguage system exactly in the way implemented in L1: null pronominal subjects in topic- continuity, overt pronominal subjects in topic-shift. In other words, the semantic distinctions made possible by the syntax of pro-drop (i.e. null pronouns denoting topic-continuity) are enforced in interlanguage systems as L2 learners resort to L1 syntactic options in structures where uninterpretable features are involved. Consequently, it can be claimed that the transfer of L1 parameterized options in L2 acquisition could be responsible for the attested optionality in pronominal subject realization. Given that the interpretable feature of [topic-shift] exploits a parametric option, residual optionality in pronominal subject realization cannot be viewed in isolation from the consistent pattern of unsuccessful L2 performance evinced through the overall use of null pronominal subjects in the English tests. Therefore, the assumptions of Interface Vulnerability Hypotheses according to which variability at the syntax-discourse is unrelated with formal features but attributed to a problem with interpretable features are questioned by our findings. Turning to the Greek Test, results have shown that the contrasts in reference assignment related to the Pronoun Type variable are detected not only in the non-biased items, as expected, but also in the biased items, which is rather interesting. In the non-biased items, data on the assignment of null pronominal subjects to the subject of the first clause allow us to strongly claim that Greek learners drop subject pronouns in English subordinate clauses because they interpret them as coreferential with the first-clause, which shows transfer of an L1 strategy. In the biased items on the other hand, antecedent preferences were

Language Acquisition 3 8 1 Major Trends in Theoretical and Applied Linguistics

clearly determined by inferences based on world knowledge, yet null pronouns were found to favor more subject antecedents and overt pronouns to choose more object antecedents within the subtypes of subject and object biased items respectively. This shows that assignment of pronominal reference in Greek is resolutely affected by the syntactic constraint associated with the null/over manifestation of the pronoun, even in the presence of factors like pragmatic bias. On the whole, our data seem to suggest that syntax and the semantic distinctions encoded by parametric choices are crucial both in L2 pronominal subject interpretation and use, as well as in the L1 resolution of pronoun ambiguity in NSLs like Greek. This implies that in the examination of L2 pronominal use and interpretation the factor of cross-linguistic influence not only is relevant but, probably, should be considered first, over factors which have to do with the interface status of the phenomenon or the executive functions involved.

3 8 2 Language Acquisition Alexandra Prentza

References

Alexiadou, A. and E. Anagnostopoulou. 1998. Parametrizing agr: word order, V-movement and EPP-checking. Natural Language and Linguistic Theory 16(3): 491-539.

Alonso-Ovalle, L., S. Fernández-Solera, L. Frazier and C. Clifton. 2002. Null vs. Overt Pronouns and the Topic-Focus Articulation in Spanish. Rivista di Linguistica, 14(2): 151-169.

Barbosa, P. 1995. Null Subjects. Ph.D Thesis, MIT.

Barbosa, P. 2009. Two kinds of subject Pro. Studia Linguistica 63(1):2-58.

Belletti, A., E. Bennati and A. Sorace. 2007. Theoretical and developmental issues in the syntax of subjects: evidence from near-native Italian. Natural Language and Linguistic Theory 25(4): 657-689.

Bini, M. 1993. La adquisición del italiano: más allá de las propiedades sintácticas del parámetro pro-drop. In J. Liceras (ed.), La lingüística y el análisis de los sistemas no natives. Ottawa: Dovehouse, 129-139.

Cardinaletti, A. and M. Starke. 2001. The typology of structural deficiency. In H. van Riemsddijk (ed.), Clitics in the languages of Europe, Vol. 8 of Language Typology. Berlin: Mouton de Gruyter.

Carminati, M.-N. 2002. The Processing of Italian Subject Pronouns. Ph.D Thesis, University of Massachusetts Amherst.

Carminati, M.-N. 2005. Processing reflexes of the Feature Hierarchy (Person>Number>Gender) and implications for linguistic hierarchy. Lingua 115(3): 259-285.

Language Acquisition 3 8 3 Major Trends in Theoretical and Applied Linguistics

Chomsky, N. 1995. The Minimalist Program. Cambridge, MA:, MIT Press.

Chomsky, N. 2001. Derivation by phase. In M. Kenstowicz (ed.), Ken Hale: A life in language. Cambridge, MA: MIT Press.

Dimitriadis, A. 1996. When pro-drop languages don’t: Overt pronominal subjects and pragmatic inference. In Proceedings of the 32nd Meeting of the Chicago Linguistics Society. Chicago, 33-47.

Geber, D. 2006. Processing Subject Pronouns in Relation to Non-canonical (Quirky) Constructions. Ottawa Papers in Linguistics 34: 47-61.

Hawkins, R. and H. Hattori. 2006. Interpretation of English multiple wh-questions by Japanese speakers: a missing uninterpretable feature. Second Language Research 22(3): 269-301.

Hulk, A. and N. Müller. 2000. Bilingual first language acquisition at the interface between syntax and pragmatics. Bilingualism: Language and Cognition 3(3): 227–244.

Jegerski, J., B. Van Patten and G. Keating. 2011. Cross-linguistic variation and the acquisition of pronominal subject reference in L2 Spanish. Second Language Research 27(3): 482-507

Liceras, J., L. Diaz and D. Maxwell. 1999. Null arguments in non-native grammars: the Spanish L2 of Chinese, English, French, German, Japanese and Korean speakers. In E. Klein and G. Martohardjono (eds.) The development of second language grammars: A generative approach. Amsterdam: John Benjamins, 109- 145.

Lozano, C. 2008. Selective deficits at the syntax-discourse interface: Evidence from the CEDEL2 corpus. In N. Snape, Y.-k.I. Leung and M. Sharwood-Smith (eds.), Representational Deficits in SLA.Amsterdam: John Benjamins, 127-166.

Lozano, C. 2009. Pronominal deficits at the interface: new data from the CEDEL2 corpus. In C.M. Bretones Callejas et al. (eds.), Applied Linguistics Now: Understanding Language and Mind/La Lingüística Aplicada Hoy: Comprendiendo el Lenguaje y la Mente. Almería: Universidad de Almería, 213-227.

Margaza, P. and A. Bel. 2006. Null subjects at the syntax–pragmatics interface: Evidence from Spanish interlanguage of Greek speakers. In M. Gratham O’Brien, C. Shea and J. Archibald (eds.), Proceedings of GASLA 2006. Somerville, MA: Cascadilla Press, 88-97.

3 8 4 Language Acquisition Alexandra Prentza

Meridor, H. 2006. An Experimental Investigation of the Antecedent Preferences of Hebrew Subject Pronouns, MA Dissertation, University of Edinburgh.

Miltsakaki, E. 2007. A rethink of the relationship between salience and Anaphora Resolution. In Proceedings of the 6th Discourse Anaphora and Anaphor Resolution Colloquium. Lagos, Portugal.

Montrul, S. and C. Rodríguez Louro. 2006. Beyond the syntax of the Null Subject Parameter: A look at the discourse-pragmatic distribution of null and overt subjects by L2 learners of Spanish. In V. Torrens and L. Escobar (ed.). The Acquisition of Syntax in Romance Languages. Amsterdam: John Benjamins, 401- 418.

Oxford Quick Placement Test 2001, version 1. UCLES 2001

Papadopoulou, D., E. Plemenou, T. Marinis and I.-M. Tsimpli. 2007. Pronoun ambiguity resolution: evidence from L1 Greek children. Clinical Language Sciences, University of Reading.

Pérez-Leroux, A. and W. Glass. 1999. Null anaphora in Spanish second language acquisition: probabilistic versus generative approaches. Second Language Research 15(2): 220-249.

Parodi, T. and I.-M. Tsimpli. 2005. ‘Real’ and apparent optionality in second language grammars: finiteness and pronouns in null operator structures. Second Language Research 21(3): 250-285.

Roussou, A. and I.-M. Tsimpli. 2006. On Greek VSO again! Journal of Linguistics 42(2): 317-354.

Smith, N. and I.-M. Tsimpli. 1995. The mind of a savant. Language Learning and Modularity. Oxford: Blackwell.

Sorace, A. and F. Filiaci. 2006. Anaphora resolution in near-native speakers of Italian. Second Language Research 22(3): 339-368.

Sorace, A. and L. Serratrice. 2009. Internal and external interfaces in bilingual language development: Beyond structural overlap. International Journal of Bilingualism 13(2): 195-210.

Spyropoulos, V. and I. Philippaki-Warburton. 2001. Subject and EPP in Greek: the discontinuous subject hypothesis. Journal of Greek Linguistics 2: 149-186.

Language Acquisition 3 8 5 Major Trends in Theoretical and Applied Linguistics

Tsimpli, I.-M. and M. Dimitrakopoulou. 2007. The Interpretability Hypothesis: evidence from wh-interrogatives in second language acquisition. Second Language Research 23(2): 215-242.

Tsimpli, I.-M., A. Sorace, C. Heycock and F. Filiaci. 2004. First Language Attrition and Syntactic Subjects: A Study of Greek and Italian Near-Native Speakers of English. International Journal of Bilingualism 8(3): 257-277.

Tsimpli, I.-M., A. Sorace, C. Heycock, F. Filiaci and M. Bouba. 2003. Subjects in L1 attrition: Evidence from Greek and Italian Near-Native Speakers of English. In B. Beachley et al. (eds.), BUCLD 27 Proceedings, 787-797.

3 8 6 Language Acquisition Linguistic Structure and Narrative Quality in Child L1 Production

Ianthi Maria Tsimpli1, Maria Papakonstantinou1 and Ageliki Nicolopoulou2

1Aristotle University of Thessaloniki, 2Lehigh University [email protected], [email protected], [email protected]

Abstract

The study investigates the development of referential functions in the use of linguistic forms encoding character Reference with the use of qualitative and quantitative measures in children’s narratives. The link between memory scores and narrative performance is also examined. A story telling task presented in two alternative methods of story presentation (booklet vs. card) was used. Four age groups of Typically Developing (TD) Greek children participated. The results on character Reference show a priority in the acquisition of the linguistic forms associated with the function of character Maintenance compared to character Introduction. The method of story presentation seems to affect Length of the narrative as well as the frequency of Coordination and Subordination structures. Finally, some effects of working memory scores are shown in narrative competence.

1. Introduction

Narrative discourse can indicate cognitive and linguistic development. This is so since anaphoric strategies such as character introduction and maintenance but also cognitive mechanisms such as the ability to construct episodes in a particular order and build a mental model of the narrative are all involved in creating a coherent narrative. At the same time, Theory of Mind (ToM), i.e. the ability to consider the listener’s point of view in narrative production is also instrumental in good narrative abilities (Chafe, 1976). More precisely, the development of narratives is associated with the acquisition of referential functions (e.g. character introduction and maintenance), since the given/new opposition is a universal principle of discourse organization (Vion & Colas, 1999; Hickmann & Hendriks, 1999). Other linguistic markers of narrative development such as the use of co-ordination

Language Acquisition 3 8 7 Major Trends in Theoretical and Applied Linguistics

and subordination of phrases and clauses may also be used to evaluate narrative development. Although these properties are primarily linguistic, they are also sensitive to the child’s ability to create a story structure. It has been shown that coordination typically precedes subordination in narrative discourse development (cf. de Villiers, 1982). Turning to character reference, languages differ in the mechanisms they employ to introduce and maintain reference to a character in a narrative. In languages which make a definite/indefinite distinction, a referent mentioned for the first time in discourse is introduced with an indefinite nominal (1). Character maintenance, on the other hand, usually employs a definite nominal (2a), or a pronominal (2b). In null subject languages, such as Greek, character maintenance may involve the use of a null pronominal subject (2c).

(1) Mia fora ki enan kero zuse enas lagos.

once upon and a time live-PAST-3s a hare-SING-NOM “Once upon a time there lived a hare” (2a) Mia mera o lagos pige na mazepsi karota.

one day the hare-SING-NOM go-PAST-3s to pick carrots- PL-ACC (2b) Mia mera aftos pige na mazepsi karota.

one day it go-PAST-3s to pick carrots-PL-ACC (2c) Mia mera pro pige na mazepsi karota.

one day go-PAST-3s to pick carrots-PL-ACC

All options in (2a-c) are grammatical and acceptable and all express the referential function of character maintenance.

2. Previous Studies

Previous studies on narrative development report that children are better at maintaining than introducing a new referent (Hickman et al, 1996; Wang & Johnston, 2003). Other studies observed sensitivity to reference continuity vs. discontinuity from the age of 4 years (Hickman & Schneider, 1993; Hickman & Hendriks, 1999). Previous studies on the development of subordination in Greek narratives focus on the distinction between temporal and non-temporal adverbials (Kantzou, 2010; Kati & Kantzou, 2004), but do not deal with the development of referential functions. These studies have shown that the use of connectives and narrative length significantly increase between the age of four to seven years (Kati & Kantzou, 2004). However, according to the same study frequency of use of adverbial clauses does not show a comparable increase during the same period.

3 8 8 Language Acquisition Ianthi Maria Tsimpli, Maria Papakonstantinou and Ageliki Nicolopoulou

Since the present study primarily focuses on the development of character reference we compared two methods of story presentation in order to also evaluate the effects of method on accuracy of linguistic expressions according to each referential function. In previous studies the card method is reported to reinforce referent givenness more than the booklet method (Vion & Colas, 1999). It was further suggested that in the booklet method, 7 year olds tended to view each picture as an independent entity, while 11 year olds viewed it as a part of a whole. The card method promoted more referent givenness between 7-9 years. In a more recent study, Vion & Colas (2005) observed more production of temporal & causal links, which are additional markers of narrative coherence, in the card method.

3. Research Questions

The research goals of the present study are: a) to investigate 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 and b) to examine possible interactions of narrative performance with short-term memory scores. We also investigated the question whether working memory could be shown to account for the differences in method presentation.

4. Method

Ninety-six typically developing Greek children divided in four age groups (3,6- 6,6) participated in our study. The method used was an elicited story-telling task based on four different stories comprising four pictures each. Two methods of picture presentation were employed for narrative elicitation: a) the booklet and b) the card method. In the booklet method each picture is presented in a single page, while in the card method all pictures forming a story sequence appear on one page. Three memory tasks were also administered for independent measurements of verbal memory abilities in order to examine possible relations between narrative performance and attention resources. Specifically, theDigit Span Task (WISC-III, adapted to Greek: Γεώργας κ.α. 1997) and the Animal Span Task were employed, (adapted to Greek from the Turkish original (Aksu-Koc (2009)), which test short working memory. We also administered the sentence repetition task of the DVIQ test (Stavrakaki & Tsimpli, 2000), which tests accuracy of recall of various linguistic structures.

Language Acquisition 3 8 9 Major Trends in Theoretical and Applied Linguistics

5. Predictions

The method of presentation is predicted to affect narrative coherence markers, i.e. coordination, subordination and length of narration, since there may be different strategies of constructing episodes in the booklet, as opposed to the card method. Moreover, maintaining reference may involve increased use of pronominals in the card method assuming that this method promotes referent givenness more, according to previous studies (Vion & Colas, 1999). On the other hand, differences in memory scores among children may facilitate narrative production only in the booklet method, since the child does not view the complete story at once and has to rely on memory to construct the story episodes. Irrespective of method, the development of reference is expected around 4;6, with character maintenance being more adult-like earlier than character introduction (cf. Hickman et al, 1996; Wang & Johnston, 2003). Accordingly, definiteness is expected to be more prominent in use by younger children than indefiniteness. Finally, memory is predicted to be a positive indicator of good narrative performance in the production of character maintenance and shift.

6. Results

The results of the study are presented for the tested variables both per method and irrespective of the method of story presentation. Considering the method of presentation, our results for character introduction (Figure 1) show significantly enhanced performance in the booklet method only in the group of 4;6 year old children (x2 =21.333, df=1, p=.000) . Developmentally, there is a significant change between the 3;6 and the 5;6 groups (x2=3.802, df=1, p=.051) and between the 4;6 and the 6;6 year old children (x2=6.641, df=1, p=.010). Irrespective of method of presentation (Figure 2), as predicted there is a significant development between the ages of 4;6 and 5;6(x2=8.277, df=1, p=.004) for the function of referent introduction. Within-group performance is significantly more target than non-target like for all groups tested (3;6: x2=4.667, df=1, p=.031; 4;6: x2=18.305, df=1, p=.000; 5;6:x2=64.076, df=1, p=.000; 6;6: x2=68.283, df=1, p=.000). Reference maintenance (Figure 3) with the use of a definite nominal is significantly more target-like in the booklet method in the youngest group (x2=18.864, df=1, p=.000), while the 5;6 year olds preferred the card method (x2=10.777, df=1, p=.001). Developmentally, there are significant changes up to 5;6 years (3;6 vs. 4;6: x2=6.079, df=1, p=.014; 4;6 vs. 5;6: x2=12.685, df=1, p=.000). On the other hand, irrespective of method and across age groups, there is development of maintenance from 3;6 to 4;6 (x2=12.193, df=1, p=.000) and

3 9 0 Language Acquisition Ianthi Maria Tsimpli, Maria Papakonstantinou and Ageliki Nicolopoulou

Figure 1. Use of indefinite NPs for character introduction per method

Figure 2. Use of indefinite NPs for character introduction across methods

Figure 3. Use of definite NPs for character maintenance per method

Language Acquisition 3 9 1

Major Trends in Theoretical and Applied Linguistics

between the ages of 5;6 and 6;6 (x2=9.085, df=1, p=.003) (Figure 4). Within- group data indicate that all groups produced significantly more target than non-target overt DPs to mark maintenance (3;6: x2=47.867, df=1, p=.000; 4;6: x2=10.625, df=1, p=.000; 5;6: x2=118.140 df=1, p=.000; 6;6: x2=156.568, df=1, p=.000). Maintenance with the use of a null pronoun shows a developmental trend up to 5;6 years (3;6 vs 4;6: x2=5.505, df=1, p=.019; 4;6 vs. 5;6: x2=4.050, df=1, p=.044). All groups distinguish between ambiguous and unambiguous null pronoun uses (3;6: x2=5973, df=1, p=.015, 4;6: x2=40.055, df=1, p=.000, 5;6: x2=84344, df=1, p=.000, 6;6: x2=81.000, df=1, p=.000) (Figure 5). The method of story presentation did not affect performance on pronominal use in any group, contrary to the prediction that the card method would promote referent givenness more than the booklet method.

Figure 4. Use of definite NPs for character maintenance

Figure 5. Appropriate and inappropriate (ambiguous) use of null pronouns for character maintenance across methods

3 9 2 Language Acquisition

Ianthi Maria Tsimpli, Maria Papakonstantinou and Ageliki Nicolopoulou

Maintenance with a null pronominal is not significantly better than the use of indefinite DPs for character introduction. As predicted maintenance and definiteness is achieved more accurately than reference introduction only when expressed with a DP (3;6: x2= 14.434, df=1, p=.000; 4;6: x2=38.908, df=1, p=.000; 5;6: x2=13.232, df=1, p=.000; 6;6: x2=31.420, df=1, p=.000) (Figure 6). Between group comparisons do not reveal any significant changes in the pattern of acquisition of the introduction of a referent when compared to the maintenance of a referent. In character maintenance, definite nominals are significantly more accurately produced than pronominals in all groups tested (Figure 7).

Figure 6. Comparison between accurate production of DPs for character introduction vs. character maintenance

Figure 7. Character maintenance with nominal vs. pronominal DPs

Language Acquisition 3 9 3

Major Trends in Theoretical and Applied Linguistics

Within group comparisons revealed a high significance in the ease of target production of nominal DPs when compared to pronominal DPs (3;6: x2=19.060, df=1, p=.000; 4;6: x2=38.673, df=1, p=.000; 5;6: x2=24.949, df=1, p=.000; 6;6: x2=52.910, df=1, p=.000). Turning to the other properties of narratives examined, length of the narrative (Figure 8) which was measured on the basis of the verbs (clauses) used, revealed a developmental trend between the ages of 3;6 and 4;6 (x2=28.351, df=1, p=.000) and between the ages of 5;6 and 6;6 (x2=10.343, df=1, p=.000). This result is in line with the results of previous studies (cf. Kati & Kantzou, 2004) and justifies our hypothesis that there is a development in the number of clauses used in narratives. Considering the length of narrative per method (Figure 9), in the booklet method children engaged in longer narratives in the 4;6 (x2=43.668, df=1, p=.000) and the 5;6 year old groups (x2=110.124, df=1, p=.000), but there was an advantage of the card method in the oldest group of 6;6 year olds (x2=6.016, df=1, p=.014). A developmental trend in the production of loose linking structures (Figure 10) was observed between the ages ages 3;6 and 4;6 (x2=6.016, df=1, p=.014) and between the 5;6 and 6;6 year olds (x2:16.206, df=1, p=.000). The frequency of use of loose linking structures does not seem to be affected by the method of story presentation. Regarding the use of coordination by method of story presentation (Figure 11) we notice a developmental change affecting coordination between 3;6 and 4;6 year olds (x2=17.126, df=1, p=.000). Within group differences reveal preference for the booklet method for all groups except the youngest. Preference 2 is for the booklet method for the 4;6 and the 5;6 group (4;6: x =.26.472, df=1, p=.000; 5;6: x2=7.298, df=1, p=.007) while the 6;6 year olds preferred the card

Figure 8. Length of narrative (no of verb/clauses) across methods

3 9 4 Language Acquisition

Ianthi Maria Tsimpli, Maria Papakonstantinou and Ageliki Nicolopoulou

method (x2=4.489, df=1, p=.034). The results showing enhanced production of coordination in the booklet method are in line with previous studies that report less referent givenness in the booklet method when compared to the card method. Referent givenness may promote more the production of subordination. Irrespective of method, coordination develops up to 5;6 (3;6 vs. 4;6: x2=17.526, df=1, p=.000; 4;6 vs. 5;6: x2=9.648, df=1, p=.000). As regards subordinate structures the results were differentiated by method only for the oldest group, who used more subordination in the card method (x2=4314, df=1, p=.038) (Figure 13).

Figure 9. Length of narrative (number of verbs/clauses) per method

Figure 10. Frequency of loose linking structures

Language Acquisition 3 9 5

Major Trends in Theoretical and Applied Linguistics

Figure 11. Frequency of coordinate structures per method

Figure 12. Frequency of coordinate structures across methods

Embedded structures, as are the subordinate structures means that referent givenness is well established. This result is in line with previous studies (Vion & Colas, 1999) reporting that the card method reinforces definiteness between 7-9 years old. As regards subordination irrespective of method, it develops up to 4;6 (x2=11.082, df=1, p=.000). Then there are no significant differences with the older groups (Figure14). As for the priority in the development of the coordinate structures over the subordinate, the 4;6 age group seems to be showing a clear developmental

3 9 6 Language Acquisition

Ianthi Maria Tsimpli, Maria Papakonstantinou and Ageliki Nicolopoulou

Figure 13. Frequency of subordinate structures per method

Figure 14. Frequency of subordinate structures across methods

turning point for both coordination and subordination. As was predicted,

the comparison of the frequency of production of the coordinate vs. the subordinate structures (Figure 15) within groups revealed that all groups produced more coordinate than subordinate structures (3;6: x2=6.630, df=1, p=.000; 4;6: x2=7.476, df=1, p=.006; 5;6: x2=16.399, df=1, p=.000; 6;6: x2=6.770, df=1, p=.000). Between groups there were not significant developmental changes in the pattern of production. Studies for other languages also support the precedence of coordination over subordination (cf. de Villiers, 1982).

Language Acquisition 3 9 7

Major Trends in Theoretical and Applied Linguistics

Figure 15. Frequency of coordinate vs. subordinate structures

Finally, to examine possible links between good narrative ability and memory, children were tested with two short memory tests, the Digit Span Task and the Animal Span Task (Table 1) and the sentence repetition DVIQ test (Table 2). The Digit (WISC-III, adapted to Greek: Γεώργας κ.α. 1997) and the Animal Span Task (adapted to Greek from the Turkish original (Aksu-Koc, 2009)), measure short-term memory involving forward and backward digit recall and similar patterns with animal names respectively. In the table below we present mean, upper and lower scores per age group.

Table 1. Scores from the Short Memory Tests

Digit Span Task

Age groups Mean scores Upper scores Lower scores

3;6 4.33 7.39 1.27

4;6 6.04 10.12 1.96

5;6 7.83 11.15 4.51

6;6 8.57 12.47 4.67

Animal Span Task

Mean scores Upper scores Lower scores

3;6 5.33 8.75 1.91

4;6 6.71 10.13 3.29

5;6 8.05 11.17 4.93

6;6 8.27 10.77 5.77

3 9 8 Language Acquisition Ianthi Maria Tsimpli, Maria Papakonstantinou and Ageliki Nicolopoulou

Table 2. Scores from the DVIQ sentence repetition task

DVIQ

Age groups Raw numbers % Norms

3;6 41/48 85% 39.4

4;6 44/48 92% 39.4

5;6 47/48 98% 41.8

6;6 25/30 83% -

The results of the two tasks are comparable. The Animal Span Task though reached slightly higher mean scores within all age groups tested. As for the sentence repetition task (Table 2) results of the 3;6 to 5;6 groups were within the standardized norms of the DVIQ preschoolers Test (Stavrakaki & Tsimpli, 2000). Additionally, significant difference is found in the comparison between the 3;6 and the 5;6 year olds (x2=14.909, df=1, p=.027). As for the 6;6 year olds notice that the scale is different. This group took the school age edition of the Test for which there are no standardized norms. Therefore, no direct comparisons can be made between the oldest group of children and the rest of the groups. The results in the three memory tests presented above were compared to performance in the narrative task. The ability to use the appropriate DP to mark reference was examined in relation to memory scores. The comparison revealed that children with high memory scores seemed to show better narrative performance in the target use of nominals and pronominals to introduce and maintain reference. In a first attempt to relate memory with referential functions Figure 16 shows the 5;6 group’s performance in the Animal Span Task. Each point stands for each child’s performance. The 8th dot is the case of a child who scored very low in the task. His performance lies in the lower outliers. On the other hand, the 11th dot presents the score of a child with high memory scores belonging to the upper outliers. What is of interest is to see whether low or high level performance in the memory task influences results in referential scores. In figure 17 we present the 5;6 group’s scores in the function of referent maintenance with a nominal or a pronominal DP. We observe that the scores of both the 2nd and the 11th dot which correspond to the same children as in Figure 16 are very high. In particular, the 11th child who had a high memory score also had a very high target performance in referent maintenance. What can be implied from this first attempt to relate memory with reference is that good memory ability boosts narrative skills but it does not seem to be a necessary condition for good narrative ability.

Language Acquisition 3 9 9 Major Trends in Theoretical and Applied Linguistics

15

10 child's score mean scores 5 upper outliers 0 lower outliers 0 5 10 15 children

Figure 16. Results of the Animal Span Task

100%

80% 60% child's score

40% chance level

scores 20% 0%

0 5 10 15

age groups

Figure 17. Target performance in maintenance with an overt DP

We assume that the reason why the memory test results were not illuminating with respect to their role in narrative discourse development has to do with the fact that what was tested with the Digit Span Task was STM (short term memory). A test measuring working memory at the sentence level would probably be more revealing in this respect.

7. Discussion

For all groups except the youngest, referent maintenance was easier when realized with an overt nominal, while referent introduction as well as maintenance with a null or an overt pronominal subject were less frequently used. The difference between lexical DPs in introduction and maintenance could be associated with the delay in the development of indefinites, possibly due to their more elaborate semantics/pragmatics. As a result, indefinite productions appear to be better markers of narrative development (see also Schneider and Hayward, 2010). That pronouns are more demanding than definite noun phrases has also been shown in adult data. Arnold and Griffin (2007) found that adults too prefer to use lexical DPs over pronouns even when gender differences in characters

4 0 0 Language Acquisition

Ianthi Maria Tsimpli, Maria Papakonstantinou and Ageliki Nicolopoulou

would make pronominal use an unambiguous marker of reference. Pronominal use is related to discourse anaphora and therefore, their target production is considered more complex and presupposes more advanced pragmatics or more resources in information processing and integration. The different performance of the 4;6 age group indicates that this age is critical for the development of narratives since linguistic variables associated with it, develop at that stage. In particular, length of narration measured in number of verbs used, loose linking, coordination and subordination show significant development at 4;6. Coordinate structures are more frequent than Subordinate ones, however, for all groups tested. There is not a clear effect of method on the development of Reference, but, overall, the booklet method elicits better performance up to 6;6 with respect to coordination and number of verbs used. The oldest group had a better performance in the card method for all linguistic properties examined except for referential functions. With respect to memory ability, low memory scores did not appear to entail low linguistic scores. However, STM (Short-Term Memory) is different from Working Memory (Baddeley, 2007) and possibly, narrative production would be sensitive to working memory scores (attentional control and resource allocation than phonological short term memory) (cf. Montgomery et al 2009 for such results on narrative comprehension).

8. Conclusion

Our study reveals that of the various aspects of Reference examined, introduction and pronominal use for Maintenance are slower in development than Maintenance with the use of full noun phrases. Accordingly, the referential function of Character Introduction seems a more reliable marker of the development of referential abilities in narratives. The method of story presentation affects quantity more than quality of narrative production. Memory span is an indicator of narrative ability only for older children.

Language Acquisition 4 0 1 Major Trends in Theoretical and Applied Linguistics

References

Arnold, J. and Z. Griffin. 2007. The effect of additional characters on choice of referring expression: Everyone counts. Journal of Memory and Language 56, 521- 536.

Baddeley, A. 2007. Working memory, thought and action. Oxford: Oxford University Press.

Bower, G.H. and D.G. Morrow. 1990. Mental models in narrative comprehension. Science 247, 44-48.

Chafe, W. 1976. Givenness, contrastiveness, definiteness, subjects, topics and point of view. In C.N. Li (ed.), Subject and topic. New York: Academic Press.

De Villiers, J. 1982. The development of sentence coordination. Language Development Vol.1: Syntax and Semantics.

Γεώργας Δ.Δ., Ι.Ν. Παρασκευόπουλος, Η.Γ. Μπεζεβέγκης και Ν.Δ. Γιαννίτσας. 1997. Ελληνικό WISC- III: Wechsler Κλίμακες Νοημοσύνης για Παιδιά. Ελληνικά Γράμματα.

Hickmann, M. 1996. Information Status and Grounding in Children’s Narratives: A crosslinguistic Perspective. Processing Interclausal Relationships. Studies in the Production and Comprehension of Text, 221-243.

Hickmann M. and H. Hendriks. 1999. Cohesion and Anaphora in Children’s Narratives: a comparison of English, French, German and Mandarin Chinese. Journal of Child Language 26, 419-452.

Hickmann, M., H. Hendriks, F. Roland and J. Liang. 1996. The marking of new information in children’s narratives: a comparison of English, French, German and Mandarin Chinese. Journal of Child Language 23, 591-619.

4 0 2 Language Acquisition Ianthi Maria Tsimpli, Maria Papakonstantinou and Ageliki Nicolopoulou

Hickmann, M. and P. Schneider. 1993. Children’s ability to restore the referential cohesion of stories. First Language 13, 169-202.

Κάντζου, Β. 2010. Η χρονική οργάνωση του αφηγήματος στην κατάκτηση της Ελληνικής ως πρώτης και ως δεύτερης γλώσσας. Διδακτορική Διατριβή. Εθνικό και Καποδιστριακό Πανεπιστήμιο Αθηνών.

Κατή, Δ. και Β. Κάντζου. 2004. Η συνδετικότητα αφηγημάτων: Αναπτυξιακή και διαγλωσσική προσέγγιση. Πρακτικά της 24ης Ετήσιας Συνάντησης του Τομέα Γλωσσολογίας της Φιλοσοφικής Σχολής του Α.Π.Θ. Θεσσαλονίκη: Κυριακίδης, 246-257.

Stavrakaki, S. and I.M. Tsimpli. 2000. Diagnostic verbal IQ test for school and preschool children: Standardization, statistical analysis, and psychometric properties. Proceedings of the 8th conference of the Panhellenic Association of Speech and Language Therapists. Athens: Ellinika Grammata, 95-106.

Vion, M. and A. Colas. 1999. Expressing Coreference in French: Cognitive Constraints and Development of Narrative Skills. Journal of Psycholinguistic Research 28 (3).

Language Acquisition 4 0 3 Major Trends in Theoretical and Applied Linguistics

4 0 4 Section 5: Language Disorders

4 0 5 Major Trends in Theoretical and Applied Linguistics

4 0 6 Language Disorders Modality-Specific Noun–Verb Dissociations in L1 but No Effect in L2: Evidence from Bilingual Aphasia

Maria Kambanaros

Cyprus University of Technology [email protected]

Abstract

The spoken and written productions of object and action names in two languages of a multilingual speaker with aphasia were compared. The results showed preserved written naming for action words in L1 (Greek) in light of severe difficulties with spoken action names, whereas for object names the opposite pattern was observed. For L2 (English), severe difficulties retrieving action and object names for spoken and written naming were found. The finding supports the widely accepted view that the cognitive processes used for spoken and written naming are relatively independent components that can be selectively impaired after brain injury. In the case of bilingual speakers, the cognitive processes can involve both languages.

1. Introduction

Not much is known about spoken and written breakdown processes, particularly for verbs and nouns, in bilingual speakers after brain impairment. In this study, the spoken and written naming abilities of a simultaneous trilingual fluent aphasic individual in two of his (three) languages, Greek and English, which differ markedly in morphological complexity, orthographic transparency, and alphabetic script on action and object retrieval at the single-word level are reported.

2. Background

2.1. Verb-noun Dissociations in Aphasia

A substantial number of studies have dealt with dissociations between nouns and verbs using confrontation naming tasks depicting pictures of actions and

Language Disorders 4 0 7 Major Trends in Theoretical and Applied Linguistics

objects. A small number of studies have identified dissociations between action and object naming in bilingual speakers with aphasia (see Kambanaros, 2009, for a detailed review). Taken together with evidence of dissociations from monolingual speakers with aphasia, these findings suggest that words are organized in the brain according to semantic and grammatical categories. A different though related issue for grammatical class dissociations in aphasia is modality specificity. Research shows that brain damage can selectively impair verb and noun processing in opposite ways at different levels of the lexical system. The most commonly reported pattern has been selective impairment for action or object spoken word naming. These impairments can result from damage to the speech production system at one or more levels: the conceptual level, semantic-syntactic level, including lemma retrieval, or the phonological level including lexeme retrieval and at the level of articulation (Levelt, 1989). In the case of patients with bilingual aphasia, the locus of impairment at one or more levels of processing may be greater for one language compared to another depending on factors such as age of acquisition, familiarity and language dominance and moreover the effects of grammatical class may be different in each language (see Kambanaros, 2009 and references within). Reports of impairment to written picture naming of nouns and verbs in aphasia are less common but nevertheless quite revealing. Caramazza and colleagues investigated lexical access for action and object names in a number of monolingual English speakers with aphasia and compared performance with the same items on oral and written word production tasks. In some cases, written word performance contrasted remarkably with spoken production (Rapp & Caramazza, 2002). For example, a selective deficit with verbs might be restricted to written word production in some patients and to speech production in others (Caramazza & Hillis, 1991) and in the case of a modality effect, a selective deficit might be found for one grammatical class only, e.g. impaired production for nouns in speech only (Hillis & Caramazza, 1995). Most intriguingly, Rapp and Caramazza (2002) established that the same patient may have greater difficulty with verbs in spoken word production and with nouns in written word production. The results strengthen the suggestion that not only are grammatical categories distinctive in the language system, the lexical processing system consists of modality-specific input and output components that reflect these distinctions at least for monolingual speakers of English (Hillis & Caramazza, 1995). Although the properties of the English grammatical system are unique (as in all languages), these patterns of modality-specific grammatical processing raise a question of whether bilingual speakers with aphasia would show modality-specific grammatical processing in one or both of their languages. A three-way interaction between grammatical class, modality of production, and language status, i.e. between the first acquired (L1) and the second acquired (L2) language, would suggest that the lexical processing system is highly modular

4 0 8 Language Disorders Maria Kambanaros

(Forster & Jiang, 2001). Furthermore, the neural organization of verbs and nouns is suggested to be highly independent as evidenced by extant studies in aphasia and brain imaging (see Obler et al., 2007, and references within).

2.2. Greek Versus English

Modern Greek is a stem-based language, unlike English (word-based), with a more complex morphology than English (Ralli, 2003). Morphophonological word forms are inflected according to grammatical category, for instanceskoup-izi ‘he/ she sweeps’ is a verb and ‘skoup-a’ ‘broom’ is a noun. Thus, nouns and verbs are differentiated by different suffixes and are also marked for person and gender. Each stem in Greek is bound and only projected at the phonological word level after the correct inflectional suffix is attached to the stem at morphological level. Furthermore, the Greek gender system distinguishes masculine, feminine, and neuter. Nouns inflect for gender at the morphological level. However, nouns are not only assigned gender on the basis of word meaning, but also on the basis of more general morphological rules. For instance, nouns ending with the nominative singular suffix –as (o papas ‘the priest’) and with the accusative singular –a (ton papa) are masculine, whereas nouns ending with the nominative singular –as (to kreas ‘the meat’]) and the accusative singular –as (to kreas) are neuter. Gender agreement is also marked in Greek throughout the whole noun phrase, including determiners and adjectives. Information about the grammatical category and about morpho-syntactic features, such as person, tense, aspect and mood for verbs or gender and case for nouns, are prominent aspects in Greek as they must be accurately projected, marked and expressed during single word production. In this manner, grammatical information also plays a crucial role in the selection and retrieval of the appropriate phonological and/or orthographic representation or lexeme via language-specific processes during morpho-phonological encoding.

3. Methodology

3.1. Participant

AA is a 25-year-old right-handed male university student who sustained a left parieto-occipital lesion verified by neuroimaging (CT) as a result of cerebral hemorrhage on February 19, 2006, while sleeping. An angiography conducted on February 21, 2006, revealed an arteriovenous malformation (AVM), that is, an abnormal tangle of blood vessels in the brain. AA has a right hemi paresis

Language Disorders 4 0 9 Major Trends in Theoretical and Applied Linguistics

involving the arm and leg. Although AA was right-handed before his stroke he now writes only with his left hand. AA was born in Athens, Greece and was raised as a simultaneous bilingual hearing and speaking Greek, English and Arabic from infancy. However, he had been schooled all his life in Greek, with many years of English language instruction throughout his schooling in Greece, but has received no formal education in Arabic. AA’s mother is Greek, his father Palestinian, and the home language is English. He has two younger brothers. AA was asked to rate his abilities in his three languages prior and post stroke in response to a self-rating language scale (adapted from Li et al., 2006) using a seven point scale (7 =excellent, 6=very good, 5=good etc) in the following modalities: understanding; communicating/speaking; reading; writing. There was no significant difference between prior and post-stroke abilities. Hewas also required to quantify his current domain of use for each language. Family members verified his responses. The results are presented in Table 1.

Table 1. AA’s self-rated estimates of proficiency in his different languages across linguistic domains andof language use (percentage).

L1 (Greek) L2 (English) L3 (Arabic)

Estimates of proficiency (on a scale from 1 to 7)

Understanding 7 6 5

Speaking 7 7 5

Reading 7 5 —

Writing 7 6 —

Language use background (percentage)

Home life 75 12.5 12.5

Tertiary Education 100 — —

Socializing 75 12.5 12.5

Reading 75 25 —

Television/Radio 12.5 75 12.5

AA considers Greek to be his predominant language and English and Arabic second languages with a greater ability in English, since he is unable to read and write Arabic. Hence Greek was classified as L1 and English as L2. AA had no premorbid writing disorder and his vision and hearing are normal. He was in his third year of a 4-year undergraduate degree in Building Renovation and Restoration at the Technological Educational Institute Patras, Greece, and he reads and writes Greek on a daily basis. This course also has three English language modules, of graded difficulty (intermediate–high–terminology), which AA has successfully completed.

4 1 0 Language Disorders Maria Kambanaros

On the basis of the language assessment using the Boston Diagnostic Aphasia Examination (BDAE), AA presented with anomic aphasia characterized by fluent, well-articulated speech but mild to moderate anomia (word-finding difficulties) in both Greek (L1) and English (L2). His results on the BDAE are presented in Table 2. His auditory comprehension across language tasks was within normal limits and he showed intact conceptual abilities in both languages.

Table 2. AA’s performance on the BDAE in Greek and English.

Greek English (L2) (L1)

BDAE subtests Word discrimination 72/72 72/72 Body parts 20/20 20/20 Commands 15/15 15/15 Complex ideation 10/12 8/12 Responsive naming 24/30 0/30 Confrontation naming 92/114 78/114 Animal naming 2 0

BDAE: English version: Goodglass, H., & Kaplan, E. (1983); Greek version: Papathanasiou et al., (2008).

3.2. Materials

Subtests of the Greek Object and Action Test (GOAT; Kambanaros, 2003) were administered to assess spoken and written retrieval of object and action names. The GOAT contains 84 items: 42 action and 42 object coloured photographs measuring 10x14cm in size. Subtests were pilot tested on a group of twenty non-brain injured, Greek-English bilingual speakers aged between 55 and 75 years; the GOAT was originally designed to assess noun and verb retrieval in bilingual aphasic speakers of Greek and English. Only items named with 80% accuracy or more were included in the test. None of the Greek words in the test were English cognate words. Object pictures were concrete inanimate nouns and included manipulated instruments used for activities of daily living such as garage tools, garden equipment, kitchen utensils, household items, office and personal implements and included no body parts. All verbs were monotransitive and pictured actions were restricted to past stereotypical roles, that is, a woman was shown performing household activities (e.g. sweeping), and a man was performing more manly duties (e.g. hammering). All action names corresponded to either an instrumental verb (where an instrument is part of the action e.g. sweeping) or to a non-instrumental verb (e.g. climbing). There were no significant differences between the mean word frequencies for nouns (mean 89.31 per million) and verbs (mean 69.95 per million) across all English

Language Disorders 4 1 1 Major Trends in Theoretical and Applied Linguistics

(L2) subtests of the GOAT. For Greek, lemma frequencies for object and action names were calculated based on the printed word frequency count for Greek (see Hatzigeorgiou et al., 2000). A Mann-Whitney Test revealed no significant difference between object and action lemmas [z = –0.154, p = 0.878]. In addition, there was no significant difference in syllable length between object and action names [z = –0.610, p = 0.542]. Furthermore, object and action names were measured for key psycholinguistic variables, including age of acquisition (AoA), imageability and picture complexity. A Mann-Whitney test revealed that object and action names were not significantly different on AoA [z = –1.168, p = 0.243], but there was a significant difference in ratings for word imageability [z = –2.978, p = 0.003] and picture complexity [z = –2.331, p = 0.20] with higher ratings for object compared to action names. Of the object names, 7 were stressed on the ultimate syllable, 30 on the penultimate, and 5 on the antepenultimate. The action names were first conjugation (or paroxytone) verbs that are by far the most numerous in Greek, and all (but one) were stressed on the penultimate syllable. The internal word structure of verbs in L1 consisted of [root + affix] for simple forms and [root + affix + affix] for more complex ones. Nouns in L1 were not controlled for gender. Of the total nouns (42), 6 were masculine, 15 were female and 21 were neuter. All responses were recorded in writing and also on audiotape.

3.3. Procedures

For spoken naming AA was shown the GOAT photographs individually (42 actions and 42 objects) and was asked to name the object or action in the picture using one word either in Greek or English according to the language investigated at the time. Each language was tested at least one week apart and subtests were given in random order. For written naming tested 10 days later, AA was asked to write the name of the object or action in the picture using one word either in Greek or English. Each language was tested at least one week apart and subtests were given in random order.

4. Results

4.1. Spoken Naming

AA was able to name 36/84 (42.8%) pictures correctly in L1 (19/42 or 45.2% action names and 17/42 or 40.5% object names), and 24/84 (28.5%) pictures correctly in L2 (12/42 or 28.5% action names and 12/42 or 28.5% object names). A chi-

4 1 2 Language Disorders Maria Kambanaros

square analysis revealed a significant difference in verbal naming performances between L1 and L2 (x2 = 21.429; p < .001), with AA showing a better overall performance in L1 compared to L2, for both action names (x2 = 8.679; p = .003) and object names (x2 = 27.429; p < .001). Nevertheless, oral naming performances for action and object names were severely impaired in both languages. Furthermore, there were non-significant differences between AA’s action and object naming performances both in L1 and L2. Moreover, similar error types were observed in both languages for action but not for object names. Specifically, no responses or some equivalent thereof (e.g., “Don’t know”) were the most prominent error type for action names in L1 (13/23 or 56.5%) and L2 (16/25 or 64%) followed by semantic errors, 7/23 (30.4%) in L1 and 6/25 (24%) in L2. Object names in contrast produced the largest number of omissions in L2 (24/29 or 83%), while in L1 errors for object names were divided between semantic errors (8/25 or 32%), no responses (7/25 or 28%), and code-switching responses (5/25 or 20%).

4.2. Written Naming

AA was able to correctly write the names of 62/84 (73.8%) pictures in L1 (34/42 or 81.0% action names and 28/42 or 66.6% object names), and 16/84 (19.0%) correct picture names in L2 (8/42 or 19.0% action names and 8/42 or 19.0% object names). A chi-square analysis revealed a significant difference in written naming performances between L1 and L2 (x2 = 3.429; p < .001), with AA showing a better written performance in L1 compared to L2, for both action names (x2 = 5.250; p = .022) and object names (x2 = 42.857; p < .001). Furthermore, there was a significant difference in AA’s performance for written action names compared to written object names in L1 (x2 = 5.950; p < .001), with a significantly better performance for action words. There was a non-significant difference in written naming between action and object words in L2. Furthermore, error types differed between the two languages for action and object names. Given the poorer performance in L2, both word types produced the same error type for action and object words viz. mainly no written responses (omissions) to the picture presented. On the other hand, for Greek (L1) the predominant error type for written action and object naming was mainly letter substitutions when writing action (7/7) and object (13/14) words. All spelling errors for action and object names in Greek involved a phonological-orthographic substitution of a single vowel in either word initial (3/7) or word medial position (4/7) for action and word final (8/13), initial (3/13), and both initial and final positions in the same word (2/13), for object words. All incorrect written naming attempts for action and object pictures were phonologically plausible alternatives for Greek (L1). All responses with coded error types are presented in Table 3.

Language Disorders 4 1 3 Major Trends in Theoretical and Applied Linguistics

Table 3. AA’s responses in L1 for action and object words in spoken and written naming.

Picture Written naming Spoken naming

Action name Ανακατεύει (mixing) + NR Μαζεύει (raking) + NR Ψαρεύει (fishing) + NR Σερβίρει (serving) + NR Χτενίζει (combing) + NR Τραβάει (pulling) + NR Δένει (tying) + NR Φοράει (wearing) + NR Φουσκώνει (blowing) + NR Ξύνει (sharpening) + NR Κόβει (cutting) + Σκίζει (ripping) Πλένει (washing) + Καθαρίζει (cleaning) Ζυγίζει (weighing) + Ζυγοστάθμιση (aligning) Σκουπίζει (sweeping) + Ξεσκονίζει (dusting) Κουρδίζει (winding) + Αλλάζει (changing) Διαβάζει (reading) + Κοιτάει (looking) Καρφώνει (hammering) + Σφυρί (hammer) Ταχυδρομεί (posting) NR Δώσει (giving) Μαγειρεύει (cooking) Μαγηρεύει + Κλειδώνει (locking) Κλειδόνι + Σφυρίζει (whistling) Σφοιρίζει + Χτίζει (building) Χτύζει DK Λιμάρει (filing) Λυμάρει DK Σιδερώνει (ironing) Σιδερόνει iron Τρυπάει (drilling) Τριπάει Ανοίγει τρύπες (Opening holes) Object name Τρίφτης (grater) + Τριφτήρι Κατσαρόλα (saucepan) + Φούρνος Πινέλο (paint brush) + Μπογιά Βελόνα (needle) + Βελονάκι Σκάλα (ladder) + Ανεβατήριο Μικρόφωνο (microphone) + Ομιλία Σίδερο (iron) + iron Σφουγγαρίστρα (mop) + mop Φάκελος + envelope Σφυρί (hammer) + hammer Τσουγκράνα (rake) + NR Ξύστρα (sharpener) + NR Κόλλα (glue) + Κολλάει

4 1 4 Language Disorders Maria Kambanaros

continuedTable 3. AA’s responses in L1 for action and object words in spoken and written naming.

Picture Written naming Spoken naming

Ξυράφι (razor) + Ξυρίζομαι Σκούπα (broom) + Κλειδί (key) Κλειδή + Σφυρίχτρα (whistle) Σφηρίχτρα + Κρεββάτι (bed) Κρεββάτη + Πολυθρόνα (couch) Πολυφρόνα + Μυστρί (trowel) μιστρή NR Σκοινί (rope) σκοινή NR Γάντι (glove) γάντη NR Μπαλόνι (balloon) μπαλόνη NR Ψαλίδι (scissors) ψαλίδη NR Κουδούνι (bell) κουδίνη Ντιν-ντιν Αναπτήρας (lighter) αναπτύρας Τσιγάρο Τρυπάνι (drill) τρυπάνη μπλακεντέκερ Μολύβι (pencil) μολύβη pencil Ποτιστήρι (watering can) ποτιστήρη Ποτίζω

Table 3 (a). AA’s responses in English (L2) for action and object words in spoken and written naming.

Picture Written naming Spoken naming

Action name stirring + ΝΡ cutting NR + reading NR watching singing cinging + tying NR NR weighing NR NR drawing drowing + ringing NR ticking combing NR NR sitting siting + building blooing NR sweeping NR NR watering NR NR sleeping sleening + pulling NR + serving cerving + drilling NR NR

Language Disorders 4 1 5 Major Trends in Theoretical and Applied Linguistics

continuedTable 3 (a). AA’s responses in English (L2) for action and object words in spoken and written naming.

Picture Written naming Spoken naming

painting NR brushing sewing NR NR lighting NR NR grating NR NR shaving NR NR glueing NR + sieving NR NR climbing NR + watching wacing + hammering NR NR raking NR gardening wearing NR dressing sharpening NR NR winding NR ticking locking NR closing whistling bloei NR filing NR NR Object names spoon NR NR sponge spang Scotch brite (fishing) rod NR + tray NR serve scales NR NR comb NR + sieve NR NR tie NR + watch wach + newspaper NR NR file NR NR grater NR NR saucepan NR NR (paint) brush NR NR needle NR pin ladder NR + microphone microfon + mop NR NR envelope NR + hammer hower + rake NR gardening sharpener NR NR

4 1 6 Language Disorders Maria Kambanaros

continuedTable 3 (a). AA’s responses in English (L2) for action and object words in spoken and written naming.

Picture Written naming Spoken naming

glue + + razor NR NR broom brom NR key + + whistle NR NR bed NR NR couch koch + trowel NR NR rope NR pull glove NR NR couch koch + trowel NR NR rope NR pull glove NR NR balloon ballon + scissors scezer + lighter NR NR drill NR NR (watering) can + NR

Key: + = Correct response. Written responses are organised according to error type. Corresponding spoken responses are classified as either correct, Omission (NR= No response, Don’t know responses or some equivalent thereof), Code switches (CSC=code switch correct), Semantic errors (single word or descriptions).

5. Discussion

AA exhibited a clear dissociation between (impaired) spoken naming and (relatively spared) written naming in his L1 (Greek) supporting the view that the grammatical categories of verbs and nouns are independent in the lexical processing system as argued by Caramazza and colleagues. With regards to spoken picture naming we assume for AA, sustained damage at the level of the phonological output lexicon in L1 (Greek) and L2 (English) with his comprehension for action and object names in both languages preserved. AA was profoundly anomic on (single-word) picture naming tasks for both word categories across the two languages. This was confirmed by the large number of first, omission errors (e.g., “don’t know” answers or no responses), and second, semantic errors, for both action and object names in L1 (Greek) and L2 (English). We assume that his spoken naming difficulties arose from a severe impairment

Language Disorders 4 1 7 Major Trends in Theoretical and Applied Linguistics

of lexical-phonological representations or access to them. Semantic errors arose when the target word node was relatively unavailable and semantically related ones were activated and produced instead (Caramazza, 1997). This interpretation of a word retrieval deficit at the level of the lexical form is typical for aphasic patients diagnosed with anomia. Furthermore, he presented with no grammatical class dissociations in spoken (action/object) naming neither for L1 nor for L2. This finding is in tune with other studies in the (monolingual) aphasiology literature showing that fluent anomic aphasic individuals can suffer from undifferentiated grammatical word class dissociations on picture naming tasks (see Kambanaros, 2009, and references within). However, the present finding is in contrast to previous studies of picture naming involving bilingual anomic aphasic speakers of Greek (L1) and English (L2), that show dissociations between performance on lexical retrieval tasks with nouns and verbs in both languages (Kambanaros & van Steenbrugge, 2006). Selective impairment when accessing the phonological representations of familiar words due to damage at the level of the phonological output lexicon allow impaired oral production of the names of pictures, as in AA’s case, but a spared ability in writing the name of the same pictures (see Hillis & Caramazza, 1991, 1995, for a complete discussion). However, this was observed for AA only in L1 (Greek) for written picture naming of action and object words (albeit with a significantly higher performance for action words). This means that on the written picture-naming task AA was able to select an orthographic representation based on the corresponding phonological representation and input from the intact semantic system. With regards to object names, the predominant written error was letter substitutions mainly orthographic errors, involving the grapheme [ι] /i/ (iota) in word-final position being replaced by the grapheme [η] (ita) /i/. The grapheme [ι] (iota) is the most common noun ending for singular neuter nouns while [η] (ita) is the most common noun ending for singular feminine nouns. AA’s written error on (neuter) object words, in Greek, could be considered a morphological error in the sense that he assigned feminine gender to neuter nouns but we know this was not the case for two reasons: first, he retrieved the gender of every noun correct (100%) either on the spoken naming task or when asked to provide noun gender by the examiner (author) on spoken nouns (GOAT), and second, he was able to write several other neuter nouns ending with “i” (iota) (e.g., κουτάλι “spoon”, σφουγγάρι “sponge”, καλάμι “fishing rod”, ρολόι “watch”, correctly. Furthermore all remaining errors for object words in Greek involved letter substitutions or orthographic replacements of graphemes representing the same vowel (e.g., αναπτύρας instead of αναπτήρας “lighter”. Given that Greek is considered to have a relatively transparent orthography one possible explanation for his performance on written picture naming in Greek of neuter nouns is that he produced, via the

4 1 8 Language Disorders Maria Kambanaros

non-lexical spelling or phonological route, used in transparent languages, written word-forms that were phonologically plausible alternatives, although incorrect in terms of word-specific orthography. There are several alternative spellings of the vowel “i” (and other vowels) in Greek therefore word- specific knowledge is required to write correctly any word containing (vowel) orthographic ambiguity even in Greek. Moreover, it is possible that access to the morphological rule for neuter noun declensions was partially available to AA (after stroke) and as such explains his mixed written picture naming performance mainly on (neuter) nouns. In contrast, the most plausible account of AA’s successful written naming of actions in Greek, is that it was not mediated by lexical-phonological knowledge but was instead facilitated by an independent pathway leading from the meaning of action words to their lexical-orthographic representations. It is possible that AA has developed some sort of visual orthographic memory that allows him to write correctly action words, including those containing phonemes that could be transcribed with more than one grapheme. All errors with written naming of action words involved phonological-orthographic substitutions of word-medial or final vowels (e.g., κλειδόνει for κλειδώνει “locking”, τριπάει for τρυπάει “drilling”, that were phonologically plausible alternatives.

6. Conclusion

AA presents with preserved written spelling for action words regardless of modality in Greek (L1). Notably, this contrasts with his severe difficulties with spoken picture naming of the same action words. AA’s performance when naming objects across written naming tasks revealed a similar performance in L1; he was significantly less impaired on writing object words compared to retrieving object names on a spoken naming task. In English (L2), AA showed similar performance across both tasks for action and object words, i.e. severe difficulties retrieving action and object names for spoken and written naming. The unique finding in the present study is that this pattern of dissociation for action over object names was observed in L1 (Greek) but not in L2 (English). In L2 (English) there were similar levels of impairment across tasks (spoken and/ written naming) for both word types. This finding supports the widely accepted view that the cognitive processes which are used for naming and spelling tasks are relatively independent components that can be selectively impaired and/ or spared by brain impairment. In the case of bilingual speakers, the cognitive processes can involve both languages. However, the results show for the first time that language status (L1 versus L2) interacts with grammatical category and task in the patterns of acquired naming disorders manifest in bilingual speakers with aphasia.

Language Disorders 4 1 9 Major Trends in Theoretical and Applied Linguistics

References

Caramazza, A. 1997. How many levels of processing are there in lexical access? Cognitive Neuropsychology 14, 177-208.

Caramazza, A. and Hillis, A. 1991. Lexical organization of nouns and verbs in the brain. Nature 349, 788-790.

Forster, K. and Jiang, N. 2001. The nature of the bilingual lexicon: Experiments with the masked priming paradigm. In J.N. Nicol (ed.), One Mind, Two Languages Bilingual Language Processing. Oxford: Blackwell Publishers, 72-83.

Goodglass, H. and Kaplan, E. 1983. The Assessment of Aphasia and Related Disorders (2nd edition). Lea and Febiger: Philadelphia.

Hatzigeorgiou, N., Gavrilidou, M., Piperidis, S., Carayannis, G., Papakostopoulou, A. and Spiliotopoulou, A. 2000. Design and implementation of the online ILSP corpus. Proceedings of the Second International Conference of Language Resources and Evaluation (LREC), vol. 3, 1737-1740.

Hillis, A. and Caramazza, A. 1991. Category-specific naming and comprehension impairment: A double dissociation. Brain and Language 114, 2081-2094.

Hillis, A. and Caramazza, A. 1995. Representation of grammatical categories of words in the brain. Journal of Cognitive Neuroscience 7, 457-458.

Kambanaros, M. 2003. Verb and noun processing in late bilingual individuals with anomic aphasia. Doctoral dissertation, Flinders University, Adelaide.

Kambanaros, M. 2009. Grammatical word class distinctions in bilingual aphasia. In G. Ibanescu and S. Pescariu (eds.), Aphasia: Symptoms, Diagnosis, and Treatment. New York: Nova Science Publishers.

4 2 0 Language Disorders Maria Kambanaros

Kambanaros, M. and van Steenbrugge, W. 2006. Noun and verb processing in Greek-English bilingual individuals with anomic aphasia and the effect of instrumentality and verb-noun name relation. Brain and Language 97, 162-177.

Levelt, W. 1989. Speaking. Cambridge, MA: MIT Press.

Li, P., Sepanski, S. and Zhao, X. 2006. Language history questionnaire: A webbased interface for bilingual research. Behavior Research Methods 38, 202-210.

Obler, K.L., Hyun, J.M., Conner, S.P., O’Connor, B. and Anema, I. 2007. Brain organization of language in bilinguals. In A. Ardila and E. Ramos (eds.), Speech and Language Disorders in Bilinguals. New York: Nova Science Publishers, 21-46.

Papathanasiou, I., Papadimitriou, D., Gavrilou, B. and Mihou, A. 2008. Psychometric data on the BDAE in a healthy adult population: the role of age and gender. Psychology 15, 398-410.

Ralli, A. 2003. Morphology in Greek linguistics: The state of the art. Journal of Greek Linguistics 4, 77-129.

Rapp, B. and Caramazza, A. 2002. Selective difficulties with spoken nouns and written verbs: A single case study. Journal of Neurolinguistics 15, 373-402.

Language Disorders 4 2 1 Major Trends in Theoretical and Applied Linguistics

4 2 2 Language Disorders Can Vocabulary Size Predict Narrative Abilities in Children with SLI?

Maria Kambanaros1, Kleanthes K. Grohmann2, Eleni Theodorou2 and Michalis Michaelides2

1Cyprus University of Technology, 2University of Cyprus [email protected], [email protected], [email protected], [email protected]

Abstract

Correlating expressive vocabulary skills with narrative microstructure measures such as morphosyntactic skills revealed that both are powerful tools to assess language-impaired children. However, for expressive vocabulary to have a predictive value on narrative performance, younger children with SLI should be investigated. Our results lend support to a unified model of SLI (Rice 2003) revealing that both expressive vocabulary and narrative re-tell microstructure are delayed in school-aged Greek Cypriot children with SLI compared to their typically developing peers.

1. Introduction

The larger research agenda pursued by the Cyprus Acquisition Team is to investigate typically and atypically language-developing children in Cyprus from mono- but also bi- and multilingual backgrounds. Our research so far has shown, among other things, that children with specific language impairment (SLI) have (i) smaller vocabularies, especially in relation to common, concrete object and action words (Kambanaros & Grohmann 2010), and (ii) difficulties producing oral narratives (Theodorou & Grohmann 2010, Theodorou 2012) compared to age-matched peers with typical language development (TLD). Taken together, then, narratives and vocabulary and/ or naming abilities do separately serve as diagnostic indexes for SLI. The present study aims to explore vocabulary size as a possible predictor of narrative ability in two groups of children, those with TLD and children with SLI. The study was carried out in the Republic of Cyprus and “Greek at large” is the language under investigation. As native acquirers of Cypriot Greek, the local variety of the language, Greek Cypriot children are not exposed to Standard Modern Greek before school entry, at least not within typical everyday life (other

Language Disorders 4 2 3 Major Trends in Theoretical and Applied Linguistics

than cartoons on television, for example). Cypriot Greek differs from the standard variety in substantial ways, which do not play a role here (but see e.g. Grohmann et al. 2006, Grohmann 2011, Grohmann & Leivada 2012, and Grohmann et al. 2012). The linguistic situation of (Greek-speaking) Cyprus is described as diglossia, where Standard Modern Greek enjoys sociolinguistically “high” status, while Cypriot Greek is the “low” variety. For the purposes of the present study, we did not look for differences between the two varieties but counted all Greek responses equally well regardless of their origin (whether lexical, phonological, or morphosyntactic). As such, we refer to the language under investigation as “Greek” throughout.

2. Background

The description of language disorders in SLI is usually based on (i) the characteristics of children’s spontaneous speech output and (ii) children’s performances on specific linguistic tasks tapping into the different language components under investigation (such as syntax, semantics, phonology, and morphology, but also pragmatics as well as lexicon and word finding).

2.1. Narrative and SLI

Many cross-linguistic studies have shown that children with SLI are less skilled both in composing and in producing oral narratives compared to age-matched peers with typical language development. In fact, the narratives of children with SLI are described as less competent and more immature with respect to length, lexical diversity, sentence complexity, and content (Gillam & Pearson 2004). Furthermore, difficulties with narratives is considered an area of spoken language in which delay is less likely to resolve over time for children with SLI (Manhardt & Rescorla 2002). Concerning the analysis of narratives, researchers typically differentiate between a micro- and a macrostructural analysis. Microstructure incorporates the semantic and syntactic complexity of narratives, using measures such as mean length of utterance, number of clauses per T-unit, quantity of vocabulary, and diversity of vocabulary. Macrostructure refers to the general, global characteristics of a narrative focusing on coherence and organization of the narrative.

2.2. Vocabulary and SLI

Vocabulary development in children with SLI has been examined both for a general delay of vocabulary overall and for specific delays within different

4 2 4 Language Disorders Maria Kambanaros, Kleanthes K. Grohmann, Eleni Theodorou and Michalis Michaelides

word classes, such as a particular difficulty with acquiring verbs compared to typically language-developing peers (see Rice 2003 on a unified model of SLI). A number of investigations (mainly for the English language) have shown that difficulties with word knowledge including storing, organizing and/or accessing lexical information is prominent in SLI. Several explanations have been put forward including the storage hypothesis (Leonard 1998) which suggests that children with SLI are slower to learn new words, have less elaborate semantic representations, and fewer lexical connections with words within their lexicon. Alternatively, the retrieval hypothesis (Newman & German 2002) posits that children with SLI have difficulties with lexical access because of inefficient word retrieval mechanisms and the difficulties do not lie within lexical entries or their connections.

3. Methodology

3.1. Participants

Two groups of children participated in this study: 11 children with TLD (aged between 4;6–6;11 years;months) and 10 children diagnosed with SLI (aged between 5;2–6;10). Participant demographics are presented in Table 1.

Table 1. Child demographics according to group.

TLD (n=11) SLI (n=10)

Boys (number of participants) 5 8

Girls number of participants) 6 2

Mean age (in months) 70 71

Range of ages (in months) 54–83 62–82

Standard Deviation 8.02 7.18

The children with TLD were recruited randomly from public pre-primary schools in Limassol after approval from the Ministry of Education and upon written parental consent. Children with SLI were recruited from speech and language therapists working in public schools and/or private practice in Limassol. No participating child presented with evidence of hearing loss, neurological impairment, gross motor or articulation difficulties, social, emotional or behavioral problems.

Language Disorders 4 2 5 Major Trends in Theoretical and Applied Linguistics

All children with language impairment were receiving speech and language therapy and were diagnosed with SLI by their speech and language therapists, prior to the study. A large testing battery developed to identify SLI in Cypriot Greek-speaking children (Theodorou 2013) was used in order to assess and confirm the earlier diagnosis of the participants of this study. These measures included the Raven’s Coloured Progressive Matrices (RCPM; Raven et al. 2000), which assesses non-verbal performance; the Diagnostic Verbal IQ Test (DVIQ; Tsimpli & Stavrakaki 2000), which consists of five subparts that test vocabulary (naming), comprehension and production of morphosyntax, comprehension of metalinguistic concepts, and sentence repetition; the Peabody Peabody Picture Vocabulary Test (PPVT; Dunn & Dunn 1981), which assesses children’s receptive vocabulary abilities; and the Word Finding Test (WFT; Vogindroukas et al. 2009), which assesses the extent to which pictures of objects can be named correctly. The participants’ mean scores are shown in Table 2.

Table 2. Mean scores (standard deviations) on the testing battery for SLI diagnosis.

Measures Group TLD SLI

RCPM 85.4 (standard)

DVIQ – Production of Morphosyntax 19.9 (2.11) 12.3 (2.09)

DVIQ – Comprehension of Morphosyntax 26.4 (2.46) 22.4 (1.84)

DVIQ – Sentence Repetition 46.8 (1.8) 40.8 (2.70)

DVIQ – Vocabulary Naming 22.3 (1.58) 15.7 (2.2) DVIQ – Comprehension of Metalinguistic 20.1 (2.45) 17.5 (1.29) Concepts PPVT 79.3 (32.02) 69.3 (16.63)

WFT 26.5–33.2 30.2 (SD 7.9) 27.5 (3.83)

3.2. Materials

In the present study, two tasks were correlated in order to investigate whether the vocabulary size of children with SLI can predict their narrative abilities: (i) vocabulary scores on two confrontation naming tasks, the Cypriot Object and Action Test (COAT: Kambanaros et al. 2013) and the vocabulary subtest of the DVIQ, and (ii) a narrative re-tell task using the Bus Story Test (BST: Renfrew 1969 [1997]). All tasks are described below.

4 2 6 Language Disorders Maria Kambanaros, Kleanthes K. Grohmann, Eleni Theodorou and Michalis Michaelides

3.2.1. Vocabulary Measures

The COAT, consisting of seventy-four coloured photographs measuring 10x14cm, was administered to evaluate lexical retrieval of object (35 pictures) and action names (39 pictures). Object names were single, concrete inanimate nouns and included manipulated instruments such as garage tools, garden equipment, kitchen utensils, household items, and office and personal implements used for activities of daily living. Object names were not controlled for gender: 5 nouns were masculine, 14 feminine, and 16 neuter. All selected verbs were monotransitive pictureable actions with either simple internal word structures of [root + affix] or more complex ones of [root + affix + affix]. Actions were restricted to (perhaps outdated) stereotypical roles, that is, a woman is shown performing household activities (e.g. sweeping), and a man is performing more manly duties (e.g. hammering). All action names corresponded either to an instrumental verb (where an instrument is part of the action, e.g. cutting) or to a non-instrumental verb (e.g. climbing). The vocabulary subtest of the DVIQ consists of 27 black and white drawings: 13 drawings representing common nouns that could be divided into high (e.g., book, car hat, spoon) and low frequency items (e.g., violin, bridge, binoculars, thermometer), and 14 drawings representing verb phrases (e.g., pushed the boy).

3.2.2. The Bus Story

The BST is a screening test of verbal expression which examines story retelling with picture support. It can demonstrate difficulties with verbal expression as well as phonological, semantic, grammatical, and sequencing problems. The test consists of a wordless picture book which describes the story of a naughty red bus. It is a multi-episode story. The scenario is designed to motivate the need to identify different characters and to express temporal and causal relationships between events and between characters.

3.3. Procedures

The COAT, the DVIQ vocabulary subtest, and the BST were presented in one or two sessions and in random order. All children were tested in their schools, in speech and language clinics, or in their houses in a quiet room. Testing was conducted individually. The procedure for the administration of each test follows. For the COAT, children were asked to name the object or action represented in the photograph in a single word (one-word target response). The stimulus question for the action pictures was: “Tell me in one word, what is he/she doing?”. Action names were required in the third person singular. The stimulus

Language Disorders 4 2 7 Major Trends in Theoretical and Applied Linguistics

question for the object pictures was: “Tell me in one word, what is this?”. The stimulus question was repeated once if needed. If no response was given, the item was scored as incorrect. No time limits were placed and self-correction was allowed. Responses were recorded and transcribed verbatim. For the purposes of this study errors were scored either as correct or incorrect. For the DVIQ vocabulary subtest, the “What’s this?” question was used to elicit naming of the nouns and foe the verbs the question incorporated the subject of the verb phrase (e.g., “What’s the girl/child/man/woman/grandmother doing?”). For the BST, the experimenter told each child the short story about a red bus, while the child was looking through the picture book illustrating the story. Then the experimenter asked the child to retell the story as close to the original as possible. The child then retold it, using the pictures as prompts. During the retellings, no reinforcement was provided other than encouragement with head nods and fillers. All narrations were audio-recorded and transcribed using the IPA (International Phonetic Association 1999). Utterances were then divided into sentences and evaluated with respect to five measures, three that the BST suggests and two additional ones: the amount of original information included (Information), the number of subordinate clauses (Subordinates), the mean sentence length of the longest five sentences (A5LS), the mean length of utterances based on words (MLU–word), and the total amount of sentences used (Sentences). All but the first of these are based on microstructural analysis. The detailed scoring scheme used is the following:

A. Information: The semantic complexity is measured using a norm-referenced information score that the BST provides. The number of relevant information included in the story are indicated following Renfrew’s (1969 [1997]) criteria, where ‘essential’ information gets two points and ‘subsidiary’ information gets one point. The Information score is the total number of points thus accumulated. An example would be the utterance: Το λεωφορείο έφυγε. /to leoforio efije/ ‘The bus left.’ For this production examinees would receive 3 points. They get 1 point for re-introducing the agent (i.e. the bus) and 2 for figuring out its role in the action (i.e. that it left). B. Subordinates: After each narrative was divided into sentences, the subordinate clauses produced were counted. C. A5LS: When the narratives were transcribed, they were divided into sentences (T-units). Then the MLU–word was calculated and the mean of the five longest sentences were computed. D. MLU–word: This was calculated based on words for each narrative (MLU– word for all T-units). All words were added up and the sum was divided by sentences produced. E. Sentences: This measure forms the total number of sentences used (T-unit).

4 2 8 Language Disorders Maria Kambanaros, Kleanthes K. Grohmann, Eleni Theodorou and Michalis Michaelides

4. Results

Children with SLI were less accurate than age-matched TLD peers on expressive vocabulary measures using the DVIQ. The mean scores on the DVIQ vocabulary subtest are presented in Table 2 above. For the COAT, the percentages of correct responses were calculated for object and action names as provided by all children and a summary of the results is given in Figure 1 according to picture type.

Figure 1. Mean percentages correct for object and action names by both groups.

Independent samples t-tests were performed to test for mean diff erences between the children with TLD and those with SLI. The latter showed signifi cantly

more diffi culties retrieving object 19(t =–2.735, p=0.013) and action (t19=–3.940, p=0.001) names compared to TLD peers. For the BST, the mean scores for all fi ve measures are displayed in Table 3. Children with SLI scored signifi cantly worse in three out of the fi ve measures than the children with TLD: Subordinates

(t19=–2.815, p=0.011), A5LS (t20=–2.377, p=0.028), and MLU–word (t20=– 2.942, p=0.008). For the other two measures, the mean sores indicated worse performance for the SLI group but the diff erence failed to reach signifi cance

(Information: t20=–1.663, p=0.112 and Sentences: t20=–0.437, p=0.667). Similar results were obtained for all pair-wise comparisons using a non-parametric Mann-Whitney U test. The scores on the expressive vocabulary subtest of the DVIQ and on the COAT were correlated with the number of subordinates, A5LS, and MLU-word based on the BST to determine if expressive vocabulary skills can predict children’s narrative performances at the microstructure level. The results are reported in the next section.

Language Disorders 4 2 9

Major Trends in Theoretical and Applied Linguistics

Table 3. Mean scores for the five measures.

Child TLD (n=11) SLI (n=10) Age 4;6–6;11 5;2–6;10 Mean St. Dev. Mean St. Dev. Inf. Total 29.8 7.264 23.5 9.419 A5LS 8.8 2.581 6.4 2.047 Sub. Cl. 4.1 2.071 2.0 1.054 MLU – words 5.1 1.223 3.8 0.685 No. of sentences 18.4 3.957 17.7 3.974

5. Correlations

The Pearson correlation coefficients between vocabulary measures and variables extracted from the Bus Story (MLU, A5LS, and subordinate clauses) appear in Table 4. The table contains values for the TLD group; correlation coefficients for the SLI group appear in parentheses. There is a positive relationship between the sub-scores of the COAT for both groups (significant only in the TLD group) and a non significant positive trend between the noun component of the COAT and the DVIQ-Vocabulary measure. High positive associations exist among the three BST measures; all are significant in the case of the TLD group, but only the A5LS and MLU correlation is significant for the SLI group. Not many of the estimated coefficients turned out to be significant, probably due to the small sample size of the two groups examined. In addition, scores in the SLI sample are lower with a more restricted range than in the TLD sample, hence correlations for the SLI turn out to be even smaller. Overall, no significant relationships were found between vocabulary and BST measures.

Table 4. Correlation coefficients between the vocabulary and BST measures.

DVIQ COAT COAT Subordinate A5LS MLU Vocabulary Nouns Verbs clauses 0.275 –0.007 0.098 –0.028 0.178 DVIQ Vocabulary (0.388) (–0.139) (–0.458) (–0.186) (–0.549) 0.752* 0.335 0.157 0.265 COAT Nouns (0.139) (–0.540) (–0.294) (–0.519) 0.011 –0.288 –0.002 COAT Verbs (0.200) (0.241) (0.330) 0.767** 0.946** A5LS (0.525) (0.939**) 0.669* Subordinate clauses (0.592) * Correlation is significant at the 0.05 level (2-tailed). ** Correlation is significant at the 0.01 level (2-tailed).

4 3 0 Language Disorders Maria Kambanaros, Kleanthes K. Grohmann, Eleni Theodorou and Michalis Michaelides

6. Discussion

In the present study, two expressive vocabulary measures (the vocabulary subtest of the DVIQ and the action and object naming subtest of the COAT) were correlated with three microstructure measures from the BST (subordinate clauses, A5LS, and MLU-word) in order to determine if single word expressive vocabulary can predict morphosyntactic abilities in school-aged children with TLD and those with SLI. The assumption is that children with larger expressive vocabulary scores would have higher scores on the narrative microstructure measures and children with lower expressive vocabulary skills would show the opposite pattern. Overall, our results showed no significant relation between the Bus Story measures and the expressive vocabulary scores for normal and language-impaired children. It is possible that the limited size of the present sample might have obscured some significant findings. From our search of the literature, the correlation between expressive vocabulary and morphosyntactic ability as related to narrative retell microstructure has been found significant in some studies but not in others. Specifically, for monolingual English-speaking children, this correlation was reported significant in typically developing 5-year-olds (Adams & Gathercole 1996) and language-impaired children with a mean age of 8;4 (Howlin & Kendall 1991). In contrast, studies investigating the relation between the BST measures and vocabulary scores in bilingual populations with TLD report a lack of correlation (Schelletter & Parke 2004, Simon-Cereijido & Guitierrez-Clellen 2009). (Expressive vocabulary was measured using the Word Finding Vocabulary Test for English (Renfrew 1995).) Both groups of researchers claim that vocabulary skills in bilingual children do not predict syntactic ability (at least in typical development). It thus appears that our results are comparable with studies from the bilingual literature. This perhaps unexpected finding raises the interesting topic of linguality in Greek Cypriot children. As mentioned above, Cypriot Greek does differ from Standard Modern Greek in many ways, which go beyond the sound system and lexical inventory. With clear morphosyntactic differences, such as clitic placement (Terzi 1999 and subsequent work), wh-question formation (e.g. Grohmann et al. 2006), and a host of other structural properties of the grammar (Tsiplakou et al. forthcoming), one possible conclusion might be that the two varieties are so much different as to allow classification as two separate languages. In that most extreme case, Greek Cypriots could then be characterized as bilingual speakers with Cypriot Greek as their native first language and Standard Modern Greek as possibly simultaneous bilingual acquisition or early child second language acquisition. If that were the case, the results from the present study would be in line with similar findings from bilingual acquisition cited above. However, we are not yet at a point at which we can unequivocally sustain such a stance. After all, Cypriot and Standard Modern Greek are also very close

Language Disorders 4 3 1 Major Trends in Theoretical and Applied Linguistics

to one another. In addition, qualitatively the standard variety spoken in Cyprus differs distinctly from that spoken in mainland Greece, for example. Lastly, in the context of diglossia, the distinction between the two varieties, for all practical and theoretical purposes, seems to lie in the sociolinguistic domain more so than in the formal grammatical domain, also for political reasons. We will not review the continuously growing body of work here but confine ourselves to a simple point: With no native speakers of Standard Modern Greek in the Greek Cypriot population, the bilingual classification in the traditional sense faces significant obstacles. Likewise, if the sociolinguistic situation is indeed that of diglossia, the “high” variety can hardly be a separate language (again, without changing the respective definitions accordingly). We therefore work on the assumption that in terms of linguality, Greek Cypriot speakers are “bi-x”, where ‘x’ needs to be further determined, possibly beyond ‘varietal’, ‘lingual’, ‘dialectal’, etc. (Grohmann 2011, Grohmann & Leivada 2012). One possible value for x could be ‘lectal’ (Rowe & Grohmann 2013): Greek Cypriots are bilectal speakers of (at least) two Greek lects. We hope to flesh out this idea in the near future. Whatever bi-x may turn out to be, it looks, however, as if bi-x such as bilectal speakers behave, with respect to the task at hand, more like bilingual than monolingual speakers. That is to say that competence in another lect may indeed contribute to (meta)linguistic abilities that go beyond monolingual speakers. In order to support such an approach, similar studies could be carried out in less strong settings than Cyprus, perhaps with Greek speakers from the islands such as the Dodecanese (which shows similarities to the grammar of Cypriot Greek) or Crete and compare them with mainland Greek speakers or investigate urban versus rural settings in Greece — or other countries for other languages with significant and less significant dialectal variation (German, French, English, etc.). This project is clearly one for the future, hence our conclusion at this point can only be tentative: Greek Cypriot children pattern more with bilinguals than monolinguals with respect to a possible correlation between expressive vocabulary and morphosyntactic ability as related to narrative retell microstructure.

4 3 2 Language Disorders Maria Kambanaros, Kleanthes K. Grohmann, Eleni Theodorou and Michalis Michaelides

References

Adams, A.-M. and S.E. Gathercole. 1996. Phonological working memory and spoken language development in young children. Quarterly Journal of Experimental Psychology Section A: Human Experimental Psychology 49(1), 216- 233.

Dunn, L. and L. Dunn. 1981. Peabody Picture Vocabulary Test. Circle Pines, MN: American Guidance Service.

Gillam, R.B. and N. Pearson. 2004. Test of Narrative Language. Austin, TX: Pro-Ed.

Grohmann, K.K. 2011. Some directions for the systematic investigation of the acquisition of Cypriot Greek: A new perspective on production abilities from object clitic placement. In E. Rinke and T. Kupisch (eds.), The Development of Grammar: Language Acquisition and Diachronic Change. Amsterdam: John Benjamins, 179-203.

Grohmann, K.K. and E. Leivada. 2012. Interface ingredients of dialect design: Bi-X, socio-syntax of development, and the grammar of Cypriot Greek. In A.M. Di Sciullo (ed.), Towards a Biolinguistic Understanding of Grammar: Essays on Interfaces. Amsterdam: John Benjamins, 239-262.

Grohmann, K.K., P. Panagiotidis and S. Tsiplakou. 2006. Properties of wh-question formation in Cypriot Greek. In M. Janse, B.D. Joseph and A. Ralli (eds.), Proceedings of the 2nd International Conference on Modern Greek Dialects and Linguistic Theory (Mytilene, Greece: 30 September–3 October 2004). Patras: University of Patras, 83-98.

Grohmann, K.K., E. Theodorou, N. Pavlou, E. Leivada, E. Papadopoulou and S. Martínez-Ferreiro. 2012. The development of object clitic placement in Cypriot Greek and the Romance connection. In S. Ferré, P. Prévost, L. Tuller and R. Zebib

Language Disorders 4 3 3 Major Trends in Theoretical and Applied Linguistics

(eds.), Selected Proceedings of the Romance Turn IV Workshop on the Acquisition of Romance Languages. Newcastle-upon-Tyne: Cambridge Scholars Publishing, 128-152.

Howlin, P. and L. Kendall. 1991. Assessing children with language tests — which tests to use? British Journal of Disorders of Communication 26, 355-367.

International Phonetic Association. 1999. Handbook of the International Phonetic Association: A Guide to the Use of the International Phonetic Alphabet. Cambridge: Cambridge University Press.

Kambanaros, M. and K.K. Grohmann. 2010. Patterns of object and action naming in Cypriot Greek children with SLI and WFDs. In K. Franich, L. Keil, K. Iserman and J. Chandlee (eds.), Proceedings of the 34th Boston University Child Language Development – Supplement.

Kambanaros, M., K.K. Grohmann and M. Michaelides. 2013. Lexical retrieval for nouns and verbs in typically developing bilectal children. First Language 33, 182-199.

Leonard, L. 1998. Children with Specific Language Impairment. Cambridge, MA: MIT Press.

Manhardt, J. and L. Rescorla. 2002. Oral narrative skills of late talkers at ages 8 and 9. Applied Psycholinguistics 23, 1-21.

Newman, R.S. and D.J. German. 2002. Effects of lexical factors on lexical access among typical language-learning children and children with word-finding difficulties. Language and Speech 45, 285-317.

Raven, J., J.C. Raven and J.H. Court. 2000. Manual for Raven’s Progressive Matrices and Vocabulary Scales. San Antonio, TX: Harcourt Assessment.

Renfrew, C. 1969 [1997]. The Bus Story Test: A Test of Continuous Speech, 4th ed. Oxford: North Place, Old Headington.

Renfrew, C. 1995. Word Finding Vocabulary Test. Bicester: Winslow.

Rice, M.L. 2003. A unified model of specific and general language delay: grammatical tense as a clinical marker of unexpected variation. In Y. Levy and J. Schaeffer (eds.),Language Competence across Populations: Toward a Definition of Specific Language Impairment. Mahwah, NJ: Lawrence Erlbaum, 63-94.

4 3 4 Language Disorders Maria Kambanaros, Kleanthes K. Grohmann, Eleni Theodorou and Michalis Michaelides

Rowe, C. and K.K. Grohmann. 2013. Discrete bilectalism: Towards co-overt prestige and diglossic shift in Cyprus. International Journal of the Sociology of Language 224: 119–142.

Schelletter, C. and T. Parke. 2004. Using story re-tell in bilingual assessment. Academic Exchange Quarterly 8, 158-162.

Simon-Cereijido, G. and V.F. Gutiérrez-Clellen. 2009, A cross-linguistic and bilingual evaluation of the interdependence between lexical and grammatical domains. Applied Psycholinguistics 30, 315-338.

Stavrakaki, S. and I.M. Tsimpli. 2000. Diagnostic verbal IQ test for Greek preschool and school age children: Standardization, statistical analysis, psychometric properties [in Greek]. Proceedings of the 8th Conference on Speech Therapy. Athens: Ellinika Grammata, 95-106.

Terzi, A. 1999. Cypriot Greek clitics and their positioning restrictions. In A. Alexiadou, G. Horrocks and M. Stavrou (eds.), Studies in Greek Syntax. Dordrecht: Kluwer, 227-240.

Theodorou, E. 2012. Narratives in Cypriot Greek children with SLI. In K.K. Grohmann, E. Shelkovaya and D. Zoumbalides (eds.), Linguists of Tomorrow: Selected Papers from the 1st Cyprus Postgraduate Conference in Theoretical and Applied Linguistics. Newcastle-upon-Tyne: Cambridge Scholars Publishing, 205- 231.

Theodorou, E. 2013. Diagnosing specific language impairment: The case of Cypriot Greek. PhD thesis, University of Cyprus.

Theodorou, E. and K.K. Grohmann. 2010. Narratives in Cypriot Greek mono- and bilingual children with SLI. In A. Botinis (ed.), Proceedings of ISCA Tutorial and Research Workshop on Experimental Linguistics 2010 – 25-27 August 2010, Athens, Greece. Athens: ISCA and the University of Athens, 185-188.

Tsiplakou, S., P. Coutsougera, and P. Pavlou. Forthcoming. A Contemporary Grammar of Cypriot Greek. Munich: Lincom Europa.

Vogindroukas, I., A. Protopappas, and G. Sideris. 2009. Word-Finding Test. Chania: Glafki.

Language Disorders 4 3 5 Major Trends in Theoretical and Applied Linguistics

4 3 6 Language Disorders Testing the Relativized Minimality Approach. Evidence from Wh-Question Production and Comprehension in Greek Aphasia

Michaela Nerantzini1, Despina Papadopoulou2 and Spyridoula Varlokosta1

1University of Athens, 2Aristotle University of Thessaloniki [email protected], [email protected], [email protected]

Abstract

The present study aims at investigating the abilities of Greek agrammatic speakers in producing and comprehending wh-questions, in light of the Relativized Minimality approach to locality in syntax (Garraffa and Grillo, 2008; Grillo, 2009). Our findings support the predictions of Relativized Minimality in that movement structures which involve an intervening NP are deficient and the class of features associated with the moved and the intervening element plays a significant role in intervention. However, Relativized Minimality cannot account for the fact that intervention was selective in our data, namely, it did not arise in the production of object referential questions. 1. Introduction

Typically, agrammatic Broca’s aphasia is characterized by severe morphosyntactic deficits across modalities, especially in cases where structures with non-canonical word order are involved (Bastiaanse and Grodzinsky, 2000; Caramazza and Zurif, 1976; for a review). This study focuses on one of such cases, namely the production and comprehension of wh-questions. Several cross-linguistic studies have shown that agrammatic speakers have selective difficulties when producing and comprehending wh-questions, one of which concerns better performance on subject compared to object questions (Garraffa and Grillo, 2008; Hickok and Avrutin, 1996; Martinez-Ferreiro, 2010; Neuhaus and Penke, 2008; Salis and Edwards, 2008; among others). The subject/object asymmetry has been recently interpreted within the Relativized Minimality (RM) approach to locality in syntax (Garraffa and Grillo, 2008; Grillo, 2005; 2009). RM predicts that structures which involve formation of a movement dependency over an intervening NP will be more difficult to process and thus more likely to be deficient in agrammatic aphasia. Thus, object questions

Language Disorders 4 3 7 Major Trends in Theoretical and Applied Linguistics

will be harder than subject questions, since only in the former does an NP intervene between the moved element and its trace/copy. Given that only some wh-question types have been investigated in a relatively small group of Greek speakers with aphasia so far (Fyndanis et al., 2010; Jovanov, 2011; Stavrakaki and Kouvava, 2003), the present study aims at addressing the following questions: (a) which wh-types are more deficient in Greek agrammatic aphasia; (b) are findings from Greek in line with previous cross-linguistic data; (c) do discrepancies appear across modalities; and (d) can RM account for the observed patterns.

2. Referential and Non-Referential Wh-Questions

The present study focuses on referential and non-referential subject and object questions (see 1 and 2, respectively), which are derived by movement of the wh-operator from its θ-position to the Specifier position of the complementizer phrase (Spec, CP) (Chomsky, 1981; for Greek: Horrocks and Stavrou, 1987; Theophanopoulou-Kontou, 2002).

(1a) Pjos andras kiniγa ti γineka? Referential subject

who_NOM man_NOM is-chasing the_ACC woman_ACC

[Which man]i ti is chasing the woman? (1b) Pjon andra kiniγa i γineka? Referential object

who_ACC man_ACC is-chasing the_NOM woman_NOM

[Which man]i is the woman chasing ti ? (2a) Pjos/Ti kiniγa ton andra? Non-referential subject

who_NOM/what is-chasing the_ACC man_ACC

[Who/what]i ti is chasing the man? (2b) Pjon/Ti kiniγa o andras? Non-referential object

who_ACC/what is-chasing the_NOM man_NOM

[Who/what]i is the man chasing ti?

Pjos is the interrogative pronoun used for both non-referential who and which- NP question types; it is inflected for gender, number, and case, and in referential questions it agrees with the noun it modifies. Non-referential who (pjos/pjon) and what (ti) questions differ with respect to their φ-feature properties; who (pjos/pjon) is inflected for gender, case, and number features, whereas what (ti) is not. Although the feature [+/–animate] distinguishes who and what questions in English (Alexopoulou and Keller, to appear), in Greek, both wh-words can be used to refer to animate1 and inanimate entities.

1 However, the use of what for reference to animate entities is rather marked.

4 3 8 Language Disorders Michaela Nerantzini, Despina Papadopoulou and Spyridoula Varlokosta

Referential questions, such as (1), differ from non-referential questions, such as (2), with respect to the presuppositions they make. While (1) presupposes a pre-established set of referents in the discourse representation of the speaker and the hearer (D(iscourse)-linked question; Avrutin, 2000; Pezetsky, 1987), such presupposition does not exist in (2) on the part of the speaker or the hearer. It has been proposed that D-linked phrases form binding chains, since they do not qualify as operators, and are interpreted via unselective binding, while non- D-linked phrases form chains of antecedent government relations (for details, see Cinque, 1990; Pesetsky, 1987; among others).

3. Referential and Non-Referential Wh-Questions in Agrammatism

Previous findings on wh-questions have revealed a selective deficit in agrammatism with better performance observed in (a) non-referential (who) questions compared to referential (which-NP) ones (Avrutin, 2000; Hickok and Avrutin, 1996; Jovanov, 2011; Salis and Edwards, 2008; Stavrakaki and Kouvava, 2003; but for a divergent pattern, see Goodluck, 2005; Goodluck, Stojanovic and Carter, 2008; Thompson et al., 1999); (b) referential subject questions compared to referential object ones (Garaffa, 2007; Hickok and Avrutin, 1996; among others); (c) non-referential subject questions compared to non-referential object ones (Garaffa and Grillo, 2008); and (d) what questions compared to who questions (Garaffa, 2007; Garaffa and Grillo, 2008). Recently, Grillo (2005; 2009) introduced the RM approach, based on Rizzi’s (1990, 2004) and Starke’s (2001) work, as a possible explanation for the aforementioned asymmetries in agrammatic speakers. As formally expressed in (3), RM captures the fact that a syntactic relation in natural languages is satisfied in the smallest possible environment in which it can be satisfied (Garaffa and Grillo, 2008; Grillo, 2005; Rizzi, 2004).

(3) Minimal Configuration: . . . X. . . Z. . . Y. . . Y is in a minimal configuration (MC) with X iff there is no Z such that i. Z is of the same2 structural type as X, and ii. Z intervenes between X and Y (in terms of c-command) (from Rizzi, 2004: 225)

2 Similarity in structural type is defined by a set of morphosyntactic features that each syntactic position can occupy.

Language Disorders 4 3 9 Major Trends in Theoretical and Applied Linguistics

RM essentially states that a local relation cannot be established between two elements X and Y, when an intervening element Z is of the same type as X and Y, i.e., when it carries the same morphosyntactic features, since Z will be recognized as a possible candidate for the establishment of the local relation. Grillo (2005; 2009) suggests that due to processing limitations, agrammatic speakers lack the ability to activate (or maintain) a complete array of φ-features in their syntactic representations. Consequently, agrammatic speakers are expected to perform worse in structures that involve movement of an element from its original/base generated position to another position, when movement crosses over a potential intervening element (Grillo, 2005: 107). Crucially, RM effects are expected to be enhanced when the features of the intervening elements belong to the same class (argumental, quantificational, modifiers, topic), as opposed to cases where their set of features belongs to a different class (Garraffa and Grillo, 2008: 181). Garraffa and Grillo (2008) observed two asymmetries in their data from an Italian-speaking agrammatic speaker; higher accuracy (a) on non-referential subject questions than on object ones and (b) on object what questions than on object who questions. The interpretation they provide is within the framework of RM. Regarding the first asymmetry, in non-referential object questions (2b), a processing breakdown affects agrammatic speakers’ access to φ-features associated with the moved element, leading to a chain formation blockage, since the operator class element (who) is considered as an argumental class element (Garaffa and Grillo, 2008: 191). With respect to the latter asymmetry, in object what questions, the moved [-animate] object differs from the intervening [+animate] NP in animacy, helping agrammatic speakers to construct correct syntactic representations. In contrast, in object who questions, the same [+animate] feature is involved in the moved wh-word and the intervening NP, causing severe difficulties to agrammatic speakers (see Garraffa and Grillo, 2008).

4. The Present Study

In the present study, we assessed referential and non-referential wh-questions across modalities by using an elicitation and a picture-pointing task, in order to investigate whether RM could effectively account for the Greek agrammatic data.

4.1. Participants

Three monolingual male Greek-speaking agrammatic individuals, aged 56-61 years (mean age: 58), and a control group of non brain-damaged participants, matched to the brain-damaged speakers on age, sex, and educational background, participated in this study. The brain-damaged participants had

4 4 0 Language Disorders Michaela Nerantzini, Despina Papadopoulou and Spyridoula Varlokosta

suffered a left ischemic CVA at least 12 months post onset (see Table 1) and were selected for inclusion on the basis of the Boston Diagnostic Aphasia Examination (BDAE) (Goodglass and Kaplan, 1983; Greek version by Papathanasiou, Feidatsi, Katsantoni, Panagiotopoulou, and Malefaki, 2004), in addition to the clinical consensus of both a neurologist and a speech and language therapist.

Table 1. Participants’ profile

Education Post-onset Gender Age Type of lesion (years) (years) P1 Male 61 8 1 Left CVA

P2 Male 56 17 4 Left CVA, IFG included

P3 Male 58 12 2,5 Left CVA, IFG included

C1 Male 57 10

C2 Male 55 16

C3 Male 56 12

4.2. Elicitation task

4.2.1. Materials and Procedure

An elicitation task was designed, which consisted of 52 questions in total, 20 non- referential pjos/pjon (=who/whom) and 12 non-referential ti (=what) (4), as well as

20 referential pjos jatros/pjon jatro (=which_nom doctor_nom/which_acc doctor_acc) (5), half with a subject and half with an object dependency. The task was implemented by means of a computer; the participants were shown a picture with one (for non- referential questions) or two (for referential questions; see Fig.1) reversible hidden agents on a computer screen, while a sentence was orally presented to them. In particular, non-referential (subject/object) questions were elicited by using an indefinite pronoun, i.e. kapjos/kapjon “someone”, kati “something” (see 4). It is important to note that the sentences provided by the experimenter did not prompt speakers to the target wh-words. Participants were instructed to complete the missing part of the presented sentences by providing the correct question type.

(4) a. Non-referential (who) question elicitation

Subject Experimenter: Kapjos kiniγa ton zoγrafo. Θelume na maθume afton pu kiniγa ton zoγrafo. Etsi rotame…

Language Disorders 4 4 1 Major Trends in Theoretical and Applied Linguistics

“Someone is chasing the painter. We want to know the one that is chasing the painter. Thus, we ask…” Target response: Pjos kiniγa ton zoγrafo?

who_NOM is-chasing the painter_ACC “Who is chasing the painter?”

Object Experimenter: O stratiotis kiniγa kapjon. Θelume na maθume afton pu kiniγa o stratiotis. Etsi rotame… “The soldier is chasing somebody. We want to know the one that the soldier is chasing. Thus, we ask…” Target response: Pjon kiniγa o stratiotis?

who_ACC is-chasing the soldier_NOM “Who is the soldier chasing?”

b. Non-referential (what) question elicitation

Subject Experimenter: Kati epese apo to δentro. Θelume na maθume afto pu epese apo to δentro. Etsi rotame… “Something fell off the tree. We want to know the thing that fell off the tree. Thus, we ask...” Target response: Ti epese apo to δentro?

what fell off the_ACC tree_ACC “What fell off the tree?”

Object Experimenter: O papus δjavazi kati. Θelume na maθume afto pu δjavazi o papus. Etsi rotame… “The grandpa is reading something. We want to know the thing that the grandpa is reading. Thus, we ask...” Target response: Ti δjavazi o papus?

what is-reading the_NOM grandpa_NOM “What is the grandpa reading?”

Figure 1. Referential question elicitation

4 4 2 Language Disorders Michaela Nerantzini, Despina Papadopoulou and Spyridoula Varlokosta

(5) Referential question elicitation

Subject Experimenter: Enas zoγrafos kiniγa ton stratioti. Θelume na maθume ton zoγrafo pu kiniγa ton stratioti. Etsi rotame… “A painter is chasing the soldier. We want to know the painter that is chasing the soldier. Thus, we ask…” Target response: Pjos zografos kiniγa ton stratioti?

who_NOM painter_NOM is-chasing the soldier_ACC “Which painter is chasing the soldier?”

Object Experimenter: O stratiotis kiniγa enan zoγrafo. Θelume na maθume ton zoγrafo pu kiniγa o stratiotis. Etsi rotame… “The soldier is chasing a painter. We want to know the painter that the soldier is chasing. Thus, we ask…” Target response: Pjon zoγrafo kiniγa o stratiotis?

who_ACC painter_ACC is-chasing the soldier_NOM “Which painter is the soldier chasing?”

The same agents, with the same 10 verbs3 were used in all conditions in order to eliminate possible lexical effects. In addition to that, a naming task was conducted prior testing in order to ensure that participants were familiar with the vocabulary denoting the agents of each picture. Four practice trials were included, as well. Repetition of the cueing sentence was accepted upon request, although feedback as to the accuracy was not given during the administration of the task. Additionally, self-corrections were allowed and only the final answer was counted as correct response.

4.2.2. Scoring

Two different analyses were adopted. In the first analysis, all well-formed wh- questions, including even partially elliptical structures, and all pragmatically felicitous answers based on the introductory sentence presented by the experimenter (6a) were considered as correct. The second analysis consisted of more strict scoring criteria; thus, only target responses were counted as correct (6b).

3 All verbs included were transitive ones. In order to eliminate any frequency effects, the same verbs and agents were used across conditions.

Language Disorders 4 4 3 Major Trends in Theoretical and Applied Linguistics

(6) O stratiotis kiniγa enan zoγrafo. Θelume na maθume ton zoγrafo pu kiniγa o stratiotis. Etsi rotame… “The soldier is chasing a painter. We want to know the painter that the soldier is chasing. Thus, we ask...” (a) Answer: O stratiotis pjon kiniγa?

the soldier_NOM who_ACC is-chasing “Who is the soldier chasing?” (b) Answer: Pjon zoγrafo kiniγa o stratiotis?

who_ACC painter_ACC is-chasing the soldier_NOM “Which painter is the soldier chasing?” Target response: Pjon zoγrafo kiniγa o stratiotis?

who_ACC painter_ACC is-chasing the soldier_NOM “Which painter is the soldier chasing?”

4.2.3. Results

Since the control group performed at ceiling, their data will not be discussed. With respect to the agrammatic data, we will initially describe the data obtained from the first analysis (see Table 2), in which well-formed and pragmatically felicitous wh-questions were considered as correct. No significant differences were observed between subject and object dependencies across question types (referential: χ2 = 1.491, p = .222; η2 = .158; who/what: χ2 = .000, p = 1.000; η2 = .000). Moreover, there were no significant differences between object referential and object who questions (χ2 = 3.750, p = .053; η2 = .250) or between object who and object what questions (χ2 = .291, p = .590; η2 = .078).

Table 2. Overall scores on wh-question elicitation (well-formed questions)

which- which- who- who- what- what- NP_subject NP_object subject object subject object

Mean 0.83 0.70 0.90 0.90 0.94 0.94 (Sdv) (0.38) (0.47) (0.31) (0.31) (0.24) (0.24)

In the second analysis (see Table 3), where only target responses were considered as correct, the main finding was that the brain-damaged participants exhibited severe difficulties in the production of certain wh- question types. Subject questions were significantly more accurate than object questions in the case of non-referential who questions (χ2 = 18.373, p = .000; η2 = .553) but not in the case of referential ones (χ2 = .067, p = .795; η2 = .034) or in the case of what questions (χ2 = 1.125, p = .289; η2 = .177). Although the difference between object referential and object who

4 4 4 Language Disorders Michaela Nerantzini, Despina Papadopoulou and Spyridoula Varlokosta

questions was not significant (χ2 = 1.684, p = .194; η2 = .168), the difference between object who and object what questions turned out to be significant (χ2 = 15.451, p = .000; η2 = .467).

Table 3. Overall scores on wh-question elicitation (target responses)

which- which- who- who- what- what- NP_subject NP_object subject object subject object

Mean 0.57 0.53 0.90 0.37 0.83 0.94 (Sdv) (0.50) (0.51) (0.31) (0.49) (0.38) (0.24)

The individual data replicate these findings to a great extent. As shown in Table 4, low performance was attested in both object referential and non- referential who questions across participants. All three participants performed more accurately on subject non-referential questions compared to object counterparts, but the preponderance of subject movement was not evident in the case of what questions or in the case of referential questions. Last, object what questions were more accurately produced than object who questions by all participants.

Table 4. Individual data on wh-question elicitation

which-NP_ which-NP_ who- who- what- what- subject object subject object subject object

P1 4/10 3/10 9/10 2/10 6/6 6/6

P2 4/10 7/10 10/10 4/10 6/6 6/6

P3 9/10 6/10 8/10 5/10 3/6 5/6

Furthermore, the participants’ errors reveal three consistent patterns:

(a) non-referential questions were produced instead of referential questions (see (6a) above); (b) a relative clause with a wh-word in nominative case was produced instead of an object wh-question (7a); (c) a what question headed by the wh-word ti, which is unmarked for case, number, and gender, was produced instead of a who object question (7b).

Language Disorders 4 4 5 Major Trends in Theoretical and Applied Linguistics

(7) O stratiotis kiniγa kapjon. Θelume na maθume afton pu kiniγa o stratiotis. Etsi rotame… “The soldier is chasing somebody. We want to know the one that the soldier is chasing. Thus, we ask…” Answers: (a) pjos ine aftos pu kiniγa o stratiotis?

who_NOM.MASC.SG is this_NOM.MASC.SG that is-chasing the soldier_NOM “Who is the one that the soldier is chasing?” (b) ti kiniγa o stratiotis?

what is-chasing the soldier_NOM “What is the soldier chasing?” Target response: pjon kiniγa o stratiotis?

who_ACC is-chasing the soldier “Who is the soldier chasing?”

4.3. Picture-Pointing Task 4.3.1. Materials and Procedure

The picture-pointing task consisted of 90 semantically reversible sentences in different question types: 30 non-referential pjos/pjon (=who/whom), 30 non-

referential ti (=what), and 30 referential pjos jatros/pjon jatro (=which_nom doctor_

nom/which_acc doctor_acc), half subject and half object questions. The task was implemented on a computer; the participants were presented with one black- and-white drawing on a computer screen, such as the one illustrated in Figure 2, while a question was orally presented to them (cf. 8 and 9). The participants were asked to point to the correct agent of the action. The drawings depicted semantically reversible actions performed by two out of three animate agents (people or animals) always of the same gender (grammatical and semantic) (Fig. 2). All 90 experimental items were randomized and presented to each participant in the same order. The same 10 verbs with the same agents were used in all conditions in order to eliminate erroneous picture identification effects.

Figure 2. (Non-)Referential question comprehension

4 4 6 Language Disorders Michaela Nerantzini, Despina Papadopoulou and Spyridoula Varlokosta

(8) Non-referential questions: who subject: Pjos trava ton aθliti? “Who is pulling the athlete?” who object: Pjon trava o aθlitis? “Who is the athlete pulling?” what subject: Ti trava ton aθliti? “What is pulling the athlete?” what object: Ti trava o aθlitis? “What is the athlete pulling?” (9) Referential questions: which-NP subject: Pjos jatros trava ton aθliti? “Which doctor is pulling the athlete?” which-NP object: Pjon jatro trava o aθlitis? “Which doctor is the athlete pulling?”

4.3.2. Results

The comprehension task revealed that all brain-damaged participants performed relatively high across conditions, except in the referential which-NP object questions (see Table 5). The object conditions were significantly less accurate than the subject conditions for both referential (χ2 = 5.414, p = .020; η2 = .245) and non-referential who questions (χ2 = 4.939, p = .026; η2 = .234). However, in the case of what questions, the object condition was significantly more accurate than the subject condition (χ2 = 4.186, p = .041; η2 = .216). Furthermore, although no significant difference was observed between object referential and object non-referential who questions (χ2 = 1.111, p = .292; η2 = .111), performance on object who questions was significantly worse than performance on object what questions (χ2 = 7.590, p = .006; η2 = .290).

Table 5. Overall scores on wh-question comprehension

which-NP_ which-NP_ who- who- what- what- subject object subject object subject object

Mean 0.93 0.76 0.98 0.84 0.91 1.00 (Sdv) (0.25) (0.44) (0.15) (0.37) (0.29) (0)

The individual data replicate the observed overall pattern to a great extent. All three participants had lower performance on object referential questions compared to subject counterparts, whereas P1 had also lower performance on object non-referential who questions compared to subject ones. Last, all three participants performed equally well on subject and object what questions. In comparing modalities, our data revealed that difficulties that agrammatic speakers have with wh-questions are more pronounced in production than in comprehension, since overall better performance was attested in the comprehension task than in the elicitation task (χ2 = 16.784, p = .000; η2 = .198).

Language Disorders 4 4 7 Major Trends in Theoretical and Applied Linguistics

Table 6. Individual data on wh-question comprehension

which- which- who- who- what- what- NP_ NP_ subject object subject object subject object P1 15/15 12/15 15/15 9/15 13/15 15/15

P2 14/15 12/15 14/15 15/15 15/15 15/15

P3 13/15 10/15 15/15 14/15 13/15 15/15

5. Discussion

The present study investigated the ability of three Greek-speaking agrammatic participants to produce and comprehend wh-questions in order to evaluate (a) which question types are harder for Greek patients; (b) whether these findings are in line with previous cross-linguistic data; (c) whether similar performance emerges across modalities; and (d) whether RM can efficiently account for the observed patterns. Overall, production was found severely impaired in comparison to comprehension, suggesting that the two modalities may not be affected to the same extent in agrammatism. With respect to production, our data revealed that participants were able to produce wh-questions, by developing error-reducing strategies in cases where they could not produce the target question type (see analysis 1, Table 2). Accuracy on the referential questions was considerably lower in the second analysis compared to the first one (76.50% and 55% correct performance in the first and in the second analysis, respectively). This finding indicates that agrammatic speakers avoid using referential questions and falls in with previous studies that revealed particular difficulties with referential questions (Avrutin, 2000; Hickok and Avrutin, 1996; Jovanov, 2011; Salis and Edwards, 2008; among others, for comprehension data). The production data pointed to two further asymmetries when only target responses are analyzed (see analysis 2, Table 3): one between non-referential subject/object who questions, with better performance on subject questions compared to object ones, and another one between object who/what questions, with better performance on what compared to who questions. Moreover, accuracy scores on (both referential and non-referential who) object questions in the second analysis are substantially reduced compared to those in the first analysis (80% and 45% for the first and for the second analysis, respectively), a fact which suggests that the Greek agrammatic speakers eluded producing a wh-word

marked with accusative case, i.e. pjon (who_ACC). This is further corroborated

4 4 8 Language Disorders Michaela Nerantzini, Despina Papadopoulou and Spyridoula Varlokosta

by the participants’ higher performance on what than on who questions (in line with Garaffa, 2007; Garaffa and Grillo, 2008; but contra Salis and Edwards, 2008), which are headed by a wh-word that does not carry any φ-features, i.e. ti (=what). The comprehension data revealed an asymmetry between subject and object questions, with the former being significantly better than the latter for both referential and non-referential who cases. Hence, our findings suggest that the difficulties the Greek agrammatic speakers experience with object questions are not constrained in the domain of comprehension (for comprehension difficulties with object referential questions in other languages, see Avrutin, 2000; Hickok and Avrutin, 1996; Salis and Edwards, 2008; among others) but are evident in the domain of production as well. In fact, the difficulties of Greek agrammatic speakers in the production of referential questions were present in subject dependencies as well, eliminating subject/object asymmetries in this modality. Last, the advantage of object what over object who questions attested in production was also replicated in comprehension. Despite the differences between the production and comprehension of referential questions, a clear generalization has emerged across modalities, namely better performance on questions that involve subject extraction than those involving object extraction. Let us now see whether RM can provide a unified account of our data. The severe problems the agrammatic speakers face with object questions can easily be accommodated within the framework of RM. This theory predicts better performance on subject questions instead of object ones, given that only in the latter case an NP intervenes between the moved element and its trace/ copy. Importantly, this effect is expected in referential and non-referential who questions. In both cases, the activation of φ-features in agrammatic speakers is slowed down and, hence, the set of features associated with the syntactic element to be moved changes from Operator to Argumental, leading to an erroneous performance. Grillo (2009) provides an account for the deficient performance observed in referential questions in comparison to non-referential ones (see Avrutin, 2000; Hickok and Avrutin, 1996; Salis and Edwards, 2008). In particular, he claims that referential questions are encountered as inherently more complex structures (op. cit: 1436), since, if Starke’s (2001) feature hierarchy is adopted, referential which-x phrases are associated with the S[pecific]Q feature that constitutes a subset of the Q feature class of who elements. By hypothesizing that processing limitations will lead agrammatic speakers to the most economical (related to markedness) interpretation, it is expected that more general features, i.e. Q which is related to who wh-words, will posit less difficulties, instead of more specific ones, such as SQ linked with which wh-words (see Grillo, 2005; 2009). However, it is not clear how such an interpretation can account for the lack of

Language Disorders 4 4 9 Major Trends in Theoretical and Applied Linguistics

a dissociation between object referential and object who questions in both modalities. Finally, RM predicts better performance on object what rather than on object who questions, given that the mismatch in animacy between the moved item and the intervening subject in what questions helps agrammatic speakers differentiate the two NPs and succeed in forming the syntactic chain. This prediction is confirmed by the production and comprehension data. In Greek, it is not animacy per se that causes the mismatch in non-referential questions; it is rather the lack of φ-features associated with ti (=what), which clearly distinguishes ti from the intervening NP. To conclude, the present findings from Greek agrammatic speakers on the production and comprehension of wh-questions support the predictions of RM in that (a) movement structures which involve an intervening NP are deficient and (b) the class of features associated with the moved and intervening element plays a significant role in the formation of the syntactic chain. Nevertheless, it is not clear how RM can account for the fact that the blockage effect of the intervening NP is selective in our data, namely, why it does not emerge in the production of object referential questions We leave this issue open to be further investigated with a larger number of agrammatic speakers.

Acknowledgements

This research has been co-financed by the European Union (European Social Fund – ESF) and Greek national funds through the Operational Program “Education and Lifelong Learning” of the National Strategic Reference Framework (NSRF) – Research Funding Program: THALIS – UOA – “Levels of impairment in Greek aphasia: Relationship with processing deficits, brain region, and therapeutic implications” – Principal Investigator: Spyridoula Varlokosta. The authors are listed alphabetically.

4 5 0 Language Disorders Michaela Nerantzini, Despina Papadopoulou and Spyridoula Varlokosta

References

Alexopoulou, T., and Keller, F. to appear. What vs. who and which: Kind-denoting fillers and the complexity of whether-islands. In N. Hornstein and J. Sprouse (eds.), Experimental Syntax and Island Effects. Cambridge: Cambridge University Press.

Avrutin, S. 2000. Comprehension of discourse-linked and non-discourse-linked questions by children and Broca’s aphasics. In Y. Grodzinsky, L. Shapiro, and D. Swinney (eds.), Language and the Brain: Representation and Processing. San Diego: Academic Press, 295-313.

Bastiaanse, R., and Grodzinsky, Y. 2000. Grammatical Disorders in Aphasia. A Neurolinguistic Perspective. London and Philadelphia: Whurr Publishers.

Caramazza, A., and Zurif, E. 1976. Dissociation of algorithmic and heuristic processes in language comprehension: Evidence from aphasia. Brain and Language 3: 572-582.

Chomsky, N. 1981. Lectures on Government and Binding. Dordrecht: Foris.

Cinque, G. 1990. Types of A’-Dependencies. Cambridge, MA: MIT Press.

Fyndanis, V., Varlokosta, S., and Tsapkini, K. 2010. Exploring wh-questions in agrammatism: Evidence from Greek. Journal of Neurolinguistics 23: 644-662.

Garraffa, M. 2007. Impoverishment of grammatical features in a non-fluent aphasic speaker: The grammatical nature of minimal structures. Ph.D. dissertation. University of Siena.

Garraffa, M., and Grillo, N. 2008. Canonicity effects as a grammatical phenomenon. Journal of Neurolinguistics 21: 177-197.

Language Disorders 4 5 1 Major Trends in Theoretical and Applied Linguistics

Goodglass, H., and Kaplan, E. 1983. Boston Diagnostic Aphasia Examination. Malvern: Lea and Febinger.

Goodluck, H. 2005. D(iscourse)-linking and question formation: Comprehension effects in children and Broca’s aphasics. In A.M. Di Sciullo (ed.), UG and External Systems: Language, Brain and Computation. Amsterdam: John Benjamins.

Goodluck, H., Stojanovic, D., and Carter, D. 2008. Comprehension of D(iscourse)- linked questions by aphasics: New evidence. In Proceedings of the Annual Conference of the Canadian Linguistics Association.

Grillo, N. 2005. Minimality effects in agrammatic comprehension. In S. Blaho, E. Schoorlemmer, and L. Dicente (eds.), Proceedings of XIV Console.

Grillo, N. 2009. Generalized minimality: Feature impoverishment and comprehension deficits in agrammatism. Lingua 119: 1426-1443.

Hickok, G., and Avrutin, S. 1996. Comprehension of wh-questions in two Broca’s aphasics. Brain and Language 52: 314-327.

Horrocks, G., and Stavrou, M. 1987. Bounding theory and Greek syntax: Evidence for wh-movement in NP. Journal of Linguistics 23: 79-108.

Jovanov, N. 2011. A morpho-syntactic study of patients with Broca’s aphasia: A comparative study of Serbian- and Greek-speaking patients. Ph.D. dissertation, Aristotle University of Thessaloniki.

Martinez-Ferreiro, S. 2010. The cartography of Ibero-Romance agrammatic deficits. Biolinguistics 4.4: 324-355.

Neuhaus, E., and Penke, M. 2008. Production and comprehension of wh-questions in German Broca’s aphasia. Journal of Neurolinguistics 21: 150-176.

Papathanasiou, I., Feidatsi, M., Katsantoni, M., Panagiotopoulou, E., and Malefaki, S. 2004. The validation of Boston Diagnostic Aphasia Examination (BDAE-3) in Greek. Paper presentation at the 26th World Congress of the International Association of Logopaedics and Phonetics, 29 August-2 September 2004, Brisbane, Australia.

Pesetsky, D. 1987. Wh-in-situ: Movement and unselective binding. In E.J. Reuland and A.G.B. ter Meulen (eds.), The Representation of (In)definiteness. Cambridge, MA: MIT Press.

4 5 2 Language Disorders Michaela Nerantzini, Despina Papadopoulou and Spyridoula Varlokosta

Rizzi, L. 1990. Relativized Minimality. Cambridge, MA: MIT Press.

Rizzi, L. 2004. Locality and the left periphery. In A. Belletti (ed.), Structure and Beyond. New York: Oxford University Press.

Salis, C., and Edwards, S. 2008. Comprehension of wh-questions and declarative sentences in agrammatic aphasia: The set partition hypothesis. Journal of Neurolinguistics 21: 375-399.

Starke, M. 2001. Move dissolves into merge, a theory of locality. Ph.D. dissertation, University of Geneva.

Stavrakaki, S., and Kouvava, S. 2003. Functional categories in agrammatism: Evidence from Greek. Brain and Language 86:129-141.

Theophanopoulou-Kontou, D. 2002. Geneticí Sídaksi: To Prótipo tis Civérnisis ce Anaforikís Désmefsis. [Generative Syntax: Government and Binding Theory]. Athens: Kardamitsa Press.

Thompson, C.K., Tait, M., Ballard, K., and Fix, S. 1999 Agrammatic aphasic subjects’ comprehension of subject and object extracted wh a questions. Brain and Language 67: 169-187.

Language Disorders 4 5 3 Major Trends in Theoretical and Applied Linguistics

4 5 4 Language Disorders Linguistic Processing and Executive Control: Evidence for Inhibition in Broca’s Aphasia

Eleni Peristeri and Ianthi Maria Tsimpli

Aristotle University of Thessaloniki [email protected], [email protected]

Abstract

We investigated the hypothesis that inhibition is impaired in a group of Broca’s aphasics relative to a group of language-unimpaired controls. We focused on distractor interference and spatial negative priming (NP) effects, which involve the ability to inhibit (i) distracting information and (ii) irrelevant information that exerts its effect from one trial to the next, respectively. The results reflect a distractor interference effect which is in line with suggestions that inhibition is impaired in aphasia. The robust NP effect in the aphasics is accounted for in terms of a deficit in processing information on a local (vs. global) level of analysis.

1. Introduction

There is an increasing recognition that the communication problems one observes in persons with aphasia (henceforth, PWA) extend beyond verbal deficits and that the myriad of symptoms observed are not solely due to a faulty linguistic system. Rather, there exists a coalition of causal elements resulting in a wide range of communicative deficits. There is some preliminary evidence suggesting that communicative success of PWA may depend on the integrity of executive function skills (Ramsberger 1994). Executive functions are called into play when an individual is involved in a complex, novel activity. They allow us to plan, sequence, organize, and monitor goal-directed activities in a flexible manner as demanded by situational and environmental changes (Lezak 1995; Phillips 1997; Ylvisaker & Feeney 1998). Ramsberger (2000: 93) suggests that ‘‘executive functions may serve an important mediating role in the complicated task of human communication especially when routine processing schemas are no longer viable due to primary speech and language processing disorders’’.

Language Disorders 4 5 5 Major Trends in Theoretical and Applied Linguistics

When linguistic skills are impaired, individuals need to rely on other cognitive skills in order to communicate. For example, they need to be able to monitor their attempts to communicate and judge whether or not their communicative partner understood the message based on feedback the partner provides. In addition, cognitive flexibility is required to switch to another modality such as gesture or writing to get their point across when verbal attempts fail. Glosser and Goodglass (1990) were among the first researchers to specifically examine executive functioning ability in PWA. They administered four experimental executive function procedures to 22 left-brain-damaged, 19 right- brain-damaged, and 49 healthy controls. Brain-damaged subjects were divided into groups according to site of lesion: prefrontal, retrorolandic, and mixed. The test procedures included the non-verbal continuous performance test, graphic pattern generation, sequence generation task, and the Tower of Hanoi. Results indicated that subjects with left frontal lobe lesions were significantly more impaired than subjects with left retrorolandic or mixed left hemisphere lesions. In addition, results suggested that the observed impairments were independent of the subjects’ linguistic deficits. Other studies have demonstrated that PWA have difficulty in various aspects of cognition that fall under the umbrella of executive functioning, such as working memory, attention, and problem solving. Working memory may be considered a foundation for executive abilities (Conner, MacKay & White 2000). It provides storage and workspace for information, thus, permitting interactions between attention, perception, and memory (Baddeley 1992). The executive controller performs operations on information held in working memory so that this information may be used effectively. Several researchers have found memory deficits in PWA (Beeson, Bayles & Rubens 1990; Beeson et al. 1993; Burgio & Basso 1997; Risse, Rubens & Jordan 1984; Ween, Verfaellie & Alexander 1996). Beeson and colleagues (1993) assessed memory using a semantic encoding procedure to provide a means to examine cued recall in addition to free recall, since a comparison of free recall versus cued recall provides information regarding the integrity of executive control processes that direct the encoding and retrieval of long-term memories. They found differences in performance of subjects with anterior versus posterior lesions on the free recall test (long-term memory) but not the cued recall test. They suggested that cued recall tasks do not require the same self-directed retrieval demands as free recall tasks and that the poor verbal long-term memory performance in subjects with anterior lesions may best be explained as a reflection of executive control deficits associated with frontal lobe damage. Attentional processes are also embedded in the construct of executive functioning. LaPointe and Erickson (1991) used a dual task paradigm to address allocation of attentional processes in subjects with aphasia. They presented their subjects with an auditory work identification task under two separate

4 5 6 Language Disorders Eleni Peristeri and Ianthi Maria Tsimpli

conditions. Under the first condition, the subjects simply indicated when they heard the target word. Under the second condition, the subjects listened for the target word while simultaneously sorting cards according to color. Results showed that all subjects performed with high accuracy under the first condition. However, under the dual task condition, subjects with aphasia performed with significantly less accuracy than healthy controls. Poorer performance during the dual task condition was believed to be due to the increased demands placed on their attentional system, which LaPointe and Erikson believed to be limited in capacity. Thus, LaPointe and Erickson concluded that variability in linguistic performance may be accounted for by inadequate attentional resources and/ or misallocation of attentional resources. Prescott et al. (1987) investigated complex problem solving in 10 subjects with aphasia using the Tower of Hanoi. In this task, subjects were given rules regarding how they may or may not go about moving disks on a peg to reach a stated goal. Results showed that 30% of the subjects with aphasia did not complete the task. Those that did required more moves and significantly more time compared to healthy control subjects. Finally, a number of Stroop-like tasks have been used to measure automatic activation, suppression/inhibition, vigilance/goal maintenance and attentional switching (engagement/disengagement) processes in normal and aphasic pathological populations. Two group studies in PWA have been identified so far (Wiener, Connor & Obler 2004; McNeil et al. 2010). Wiener and colleagues investigated the inhibition process at the lexical-semantic level of language processing in 5 participants with Wernicke’s aphasia and 12 normal adults using a numerical Stroop-like task. They found a Stroop effect for both groups; however, the interference effect for the PWA was significantly larger than for the controls. They concluded that persons with Wernicke’s aphasia had normal automatic activation and a selective deficit of inhibitory control. However, their study had a very small sample size and did not directly test the alternative hypothesis that the overall slowness of the PWA might be due to slow lexical activation in addition to impaired inhibition. In the second study McNeil et al. (2010) investigated the effects of several color-word congruent and incongruent Stroop tasks, within the context of a reading comprehension test in 25 PWA and 29 normal adults. Reading times, percentage of correct responses and reading comprehension scores were examined for the color words. The PWA demonstrated RT ratios that reflected costs in sustained attention, attentional switching mechanisms and interference/suppression effects for the specific language comprehension task. The majority of the aforementioned studies have addressed either the global construct of executive functioning, or isolated cognitive skills (like working memory, attention, and problem solving) under the domain of executive functioning, via analysis of accuracy of performance at a rather conscious, more ‘controlled’ level of processing. In comparison, the inhibitory mechanism

Language Disorders 4 5 7 Major Trends in Theoretical and Applied Linguistics

responsible for the reduced allocation of resources in aphasia and the resultant selective processing impairments has so far received little attention. The purpose of the present study is to further explore the inhibitory mechanism that serves to limit the generation and maintenance of irrelevant information in a group of Greek-speaking individuals diagnosed with Broca’s aphasia. It is hypothesized that PWA will not only perform less accurately on the task testing inhibition of distracting stimuli compared to neurologically normal controls, but that their performance will also be less efficient in terms of their reaction time (RT) performance.

2. Method

2.1. Subjects

A group of nine Greek-speaking agrammatic aphasic subjects (all males, mean age: 65.6 years – SD: 16.7) were tested for this study. The patients were recruited from chronic referrals to a neurohabilitation unit situated in the Greek province. Their diagnosis as Broca’s aphasics was based on the administration of the Boston Diagnostic Aphasia Examination Battery-Short Form (BDAE-SF) translated and adapted into Greek by Tsapkini, Vlahou and Potagas (2010), the Bilingual Aphasia Test translated into Greek by Paradis and Kehayia (1987), as well as a series of sentence-picture matching tasks testing the patients’ comprehension of various non-canonical syntactic structures (Peristeri 2007). All the aphasic subjects were right-handed, with the exception of a single patient who was bimanual. None of the subjects had a history of prior neurological disease, drug or alcohol abuse, psychiatric disorders, developmental speech/language disorders or learning disabilities. What should also be mentioned is that the aphasic group was relatively heterogeneous with respect to the educational background of its patients (mean years of education: 9.6; SD: 4.8), who were distributed as follows: four primary school, one secondary school, and four high school and beyond. Eight of the patients had a documented pathology confirmed by CT, and they were clinically assessed as Broca’s aphasics in terms of impairment and disability. More specifically, the CT scans revealed that seven of the individuals had suffered from a left hemisphere, ischemic stroke leaving a deep lesion in Broca’s area; two of the same patients have also suffered from a second incident at the same brain area a few years after the first ischemic stroke. The eighth and the ninth patient suffered from an extensive haemorrhage of the left basal ganglia caused by an aneurysm and a surgical intervention, respectively. All

4 5 8 Language Disorders Eleni Peristeri and Ianthi Maria Tsimpli

subjects were tested between one and twenty-one years post-stroke, while none had received language treatment in the past. Seven of the patients also had right hemiparesis. Finally, an informal evaluation of each patient’s audio- recorded motor speech samples collected during the study revealed that five of the patients presented with apraxia of speech, i.e. they tended to produce phonetic transformations and pseudo-phonemes, while one of them presented with extensive distortions of sounds and neologisms. Details on each patient’s biographical and neurological profile are reported in Tables 1(A&B) below. The control group consisted of eighteen adult males (mean age: 66 years – SD: 13.1), all of them native speakers of Greek. The specific adults were purposefully selected so as to match most closely the mean age and educational level of the aphasic experimental group in order to ensure that any deviation observed between the performance of the controls and the nine aphasic patients would not be biased by factors such as age and educational level.

2.2. Stimuli, Apparatus, Design, and Procedure

The online non-verbal task which was used to test the inhibitory mechanism in the group of the agrammatic patients in comparison to the language-unimpaired controls was adapted from a negative priming (henceforth, NP) paradigm initially designed (Treccani et al. 2009) to test the hypothesis of balanced bilinguals’ greater efficiency in inhibitory control. NP emerges when the irrelevant information in the prime trial becomes relevant in the next, probe trial. Slower or/and inaccurate responses are usually observed in this condition due to the fact that inhibition of the irrelevant information tends to exert its effect from the prime to the probe trial. NP has already been found to be reduced in people with deficits in inhibitory abilities, like patients with Alzheimer disease (Amieva et al. 2004). Most of the information regarding the design of the specific task is derived from Treccani et al.’s (2009) paper with balanced bilinguals. Each trial was initiated by pressing the space bar of the computer keyboard. Immediately afterward, four horizontal white lines were drawn on the black background of the screen: two lines on the top and the other two on the bottom. These lines remained on the screen throughout the trial and marked the locations at which the target and the possible distractor could appear. After 1500 msec from the lines’ onset, the prime display appeared for 150 msec: Two symbols, an X and an O were drawn on the screen, each in one of the four marked locations. Responses consisted of pressing one of four red-labeled keys (“d,” “c,” “k,” and “m” characters on the computer keyboard), each spatially compatible with one of the four marked positions on the screen. Half of the participants in each group were asked to press the key the position of which corresponded to the X (i.e.,

Language Disorders 4 5 9 Major Trends in Theoretical and Applied Linguistics Subject#5 M 79 6.1 R weakness 17 R/L - Left-hemisphere ischemic Left-hemisphere area, all Broca’s stroke: with deep extension to white matter, of some involvement subcallosal fasciculus, insular structure, putamen, anterior limb of internal capsule; superior lower involving extension motor and sensory 2/3 of anterior and post cortex, gyrus and supermarginal angular gyrus part of Subject#4 M 87 6.1 R weakness 4 R - Left-hemisphere ischemic Left-hemisphere left of (occlusion stroke artery): middle cerebral area, Broca’s most of fasciculus, arcuate left anterior and post gyrus, and supermarginal angular gyrus part of Subject#3 M 68 20 R weakness 9 R mild oral apraxia apraxia mild oral effortful by characterized speech, articulatory distortions and repetitions Extensive haemorrhage haemorrhage Extensive basal ganglia: the left of area Broca’s most of Subject#2 M 47 4.3 R weakness 16 R mild oral apraxia apraxia mild oral effortful by characterized speech with attempts to and abnormal self-correct prosody Left-hemisphere Left-hemisphere large ischemic stroke: area, lesion in Broca’s with deep extension subcallosal involving lesion in fasciculus, across insular structure isthmus, and temporal to some super extension gyrus and supermarginal angular gyrus Subject#1 M 57 3.5 None 12 R mild oral apraxia apraxia mild oral mostly by characterized articulatory omissions and additions Left-hemisphere ischemic Left-hemisphere lesion in Broca’s stroke: with deep extension area and cortical involving the gyrus of subcortical temporal- inferior left parietal lobe and left basal ganglia Patient Gender Age (Y) Time Since Onset (Y.M) Hemiparesis Education (Y) Handedness AOS of Speech) (Apraxia Lesion Table 1a. Table 1-5). (Patients the participants in aphasic group of information and neurological Demographic

4 6 0 Language Disorders Eleni Peristeri and Ianthi Maria Tsimpli 10.8 None Subject#9 12 M R 34 - Extensive haemorrhage of of haemorrhage Extensive basal ganglia: most of the left area Broca’s 2.2 R weakness Subject#8 6 M R 69 severe oral apraxia apraxia oral severe distortion of by characterized speech sounds and neologisms Left-hemisphere ischemic Left-hemisphere ½ left area, all Broca’s stroke: ½ temporal fasciculus, arcuate area; isthmus, ½ Wernicke‘s white matter, periventricular putamen, insular structure, global pallidus, anterior limb super internal capsule, of premotor, lesion involving motor and sensory cortex, gyrus; anterior supermarginal supplementary motor area, 24 cingulated gyrus area incid. 10.9 incid. 7.9 st nd 1 2 R weakness Subject#7 7 M R 76 - Left-hemisphere ischemic Left-hemisphere area, Broca’s frontal left stroke: to left with deep extension motor horn and lower frontal cortex incid. 1.1 incid. 3.1 st nd 2 R weakness Subject#6 4 M R 74 mild oral apraxia characterized characterized apraxia mild oral articulatory substitutions, by and distortions omissions 1 Left-hemisphere ischemic Left-hemisphere hemisphere patchy left stroke: temporal lesion involving isthmus and posterior portion putamen and insular area; of and posterior supermarginal with deep angular gyrus areas, body of of to border extension ventricle lateral left Hemiparesis Patient Education (Y) Gender Handedness Age (Y) AOS of Speech) (Apraxia Time Since Onset Time Since (Y.M) Lesion Table 1b. Table 6-9). (Patients the participants in aphasic group of information and neurological Demographic

Language Disorders 4 6 1 Major Trends in Theoretical and Applied Linguistics

the target) location and to ignore the O (i.e., the distractor) location, whereas the opposite mapping was assigned to the rest of the participants. After either the response was made or the time granted for responding elapsed (3000 msec after the prime’s onset), there was a 350-msec interval before the probe display appeared. This display could exhibit either the target and the distractor or just the target. Again, the probe display remained visible for 150 msec; a maximum of 3000 msec from the probe’s onset was given to respond, and a 350-msec interval ended the trial. The screen was then cleared, and a prompt to begin the next trial was presented. The experiment was run using the E-Prime software (Schneider et al. 2002). On the whole, there were 144 possible prime–probe combinations. Each combination was presented once, and the 144 prime–probe trials were presented randomly in two blocks of 72 trials. Critical conditions were those in which both the probe target and distractor appeared in previously vacant locations (i.e., target+distractor/unrelated conditions; 24 trials), those in which the probe target appeared in a previously vacant location and no distractor was presented in the probe display (i.e., only-target/unrelated conditions; 24 trials), and those in which the probe target appeared in the previous distractor location— either when a probe distractor appeared in a previously vacant position (i.e., target+distractor/D→T conditions; 24 trials) or no probe distractor was present (i.e., only-target/D→T conditions; 12 trials). Examples of these four conditions are shown in Figures 1-4 below.

(prime) (probe)

Figure 1. Critical condition 1: the target and the distractor in the probe slide appear in previously vacant positions (P(rime)-P(robe)_unrelated)

(prime) (probe)

Figure 2. Critical condition 2: the probe target appears in a previously vacant location and no distractor is present in the probe display (P-P_unrelated)

4 6 2 Language Disorders

Eleni Peristeri and Ianthi Maria Tsimpli

(prime) (probe)

Figure 3. Critical condition 3: the probe target appears in the previous distractor location and a probe distractor appears in a previously vacant position (D→T, P-P_related)

(prime) (probe)

Figure 4. Critical condition 4: the probe target appears in the previous distractor location and no probe distractor is present (D→T, P-P_related)

2.3. Data Analysis

Though the specifi c task provided data on a great number of experimental conditions, due to limitations of space we are only going to focus on the analysis of the distractor and the NP eff ect in both the agrammatic and the language- unimpaired population. More specifi cally, in order to evaluate the distractor eff ect, both groups’ RTs in the target+distractor trials were contrasted with those in only-target trials, separately for the two critical prime-probe relationship conditions (P-P related, P-P unrelated), i.e., Condition 1 vs. Condition 2 for the P-P unrelated contexts, and Condition 3 vs. Condition 4 for the P-P related contexts. On the other hand, in order to evaluate NP eff ects in the aphasic and the language-unimpaired group, the participants’ RTs in prime-probe related trials were contrasted with these in prime-probe unrelated trials, separately for the two distractor presence conditions (present, absent), i.e., Condition 1 vs. Condition 3 for the contexts whereby the distractor is present in the probe slide, and Condition 2 vs. Condition 4 for the contexts whereby the distractor is absent. Apart from the subjects’ RTs, both analyses were performed on the participants’ percentages of errors as well.

Language Disorders 4 6 3 Major Trends in Theoretical and Applied Linguistics

3. Results

Table 2 shows mean RTs and error percentages for the probe trials as a function of the distractor presence and prime–probe (P–P) location relationship. Planned comparisons were performed on the data from the four critical conditions to test possible differences in the distractor and NP effects between the two groups. Between-groups differences in the effect of each of the two relevant factors (distractor presence and P–P location relationship) were tested for each level of the other factor.

Table 2. Spatial NP task: mean Reaction Times (RTs in msecs) and percentages of errors of the group of control and Broca participants as a function of distractor presence and Prime-Probe location relationship

Distractor Prime-Probe RTs Percentages of Group Presence Relationship (SDs) Errors (SDs)

417.4 0.44% Controls Target + Distractor Unrelated (128.3) (1.8) D→T (related 498.6 0.89% positions) (181.2) (2.5) 369.0 1.78% Only-target Unrelated (122.8) (3.4) D→T (related 443.5 0.94% positions) (179.0) (4.0) 1031.6 9.00% Broca Target + Distractor Unrelated (558.1) (15.7) D→T (related 1300.0 20.67% positions) (762.9) (26.7) 983.7 4.44% Only-target Unrelated (480.9) (13.3) D→T (related 1430.3 9.67% positions) (843.7) (22.4)

In order to evaluate the distractor effect, language-unimpaired subjects’ and aphasic individuals’ RTs in target+distractor trials were contrasted with those in only-target trials, separately for the two critical P–P location relationship conditions (unrelated, D→T). There was a marginally significant interactive effect only for the D→T trials (F (1, 25) = 3.964, p = .055) and a main effect of distractor (absence vs. presence of distractor) (F (1, 25) = 4.155, p = .050) for the unrelated P-P location relationship trials. We then conducted separate repeated measures ANOVAs for each group. The controls showed a significant distractor effect in both P-P related and unrelated location relationship trials (F (1, 17) = 7.776, p = .013, and F (1, 17) = 4.552, p = .050, respectively), while the Broca patients did not exhibit significant distractor effects in either condition (p = .395 for the P-P related, and p = .298 for the P-P unrelated trials).

4 6 4 Language Disorders Eleni Peristeri and Ianthi Maria Tsimpli

The same analyses were performed on percentages of errors. With respect to the P-P related trials, a main effect of distractor was observed (F (1,25) = 4.388, p = .046), as well as a significant interaction between group and distractor (F (1, 25) = 4.477, p = .044). No significant interaction or effect was observed for the P-P unrelated trials. The repeated measures ANOVAs which were conducted separately for the two groups have not revealed a significant distractor effect for either population. To compare NP effects in the two groups, language-unimpaired subjects’ and Broca’s aphasic individuals’ RTs in D→T trials were contrasted with those in unrelated trials, separately for the two distractor conditions (absent vs. present). There was a significant main effect of NP and a significant interaction between NP and group for both distractor conditions, i.e., independently of whether the distractor was present or not (F (1, 25) = 18.123, p = .000) for the NP effect - (F (1, 25) = 5.199, p = .031) for the two-way interaction between NP and group for the [+distractor] trials, and (F (1, 25) = 24.456, p = .000) for the NP effect - (F (1, 25) = 12.475, p = .002) for the interaction between NP and group for the [-distractor] trials). Separate repeated measures analyses for each group revealed that the controls exhibited a NP effect only in the trials whereby the distractor was present, while the Broca group tended to show a significant NP effect independently of whether the distractor was present or not(controls: (F (1, 17) = 11.248, p = .004 for the [+distractor] trials, Broca group: (F (1, 8) = 6.233, p = .037) for the [+distractor] trials, and F (1, 8) = 11.899, p = .009 for the [-distractor] trials). Percentages of errors underwent the same analyses that were performed on RTs. There was a significant main effect of NP (F (1, 25) = 8.467, p = .007) and a significant interaction between NP and group (F (1, 25) = 7.270, p = .012), yet, only for the trials whereby the distractor was present. Separate repeated measures analyses for each group revealed that it was only the Broca group that has exhibited a marginally significant NP effect for the trials whereby the distractor was present in the probe slide (F (1, 8) = 4.066, p = .070).

4. Discussion

In the current study, a NP non-verbal task was chosen to represent rather subconscious inhibitory processes within executive functioning in an attempt to further understand executive function skills in individuals with Broca’s aphasia. The considerably greater distractor interference effect observed in the Broca group relative to the language-unimpaired group supports the hypothesis that Broca patients do demonstrate characteristics of an impaired inhibitory mechanism, i.e. inability to sufficiently inhibit distracting and irrelevant information. Such deficit was identified primarily in their speed of performance measured to

Language Disorders 4 6 5 Major Trends in Theoretical and Applied Linguistics

discover how quickly subjects could identify the target and inhibit the distractor at the same time; the patients’ high error rates in the target+distractor trials (e.g. 20.67% errors for the Broca group in the D→T trials vs. 0.89% errors for the control group in the same experimental condition), together with the patients’ RT ratios that reflected costs in interference/suppression effects and, possibly, the attentional switching mechanisms for this online non-verbal task evince that the Broca’s aphasic individuals had a less efficient inhibitory mechanism than the language-unimpaired subjects. The overall pattern of results from the experimental conditions testing NP across the two groups revealed that the Broca group exhibited a considerably more robust NP effect relative to the controls, i.e. their performance tended to slow down and be considerably more erroneous for the probe trials wherein the target appeared in the previous distractor location independently of whether a distractor was present in the probe trial or not. Interestingly, the same finding was reported for the balanced bilingual (vs. monolingual) group in Treccani et al.’s (2009) study and it was accounted for in terms of the notorious bilingual advantage in reducing cost of conflict resolution; early, simultaneous consolidation of two language systems has been widely argued to bring about enhanced monitoring processes and reduced conflict costs in bilingual populations (e.g. Tao et al. 2011). It seems that the Broca patients’ accentuated NP effect may stem from the patients’ over-reliance on the distractor position (rather than the content of the activated information) which prevented them from monitoring and timely detecting the target information in the probe trial. By this account, the aphasic individuals might have been more focused on the location of the lines being filled with a target or/and a distractor rather than on the type of the critical information in question and, therefore, they proved to be less efficient in executing correct responses and performing within normal time limits in their responses. Hence, although at this point it seems rather difficult to disentangle the exact factors that may have led to specific enhancement of the NP effect in the Broca group, the present study shows the necessity to investigate the same patients’ performance in tasks that would show a similar local (vs. global) processing deficit in linguistic phenomena. Such structures could involve discourse comprehension or/and grammaticality judgment tests with morphological case violations, thus, testing the patients’ performance in phenomena supporting processing at a more global and local level, respectively. Most interestingly, statistical analyses conducted on an individual basis (not reported here due to limitations of space) revealed that the NP effect was significant for eight out of the nine aphasic individuals that have participated in the present study. Such finding strongly implies that the specific performance is generalizable to the whole Broca group irrespective of neurological correlates, age, educational level and localization of injury. Of course, a relatively small subject pool was used in this study. A larger sample of individuals with Broca’s

4 6 6 Language Disorders Eleni Peristeri and Ianthi Maria Tsimpli

aphasia would allow us to examine the inhibitory mechanism and the NP effect in Broca’s aphasia more effectively.

5. Conclusions

Subjects with Broca’s aphasia in this study did demonstrate deficits consistent with an impaired inhibitory mechanism. Furthermore, the Broca group has exhibited a strong NP effect which may be accounted for in terms of a local processing deficit forcing the patients to ignore crucial informational cues related to the content of both relevant and irrelevant information at a non-verbal level. However, continued research is needed to more systematically explore the nature of the NP effect in Broca’s aphasia. Additionally, studies should strive to operationally define the subprocesses of NP in Broca’s aphasia and its interrelationships with the mechanism of inhibition, as well as with linguistic processing.

Language Disorders 4 6 7 Major Trends in Theoretical and Applied Linguistics

References

Amieva, H., Phillips L.H., Della Sala S. and Henry J.D. 2004. Inhibitory functioning in Alzheimer’s disease. Brain, 127: 949-964.

Baddeley, A. 1992. Working memory. Science, 255: 556-559.

Beeson, P., Bayles K. and Rubens A. 1990. Memory impairment associated with stroke and aphasia. Paper presented at the Annual Meeting of the Speech- Language-Hearing Association, Seattle, WA.

Beeson, P., Bayles K., Rubens A. and Kaszniak A. 1993. Memory impairment and executive control individuals with stroke-induced aphasia. Brain and Language, 45: 253-275.

Burgio, F. and Basso A. 1997. Memory and aphasia. Neuropsychologia, 35: 759- 766.

Conner, L., MacKay A. and White D. 2000. Working memory: A foundation for executive abilities and higher-order cognitive skills. Seminars in Speech and Language, 21(2): 109-119.

Glosser, G. and Goodglass H. 1990. Disorders in executive functions among aphasic and other brain damaged patients. Journal of Clinical and Experimental Neuropsychology, 12(4): 485-501.

LaPointe, L. and Erickson R. 1991. Auditory vigilance during divided task attention in aphasic individuals. Aphasiology, 5(1): 1-9.

Lezak, M. 1995. Neuropsychological assessment. New York: Oxford University Press.

4 6 8 Language Disorders Eleni Peristeri and Ianthi Maria Tsimpli

McNeil, M., Kim A., Lim K.Y., Pratt S., Kendall D., Pompon R.H., Szuminsky N., Fassbinder W., Sung J.E., Kim H.S., Hamer K. and Dickey M. 2010. Automatic activation, interference and facilitation effects in persons with aphasia and normal adult controls on experimental CRTT-R-Stroop tasks. Paper presented at the40th Clinical Aphasiology Conference (CAC), Isle of Palms, U.S.A.

Paradis, M. and Kehayia E. 1987. The assessment of bilingual aphasia (Greek version). Lawrence Erlbaum.

Peristeri, E. 2007. Agrammatism in Greek: aspects of production and comprehension. In E. Agathopoulou, M. Dimitrakopoulou and D. Papadopoulou (ed.), Selected Papers on Theoretical and Applied Linguistics. School of English, Aristotle University of Thessaloniki, 464-475.

Phillips, L.H. 1997. Do ‘‘frontal tests’’ measure executive function? Issues of assessment and evidence from fluency tests. In P. Rabbit (ed.), Methodology of frontal and executive function. Hove, UK: Psychology Press, 191-213.

Prescott, T., Gruber J., Olson M. and Fuller K. 1987. Hanoi revisited. In R. Brookshire (ed.), Clinical aphasiology. Minneapolis, MN: BRK Publishers, 249-325.

Ramsberger, G. 1994. Functional perspective for assessment and rehabilitation of persons with severe aphasia. Seminars in Speech and Language, 15: 1-16.

Ramsberger, G. 2000. Preface. Seminars in Speech and Language, 21(2): 93.

Risse, G., Rubens A. and Jordan L. 1984. Disturbances of long-term memory in aphasic patients. Brain, 107: 605-617.

Schneider, W., Eschman A. and Zuccolotto A. 2002. E-Prime user’s guide. Pittsburgh: Psychology Software Tools, Inc.

Tao, L., Marzecova A., Taft M., Asanowicz D. and Wodniecka Z. 2011. The efficiency of attentional networks in early and late bilinguals: the role of age of acquisition. Frontiers in Psychology, 2 (123): doi: 10.3389/fpsyg.2011.00123.

Treccani, B., Argyri E., Sorace A. and Della Sala S. 2009. Spatial negative priming in bilingualism. Psychonomic Bulletin and Review, 16(2): 320-327.

Tsapkini, K., Vlachou C. and Potagas C. 2010. Adaptation and validation of standardized aphasia tests in different languages: Lessons from the BDAE-Short Form in Greek. Behavioral Neurology, 22: 111-119.

Language Disorders 4 6 9 Major Trends in Theoretical and Applied Linguistics

Ween, J.E., Verfaellie M. and Alexander M. 1996. Verbal memory function in mild aphasia. Neurology, 47: 795-801.

Weiner, D.A., Connor L.T. and Obler L.K. 2004. Inhibition and auditory comprehension in Wernicke’s aphasia. Aphasiology, 18: 599-609.

Ylvisaker, M. and Feeney T.J. 1998. Collaborative brain injury intervention: Positive everyday routines. San Diego: Singular.

4 7 0 Language Disorders