Quick viewing(Text Mode)

SKY Journal of Linguistics Is Published by the Linguistic Association of Finland (One Issue Per Year)

SKY Journal of Linguistics Is Published by the Linguistic Association of Finland (One Issue Per Year)

SKY Journal of is published by the Linguistic Association of Finland (one issue per year). Notes for Contributors Policy: SKY welcomes unpublished original works from authors of all nationalities and theoretical persuasions. Every manuscript is reviewed by at least two anonymous referees. In addition to full-length articles, the journal also accepts short (3-5 pages) ‘squibs’ as well as book reviews. of Publication: Contributions should be written in either English, French, or German. If the article is not written in the native language of the author, the language should be checked by a qualified native speaker. Style Sheet: A detailed style sheet is available from the editors, as well as via WWW at http://www.ling.helsinki.fi/sky/skystyle.shtml. Abstracts: Abstracts of the published papers are included in Linguistics Abstracts and Cambridge Scientific Abstracts. SKY JoL is also indexed in the MLA Bibliography. Editors’ Addresses (2005): Pentti Haddington, Department Finnish, Information Studies and Logopedics, P.O. Box 1000, FIN-90014 University of Oulu, Finland (e-mail [email protected]) Jouni Rostila, German Language and Culture Studies, FIN-33014 University of Tampere, Finland (e-mail [email protected]) Ulla Tuomarla, Department of Romance , P.O. Box 24 (Unioninkatu 40B), FIN-00014 University of Helsinki, Finland (e-mail [email protected]) Publisher: The Linguistic Association of Finland c/o Department of Linguistics P.O. Box 4 FIN-00014 University of Helsinki Finland http://www.ling.helsinki.fi/sky The Linguistic Association of Finland was founded in 1977 to promote linguistic research in Finland by offering a forum for the discussion and dissemination of research in linguistics, both in Finland and abroad. Membership is open to anyone interested in linguistics. The membership fee in 2005 was EUR 25 (EUR 15 for students and unemployed members). Members receive SKY Journal of Linguistics gratis. Cover design: Timo Hämäläinen 1999

SKY Journal of Linguistics 18

Suomen kielitieteellisen yhdistyksen aikakauskirja Tidskrift för den Språkvetenskapliga föreningen i Finland Journal of the Linguistic Association of Finland

Editors: Pentti Haddington Jouni Rostila Ulla Tuomarla

Advisory editorial board: Raimo Anttila Juhani Härmä Martti Nyman UCLA University of Helsinki University of Helsinki

Markku Filppula Esa Itkonen Mirja Saari University of Joensuu University of Turku University of Helsinki

Auli Hakulinen Fred Karlsson Helena Sulkala University of Helsinki University of Helsinki University of Oulu

Orvokki Heinämäki Ulla-Maija Kulonen Marketta Sundman University of Helsinki University of Helsinki University of Turku

Marja-Liisa Helasvuo Marja Leinonen Kari Suomi University of Turku University of Tampere University of Oulu

Tuomas Huumo Jussi Niemi Maria Vilkuna University of Turku University of Joensuu Research Institute for the Languages of Finland Irma Hyvärinen Urpo Nikanne Jan-Ola Östman University of Helsinki Åbo Akademi University University of Helsinki

2005

ISSN 1456-8438

Tampere University Print - Juvenes Print Tampere 2005 Contents

A Note from the Editors...... 5

Pauli Brattico A Category-free Model of Finnish Derivational ...... 7

Ann Delilkan Head-dependent Asymmetry: Feet and Fusion in Malay...... 47

Agurtzane Elordui & Igone Zabala Terminological Variation in Basque: of Texts of Different Degrees of Specialization...... 71

Heidi Koskela Invoking Different Types of Knowledge in “Personal Interest” Interviews ...... 93

Donna L. Lillian Homophobic : A ‘Popular’ Canadian Example ...... 119

Francesca Masini Multi- Expressions between and the : the Case of Italian -particle Constructions...... 145

Ana Ibáñez Moreno A Semantic Analysis of Active Accomplishment of Movement...... 175

Maarit Niemelä Voiced Direct Reported in Conversational Storytelling: Sequential Patterns of Stance Taking...... 197

Timothy Osborne Coherence: A Dependency Analysis ...... 223

Michael Putnam An Anti-Local Account of Why Scrambled Datives in German Can’t Bind Anaphors...... 287

Jari Sivonen An Exercise in Cognitive Lexical : The Case of the Finnish Motion Verb Kiertää ...... 311 Remarks and Replies:

Fred Karlsson Nature and Methodology of Grammar Writing ...... 341

Esa Itkonen Concerning the Synthesis between Intuition-based Study of Norms and Observation-based Study of Corpora ...... 357

Book Reviews:

Aikhenvald, Alexandra (2004) . Reviewed by Heiko Narrog...... 379

Etymologie, Entlehnungen und Entwicklungen. Festschrift für Jorma Koivulehto zum 70. Geburtstag. Herausgegeben von Irma Hyvärinen, Petri Kallio und Jarmo Korhonen. Reviewed by Marc Pierce ...... 389

Kaisa Häkkinen (2004) Nykysuomen etymologinen sanakirja. Reviewed by Ante Aikio...... 393

Anne-Marie Simon-Vandenbergen, Miriam Taverniers and Louise J. Ravelli (2003) Grammatical Metaphor: Views from systemic . Reviewed by Asunción Villamil ...... 407

Anna Vogel (2004) Swedish Dimensional . Reviewed by Heli Tissari ...... 417

A Note from the Editors

This year’s SKY JoL comes thicker and covers a more varied range of topics than perhaps ever. The papers deal with , , critical , dependency theory, , and prosody, to but a few areas; and with languages as varied as Basque, English, Finnish, German, Italian, and Malay. Despite our forebodings expressed in the bulletin of the Linguistic Association of Finland a few months ago, there was no shortage of papers by Finnish authors; indeed, there was a significant increase in their number. Papers of Finnish origin are important in that their topics often reflect what is current in the field of linguistics in Finland, and are therefore likely to be interesting to our readers. We have also had the opportunity to introduce a Remarks and Replies section, with the exchange between Fred Karlsson and Esa Itkonen paving the way for further contributions of this kind. The papers published in this section are not peer-reviewed, and contributions to it may also take up less broad questions than is the case in this year’s issue. We have been delighted to see that our Book Review section has kept growing and has attracted reviews of highly interesting and varied works. The year’s balance has not, however, been solely positive. Low-quality submissions have also kept flowing in, and too many contributors seem to think that it is the editors’ job to polish their papers into publication shape. Due to attitudes of this kind, and the continually rising number of submissions, the editors’ work load has steadily increased – even to the extent that three editors no longer seems enough. Fortunately, the above criticism does not apply to all our contributors. We editors have again had the privilege of collaborating with many authors and referees dedicated to their work.

Pentti Haddington Jouni Rostila Ulla Tuomarla

Pauli Brattico

A Category-free Model of Finnish Derivational Morphology

Abstract

The lexicon is traditionally understood as consisting of lexical items, which are categorized in lexical categories such as verbs, or adjectives. Recently, this assumption has been challenged by a theory which posits no lexical categories in the lexicon. Rather, lexical items are taken to be categorially underspecified roots. This article presents a theory of Finnish word formation which supports, and is based on, a category-free model of the lexicon. It is argued that the category neutral layer of word formation in Finnish is recursive, hence likely to be part of the syntax proper. Some implications are discussed.

1. Introduction Recently, some linguists have argued that the lexicon contains categorially underspecified roots, rather than nouns, verbs and adjectives (Adger 2003, Alexiadou 2001, Barner & Bale 2002, Chomsky 1970, Chomsky 2004, Giegerich 1999, Farrell 2001, Jeanne & Hale 2000, Marantz 1997, 1999, 2000, Hale & Keyser 2002, Harley & Noyer 1999, Pesetsky & Torrego 2004, Salo 2003, Whitman 2004).1,2 According to one such view (e.g.,

1 Preparation of this paper was supported by Finnish Cultural Foundation and the Academy of Finland (project number 106071). The basic ideas presented here were first formulated in my PhD thesis (Salo 2003). The present paper was first presented at the SKY symposium “The lexicon: its status in the ” (2004, Turku). I thank the audience of the symposium for comments. Saara Huhmarniemi, Christina Krause, Markus Mattsson, Jouni Rostila and two anonymous referees gave me valuable comments at various aspects of this work. I can’t imagine what this work would be like without their time and patience. Julie Uusnarkaus helped me with the English language, although all the remaining mistakes are of course mine. 2 Jouni Rostila (p.c.) pointed out that this proposal is essentially compatible with various Construction Grammar approaches. Within Construction , lexemes can be conceived of as having rich frame-semantic meanings, and their categorial properties

SKY Journal of Linguistics 18 (2005), 7–45

8 PAULI BRATTICO

Marantz 1997, 1999, 2000), categorially underspecified roots become nouns, verbs and adjectives when inserted into specific syntactic contexts. The question then arises whether this new model can be applied to Finnish or, more interestingly, whether properties of Finnish word formation support or reject such a view. Furthermore, we want to know whether the category free model could help us to solve some of the remaining mysteries of Finnish word formation. I argue that the answer to both questions is positive. Section 3 presents one category free model of Finnish word formation and section 4 applies the theory to a number of open problems in Finnish. What emerges is not so much a completely new model of Finnish word formation, but a fine-tuning of the standard theory (e.g. Karlsson 1983) under a somewhat different theoretical orientation. Before going to the main business of this article, I want to clarify some terminological matters and explicate certain features of my theoretical orientation. I take “lexicon-as-listedness” to refer to elements which are the output of no grammatical rules. It is a storage of all primitive elements, often referred to as “linguistic features.” These can include derivational , inflectional morphemes, lexical category features, agreement features, semantic features, concepts, wider cognitive categories and phonological features − in short, all the grammatical elements that are primitive. This list has to be composed by means of empirical, not conceptual investigation. Some of these feature combinations produce possible and phrases in a language. Lexicon as ‘derivational morphology’ refers to the output of applying word formation rules to the elements in the lexicon-as-listedness, so that the resulting objects constitute the domain of syntax proper. To follow standard terminology, I call them lexemes. Intuitively, these constitute ‘possible words’ in a language, assuming that inflection takes place in syntax. ‘Psycholinguistic lexicon,’ or lexicon-in-use, is the domain of cognitive processes involving actual language use, , the effects of word frequency, automatization, and so forth. This list contains a catalogue of words memorized by an individual speaker or a list of words shared by a community of speakers. These are potentially very large feature bundles chunked together. Take the word juoksu-tta- ‘run-CAUSE.’ This is a complex lexeme from the perspective of derivational morphology, but it might constitute

arise when they are unified with schematic constructions. Since schematic constructions are Construction Grammars' means of capturing syntactic phenomena, this means in practice that categorial properties arise in the syntax. A CATEGORY-FREE MODEL OF FINNISH DERIVATIONAL MORPHOLOGY 9 one atomistic unit in the psycholinguistic sense if, for example, it is used with considerable frequency. It might, furthermore, be accessed by using a whole-word strategy instead of a decompositional strategy in an actual context of use (see Baayen 1993). We expect and will later find a lot of variation between speakers concerning the constitution of their lexicon-in- use. From the perspective of lexicon-as-listedness, juoksuttaa contains at least three elements: juokse- (the base), -tta- (causative ) and -a (the marker for the first ). It could be found to contain much more, if we are allowed to penetrate deeper into its structure. The word hamma- sta-a ‘teeth-CAUSE’ is a potential word in Finnish as well, yet it is seldom (if ever) used in modern Finnish, so it belongs to the lexicon as derivational morphology, but possibly not to the typical lexicon-in-use. It might belong to the lexicon-in-use of a dentist specialized in gerontology, for example. Unless otherwise stated, ‘lexicon’, as it is used here, does not mean lexicon in the psycholinguistic sense, but in the linguistic sense. I recognize that this orientation differs significantly from those of many others. For what it is worth, my motivation for separating the lexicon-in-use from the two other linguistic notions is the fact that no theory of language use can be restricted only to the linguistic domain. For instance, almost any type of cognitive material can be stored and manipulated as a single item or by means of a “whole-item strategy,” given enough practice and repetition (Logan 1988). Thus, to understand the notion of lexicon-in-use, we would need a general psychological theory of automatization, rather than only a linguistic theory of word formation. To assume that such a general theory could succeed without the more fundamental linguistic theory is as mistaken as the assumption that chess psychology could start without acknowledging the rules of chess, but equally it makes little sense to try to explain such general psychological matters only inside of the linguistic domain. Hence, some current trends notwithstanding, I think that we have to keep the two domains distinct. Ultimately, the issue is empirical. We will have the chance to return to this matter on several occasions. Since the lexicon-as-listedness contains only primitive features, it cannot contain linguistic elements which have been assigned to some lexical category; such elements are automatically ‘complex’ by virtue of being composed out of something plus a lexical category feature such as V, N or A. The interesting question here is whether morphemes, derivational or inflectional, are complex elements in the sense of being provided with one 10 PAULI BRATTICO of these categories. I will argue in this paper that derivational morphemes and the resulting complex lexemes are categorially underdetermined. To introduce some terminology, by “categorially underspecified Root” I mean a morpheme, stem, or lexeme which does not belong to any lexical category. This terminology comes from Giegerich (1999). Intuitively speaking, these are constituents of words that do not contain features such as +N, V, A. To follow Pesetsky (1995), I use the notation √ROOT to refer to Roots. Roots are lexical elements which can become verbs, nouns and adjectives when they become Words. Words are lexical items which are, in themselves and without further ado, pronounceable as grammatically well- formed units.3 Whether a given Root becomes a verb, or an depends on its syntactic context. For instance, the Finnish Root √BUY can become a verb (osta-a), noun (osta-minen, ost-o) or adjective (osta-va). Osta- is the phonological exponent of a categorially underspecified Root that cannot be pronounced alone without certain minimal inflectional markers and/or a marked stress pattern.4 Each Root projects thematic roles, which are associated with argument DPs (determiner phrases). The syntactic realization of these DPs depends on the categorial status of the Root, as shown in these Finnish examples (1a−d) (all examples, unless otherwise stated, are from Finnish). When the Root is combined with relevant inflectional material, they become phonological Words, which are pronounceable as such.

(1) a. isä osta-a auto-n (V) father-NOM buy-3SG car-ACC ‘the father buys a/the car’

b. isä-n auto-n osta-minen (N) father-GEN car-GEN buy-N.NOM ‘the buying of a/the car by the father’

3 While also Roots, like cranberry morphemes, have pronounceable phonological exponents, these elements are not grammatical and understandable without further affixation. 4 The root osta- by itself can be used as an imperative form in Finnish. However, in the case of derivationally complex Roots, the imperative form and the Root form differ. Thus we have osta-tta- ‘to cause to buy’ as a causative Root, whereas the imperative form of the same Root is osta-ta. The causative Root osta-tta- cannot be used as a Word in any context. A CATEGORY-FREE MODEL OF FINNISH DERIVATIONAL MORPHOLOGY 11

c. isä-n osta-ma auto (A) father-GEN buy-A car ‘a/the car bought by the father’

d. auto-n osta-va isä (A) car-GEN buy-A father ‘a/the father who buys a car’

Note that according to my definition, Roots can be either complex or simple. Indeed, I will argue that there are complex Roots along with simple roots. This terminology comes from Giegerich (1999). The notion of Root is almost identical to the notion of ‘categorially underdetermined lexeme,’ except that it includes the individual morpheme constituents of lexemes in addition. For instance, I will argue that such as osta-tta- ‘buy- CAUSE’ are category-free Roots, but so is the causative morpheme itself. This lexical element has been composed by merging two Roots together, which together constitute another, complex Root (√BUY + √CAUSE = √BUY + CAUSE). Furthermore, I do not reject the reality of lexical categories altogether; rather, I assume that they are part of the conversion process of Roots into Words. Inside of Roots, they do not have any status. For instance, it is the Root−Word conversion process which explains why √FISH appears as a zero-derived form inside of an NP and with a copula inside of a VP, while for √RUN the situation is the converse in that the VP context gives the zero- derived form, and the context requires the presence of overt morphemes. In the case of complex stems such as osta-tta- ‘buy-CAUSE’, categorization is always overt, since ostatta- does not constitute a Word in any context, so there is no asymmetry in its phonological form with respect of the lexical categories. I assume that any complex linguistic element, be it a word or a phrase, can obtain an idiomatic semantic interpretation without losing its syntactic/morphological complexity. Thus, kick the bucket means in some contexts ‘to die,’ but it is inflected as if it were a complex phrase (kicked the bucket, not *kick the bucket-ed). Moreover, it still has the literal interpretation. Thus, semantic opaqueness is by no means a good argument for syntactic atomicity (Marantz 1997). The same is true of words: girlfriend is a complex word even if it means (in some contexts) something else than ‘a friend, who is also a girl.’ Here the morphologically and syntactically complex element has obtained idiomatic semantic features, which are stored in the lexicon-in-use. This assumption is motivated also 12 PAULI BRATTICO by the fact that in some cases my informants were aware of both the idiomatic and the compositional meanings of complex words, which thus exists side-by-side.

2. Cross-linguistic evidence for category neutral roots

Before developing and testing the model on Finnish word formation, I will explain briefly why many linguists have been interested in the category- free theory. These constitute in my mind persuasive arguments in favor of the model, which do not rely specifically on Finnish word formation. In some languages there is direct evidence of category neutral Roots. For instance, in Semitic various semantically related words can be produced by altering the vowels between a sequence of consonants. The consonant sequence itself is not associated with any lexical category. Example (2) comes from Hebrew.

(2) a. g.d.l − ‘big’ as a root that is never used in isolation b. gadol − ‘big’ as an adjective c. giddel − ‘be magnified’ as a verb

In addition to the Semitic languages, several other languages such as Wintu (Pitkin 1984), Tagalog (Gil 1995), Jimgulu (Pensalfini 1997), Tuscarora (Williams 1976), Salom (Mosel & Hovdhaugen 1992), Wolof (Mc Laughlin 2004) and Cayuga (Sasse 1993) are arguably best described as containing categorially underspecified roots. Mosel and Hovdhaugen (1992) describe Samoan as follows:

Many, perhaps the majority of, roots can be found in the function of verb phrase and NP nuclei and are, accordingly, classified as nouns and as verbs. This does not mean that a noun can be used as a verb or a verb as a noun or that we have two homophonous words, one being a noun and the other being a verb. Rather, it means that in Samoan the categorization of full words is not given a priori in the lexicon. It is only their actual occurrence in a particular environment which gives them the status of a verb or a noun […] What is given in the lexicon, is not a particular word class assignment, but the potential to be used in certain syntactic environments as a noun or a verb. (Mosel & Hovdhaugen 1992: 77)

In Finnish, in contrast, many complex stems are non-words, which then leads essentially to the same kind of model. Secondly, from the typological perspective it is reasonably clear nowadays that dichotomous features such as N, V and A do not exist (Baker 2003, Dixon & Aikhenvald 2004, A CATEGORY-FREE MODEL OF FINNISH DERIVATIONAL MORPHOLOGY 13

Hengeveld, Rijkhoff & Siewierska 2004, Rijkhoff 2002, Vogel & Comrie 2000). Perhaps nowhere is the need to reject the dichotomous lexical categories as evident as in the case of adjectives (Dixon & Aikhenvald 2004). Rather, there is a loose category of nounhood and verbhood, which are constituted by prototypes listing the typical properties of nouns and verbs. Lexical categories must be dissolved into several independent properties which correlate with each other. In statistical terms, a lexical category is like a factor: a cluster of properties correlating with each other. Thus, labeling lexical elements with the dichotomous categories N/V/A prior to syntactic computations does not seem to lead to a tenable theory. Studies of agrammatic patients with selective Noun−Verb dissociations show that while it is true that agrammatic patients can have selective problems with verbs and nouns, these deficits seem to extend also to pseudonouns and pseudoverbs (Caramazza & Shapiro 2004). This indicates that the deficit has to do with some form of productive morphosyntax, suggesting that categorization is part of some rule-based component of the grammar.5 More importantly, virtually all Words are associated with a lexical category. If lexical categories were not be associated with lexical elements in some rule component, the fact that there are twenty thousand nouns in somebody’s lexicon would be a miracle, much as if all nouns in somebody’s lexicon would happen to represent entities which are all red (blood, the flag of former Soviet Union, fire truck, etc.) and of no other color. Because lexical categories are based on grammatical rules, they must be, in principle, dissociable from the lexicon- as-listedness which is the domain of all rule-like processes. To put it simply: if the standard theory of the lexicon says that lexical elements are constituted by structure [√CAT+N], then these complex lexical elements must have been composed somewhere out of the feature +N and a sublexical feature √CAT, as follows:

5 It is also worthwhile to note that one type of mistake children commonly make is that they nominalize and verbalize beyond adult grammars (I’m going to basket those apples), use verbs in nominal contexts (Where’s the shoot) and vice versa (Mommy trousers me) (Barner & Bale 2002, Berman 1999). As noted by an anonymous referee, the relevance of this evidence is weakened due to the fact that English has much categorial homonymy. 14 PAULI BRATTICO

(3)

Here one may assume that √CAT and +N originate from the lexicon-as- listedness, whereas the complex entry [√CAT+N] belongs to derivational morphology, being a possible feature combination in many languages. But then √CAT must initially be a category-free element. It thus looks as if the postulation of category-free roots is simply inevitable, at least in relation to the lexicon-as-listedness. Furthermore, it would make no sense to say that the category-free features are excluded from the domain of linguistics, since (3) is a linguistic rule on the basis of its inputs and outputs. To me, the only controversial question is then the issue of how the category-free elements interact with derivational and inflectional morphology, and especially with syntax, not whether they exist or whether they are “linguistically relevant.” For instance, Karlsson’s (1983) model of Finnish word formation does not acknowledge any status to rule (3), but there is no convincing case to be made, I think, for the hypothesis that rule (3) is not “linguistic” or otherwise irrelevant. There is a prima facie argument which supports the contention that lexical category is attached to the root only in the syntactic component of the grammar, so that process (3) becomes part of the core syntax, as argued by Marantz (1997, 1999, 2000). Nominalization, verbalization and adjectivization are productive and systematic processes which seem to take place in syntax, at least in some cases. As we will see, this is so in the case of the Finnish -minen nominalization, for instance (Hakulinen & Karlsson 1979). If it is true that the category of a stem/word can be determined in the syntactic component, the question arises whether we need to assume, on top of this, that it can be assigned also in the lexical component. Since there are independent reasons to assume that the lexical category of a word can be determined in the syntax, the null hypothesis is to assume that it is determined only in the syntax. As it turns out, Finnish word formation provides further evidence for this hypothesis. The fact that the X-bar theory, created in the 70s for purposes of describing certain generalizations concerning syntax, seems descriptively correct, represents one of the strongest cases in favor of the category-free lexicon. The leading idea of the X-bar theory is that a substantial amount of regularity in syntax is insensitive to lexical categories, but refers only to A CATEGORY-FREE MODEL OF FINNISH DERIVATIONAL MORPHOLOGY 15 abstract relational notions such as complement-of-X and specifier-of-X where X can be a head of any category − thus, for example, either a verb, noun, preposition, or an adjective. The following list provides examples of a verb phrase, an adjective phrase and a , each instantiating the same underlying structure containing a head (envy), complement (Mary) and a specifier (John).

(4) a. John envies Mary (VP) b. John’s enviousness of Mary (NP) c. John’s envy of Mary (NP) d. John is envious of Mary (AP)

This raises another question of no less importance: why is this true of human language(s)? The assumption that the syntax does not see lexical categorial information in the first place provides one explanation: if lexical categorial information is invisible at the level of syntax, then substantial evidence in favor of the X-bar theory would emerge, X0 being the category neutral lexical element. Thus, extracting categorial information out of the lexicon provides a way to explain some properties of the X-bar theory itself (Salo 2003: 106−107). Finally, categorially underspecified roots are semantically relevant, as each of the envy morphemes in (4a-d) are semantically related: basically, they represent enviousness. The root √ENVY can capture this common conceptual meaning behind the various words. I propose in this article that lexical roots are the linguistic counterparts of concepts: mental symbols which are constituted over and above by their meaning (Chomsky 2005, Fodor 2003: 152−158, Salo 2003: 69−76, 123−126). This is the standard assumption in formal semantics literature, which piles up intransitive verbs, nouns and adjectives all into the semantic category of one-place predicates and thus claims that they all correspond to the same category-neutral semantic entity (e.g., Heim & Kratzer 1998). To sum up, there are a number of reasons to assume that there is a notion of lexicon which contains only categorially underspecified roots, not verbs, nouns or adjectives. I will argue next that properties of Finnish word formation support the same conclusion. 16 PAULI BRATTICO

3. The proposal

In this section I will define the proposal that describes the properties of Finnish word formation in the best possible way and is in agreement with the kind of category-free lexicon that the evidence cited in the previous section supports. I present a few preliminary arguments in favor of the theory and show how it organizes the facts around Finnish derivational morphology. I begin with Karlsson’s (1983) comprehensive model of Finnish derivational morphology. According to Karlsson’s theory, suffixes can be located in specific positions after the root. The root is specified for its category, so (3) falls out of the domain of Finnish word formation. There are ten positions, in the following order: three for V affixes (1–3), one for passive (4), two nominal positions (5–6), two adjective positions (7–8), one nominal position (9) and, finally, all inflections (10) (Karlsson 1983: 244). Each position can be left empty, and each position can contain a closing suffix so that the derivation stops at that point. For example, nominal marker -minen occurs in the first nominal position (5) but positions (6–9) must be left empty. To illustrate the model, consider a nominalized form of a triple causative:

(5) tee-tä-ty-ttä-minen do-cau-cau-cau-n ‘causing ... to do’

The three causative suffixes fill the three verbal positions, after which comes the nominalizer from position 5.6 From this position, one cannot continue derivation. Yet Karlsson admits (ibid.: 241) that more than three verbal affixes can be stacked on top of each other; see also section 4.3 of this paper. Even four causative morphemes can be added to a root, resulting in only a sense of marginality:

(6) ?tee-tä-ty-tä-ttä-minen do-cau-cau-cau-cau-n ‘causing ... to do’

6 To be exact, the correct analysis of tee-tä-ty-ttä-minen seems to be tee-tä-tä-y-ttä- minen ‘do-CAU-CAU-REFL-CAU-N,’ where -y- is the exponent of reflexivization. I will ignore this detail here. A CATEGORY-FREE MODEL OF FINNISH DERIVATIONAL MORPHOLOGY 17

Karlsson excludes such words from the model because they are “marginal” albeit not “ungrammatical” (241) and because their semantics is no longer transparent (238). Similarly, Hakulinen et al. (2004: § 306) found that only three V affixes can be stacked on top of each other if the actual use of the language is used as a criterion. Longer words become marginal, pragmatically deviant, more difficult to understand, hence such words are unusable (compare under-undercover agent and anti-anti-missile), but the combinatorial process itself is without doubt recursive. I will allow recursive stacking of V affixes instead of three independent positions, but there is hardly any empirical issue at stake here, given my ignorance of the lexicon-in-use.7 In Karlsson’s model, nominal suffixes follow verbal suffixes. Some of these suffixes are closing suffixes, which is a stipulated fact in Karlsson’s model. Karlsson however claims that, excluding some exceptions, complex nominals cannot be verbalized (ibid., 236−237). Because of this, he ends up modeling Finnish word formation by means of rigid positions (1–10) without loopback. Assuming that the three V positions are actually filled with a recursive loop of verbal affixes, such as causatives, this implies that nominal affixes are merged after the verbal loop. Yet as Karlsson himself admits, there are several complex nominals which allow recursion back to the beginning. Furthermore, in some cases complex nominals can be causativized, which brings the derivation back to the inner recursive V layer. Potential examples are provided in (7a-k).8

7 In chess psychology, to pursue again another well-studied example, we have to make a difference between possible positions which do not make sense to experienced chess players and positions which do (Saariluoma 1995). The distinction is largely irrelevant for novice players. The difference is a function of the frequency which such positions are encountered in actual games and in the chess literature, more exotic variations being infrequent and largely unexplored territory. Here we can choose to study chess either as a dynamical, recursive of rules, or as something represented in the mind of experienced chess players. The former offers a more abstract but at the same time more fundamental basis of inquiry, whereas the latter is a function of the particular experiences of the given players. Both views are equally important; which one we choose to study does not, in and itself, commit us to any empirical claims about chess or chess psychology. 8 When a particular example did not exist in the current Finnish lexicon-in-use, I sought an analoguous derivation. An actual record of usage was obtained from the Finnish corpus composed by the Research Institute for the Languages of Finland, the Finnish IT Centre for Science and Department of General Linguistics, University of Helsinki. The 18 PAULI BRATTICO

(7) a. hampa-, hampa-isto, hampa-isto-ttaa (hampaistottaa) (>puistottaa) teeth, teeth-COL, teeth-COL-CAUSE (tree-COL-CAU) ‘teeth, a collection of teeth, to cause a collection of teeth (to cause to have parks)’

b. hiki, hiki-tys, hiki-tys-tää (hiestystää) sweat, sweat-N, sweat-N-CAUSE ‘a sweat, causing to sweat, to cause to sweat’

c. jyrsi-, jyrsi-in, ?jyrsi-in-tää (jyrsintää) (>viestintää) bite, bite-IN, bite-IN-CAUSE (message-N-V) ‘to bite, the thing that is used for biting, to cause (to have) the things that are used for biting (to broadcast)’

d. toimi-sto-tta- (toimistottaa), kalu-sto-tta-, vesi-stö-ttä-, >pui-sto-tta- act-N-V, thing-N-V, water-N-V, tree-N-V ‘to cause to have/be an office, to cause to have/be furniture, to cause to have/be water, to cause to have many trees/cause to have parks’

e. hampai-stus-ta- (hampaistustaa), kalu-stus-ta-, kala-stus-ta-, >avu-stus-taa teeth-N-V, thing-N-V, fish-N-V, help-N-N ‘to cause to be/have a collection of −’

f. juoksu-(t)t(a)-in- (juoksutin), paalu-(t)t(a)-in-, syö-(t)t(ä)-in- run-V-N, pole-V-N, eat-V-N ‘an instrument for causing to −’

g. tutki-nto-ttaa (tutkinnottaa) research-N-V ‘to cause (academic etc.) degrees’

h. lomaile-u-ttaa (lomailuttaa), arvele-u-ttaa (arveluttaa) holidaying-N-V, suppose-N-V ‘cause to have a vacation, to cause to suppose/wonder’

i. laahaa-us-taa (laahustaa), etu-us-taa (edustaa) drag-N-V, front-N-V ‘crawl, represent/stand for’

j. pehmeä-us-taa (pehmustaa), helma-us-taa (helmustaa) soft-N-V, hem-N-V ‘soften, to cause to have hems’ corpus was used through WWW-Lemmie 2.0 at the Finnish IT Centre for Science, obtainable from www.csc.fi/kielipankki. A CATEGORY-FREE MODEL OF FINNISH DERIVATIONAL MORPHOLOGY 19

k. ime-u-ri-oida (imuroida), puske-u-ri-oida (puskuroida) suck-N-N-V, buck-N-N-V, ‘vacuum-clean, to buffer’

Based on Karlsson’s own examples and data such as this, there appear to be two kinds of nominal affixes: those which allow the derivation to continue (e.g., -o in tule-o-ttaa) and those which do not (e.g., -inen, -minen, -ma, - na; *punainentaa ‘red-CAUSE’). An alternative explanation for the data in (7) is to claim that the nominal stems in (7) are not just complex nominals, but lexicalized complex nominals. It is well known that causativization can apply to bare nominals in Finnish, as well as in many other languages (paalu-tta- ‘to pole’). This theory is supported by the fact that many causatives of complex nominals, of which there exists a record of actual usage in my sample, are highly lexicalized (Hakulinen et al. 2004: § 306). The word puistottaa ‘to cause to have parks’ (7b) is derived from puisto, which is a lexicalized noun in Finnish referring to parks. But it is itself composed from puu ‘tree’ and -isto ‘a collection of −,’ so that the compositional meaning of the word is ‘a collection of trees.’ Because of its lexicalized use, there can be a puisto even if there are no trees, which contradicts the compositional reading of the term. The question is then if it is possible to form a novel nominal and causativize it as well. To me, words such as hampa-isto-ttaa ‘to cause a collection of teeth’ are novel, understandable and possible words in Finnish, especially when given some extra-linguistic context. Therefore, it is necessary to look also at possible but nonactual words, since these are items for which the effects of lexicalization can be best controlled. Because the rule which causativizes complex nominals works even in the case of nonactual, but possible words, the data cannot be explained solely by relying upon lexicalization. These considerations bring us to the problem of saying what counts as a “possible word” in a language. The above judgments are based on the present author’s judgment of and semanticality, which could be contaminated by theoretical bias. On the other hand, in order to study the combinatorial potential in a language, whether in the domain or syntax or lexicon, we cannot rely solely on instances that the has actually heard and used frequently, if only because such instances are also contaminated by properties of the lexicon-in-use which we have to control experimentally. To investigate the matter further, I presented the above data to a range of Finnish-speaking informants to obtain their judgments on 20 PAULI BRATTICO semanticality and grammaticality.9 The subjects were asked first to rate the words in the test by their grammaticality (on a scale of 1−5: 1 completely impossible word, 3 possible word, 5 very possible word) and then to provide a semantic interpretation (if it had any). Semantic interpretation was encoded quantitatively so that 0 = no interpretation, 1 = interpretation. Four categories of words were used: (1) monomorphemic Finnish words as fillers and control items (virsi ‘gospel’, talo ‘house’, Suomi ‘Finland’, matto ‘carpet’), (2) derivationally complex words which are used idiomatically in current Finnish (imuroida, laahustaa, rokottaa, tulospalvelu), (3) derivationally complex words which are possible according to the above model but not idiomatic (hampaistottaa, juoksutin, hikistyttää, kalustottaa, puistottaa, juoksuttaa, tutkinnottaa, puistollinen, puistollistaa, hampaallistaminen) and (4) derivationally complex words which involve an ungrammatical lookback from a closing N suffix (*hyppäämäminen, *juokseminentaa, *hyppimineninen, *punainentaa, *puistotintaa). The present model predicts that items on the group 4 should not have a coherent semantic interpretation, whereas items on the three other lists should have. Furthermore, the model predicts that items in category 2 should be classified on average as “fully possible words” (belonging to the lexicon-in-use), items in category 3 as “possible words” (belonging to the class of possible words only) and items in category 4 as “impossible words.” The results agreed with these predictions. Average grammaticality and semanticality judgments obtained from this test were as follows:

9 All informants (7 male, 9 female) were above 20 years of age, most but not all studying or working at the university. Their judgments were obtained by means of a questionnaire. A CATEGORY-FREE MODEL OF FINNISH DERIVATIONAL MORPHOLOGY 21

Table 1. Average grammaticality and semanticality judgments obtained in the present study. The behavioral semanticality and grammaticality data (dependent variables) were analyzed separately. These data were entered into repeated measures analyses of variance (rmANOVA) with one independent factor (experimental category; four levels: 1, 2, 3, 4). The main effect for the factor experimental category was statistically significant for both semanticality (F(2, 14) = 118.7, p = .000) and grammaticality (F(2, 24) = 287.4, p = .000). These effects arose due to differences between groups 2, 3 and 4 p <. 01 for both semanticality and grammaticality, specifically. Note that for judgments of semanticality, the novel words were very similar to the established idioms (groups 2, 3). This seems to confirm my own intuitions about the grammaticality and semanticality of these words, and thus the distinction between the two kinds of nominal suffixes. I therefore hypothesize that there is a first layer of processes which turns stems into new stems, and which may include both V affixes and N affixes, as traditionally understood. The resulting stems have the property that they are still free to turn into adjectives, nouns or verbs when suffixed with appropriate suffixes or inflectional markers. All causatives can be verbalized (juoksu-tta-a), nominalized (juoksu-tta- minen) or adjectivized (juoksu-tta-va, juoksu-tta-ma). These causative stems as such cannot be used as Words, since they require inflectional markers (tense/aspect/case) or further suffixes (nominalization, adjectivization). But when the nominalization, adjectivization or verbalization is attached to the stem, no further derivation is possible. These level 2 suffixes are thus closing suffixes, in that they block further derivation (group 4 violations). Because level 1 stems often require further suffixes in order to be used as Words, I therefore make a distinction between level 1 affixes which turn Roots into Roots, and level 2 affixes which turn Roots into Words. Words are the exit points from the derivation. The model that will emerge as we proceed is illustrated in (8).

22 PAULI BRATTICO

(8)

Examples (9a−h) give a general outlook of the model with some concrete examples.

(9) a. [ AfR + AfR + ... + AfR ] + n/v/a + inflection +

b. [ osta+tta- ] + misen + han [ buy-CAU ] + N.case + Cl ‘causing to buy’.

c. [ osta+tta- ] + nen + ko [ buy-CAU ] + V.Mood.1SG + Cl ‘whether I should cause somebody to buy’.

d. [ puu+isto+tu+tta- ] + minen + han [ tree-col-cau-cau ] + N.case + Cl. ‘the phenomenon of causing a collection of trees’.

e. [ puu+isto+tu+tta- ] + va + mpi + han [ tree-col-cau-cau ] + A + COMP + Cl. ‘the property of causing collections of trees’ (comparative).

f. [ pu(u)+isto+t(a)+in+ta ] + a [ tree-col-cau-instr-cau ] + V.1inf ‘to cause to have an instrument of causing a collection of trees’.

g. [ pu(u)+isto+ll ] + inen + ko [ tree-COL-LL ] + N.case + Cl ‘whether it is something that has a collection of trees’.

A CATEGORY-FREE MODEL OF FINNISH DERIVATIONAL MORPHOLOGY 23

h. [ pu(u)+isto+ll+sta- ] + A + ko [ tree-COL-LL-CAU ] + V.1inf + Cl ‘whether to cause something to have a collection of trees’.

AfR refers to categorially underspecified Root affixes, which together form a Root stem marked by the brackets. After that comes inflection: first the category label, then category-specific inflection (case for nominals, transitivity/tense/aspect for verbs), then clitics. I assume that the causative morpheme is a level 1 Root, based on the fact that it does not produce phonological words and that it requires separate morpheme pieces to be used as a noun, verb, or an adjective. This provides the diagnostic tool to detect other Roots under model (8). If the causative morpheme can be suffixed to a stem S, then this gives a sufficient condition for S being a level 1 Root. On the other hand, if the causative morpheme cannot be merged to S, then either (i) S is a level 2 word (*punainentaa) or (ii) some independent constraint prevents the combination (*hammas-llinen). A necessary property of conclusion (i) would be that S as such can be used as a phonological word. Hypothesis (ii) can be tested by using proper controls on the independent constraints, for example, by controlling the number of syllables in the word, its pragmatical plausibility, phonological output, and so forth. It is well-known that many semantic attributes correlate with syntactic verbhood and nounhood. I will later discuss two such features in this study, eventiveness and referentiality (see Baker 2003: Ch. 3). Eventiveness refers to the property that the phenomenon represented by a word has a ‘temporal contour’ or that it evolves over time. Referentiality captures the property that the word refers to a complete thing, either abstract or concrete, in the world. The first grammarians circa 100 B.C noted that eventiveness is typical of verbs and referentiality of nouns (Robins 1989). This is the default view in functionalist literature (Givón 1984, Langacker 1986). However, it is not possible to correlate these notions one-to-one with the lexical categories. For example, there are eventive nouns, such as juokseminen ‘running,’ and non-eventive verbs (or verb phrases), such as olla talo ‘to be a house.’ I will return to this problem later. I therefore conclude that lexical categories cannot be reduced to features such as eventiveness or referentiality, although they play an important role in grammar. More specifically, there is evidence that they are semantic features of the Roots, unlike lexical categories, which reflect the syntactic contexts of Roots. 24 PAULI BRATTICO

An anonymous referee has pointed out that it is not certain that the complex words listed in (7) are interpreted compositionally in the manner that I have decomposed them. The little experiment I conducted reveals that speakers of Finnish are quite good at decomposing the words semantically. Consider the word juoksutin. From the perspective of feature combination, this word can be decomposed into juokse-u-ta-in ‘run-N-CAU- INST,’ which thus involves elements of running, causation, instrument, and nounhood. When the word did not exist in their lexicon-in-use, the subjects gave precisely this meaning.10 Much of the same is true of more complex words, such as hampaallistaminen ‘the causing of somebody to have a collection of teeth’.11 Finally, even in the case of fully lexicalized complex

10 The subjects’ interpretations were as follows (without glossing): juoksettamiseen käytettävä ainesosa, vrt. juoksute, miksei myös juoksumatto; jokin mekaaninen laite tai ihminen, joka juoksuttaa tyyppejä; kyseessä on jokin juoksumaton tyyppinen laite, joka kuittaa ilkeillä sähköiskuilla tms. huonon harjoittelun; eikö ruuanvalmistuksessa käytetä juoksutinta, että saadaan aine juoksevaksi; juoksupyörä; Jonkinlainen houkutuslintu, esim. juoksukoiraradoilla. Tai sitten käännös tietotekniikkatermille ‘iterator’; eläimen juoksutin, esim. kilpakoiran juoksutin on jänis; Juoksuliina koiralle; Juustoa tehdessä tarvitaan juoksutin (naudoista saatava tai synteettinen), joka juoksettaa juustomassan eli hera erottuu; juuston valmistuksessa käytettävä ainesosa; Koiran/hevosen liikunta- avustin; laite jonka tehtävä on juoksuttaa; Juustonjuoksutin ainakin on olemassa; Aine tai väline, jolla esim. maidon tai jonkun muun valkuaisperäisen aineen voi juoksuttaa niin että siitä tulee rakenteeltaan epätasaista; Jonkinlainen kone jonka läpi neste virtaa eli juoksee. Tulee mieleen jonkinlainen tuotantovaihe meijerissä, varsinkin juustontuotannossa...; aine jolla voi juoksettaa toisen aineen; käsittääkseni juuston valmistuksessa käytetty aine, jolla maito saadaan juoksettumaan. 11 The subjects’ interpretations were as follows: tehdä jollekin hammas (jolla ei aiemmin ole); lähes sama kuin hampaistottaa eli laittaa jollekin hampaat; hampaiden suuhunlaitto, esim. vanhuksia voidaan hampaallistaa valtion toimesta. Toisaalta substantiivia voidaan käyttää myös kuvaannollisessa merkityksessä, jolloin se tarkoittaa esim. henkilön mielipiteiden terävöittämistä. Vrt. ylioppilaskunnan vaalien ehdokkaan mielipiteitä hampaallistettiin; ei hampaistoiteta koko suuta kuten aijemmin, vaan vain pari kolme hammasta hampaallistetaan uudestaan. se on hampaallistamista; hampaat suuhun; Ehkä uusien hampaiden asentaminen hampaattomalle henkilölle; tehdä hampaalliseksi joku, jolla ei ole hampaita. voisin ehkä kuvitella sanottavan, että "tällä papalla on menossa hampaallistamisprosessi”; Asettaa hampaat johonkin (lähinnä koneeseen); Tehdä hampaita eli "hampaistottaa". Prosessin nimi on siis hampaallistaminen, ehkä; hampaisiin liittyvää toimintaa epäilemättä. Ehkä meidän kulttuurissa tätä sanastoa ei ole tosiaan hiottu tarpeeksi nyansoidulle tasolle?; Virkamies hammastaa; hampaiden lisääminen johonkin; Tapahtuma, jossa jollekin annetaan hampaat suuhun; tehdä hammasrattaan muotoiseksi; tehdä jotain sellaista jonka ansiosta jollekin kohteelle tulee hammas/hampaita; hampailla varustaminen, tuskin kuitenkaan A CATEGORY-FREE MODEL OF FINNISH DERIVATIONAL MORPHOLOGY 25 stems such as puisto ‘a park’ (lit. a collection of trees), subjects were sometimes spontaneously aware of the compositional reading, as in the following interpretation of puistollistaa ‘tree-COL-POSS-CAU-V’:

(10) alueen varustaminen puilla ‘to supply the area with trees.’

This is because the compositional and idiomatic meanings can exist side- by-side. Although a larger and better controlled psychometric study should be considered for the future, I think that these facts speak in favor of the two-layer model: words are interpretable compositionally as long as we do not merge V affixes after the layer 2 closing suffixes (or as long as we do not violate some independent constraint).

4. Finnish word formation without lexical categories

4.1 General remarks

In this section, I will apply the category-free theory of word formation to Finnish by concentrating on certain (to me, at least) interesting features of Finnish word formation. This analysis relies on a few essential differences from the standard description of Finnish word formation. Most importantly, I do not assume that suffixation must create phonological Words. This leads to a more fine-grained and, I think, more simple analysis. Second, as we have much word formation which does not generate phonological Words, we can move lexical categories from derivational morphemes into syntax. This is in agreement with the kind of independent evidence briefly reviewed in section 2. Due to limitations of space, I have to leave many interesting issues untouched; what follows is rather a collection of what I take to be the most revealing aspects of Finnish word formation.

4.2 Nominalization and adjectivization

In this section I will look more closely at Finnish nominalization. In Finnish, any verb can be nominalized by suffixing it with -minen ihmisestä, vaan vaikkapa hammaspyörästä tms. kuvaannollisesti myös ihmisestä, siis antaa aseet käteen jossain tilanteessa, jolloin joku on hampaaton, ei osaa puolustautua sanallista hyökkäystä vastaan. 26 PAULI BRATTICO

(Hakulinen & Karlsson 1979: § 14.9, Koski 1978). This is part of a larger group of deverbal nominalizers, of which there are about 15 in Finnish, and it is one of the most productive; most of the others are conditioned by morphological and phonological properties of the stem they attach to (Hakulinen et al. 2004: § 222, 227). Interestingly, -(i)nen, which seems to be a part of -minen, is nevertheless a very common suffix that appears either alone as a nominal affix in a wide range of both nouns and adjectives, or in combination with other material whose status has been so far unclear. Some examples include -(h)inen, -iainen, -i(m)mainen, -kainen, -kalainen, -kkainen, -(k)ko(i)nen, -lainen, -llinen, -lloinen, -mainen, - moinen, -nainen, -nkainen, -noinen, -rainen, -ttainen, -tuinen, -uainen, - uinen, -jainen, and so forth (see Hakulinen 2000: § 3, Hakulinen et al. 2004: § 261–283, Karlsson 1983: 232–243). Many seemingly simple nouns end with -nen, such as hevonen, ihminen, työläinen ‘horse, human, worker.’ The suffix -nen can also be attached productively to a noun. In this case its meaning can be best described as diminutive. Thus, kirja-nen (‘book-let’) means ‘little book’. The distribution of -i- in the affix seems to be conditioned by morphological and phonological properties of the stem (Hakulinen et al. 2004: § 263). This data raises a series of questions: Are all these affixes distinct atoms? Why is -inen so common in the Finnish lexicon? As a way into what I think is the correct answer, consider the suffix -llinen (11a) and a closely related variation (11b):

(11) a. hampa-(a)llinen teeth-A ‘something that has teeth’

b. hampa-(a)llista-minen12 teeth-R-N ‘the property of causing to have teeth’

The suffix -llista in (11b) carries the same meaning ‘something that has −’ with (11a), plus an identical phonological shape -ll-.13 The nominal form contains -(i)nen, the V form causative -sta-:

12 Other examples are: teollistaminen ‘industrialization,’ kansallistaminen ‘nationalization,’ ennallistaminen ‘restoration,’ kaupallistaminen ‘commercialization,’ koneellistaminen ‘mechanicalization.’ 13 According to Rintala (1980a, b), the meaning associated with -llinen is heterogenous. She cited the following possible relations: ‘x which is y,’ ‘x which belongs to y,’ ‘x A CATEGORY-FREE MODEL OF FINNISH DERIVATIONAL MORPHOLOGY 27

(12) a. -ll-inen ‘something that has −’(nominal)

b. -ll-ista ‘to cause something to have −’ (neutral: can be nominalized or verbalized)

It thus looks as if -llinen contains actually two components, -ll- (or -lli-), expressing the meaning ‘something that has−,’ which can be nominalized and verbalized, plus the nominalization or adjectivization -(i)nen. Furthermore, -(s)ta is a neutral morpheme piece that can be verbalized and nominalized.14 Yet there is one problem in dissolving -llinen and -llista into two morphemes: -ll- is an affix that cannot alone produce a word. Thus, according to a traditional word-based theory, it would be a problem to separate -ll- from -llinen. Assuming that word formation is based on morphemes, not words, provides a solution. Suppose that -ll(i)- is a categorially indeterminate morpheme piece with its own meaning ‘something that has −’. Because it is a categorially indeterminate morpheme, it does not, by itself, constitute a Word. Because it is not a Word, it cannot be pronounced alone. In this way, it is possible in principle to divide many of the -(i)nen forms listed above into two or more morpheme pieces, and thereby simplify Finnish word formation. For instance, there is only little reason to assume that -lliste, -llistin, -llisto, - llista, llistusm, and so forth, are unrelated atomistic affixes, and no need to stipulate that some of them are terminating 2 affixes.15 To test this hypothesis, we can try to combine -ll- with other Root affixes according to (9a). Here I consider some of the affixes present in (7). Example (13a) shows how to combine other Root affixes to -ll- and (13b) which is in y’s possession,’ ‘x which is like y,’ ‘x where y is,’ ‘x which produces y,’ among others. 14 The affix -sta- is composed from two elements, -s- and causative -ta-. This is related to the fact that -inen is replaced with -s- in certain contexts; see below. I thank an anonymous reviewer for pointing this out. 15 When the ll-material is separated from the suffixes, we get more fine-grained analysis, e.g. -ll-is-ta-e poss-is-cau-res = -lliste. Other -(i)nen nominalization affixes that are, according to Hakulinen (2000), similarly formed from two affixes are the following: - kainen (-kka-(i)nen, -nkainen (-nka-(i)nen), -noinen (pronominal -noin-(i)nen)); likewise affixes such as -rainen, -ttainen, -uinen, -tuinen. The analysis is here likewise simplified if -(i)nen is separated from the other material. Space limitation prevents me from going into the details of these affixes and analyses; they have to be studied one-by- one. What matters for present concerns that the category-free theory leads potentially into a more fine-grained and, I think, more simple analysis. 28 PAULI BRATTICO shows the vice versa. Word forms which are phonologically impossible in Finnish, creating ungrammatical CCC combinations /nll/ and /sll/ (Karlsson 1983: § 4.2.2.2), are of course impossible.

(13) a. hampaa-lli-ste (työlliste, lailliste) hampaa-lli-stin (>paika-lli-stin), hampaa-lli- sto, hampaa-lli-staminen (>kansallistaminen, kaupallistaminen), hampaa-lli- sta-ma (>ehdo-lli-sta-ma, lai-lli-sta-ma), hampaa-lli-stus (>enna-lli-stus, havainno-lli-stus), hampaa-lli-sti, hampaa-lli-staa (>paikallistaa, kohtuullistaa, kansallistaa, kaupallistaa, alueellistaa), hampaa-lli-stella, hampaa-lli-stua (>osallistua, kaupallistua, pinnallistua, koneellistua, liikunnallistua, ammatillistua, taiteellistua)

b. hammas-ste-lla (>arvostella, perustella, varmistella, kummastella, kauhistella, ujostella), *hammas-tin-lla, hamma-sto-ll-inen (>yliopistollinen, laivastollinen, osastollinen, linnustollinen), ??hamma-sta-ll-inen, *hammast-us-ll-inen, hamma-st-e-ll-inen (>ennusteellinen), ??hamma-st-u-ll-inen.16

But what kind of affix is -inen? Aside from the diminutive meaning in some cases, it has no clear meaning in itself, and it is connected only to nominal forms (including adjectives, which are discussed below). The category neutral theory provides a straightforward answer: -(i)nen is one exponent of the syntactic marker for nounhood (Marantz 1997) and adjectivehood. The same reasoning can be applied to monomorphemic words which end with -(i)nen. Many seemingly monomorphemic nouns are affixed with -(i)nen, among them hevonen ‘horse’ and ihminen ‘human.’ Separating - (i)nen from these words creates stems ihmi- and hevo-, which cannot be used as Words. This is not a problem for the morpheme-based theory, since these forms are Roots: morphemes which, by themselves, cannot form Words. We can avoid repetition in the Finnish lexicon by assuming that - (i)nen is the exponent for n (see Koskenniemi 1983: § 1.8, for the same

16 The words hammastallinen and hammastullinen are impossible, because -ll- only attaches to N affixes; I return to categorial selection in section 4.3. Note that if (13) is the correct analysis of words such as perustella, perustele-n, then the verbal reflexive morpheme -ele-, discussed in section 4.3, is composed out of arvo-s-ta-e-ll-a ‘value-S- CAU-E-LL’ and the marker -a for the infinitivial verb. Similarly, we have tule-o-s-ta-e- lla ‘come-n-s-cau-e-ll-inf’. If we produce a separate Word for each of these phases, we get tule ‘come!’, tulo ‘coming’, tulos ‘the result of coming, i.e., a result’, tulos-taa ‘to print’, tulos-te ‘a printout’, tulos-tee-ll-inen ´something that has printouts’ and tulos-te- ll-a ‘to do printing in a specific, casual manner’. It also follows that perustella is related to perustee-ll-inen. Needless to say, this hypothesis is controversial. A CATEGORY-FREE MODEL OF FINNISH DERIVATIONAL MORPHOLOGY 29 argument). For instance, the morphological analysis of ihminen is √ ihmi+n, hevonen = √ hevo+n. But then phonological material such as ihmi- and hevo cannot constitute Words, hence there is little reason to say that they belong to any of the lexical categories. In other words, they are Roots. There is morphosyntactic evidence that bare nouns ending with -(i)nen should be treated similarly to complex nouns having the -(i)nen ending. To begin with, all nominals ending with -(i)nen have exceptional case-marking properties: -(i)nen is turned into -s(e) + case. This is not the typical case- marking pattern for Finnish nominals (Nelson 1998). The following data illustrates the difference with the words talo ‘house’ and hevonen ‘horse’:

Table 2. Case marking for two kinds of Finnish nominals, bare nominals and (i)nen nominals. Table 2 shows that the case-marking properties of -inen nouns and bare nouns are different. The original -nen is replaced with -s- or -se-, and then the regular case affix rules are applied.17 If so, lexicalized nominals ending with -(i)nen must be marked overtly for n, since the same rule applies to these cases as well, as shown in (14). More generally, this suggests that there is some syntactic structure inside of hevonen comparable to juokseminen, namely, Marantz’s (1997) n(ominalization) head. One alternative hypothesis would be that the rule is phonological, applying to all words ending with -(i)nen. There is evidence that this is not the case. Note, first, that complex nominals ending with -(i)nen cannot occur as the first constituent in a compound, but nouns containing covert n can. Thus liitu#taulu ‘blackboard’ is an acceptable compound in Finnish, but *liitu+mainen#taulu, *työ+läinen#mies are not. Instead of a compound, a word boundary must be used here (liitumainen taulu). The

17 Affixes -s- and -se- are layer 1 affixes, since they occur in causatives as well. Thus, this is different affix than -(i)nen (See Hakulinen 1979: 124−5). The -e- material seems to be there to prevent outputs such as *hevosn. 30 PAULI BRATTICO same is true of lexicalized forms ending with -(i)nen, thus *hevonenmies ‘horseman’ is not a possible word in Finnish. Suffix -nen must be replaced with -s generating hevosmies. The same truncation occurs in the case of any suffix; thus, a causative of hevonen is hevo-s-taa not *hevonentaa. Barring for now the analysis of the emerging -s-, the general rule is to ban further derivation after overt n, as n behaves like a closing affix. This explanation is in line with the category-free model. This rule then automatically explains the behavior of lexicalized -(i)nen nominals if it is assumed that they contain the syntactic constituent n. Now consider the word onnen ‘of luck.’ This genitive word can be used as a first member of a compound, as in onnenpäivät ‘lucky days.’ The word ends with -(i)nen, yet a compound is possible. If the rules regulating -(i)nen nominals were phonological, then onnenpäivät should not be a possible word. There is thus evidence that Finnish has an overt marker for a functional head n. There are at least two alternative hypotheses. According to the first hypothesis, the distribution of -(i)nen throughout the Finnish lexicon is a matter of diachronic facts only. This hypothesis makes it hard to explain why the suffix is still in productive use. But note that in some cases the -inen could well be frozen inside of a larger suffix. Another problem with this hypothesis is how the exceptional properties of -(i)nen words are to be explained.18 Another hypothesis, to my mind much more plausible, is to claim that many nominals end with -(i)nen because nominals have “prototypical -(i)nen instances” which are used when new words are coined. Here -(i)nen would not be its own morpheme piece, but part of a phonological template which is used to coin new words, perhaps by relying on “analogy”. This leaves unexplained why -(i)nen could nevertheless be separated from the stem, leaving semantical and formal material (‘morphemes’) that can enter into other contexts as well. Furthermore, there is direct evidence (case suffixation etc.) that the behavior of (i)nen nominals cannot be accounted for in terms of phonological rules only. I conclude that -(i)nen seems to be a morpheme piece on its own. If this is so, there is a reason to believe in the existence of Roots, since what is left of many morphemes when -(i)nen has

18 Finally, if the distribution of -(i)nen would be a matter of diachrony, being completely lost in modern Finnish as a separate morpheme, it would not remove the explanatory burden: at least -(i)nen had been some kind of morpheme. If so, what kind? Why does it have such a distribution? A CATEGORY-FREE MODEL OF FINNISH DERIVATIONAL MORPHOLOGY 31 been separated does not, in itself, create a Word but a unpronounceable Root.19 The affix -(i)nen can be used to derive adjectives as well; indeed many -(i)nen forms are adjectives. There are two possible hypotheses with respect to this data. First, one could say that -(i)nen is homonymous between two possible underlying syntactic representations, being the exponent of n(ominals) or a(djectives). Another hypothesis says that adjectives are derived from nouns in Finnish. The latter hypothesis is more interesting, given the cross-linguistic observation that adjectives tend to cluster either with nouns or verbs in languages around the world (Dixon & Aikhenvald 2004), suggesting that adjectives can be either ‘noun-like’ or ‘verb-like’. Not surprisingly, Finnish adjectives are noun-like: they inflect like nouns, they require the supporting copula in the position of a predicate, and like nouns, they take genitive modifiers (valtava-n iso ‘extremely-GEN big-A’). This pattern becomes intelligible at once if we regard adjectives as being constructed out of n. Yet apart from their noun-like properties, adjectives clearly have some properties that nouns do not have. For instance, they occur in the modifier position inside of the NP in their bare forms, whereas nouns occur in genitive forms (14a−b), but note that both nouns and adjectives are capable of occurring in the argument position (14b−c). Adjectives and nouns differ in some of the syntactic contexts they may appear (e.g., 14d−e).

(14) a. hevo-sen satula (nominal) horse-N.GEN saddle ‘a/the saddle of a horse’.

b. puna-inen satula (adjective) red-N.NOM saddle ‘a/the red saddle’.

c. puna-inen on väri (adjective) red-N.NOM is color ‘red is a color’.

19 As I said before, all this leaves room for the possibility that -(i)nen is lexicalized in some cases. From the fact that the causative morpheme -tta is frozen inside of a word such as roko-ttaa ‘to vaccinate, lit. to cause to have pox’ we cannot conclude that it is not in productive use at all. 32 PAULI BRATTICO

d. hyvin punainen very red

e. *hyvin satula very saddle

The difference between adjectives and nouns in Finnish is grammatically very subtle. There is nonetheless a quite dramatic semantic difference between words which occur in a modifier and head position inside of an NP. The modifier attributes a general property to the head, whereas the head, perhaps together with an overt or covert determiner, is capable of picking a concrete or abstract referent from the world (Baker 2003). The difference between adjective-like punainen ‘red’ and noun-like punainen ‘redness’ is thus that in the nominal use, it refers to the redness as a whole concept/property (i.e., that the redness as such is a color), whereas in the attributive use it attributes one particular shade or one particular piece of redness to the saddle (Salo 2003: 69−73). We can then say that an nP bears a “referential index,” to use Baker’s (2003) terminology. This has the semantic consequence that it refers to a whole entity in the world (and in turn explains how it interacts with quantifiers, determiners, binding theory, number, and the rest of it) and cannot occur in the modifier position without undergoing the grammatical alteration from NOM to GEN. Other morphemes have the same effect, for example, the -ll- affix discussed above. The fact that those nPs which occur in the modifier position in their zero-derived forms possess comparative forms can be explained on the same semantic basis, since comparatives produce gibberish when combined with words possessing a referential index (??talompi, ‘house-COMP’). This hypothesis, which I assume tentatively here, is in agreement with Baker’s claim that adjectives are characterized by the lack of referential capacity belonging to nouns. In a language where adjectives behave in a verb-like manner, they are then created from vPs and thus they bear many verbal properties, such as verbal inflection and a lack of copula in predicative construction (Baker 2003: 249−263). I will summarize these ideas in the next section, after first discussing verbs and eventiveness.

4.3 Verbal affixes

In this section I will look at the properties of “verbal affixes” (as traditionally understood) of Finnish derivational morphology and argue that A CATEGORY-FREE MODEL OF FINNISH DERIVATIONAL MORPHOLOGY 33 they are best described as category neutral Root suffixes which are eventive in their semantics. Furthermore, this section presents more detailed data concerning verbal affixes, causatives included, which I discussed in a more preliminary sense in section 2. According to the model presented here, eventive ‘verbal affixes’ (V affixes from here on) in Finnish morphology would belong to category neutral affixes as they can be verbalized, nominalized and causativized. Recall from section 2 that causatives were seen as a core example of this phenomenon. What sets these ‘verbal’ affixes apart from the nominal ones is that they are semantically eventive. Nominals are anchored to the spatio- temporal word by means of spatially bounded ‘things’; verbs are anchored by means of events (e.g., Rijkhoff 2002). But, as I will argue presently, eventiveness does not imply syntactic verbhood. In fact, the verbal category label +V is completely redundant in the case of Finnish V affixes. To begin with, Finnish has four types of simple V affixes: causatives CAU ‘to cause,’ reflexives REF ‘to do by oneself,’ frequentives/continuatives FRE ‘to do continuously, frequently’ and momentives MOM ‘to do momentarily, fast’ (Hakulinen et al. 2004: § 303, Karlsson 1983: 237ff., Wiik 1975). These semantically aspectual morphemes are derived from the repertoire of the universal grammar, as they are in no way limited to Finnish (Cinque 1999). Causatives and reflexives are valency changing morphemes, and frequentives and momentives express the manner of performing the action. Several morphological forms in Finnish can correspond to one or another of these eventive semantic attributes, which means that the classification is not based on form but follows current tradition by using semantic classification. The allomorphy of FRE includes at least skele, ele, ile; MOM ahta, aise, alta; CAU ista, sta, ta, tta; REF istu, itu, stu, tu, u, utu, Vntu; see Karlsson (1983: 276). This allomorphy seems to simplify under the category-free theory, as these endings dissolve further (see § 4.2 and fn. 16), but I will put that aside for now. Importantly, each of these affixes carries an eventive or aspectual meaning. There are other nonproductive affixes that can belong to these groups, but which are not addressed here. Combinatorial possibilities of these affixes are shown in Table 3 (adopted from Karlsson 1983) with concrete examples in (15). 34 PAULI BRATTICO

Table 3. Combinatorial possibilities of Finnish V affixes according to Karlsson (1983).

(15) tee-tä-ttä-ä, laula-tt-ele, lanno-itt-u, juo-vu-tta, do-CAU-CAU, sing-CAU-FRE, fertilizer-CAU-REF, drink-REF-CAU, ‘to cause to cause to do’, ‘to cause to sing frequently’, ‘to be fertilized’, ‘to get drunk’, vaiht-u-ile, heitt-el-yttä, suoj-el-tu change-REF-FRE, throw-REF- CAU, protect-FRE-REF ‘to change oneself frequently’, ‘to cause to throw frequently’, ‘to become protected frequently’.

The table shows only three positions, but as many as five of these affixes can be combined together to form such sequences as (luetuttelutella (√READ−CAU−CAU−FRE−CAU−FRE), luettelutella (√ READ−CAU−FRE−CAU −FRE), lueskeluttelutella (√READ−FRE−CAU−FRE−CAU−FRE) and lukaisuttelutella (√READ−MOM−CAU−FRE−CAU−FRE) (see Ojanen & Uotila−Archelli 1979, Karlsson 1983). Although these forms are quite unusable, they are possible words in Finnish. Category neutral V affixes combine rather freely, to a point where it looks clear that the process is recursive and turns a stem into another, more complex stem. Much like the causative morpheme, I would like to suggest that all V affixes are part of the categorially neutral level 1 derivation. This is because (i) they cannot be pronounced alone as words, and because (ii) all stems formed by using any of these affixes can be again verbalized, adjectivized and nominalized. In short, they sit in the category of level 1 Roots. Verbal affixes CAU, MOM, FRE and REF, in the traditional sense, are called “verbal” because when merged with appropriate verbal inflectional elements they possess properties of verbs: the ability to transitivize, to bear tense/aspect/mood, and so forth. Without these inflectional elements they are verbal in the sense that they are eventive in their meaning. However, there is evidence that semantic eventiveness and the syntactic category of verbs should be separated from each other. Many nominals are also A CATEGORY-FREE MODEL OF FINNISH DERIVATIONAL MORPHOLOGY 35 eventive and have a ‘process reading’ (Alexiadou 1999, Fu, Roeper & Borer 2001, Grimshaw 1990, Vendler 1967, 1968). Nominalized verbal stems can behave in this way in Finnish. For instance, juokseminen ‘running’ or juoksu ‘a run’ are more eventive in their meaning than talo ‘house’ or Pekka (proper name). The former allows modification by aspectual much better than the noneventive nominals:

(16) a. isä-n juokse-minen nopeasti hämmensi muita kilpailijoita father-GEN run-N.NOM fast confused other competitors ‘Father’s running fast confused other competitors-’

b. *isä-n talo nopeasti oli punainen father-GEN house fast was red

In Finnish, deverbal adjectives have the same property; they can be combined with eventive adverbs, as shown below:

(17) a. nopeast juokse-va isä fast run-A father ‘a/the fast running father’

b. ?*nopeasti puna-inen talo fast red-A house ‘a/the fast red house’

Finally, as pointed out by Peter Bosch (p.c.), certain verbal contexts probe eventive readings of simple nouns:

(18) a. Pekka began the book ‘Pekka began to read/write the book’

b. Pekka began the tunnel ‘Pekka began to drive/build/walk the tunnel’

Nouns book and tunnel can be seen as referring to an event, not a concrete thing. Thus, eventiveness is not the same thing as the syntactic category of verbs. Furthermore, if we follow Marantz (1997) and assume that verbhood is associated with words only in syntax, i.e., in a component which generated phrases, we need to look at verb phrases as well. Here we find examples of noneventive verb phrases, which are generated from non- eventive words: 36 PAULI BRATTICO

(19) a. ?*Pekka on ihminen nopeasti Pekka is human fast

b. Pekka juoksee nopeasti Pekka run fast

The VP on ihminen ‘be a human’ functions in the role of the predicate, exactly as the verb run ‘to run’. Depending on the theory, the copula can be taken as an inflectional element which realizes verbal inflection (Salo 2003: 247−256) or the phonological exponent of a functional head signaling predicatehood (see Baker 2003). I will thus assume that verbal affixes, so called in the traditional theory, are categorially underdetermined, but marked with a feature [+eventive] (or with a more complex eventive structure, if necessary). Consider the fact that Finnish nominalization affixes typically select for ‘verbal affixes,’ so that there are very rarely two consecutive N affixes (Karlsson 1983: 241).20 We do not need to assume that they select words which are syntactically verbs; rather, these affixes are either closing affixes, or they select words which are marked as being eventive. In this way, selectional restrictions that were originally explained by relying upon an ultimately redundant syntactic category can be explained in a functionalist fashion by using a feature [±eventive]. A word-internal verbal category is thus redundant; rather, it obscures the fact that syntactic verbhood and eventiveness do not always occur in tandem. In sum, Finnish verbal affixes are not yet syntactically verbal: they cannot be used as verbs without certain further inflectional markers. Moreover, they can be used as nouns as well. They are verbal in the sense that they are semantically eventive, yet many nominals and adjectives are also eventive in their meaning, namely, when they contain an eventive Root. The following table lists all possible feature combinations of nP−vP and referential−eventive in Finnish, as discussed so far.

20 In Finnish and likewise in English (Fabb 1988), few nominalizators can apply to already nominalized forms. One of them is -lainen (Rintala 1972: 4.3.2). The affixes −la (‘place of−’), −tar (‘female member of −’) and −kas (semantics more obscure) can be affixed to certain nominalized forms. A CATEGORY-FREE MODEL OF FINNISH DERIVATIONAL MORPHOLOGY 37

Table 4. Various combinations of the two grammatical dimensions discussed here: syntactic category (nP−vP) and semantic features eventive and referential. The syntactic category reflects the position of the Root in the syntax, nP correponding to the position of arguments (or logical subjects), vP corresponding to the position of predicates. Semantic features ±eventive (i.e., the existence of ‘temporal contour’), and ±referential (whether the element refers to a complete entity) are features of the Roots. Adjectives are generated from nPs in Finnish. The distinction between nP−vP corresponds to the syntactic position of the element (argument−predicate), whereas features ±eventive and ±referential are properties of the Roots. Intuitively, lexical category is determined from the context of the word, semantic features from within.

4.4 , roots and words

The categorially underspecified theory of the lexicon postulates the existence of abstract, completely neutralized roots that cannot be pronounced alone. In the case of complex verbs, there is no special difficulty in postulating such elements. In the case of nouns, the situation is more complex. For instance, the word vesi ‘water’ is derived from an abstract Root √WATER, for which there does not exist any kind of vocabulary entry. This means that √WATER is a completely neutralized item: the minimal pronounceable word is the nominal, nominative singular form vesi. One could argue that the postulation of completely neutralized items is too risky and perhaps too speculative. However, there is evidence for the existence of the abstract Root. Consider example (22), which illustrates complex words derived from the 38 PAULI BRATTICO

Root √WATER (the d−t alteration is predicable by a general rule):

(20) vede-ssä, vete-en, vede-ttä, vede-llä, vede-n water-INE, water-ILL, water-ABE, water-ADE, water-ACC ‘in the water, to the water, without water, at water, water’

This is an instance of well-known stem alteration in Finnish. Affixation is based on the word vede-, which is never pronounced as such, whereas the nominative singular form is vesi. Thus we have to regard vede- as a Root. Accordingly, vesi is generated by a -like process for nominative singular nominals. This model predicts that all syntactic derivations operating with a nominative nP should attach to the exceptional nominal form vesi, not vede-. This prediction is borne out in the case of clitics, which indeed are adjoined to the nominal singular form vesi+hän (*vedehän). Clitics are clearest examples of morphemes which are not part of the derivational morphology, as they are suffixed to phonological words even after inflectional markers. Furthermore, Root affixes trigger the regular form:

(21) a. vesi ‘water’ vede-ll-inen, vede-ttää, vede-stää, vede-stö, vede-tär water-LL-INEN, water-CAU, water-CAU, water-COL, water-TAR ’something containing water, to cause to contain water, to cause to contain water, a collection of waters, a female related to water, such as a mermaid.’

b. susi ‘wolf’ sude-ll-inen, sude-ttaa, sude-staa, sude-sto, sude-tar wolf-LL-INEN, wolf-cau, wolf-cau, wolf-col, wolf-tar ‘something containing a wolf, to cause to have/be a wolf, to cause to have/be a wolf, a collection of wolfs, a female wolf.’

c. käsi ‘hand’ käde-ll-inen, käde-ttää, käde-stää, käde-stö, käde-tär hand-LL-INEN, hand-CAU, hand-CAU, hand-COL, hand-TAR ‘something having a hand, to cause to have a hand, to cause to have a hand, a collection of hands, a female related to a hand (such as a female having a hand).’

Root affixes thus select the Root form since they are merged to the Root before n, whereas merging of the n with features [+nominative, singular] A CATEGORY-FREE MODEL OF FINNISH DERIVATIONAL MORPHOLOGY 39 triggers suppletion.21 Kiparsky (2004) notes that stems in Finnish must end with a vowel before affixation, whereas this constraint does not apply to Words. Thus we have alterations such as:

(22) a. tuhanne-ssa tuhat thousand-INE thousand-SG.NOM ‘in thousand, thousand’.

b. avaime-ssa avain key-INE key-SG.NOM ‘in the key, a/the key’.

c. miehe-ssä mies man-INE man-SG.NOM ‘in the man, a/the man’.

The word form on the right occurs only in the nominative and partitive singular form, so that morphophologically the regular form ending with -e occurs at the Root level, whereas the exceptional form is supplied at the Word level. This makes sense if we identify Stem morphology with Roots, and assume that the nominative singular is provided by suppletion, as argued in the case of vesi−vede. Again, clitics and other closing affixes attach to the nominal singular forms which suggest that these forms are Words, thus categorially nominals:

(23) a. tuhat-han, avain-pa, mies-kö thousand-CL, key-CL, man-CL ‘thousand, (it is) a key, is it a man?’

b. tuhat-mainen, avain-mainen, mies-mäinen thousand-MAINEN, key-MAINEN man-MAINEN ‘a thousand-like, a key-like, a man-like’

We can thus explain part of the curious stem alterations in Finnish word formation by making a distinction between Roots and Words: certain stems are phonological exponents of Roots, others are phonological exponents of Words. What applies to verbs applies to many nouns, namely, there is a component inside of the phonological word that does not make a word itself.

21 Furthermore, vesi- is used for the exponent of the Root plural forms. Thus vesi-ä, käsi-ä ‘water-PRT, hand-PRT’ means ‘(many) waters’ or ‘many hands’, respectively. 40 PAULI BRATTICO

Cross-linguistically, the idea that word formation can be dissolved into levels or strata is not anything new (Allen 1978, Kiparsky 1982, Seigel 1974; for a recent review, see McMahon 2000: 1−53). First, stratum 1 affixes (e.g. English -y, -al, -ic, -ize, -ous, -ive, -ese) constitute a prosodic word, whereas stratum 2 affixes do not have similar effects in that they are not part of the prosodic word they attach to (-ness, -less, -ful, -hood, -ship, - ly, -ish, -dom). Among the concrete differences between stratum 1 and stratum 2 affixes are the following:

Stratum 1 affixes can alter the syllabic and stress patterns of the stems, whereas stratum 2 affixes do not have this property.

Stratum 1 affixes are often not productive morphologically and semantically, whereas stratum 2 affixes are.

Stratum 1 affixes cannot be attached outside of stratum 2 affixes (the so-called Affix Ordering Generalization), while stratum 1 affixes are the only affixes which can be attached to bound morphemes.

Although this model itself is not uncontroversial, it is possible that the traditional distinction between stratum 1 and stratum 2 affixes coincides with the distinction between Roots and Words. Giegerich (1999), drawing on Selkirk (1982), provides such a theory and a number of arguments in favor of it.22 The Finnish -minen affixation is extremely productive, nearly unrestricted, whereas reflexivization, frequentivization, and likewise many N affixes are productive but not unrestricted (Karlsson 1983). Similarly, stratum 1 affixes are fusioned with their hosts more tightly than stratum 2 affixes (Hakulinen et al. 2004: § 159):

(24) a. tule+o = tulo (stratum 1) tule+minen = tuleminen (stratum 2)

b. juoksu+tta+in = juoksutin (stratum 1) juokse+minen = juokseminen (stratum 2)

22 Giegerich describes his model as follows: “Under this model, the inputs to all stratum- 1 affixation are members of the category Root. Given the multiple stratum-1 affixation is possible […] the outputs of stratum-1 affixation must again be members of the category Root. Root must therefore be a recursive category: unlike in the traditional understanding of the term […] roots may be morphologically complex as long as such complexity is the result of the stratum-1 morphology” (73). In other words, R-affixation applies to Roots and creates new Roots as assumed here. The same idea seems to apply well to Finnish. A CATEGORY-FREE MODEL OF FINNISH DERIVATIONAL MORPHOLOGY 41

c. puu+isto = puisto (stratum 1) puu+mainen = puumainen (stratum 2)

d. perus+ta+e+ll+inen = perusteellinen (stratum 1) perusta+minen = perustaminen

The Root-Word hypothesis can explain some otherwise stipulated properties of the two stratum model. Siegel’s (1974) Affix Ordering Generalization, which states that stratum 1 affixes are applied before stratum 2 affixes, now follows automatically: R affixes apply by default to Roots which are categorially underspecified. Stratum 1 affixes are morphologically more unproductive and semantically more non- compositional than stratum 2 affixes, which can perhaps be explained on the basis of their different functional roles: whereas level 1 affixes are more relevant to the semantics of the complex predicate, level 2 affixes encode parts of speech and hence syntactic context, either alone or in addition to some semantic function. Productivity is a necessary feature of the latter for the same reason as inflection is, whereas productivity is compromised in the case of the former. Also, one major weakness of the two-stratum model has always been the fact that the postulation of the strata has been an isolated property, not derivable from independently motivated general principles. The present model can avoid this problem, since the distinction between the two strata is principled on an independent basis, as it can be derived from the Root-Word distinction.

5. Conclusions

There is at least some evidence for the hypothesis that linguistic theory cannot do without studying category neutral Roots, along with nouns, verbs and adjectives. Somewhere in the depths of grammar there is a rule which combines category features +N, +V and +A with Roots that do not bear such features. The standard theory says that this combination is relevant neither for word formation nor for syntax. I argued that, on the contrary, it is potentially important for both. While many specific aspects of the model presented here are controversial, and many issues had to be left out due to limitations of space, I still feel that the more fine-grained analysis along these lines could provide solutions to some of the problems concerning Finnish word formation. 42 PAULI BRATTICO

Because category neutral lexical Roots are almost completely devoid of syntactic information, they are best understood as “concepts” in the psychological sense, thus some kind of mental symbols which are individuated based on their semantics alone. For example, the lexical Root √OSTA- ‘buy’ is a lexical item which is “related to or perhaps identical with the concepts that are the elements of the ‘cognoscitive powers,’ sometimes now regarded as a ‘language of thought”’ (Chomsky 2005: 19). It obtains syntactic properties when the concept is used to construct a linguistic expression referring to buying. When buying is the logical subject of the proposition, it obtains the properties of nouns (osta-minen ‘buy-ing’), and when it is used in the role of a logical predicate, it obtains properties of verbs (osta-a ‘to buy’). When it is used attributively in a non-predicate position, it becomes an adjective (osta-va). Lexical Roots thus serve as the interface between psychology and linguistics: they are both conceptual, being associated with meaning but not syntactic features, and linguistic, serving as the basic building blocks of linguistic expressions.

References

Adger, David (2003) Core syntax: A minimalist approach. Oxford: Oxford University Press Alexiadou, Artemis (1999) On the syntax of nominalization and possession. Doctoral dissertation, University of Potsdam. —— (2001) Functional Structure in Nominals. Nominalization and Ergativity. Amsterdam: John Benjamins. Allen, Margaret (1978) Morphological Investigations. Dissertation, University of Connecticut. Aronoff, Mark (1976) Word formation in generative grammar. Cambridge, MA.: MIT Press. Baayen, Harald (1993) On frequency, transparency and productivity. In Geert Booij & Jaap van Marle (eds.) Yearbook of Morphology 1992, pp. 181−208. Dordrecht: Kluwer. Baker, Mark (2003) Lexical categories: Verbs, Nouns, and Adjectives. Cambridge, MA.:MIT Press. Barner, David & Bale, Allen (2002) No nouns, no verbs: psycholinguistic arguments in favor of lexical underspecification. 111: 771−791 Berman, Ruth (1999) Children’s innovative verbs versus nouns: structured elicitations and spontaneous coinages. In Lise Menn & Nan Bernstein Ratner (eds.) Methods for Studying , pp. 69−93. New Jersey: Erlbaum. Caramazza, Alfonso & Shapiro, Kevin (2004) Language categories in the brain: Evidence from Aphasia. In Adriana Belletti (ed.) Structures and Beyond: The A CATEGORY-FREE MODEL OF FINNISH DERIVATIONAL MORPHOLOGY 43

cartography of syntactic structures, Volume 3, pp. 15−30. New York: Oxford University Press. Chomsky, Noam (1970) Remarks on Nominalization. In Roderick A. Jacobs & Peter S. Rosenbaum (eds.), Readings in English , pp. 184−221. Waltham, Mass.: Ginn & Co. —— (2004) Beyond Explanatory Adequacy. In Adriana Belletti (ed.) Structures and Beyond: The cartography of syntactic structures, Volume 3, pp. 104−131. New York: Oxford University Press. —— (2005) Three Factors in Language Design. 36: 1−22. Cinque, Guglielmo (1999) Adverbs and Functional Heads. A Cross-linguistic perspective. New York: Oxford University Press. Dixon, Robert M. V. & Aikhenvald, Alexandra (eds.) (2004) Adjective classes: A cross- . New York: Oxford University Press. Fabb, Nigel (1988) English suffixation is constrained only by selectional restrictions. Natural Language and Linguistic Theory 6: 527−539. Farrell, Patrick (2001) Functional shift as a category underspecification. English Language and Linguistics 5: 109−130. Fodor, Jerry (2003) Hume Variations. Oxford: Oxford University Press. Fu, Jingqi, Roeper, Thomas & Borer, Hagit (2001) The VP within process nominals: evidence from adverbs and the VP anaphor Do-so. Natural Language & Linguistic Theory 19: 549−582. Giegerich, Heinz J. (1999) Lexical Strata in English. Morphological Causes, Phonological Effects. Cambridge: Cambridge University Press. Gil, David (1995) Parts of speech in Tagalog. In M. Alvez (ed.) Papers from the Third Annual Meeting of the Southeastern Asian Linguistics Society, pp. 67−90. Tempe, AZ: Arizon State University Program for Southeast Asian Studies. Givón, Tom (1984) Syntax: a functional-typological introduction. Amsterdam: John Benjamins. Grimshaw, Jane (1990) Argument Structure. Cambridge, MA.: MIT Press. Heim, Irene & Kratzer, Angelika (1998) Semantics in Generative Grammar. Malden & Oxford: Blackwell. Karlsson, Fred (1983) Suomen kielen äänne- ja muotorakenne. Juva: WSOY. Kiparsky, Paul (1982) From Cyclic to Lexical Phonology. In Harry van der Hulst & Norval Smith (eds.) The structure of phonological representations. Vol 1. Dordrecht: Foris, 131−175. —— (2004) Finnish Noun Inflection. In Diane Nelson & Satu Manninen (eds.) Generative Approaches to Finnic and Saami Linguistics, pp. 109−162. Stanford: CSLI Publications. Koskenniemi, Kimmo (1983) Two-level Morphology: A general computational model of word-form recognition and production. Helsinki: Suomen kielen laitos, Helsingin yliopisto. Koski, Mauno (1978) Suomen sananjohdon perustyypit. Papers from the conference on general linguistics, Suomen kielitieteellisen yhdistyksen julkaisuja, pp. 103−117. Hale, Ken & Keyser, Samuel J. (2002) Prolegomenon to a theory of argument structure. Cambridge, MA.: MIT Press 44 PAULI BRATTICO

Hakulinen, Auli, Vilkuna, Maria, Korhonen, Riitta, Koivisto, Vesa, Heinonen Tarja- Riitta & Alho, Irja (2004) Iso Suomen kielioppi. Helsinki: SKS. Hakulinen, Auli & Karlsson, Fred (1979) Nykysuomen lauseoppia. Helsinki: SKS. Hakulinen, Lauri (2000). Suomen kielen rakenne ja kehitys (5th edition). Vantaa: Helsingin yliopiston suomen kielen laitos. Harley, Heidi & Noyer, Robert (1999) Licensing in the non-lexicalist lexicon: nominalization, roots and the encyclopedia. MIT Working Papers in Linguistics 32: 119−137. Hengeveld, Kees, Rijkhoff, Jan & Siewierska, Anna (2004) Parts-of-speech systems and . Journal of Linguistics 40: 527−570. Jeanne, LaVerne M. & Hale, Ken (1998) Transitivization in Hopi. In Memorias del Quinto Encuentro Internactional de Lingüística en el noroeste. Tomo l. Hermosillo: Universidad de Sonora. Langacker, Ronald W. (1987) Foundations of cognitive grammar. Stanford, California: Stanford University Press. Logan, Gordon D. (1988) Towards an instance theory of automatization. Psychological Review 95: 492−527. Marantz, Alec (1997) No escape from syntax: Don’t try morphological analysis in the privacy of your lexicon. In Alexis Dimitriadis, Laura Siegel, Clarissa Surek-Clark & Alexander Williams (eds.), Proceedings of the 21st Annual Penn Linguistic Colloquium, pp. 201–225. (Pennsylvania Working Papers in Linguistics 4.2). Philadelphia: Penn Linguistics Club. —— (1999) Creating verbs above and below little v. Manuscript, MIT. —— (2000) Reconstructing the lexical domain with a single generative engine. Unpublished ms., MIT. Mc Laughlin, Fiona (2004) Is there an adjective class in Wolof? Robert Malcolm Ward Dixon & Alexandra Aikhenvald (eds.) (2004). Adjective classes: A cross- linguistic typology, pp. 242−261. New York: Oxford University Press. McMahon, April (2000) Lexical phonology and the history of English. Cambridge: Cambridge University Press. Mosel, Ulrike & Hovdhaugen, Even (1992) Samoan reference grammar. Oslo: Universitetsforlaget AS. Nelson, Diane C. (1998) Assignment in Finnish. London: Garland. Ojanen, Jussi & Uotila-Arcelli, Eeva (1977) Suomen kielen frekventatiivi- ja momentaaniverbien derivaatio. Virittäjä 81: 245−264. Pensalfini, Robert (1997) Jingulu grammar, dictionary and texts. Dissertation, MIT. Pesetsky, David (1995) Zero Syntax. Cambridge, MA.: MIT Press. Pesetsky, David & Torrego, Esther (2004) Tense, Case, and the Nature of Syntactic Categories. In Jacqueline Guéron & Jacqueline Lecarme (eds.) The Syntax of Time, pp. 495–537. Cambridge, MA.: MIT Press. Pitkin, Harvey (1984) Wintu grammar. Berkeley: University of California Press. Rijkhoff, Jan (2002) The Noun Phrase. Norfolk: Oxford University Press. Robins, Robert H. (1989) A Short History of Linguistics. London: Longman. Saariluoma, Pertti (1995) Chess Players’ Thinking. London: Routledge. A CATEGORY-FREE MODEL OF FINNISH DERIVATIONAL MORPHOLOGY 45

Salo, Pauli (2003) Causatives and the empty lexicon: A minimalist perspective. PhD Dissertation, University of Helsinki. Sasse, Hans-Jürgen (1993) Das Nomen − eine universale Kategorie? Sprachtypologie und Universalienforschung 3: 187−221. Scalise, Sergio (1998) Inflection and derivation. Linguistics 26: 561−581. Selkirk, Elisabeth O. (1982) The Syntax of Words. Cambridge, MA.: MIT Press. Siegel, Dorothy (1974) Topics in English morphology, dissertation, MIT. Sulkala, Helena & Karjalainen, Merja (1992) Finnish. London: Routledge. Rintala, Päivi (1976) Suomen nominikantaiset mainen-johdokset. Virittäjä 80: 163−91. —— (1980a) Adjektiivisten llinen-johdosten semantiikkaa ja syntaksia. Virittäjä 84: 212−29. —— (1980b) Adjektiivisten llinen-johdosten semantiikkaa ja syntaksia. Virittäjä 84: 302−18. Vendler, Zeno (1967) Linguistics in Philosophy. New York: Cornell University Press. —— (1968) Adjectives and Nominalizations. The Hague: Mouton. Vogel, Petra M. & Comrie, Bernard (eds.) (2000) Approaches to the Typology of Word Classes. Berlin, New York: Mouton. Whitman, Neal. (2004) Category Neutrality: A Type-Logical Investigation. London: Routledge. Wiik, Kalevi (1975) Suomen frekventiivi ja kontinuatiivi. Virittäjä 79, 153−166. Williams, Marianne (1976) A grammar of Tuscarora. New York: Garland.

Contact information:

Pauli Brattico Unit/Department of Psychology Siltavuorenpenger 20 C 00014 University of Helsinki e-mail: [email protected]

Ann Delilkan

Head-dependent Asymmetry: Feet and Fusion in Malay1

Abstract

The variegated data relating to fusion in Malay have long escaped unified analysis. Past analyses assume that the conditioning environment is morphological. I suggest the question to ask is not why fusion occurs where it does but why it does not occur elsewhere. I show, too, that the answer to this question lies unquestionably in the consideration of prosody, not morphology. Crucially, I claim that the demand for feature recoverability interacts with the prosodic specifications of the language. While seeking independent support in the language for this claim, I find evidence that, far from demanding special treatment, fusion is more appropriately seen as a natural impulse in the language, one of many segmental changes that occur to effect greater prosodic unmarkedness. Within my description of the parameters of fusion, all data relevant to the process that have hitherto been cited as exceptions are, instead, predictable. Fusion in Malay is, in fact, learnable.

1. Introduction: Background and data

The complex data relating to fusion (or ‘nasal substation’) in Malay have long escaped unified analysis. Most past analyses assume that the crucial conditioning environment for the process is purely morphological. Thus, it is claimed that root-initial voiceless obstruents trigger fusion when preceded by nasal-final prefixes (Teoh (1994), Onn (1980), Zaharani (1998), and, on related data in Indonesian, Pater (1996)). All analyses, however, suffer from incomplete coverage of the relevant data, numerous instances set aside as ‘exceptions’, if mentioned at all. In (1), which derives from Delilkan (1999, 2000), ‘•’ denotes data previously considered exceptional. Data not hitherto discussed in

1 An earlier version of this paper, ‘Feet and Fusion: The Case of Malay,’ appears in the proceedings of the Southeast Asian Linguistics Society's 13th Annual Conference.

SKY Journal of Linguistics 18 (2005), 47–70 48 ANN DELILKAN connection with fusion are superscripted ‘ + ’ . Fusion occurs in all (1a) cases but is blocked throughout (1b). (Henceforth, x is root material, x the output of fusion, and ‘.’ a syllable break. All data in this paper derive from Standard Malaysian Malay.)

(1a) Fusion i. m´N + pukol m´.mu.kol, *m´m.pu.kol ‘hit’, v.t. + ii. m´N + p´reksa m´.m´.rek.sa, *m´m.p´… . ‘examine’, v.t. iii. p´N + p´r + alat + an• p´.m´.ra.lat.tan,*p´m.p´… ‘utilization’, n. iv. m´N + k´ + muka + kan• m´.N´Ι.mu.ka.kan,*m´N.k´… ’reveal’, v.t.

(1b) No Fusion

+ i. m´N + p´r + tadZam + kan m´m.p´r.ta.dZam.kan ‘sharpen’, v.t. * m´.m´r.ta.dZam.kan ii. m´N + p´lbagaj + kan• m´m.p´l.ba.gaj.kan ‘diversify’, v.t. * m´.m´l.ba.gaj.kan iii. m´N + proses m´m..ses ‘process’, v.t. * m´m.pro.ses iv. m´N + pam• m´.N´.pam ‘pump’, v.t. * m´.mam v. tanam + kan tanam.kan ‘plant’, v.t. * ta.na.Nan vi. kan + kah• (suffixes) kan.kah transitive + interrogative suffixes * ...ka. Nah vii. tilam k´tSil tilam k´tSil’ small mattress’ (noun phr.) * tila N´tSil viii. kampoN kampoN ’village’, n. * ka.moN + viii. t´mpojan t´mpojjan ’urn’, n. * t´.moj.jan ix. p´N + dapat p´n.dapat ‘opinion’, n. * p´. na.pat x. m´n + tSipta• m´n. tSip.ta ‘create’, v.t. * m´.¯ip.ta xi. m´N + rawat m´.ra.wat ‘nurse/treat’, v.t. * m´.na.wat

HEAD-DEPENDENT ASYMMETRY: FEET AND FUSION IN MALAY 49

A complete analysis must clearly discriminate between these two sets of cases. My paper shows that, and how, prosodic structure crucially conditions fusion. This fact provides crucial support for seeing a unity between fusion and a range of other segmental processes (Delilkan 2002).

2. Prosodic word structure

As per Delilkan (1998), I first assume that the Malay prosodic word comprises prefix(es) and root, suffixes each projecting their own prosodic word domain, as depicted in (2), below. This differs from claims about prosodic words in Indonesian, a closely related language, to the effect that roots and suffixes together project a prosodic word (Cohn 1989). I assume further that Malay prosodic words are right-headed and that feet are trochaic, i.e., left-headed (Delilkan 1999 and forward). (‘FT’ denotes a head foot, ‘ft’ a dependent.2)

(2) Wd’ / | Wd’ Wd / | | Wd Wd | / | | | ft FT FT FT | \ | \ | | σ σ σ σ σ σ m´n t´r ba lek kan ¯a ‘overturn it’

The novel prosodic structure that I propose is independently motivated via reference to stress facts, asymmetries in affix vowel inventories, and the correlation between root size and prefixation potential (Delilkan 2002: Ch3). It is corroborated further by the distribution of other segmental

2 In invoking such terminology, I assume that head foot syllables are prominent relative to dependent feet, where prominence is defined in terms of asymmetry (Liberman and Prince 1977). Certainly, a manifestation of this prominence is the ability of head foot syllables to bear primary stress. If a head foot syllable must be able to bear primary stress, it is then also reasonable to assume that a closed syllable is a better vehicle for such stress than an open one. Thus I claim that head feet in Malay are ideally composed of the heaviest syllables possible in the language, closed syllables. I discuss the ramifications of this claim further in §3, below. 50 ANN DELILKAN processes besides fusion (such as nasal assimilation, sonorant deletion, and gemination) (Delilkan 1999, and 2002: Ch 5).

3. Prosodic location of fusion

3.1 Fusion and the unmarked dependent foot

3.1.1 Heterogeneous facts between prefixes

In traditional accounts, the failure of fusion between prefixes in forms such as those in (3) is cited as an exception to claims that fusion occurs at the prefix right edge (Onn 1980), since the juncture between prefixes is readily construed as the right edge of the first prefix.

(3) m´N + p´r + tadZam + kan [[(m´m.p´r) (ta.dZam)] (kan)] ‘sharpen’, v.t. * [[(m´.m´r) (ta.dZam)] (kan)]

Assuming the prosodic structure the current analysis imposes on such forms, as depicted in (3), Delilkan 1999 suggests instead that fusion occurs only between feet within a prosodic word, and is blocked between two prefixes because no foot boundary intervenes between them. In the closing remarks of his ground-breaking Optimality theoretic account of the morphophonology of Malay, Zaharani (1998: 271) observes that, contrary to such discussions about the juncture between prefixes, double prefixes do in fact exhibit fusion between them, but only when the second prefix is light, as in (4).

(4a) m´N + k´ + muka + kan [[(m´.NΙ) (mu.ka)](kan)] ’reveal’, v.t. * [[(m´N.k´) (mu.ka)](kan)] (4b) p´N + s´ + ragam + an [[(p´.¯´)(ra.gam)](man)] ‘homogenization’, n. * [[(p´n.s´)(ra.gam)](man)]

Zaharani then leaves this work ‘for future OT analyses’. I hasten to add to this point that fusion will also occur between prefixes when the second prefix is closed if the root that follows begins with a vowel, as seen in (5).

(5) p´N + p´r + alat + an [[(p´.m´) (ra.lat)](tan)] ‘employing as a tool’, n. * [[(p´m.p´) (ra.lat)](tan)] HEAD-DEPENDENT ASYMMETRY: FEET AND FUSION IN MALAY 51

Furthermore, the prefix-trisyllabic root juncture will at times permit fusion as well, as seen in (6a), blocking it at others, as shown in (6b). Heterogeneous fusion behavior is evidently even more widespread than Zaharani’s observation suggests.

(6a) m´N + p´reksa [(m´.m´) (rek.sa)] ‘examine’, v.t. * [(m´m.p´) (reksa)] (6b) m´N + p´lbagaj + kan [[(m´m.p´l) (ba.gaj)] (kan)] ‘diversify’, v.t. * [[(m´.m´l) (ba.gaj)] (kan)]

Neither a foot juncture hypothesis nor reference to the prefix right edge as the conditioning environment of fusion will accommodate these facts. Clearly, a new account is required. The relevant heterogeneous fusion facts are summarized in (7), for ease of reference. (As before, italics denote a fusion segment, and underlining indicates roots. Boldface marks the dependent foot. Output shapes refer to the prefix-root complex only, on the assumption that, together, they project a prosodic word that excludes suffixes.)

(7a) Fusion occurs between two prefixes when the second prefix is closed and the root vowel-initial.

p´N + p´r + oleh (+ an) p´. m´. ro. leh. han ‘acquisition’, n. Output: C´. C´. CV(C). CV(C). CV(C)

(7b) Fusion occurs between two prefixes when the second prefix is light and the root consonant-initial.

m´N + k´ + bumi + an m´. N´. bu. mi. kan ‘bury’, v.t. Output: C´. C´. CV(C). CV(C). CV(C) 52 ANN DELILKAN

(7c) Fusion occurs between an N-final prefix and a trisyllabic root with an open first syllable. 3 m´N + p´rentah m´. m´. ren. tah ‘rule’, v.t. Output: C´. C´. CV(C). CV(C)

(7d) Fusion is blocked between an N-final prefix and a trisyllabic root with a closed first syllable.

p´n + t´rdZ´mah p´n. t´r .dZ´. mah ‘interpreter’, n. Output: C´C. C´C. CV(C). CV(C ) * p´. n´r. dZ´. mah * C´.C´C. CV(C). CV(C )

(7e) Fusion is blocked between two prefixes if the second is closed and the root is consonant-initial.

m´N + t´r + balek (+ kan) ‘overturn’, v.t. m´n.t´r.ba.lek… Output: C´C.C´C .CV(C).CV(C ) * m´.n´r.ba.lek.. * C´.C´C .CV(C).CV(C )

For ease of reference, I repeat the output shapes of the prefix-root complex in (7d-e) and (7a-c) as (8a) and (8b), respectively.

(8a) [( C´C.C´C)(CV(C).CV(C))] (No fusion) * [( C´.C´C)(CV(C).CV(C))] (Ungrammatical fusion)

(8b) [( C´.C´) (CV(C).CV(C))] (Fusion result)

The shapes in (8) merit discussion and are the focus of the following section.

3 Other examples are as follows: /m´m´reksa/ (memereksa, ‘examine’, v.t.), /m´N´lirukan/ (mengelirukan, ‘discombobulate’, v.t.) /m´m´roses/ (memeroses, ‘process’, v.t.), /m´m´randZat/ (memeranjat, ‘shock’, v.t.), /m´¯´limotkan/ (menyelimutkan ‘cover (as with a blanket’, v.t.), /m´¯´lidek/ (menyelidek, ‘investigate’, v.t.) HEAD-DEPENDENT ASYMMETRY: FEET AND FUSION IN MALAY 53

3.1.2 Fusion and the prosodically weak domain of a word

Based on the shapes in (8), past accounts of the locus of fusion can no longer be upheld. To begin elucidation of a new description of the location of fusion, I make the claims in (9).

(9a) CLAIM 1: The target dependent foot in Malay is composed of two open ‘schwallables’ (i.e. schwa-headed syllables, van der Hulst, p.c. 1999), i.e., light syllables.

4 [[(C´.C´).CV(C).CV(C)]

(9b) CLAIM 2: The dependent foot in Malay may not, however, be light-heavy, shapewise.

* [(C´.C´C)(CV(C).CV(C))]

According to Mascaro’s (1976) Derived Environment Effect and Kiparsky’s (1973) Revised Alternation Condition, processes fail to apply within a morpheme but apply at morphological junctures only if they produce more unmarked forms. I suggest that the salient dimension of for fusion in Malay is prosodic. Thus fusion occurs within the prosodically weak domain of the word, i.e., the dependent foot, triggered by the need to achieve the target output prosodic shape in (9a). In short, I claim that fusion should occur in the prosodically weak domain because it leaves dependent foot syllables open. However, although (9a) shows the target shape for a dependent foot, fusion may not apply if, in the process of creating a light syllable in a dependent foot, the entire foot has the shape shown in (9b). The question that needs to be answered is why the dependent foot in (9b) is undesirable. It has one coda fewer than non-fusion produces, yet it is not the correct output. The restriction here relates to a further generalization about foot typology in the language. In the next section, I take a closer look at the relevant facts.

4 See Appendix 3 in Delilkan (2002) for a list of trisyllabic roots (including fossilized reduplicants) that begin with schwallables. 54 ANN DELILKAN

3.1.3 Foot typology

I argue in Delilkan (2002: §3.1) that the default foot in Malay is a trochee. I claim also that a sequence of prefixes together projects a dependent foot that ultimately rests on two schwallables. Since the syllables in this foot are both schwa-headed, there is no qualitative difference in their nuclei that would signify ‘trochaicness’. Consequently, acknowledgement of the trochaic foot form in the language is achieved between the two prefixes in question in the avoidance of a sequence that is light-heavy, a sequence that would be in direct contradiction with the requirement that a trochaic foot should be strong-weak. I refer here to Winstedt’s (1927, cited in Hayes 1995b) claim that only an open schwallable in Malay counts as a light syllable.5 Since I assume that two prefixes together project a foot, fusion between two closed schwa-headed prefixes followed by a C-initial root would produce a light-heavy foot (cf., (9b)), in direct conflict with the idea that, in a trochee, the first element is stronger than the second. The sequence in (10) illustrates the undesirable foot shape in question. (‘L’ means ‘light’, ‘H’ ‘heavy’.)

(10) *(C´.C´C), or *(L-H)

By the restriction embodied in (10), fusion is blocked in (11). (Boldface marks the unacceptable foot.)

(11) m´N + p´r + (disyllabic C-initial root) [(m´m. p´r) (C…)] [*(m´. m´r) (C…)]

In drawing the preceding conclusion, I rely on the claim that foot typologies across languages differ in terms of their definition of what counts as an acceptable foot for the language (Hayes (1995a)). Thus, some languages might require that well-formed trochaic feet be only precisely heavy-light, so that light-heavy, light-light and heavy-heavy are equally ill- formed feet. Less restrictive typologies may require only that trochaic feet begin with a heavy syllable, so that both heavy-heavy and heavy-light sequences are acceptable. Still others allow all sequences except one that is

5 This explains why initial stress is possible in a disyllabic root that begins with a closed schwa-headed syllable, but is impossible if the first syllable were an open schwallable. (Winstedt 1927, cited in Hayes 1995b: 263). HEAD-DEPENDENT ASYMMETRY: FEET AND FUSION IN MALAY 55 light-heavy, in direct contradiction of the classic heavy-light imbalance that makes a trochee. Such languages are termed “mismatch” languages (van der Hulst (1984)). Malay shows evidence of being a mismatch language. The lack of fusion in (11) betrays the avoidance of a mismatch between the intrinsic relationship of the prefix syllables at the level of the syllable, on the one hand, and their extrinsic relationship as it relates to their shared existence in a foot, on the other. Consider (12), which shows all possible shapes of disyllabic dependent feet in Malay. (‘L’denotes a light syllable, i.e., an open schwallable. ‘H’ denotes all other syllables.)

(12a) √ (L - L) [(s´. k´) (h´ndak)] ‘possessing the same needs’, adj. [(m´.m´)(ro.leh)] (fusion permitted) ‘acquire’, v.t. 6 (12b) √(H – L)[(b´r.k´)(h´n.dak)] ‘with intent’, adj. (12c) √(H - H) [(m´m.p´r)(tiN.gi)](kan)] (no fusion) ‘raise’, v.t. (12d) *(L - H) * [(m´.m´r)(tiN.gi)](kan)] (fusion blocked)

Only a pair of dependent foot syllables displaying an L-H pattern is undesirable (cf. the impossible (12d)).

6 It ought to be noted that there are exceptions to the basic pattern of fusion occurring between prefixes if the second is open. For example, ‘mengkebumikan’, pronounced without fusion between the first two (schwa-headed prefix) syllables, is a variant of ‘mengebumikan’ (/m´N´bumikan/). Note, Mismatch restrictions would not block these forms, since Heavy-Light sequences do not violate the trochaic form of the language. What would be the dependent foot in the non-fusion pronunciation is not as light as it could be, though. Given the current proposal, this means that the form is marked because a coda that would normally be targeted for removal has been preserved instead. Forms like these are mentioned in grammar textbooks as highly marked exceptions, though, and a child is expected to learn them as such. I take the acknowledged exceptionality of such forms as indirect evidence of the ‘rule’ I have proposed about the undesirability of codas in dependent feet. 56 ANN DELILKAN

3.1.4 Trisyllabic roots : Fusing and non-fusing

The heterogeneous fusion behavior of prefixed trisyllabic roots can now be accounted for in like fashion. Assuming that doubly prefixed disyllabic roots and singly prefixed trisyllabic roots have the same prosodic structure, I now claim that the prefix /m´N/ – and the first syllable of a trisyllabic root like terjemah (‘translate’,v.t.) would be a closed-closed sequence (cf. 13), making fusion as undesirable an option as it was in the double prefix case.

(13a) p´N + terdZ´mah [(p´n.t´r)(dZ´.mah)] ‘translator, n. * [(p´.n´r)(dZ´.mah)] (13b) m´N + p´lbagai(kan) [[(m´m.p´l)(ba.gai)](kan)] ‘diversify’, v.t. * [[*(m´.m´l)(ba.gai)](kan)] (13c) m´N + p´rdana +kan [[(m´m.p´r)(da.na)](kan)] ‘give primacy’, v.t. * [[*(m´.m´r)(da.na)](kan)]

By contrast, in (14), fusion is permitted between a prefix and a trisyllabic root that begins with an open schwallable—a light syllable.

(14) p´N + p´reksa + an [[(p´.m´)(reksa)(/an)] ‘examination’, n. * [[(p´m.p´)(reksa)](/an)]

The result in (14) thus patterns with the prefix pairs that are ‘closed-open’ (or ‘heavy-light’), i.e., those that do routinely permit fusion. The resulting foot in either case is ‘light-light’, a shape that presents no mismatch. Certainly, the resulting first foot is also desirable as a dependent foot, 7 which I claim is unmarked if composed of light, /C´/, syllables. It is worth noting, therefore, that the shape of the root involved is a critical factor in determining whether or not fusion will occur between two prefixes. A vowel-initial disyllabic root undoes the closed-closed shape in the dependent foot that blocks fusion. The final consonant of the second prefix is syllabified to provide an onset to the first syllable of the root, and the combination of fusion and this syllabification produces light syllables in the dependent foot. The relevant syllabification is shown in (15).

7 I refer to Gafos (1996), who states that syllable weight is scalar, and that C´ is the lightest possible syllable in South East Asian languages. HEAD-DEPENDENT ASYMMETRY: FEET AND FUSION IN MALAY 57

(15a) m´N + p´r + oleh [(m´.m´)(ro.leh)] ‘procure’, v.t. *[(m´m.p´)(ro.leh)] (15b) p´N + p´r + alat + an [[(p´.m´) (ra.lat)][(tan)] ‘using (as a tool)’, n. * [[(p´m.p´)(ra.lat)][(tan)]]

This pattern is echoed in trisyllabic roots, including loanwords. The English word program is pronounced ‘[p´rogram]’ in Malay, where schwa epenthesis eliminates a potential cluster in the first syllable.8 When prefixed with /m´N-/, the result is (16a), a form in which fusion has been permitted. Below, (16b) shows the same result in the case of a native trisyllabic root.

(16a) m´N + p´rogram + kan [[(m´.m´)(ro.gram)](kan)] ‘program’, v.t. * [[(m´m.p´)(ro.gram)](kan)] (16b) m´N + s´limot [[(m´.¯´)(li.mot)] (ti)] ‘cover, as with blanket’, v.t. * [[(m´¯.s´)(li.mot)](ti)]

Mismatch restrictions will not block fusion between the first two syllables in (16a) and (16b), as no ‘L-H’ sequence would be produced by permitting it. Fusion is therefore free to occur in both cases. The current analysis thus provides a single account for the two environments that display heterogeneous fusion behavior—the juncture between prefixes and that between a prefix and a tri-syllabic root. The restriction on foot shape is not at odds with the broader prosodic analysis of fusion offered thus far. Fusion occurs if it produces a dependent foot with unmarked (light) syllables, but only if ‘Mismatch’ restrictions at the level of the foot are not violated in the process. I turn to the next piece of evidence in support of the current prosodic account of the motivation for fusion.

8 The fact that the second cluster in the word remains relates to the four-syllable maximum on the total number of syllables a prefix and root may together comprise. See Delilkan (2002: §3.1) for discussion of other evidence for this limit, which I take as evidence of prosodic word maximal binarity. Resolution of the first cluster in ‘program’ is, in turn, driven by the target shape I claim for dependent feet. 58 ANN DELILKAN

3.2 Fusion is blocked by clusters

Cluster-initial loanword roots resist fusion with N-final prefixes. Consider (17). (I include the prosodic structure the current account assumes, to aid discussion of the ungrammatical forms.)

9 (17a) m´N + proses [(m´m)(proses)], *[(m´m)(ro.ses)] ‘process’, v.t. (17b) m´N + program [(m´m)(program)], *[(m´m)(rogram)] ‘program’, v.t. (17c) p´N + klorin+an [(peN)(klorin)][(nan)], *[(peN)(lo.rin)](nan)] ‘chlorination’, n. (17d) p´N + frasa +an [(p´m)(fra.sa)][(/an)], [( *p´m)(ra.sa)][(/an)] ‘phrasing’, n.

The lack of fusion in such cases could perhaps be given a functionalist explanation, as the way in which the language marks loanwords. The fact of nonfusion is, however, readily explained by referring to the restrictions seen to be active elsewhere in the language. The ungrammatical outputs in (17) would have unlicensed codas in their dependent feet, codas that violate Syllable Contact Law (Venneman 1988, Kaye 1990), by which an onset must be less sonorous than its preceding coda in order to license it. Thus, for instance, in the ungrammatical forms of (17a), (17b) and (17d), /m/ in the dependent foot coda is less sonorous than /r/ in the adjacent onset. Further, fusion in the ungrammatical cases does not even produce the open syllable in the dependent foot that I have argued the process always otherwise does. In every one of the ungrammatical cases in (17), the dependent foot has closed syllables, fusion notwithstanding. Recall that in my discussion of Mismatch effects, I claimed that fusion will occur in the dependent foot only if it will not produce a marked form. This would never be the case if fusion were the strategy adopted in any of the forms in (17). Thus a section of the data relevant to fusion that has consistently been cited as exceptional in the literature finds explanation in the current account, by which markedness restrictions apply to dependent feet in Malay. Such feet are shown to be subject to coda licensing restrictions, too, the effects of which are seen in consideration of sonorant deletion facts in the language (cf. Delilkan 2002: Ch5).

9 The lack of fusion here contrasts with the pronunciation variant of the loanword root that begins with a simple onset, peroses (/p´roses/). This form permits fusion when prefixed with /m´N-/, resulting in /m´m´roses/. HEAD-DEPENDENT ASYMMETRY: FEET AND FUSION IN MALAY 59

4. No fusion in the strong domain: Head feet and closed syllables

I have assumed thus far that fusion serves to produce open syllables in the dependent foot. This claim replaces the foot juncture hypothesis proposed in Delilkan (1999), but is not a sufficient description of the facts. Accordingly, I claim further that fusion is blocked between roots and suffixes, and between suffixes because in each of these cases, it would produce an open syllable in a head foot (Delilkan 2002: Ch2). Thus I assert that, unlike the case in the dependent foot, syllables in the head foot in Malay are unmarked when closed. (18) states this claim.

(18) CLAIM 3: Closed syllables are unmarked in head foot positions.

The claim in (18) relies on the assumptions about the prosodic structure of the language outlined in §2.0. By this structure, first posited and defended in Delilkan (1999), disyllabic roots and suffixes project head foot syllables. I repeat the relevant structure in (19). (As before, ‘FT’ denotes a head foot, ‘ft’ a dependent.)

(19) Wd’ / | Wd’ Wd / | | Wd’ Wd | / | | | Wd Wd | | / | | | | ft FT FT FT FT | \ | \ | | | σ σ σ(σ) σ σ σ (root) suff . suff. suff.

Since head foot syllables correspond to the prosodically strong domain of a word, it can be assumed that they are prominent, where prominence is defined in terms of asymmetry (Liberman and Prince 1977). Thus, head foot syllables are prominent relative to dependent feet. If dependent feet are unmarked when light (where lightness in Malay corresponds to the combination of openness and schwa-headedness), it is reasonable to suppose that relative prominence is achieved in the head foot via syllables that are as heavy as possible. 60 ANN DELILKAN

Certainly, a manifestation of this prominence is the ability of head foot syllables to bear primary stress. If a head foot syllable must be able to bear primary stress, it is then also reasonable to assume that a closed syllable is a better vehicle for such stress than an open one. There is no lack of claims in literature to the effect that languages display a correlation between syllable weight and stress bearing ability (cf. Peak Prominence (Prince and Smolensky 1993), Stress-to-Weight Principle (Riad 1992), Myers (1987)). Thus I claim that head feet in Malay are ideally composed of the heaviest syllables possible in the language, closed syllables. Fusion is triggered in the dependent foot in Malay because it creates light dependent foot syllables. It is, however, blocked in head foot positions, because it produces open syllables in the strong domain, syllables that are marked for that domain. (20) shows the result of ungrammatical fusion in the relevant positions, boldface denoting the open syllable produced by fusion. For completeness, I include (in (20c) and (20d), respectively) a compound and noun phrase, each a sequence of two prosodic words.10

(20a) di + rakam + kan [[(di)(ra.kam)](kan)] ‘recorded’, pass. * [[(di)(ra.ka)](Nan)] (20b) di + rakam + kan + kah ..](kan)](kah)] ‘recorded’, pass./interr. * ..](ka)](Nah)] (20c) tulaN puteh [(tu.laN)][(pu.teh)] ‘white bone’ i.e.,‘dead’ * [(tu.la)][(mu.teh)] (20d) tilam k´tSel [(tilam)][(k´tSel)] ‘small mattress’, n. * [(tila)](N´.tSel)]

The prosodic analysis of fusion I have presented thus far accounts for all the environments discussed thus far. I have claimed that fusion occurs when it will produce unmarked prosodic structures—open syllables in a dependent foot. It is blocked when it produces prosodically marked structures—an open head foot syllable, or a dependent foot that violates Mismatch.

10 These last two forms were mentioned in Delilkan (2002: Ch 3), where I took the fact that two primary stresses attended each as evidence that each comprised a pair of prosodic words. HEAD-DEPENDENT ASYMMETRY: FEET AND FUSION IN MALAY 61

5. Monosyllabic roots and schwa epenthesis—another failure of fusion?

5.1 Consequence for dependent foot

As shown in (21), monosyllabic roots do not permit fusion at their left edge when prefixed with an N-final prefix, not even when the root begins with a T (a voiceless obstruent).

(21) m´N + pam m´N´pam, *m´mam ‘pump’, v.t.

According to Pater (1996) vowel epenthesis is unattested as an NT avoidance strategy in any natural language that he has investigated. At first glance, the Malay data in (21) might be taken as evidence that epenthesis is an active strategy that eliminates an NT sequence. However, it should be noted that ND sequences are subject to the very same schwa epenthesis, as are N-sonorant and N-fricative sequences as shown in (22).

(22a) m´N + bom m´N´bom, *m´mbom ‘bomb’, v.t. (22b) m´N + lap m´N´lap, *m´lap ‘wipe’, v.t. (22c) m´N + waN m´N´waN, *m´waN ‘finance’, v.t. (22d) m´N + had m´N´had, *m´Nhad (?) ‘restrict’, v.t.

It follows that the epenthesis in (22) is not just another case of avoiding an NT sequence. The fact of nonfusion in forms like (21), and the schwa epenthesis that unites the entire set of data in (21) and (22), relates directly to the length of the root. This descriptive point is well documented in the literature (see Teoh 1994, Farid 1980). I argue for an approach to this epenthesis that unites it with the main claims of this paper, focusing on the prosodic structures the process targets---disyllabic open-syllabled dependent feet. In the case of monosyllabic roots that are N-finally prefixed, fusion does not occur, but then neither does sonorant-deletion, the process that would normally occur in the language, given prefixation of any other sonorant-initial root (Delilkan 2002: §5.1). Instead of either fusion or sonorant deletion, in both cases, schwa epenthesis occurs and a default velar place specification is inserted, just as in the case where an N-final prefix is followed by a vowel-initial root. Consider (23).

62 ANN DELILKAN

(23a) m´N + pam m´N´pam i) *[(m´)(mam)] (fusion) ‘pump’, v.t. ii) *[(m´m)(pam)] (NPA) (23b) m´N + lap m´N´lap i) *[(m´)(lap)](sonorant del’n) ‘wipe’, v.t. ii) *[(m´n)(lap)] (NPA)

If the prefix and root are assumed to inhere in separate feet, the ungrammatical forms in (23aii) and (23bii) follow directly from the current claim that codas are dispreferred in dependent feet. One possible explanation is that Mismatch restrictions pertain to (23). It is possible, for instance, that the ungrammatical forms in (23ai) and (23bi) would be, both stresswise and shapewise, L-H feet, as indicated in (24).

(24a) *[(m´.maêm)] (24b) *[(m´.lap)]ê *?(L H) *?(L H)

Delilkan (2002: Ch3) argues for a binary word as the appropriate prosodic structure or prefixed monosyllabic roots, whether the prefixing involves an open or closed prefix. The arguments employed relate to foot level trochaic stress as well as word level iambic stress, and take into consideration the fact that a monosyllabic root itself meets word minimality requirements. Assuming the correct prosodic structure for prefixed monosyllabic roots is a binary word, the contrast between grammatical and ungrammatical forms is seen in the prosodic structures in (25). Note that the ungrammaticality of (25ai) and (25bi) does require further explanation, since the language routinely produces right-headed words, and since by my current claims, a coda-less dependent foot is otherwise highly desirable, as my investigation thus far indicates. HEAD-DEPENDENT ASYMMETRY: FEET AND FUSION IN MALAY 63

(25a-i) Wd (25a-ii) Wd / | / | Ft Ft Ft Ft | | | \ |

* m´ mam m´ N´ pam

(25 b-i) Wd (25b-ii) Wd / | / | Ft Ft Ft Ft | | | \ | *m´ lap m´ N´ lap

One option would be to say that the first foot in each of the ungrammatical forms in (25) is degenerate and therefore undesirable. If this argument is used, though, there would be no way to guarantee that, for example, it is not schwa epenthesis but fusion that occurs in /m´N + pukol/, resulting in the grammatical /m´mukol/ (‘hit’, v.t., from /m´N/ + /pukol/) and which, by the current analysis would begin with a degenerate foot, as in (26), below.

(26) m´N + pukol [(m´)(mukol)], *[(m´.N´)(pukol)] ‘hit’, v.t.

As mentioned earlier, it is clear that the length of the root must play a part in the explanation for the schwa epenthesis in (25). Accordingly, I propose that there are two motivations for schwa epenthesis, and that the confluence of these impulses in the precise environment relating to prefixed monosyllabic roots is what sanctions the epenthesis. Both reasons have a prosodic basis. I propose that there is a tendency in the language to disprefer sequences of word-internal monosyllabic feet. This could be interpreted as a form of “anti-clash” or “lapse” avoidance (Kager 1999: 175, Eisner (1997)). If this restriction holds, the schwa insertion at the point of clash, i.e., between the feet, becomes predictable. The fact that it is a schwa that is inserted is readily explained by referring to a further unity between this epenthetic vowel and the default glottal stop insertion that occurs to resolve a hiatus in derived environments in the language (Delilkan 2002: §5.2). Both are placeless, so that there is the minimal addition of featural material associated with their insertion. This epenthesis has the added advantage of 64 ANN DELILKAN creating not only an open syllable in the dependent foot of the word, but also the most unmarked foot possible in such a position. It is binary, it is very light/weak, and it avoids Mismatch, as (27) displays. (“L’’ denotes a light syllable.)

(27) m´N´pam Wd / | Ft Ft | \ | σ σ σ | \ | Target Dep foot: (C´.C´)(CVC) m´ N´ pam L L

This account of epenthesis unites the non-fusion case with a non-deletion of a sonorant and avoids compelling schwa epenthesis between a single prefix and a disyllabic root. Epenthesis is permitted only because of a confluence of two impulses in the language – a preference for open syllables in the dependent foot and an avoidance of lapse avoidance at the level of the word.11 The potential of violating only one of these requirements is not sufficient to compel the epenthesis. Thus, epenthesis does not occur between a prefix and a disyllabic root since there is no lapse violation avoided by it. Between an open prefix and a monosyllabic root, schwa epenthesis could not be motivated by a need to produce an open syllable in a dependent foot since the prefix is not consonant-final (cf. (28)).

(28) di + pam [(di)( pam)], *[(di.´)(pam)] ‘pumped’, v. pass.

11 Van der Hulst (p.c. 2002) points out that the current analysis cannot, however, be extended to explain the fact that schwa epenthesis does not occur between an r-final prefix and a monosyllabic root. He suggests that the lack of epenthesis in such cases could be the fact that a liquid like /r/ is a less desirable onset than a nasal might be, since less sonorous segments make better onsets, where the current type of epenthesis casts the prefix-final segment as an onset. I propose the further possibility that prefix final /r/ is syllabic in such cases, so that a form like berhad (‘limited’, adj.), from b´r + had , is realized as [(br) (had)]. In such a case, the dependent foot does not contain a closed syllable that epenthesis would ‘open’. Recall also, that r-deletion is also common in rapid speech, so that [(b´)(had)] may be the output of prefixation, again with an open dependent foot syllable, and again irrelevant for epenthesis. HEAD-DEPENDENT ASYMMETRY: FEET AND FUSION IN MALAY 65

In the ungrammatical form in (28), an undesirable hiatus would be produced in by such epenthesis. Hiatus avoidance is a priority in the language, as shall be seen in the facts of glide formation and glottal stop insertion (cf. Delilkan 2002: §5.2).12

5.2 Head feet or elsewhere?

The precise location of schwa epenthesis follows from the prosodic structure assumed for the relevant forms. If, as argued in Delilkan (2002: Ch.3), a schwa-headed prefix projects a degenerate foot, it is boosted by such epenthesis to form a more stable disyllabic foot. There would be little sense in the epenthesis occurring elsewhere in the word to achieve lapse avoidance. Any other position in the word would produce hiatuses (cf. (29a-b)), onsetless syllables (cf. (29a-c), or – perhaps worse still – open syllables in the head foot (cf. (29d)).

(29) m´N + pam [(m´N´)(pam)] a. *[(m´.´N)(pam)] b. * [(m´N)(p´.am)] c. *[[(´m´N)(pam)] d. * [(m´N)(pam´)]

Apart from the schwa epenthesis facts of this section, the tendency in the language to favor schwallables at the left edge of a word is reflected in the fact that a large number of trisyllabic roots begin with schwallables, many of them open (cf. Delilkan 2002: App. 3). This pattern coheres with my general claims about prosody in the language. If trisyllabic roots permit single prefixation and if most prefixes are schwallables, this means that the first foot formed in such cases will be double schwa-headed—a relatively weak foot in the prosodically weak domain of the word. A final fact relating to monosyllabic roots is that no epenthesis accompanies their suffixation. Thus, prefixed monosyllabic roots undergo schwa epenthesis whether or not a suffix is present, as in (30) (over). The facts in (30) could be taken as evidence that lapse restrictions are assessed only on the prefix-root combination, i.e. only within a prosodic word.

(30a-i) m´N + waN m´N´waN ‘provide financial support’, v.t.

12 In the case of glottal stop insertion, Delilkan (2002: §5) shows that epenthesis is in fact quite clearly a costly move, reserved as a last resort to avoid hiatus. 66 ANN DELILKAN

(30a-ii) m´N + waN+ kan m´N´waNkan ‘convert to cash’, v.t. (30b-i) m´N + tSam m´N´ t Sam ‘recognize’,v.t./invol. (30b-ii) m´N + tSam+kan m´N´tSamkan ‘seek to identify’, v.t.

The suffix does not, for instance, combine with the monosyllabic root and thereby remove the lapse violation that an unepenthesized prefix-root sequence poses. However, a singly suffixed but unprefixed monosyllabic root undergoes no epenthesis either, as (31) shows. (I insert prosodic structure to aid discussion.) The lack of epenthesis in (31) directly supports my claim that head foot syllables are unmarked when closed, for in every case, ungrammatical epenthesis results in an open syllable in a head foot.

(31a) bom+kan [(bom)](kan)], *[(bo.m´)](kan)],*[(bom)](ka.n´)] ‘to bomb’ (31b) lap + ¯a [(lap)](¯a)], *[(la. p´)](¯a)], *[(lap)](ka.n´)] ‘wipe it’

My prosodic explanation for epenthesis over non-fusion and sonorant deletion in singly-prefixed monosyllabic roots thus coheres with the larger prosodic structure proposed in the current analysis.

6. Roots and fusion

The fact that fusion is blocked within roots is the only evidence that morphology plays a role in the distribution of fusion in Malay. Fusion can thus be termed a ‘derived environment’ effect, per Mascaro (1976). Traditional analyses of fusion and more recent ones alike have claimed that the root-internal environment resists fusion (Teoh 1994, Farid 1980, and, on Indonesian, Pater 1996). What is of relevance to the current analysis, however, is the interaction between so-called ‘Root Integrity’, on the one hand, and the posited preference for dependent feet resting on open syllables, on the other. For the interaction between these two forces, consider the facts in (32). Once again, I insert the relevant prosodic information.

(32) s´ + umpama a) [(s´./um) (pa.ma)] b) *[(s´./u) (ma.ma)] ‘likened to’

The fact that fusion is ungrammatical in (32b), even though a dependent foot resting on open syllables results from its application, shows that Root Integrity outweighs the prosodic drive to produce unmarked dependent feet. HEAD-DEPENDENT ASYMMETRY: FEET AND FUSION IN MALAY 67

7. Summary

In (33), I now recast the dataset with prosodic information superimposed.

(33a) Fusion and prosodic location i. m´N + pukol [(m´)(mu.kol)], *[(m´m) (pu.kol)] + ii. m´N + p´reksa [(m´.m´) (rek.sa)], *[(m´m.p´) (reksa)] iii. p´N + p´r + alat + an• [[(p´.m´) (ra.lat)](tan)] * [[(p´m.p´) (ra.lat)](tan)] iv. m´N + k´ + muka + kan• [[(m´.N´) (mu.ka)](kan)] * [[(m´N.k´) (mu.ka)](kan)]

(33b) No Fusion, and concomitant prosodic information

+ i. m´N + p´r + tadZam + kan [[(m´m.p´r) (ta.dZam)] (kan)] * [[(m´.m´r) (ta.dZam)] (kan)] ii. m´N + p´lbagaj + kan• [[(m´m.p´l) (ba.gaj)] (kan)] * [[(m´.m´l) (ba.gaj)] (kan)] + iii. m´N + proses [(m´m) (pro.ses)] *[(m´m)(ro.ses)] iv. m´N + pam• [(m´.N´) (pam)], *[(m´) (mam)] v. tanam + kan [[(ta.nam)](kan)], *[[(ta.na)] (Nan)] vi. kan + kah• (suffixes) ..] (kan)] (kah)] *...] (ka) ] (Nah)] vii. tilam k´tSil [(ti.lam)] [(k´.tSil)], [(ti.la)][N´.tSil)] viii. kampoN [(kam.poN)], *[(ka.moN)] ix. t´mpojan [(t´m)(poj.jan)], *[(t´)(moj.jan)] x. p´N + dapat [(p´n) (da.pat)], *[(p´) (na.pat)] xi. m´n + tSipta• [(m´n) (tSip.ta)], *[(m´) (¯ip.ta)] + xii. m´N + rawat [(m´) (ra.wat)], *[(m´)(na.wat)]

From the body of data in (33a), it can be said that fusion produces open syllables in a dependent foot, whether it occurs between a prefix and a root or between prefixes. The fact that fusion is blocked between prefixes, in (33bi), and between a prefix and a trisyllabic root, in (33bii), even though in each case an open syllable is produced in the dependent foot as a result, in turn shows a further restriction on dependent feet. They may not begin with an open syllable and end in a closed one. I suggest such feet are 68 ANN DELILKAN undesirable because they represent a ‘Mismatch’ (van der Hulst 1984) with the trochaic foot form of the language. Further evidence that fusion aims at producing dependent feet resting on open syllables is seen indirectly in the fact that cluster-initial roots block fusion, as seen in (3biii). Fusion here would have failed to produce an open syllable in the dependent foot and is therefore not sanctioned. The ungrammatical output here also violates Syllable Contact Law, since the coda nasal resulting from fusion is followed by a more sonorous onset, the liquid /r/.) I have also maintained that fusion is blocked between roots and suffixes (as in (33bv)), between suffixes (as in (33bvi)) and between roots (as in (33bvii)) because, in each case, the result of fusion would be an open head foot syllable. As the last piece of evidence of the relationship between prosodic structure and fusion, the process is blocked in (33bvi), between a prefix and a monosyllabic root. I suggest that the language disfavors sequences of word-internal monosyllabic feet, the schwa epenthesis that does in fact obtain representing a strategy to avoid such a sequence. Unlike the epenthesis, fusion would have failed to produce a prosodically desirable dependent foot, and is therefore blocked. The fact that the epenthesis that does obtain occurs precisely where it does and not, for instance, at the right edge of the monosyllabic root, in turn provides support for my corollary claim that head foot syllables are ideally closed. In (33bviii) and (33bix) is seen evidence that roots resist fusion, and that this resistance overrides the requirement that dependent feet have open syllables. The fact that (33bx) through (33bxii) do not permit fusion, on the other hand, relates to a broader claim I make in Delilkan 2002, that the featural information of input segments associated with head foot segments may not be obscured. It is beyond the scope of this paper to discuss these last three forms in detail, however. All other sections of the dataset in (33) do follow from the main tenets of this paper, that fusion is prosodically constrained, that dependent feet are ideally composed of open schwallables, and that head feet, by contrast, ideally rest on closed syllables.

References

Cohn, Abigail (1989) Stress in Indonesian and bracketing paradoxes. Natural Language and Linguistic Theory 7: 167–216 Delilkan, Ann (1998) When fusion alone just isn’t enough. Presented at LSA annual meeting. January 8th. New York. HEAD-DEPENDENT ASYMMETRY: FEET AND FUSION IN MALAY 69

—— (1999) Prosody drives segmental phonology: The case of fusion in Malay. Unpublished ms. (Dissertation proposal. NYU Dept of Linguistics March 24th.) —— (2000) Fusion in Malay: Why Not. Invited Speaker at The International Linguistics Association’s Monthly Talks Series. Hunter College, New York (May 13th). —— (2002) Fusion and other segmental processes in Malay: The crucial role of prosody. Doctoral dissertation. New York University. —— (forthcoming) Feet and Fusion: The Case of Malay. Proceedings of the Southeast Asian Linguistics Society’s 13th Annual Conference. Eisner, Jason. (1997) Foot form decomposed. In MITWPL 31 (1998). Proceedings of SCIL VIII Gafos, Adamantios (1996) On ‘minor’ syllables in South-East Asia. Paper presented at the GLOW Workshop on Weight Effects. Athens. Hayes, Bruce (1995a) A phonologically-driven, Optimality theoretic account of post- nasal voicing. Handout from MIT colloquium. April 28th. —— (1995b) Metrical stress theory: Principles and case studies. University of Chicago Press. Kager, René (1999) . Cambridge: Cambridge University Press. Kaye, Jonathan (1990) Coda licensing. Phonology 7: 301–30. Kiparsky, Paul (1973) Abstractness, opacity, and global rules. In Osamu Fujimura (ed.) Three dimensions of linguistic theory. pp 57–86. TEC: Tokyo. Liberman, Mark and Alan Prince (1977) On stress and linguistic rhythm. Linguistic Inquiry 8: 249–336 Mascaro, J. (1976) Catalan phonology and the phonological cycle. Doctoral dissertation, MIT, Cambridge, Mass. Myers, Scott (1987) Vowel shortening in English. Natural Language & Linguistic Theory 5: 485–518. Onn, Farid (1980) Aspects of Malay phonology and morphology. Universiti Kebangsaan Malaysia. Bangi. (Published version of 1976 University of Illinois dissertation) Pater, Joe (1996) Austronesian nasal substitution and other *NC effects. In René Kager, Harry van der Hulst, and Wim Zonneveld (eds.) 1999. The prosody-morphology interface Cambridge: Cambridge University Press. pp. 310–343. [Edited version in J. McCarthy, (ed.) (2004) Optimality Theory in Phonology: A Reader. Oxford and Malden, MA: Blackwell.] Prince, Alan and Smolensky, Paul (1993/2004) Optimality theory: Constraint interaction in generative grammar. Technical Report, Rutgers University and University of Colorado at Boulder (1993) Revised version published by Blackwell, 2004. Rutgers Optimality Archive 537. Riad, Tomas (1992) Structures in Germanic prosody: A diachronic study with special reference to Nordic languages. Doctoral dissertation, Stockholm University. Teoh Boon Seong (1994) The sound system of Malay revisited. Kuala Lumpur: Dewan Bahasa dan Pustaka. Van der Hulst, Harry (1984) Syllable structure and stress in Dutch. Dordrecht: Foris. Venneman, Theo (1988) Preference laws for syllable structure. Berlin, NY, Amsterdam: Mouton de Gruyter. 70 ANN DELILKAN

Winstedt, Richard O. (1927) . Oxford: Oxford University Press. Zaharani Ahmad (1998) Phonology and morphology interface in Malay: An Optimality theoretic account. Doctoral dissertation. .

Contact information:

Ann Delilkan City University of New York tel: 1–718–260–5018 e-mail: [email protected] Agurtzane Elordui and Igone Zabala

Terminological Variation in Basque: Analysis of Texts of Different Degrees of Specialization1

Abstract

The aim of this paper is to contribute to the evaluation of the present state of lexical- discursive development in scientific texts written in Basque. More specifically, it focuses on the study of terminological variation in texts on genetic engineering of different degrees of specialization. It describes the patterns of variation found in those texts, and attempt to relate such patterns to different types of sociolinguistic and discourse factors. The sociolinguistic factors are mostly associated with the incomplete normalization of Basque. For that reason, we also discuss the issue of terminological variation in languages undergoing normalization and we propose a set of recommendations based in this study, which we feel should be borne in mind during the normalization of terminology of those languages.

1. Introduction

When a language has been undergoing a process of normalization for various decades, as in the case of Basque, it becomes necessary to evaluate how well the objectives set at different language planning stages have been met. There are numerous aspects that such an evaluation should cover, but our interest focuses specifically on analyzing the present state of lexical expansion and discursive development in fields where Basque was not used until recent times. This general interest has led us to analyze the lexical-discursive development in science, as one of the areas of Basque language use to be incorporated most recently. Within this broad field, we have chosen to concentrate on the subject of genetic engineering due to its wide presence in texts of different degrees of specialization, ranging from newspaper and

1 This work is part of the investigation project EHU 1/UPV 00113 310-H-15921/2004 developed in the University of the Basque Country.

SKY Journal of Linguistics 18 (2005), 71–91

72 AGURTZANE ELORDUI AND IGONE ZABALA magazine articles to high school and university textbooks or encyclopedia articles.2 The functional variety of these texts enables us to analyze lexical and discursive development in different communication contexts, examining in particular whether or not the texts contain linguistic features that point to progress in the normalization of specialized discourse. Specifically, we have studied the degree of systematic variation developed to date in written Basque. Moreover, since the texts chosen for our study form part of the input required to build the science-related lexical skills and discourse strategies of different types of speakers—including secondary and university students, scientists and the general public—they enable us to analyze how these texts contribute to the lexical knowledge shared by the linguistic community, thus helping to make the community more compact. The aim of this paper is to contribute to the evaluation of progress in scientific writing in Basque through a descriptive study of terminological variation. More specifically, it describes the patterns of variation found in different texts and attempts to relate such patterns to different types of sociolinguistic or discourse factors. To further the reader’s comprehension of the sociolinguistic factors at work, the first part of this paper will be devoted to a brief description of the ongoing normalization process under way since the 1960s, when Basque began to be standardized. In describing the current situation, we will briefly explain the major sociolinguistic problems faced by the Basque linguistic community and how they affect the scientists and writers producing the texts concerned in this study. Sections 3 and 4 will then be devoted to terminological variation. First we discuss the issue of terminological variation in languages undergoing normalization, and argue that variation should actually be an objective to strive for within the normalization process. We then present the results of our descriptive study of terminological variation in Basque. We discuss each of the variation patterns observed in our corpus of Basque texts of different degrees of specialization. First we show the patterns of designative variation identified, and then attempt to account for each

2 The texts examined are: texts from the encyclopedic dictionary LUR Hiztegi Tematikoa (~ 16,000 words); Basque high school texts from different publishers (~ 15,600 words); articles appearing in the science magazine Elhuyar (~ 4,000 words) and newspaper articles appearing in Egunkaria and Berria (~ 10,000 words). To help complete our study of the use of this terminology at higher levels of specialization, we consulted directly with experts in genetics at the University of the Basque Country, who kindly provided teaching materials used in fifth year genetics classes.

TERMINOLOGICAL VARIATION IN BASQUE 73 through explanations of the underlying sociolinguistic and discourse factors that may have motivated them. Finally, we propose a set of recommendations based on this study, which we feel should be borne in mind during the modernization (cultivation or elaboration) of languages undergoing normalization and, more specifically, in the normalization of terminology in Basque.

2. Language normalization and scientific writing in Basque

In the middle of the 20th century, Basque was in a state of diglossia and incipient displacement to Spanish and French.3 Relegated for centuries to use almost entirely in the home,4 it was practically absent from domains of public use, particularly areas concerned with technology and science. Written Basque was in practice almost nonexistent for common speakers, since only religious, and to a lesser extent literary, texts were available, and then only in very small numbers. What is more, even this functional compartmentalization was breaking down to the detriment of Basque, with Spanish and French tending to be used more and more with family and friends. The lack of prestige and functionality of the language were, together with other political, cultural and economic factors (Elordui 1995), major reasons for abandoning use of the language. In a period of mass migration to cities and industry, with on the rise and formal language domains gaining in importance — particularly those areas concerned with technology and science— Basque was not considered adequate for meeting the new functional requirements of Basque speakers (Elordui 1995). Moreover, the dialect fragmentation afflicting Basque through the centuries, together with the narrow range of text types, seriously hindered readership and the development of the language in formal written contexts. In the 1970s, however, a new and promising period began for Basque, particularly as concerns the development and diffusion of the language in

3 Basque is spoken in territories to the north and south of the Spanish-French border. In the middle of the 20th century, there was still a large proportion of monolingual Basque speakers on both sides of the border. Today however, in both cases the Basque-speaking community is mostly bilingual (Spanish-Basque/French-Basque) and shares the territory of the Basque Country in the south with a monolingual Spanish-speaking community, and in the north with a monolingual French-speaking community. 4 Basque was also commonly used in social activities of a religious nature and at work among farmers and fishermen. 74 AGURTZANE ELORDUI AND IGONE ZABALA formal domains. The three most important factors accounting for this change were: the codification and standardization of Basque;5 its introduction into the educational system,6 in many cases as the vehicular language; and the use of Basque in the media. Importance must also be given to the progress and development that Basque has enjoyed in other public contexts (government, health, technology, etc.). According to a Basque Government sociolinguistic survey conducted in 2001, the more public services offered in Basque, the greater the demand for this language in such domains (www.euskadi.net). These facts, together with other standardizing and normalization initiatives undertaken by individuals and institutions, have made it possible for written Basque to develop and for the language itself to spread to new areas of use. The result has been a spectacular increase in the publication of texts of different functional types and the creation of a growing community of consumers of formal written and oral texts in Basque.7 This situation, while undoubtedly promising, is somewhat offset by the difficulties encountered daily by Basque speakers in different usage areas. Moreover, apart from these problems common to all speakers of the language, the communities that produce texts in specialized fields suffer the particular difficulties associated with developing a type of discourse that was until lately non-existent in the linguistic repertoire of Basque. Finally,

5 At the congress held in Arantzazu in 1968, the Royal Academy of Basque Language (Euskaltzaindia) embarked on a process of codifying and standardizing Basque in order to overcome the diglossia to which the language was subjected and curb its displacement (Lekuona 1968). The design of standard Basque was based primarily on sociolinguistic criteria (Mitxelena 1968) which sought not only to curb abandonment of the language and gain new speakers, but also to endow the language with linguistic resources enabling its use in new areas. 6 In France, Basque does not enjoy recognition as an official language, and therefore its presence in education is limited to private schools. In Spain, Basque enjoys official status in the Basque Autonomous Community and in the northern part of the Autonomous Community of Navarra. In regions where Basque is an official language, schools offer students the possibility of studying in Basque (linguistic immersion model), or else studying in Spanish with Basque taught as a subject. The fastest- growing of the two is the immersion model, where Basque is the vehicular language for most subjects. The language is also gaining ground at the university level, although to a lesser extent. It is worth noting in the context of this article that a degree course taught entirely in Biology has been offered for more than a decade. 7 The Basque community gained 110,000 speakers in the decade between 1991 and 2001. What is more, the population segment accounting for most of this increase is children and young people (www.euskadi.net), the prime consumers of formal writing. TERMINOLOGICAL VARIATION IN BASQUE 75 we must not overlook the fact that the development of specialized language has occurred simultaneously with the ongoing codification and standardization of Basque, which has made it all the more difficult for specialized language to develop and become established.8 Indeed, deficiencies in standardization affect these functional registers more than the registers reserved for interactions with family and friends. This is so because most specialized languages take the formal standard style as their reference point, which in the case of Basque is euskara batua or unified Basque. One of the main obstacles to be overcome by people wishing to speak or write about science in Basque is the lack of a model to use in developing discourse. Until now, in most fields of study, Basque has had no texts that could be used as a reference and from which people could learn proper discourse strategies for that field. Basque scientists therefore have had no other option than to create their own scientific discourse, and in nearly all cases have had to deal with the problems arising from having, as their only reference, some other language, normally Spanish, French or English. To top things off, the option to borrow terms from one of these languages has been severely frowned upon, there being strong sociolinguistic pressure to create neologisms instead so as to shield the language from outside influence. Another problem for the development and diffusion of scientific language is that the Basque science community is not a compact one. Communication networks among scientists are rare and discontinuous. Exchanges in Basque about science are mainly limited to academic contexts and then mostly to teacher-student relations.9 This gives rise to two further problems affecting the development of scientific language. First, it seriously hinders the development and use of scientific Basque at higher levels of expertise, since it is through interactions among experts that specialized terminology is created. Secondly, the fact that specialists fail to share terminology and discourse strategies in their field means that the input received by their students —the potential future members of the science community— will, instead of making the community more

8 Specifically, the first volume of the Unified Dictionary of the Basque Language was published in the year 2000 (Euskaltzaindia 2000). 9 In these spheres as well, the input received by students is produced by teachers and professors whose own access to specialized knowledge has been mostly through Spanish or English. It would be interesting to study how this fact has affected the development of specialized discourse to date. 76 AGURTZANE ELORDUI AND IGONE ZABALA compact, foster further free variation in the community’s use of terminology. Basque, therefore, does not yet have a compact community of scientists sharing a common fund of scientific discourse to which they can turn for guidelines when producing new text. That is, there are no discourse strategies or established terminologies shared by this speech community. The creators of the specialized texts in our corpus find themselves without reference models, as happens as well with authors of teaching materials and articles intended for the general public. In the case of textbooks and teaching materials, there is the additional problem that a large number of them are translations from Spanish. Having explained some of the major problems affecting producers of specialized texts and in general those involved in modernizing Basque, brief mention should be made of the important contribution that the type of texts chosen for this study can make in the development of specialized vocabulary or terminology in Basque. As pointed out above, the genetic engineering texts analyzed are intended for different levels of expertise and help in different ways to make the language community in this field more compact. Textbooks and teaching materials provide students with the terminology that will eventually enable them to share their specialized knowledge, and that will serve as the first rung on the ladder leading to higher and higher levels of specialization. The terminology acquired during these early stages is therefore essential to the building of terminology in any field of knowledge. The lack of a shared standard lexicon at these basic levels makes it all the more difficult for normalized terminology to exist at higher levels of specialization. Consequently, these texts are crucial for the contribution they make to the degree of compactness of the discourse community in each field of specialization. Encyclopedic articles, on the other hand, help in that they both reflect and secure the encyclopedic-lexical knowledge of a linguistic community. They can play an important role during phases in which the language is not sufficiently normalized or stabilized in certain areas of study, establishing terminology and offering it besides in a more or less educational context. Finally, texts intended for the general public extend some of the vocabulary of different specialties among the linguistic community. Such texts are particularly important for filling in the gaps for speakers of minority languages excluded from the educational sphere, and they therefore can also play a large role in making the speech community more compact. TERMINOLOGICAL VARIATION IN BASQUE 77

3. Terminological variation in normalized languages

Terminology can be approached from a standardizing, normalizing or descriptive standpoint, and each approach sees variation in a different light. Followers of the Wüster school (Wüster 1998 [1979]) and the Soviet school (Lotte 2001 [1948]) attempt to combat synonymy through standardization, since they feel that the proliferation of variants hinders communication among specialists. The normalization school centered around the recuperation of French in Quebec sees synonymy as a linguistic variation phenomenon that must be addressed, but not necessarily eradicated (Auger 1994, Boulanger 1983). Finally, Socioterminology and Communicative Theory — the descriptive schools of terminology — criticize the Wüster school (General Theory of Terminology, GTT) for its reductionism primarily as far as the tendency to obviate variation in terminological units is concerned. Socioterminological (Boulanger 1991; Gaudin 1993; Guespin 1991) and communicative (Cabré 1999) approaches, on the other hand, see terminological variation as one of their main subjects of study. They describe terminological units in discourse and analyze the sociological and discourse conditions that give rise to different types of texts. These approaches enable us to relate inter- and intratextual variations in terminology to the degree to which discourse has developed in the language. In line with these latter schools, we assume that there is designative and conceptual variation in terminological units within and between texts, and that these two types of variation depend both quantitatively and qualitatively on the characteristics of the text, particularly on its degree of specialization (Freixa 2001, 2003). In a given specialized discourse community, there is implicit or explicit agreement on the notion- designation relationship established within each terminological unit, although the discourse community also shares a system of variation that allows a) the terminology to adapt to communicative situations of different degrees of specialization and abstraction, and b) specialized discourse to be ordered using different discourse strategies (paraphrase, reformulations, metaphors, etc). In analyzing variations in different types of texts, we must first draw a distinction between variants found in texts written by the same author (individual or intra-variation) and variation in texts written by different authors (inter-variation) (Freixa 2003). In the case of individual variation, stylistic factors (e.g. the desire to avoid repetition, or find more expressive, 78 AGURTZANE ELORDUI AND IGONE ZABALA succinct, emphatic or erudite ways of expressing something) often underlie designative variants and should therefore be taken into special account. Individual speakers may also vary the expressions they use, including scientific terminology, for functional reasons (e.g. to adapt to their listener’s level of specialization or the degree of formality required by the occasion). All such causes are discursive in nature. Generally speaking, efforts to enhance precision, systematicity and conciseness can be seen in the linguistic choices made by the authors of texts of different degrees of specialization. However, in comparing texts written by different authors, other types of causes may be at work. Variations can occur due to dialect differences of geographical, chronological or social origin, or they may arise for cognitive reasons associated with e.g. concurrent theories, conflicting concepts of different fields of study, etc. Geographical dialect differences are found primarily in texts produced in different countries, whereas chronological dialect variation mostly affects developing where the level of established vocabulary is still very low (Guilbert 1973).

4. Patterns of terminological variation in Basque

According to a number of studies, the patterns of variation found in sociolinguistically unstable, minority languages that have not been fully normalized such as Basque, differ considerably from those found in normalized ones. The two major distinctions are a greater frequency of free variation and the difference in factors accounting for this variation (Dorian 1981; Dressler 1981; Elordui 2003; King 1989). With these findings in mind and on the premise that the more than 25 years of use of scientific language in academic settings and the press must have produced some progress in discourse, in our study we formulated the following hypotheses: a) the terminological units of the corpus will exhibit variation patterns which can, as in normalized languages, be accounted for by discourse factors, b) but there will also be other variation patterns that can only be accounted for by sociolinguistic factors stemming from the current instability of Basque. This latter type of variation will be highly unsystematic. In this article we are concerned with designative variation — i.e., the phenomenon of designating in different ways a single item having specialized meaning. Designative variation is much easier to detect than conceptual variation, and besides is often reflected directly in the TERMINOLOGICAL VARIATION IN BASQUE 79 alternative designation itself. Classified under this heading are lexical variants (variants that could be considered synonyms), and graphic, spelling and morphosyntactic variants. Designative variants were extracted manually from the corpus and classified according to morphosyntactic pattern.10 The resulting variation patterns were analyzed from two standpoints: a) intratextually, comparing variants within a single text to analyze the author’s fund of terminology and discourse strategies, and b) intertextually, comparing variants found in different texts to find patterns of variation motivated by different discourse conditions (correlating variants to degree of specialization, aurthor’s objectives, etc).

4.1 Analysis of designative variation patterns

To identify and analyze morphosyntactic variation patterns, we used the classification proposed by Freixa (2001, 2003), adjusted to suit the characteristics of Basque and supplemented by other patterns not included in Freixa’s classification. Examples of the patterns found are set out below:

a) Graphic variants: spelling variants (ingeniaritza genetiko/injinerutza genetiko ‘genetic engineering’); variant order of letters in abbreviations (ADN/DNA); hyphen-space alternation (DNA- kate/DNA kate ‘DNA chain’); typographical alternation of quotation marks («ADN zunda»/ADN zunda ‘DNA probe’); term-acronym alternation (Polimerasaren katea-erreakzioa/PCR ‘polymerase chain reaction’); term-abbreviation alternation (ARNm/ARN mezulari ‘messenger RNA’); capital-small letter alternation (Giza Genoma Proiektua/Giza genoma proiektua ‘Human Genome Project’). b) Morphosyntactic variants: subordinant compound N1-N2/attributive compound N2 N1 alternation (ama-zelula/zelula ama lit. mother- cell/cell-mother ‘stem cell’); N-N compound/PP+N sequence alternation (zelula-nukleo/zelularen nukleoa lit. cell-nucleus/cell-

10 Automatic extraction tools for Basque are still in their infancy and took their first steps only recently in the field of variation (Alegría et al. 2004), focusing almost exclusively on morphosyntactic patterns. The relative scarcity of technical dictionaries and texts hinders statistical methods, and for the moment lack of knowledge of discourse markers makes it impossible for automatic detection tools to apply them. 80 AGURTZANE ELORDUI AND IGONE ZABALA

GEN nucleus ‘cell nucleus’); N-N compound/N+Rel.Adj. type syntactic sequence alternation (gene-terapia/terapia geniko lit. gene- theraphy/theraphy genic ‘gene therapy’); N-N compound / derivative alternation (landare-hobekuntza/fitohobekuntza ‘plant improvement/ phytoimprovement’); N+Adj.11/PP+N alternation (gaixotasun hereditario/herentziazko gaixotasun lit. disease hereditary/heredity- GEN disease ‘hereditary disease’). c) Reduction: of extension (izaki bizidunak/izakiak lit. organism life- having/organisms = ‘(living) organisms’); of base (izaki bizidunak/bizidunak lit. organism life-having/life-having ‘(living) organisms’); other types of reduction changing the category of one of the elements (eragile kartzinogeniko/kartzinogeno lit. agent carcinogenic/carcinogen ‘carcinogenic agent/carcinogen’); or Basque term vs. borrowed term alternation (zelula hartzaile/errezeptore lit. cell receptor/receptor ‘receptor cell/receptor’). d) Lexical variants: between two simple Basque terms ( txertatu/erantsi ‘insert’) or a simple vs. a complex Basque term (hari/harizpi lit. thread/ thread-strand ‘strand’); borrowed term vs. complex Basque term (bektore/garraiatzaile ‘vector/transporter’; organismo/bizidun lit. organism/ live-being ‘organism’); scientific term vs. Basque vernacular (zigoto/obulu ernaldu ‘zygote/fertilized egg’); in components of complex units, such as variation in prefix (birkonbinatu/berkonbinatu ‘recombine’) or suffix (klonatze/ klonaketa ‘cloning’), variation in nucleus of a compound (DNA zati/DNA segmentu lit. DNA piece/ DNA segment ‘DNA segment’), variation in compound modifier (hazkuntza-ingurune/kultibo- ingurune ‘growth medium/ culture medium’); in polylexical units such as variation in the base (aztarna genetiko/arrasto genetiko lit. footprint genetic ‘genetic footprint’), variation of extension (aztarna genetiko/aztarna geniko lit. footprint genetic/ footprint genic ‘genetic footprint’; gene azpirakor/gene errezesibo ‘recessive gene’; transkriptasa alderantzikatua/alderantzizko transkriptasa lit. transcriptase reversed/reverse-GEN transcriptase ‘reverse transcriptase’); variation in base and extension (bakterio- genoma/bakterioaren DNA lit. bacteria-genome/bacteria-GEN DNA

11 Note that in Basque, like in Spanish, adjectives are commonly placed to the right of the noun. TERMINOLOGICAL VARIATION IN BASQUE 81

‘bacterial genome/ bacterial DNA’; ernaldutako obozitoa/obulu ernaldua lit. fertilized-GEN oocyte/oocyte fertilized ‘fertilized oocyte’). As noted in previous sections, there are different motivating factors behind these variation patterns. Intratextual and intertextual analysis reveals some that appear to be due to discourse factors. However, many of the patterns detected in the two analyses cannot easily be accounted for by this type of functional motivation, and instead must be explained in terms of sociolinguistic instability. Examples of each type are discussed below.

4.2 Variation patterns due to discourse factors

Intratextual analysis reveals that certain variation patterns are due to discourse factors. For example, authors often resort to graphic alternation to achieve conciseness, as is the case when they use both a term and its abbreviation in the same text: (Polimerasaren katea-erreakzioa ‘polymerase chain reaction’/PCR; Giza Genoma Proiektua ‘Human Genome Project/HGP). Once an author has given the meaning of the abbreviated form, he uses it thereafter in the text for reasons of economy.12 Morphosyntactic variants can be explained in similar terms: e.g. compound/derivative alternation (landare-hobekuntza ‘plant improvement’ /fitohobekunta ‘phytoimprovement’). The effort to be concise and avoid repetition leads authors to use forms of different degrees of succinctness, once they have established the meaning equivalence between variants. Different types of reductions (izaki bizidunak lit. organism live-having ‘live beings’/ izakiak lit. beings ‘organisms’/bizidunak lit. live-having ‘organisms’) can be attributed to the same discourse and stylistic causes — i.e. the desire to be concise and avoid repetition. Somewhat different are patterns involving lexical variants of different types: e.g. one Basque term vs. another (txertatu/erantsi ‘to insert’); simple vs. complex Basque terms (hari ‘thread’/harizpi ‘thread-strand = strand’); Basque term vs. borrowed term (garraiatzaile ‘transporter’/bektore ‘vector’); or Basque vernacular vs. scientific term (obulu ernaldu lit. ovule fertilized ‘fertilized ovule’/zigoto ‘zygote’). Underlying such instances is

12 On some occasions authors resort to a reference language other than Basque for abbreviations (PCR), while on other occasions, the abbreviation stems directly from the Basque sequence (GGP).

82 AGURTZANE ELORDUI AND IGONE ZABALA also the desire to avoid repetition, to sound more erudite or to improve style. Borrowed and scientific terms vs. their vernacular counterparts, and complex vs. simple forms13 are all accorded greater prestige within the linguistic community. There is a different type of discourse reason for quotation marks variation («ADN zunda»/ADN zunda ‘DNA probe’). This pattern seems to be due to the author’s desire to establish the terminological value of a given sequence within the text: the first time the term appears, the author sets it off in quotation marks, but does not use them thereafter once the terminological value of the sequence has been established. Finally, morphosyntactic alternation such as ‘compound vs. syntactic sequence’ or alternation between different types of syntactic sequence may often be due to syntactic-semantic requirements of the language itself. For example, alternation between a compound and a syntactic sequence of the type PP+N (zelula-nukleo ‘cell nucleus’/zelularen nukleoa cell-GEN nucleus ‘cell’s nucleus’) is essential for distinguishing between generic and specific reference. In Basque, the modifier in subordinative noun compounds always has generic reference, unless specific reference is shown syntactically by adjoining a case marker and article, giving rise to a PP+N noun phrase. Another example would be PP+N vs. N+Rel.Adj. (herentziazko gaixotasun lit. inheritance-GEN disease ‘hereditary disease’/gaixotasun hereditario lit. disease hereditary ‘hereditary disease’), an alternation that is often due to the requirement that coordinated elements must belong to the same syntactic category: gaixotasun hereditario eta sexualak lit. disease hereditary and sexual ‘hereditary and sexual diseases’/herentziazko eta transmisioko gaixotasunak lit. inheritance-GEN and transmission-GEN diseases ‘hereditary and transmitted diseases’. Analysis from the intertextual point of view reveals a rise in established discourse rules associated with a text’s degree of specialization and the discourse community to which the author belongs. For example, the presence or absence of hyphens in dependent noun compounds can show whether or not the author belongs to the science community. The standard code allows subordinative noun compounds to be written with or without a hyphen (telefono-zenbaki/telefono zenbaki = ‘telephone(-)number’) and this

13 The choice of a Basque complex form rather than a simple form as a technical term is particularly evident in the case of tautological compounds where both elements have the same semantic content (larruazal ‘skin-skin’, lurzoru ‘ground-ground’), but it is frequent as well in other compounds such as gainazal ‘above-surface = surface’. TERMINOLOGICAL VARIATION IN BASQUE 83 has given rise to different stylistic options. Novelists and journalists tend to avoid this use of the hyphen as much as possible, whereas scientists feel that it not only can, but should be used for the sake of accuracy. So, for example, they write gene-emaile ‘gene-donor = donor of genes’ with a hyphen to show that they are referring to a type of donor, whereas they write zelula hartzaile (lit. cell receptor = ‘receptor cell’) without a hyphen to show that they are referring to a type of cell — i.e. one that is a receptor (of genes). The textbooks and Elhuyar magazine articles analyzed for our study were found to follow the scientists’ policy. The encyclopedic dictionary LUR, on the other hand, and newspaper articles normally do not use the hyphen. We conclude therefore that the presence of hyphens in this type of compound indicates the discourse community to which the author or editor of the article belongs. When we compared different types of texts written by authors all clearly belonging to the same discourse community, since they systematically used the hyphen in subordinative noun compounds, we found inconsistencies in its use with certain other types of compounds (DNA sekuentzia/DNA-sekuentzia ‘DNA(-)sequence’; DNA kate/DNA-kate ‘DNA(-)chain’). Interviews with these authors revealed that such differences are due to conceptual variation. That is, those that use the hyphen see a dependency relationship in the compound (‘the sequence/ the chain is formed by DNA’), while those who leave it out see a relationship of apposition (‘the sequence/ the chain is (called) DNA’).

4.3 Variation patterns due to sociolinguistic factors

Many of the terminological variation patterns found in the corpus were not motivated by functional causes or the exigencies of discourse, but arose rather from factors of a sociolinguistic nature. We feel that the consolidation of specialized terminology is influenced mainly by four such factors. a) First, the instability of the standard code, plus b) the linguistic dependency arising from the language’s minority status, and its hitherto infrequent use in specialized contexts. c) This latter factor causes variation particularly if combined, as in the case of Basque, with strong sociolinguistic pressure on the part of normalizing agents. d) Finally, to the free variation arising from all these factors must be added the lack of fluid communication networks which could contribute to the fixing of specialized vocabulary. 84 AGURTZANE ELORDUI AND IGONE ZABALA

a) Some types of graphic variation are associated with the instability of the code itself. The spelling differences we detected are not due to geographical dialect, as happens in languages such as Spanish or English, which are spoken in vast, well-differentiated geographical areas. In the case of Basque, these variation patterns include recently standardized variants (kate ‘chain’, ingeniaritza ‘engineering’, zientzialari ‘scientist’, xurgatu ‘to absorb’) vs. their non-standard variants (katea, injinerutza, zientzilari, zurgatu), which were used extensively in the linguistic community prior to the recommendations of the official unified dictionary Hiztegi Batua (Euskaltzaindia 2000). Our analysis of the corpus shows that the non- standard variants appear particularly in textbooks and encyclopedia articles (all published before 2000), but are much less frequent in magazine and newspaper articles. Since the latter are published at regular intervals, they can adapt their style-sheets and update them more often, whereas textbooks and encyclopedias are re-edited only after a certain number of years. Therefore they remain in circulation long after their spelling and style- sheets have been rendered outdated by the publication of new standard recommendations. b) As for linguistic dependency, our study of the corpus revealed a tendency to borrow lexical units from the reference language (mainly Spanish) and to retain their original structure. This gives rise to borrowed terms (bektore, organismo) and sequences patterned exactly after the Spanish (terapia geniko ‘terapia génica’ = ‘gene therapy’, aztarna genetiko ‘huella genética’ = ‘genetic footprint’, plasmido birkonbinatzaile ‘plásmido recombinante’ = ‘recombinant plasmid’). A second consequence of linguistic dependency, and more concretely of the failure to use Basque in specialized spheres, is the clear lack of writing skills or discourse competence evident in some of the texts studied. Let us take for example cases in which it is necessary to construct or deconstruct complex terminological units by binding or separating the elements of which they are composed, depending on the exigencies of discourse. In Basque, speakers must be able, for example, to dissociate the items in a subordinative noun compound when one of the elements is modified. Basque nominal compounds are usually formed by two elements. The introduction of a third element asks for a postpositional nexus: transmisio-gaixotasunak lit. transmission-diseases ‘transmitted diseases’/ transmisio sexualeko gaixotasunak lit. transmission sexual-GEN diseases ‘sexually-transmitted diseases’. However, in our corpus we detected cases where authors failed to make the required change and simply formed a TERMINOLOGICAL VARIATION IN BASQUE 85 sequence of three elements without any kind of postpositional nexus, thereby revealing a gap in discourse competence: azido nukleiko harizpi lit. acid nucleic strand ‘nucleic acid strand’; gene-transferentzi sistema ‘gene- transfer system’. Problems with word formation also appeared to stem from poor discourse competence. Here is a small example:

(1) Proteina onkogenikoek minbizia sortzen dute proteina normalen funtzioa partzialki edo modu desegokian imitatzen dutenean, zelula normal bat zelula kantzeroso bihurtuz... ‘Oncogene proteins produce cancer when they partially or inappropriately imitate the function of normal proteins, thus turning the normal cell into a cancer cell …’

In the explanation the author uses the well fixed Basque term. (minbizi ‘cancer’), instead of the loan word (#kantzer). However, when it comes to using the polylexical term ‘cancer cell’, the author resorts to the borrowed Spanish term (zelula kantzeroso), instead of following normal word- formation rules in Basque to produce zelula minbizidun. Such a solution would also have been much more consistent with the development of the discourse. Deficient mastery of the discourse strategies available in Basque to achieve conciseness and avoid repetition is also evident from the way authors use reductions of the base or the extension in polylexical units, as shown in the following example.

(2) Bioteknologiak Injinerutza Genetikoaren eta beste alor batzuen (…) teknikak bildu eta izaki bizidunei ezartzen dizkie gizadiarentzat onurak lortzeko helburuz(…) Azkeneko bi kasuetan, organismo batean material genetikoa sartzeak haren eduki genetikoa aldatzen du. ‘Biotechnology combines the techniques of Genetic Engineering and other fields (…) and applies them to living beings to benefit human life (…) In the last two cases, the introduction of genetic material in an organism alters its genetic content.’

The author resorts to the borrowed term organismo, despite the availability of two reduced Basque forms (izaki ‘being = organism’ and bizidun (lit. life-having = ‘organism’) which would solve the author’s discourse objectives of conciseness and avoidance of repetition of the term izaki bizidun ‘living being’. Note that in Spanish there is no reduced variant for the polylexical term ser vivo (‘living being’) — i.e. *ser or *vivo — and that the only variant available is the monolexical term organismo. 86 AGURTZANE ELORDUI AND IGONE ZABALA

Presumably the author of (2) simply copied this discourse strategy from Spanish. c) As in other languages, and more particularly in minority languages where there is strong sociolinguistic pressure, there are, in addition to variants stemming from the reference language as described above, variants that have been proposed by the different normalizing agents for the purposes of counteracting lexical dependency on Spanish. A case in point is that of N + Rel.Adj. sequences,14 which have been officially labeled foreign to Basque by the Royal Academy of the Basque Language (Euskaltzaindia 1992). From that moment on, numerous linguists, translators and editors have almost systematically replaced this type of sequence with noun compounds. Examples from our corpus include: terapia geniko/gene-terapia ‘gene therapy’; plasmido bakteriano/bakterio- plasmido ‘bacterial plasmid’; ugalketa sexual/sexu-ugalketa ‘sexual reproduction’. Other cases of sociolinguistic pressure include the effort to avoid ill- formed derivatives in Basque through the formal recommendation of supposedly more appropriate terms. Here, as in the previous case, the consequence is terminological dispersion, occasioned by the coexistence of variants that emerged during different phases of linguistic intervention (Gutiérrez 1998: 98). As an example from our corpus, let us take the term recombinant plasmid, for which the following Basque variants exist: plasmido birkonbinatzaile/plasmido birkonbinagai/plasmido birkonbinatu. d) As noted above, in addition to all these factors motivating variation, in the case of Basque there is also the further problem of a lack of fluid communication networks which could help to fix and consolidate specialized vocabulary. In normalized languages word formation processes (derivation and composition) tend to block when a lexical unit for a given notion already exists. In Basque, however, there is an overproduction of these formation processes. In our corpus we have encountered numerous examples of variation associated with this phenomenon: e.g.: klonatze/klonazio/klonaketa for the term cloning, or ama-zelula/zelula ama for stem cell. In some cases, the Unified Dictionary makes a distinction between variants, as for example haziera ‘growth’/hazkuntza ‘offspring, culture’. However, the actual uses that we find in the texts in our corpus show that

14 See Zabala (1997) on Basque equivalents of different types of Spanish relational adjectives. TERMINOLOGICAL VARIATION IN BASQUE 87 the unification recommendations are not always known or followed by writers, who use such variants in free variation. Finally, brief mention should be made of the fact that some authors of texts in our corpus are aware of the lack of an established, specialized vocabulary and attempt to offset this deficiency by resorting to certain discourse strategies. The following example illustrates one available resource — the use of quotation marks for metalinguistic purposes, in this case to show the reader that the term in quotes, eraginpean ‘effect-under = undergo’, is a fixed technical term and is not used by accident or for stylistic reasons:

(3) Sagu-kume hau terapia geniko baten «eraginpean» dago. mouse-baby this therapy genetic DET-GEN «effect-under» is ‘This baby mouse is undergoing gene therapy.’

5. Conclusions

Analysis of the designative variation patterns found in our corpus confirms our opening hypotheses. In both intratextual and intertextual analyses we have found variation patterns motivated by discourse factors — i.e. variants within a single text due to different discourse requirements, and differences within a range of texts that are due to the variety of discourse conditions in each type of writing. These patterns show the progressive establishment of discourse rules both at the intratextual and intertextual levels. However, we have also observed variation patterns that cannot be accounted for by discourse motivations and which seem to be due instead to sociolinguistic factors associated with the incomplete normalization of Basque. As was to be expected, such variants are mostly in free variation. However, we should note that we have also found systematic variation associated with incomplete normalization, more specifically, variation patterns showing that authors attempt to compensate in discourse for the lack of established terminology. There are various sociolinguistic factors that can account for the free variation found in the corpus. a) First, free variation is in many cases motivated by the relative instability of the standard code. As noted above, the use of Basque in specialized areas of study began years before the lexical standardization recommendations issued in the year 2000, a fact that has given rise to spelling and lexical variants fitting the ‘standard vs. nonstandard variant’ 88 AGURTZANE ELORDUI AND IGONE ZABALA pattern. The texts studied show that the standard code made its way more quickly into newspaper and magazine articles thanks to their greater frequency of publication, whereas textbooks, which have a greater influence on the sector of the speech community most likely to enter fields of specialized study, have taken much longer to adapt to the code. b) Other patterns of free variation appear to be due to the pull of opposing forces. On the one hand there is the linguistic dependency that comes from reference terms established in another language, while on the other there is the strong sociolinguistic pressure so often present in languages undergoing normalization, which prefers the creation of neologisms to borrowed terms or words patterned on the formation rules of other languages. The natural consequence of such a conflict is terminological dispersion. Different experts, translators and institutions come up with different solutions to terminology problems. Moreover, the lack of fluid communication networks between speakers in the same field of study is a further impediment. The consequence observed in our corpus is an overproduction of variants occurring in free variation. c) Finally, free variation is also related to authors’ linguistic proficiency and writing skills, and to their knowledge of the specialized vocabulary of the field in question. The corpus reveals a broad range of skill in these areas. Some authors tend to use variants stemming from the reference language and, as intratextual analysis shows, variants that reveal a certain lack of discourse competence. Such failings are particularly evident when the author attempts, for discourse reasons, to generate in Basque terms drawn from other languages that require the adaptation of components of polylexical items. To conclude, this preliminary study of terminological variation in Basque has led us to a number of criteria that should be borne in mind when undertaking the normalization of a language such as Basque. Indeed, the first two take the form of recommendations specifically for this language, while the rest should, we feel, be borne in mind in any process of linguistic normalization: 1. Variation due to instability of the standard code appears likely to lessen as the code becomes stabilized. However, we feel that a normalizing initiative designed to update all the publications in the market, particularly textbooks and teaching materials, would help to provide the community with an input that would at least not foster free variation. 2. As for terminological dispersion, we assume that the normalizing activity of the Terminology Committee recently created by the Basque TERMINOLOGICAL VARIATION IN BASQUE 89

Government Department for Linguistic Policy will help to establish reference terms, thereby attenuating the situation. For any language currently in the process of modernization and seeking to normalize the terminology of specialized fields of study, we would make the following general recommendations: First, it is crucial to recognize that incipient systematic variation due to discourse reasons should be seen as grounds for promoting such variants, particularly in the academic world. Second, we would recommend pursuing a linguistic policy flexible enough to allow different users to contribute to the discursive development of the language. The results of our study show that in cases where the standard model is flexible (use of a hyphen in subordinative noun compounds), usage rules have emerged in different Basque discourse communities, whereas overly rigid normalizing recommendations (e.g. N+Rel.Adj. sequences considered foreign to the language) simply foster terminological dispersion. Finally, we consider that normalizing activity should focus not on the eradication of free variation, but rather on channeling it. Today’s free variation could provide tomorrow’s established repertoire of functional variants for different discourse requirements.

References

Alegría, Iñaki et al. (2004) Linguistic and Stadistical Approaches to Basque Term Extraction. In Actes de GLAT-Barcelona 2004, pp. 235-246. Barcelona: Universitat Politécnica de Catalunya. Auger, Pierre (1994) Pour un modèle variationniste de l’implantation terminologique dans les entreprises au Québec. In Colloque sur la problématique de l’aménagement linguistique: Actes du Colloque, pp. 483-493. Quebec: OLF. Boulanger, Jean-Claude (1983) Synonymie, néonymie et normalisation en terminologie. In Duquet-Picard, D. (ed.) Problèmes de la définition et de la synonymie en terminologie. Actes du colloque international de terminologie, pp. 311-327. Quebec: Girsterm. —— (1991) Une lecture socioculturelle de la terminologie. Cahiers de Linguistique Sociale 18: 13-30. Cabré, Maria Teresa (1999) Elementos para una teoría de la terminología: hacia un paradigma alternativo. In Maria Teresa Cabré (1999) La terminología. Representación y comunicación, pp. 69-92. Barcelona: Institut Universitari de Lingüística Aplicada, Universitat Pompeu Fabra. Ciapuscio, Guiomar (2003) Textos especializados y terminología. Barcelona: Institut Universitari de Lingüística Aplicada, Universitat Pompeu Fabra. Dorian, Nancy (1981) Language Death: The Life Cycle of a Scottish Gaelic Dialect. Philadelphia: University of Pennsylvania Press. 90 AGURTZANE ELORDUI AND IGONE ZABALA

Dressler, Wolfgang Ulrich (1981) Language shift and language death: a protean challenge for the linguist. Folia Linguistica XV/1-2: 5-27. Elordui, Agurtzane (1995) Hizkuntz heriotza eta aldaketa morfologikoa: Hegomendebaldeko bizkaieraren kasua. Ph Dissertation. University of the Basque Country. —— (2003) Variation in the Grammar of Endangered Languages: The Case of Two Basque Dialects. SKY Journal of Linguistics 16: 7-26. Euskaltzaindia (1992) Hitz elkartuen osaera eta idazkera. Bilbao: Euskaltzaindia. —— (2000) Hiztegi Batua. Euskera 2000, II volume. Bilbao: Euskaltzaindia. Freixa, Judit (2001) Reconocimiento de unidades denominativas: incidencia de la variación en el reconocimiento de las unidades terminológicas. In Maria Teresa Cabré & Judit Feliu (eds.) La terminología científico-técnica: reconocimiento, análisis y extracción de información formal y semántica, pp. 57-65. Barcelona: Institut Universitari de Lingüística Aplicada, Universitat Pompeu Fabra. —— (2003) La variació terminòlogica. Anàlisi de la varició denominativa en textos de diferent grau d’especialització de l’àrea de medi ambient Ph Dissertation. Universitat Pompeu Fabra. Barcelona. Gaudin, François (1993) Pour une socioterminologie: des problèmes pratiques aux pratiques institutionnelles. Rouen: Publications de l’Université de Rouen. Guespin, Louis (1991) La circulation terminologique et les rapports entre science, technique et production. Cahiers de Linguistique Sociale 18: 59-79. Guilbert, Louis (1973) La spécificité du terme scientifique et technique. Langue Française 17: 5-17. Gutiérrez, Bertha M. (1998) La ciencia empieza en la palabra. Análisis e historia del lenguaje científico. Barcelona: Editorial Península. King, Ruth (1989) On the social meaning of linguistic variability in language death situations: Variation in Newfoundland Frech. In Nancy Dorian (ed.). Investigating obsolescence. contraction and death. Studies in the Social and Cultural Foundations of Language 7, pp. 139-148. Cambridge: Cambridge University Press. Lekuona, Manuel (1968) Euskeraren batasuna. The Conference of Arantzazu. Euskera XIII: 139-141. Lotte, Demitrii Semenovich (2001[1948]) Condiciones para la precisión y la claridad de la terminología. In Maria Teresa Cabré, Judit Freixa, Mercè Llorente y Carle Tebé (eds.) Textos de Terminología de la Escuela Rusa, pp. 131-150. Barcelona: Institut Universitari de Lingüística Aplicada, Universitat Pompeu Fabra. Mitxelena, Koldo (1968) Ortografia. The Conference of Arantzazu. Euskera XIII: 202- 219. Wichter, Sigurd (1994) Experten-und Laienwortschätze. Umriss einer Lexikologie der Vertikalität. Tübingen: Niemeyer. Wüster, Eugen (1998[1979]) Introducción a la teoría general de la terminología y a la lexicografía terminológica. Maria Teresa Cabré (ed.). Barcelona: Institut Universitari de Lingüística Aplicada, Universitat Pompeu Fabra. www.euskadi.net TERMINOLOGICAL VARIATION IN BASQUE 91

Zabala, Igone (1997) Argumentu-harremana/ eremu-harremana: izenondo erreferentzialen euskal ordainen bila. In International Congress on Terminology, pp. 535-556. San Sebastián: IVAP-UZEI.

Contact information:

Agurtzane Elordui Euskal Filologia Saila Gizarte eta Komunikazio Zientzien Fakultatea Euskal Herriko Unibertsitatea Sarriena auzoa z./g. 48940 Leioa Spain e-mail: [email protected]

Igone Zabala Euskal Filologia Saila Zientzia eta Teknologia Fakultatea Euskal Herriko Unibertsitatea Sarriena auzoa z./g. 48940 Leioa Spain e-mail: [email protected]

Heidi Koskela

Invoking Different Types of Knowledge in Celebrity Interviews1

Abstract

My aim in this paper is to examine how participants in celebrity television interviews invoke different types of knowledge and move between first-hand and general knowledge. The data that I use come from Finnish television interviews where foreign celebrities are interviewed. In the analysis I describe the resources the participants use in mobilizing different types of knowledge. First I describe how the interviewer’s questions invoke first-hand knowledge and then I move on to describe one way of resisting the agenda set in the interviewer’s questions – namely that of displaying general knowledge instead of first-hand knowledge in the answer. These practices serve the functions of 1) managing the intimacy-level of the topics, 2) resisting some aspect of the interviewer’s question and 3) constructing expertise that is based on general knowledge.

1. Introduction

Celebrity interviews introduce celebrities, or some aspect of them, to the television audience. Celebrity interviews as a genre require careful negotiation of the level of intimateness. Since the goal of the interviews is to reveal personal aspects of the interviewees, a certain level of intimacy has to be achieved. On the other hand, there are topics that the interviewees do not wish to talk about in a television interview. The aim of this article is to examine how the participants can invoke different types of knowledge and move between first-hand and general knowledge in a manner relevant for the activity-at-hand and for the management of topic and knowledge in celebrity interviews.

1 I would like to thank Pentti Haddington and two anonymous reviewers for their thorough and most helfpul comments on an earlier version of this paper. I am also grateful to Arja Piirainen-Marsh for her comments and discussions which have helped me with this text. All remaining errors and inaccuracies are naturally my own responsibility.

SKY Journal of Linguistics (18) 2005, 93–118 94 HEIDI KOSKELA

My data come from Finnish television interviews where foreign celebrities are interviewed. I present examples where the interviewer’s (IR) question invokes first-hand knowledge and the interviewee (IE) resists the agenda set by the question and displays general knowledge in the answer. The focus of this paper is to describe 1) how first-hand knowledge is invoked in the questioning turn and 2) one way of resisting the IR’s questions about issues that are too intimate, ‘loaded’, or in some other way problematic matters – namely that of displaying general knowledge instead of first-hand knowledge in the answer. There are other ways of resisting the interviewer’s agenda that include, for example, refusing to answer the question, providing a partial or incomplete answer, and changing the topic (see Clayman & Heritage 2002: 250–257). Some of the practices of resisting or shifting the IR agenda are quite overt (for instance explicitly refusing to answer the question) and others, such as the one that I am focusing on here, are more covert. When the IR agenda is shifted covertly, the IEs do not explicitly acknowledge the shift of the agenda (see Clayman & Heritage 2002: 269). Mobilizing different types of knowledge is a salient feature of interaction in television interviews. Different states of knowledge are always present in the questioning activity one way or another. A television interview is a question-driven form of interaction and the situation itself brings with it an asymmetry that is linked to the institutional roles of the participants. The roles of the interviewer (as a questioner) and the interviewee (as an answerer) already in themselves involve an asymmetry of knowledge. Negotiating knowledge, roles and identities are connected with each other. Each relevant role in the interactive event is linked to a relevant state of knowledge that is made available for the other participants through interactive practices. Knowledge in this study refers not to whatever mental constructions might be lodged inside the participants’ minds but to positions that are constructed in interaction. The question that interests me is how knowing and situationally appropriate or relevant ways of knowing are managed interactionally. Recently this topic has been approached by Heritage & Raymond (2005) who studied how knowledge and information are managed in affiliative assessment sequences in everyday talk. In this article the management of knowledge is studied in a different environment, namely question-answer sequences in an institutional context. The practices of managing knowledge and resisting or shifting the IR agenda that I INVOKING DIFFERENT TYPES OF KNOWLEDGE IN CELEBRITY INTERVIEWS 95 examine in this paper have not been studied in the context of celebrity interviews before. Closely linked with the asymmetry of knowledge is the asymmetry of participation rights that is also connected with the participants’ institutional roles. However, it should be remembered that the division of roles is naturally not this straightforward in the actual interactive event. The participants may not adhere to their institutional roles throughout the interview. Instead, they may orient to some other roles as more relevant ones at any particular moment.

2. Background

I approach my data from a conversation analytic perspective. I examine the interviews as interactive events constructed through social practices. The interviewers and interviewees recognize a set of social conventions that are associated with interviews. These social conventions constitute the interviews as an organized social institution (Clayman & Heritage 2002: 6). Practices used in constructing knowledge cannot be studied in isolation from the interactional context, i.e. the turns and sequences in which the practices occur. Thus a method that acknowledges the "in-progress" character of interactive practices, in other words participants analyzing turns as they unfold, is needed. Conversation analysis provides a method for capturing the array of features of interaction that are relevant for the participants and for the analyst. Conversation analysis has proved to be a useful approach in examining the practices that sustain the interview. Television interviews have been studied from a conversation analytic perspective by Clayman (1988, 1992); Greatbatch (1988); Heritage (1985); Heritage & Greatbatch (1991); Clayman & Heritage (2002) among others. Finnish television interviews have been studied by, for example, Nuolijärvi & Tiittula (2000); Berg (2001, 2003) and Kajanne (2001a, 2001b). Many conversation analytic studies on television interviews have focused on news interviews. Other types of interviews have received less attention among researchers conducting conversation analytical studies (however, see, e.g., Hutchby 2001b on talk shows and Clayman & Heritage 2002b on press conferences). There are many different genres of interviewing in addition to the news interview, e.g., the press conference, the talk show interview, the sports interview etc. The different interview genres share certain similar properties, for instance, all interviews are 96 HEIDI KOSKELA primarily organized through questions and answers and there are also similarities in the interviewer and interviewee conduct in producing talk for an overhearing audience. Moreover, similar IR techniques are used in the opening and closing phases of the interview. In the opening phase the IR produces a monologue addressed explicitly to the audience. This monologue typically includes an introduction of the guest and a statement about the topic of the program (see Clayman & Heritage 2002: 59–60). In closings, the IRs typically initiate the closing, usually by thanking the IEs for their participation (Clayman & Heritage 2002: 74). However, there are also differences in the nature of questioning in different interview genres. In political interviews the questioning is often aggressive, attempting to corner the IE or to provoke debate (see, e.g. Nuolijärvi & Tiittula 2000: 83, Heritage 1985, 2002). The politicians’ answers in turn are shaped by the questioning style and also by the institutional norms of politics. In news interviews it is central for the IRs to retain a neutralistic stance towards the IEs’ statements and opinions (Clayman & Heritage 2002: 120). In talk show interviews the function of the questions is to get the guests to talk about themselves and the questioning is often done in a way that enables the host to express their own views and share their own experiences with the guests and the television audience (Nuolijärvi & Tiittula 2000: 85–88). Typically in talk shows an audience is present in the studio, which also influences the shape the interaction takes. The purpose of a celebrity interview is different from other interview genres and that can be seen in the way the interview is organized. The practices I analyze in this paper in part organize the interview so that it meets its purpose of introducing the celebrities to the television audience. The interviewer's institutional role as the controller of the topic and the agenda has previously been studied by Clayman and Heritage (2002). Previous research on the ways in which the IEs resist IR agenda include studies by Clayman (1993, 2001) and Greatbatch (1986). Clayman and Heritage (2002: 196) describe how agendas are set in questions by setting a specific topical domain as the relevant domain in the response. If the interviewee fails to address the question's topical agenda, the failure is made noticeable and accountable (see also Schegloff 1972). According to Clayman (2001), when interviewees resist the interviewer agenda covertly they minimize the possible negative consequences of being evasive. In the data extracts that are presented here resistance is done in a similar manner, in a way that makes resistance less conspicuous. INVOKING DIFFERENT TYPES OF KNOWLEDGE IN CELEBRITY INTERVIEWS 97

3. The data

The data I use come from Finnish television interviews. The IR is Finnish and the IEs come from a variety of linguistic and cultural backgrounds. The language the participants use in the interviews is English. The data come from a larger collection of data of using English in the Finnish society in the domains of media, education and business life.2 Because the participants do not share a common native language, they use English as an international language in the interviews. However, they generally do not seem to orient to the fact that they are using an international language. Instead, they orient to the norms of the interview or ‘do interview talk’ and do similar things with language as native speakers in a similar situation. The data analysed here come from four different interviews, which are part of a series of interviews called "Yölento" (“Night Flight”), hosted by Maarit Tastula. The data are transcribed using the notation system summarized in the Appendix. The interviews consist of the participants’ talk that takes place in a television studio (without a studio audience) and video inserts that are placed within the talk. The genre could best be described as celebrity interviews. The general theme has to do with some aspect of the interviewees that they are famous for, their public roles or identities. For example, a film director is invited to talk about his films, a civil rights activist is invited to talk about civil rights etc. Whether the interviewees speak as their private selves, being 'experts' of their own life and of things that they personally have experienced, or as 'experts' of some specified field (usually a profession, but also a nationality etc.), is something that is jointly negotiated in the interaction. One means of such negotiation is the mobilization of different types of knowledge.

4. Questioning and knowledge

Questioning is an activity that has a central role in constituting the news interview as a social institution (Heritage & Roth 1995: 2). Questions are often complex and multifunctional. Questioning turns can also serve as vehicles for doing other actions besides questioning. Agendas can be set, assumptions and opinions can be expressed and presuppositions can be

2 The project English Voices in Finnish society: the use of English in media, education and professional settings is based at the University of Jyväskylä and financed by the Academy of Finland (project number 7102075). 98 HEIDI KOSKELA made in questioning turns (see Clayman & Heritage 2002). Questions can be used to accuse, challenge etc. (Heritage & Roth 1995). Because questions are multifunctional, they can also be studied from many different perspectives. The perspective I take in this paper to questioning is how different types of knowledge are mobilized in question/answer sequences. Aspects of the interviewer’s and interviewee’s institutional or public roles and also their private identities can be invoked in questions (see Clayman & Heritage 2002; Roth 1998). For example, a question can be designed in such a way that the interviewees are treated as members of a nationality, and through that membership as knowing participants in relation to their home country, or as members of a profession and through that membership knowledgeable about matters relevant to their profession. Categorization work is being done in these instances where a question invokes an identity of a member in some group and because of that membership treats the interviewee as a knowing (or unknowing) participant (see Piirainen-Marsh & Koskela 2000). A person belonging to a certain category has entitlement to certain knowledge (cf. Sharrock's (1974) idea of "ownership of knowledge"); for instance, when talking about matters concerning one participant’s home country that participant has entitlement to that knowledge. This paper focuses on instances where the interviewer’s question invokes first-hand knowledge and the interviewee displays general knowledge. I attempt to identify the resources that the participants use in mobilizing these two types of knowledge. I suggest that the function of invoking first-hand knowledge has to do with managing the intimacy level of the interview. In these data the genre of the interviews (human interest, personal matters) calls for talk about the private aspects of the interviewees’ identities. While the IRs attempt to build positions from which the IEs could answer the questions on the basis of first-hand knowledge, the IEs sometimes resist these positions for their own purposes. Something in the questions is either too intimate or somehow ‘loaded’. Instead of offering first-hand knowledge, the IEs’ answers are designed in such a way that they invoke general knowledge. Before moving on to the examples that are the focus of my analysis I will present an example of a case where the interviewer’s question that invokes first-hand knowledge is not resisted, but the interviewee produces talk that displays first-hand knowledge in response to the question. This type of question-answer sequence is very typical of celebrity interviews. In INVOKING DIFFERENT TYPES OF KNOWLEDGE IN CELEBRITY INTERVIEWS 99 this example it can be seen what happens if the question is not treated as problematic in any way and the IE designs his answer to fit the IR agenda.

(1) MT = IR, Maarit Tastula EK = IE, Emir Kusturica

Video insert

1 MT → so (.) how did you (.) react yourself (.) 2 when you first uh heard about(.) the war (.) 3 that the war had be[gun] 4 EK [it ] was very painful it was awful 5 EK I just couldn’t believe that= 6 MT =you were (.) where (.) at that time.= 7 EK =u-uh Paris. I was editing the movie and I was (.) 8 crippled up (.) you 9 know I was °hh it was (.) the (.) uh really the point 10 in which (.) my existence was (.) like under the (.) 11 strongest (.) 12 uh atomic bomb uh- i- if if (.) 13 if (.) I will (.) see atomic bomb destroying the- the the 14 earth °hh I would say (.) even (.) 15 that it happened to me (.) 16 °hh my father and mother who were in Hertsegnovy? 17 and I was (.) openly: (.) like uh dying 18 because the war started I was °hh twenty-four hours 19 on the phone, (.) trying to connect to do: (.) 20 and I´m very proud because when the war- uh-n- 21 little bit (.) uh th-th- one detail when the war started, 22 (1.0) 23 I was calling the general who was (.) 24 keeping Sarajevo trying to do (.) something (.) 25 that (.) they do in (x) and other places together now. 26 to- to do the ( . ) uh military: 27 and the police that was (.) mixed in between (.) 28 muslim serbs and croats. 29 because I’ve heard about this formula. 30 °hh but everything failed ( . ) 31 because it was not in the hands of any individual. (.) 32 it was °hh mostly in the hands of d- 33 o:f of the dogs of the war.

The IR asks about the IE’s past experiences (lines 1–3). The question about past reactions (how did you (.) react yourself (.) when you first uh heard about (.) the war…) invokes first-hand knowledge. Reactions can include feelings, emotions, or actions. All of these are something that the person who has experienced the feelings or emotions or taken the actions has first- hand knowledge of. The invoking of first-hand knowledge is further strengthened by the reflexive yourself in the question. After the IE 100 HEIDI KOSKELA starts to answer the question (lines 4–5) the IR interrupts the answer and asks another question (you were (.) where (.) at that time, line 6) This question is treated as a side-sequence and answered briefly by the IE before moving on to continue the previous answer. In his answer the IE answers from the position that is built for him in the question and tells about his feelings (crippled up in line 8, like dying in line 17) and also the actions that he took (I was twenty-four hours on the phone, lines 18–19 and I was calling the general, line 23). Example (1) shows a typical and unproblematic question-answer sequence in a celebrity interview: the IR asks a question that invokes first- hand knowledge and the IE designs his answer to accommodate the type of knowledge that is attributed to him. In the examples that are the focus of my analysis I will examine question-answer sequences that are somehow problematic and in which the IR agenda is shifted. These sequences, although not very frequent, are still recurrent and an interesting part of constructing a celebrity interview and negotiating the limits of questioning in a celebrity interview.

5. Invoking first-hand knowledge

The IRs can formulate their questioning turns so that specific knowledge positions are built as relevant for the answers. In the following I am going to introduce different practices of invoking first-hand knowledge in the questions: 1) directing topic to matters that the IE has personally experienced or otherwise has first-hand access to and 2) explicitly voicing the ‘personal’ viewpoint that is called for.

5.1 Directing topic to matters to which the IE has first-hand access

There seem to be specific ways to invoke first-hand knowledge in a questioning turn. One of these, and perhaps the most obvious, is to select the topic so that it deals with matters that the interviewee has first-hand access to. This could mean things the IE has personally experienced or witnessed in the past or knowledge that the IEs, because of their membership in a certain category, have entitlement to. In example (2) the host invokes knowledge based on personal experience by her question about the guest’s father (was your father a patriarch of the family?, lines 12–13). INVOKING DIFFERENT TYPES OF KNOWLEDGE IN CELEBRITY INTERVIEWS 101

(2) MT = IR, Maarit Tastula EK = IE, Emir Kusturica

→ 12 MT =°hh but by the way, was your (.) father 13 a patriarch of the family. 14 (.) 15 EK somehow yes. 16 (.) 17 MT in which way. 18 EK °hh >but< you know, patriarchs here are very much (.) 19 u:h uh like uh uh th- they are- 20 it’s not like (on) the west you know if you are father 21 in the family:, 22 MT mm-h 23 EK you have to deser::ve to be liked (.) to be loved 24 by your family members ↑here (.) you are very u:h 25 ( . ) comfortable with uh you know if you are father. 26 °hh you want- it- it’s understandable that everybody 27 has to like you even if you don’t do all the time 28 the best things for the family. 29 which is u:h (.) pattern of: uh of a life here. 30 so, °hh- my father was a patriarch and uh (.) 31 somehow he was u:h (.) I would say that (.) u:h uh uh 32 (.) remembering now, certainly I’m (.) over forty now 33 and I- °hh when I remember him and everything 34 that was going on it- it is uh °hh (.) uh uh 35 already memory: and a memory always has 36 (.) uh certain em:otion and certain sentiments 37 involving to this.

The question that invokes first-hand knowledge about the IE´s father (lines 12–13) is produced in the form of a “yes/no” question, which projects a certain, limited, answer type including an affirmation or a negation. The IE’s answer (line 15) is very short and produced after a pause. Clayman and Heritage (2002: 113) have found that interviewees tend to offer such brief responses when faced with questions that they to. Delaying the answer with a pause can also be seen as an indicator of disalignment with (some aspect of) the question. In this example we can see different ways in which the interviewee treats the question as somehow inappropriate: first, the answer is delayed, then the IE disaligns with the question and does not provide the answer type (affirmation or negation) that is projected by the question as such but instead answers somehow yes, which implies that the question cannot be answered simply by affirming or negating, but requires more. Although the question is treated as somehow inappropriate, the answer still accommodates both sides by including in it the projected 102 HEIDI KOSKELA answer type (even if it is not produced in the pure form of either an affirmation or a negation) and thus the risk of threatening the IR’s face is reduced. In many environments a brief answer to a “yes/no” question is appropriate. However, in television interviews, and especially in this television interview genre, a closed question followed by a brief answer does not adhere to the institutional norm. Short answers can make the interaction seem halting and disrupted, which would not support the institutional aim of producing entertaining or informative interaction for the viewers to watch. In general, questions in celebrity interviews elicit talk about the IEs themselves and introduce the IEs to the television audience and “yes/no” questions have the same function. After a micropause (line 16), which would still be an opportunity for the IE to go on with his answer, the interviewer continues with a follow-up question (in which way, line 17) that invites the IE to explicate how his father was a patriarch. In this example the IE does not treat the question (lines 12–13) entirely appropriately in its context (a celebrity interview) because he provides a very brief answer. Thus a follow-up question (line 17) needs to be added. In most cases in my data after the type of question that 12–13 represents, the IE produces an elaborated answer that contains the affirmation or negation followed by additional talk on the topic. Clayman and Heritage (2002: 245) have called this type of answering "minimal answer plus elaboration". In this answer type the orientation to the institutional requirements of the interaction are clearly oriented to. As example (2) shows, if the institutional demands are not met, the IR orients to those demands by adding a follow-up question. In example (2) the IE does mobilize first-hand knowledge and a “minimal answer plus elaboration”-type of answer later, starting in line 30 (my father was a patriarch and uh (.) somehow he was u:h (.) I would say that (.) u:h uh uh (.) remembering now,...) but does this only after displaying general knowledge about the topic (lines 18–29). The IE’s answer is designed in a way that enables him to deal with the problematic aspects of the question first, before displaying first-hand knowledge. In example (3) the participants have been talking about the fact that the IE has suffered from anorexia nervosa as a teenager. The IE has talked about developing anorexia and the feelings that he had at that time (lines 1– 4). After this the IR asks a question about the reasons for developing the eating disorder (lines 5–6). The IR smiles while asking the question and the IE answers the question also smilingly. The next question (lines 11–12) INVOKING DIFFERENT TYPES OF KNOWLEDGE IN CELEBRITY INTERVIEWS 103 contrasts the mood set by the previous question and returns the talk back to serious mode. This can be seen both in the verbal elements of the question (contrasting the question with the previous talk with but and the follow-up it was actually very serious at some moment) and the change in the IR´s facial expression (she stops smiling at this point). In this turn (lines 11–12), which is then continued further (line 14) the IR invokes first-hand knowledge and the IE starts displaying general knowledge in his answer (line 15 onward).

(3) MT = IR, Maarit Tastula SK = IE, Stephen Kuusisto

1 SK I felt very very (.) sa:d about who I was you know 2 I didn´t fit in anywhere I (.) I felt ugly 3 I felt that I didn´t belong in the world you know (.) 4 and I stopped eating. I became ↑very depressed u:h 5 MT did you start admiring (.) John Lennon, o-----IR smiles----- 6 [his ] skeletal [figure] ------o 7 SK [yeah] [well ] yeah I wanted to look (x)- o---IE smiles------8 SK I wanted to look like all those ------9 who were really skinny. you know (.) that´s how ------o 10 it started. → 11 MT but it was actually very serious at some moment (.) → 12 so you were um:: brought into hospital 13 SK ºyeah.º → 14 MT [it was only] hundred and five po[unds] 15 SK [u:hm ] [u:h ] we- we know 16 SK from psychological studies that the (.) 17 you can (0.6) uh (.) avoid (.) self-destructive tendencies (.) 18 is to: have a belief that life is possible, 19 that it´s worth living, that there´s meaning ahead (.) 20 u:h (.) you know (.) anorexic teenagers 21 the ones who stop eating (.) in fact believe (.) 22 that (.) the future will be terrible. 23 you know (.) anorexia is a disease of (.)psychological 24 dimensions >right< you don´t wanna become an adult. 25 because you think (.) boy that will be worse (.) 26 it´s bad now it´ll be worse at the next age (.) 27 >you know< I´m getting off the train right here (.) 28 you know I´m gonna stop (.) right here I´m not going on. 29 u:hm I think that´s what was going on with me, 30 I really felt ₤oh my god₤ you know I mean 31 .hh I´ve been (.) beaten up on the playgrounds 32 I´ve been called a martian, (.) 33 u:h they´ve (.) you know made me feel (.) 104 HEIDI KOSKELA

34 u:h you know small and and u:h (.) 35 treated me to cruelties and uh I don´t fit in, 36 I don´t belong,

The IR’s question (but it was actually very serious at some moment (.) so you were um:: brought into hospital [..] it was only hundred and five pounds) does not have interrogative syntax, but it is a question in the form of a declarative. Declaratives about issues which the IE has particular knowledge about have been called “b-event questions” (see Clayman & Heritage 2002: 102, also Pomerantz 1980). They function as questions that require confirmation from the IE. They are often constructed in the ”you + progressive/imperfective verb”-format (see Clayman & Heritage 2002: 102). In this example one element of the question (so you were um:: brought into hospital) is produced in this format. The question (lines 11–12 and line 14) continues to invoke first-hand knowledge, just like the previous question, but now returning back to serious mode. The question deals with an issue that the IE has first-hand access to because of his personal experience. In the question the IR displays knowledge about things that have happened to the IE, but the IE has stronger rights to this knowledge. The IE orients to the IR’s turn, produced in the form of a declarative, as a question. Instead of displaying first-hand knowledge he starts to answer the question by displaying general knowledge. There are different linguistic and interactional resources the IR can use in invoking first-hand knowledge. Often this is done through directing the topic to matters that deal with the IE’s personal history or past conduct or events that the IE has witnessed. When the IR asks such a question, it is often asked using the . This is in line with the topical content of the questions. When asking a question about matters that somebody has experienced or witnessed in the past, the past tense is naturally a logical tense to use, as the following examples will illustrate.

INVOKING DIFFERENT TYPES OF KNOWLEDGE IN CELEBRITY INTERVIEWS 105

(4) MT = IR, Maarit Tastula EK = IE, Emir Kusturica

→ 12 MT =°hh but by the way, was your (.) father 13 a patriarch of the family.

(5) MT = IR, Maarit Tastula EK = IE, Stephen Kuusisto

→ 11 MT but it was actually very serious at some moment (.) 12 so you were um:: brought into hospital 13 SK ºyeah.º 14 MT [it was only] hundred and five po[unds] 15 SK [u:hm ] [u:h ] we- we know

(6) MT = IR, Maarit Tastula HB = IE, Harry Belafonte

→ 1 MT and when did you have your first personal experience 2 with (.) racism.

While past conduct, past experiences and events that the IE has witnessed form one basis for first-hand knowledge, as has been the case in the previous examples, another basis for first-hand knowledge is subjectivity and people’s primary access to their current thoughts, opinions, feelings etc.

5.2 Explicitly voicing the “personal”

Besides directing the topic to personal matters and asking questions about personal history or past experiences, invoking first-hand knowledge can be done by explicitly voicing the “personal” point of view that is sought for. This is done both in questions on topics where the IE’s epistemic authority is based on personal experiences in the past and in questions about the IE’s current mental processes (i.e. thoughts, opinions) where the ownership of mental processes is the basis of epistemic authority. An environment where this type of action seems to occur is in questions that include a shift in topic. IRs the “personal” aspect especially when shifting the topic from a more general level to the same topic matter, but now including a personal perspective. Voicing the “personal” in the question imposes quite strong constraints on the way the question is appropriately answered. 106 HEIDI KOSKELA

(7) MT = IR, Maarit Tastula HB = IE, Harry Belafonte

1 HB much that is wrong with America today, .hh within 2 the black community. 3 °hh can be directly related to the fact that u:h (.) 4 °hh the elite of that community (.) 5 has left to go live elsewhere 6 °hh and those who cannot afford to leave, are trapped. 7 MT ºmmº= 8 HB =and they have no (.) icons they have no (.) role models 9 ºhh when I was a boy u:h all the (.) great minds (.) 10 lived right next door. 11 MT m[m ] 12 HB [now] you- they live (.) very far away. → 13 MT and when did you have your first personal experience 14 with (.) racism. 15 HB when you live within (.) the segregated society (.) 16 or in a segregated community 17 experience with race within the black community 18 is quite minimum

In example (7) the topic prior to the IR’s question in line 13 has dealt with racism in the United States. When the IR asks the question about the IE’s personal experiences with racism the topic is shifted from a general to personal level. In example (8) a video insert precedes the IR´s question about the meaning of love (lines 1–3). In the video insert (taken from one of Woody Allen´s movies), a monologue about the definition and psychological explanations of love is shown. After the video insert the IR shifts the topic so that it now includes a personal perspective. The background statement (line 1) which precedes the question is still quite general, but as the question emerges, the focus moves more and more to the personal level.

(8) MT = IR, Maarit Tastula WA = IE, Woody Allen

1 MT love is very important in your films. → 2 what is the meaning of love (.) for you in your- → 3 °hh uh in your movies, and in your (.) own life= 4 WA =well. It- it’s too big a question to ask- 5 [ to answer ] so °hh u::h uh uh succinctly at all 6 MT [°((laughs))°] 7 WA °hh y-you know but the- the- the human interaction 8 °hh u:m th- between (.) a man and a woman, 9 between ((coughs)) a mother and a chi:ld, 10 or °hh brother and sister, brother and brother, INVOKING DIFFERENT TYPES OF KNOWLEDGE IN CELEBRITY INTERVIEWS 107

What all of the above questions have in common is that they invoke first- hand knowledge. In example (2) was your father a patriarch of the family invokes first-hand knowledge that is gained through living in the family. The IE is expected to talk about the characteristics of his father that are based on his own experiences with him. In example (3) but it was actually very serious at some moment… the first-hand knowledge is based on the IE’s past experiences. In example (7) the first-hand nature of the knowledge that is invoked is emphasized through including the “personal” or first-hand aspect explicitly in the question (your...personal experience). Similarly, in example (8) the IR makes first-hand knowledge relevant through the explicit mention of your own life (line 3). Now that I have examined the IR's questions and seen how first-hand knowledge is invoked by 1) selecting the topic so that it is about things that the IE has personally experienced or has first-hand access to or 2) explicitly voicing the 'personal' aspect, I can take a look at how the IEs handle these questions.

6. Resisting IR agenda: Displaying general knowledge in the answer

Usually in celebrity interviews when first-hand knowledge is invoked in the questioning turns the IEs answer in a way that accommodates the type of knowledge attributed to them. However, they can design their answers in other ways as well. In this section I will analyze instances where the IEs display general knowledge in their answers. In some cases, if the question is treated as somehow problematic (e.g. too intimate, somehow ‘loaded’ or containing an incorrect presupposition), the IEs can first orient to the problematic aspect of the IR’s question and after doing this answer the question on a more general level. In the following I will first show how the IEs can orient to the question (or some part of it) as problematic, then I will analyze how the IEs mobilize a different knowledge type and display general knowledge in their answer.

6.1 Orienting to (some aspect of) the question as problematic

One of the resources the IEs use in resisting the IR agenda is orienting to that aspect of the question which is somehow problematic by using contrasts to mark the difference in knowledge types in questions and 108 HEIDI KOSKELA answers. In (9), instead of following the line of the IR’s questioning and talking about his personal experiences, the IE offers a general rule about fathers (and patriarchs) in his native country. When he does this, he marks the contrast between the question and the answer he is about to provide by saying but you know.

(9) MT = IR, Maarit Tastula EK = IE, Emir Kusturica

12 MT =°hh but by the way, was your (.) father 13 a patriarch of the family. 14 (.) 15 EK somehow yes. 16 (.) 17 MT in which way. → 18 EK °hh >but< you know, patriarchs here are very much (.) 19 u:h uh like uh uh th- they are- 20 it’s not like (on) the west you know if you are father 21 in the family:, 22 MT mm-h

The but in line 18 does not mark a contrast with the topical content of the other person’s talk. As a matter of fact the utterance that is being contrasted (in which way, line 17) does not have a topical content in itself, but rather invites the IE to elaborate on the topic in his answer. So in this example but marks a contrast with the activity that is done in the previous utterance, which is asking the IE to specify how his father was a patriarch. The IE does not do this, but singles out the word ‘patriarch’ as something that needs to be dealt with before answering the question. But contrasts with the follow-up question in a manner that enables the IE to continue his own answering turn. The contrast also marks resistance to the invitation to talk about personal experiences and a shift to an alternative teller role. After the contrastive “but” the IE produces the word patriarchs in first position. He resists implications that the term patriarch carries and shows that specific cultural knowledge is needed to answer the question in an appropriate way. The IE also makes visible a contrast between the west (where the IR is from) and the IE’s home country (line 20). At the same time he displays knowledge about how things are in the “west”. Thus he places himself in a position where he has enough knowledge about two different cultures to contrast them. The word here ties the talk to a specified body of knowledge – that of things about the IE's culture. The use of spatial adjuncts seems to be one way to contrast the question and the INVOKING DIFFERENT TYPES OF KNOWLEDGE IN CELEBRITY INTERVIEWS 109 answer and make relevant the different bodies of knowledge that the participants have. The contrast between the question and the answer can be voiced explicitly, as is done in example (10). After the IR’s question (lines 1–3) the IE does not produce an answer, but first makes a complaint about the question not being reasonable (it´s too big a question to ask-). He follows this complaint by stating the implications that such a question has in this context (…to answer so °hh u::h uh uh succinctly at all…). However, after making the complaint he moves on to answering the question and starts his answer with the contrasting but (line 7) and then answers in a way that displays general knowledge.

(10) MT = IR, Maarit Tastula WA = IE, Woody Allen

1 MT love is very important in your films. 2 what is the meaning of love (.) for you in your- 3 °hh uh in your movies, and in your (.) own life= → 4 WA =well. It- it’s too big a question to ask- → 5 [ to answer ] so °hh u::h uh uh succinctly at all 6 MT [°((laughs))°] → 7 WA °hh y-you know but the- the- the human interaction 8 °hh u:m th- between (.) a man and a woman,

Contrasts are sometimes used to mark the difference in knowledge types in questions and answers, but this is not always the case. Some of the examples in the following will show that it is possible to start displaying general knowledge in the answer without first explicitly orienting to the question as somehow problematic.

6.2 Displaying general knowledge

The differences of knowledge types in the question and in the answer can be seen in the lexical choices that the participants make. In example (11) the IE shifts away from a proposed participation role of teller of first-hand knowledge and starts to display more general knowledge. Like in the examples above the IE mobilizes a different type of knowledge than is asked for and, from line 5 onwards, does not talk about his personal experiences, but instead shifts to an “expert” role and displays general knowledge. The IE constructs expert knowledge through lexical choices, 110 HEIDI KOSKELA using vocabulary from the fields of psychology (self-destructive tendencies, line 7) and medicine (a disease of psychological dimensions, lines 13–14).

(11) MT = IR, Maarit Tastula SK = IE, Stephen Kuusisto

1 MT but it was actually very serious at some moment (.) 2 so you were um:: brought into hospital 3 SK ºyeah.º 4 MT [it was only] hundred and five po[unds] → 5 SK [u:hm ] [u:h ]we-we know from 6 SK psychological studies that the (.) you can 7 (0.6) uh (.) avoid (.) self-destructive tendencies (.) 8 is to: have a belief that life is possible, 9 that it´s worth living, that there´s meaning ahead (.) 10 u:h (.) you know (.) anorexic teenagers 11 the ones who stop eating (.) in fact believe (.) 12 that (.) the future will be terrible. 13 you know (.) anorexia is a disease of (.)psychological 14 dimensions >right< you don´t wanna become an adult. 15 because you think (.) boy that will be worse (.) 16 it´s bad now it´ll be worse at the next age (.) 17 >you know< I´m getting off the train right here (.) 18 you know I´m gonna stop (.)right here I´m not going on. 19 u:hm I think that´s what was going on with me,

Instead of telling his personal experiences about being brought into the hospital, the IE displays general knowledge about anorexia. The topical agenda remains the same, but a different type of knowledge is mobilized. The use of generic we (we know from psychologocial studies..., lines 6–7) implies that the knowledge the IE has is generalizable expert knowledge. Later on in the answer he does mobilize first-hand knowledge (I think that’s what was going on with me, line 19), but by constructing his answer as he does, the first-hand knowledge that is displayed later on in his answer is framed by general knowledge. This way he is able to demonstrate expertise that includes, but goes beyond, first-hand knowledge. In (12) it can be seen how the IR emphasizes the “personal” aspect of the question, and the IE can still choose to answer in a different framework of knowledge. Here the IE shifts away from the role of a teller of first-hand knowledge and adopts the role of a teller of ‘expert’ knowledge. He produces an answer that offers a fact, using general, abstract words (segregated, society, community), and is designed to be impersonal (when you live.., experience…is quite minimum). INVOKING DIFFERENT TYPES OF KNOWLEDGE IN CELEBRITY INTERVIEWS 111

(12) MT = IR, Maarit Tastula HB = IE, Harry Belafonte 1 MT and when did you have your first personal experience 2 with (.) racism. → 3 HB when you live within (.) the segregated society (.) 4 or in a segregated community 5 experience with race within the black community 6 is quite minimum. 7 (.) 8 HB (xxx) the restaurants are black, the schools are black, (.) 9 the minute you step outside that society 10 you will have your first experience with race.

By answering the question from a general point of view the IE shows how the presupposition that is included in the question (that the IE has had personal experiences with racism as a child) is not entirely appropriate. When he answers the question from a more general framework of knowledge the IE manages to correct the presupposition and let the IR and the television audience know that because the races were segregated, experiences with racism were not that common in everyday life, but something that were encountered only when going outside your own community. By subtly shifting to a different knowledge type the IE can answer the question (as he is supposed to do in an interview), not overtly disagree with the question, and still manage to point out what is ‘wrong’ in the question. In (13) the IE answers in such a way that displays general knowledge. A question about Woody Allen’s personal love life soon after his marriage to his ex-wife’s adopted daughter is a ‘loaded’ question and understandably one that might be resisted.

(13) MT = IR, Maarit Tastula WA = IE, Woody Allen

1 MT love is very important in your films. → 2 what is the meaning of love (.) for you in your- → 3 °hh uh in your movies, and in your (.) own life= → 4 WA =well. It- it’s too big a question to ask- 5 [ to answer ] so °hh u::h uh uh succinctly at all 6 MT [°((laughs))°] 7 WA °hh y-you know but the- the- the human interaction 8 °hh u:m th- between (.) a man and a woman, 9 between ((coughs)) a mother and a chi:ld, 10 or °hh brother and sister, brother and brother, 11 °hh u:m (.) you know, (.) is is o-one of the ways 12 that we have and (.) maybe our most effective way 112 HEIDI KOSKELA

13 °hh of (.) dealing with a very: (.) uh cold (.) 14 unforgiving (.) uh implacable: universe. 15 and uh (.) people (.) get a lot of comfort 16 ((swallows)) from their love relationships 17 °hh and so (.) uh they have many needs (.) 18 and it’s very complicated but 19 °hh it- uh >you know< it’s very very comforting 20 and makes life (.) a tiny bit more bearable.

The IE’s answer is not based on first-hand knowledge, but consists of generalizations such as human interaction (..) is one of the ways that we have..., and people get a lot of comfort. These types of generalizations claim knowledge of the “facts” that are stated. This is an example of what in courtrooms would not be acceptable talk from a lay witness, only from an expert witness (see Matoesian 1999). Making abstractions and generalizations requires knowledge that is organized and specialized and it is thus not seen as knowledge that a lay person would have. By displaying general knowledge instead of first-hand knowledge the IE manages to answer a ’loaded’ question in a manner that does not seem evasive. The answer follows the topical agenda set by the IR, but is designed to be very impersonal. This is achieved through the use of generic "man", and "woman" (line 8), and then moving on to categories such as “mother” and “child” (line 9) that make it explicit that the IE is not speaking about personal experiences, but his talk is to be understood on a more general level. The IE continues to use impersonal expressions throughout his answer and uses terms like “people", "we" and "they". Similar use of generic and impersonal expressions is employed in examples (11) and (12). In the examples that we have looked at the IE displays general knowledge by approaching the topic from a general level of knowledge. Besides lexical choices one way of moving to a more general framework of knowledge is the use of the present tense. In the following examples the use of the present tense is particularly visible, since it differs from the tense used in the IR's questions. The interviewer invokes first-hand knowledge and uses the past tense. In their answers the interviewees do not start telling about their experiences in the past, but instead shift to a more general framework of knowledge and accomplish this partly through the use of the present tense.

(14) MT = IR, Maarit Tastula INVOKING DIFFERENT TYPES OF KNOWLEDGE IN CELEBRITY INTERVIEWS 113

EK = IE, Emir Kusturica

12 MT =°hh but by the way, was your (.) father 13 a patriarch of the family. 14 (.) 15 EK somehow yes. 16 (.) 17 MT in which way. → 18 EK °hh >but< you know, patriarchs here are very much (.) 19 u:h uh like uh uh th- they are-

(15) MT = IR, Maarit Tastula HB = IE, Harry Belafonte

1 MT and when did you have your first personal experience 2 with (.) racism. → 3 HB when you live within (.) the segregated society (.) 4 or in a segregated community 5 experience with race within the black community 6 is quite minimum

(16) MT = IR, Maarit Tastula SK = IE, Stephen Kuusisto

4 MT [it was only] hundred and five po[unds] → 5 SK [u:hm ] [u:h ] we- we know 6 SK from psychological studies that the (.) 7 you can (0.6) uh (.) avoid (.)self-destructive tendencies (.) 8 is to: have a belief that life is possible, 9 that it´s worth living, that there´s meaning ahead (.) 10 u:h (.) you know (.) anorexic teenagers 11 the ones who stop eating (.) in fact believe (.) 12 that (.) the future will be terrible.

By selecting the present tense in their answers the IEs manage to shift the focus of talk from lived personal experiences to things that are generalizable. The IEs can talk about generalizable things as matters of fact and as something they have "expert" knowledge of. A similar practice of claiming greater certainty of knowledge in a response to a question that solicits the IE’s personal view has been identified in American news interviews (see Roth 2002: 372).3 Matoesian (1999: 491) has found that lay witnesses can testify only to facts that they have first-hand knowledge of, while expert witnesses can

3 This shows how participants who are using English as an international language in television interviews employ similar practices to those used by native speakers in television interviews. 114 HEIDI KOSKELA give opinions and explanations about facts on a more general level, based on their training, qualifications, skill, experience and knowledge. The link between first-hand knowledge and lay participants is also presented by Hutchby (2001a), who shows how ‘lay’ participants legitimate their opinions through claiming first-hand knowledge.

7. Conclusion

In the data I have analysed the interviewers can invoke first-hand knowledge through topic-selection or explicit voicing of the 'personal' viewpoint in the question. In their answers there are several resources the interviewees can use to resist the agenda in the IR’s questions and to mobilize a more general body of knowledge. The IEs can explicitly orient to the aspect of the questions that is in their eyes somehow inappropriate. To do this they use contrasts to mark the shift to a different type of knowledge. The contrastive devices used are the contrastive “but”, and other lexical elements that contrast with the previous talk (e.g. the spatial adjunct “here”). After showing the problematic aspect of the question they then change the topical content of the talk to a more general level. The IEs can also start answering the question from a more general framework of knowledge without producing contrasts or explicitly orienting to the previous question as problematic. When displaying general knowledge, the lexical choices that the IEs make in their answer are in line with the more general topic. The IEs can also use a shift to the present tense to mark the mobilization of general knowledge. In some cases the IR has used the past tense, which is the logical tense to use when asking somebody about things that they have experienced, but in his answer the interviewee uses the present tense, which is typically used when describing general facts/state of affairs. What functions might mobilizing different types of knowledge serve in television interviews? The first of the functions here involves the management of the intimacy-level of the topics. In these types of television programs the invoking of first-hand knowledge might be an attempt by the interviewer to make the interaction seem more intimate and to reveal aspects of the interviewees' private personae to the television viewers. The IEs clearly recognize (and demonstrate their understanding of) the IR's attempt to invoke first-hand knowledge. The use of the contrasting devices is an indication of this. However, in answering the questions the IEs do INVOKING DIFFERENT TYPES OF KNOWLEDGE IN CELEBRITY INTERVIEWS 115 something different instead. The interviewees' resistance to take on the role of a teller of first-hand knowledge while mobilizing a different type of knowledge serves to avoid topics that are too intimate, or topics in which some presupposition needs to be dealt with before answering, and to shift the topic away from a personal to a more general level. The resistance is done very subtly. Many elements of the topical agenda set by the interviewer remain the same – only a different type of knowledge is mobilized. This enables the interviewees to manage a shift in the agenda in a way that is not made accountable (e.g. the IR does not repeat the question) and also the IEs manage to avoid seeming evasive to the television viewers. The second function of this type of action is the construction of ‘expert’ knowledge. The IEs present themselves as experts of some field, as people who have a specified body of knowledge that is not based only on first-hand knowledge. Mobilizing a more general type of knowledge is a resource to display 'expert' knowledge of a specified field.

Appendix: Transcription conventions underlined talk emphasis CAPITALS increased volume ºhigh circlesº decreased volume ta:::lk prolongation of the preceding sound tal- cut-off word ºhhh inbreath hh outbreath (.) a micropause of less than 0.4 seconds (0.8) a pause, timed in tenths of a second ta[lk] [tal]king overlapping utterances talk= =talk latching utterances (talk) uncertain transcription (x) unintelligible item, probably one word only (xx) unintelligible items, approximately of phrase length (xxx) unintelligible items, beyond phrase length , continuing intonation . falling intonation ? rising intonation 116 HEIDI KOSKELA

↑ high pitch >fast< fast speech slow speech ₤ altered tone of voice, e.g. when quoting somebody ta(h)lk breathiness, e.g. in laughter

References

Berg, Maarit (2001) Vallan vahtikoira haukkuu mutta ei pure. Poliitikkoa syyllistävät toimittajan kysymykset televisiokeskustelussa. In Johanna Ruusuvuori, Markku Haakana & Liisa Raevaara (eds.), Institutionaalinen vuorovaikutus. Keskustelunanalyyttisia tutkimuksia. [Institutional interaction. Studies in conversation analysis], pp. 121–137. Helsinki: Suomalaisen Kirjallisuuden Seura [Finnish Literature Society]. —— (2003) Syytöksiä ja epäilyksiä. Toimittajan ja poliitikon vuorovaikutuksesta televisiokeskustelussa. [Accusations and suspicions. On interaction between journalists and politicians in televised conversation]. PhD Dissertation. Helsinki: Suomalaisen Kirjallisuuden Seura [Finnish Literature Society].. Clayman, Steven (1988) Displaying neutrality in television news interviews. Social Problems 35: 474–492. —— (1992) Footing in the achievement of neutrality: The case of news interview discourse. In Paul Drew & John Heritage (eds.), Talk at work, pp. 163–198. Cambridge: Cambridge University Press. —— (1993) Reformulating the question: A device for answering/not answering questions in news interviews and press conference. Text 13: 159–188. —— (2001) Answers and evasions. 30: 159–188. Clayman, Steven and Heritage, John (2002) The News Interview. Journalists and Public Figures on the Air. Cambridge University Press. —— (2002) Questioning Presidents: Journalistic Deference and Adversarialness in the Press Conferences of Eisenhower and Reagan. Journal of Communication 52(4): 749–775. Greatbatch, David (1986) Aspects of topical organization in news interviews: the use of agenda-shifting procedures by interviewees. Media, Culture & Society 8: 441– 455. —— (1988) A turn-taking system for British news interviews. Language in Society 17(3): 401–430. Heritage, John (1985) Analyzing news interviews: Aspects of the production of talk for an overhearing audience. In Teun A. van Dijk (ed.), Handbook of discourse analysis, vol. 3, pp. 95–119. New York: Academic. —— (2002) The Limits of Questioning: Negative Interrogatives and Hostile Question Content. Journal of 34: 1427–1446 —— (2003) Designing Questions and Setting Agendas in the News Interview. In Phillip Glenn, Curtis D. LeBaron & Jenny Mandlebaum (eds.), Studies in Language and INVOKING DIFFERENT TYPES OF KNOWLEDGE IN CELEBRITY INTERVIEWS 117

Social Interaction: In Honor of Robert Hopper, pp. 57–90. London: Lawrence Erlbaum. Heritage, John & Greatbatch, David (1991) On the institutional character of institutional talk: the case of news interviews. In Deirde Boden and Don H. Zimmerman (eds.), Talk and Social Structure: Studies in Ethnomethodology and Conversation Analysis. pp. 93–137. Oxford: Polity Press. Heritage, John & Roth, Andrew (1995) Grammar and institution: questions and questioning in the broadcast news interviews. Research on Language and Social Interaction 28(1): 1–60. Heritage, John & Raymond, Geoffrey (2005). The terms of agreement: indexing epistemic authority and subordination in talk-in-interaction. Social Psychology Quarterly 68(1): 15–38. Hutchby, Ian (2001a) `Witnessing`: the use of first-hand knowledge in legitimating lay opinions on talk radio. Discourse Studies 3(4): 481–497 —— (2001b) Confrontation as a spectacle: The argumentative framework of the Ricki Lake show. In Andrew Tolson (ed.), Television Talk Shows: Discourse, Performance, Spectacle. pp. 155–172. New Jersey: Lawrence Erlbaum Kajanne, Milla (2001a) Kansalaiset kysyjinä. Yleisön kysyminen osana vuorovaikutusta television EU-keskusteluissa. [Participating citizens. Audience questioning as a form of interaction in Finnish television discussions concerning the European Union.] PhD dissertation. Helsinki: Suomalaisen Kirjallisuuden Seura [Finnish Literature Society]. —— (2001b) Maallikot televisiokeskustelun osallistujina. In Johanna Ruusuvuori, Markku Haakana & Liisa Raevaara (toim.), Institutionaalinen vuorovaikutus. Keskustelunanalyyttisia tutkimuksia. [Institutional interaction. Studies in conversation analysis], pp. 138–158. Helsinki: Suomalaisen Kirjallisuuden Seura [Finnish Literature Society]. Matoesian, Greg (1999) The of participant roles in the constitution of expert identity. Language in Society 28: 491–521 Nuolijärvi, Pirkko & Tiittula, Liisa (2000) Televisiokeskustelun näyttämöllä. Televisioinstitutionaalisuus suomalaisessa ja saksalaisessa keskustelukulttuurissa [TV-discussion as a performance. Television institutionality in Finnish and German discourse cultures]. Helsinki: Suomalaisen Kirjallisuuden Seura [Finnish Literature Society]. Piirainen-Marsh, Arja & Koskela, Heidi (2000) Collective participation as a resource in multiparty multicultural broadcast interactions. Issues in 11(2): 243–272. Pomerantz, Anita (1980) Telling my side: “Limited access” as a “Fishing device”. Sociological Inquiry, 50: 186–198. Roth, Andrew (1998) Who makes the news? Descriptions of television news interviewees´ public personae. Media, Culture & Society 20: 79–107. —— (2002) Social epistemology in broadcast news interview. Language in society 31(3): 355–381. 118 HEIDI KOSKELA

Schegloff, Emanuel A. (1972) Notes on a conversational practice: formulating place. In David Sudnow (ed.) Studies in Social Interaction, pp. 75–119. New York: Free Press. Sharrock, Wes (1974) On owning knowledge. In R. Turner (ed.), Ethnomethodology. pp. 45–53. London: Penguin.

Contact information:

Heidi Koskela University of Jyväskylä Department of Languages / English P.O. Box 35 40014 University of Jyväskylä FINLAND e-mail: [email protected] Donna L. Lillian

Homophobic Discourse: A ‘Popular’ Canadian Example

Abstract

This paper presents a critical discourse analysis of the writings of Canadian neo- conservative author, William D. Gairdner, focusing specifically on his discourse about homosexuals and homosexuality. I first examine Gairdner’s use of lexical choices, metaphor, and transitivity. Next, I quote a short sample passage from his writings and discuss the various linguistic features Gairdner employs in presenting his views about homosexuality. Following that linguistic analysis, I discuss neo-conservatism in Canada and Gairdner’s role in that movement, and then I contextualize my study within other social and linguistic research aimed at identifying and resisting homophobia and homophobic . Finally, I conclude by illustrating the ways in which Gairdner’s discourse can be understood as part of the broader neo-conservative agenda with respect to gender, race, and sexual orientation. This paper fills a void in the published literature by focusing on a specifically Canadian example of homophobic discourse.

1. Introduction

This paper uses the techniques of critical discourse analysis to identify and explicate some of the linguistic devices prevalent in neo-conservative discussions of homosexuality and homosexuals.1 Specifically, it focuses on the writings of William D. Gairdner, a contemporary Canadian writer of political and social commentary. Gairdner’s works include numerous essays and five popular (i.e. non-academic) books: The Trouble With Canada: A Citizen Speaks Out (1990), The War Against the Family: A Parent Speaks Out (1992), Constitutional Crackup (1994), On Higher

1 This paper is based on portions of my doctoral dissertation (Lillian 2001), under the direction of Drs. Susan Ehrlich, Sheila Embleton, and Ruth King. I would like to thank them for their extensive help and guidance as I wrote the dissertation. I would also like to thank Drs. Donna Kain and Roberta Martin, whose careful reading and considerable insight guided me as I prepared this paper.

SKY Journal of Linguistics 18 (2005), 119–144 120 DONNA L. LILLIAN

Ground: Reclaiming a Civil Society (1996), and The Trouble With Democracy: A Citizen Speaks Out (2001). With the exception of Constitutional Crackup, which deals exclusively with the issue of the French separatist movement in Canada, all of Gairdner’s popular books contain at least some discussion of homosexuality. Gairdner presents his longest and most explicit discussion of homosexuality in The War Against the Family, which includes two consecutive chapters on this theme. Those two chapters constitute the core data analyzed in this paper to illustrate the discursive strategies Gairdner employs in vilifying homosexuals and homosexuality. This paper derives its theoretical and methodological framework from critical discourse analysis. Critical discourse analysis (CDA) focuses on the social, political, and cultural dimensions of discourse, not simply on its structure (van Dijk 1990: 8). It is a diverse, multi-disciplinary approach to the study of texts, not a rigidly or narrowly defined theory or discipline. While CDA borrows many of the insights and tools of other branches of formal linguistics and discourse analysis, it differs from them in being critical, that is, in explicitly engaging in an analysis of the structures of power and dominance at work in the context of particular texts (van Dijk 1994). One of the major fields contributing to CDA is critical linguistics.

Critical linguistics seeks, by studying the minute details of linguistic structure in the light of the social and historical situation of the text, to display to consciousness the patterns of belief and value which are encoded in the language – and which are below the threshold of notice for anyone who accepts the discourse as ‘natural’ (Fowler 1991: 67).

Using the techniques of critical discourse analysis, I analyze some of the linguistic devices Gairdner uses in his presentation of homosexuals and homosexuality. First, I examine Gairdner’s use of lexical choices, metaphor, and transitivity. Next, I quote a short sample passage from his writings and discuss the various linguistic features Gairdner employs in presenting his views about homosexuality. Following that linguistic analysis, I discuss neo-conservatism in Canada and Gairdner’s role in that movement, and then I contextualize my study within other social and linguistic research aimed at identifying and resisting homophobia and homophobic discourses. Finally, I conclude by illustrating the ways in which Gairdner’s discourse can be understood as part of the broader neo- conservative agenda with respect to gender, race, and sexual orientation. HOMOPHOBIC DISCOURSE: A ‘POPULAR’ CANADIAN EXAMPLE 121

An effort was made in preparing this paper to balance the negative characterization of Gairdner’s writing as homophobic with evidence that might tend to neutralize this characterization; however, no mitigating evidence was found, either within the two chapters I focused on or elsewhere in Gairdner’s texts. While my analysis strives to be academically rigorous and intellectually valid, in keeping with the critical agenda of critical discourse analysis, it does not purport to be value-neutral. This study has been undertaken not in order to foreground the discourse of William Gairdner that others might admire it, but rather, to expose it as being discriminatory and dangerous. This paper differs from most published critical discourse analyses, which focus on any number of text types including newspaper articles, news broadcasts, interviews, political , advertising, courtroom transcripts, among others, but which rarely focus on texts of the specific non-fiction genre of the books under examination in this paper. It also seeks to fill a void in the published literature by focusing on a specifically Canadian example of homophobic discourse.2

2. Lexical Choices

Among the most striking features of Gairdner’s lexical choices is the way he attacks homosexuals, particularly gay men, by attempting to define them out of existence. He employs a strategy which he dubs “the verbal substitution phenomenon” (Gairdner 1992: 450), while simultaneously criticizing his opponents for purportedly using that very same strategy. He explains the verbal substitution phenomenon in some detail:

One of the best ways to control a political or moral agenda, or simply a debate over cocktails, is to put the opposition on the defensive by being the first to define their behaviour negatively. This is achieved most effectively by forcibly substituting a negative word for an ordinary one; the word intolerant for the word disapprove, for example. Soon, if repeated often enough, the negative word becomes a cover term for any of the ordinary words previously used. Rendered speechless by this negative contamination of their vocabulary, ordinary people react by backing down. They are then effectively silenced, terrorized, and neutralized by the group with the agenda. (Gairdner 1992: 45)

2 A search on March 27, 2005 of the Linguistics and Language Behaviors Abstracts, the Modern Languages Association International Bibliography, and the Science Direct database turned up just one reference to homophobic or heterosexist discourse or language or speech in Canada. This reference was Jacobs (1995). 122 DONNA L. LILLIAN

This strategy that he claims is used by his left-wing opponents is exactly the technique that Gairdner himself uses and advocates with respect to terms such as sexual orientation, gay, and homophobic, discussed below. Since Gairdner has a graduate degree in linguistics, one must assume that he is aware that words acquire meaning not by fiat, but by usage. Thus, his attempt to dismiss homosexuality by unilaterally imposing meanings on terms associated with it is not merely ironic, but suggests a certain degree of manipulation on his part.

2.1 ‘Sexual orientation’

Gairdner signals his rejection of the concept of legitimate variations in human sexuality in part through his rejection of the term ‘sexual orientation’. In The War Against the Family, he attempts to justify this position.

Perhaps one of the most insidious victories of the homosexual movement can be seen in the broad acceptance of the term “sexual orientation”, now used widely in our schools and in sociological and government documents. It was coined to replace the phrase “sexual preference”, which suggested that homosexuals were choosing their behaviour and therefore were responsible for it -- an implication most of them strive to avoid. They needed to find a word which suggested that just as the magnetic forces of earth pull the compass needle to North, something called “sexual orientation” directs homosexuals to indulge in the behaviours they enjoy as if they were a natural fact of life . . . The public would do well to reject this word . . . and to talk about “sexual choice” instead of “orientation”. And everywhere they see the former phrase, they should insist on the use of the latter, both privately and publicly. (Gairdner 1992: 378-379)

‘OUT!SPOKEN’, a Toronto-based action and advocacy group working to end media discrimination against lesbians, bisexuals, and gays produced a style guide in the early 1990’s for the print and broadcast media (OUT!SPOKEN n.d.: 1). This guide concurs with Gairdner that the term sexual orientation is the term most often used in anti-discrimination legislation to refer to an individual’s primary gender orientation (OUT!SPOKEN n.d.: 14), but not surprisingly it does not identify the term as problematic. In fact, whereas the OUT!SPOKEN guide generally identifies as controversial terms such as queer, fag, or dyke, which are not viewed or used by all gays and lesbians with equal degrees of comfort, it presents the term sexual orientation as unproblematic and uncontroversial (see also Jacobs 1995 & 1998). Thus, Gairdner appears to be attaching HOMOPHOBIC DISCOURSE: A ‘POPULAR’ CANADIAN EXAMPLE 123 controversy to a term which is not otherwise a site of contestation in discussion of gay and lesbian rights. Contesting otherwise uncontested terms is thus a strategy that Gairdner employs in order to deflect attention away from the serious issues of discrimination faced by gays and lesbians and onto the largely moot issue of whether being gay or lesbian is an inherent or a derived state.

2.2 ‘Gay’

Sexual orientation is not the only term Gairdner rejects; he also rejects the term gay as used by and about homosexual persons. The OUT!SPOKEN Styleguide offers the following explanation of the term gay.

The term that homosexual men and some women have used to describe themselves, and the currently preferred term to describe men whose primary romantic and physical attraction is to members of their own gender . . . Most gay people prefer ‘gay’ to the clinical word ‘homosexual’. The term ‘gay’ may be used to refer to both men and women, but ‘lesbian’ is the preferred term for women. (OUT!SPOKEN n.d.: 12)

Challenging such neutral or positive uses of terms such as gay and sexual orientation, Gairdner asserts that

by altering the word-labels that describe their world, homosexuals hope to make their behaviour invisible, or at least so “normal” it won’t be noticed. This is part of their campaign to morally desensitize the public. (Gairdner 1992: 376)

This explanation imputes sinister, or at least objectionable, motivations to gays and lesbians who use these terms, an idea on which Gairdner elaborates in the following passage:

Public acceptance of the word “gay” has been a great victory for homosexuals, for the word “homosexual” had always been a negative term to denote those who preferred to copulate with their own sex. But what homosexuals wanted was a word that elevated their behaviour to an admirable status, and they achieved this by taking a perfectly good English word -- now off-limits to normal people -- and appropriating it for their specific use. (Gairdner 1992: 377)

In both of the two preceding quotations, Gairdner appropriates the voice of homosexuals, purporting to know and understand their motives and their goals. Furthermore, in both of these quotations, he isolates homosexuals 124 DONNA L. LILLIAN from the mainstream, in the first case by contrasting them with “the public”, whom they purportedly wish to desensitize, and in the second case by contrasting them with “normal people”, who purportedly have been robbed of a “perfectly good English word”. The choice of someone, particularly someone outside the gay and lesbian community, to use gay instead of homosexual to refer to gay men does, as Gairdner recognizes, have a significance, although not necessarily the significance he assigns it when he claims,

To them, the word “gay” suggests carefree, happy-go-lucky individuals who have a culture and a way of being all their own, which may in some secret way be superior to the hum-drum, sometimes ungay world of the rest of humanity. The word suggests a special in-group with access to happiness the rest of us lack. (Gairdner 1992: 377)

The word does not, in fact, necessarily connote cheerfulness or good humour on the part of gays, but it does suggest a more positive and accepting attitude towards members of the group than does homosexual in many contexts, which is undoubtedly why Gairdner admonishes his readers not to use it:

Those who wish to defend the natural family and the core values of our society should strenuously resist the use of this word, and publicly replace it with the word “homosexual”, which is more honest. After all, most of these people are the furthest thing from “gay”. (Gairdner 1992: 377)

The contestation of the word gay both within and outside the homosexual community is not unlike the evolution of terms to refer to descendants of African slaves in North America, particularly in the U.S.A. Just as gay was at first primarily in-group usage by homosexual men before it became widely recognized and used outside the homosexual community, so was African American first used and discussed within the black community before it came to the attention of most Americans, and just as gay was not at first uniformly positively received (Ashley 1979: 226), neither was African American (Baugh 1991: 135). However, by the 1970’s, gay was becoming increasingly popular as a term for homosexual men, by the 1980’s it had began to appear regularly in the mainstream press, and by the early 1990’s, the term gay was no longer even a source of debate within the homosexual community (Jacobs 1998: 195). The fact that Gairdner expends such effort over gay demonstrates that naming and the right to self-identification is a site of considerable ideological struggle, and is HOMOPHOBIC DISCOURSE: A ‘POPULAR’ CANADIAN EXAMPLE 125 contested by representatives of the dominant heterosexist patriarchy such as Gairdner.

2.3 ‘Homophobic’

In much the same way as Gairdner rejects the terms gay and sexual orientation, he also rejects homophobic.

Once they have defined themselves anew by rejecting society’s generic “homosexual” label and established the “dignity” of being “gay”, the next choice was to position normal people as frightened, bigoted, and irrational. This was achieved by inventing the nonsensical but extremely effective word “homophobic” and using it like a grenade to throw at anyone who dared to utter the slightest negative opinion. (Gairdner 1992: 377)

Gairdner attempts to argue away the reality of homophobia by rejecting the term itself. By suggesting that homosexuals invented the term, he is implying that they invented the concept that the term represents. In his statement “[m]ost people, of course, are not afraid of homosexuals at all. They’re disgusted by them” (1992: 377), Gairdner reveals that his rejection of the term homophobic is based on the assumption that it means something like “fear of homosexuals”. However, the root phobe, from which phobic is formed, is defined by The Concise Oxford Dictionary (1982) as “not fond of, (person) disliking or fearing”. Furthermore, in common usage, describing someone as homophobic is asserting that they hate, despise, and/or fear homosexuals and homosexuality. Finally, OUT!SPOKEN (n.d.: 13) defines homophobia as, “[f]ear of or aversion to homosexuals or homosexuality, including one’s own homosexual feelings, and the hatred, disgust, and prejudice that fear brings”. Gairdner cannot, by fiat, make the word stop meaning what it does mean, and he cannot erase the reality of homophobia with glib remarks such as the following. “After all, the truth is that it is not we who are homophobic, but rather homosexuals who are ‘heterophobic,’ ‘moralphobic’, and ‘familyphobic’” (Gairdner 1992: 378). In his discussion of homosexuality, Gairdner challenges the use of the terms gay, sexual orientation, and homophobic. In fact, he attempts to disallow such terms and urges his readers to refuse to use them and to try to force others to abandon them as well (Gairdner 1992: 378). Gairdner recognizes that language is a site of ideological struggle, although he appears not to acknowledge that he is engaging in just the sort of linguistic manipulation he accuses his opponents of using. By trying to force his 126 DONNA L. LILLIAN meanings onto particular lexical items, to the exclusion of all other meanings, he is manipulating not just the language, but through the language the thoughts and emotions of the readers, attempting to generate in them the same sorts of homophobic attitudes that his writing expresses.

3. Metaphor: Disease/pathology

A prevalent metaphor in Gairdner’s writings on homosexuality is that of disease and pathology. In a chapter entitled “The Gay Plague and the Politics of AIDS”, Gairdner (1992) selectively interprets data about AIDS and other (potentially) sexually transmitted diseases, presenting them separately and collectively as “the gay plague”. The word plague occurs four times in the text of this chapter, but it occurs an additional nine times if one counts the chapter title reprinted in the header of each odd-numbered page. Thus, the word and therefore the concept of plague are never far from the consciousness of the reader. In addition to the word plague, the chapter is saturated with terms denoting diseases and illnesses, among them cancer, amoebiasis, giardiasis, salmonellosis, shigellosis, hepatitis, tuberculosis, syphilis, AIDS, HIV, gonorrhea, and others. Moreover, terms denoting illness and disease are repeatedly collocated throughout the chapter with the terms gay and homosexual, including such phrases as “the homosexual disease epidemic” (1992: 397). The most explicit statements in the chapter are those that blame gay men for AIDS. For example, Gairdner writes that

homosexuality is an antifamily phenomenon, and AIDS in North America is still overwhelmingly a homosexual disease. Encouraging more of one will bring more of the other (Gairdner 1992: 401).

Gairdner contends that AIDS is primarily a gay disease spread through anal intercourse, but he also argues strenuously that it can be spread through non-sexual, casual contact (Gairdner 1992: 406-407). Rather than drawing the conclusion that since the virus can be spread through means other than gay sex, it is therefore not a gay plague, Gairdner uses the possibility of it spreading in other ways to further condemn homosexuals, claiming that they knowingly and callously pose a direct health risk to heterosexuals, and especially to children in daycare centres, all in the interests of promoting their alleged political agenda (Gairdner 1992: 404). HOMOPHOBIC DISCOURSE: A ‘POPULAR’ CANADIAN EXAMPLE 127

Not content to collocate homosexuals with physical diseases and pathologies, Gairdner also attempts to link homosexuality with the spread of social pathologies, particularly in daycare centres (Gairdner 1992: 368). Gairdner quotes psychologist Elizabeth Moberly (1983), who theorizes that homosexuality is a search for parenting that was absent or inadequate during a person’s infancy. Since Gairdner sees it as primarily a mother’s responsibility to provide the crucial nurturing to children, he exploits Moberly’s view as another opportunity to blame feminists for what he perceives to be the destruction of human society.

If Moberly is correct, then from the egalitarian/feminist drive for tax-funded universal daycare we can expect to see a horrendous unintended consequence: tax- funded centres for sexual pathology. (Gairdner 1992: 368)

What Gairdner means here by sexual pathology is homosexuality. He is echoing the disease-and-death associations he made previously when he links homosexuality with such illnesses as cancer and tuberculosis. He is also drawing a conclusion about a purported causal link between daycare centres and homosexuality, between feminism and homosexuality, and between egalitarianism and homosexuality, none of which causal links have ever been demonstrated by him or by anyone else. If he cannot frighten readers sufficiently by implying that homosexuals will infect children with physical diseases, then he will attempt to frighten them by suggesting that daycare workers will socially condition children to become homosexuals. As a final attack against homosexuals and on the excuse of doing a public service, Gairdner presents nine pages of graphic description of activities which gay men purportedly engage in, many of them violent. Although the text of Gairdner’s description of alleged homosexual behaviours could be described at length and in detail on several different levels, the following discussion of Gairdner’s use of transitivity is more than adequate to reveal the nature of his attribution of violence and disease to gay male sexual activity.

128 DONNA L. LILLIAN

4. Transitivity

The following discussion of transitivity3 draws heavily on the insights of Fowler (1985), Simpson (1993), and Sykes (1985), all of whom employ Halliday’s notion of transitivity as something much broader than the traditional notion that a verb can be either transitive (having a direct object complement) or intransitive (having no direct object complement). Rather, within critical discourse analysis, transitivity is a relative concept which is applied to a clause rather than just to a verb. Thus, rather than a clause simply being either transitive or intransitive, it can be judged as being highly transitive, somewhat transitive, relatively low in transitivity, with infinite gradations in between. Clauses that are deemed to be very highly transitive have three main characteristics: an animate subject, who deliberately performs an action, which affects a patient. The transitivity of the clause diminishes as one alters, obscures, or removes any of these three criteria. Simpson (1993) formalizes the transitivity scale by characterizing it in terms of four types of processes, listed from highest to lowest levels of transitivity: material processes, verbalization processes, mental processes, and relational processes. According to Simpson, material processes are processes of doing, typically involving an ACTOR carrying out an observable act, often one that has consequences that are also observable. Verbalization processes, or processes of saying, are observable, but they involve a SAYER rather than an ACTOR and they are thus less highly transitive than are material processes. Mental processes are internalized processes of sensing, involving perception, reaction, or cognition. As such, they are less transitive than either of the first two types of process, which are at least observable. Finally, relational processes are processes of being and having (Simpson 1993: 90-92). By applying Simpson’s taxonomy of processes to one chapter of Gairdner’s writings on gay men (Gairdner 1992, Chapter 13: “Radical Homosexuals vs. the Family”), one can demonstrate how Gairdner uses transitivity as one more tool with which to discredit homosexuals. Specifically, the images of homosexuals presented by Gairdner are chosen to maximize the image of homosexuality as violent and unhealthy. Gays are often placed as agents of transitive clauses and especially as agents of

3 This section on transitivity includes material adapted from Lillian (1997). Copyright rests with the author. HOMOPHOBIC DISCOURSE: A ‘POPULAR’ CANADIAN EXAMPLE 129 violent acts. Taken together, this paints a picture of gay men as actively carrying out material processes, especially those resulting in real or perceived harm or injury. In the chapter under analysis, there are 47 predicates that represent the category of material processes with gay men in the role of ACTOR. Eight of these represent metaphorical rather than literal actions: “have broken the implicit pact”; “are intentionally attacking this pact” (1992: 357); “to destroy the natural family” (1992: 361); “taking a perfectly good English word”; “like a grenade to throw at” (1992: 377); “to destroy all moral hierarchy”, “can escape criticism”; “have imported the notion” (1992: 379). The remaining material processes represent literal rather than figurative activities: “homosexuals commit suicide” (1992: 375); “indulge in the behaviours” (1992: 379); “perpetrate child molestations”, “homosexual teachers commit pupil molestations”, “likely to molest”, “assaulted”, “killed”, “committed”, “raped, killed, then raped them dead”, “ate”, “use whips”, “tie up”, “beat”, “punish”, “use”, “are publicly beaten and ‘sold’”, “hurt”, “scratched”, “bruised”, “bloodied”, “hurt each other”, “hurt themselves” (1992: 388-389); “seldom stop to disinfect”, “ingest”, “licking”, “inserting”, “eating”, “rubbing”, “ingest”, “urinating on”, “defecating on”, “drank”, “urinated on”, “ate”, “received sperm” (1992: 394-395). With the possible exception of the processes “escape” and “import”, all these material processes, whether they are used literally or figuratively, convey violent or at the very least unhealthy (e.g. “ingest fecal material”) acts. This chapter paints a picture of homosexual men as violent and dangerous, a picture which is not altered by the following disclaimer near the beginning of chapter 13.

So this chapter is not about homosexual individuals who are minding their own business. Nor is it intended to hurt the feelings of otherwise proper homosexual citizens, many of whom are themselves extremely distraught by homosexual behaviour and the radical agenda. (Gairdner 1992: 357)

The import of this disclaimer is lost completely since, throughout the chapter, Gairdner attributes actions simply to “homosexuals”, effectively painting all homosexuals with the same negative brush. Of the verbal processes attributed to homosexuals, several connote violence and aggression, as the following examples demonstrate: “chanting”, “threatening” (1992: 369); “‘homophobic’ . . . using it” [i.e. using the word] (1992: 377); “to justify it” (1992: 379); “promoting”, “declared war” (1992: 387); “admit to” (1992: 389). The mental processes 130 DONNA L. LILLIAN attributed to homosexuals are closely linked to the verbalization processes and tend to concern homosexuals’ purported efforts to justify themselves and their actions and to convince other people to accept their views and actions. Thus we get: “homosexuals want [to destroy]” (1992: 361); “to alter the way the public thinks”, “controlling”, “wanted . . . a word”, “appropriating it”, “defined”, “rejecting”, “to position”, “inventing” (1992: 377); “needed to find a word”, “enjoy”, “attempted [to justify]” (1992: 379); “choose” (1992: 387); “enjoy” (1992: 389). The only relational process assigned to homosexuals is “they are out in the open” (1992: 387), but even that is linked to the verbalization processes “promoting their ‘life- style’” and “declared war” (1992: 387). To summarize, in the chapter examined, homosexuals are associated 47 times with material processes, many of which represent acts of violence, 7 times with verbalization processes, 14 times with mental processes, and only once with relational processes. Gays are most frequently assigned roles in material processes which are high in transitivity and which often connote violence or aggression. The effect of this is that homosexuality is portrayed as an imminent, physical threat. Gairdner attempts to generate disdain toward all with whom he disagrees, but he is evidently out to generate in his readers a special fear and loathing of gay men. Damning gays (since he largely ignores lesbians in this discussion) as perverse, and attempting to generate feelings of revulsion among readers by his unfair and prejudicial association of gays with pedophilia, bestiality, and violence, Gairdner works to generate panic and fear among readers by portraying gays as dangerous not just to themselves, but to the health and well-being of the public at large.

5. Sample Analysis

This section presents a sample analysis of one excerpt from Gairdner (1992), demonstrating how a critical linguistic analysis can be applied to a text. This passage by no means represents the most extreme example of Gairdner’s homophobic discourse. In fact, it represents some of Gairdner’s least inflammatory anti-gay writing. It is often the more rational-sounding and seemingly ‘neutral’ discourse which is the more insidious, precisely because readers may be less likely to realize that through reasonable- sounding prose they are being led toward a radical position. The passage is taken from the middle of a 40-page chapter entitled “Radical Homosexuals vs. the Family” and constitutes one sub-section of the chapter. Prior to this HOMOPHOBIC DISCOURSE: A ‘POPULAR’ CANADIAN EXAMPLE 131 passage, Gairdner discusses homosexuality under the sub-sections: Born Homosexual?, Made Homosexual?, and Homosexual by Choice? (italics in original). In this passage, Gairdner distils from the cursory and brief presentation he has made of various theories of the origins of homosexuality his own view of the matter.

The Author’s Model My view is that any human population could be made homosexual under certain extreme conditions. For example, a group of young children with no prior sexual knowledge, raised on a desert island by a homosexual, would likely end up fancying homosexual behaviour themselves. In other words, it is quite possible to train a young population to like homosexual behaviour. What prevent this from happening on a large scale is the social sanctions against it. Societies intentionally suppress homosexual behaviour so that such extreme situations cannot arise. Secondly, there may be very small numbers of people with a predisposition, however caused, to choose homosexual behaviour. Notice that I used all the models here. Some may be predisposed, in the way an alcoholic may be predisposed to drink if drink is available. But nothing except personal desire makes the bottle (or the penis) jump into the mouth. After all, all societies have people with predispositions to all sorts of weird appetites and behaviours, whether homosexuality, suicide, drunkenness, killing, stealing, overeating, or bad temper. This has always been so, and it is not going to go away. We are only going to exhaust ourselves trying to figure out why, and we likely will always wind up in a conflict of expertise in which biased experts with warring political agendas dispute each other’s findings for eternity. It’s a dead end. (Gairdner 1992: 369- 370)

The passage opens with the noun phrase “my view” as the subject of the main clause, with the content of his view itself being relegated to a subordinate clause. The proposition of the subordinate clause, which constitutes the substance of Gairdner’s view, is that, “any human population could be made homosexual under certain extreme conditions”. On its own, this clause conveys a modality of validity, reflecting a high degree of certainty and confidence on the part of the writer. “Any population” defines the broadest possible range of humanity, potentially including not just ‘other’ people, but also the reader and even the writer himself. Gairdner thus presents homosexuality as something which his readers cannot afford to ignore, since, according to him, they themselves could be vulnerable to being made homosexual “under certain extreme conditions”. In one sense, the qualifying phrase “under certain extreme conditions” mitigates the modality slightly, but later in the chapter and in the following chapters, Gairdner argues that daycare centres constitute the 132 DONNA L. LILLIAN sorts of “extreme conditions” that generate homosexuality, so in practice, the qualifier does not significantly reduce the modal strength of the assertion Gairdner is making. On its own, Gairdner’s proposition makes a very strong claim, one for which he lacks supporting empirical evidence. Thus, he utilizes the syntax, placing his proposition in a subordinate clause, together with the personalized pronoun my in the main clause to alter the modality of the opening sentence of the passage such that he is not making an empirical claim, but rather is expressing his proposition in the form of an opinion. In the second sentence of the passage, Gairdner gives an example to illustrate the claim he makes in the first sentence. He then presents a scenario which is so unlikely as to be absurd: a group of young children raised on a desert island by a homosexual. Nevertheless, as absurd as this scenario is, it serves as part of the foundation of ideas and impressions upon which Gairdner subsequently relies in trying to argue that homosexuals working in daycare centres are likely to cause inordinate numbers of children to become homosexuals themselves. In the sample passage, Gairdner suggests a cause-and-effect relationship between the sexual orientation of a care-giver and that of the children who are being cared for, writing that this hypothetical set of children “would likely end up fancying homosexual behaviour themselves”. The degree of certainty is mitigated by the modal adverb “likely”, which allows for the possibility of doubt. Gairdner also stops short of asserting that such children would irrevocably become homosexuals, claiming instead that they would likely “fancy” homosexual behaviour, a somewhat weaker claim. However, even modalized as it is, this sentence contains some insidious implications. As he establishes the scenario, the desert island contains one adult (presumed to be male since Gairdner routinely uses homosexual for males and lesbian for females) and two or more children. The children are presumably quite young since they have no prior sexual knowledge. As the only adult, the caregiver would have no available adult sexual partner with whom the children might observe him engaging in sexual acts. That being the case, how could he model for them the “homosexual behaviours” they purportedly grow to fancy? The answer is clear: Gairdner is implying that the adult will engage in sexual acts with the children. Thus, even without saying it explicitly, Gairdner is planting in the reader’s mind an implication that homosexuals are pedophiles, and that in the hypothetical scenario, the homosexual care-giver would deliberately expose the children to homosexual acts. This implication is made more explicit in the third HOMOPHOBIC DISCOURSE: A ‘POPULAR’ CANADIAN EXAMPLE 133 sentence of the passage, “In other words, it is quite possible to train a young population to like homosexual behaviour”. The verb “to train” implies a conscious and deliberate teaching, together with practice in the behaviour. Sentences 4 and 5 turn the reader’s attention away from the hypothetical example and back to the ‘real world’. Both sentences utilize a modality of validity which is neither emphasized nor mitigated and which constitutes a sort of unmarked expository style. In terms of the content of these sentences, Gairdner may be reassuring the readers that societies prevent “this” (the child abuse implied by sentence 3), but he qualifies the assertion through the phrase “on a large scale” in such a way that he suggests that such abuse occurs on a small scale and that gay men must therefore still be considered a threat. Presumably, this suggestion is meant to justify what is asserted in sentence 8, which is that societies intentionally suppress homosexual behaviour. This amounts to justifying the persecution of homosexuals on the basis of a preposterous hypothetical scenario involving people stranded on a desert island. In the space of nine short lines, Gairdner has therefore moved the reader from an empirically unjustifiable proposition, through a preposterous hypothetical situation, to an unsubstantiated generalization about children being trained to become homosexuals, and finally to a justification for persecuting homosexuals on no grounds other than their homosexuality. In order to understand more completely Gairdner’s use of the noun “predisposition”, one must look to sentences 8 through 10. In sentence 8, a predisposition to homosexuality is equated with a predisposition for an alcoholic to drink if drink is available. By equating homosexuality with alcoholism, Gairdner pathologizes it and presents it as inherently harmful and unhealthy. Gairdner then portrays both alcoholism and homosexuality as simply matters of desire. “But nothing except personal desire makes the bottle (or the penis) jump into the mouth” (sentence 9). The assertion “nothing except” conveys a strong modality of certainty and leaves no room for evidence or argument to the contrary, such as evidence that alcoholism may be a disease with a physiological basis, or that sexual orientation may be something over which a person has no control and which is innate. Still not satisfied that he has sufficiently denigrated homosexuals, in sentence 10 Gairdner collocates “predispositions”, the noun he has used in connection with homosexuality, with “all sorts of weird appetites and behaviours”. The impression that Gairdner is engaging in an undisguised attack on homosexuals is strengthened by the list of items 134 DONNA L. LILLIAN with which he collocates homosexuality in exemplifying what he describes as weird: suicide, drunkenness, killing, stealing, overeating, and bad temper. What the items in this collocational set have in common is that they all represent negative states or activities. Beyond that, there are subsets of items from the list that may be related, but the list as a whole does not form any particularly natural or transparent set. Killing and stealing (and in some jurisdictions suicide) constitute illegal acts. Suicide and killing both pertain to death. Drunkenness and overeating both pertain to indulgence in the appetites of the palate and/or gullet. Bad temper cannot be linked in any apparent way with any of the other items in the list, nor can homosexuality. Nevertheless, there is a strong tendency for a reader/hearer to attempt to find some semantic link between items enumerated within a list. The negative judgement made about all the items in the list is the only obvious link. Still, linking homosexuality with such things as overeating and bad temper compounds the negativity by also trivializing it and holding it up to ridicule. In sentence 11, Gairdner begins to move away from the particularization implied by his use of the first person singular, toward a claim for universality, and then in sentence 12 he moves from first person singular to first person plural, as if to imply that what started out as Gairdner’s view is being extended to encompass a larger group. The context does not explicitly indicate whether the pronoun we (used twice) is intended to include or to exclude the reader, but given that just seven lines earlier, the reader was addressed in an imperative clause, the default assumption is that the reader is meant to include him/herself in the pronoun. The final sentence of the sample passage states, “It’s a dead end” (line 22). The context would lead to an interpretation that the academic debate is a dead end, but that in turn leads to an interpretation that finding out the origins or causes of homosexuality is a dead end. An interpretation that is not ruled out by the context is also that homosexuality is itself a dead end, and this interpretation is in fact consistent with Gairdner’s use elsewhere of imagery of death in connection with homosexuality. The passage analyzed in this section constitutes only a very small sample of the many pages that Gairdner has written about homosexuality, but this analysis is sufficient to demonstrate the range of devices that Gairdner employs in his anti-homosexual : lexical choices, metaphor, modality, syntax, mood, modality, pronoun choice, collocation and implication. HOMOPHOBIC DISCOURSE: A ‘POPULAR’ CANADIAN EXAMPLE 135

6. Neo-Conservative Ideology in Canada

The ideology reflected and reproduced in the popular writings of William Gairdner is that of neo-conservatism. Neo-conservatism, or the ‘New Right’, seeks to protect the interests of the dominant elites in Western countries including, but not limited to, Great Britain, the United States, and Canada. Trevor Harrison characterizes neo-conservatism as

[…] an amalgam of classical liberal economic and political theory with traditional conservative social and moral doctrine. Specifically, neo-conservatism is said to promote capitalist socio-economic structures and beliefs (the free-market system, individualism, a minimal state, and private ownership of property) while espousing a belief in natural inequalities and natural authorities (the Christian church, the family, and the state). (Harrison 1995: 20-21)

Rather than being associated directly with any one political party or one social group, the neo-conservative movement actually comprises a de facto alliance of sorts among at least three different sectors of society: the international corporations, assorted smaller forms of capital which favour a libertarian, ultra-free enterprise version of capitalism, and fundamentalist religion (Marchak 1988: 187). Particularly within North America, the ‘family values’ constituency, often associated with evangelical Christianity, makes up a significant core of the New Right, although not all who support the so-called family values agenda of the New Right are themselves evangelical or even necessarily Christian (Peele 1984: 25). In both its Jewish and Christian origins, however, it is on the cluster of social family values issues that the religious right has its strongest impact. Despite some differences in emphasis and origin, the different strands of neo-conservatism have in common that they privileging white, heterosexual, English-speaking males above all other individuals and groups (Marchak 1988). The neo-conservatism in Great Britain and in the U.S.A. peaked in its influence during the 1980’s under Margaret Thatcher (Prime Minister from 1979 - 1990) and Ronald Reagan (President from 1980 - 1988), but it still exerts a strong influence in American politics in the 1990’s and beyond, as evidenced by the Republican Presidential victories of George W. Bush in 2000 and in 2004. The neo-conservative movement was slower to gain a foothold in Canada than in either Great Britain or the United States, however, and did not begin to manifest itself as a force to be reckoned with until the 1990’s. 136 DONNA L. LILLIAN

Jeffrey attributes the slower development of neo-conservatism in Canada to four factors. First, she argues that Canada’s political culture tends to be further to the left than that of the United States, as illustrated by programmes such its public health system, making Canadians slower to embrace neo-conservatism. Second, she argues that Canadians tend to exhibit a greater respect for authority and deference to elites than have Americans (Jeffrey 1999: 45). Third, Jeffrey explains that,

[w]ith a much smaller population and far less capital for private-sector investment, state intervention in the economy and the provision of government- sponsored social programs through the welfare state came to be not only strongly supported, but expected. Many Canadians considered it a right of citizenship. Risk-taking and entrepreneurship, by contrast, never figured prominently in the Canadian identity. (Jeffrey 1999: 46)

Finally, the fact that Canadians have tended to perceive neo-conservatism as being a predominantly American phenomenon has also made them less inclined to adopt it. Jeffrey notes,

[s]ince most Canadians define their national identity as one that is emphatically not American, they have been very unwilling to import American political baggage. (Jeffrey 1999: 47)

To date, neo-conservative governments have been elected in Canada’s two wealthiest provinces, Alberta and Ontario. In 1992, neo-conservative Ralph Klein won the leadership of the Alberta Progressive Conservative Party and assumed the premiership of Alberta. Klein went on to win the 1993, 1997, and 2001, and 2005 provincial elections and at the time of this writing is still Premier of Alberta. In Ontario, Mike Harris and his far-right Progressive Conservative Party won the provincial election in 1995 and immediately began instituting the measures outlined in his party’s neo- conservative manifesto, the Common Sense Revolution. Harris’ government was re-elected in 1999 and served one more term in office before being defeated by the Liberal Party. On the national scene, during the 1980’s, an assortment of neo- conservatives based in western Canada dreamed of pursing their own agenda in Canadian federal politics. Finally, in November 1987, they founded a new party, The Reform Party of Canada, acclaiming Preston Manning as its first leader. Manning held that position until 2000, when the Reform Party of Canada morphed into the new Conservative Reform Alliance Party of Canada, which in turn merged with the Progressive HOMOPHOBIC DISCOURSE: A ‘POPULAR’ CANADIAN EXAMPLE 137

Conservative Party of Canada in 2003 to form the Conservative Party of Canada. The Reform Party embraced not only the sort of fiscal conservatism pursued by the neo-conservative governments of Ontario and Alberta, but also a very strong social and moral conservatism (Jeffrey 1999: 50). With each reincarnation of the party, it adopts less extreme policy platforms and moves closer to the centre of the Canadian political spectrum, but to date it has not succeeded in winning a federal election. William Gairdner had a strong influence on the neo-conservative movement in Canada, particularly in its early days, and his role in the Reform Party was considerable during its formative years. In fact, Trevor Harrison places Gairdner among the most influential core members of Reform (Harrison 1995: 209). For example, Gairdner was a keynote speaker at the 1991 Reform Party’s Assembly in Saskatoon. Murray Dobbin reports that The Trouble with Canada, upon which Gairdner’s speech was based, had become “the de facto manifesto for Preston Manning’s Reform Party” (Dobbin 1992: 134). Dobbin further reports that Gairdner’s 1991 Saskatoon address was only the first among many such speeches to Reform audiences and that “Gairdner is one of the party’s most frequent guest speakers at rallies” (Dobbin 1992: 118). Gairdner has not run for elected office, but through his books, speeches, and media appearances, his influence over the agenda of neo-conservatives in Canada has been considerable. Gairdner attempts to portray himself as just an ordinary Canadian, but there can be little doubt that he speaks from a position of social power, being wealthy, white, male, heterosexual, and highly educated, holding a Master’s degree in , a Master’s in English/Creative Writing, and a doctorate in English Literature, all from Stanford University. He is also a former Commonwealth and Olympic decathlete, one-time professor of English, former President and owner of The Fitness Institute of Canada, former Chair of The Gairdner Foundation, and current President and owner of Gairspar Investments, Ltd. Gairdner has social and linguistic power and he wields it in his writing in order to induce other people to adopt his political and social policies. This makes his books appropriate for critical discourse analysis, particularly because he somewhat problematically positions himself as opposing academic approaches to knowledge, even though it is his training within academia that in large measure gives him the credibility to have his books published. 138 DONNA L. LILLIAN

7. Homophobia and Homophobic Discourse

The discursive strategies apparent in Gairdner’s writings are representative of the ways in which neo-conservatives express particular attitudes, not only toward homosexuals, but also toward other groups with which they are at odds. Critical theorists have argued that neo-conservatism promotes and enables racism, sexism, and homophobia; however, fewer academic studies have been published on homophobic discourse than on either racist discourse or sexist discourse. Barbara Smith notes that

[h]omophobia is usually the last oppression to be mentioned, the last to be taken seriously, the last to go. But it is extremely serious, sometimes to the point of being fatal. (Smith 1993: 99)

The relative paucity of linguistic analyses of homophobic discourse may, at least in part, reflect this tendency for homophobia to be prioritized below racism and sexism as forms of oppression to be challenged both within and without academia. Ironically, according to Smith, while the anti-oppression forces of the left may sometimes fail to recognize the full impact of homophobia,

for the forces on the right, hating lesbians and gay men, people of color, Jews, and women go hand in hand. They make connections between oppressions in the most negative ways with horrifying results. (Smith 1993: 100)

Elizabeth Morrish, writing about Great Britain, observes that there is a certain cumulative effect to public discourses about lesbians and gays.

In the case of discourse about lesbians and gays, there is a clear cumulative effect that leads to an atmosphere in which homosexuality is deemed ‘unnatural’ and threatening to the dominant culture and in which homophobic acts are held to be justified and in the interests of the ‘general public’. (Morrish 1997: 335)

Morrish notes two recurring themes in public discussion of homosexuality, namely family life and disease (Morrish 1997: 338). She observes that, “... most often in contemporary discourse, the term family has become a codeword for the exclusion of homosexuality” (Morrish 1997: 339). Both themes, family life and disease, are also prominent in William Gairdner’s discourse on homosexuality. Gayle Rubin likewise links right-wing politics with narrow definitions of appropriate sex. HOMOPHOBIC DISCOURSE: A ‘POPULAR’ CANADIAN EXAMPLE 139

Right-wing ideology linking non-familial sex with communism and political weakness is nothing new. During the McCarthy period, Alfred Kinsey and his Institute for Sex Research were attacked for weakening the moral fibre of Americans and rendering them more vulnerable to communist influence. (Rubin 1993: 7-8)

Rubin’s term non-familial sex includes same-sex relationships. According to her analysis, dominant North American views divide sex into a hierarchy consisting of “the charmed circle”, seen as Good, Normal, Blessed Sexuality, consisting of sex that is heterosexual, married, monogamous, procreative, non-commercial, in pairs, in a relationship, same generation, in private, no pornography, bodies only, and vanilla, versus “the outer limits”, seen as Bad, Abnormal, Unnatural, Damned Sexuality, and consisting of sex that is homosexual, unmarried, promiscuous, non-procreative, commercial, alone or in groups, casual, cross-generational, in public, pornography, with manufactured objects, and/or sadomasochistic (Rubin 1993: 13). William Gairdner’s characterization of sex into acceptable and unacceptable varieties corresponds closely to the model described by Rubin. In “The Gay Agenda: Marketing Hate Speech to Mainstream Media”, Marguerite J. Moritz identifies some of the extreme anti-homosexual rhetoric of the religious right in the U.S.A. as hate speech (Moritz 1995: 58). Some of the most egregious forms of this propaganda include calls for the death penalty for homosexuals, whom these propagandists purport violate Biblical laws and principles. According to Moritz, these extreme forms of rhetoric gain little support for the anti-gay lobby outside of fundamentalist circles, so the radical right has moderated its rhetoric somewhat to appeal to mainstream audiences, often reframing their arguments as a special rights issue (Moritz 1995: 58). Marian Meyers echoes the observation about the use of a category of ‘special interest group’ to marginalize gays and lesbians and to distance them from the so- called mainstream of America (Meyers 1994: 340). Although William Gairdner’s own anti-gay rhetoric at times becomes quite extreme, his use of the special rights argument against gays and lesbians also links him to the forces Moritz refers to as the radical right. The fact that Gairdner’s anti-gay rhetoric is not isolated, but rather forms part of a wide continuum of such discourse makes it all the more potent. Many Canadians, especially those who adhere to far-right political and/or religious positions, are accustomed to reading American perspectives on issues and thus find that Gairdner’s writings resonate with ideas and even styles of language with which 140 DONNA L. LILLIAN already feel comfortable. Gairdner sows his seeds of hatred in ground that has been already tilled by the years of homophobic propaganda that various right-wing elements have been circulating throughout North America. Vasu Reddy (2002), analyzing print and electronic news sources in Africa, identifies homophobic discourse as hate speech.

The point here is that homophobia, externalized as hate speech, is not merely an emotional response but a political, social and cultural response to the increasing visibility of homosexuality as a real and lived experience of African gays and lesbians. (Reddy 2002: 164)

Reddy examines seven examples of homophobic discourse from Namibia, Zimbabwe, Kenya, Swaziland, Uganda, Zambia, and Egypt. What is clear from these examples is that homophobia and homophobic discourses are phenomena not constrained by geographical or political boundaries. Rather, homophobia manifests itself internationally and cross-culturally, threatening the physical and psychological safety of lesbians and gays worldwide.

8. Conclusion

The scorn William Gairdner expresses toward gay men as he propagates his homophobic agenda to his readers is both unmistakable and overwhelming. In reality, homosexuals reflect the same range of diversity as heterosexuals do in terms of race, ethnicity, religion, and politics. They are also diverse in their sexual practices, as are heterosexuals. Yet Gairdner dismisses the majority of homosexuals and then presents a gruesome and often violent picture of homosexual behaviour as if that represented the norm.

And I know that there are a lot of respectable, kind-hearted, talented homosexuals out there. But that’s not the issue here. The issue is that homosexuals have broken the implicit pact of a conservative society, which has always been to tolerate unnatural private behaviour, but never to approve of it; nor allow it any public weight; and certainly never to force acceptance of it upon the citizens at large. . . . So this chapter is not about homosexual individuals who are minding their own business. Nor is it intended to hurt the feelings of otherwise proper homosexual citizens, many of whom are themselves extremely distraught by homosexual behaviour and the radical agenda. (Gairdner 1992: 356-357)

These condescending and insulting disclaimers in no way counter-balance the pages of negative and violent imagery which are presented to the HOMOPHOBIC DISCOURSE: A ‘POPULAR’ CANADIAN EXAMPLE 141 readers as simply a description of ‘homosexual’ behaviour. The repeated pairing of homosexuality with pedophilia and bestiality, the images of violence and disease, and the erroneous identification of AIDS as the gay plague become indelibly imprinted in the mind of the reader even though they are distorted, exaggerated, and unrepresentative of gay practices and values. This treatment of homosexuals is consistent with one of the hate strategies identified by Rita Kirk Whillock:

Another strategy the rhetor may use is to vilify the out-class in ways that make it difficult to elicit the support of others in mounting a response. The villains in a hate story must, therefore, be made to appear as evil, opposing the values of the audience. (Whillock 1995: 40)

Homosexuals are presented by Gairdner as one-dimensional beings entirely consumed by insatiable sexual desire, not as people. The reason he presents them this way is clear: he wants to eradicate homosexuality altogether, so he works hard to discredit homosexuals. Gairdner has four main strategies for getting rid of homosexuality. His first strategy is to get rid of feminism, since he thinks that feminism causes homosexuality.

It is no simple coincidence that homosexuality is thriving in a time of sexual egalitarianism and feminism. The two go together like the two sides of a coin. (Gairdner 1992: 318)

Gairdner’s second approach to eliminating homosexuals is to ‘cure’ them (Gairdner 1992: 373). Obviously, any discussion of curing homosexuality presupposes that it is some sort of disease or addiction, analogous, perhaps, to alcoholism or drug abuse: “[s]ome may be predisposed, in the way an alcoholic may be predisposed to drink if drink is available” (1992: 370). That the evidence does not support this supposition does not deter Gairdner. A third approach Gairdner advocates in his campaign to eliminate homosexuality is to censor sex education programmes in schools so that they either teach against homosexuality and hold up married heterosexuality as the only acceptable sexual choice, or else simply avoid mentioning homosexuality entirely. Gairdner’s fourth strategy for eliminating homosexuality would be to make practising it illegal. Along with advocating making certain sexual acts illegal, Gairdner proposes actively limiting the rights and freedoms of homosexuals. Although he ‘only’ advocates repression of homosexuals, he nevertheless writes approvingly of the oppression they are subjected to in China. 142 DONNA L. LILLIAN

This [claiming that homosexuals are curable] amounts to saying that all human societies have engaged in -- and must continue to engage in -- a civilized repression of socially unsavoury behaviour as a matter of course, for the sake of self-protection and child protection. Some, like China and other socialist states, actually engage in oppression of such behaviour, which seems to do the job. (Gairdner 1992: 374)

Gairdner contends that homosexuals are not entitled to the full range of benefits available to other people in our society. He states unequivocally,

I should state my view that as citizens of a free country, homosexuals ought to have the same legal protections as any other citizen, as long as they keep their proclivities to themselves, but never the same privileges. (Gairdner 1992: 356)

He advocates, in effect, designating homosexuals as second-class citizens and repressing them such that they must hide their true identities in public.

Clearly, in the case of homosexuality, it is far better for a normal society to be unfair to homosexuals in the name of its own social health, than to be unfair to the entire society in the name of fairness to homosexuals (or any other pressure group.) (Gairdner 1992: 361)

Based on the evidence I have presented in this paper, Gairdner’s discourse can and must be identified as nothing less than homophobic propaganda. Homophobia may not get the same wide coverage as racism and ethnicism, yet the discrimination faced by sexual minorities is every bit as devastating as that faced by any other oppressed group, and we as critical discourse analysts must work every bit as hard at identifying and challenging homophobia as we do at identifying and challenging racism and sexism.

References

Ashley, L. (1979) Kings and queens: Linguistic and cultural aspects of terminology. , 3: 215-256. Baugh, John (1991) The politicization of changing terms of self-reference among American slave descendants. 66 (2): 133-146. Concise Oxford Dictionary of Current English, Seventh Edition (1982) Oxford: The Clarendon Press. Dobbin, Murray (1992) Preston Manning and the Reform Party. Halifax: Formac Publishing Company. Fowler, Roger (1985) Power. In Teun A. van Dijk (ed.), Handbook of Discourse Analysis, Vol. 4, pp. 61-82. London: Academic Press, Inc. HOMOPHOBIC DISCOURSE: A ‘POPULAR’ CANADIAN EXAMPLE 143

—— (1991) Language in the News: Discourse and Ideology in the Press. London: Routledge. Gairdner, William D. (1990) The Trouble With Canada: A Citizen Speaks Out. Toronto: Stoddard. —— (1992) The War Against the Family: A Parent Speaks Out. Toronto: Stoddard. —— (1994) Constitutional Crackup: Canada and the Coming Showdown With Quebec. Toronto: Stoddard. —— (1996) On Higher Ground: Reclaiming a Civil Society. Toronto: Stoddard. —— (2001) The Trouble With Democracy: A Citizen Speaks Out. Toronto: Stoddard. Harrison, Trevor (1995) Of Passionate Intensity: Right-Wing Populism and the Reform Party of Canada. Toronto: University of Toronto Press. Jacobs, Greg (1995) “Homosexual,” Gay,” or “Queer”: The struggle over naming and Its Real-World Effects. Linguistica Atlantica, 19: 93-114. —— (1998) The struggle over naming: A case study of ‘queer’ in Toronto, 1990-1994. World Englishes, 17 (2): 193-201. Jeffrey, Brooke (1999) Hard Right Turn: The New Face of Neo-Conservatism in Canada. Toronto: Harper Collins. Lillian, Donna L. (1997) Transitivity as an ideological tool: The discourse of William D. Gairdner. Papers from the Twenty-First Annual Meeting of the Atlantic Provinces Linguistic Association, pp. 122-131. Halifax, Nova Scotia. —— (2001) Canadian neo-conservative discourse: A critical discourse analysis. Unpublished Ph.D. dissertation, York University, Toronto. Marchak, Patricia M. (1988) Ideological Perspectives on Canada. Third Edition. Toronto: McGraw-Hill Ryerson Limited. Meyers, Marian (1994) Defining homosexuality: News coverage of the ‘repeal the ban’ controversy. Discourse & Society, 5 (3): 321-344. Moberly, Elizabeth R. (1983) Homosexuality: A New Christian Ethic. Greenwood, South Carolina: Attic Press. Moritz, Marguerite J. (1995) The gay agenda: Marketing hate speech to mainstream media. In Rita Kirk Whillock and David Slayden (eds.), Hate Speech, pp. 55-79. London: Sage. Morrish, Elizabeth (1997) Falling short of God’s ideal: Public discourse about lesbians and gays. In Anna Livia and Kira Hall (eds.), Queerly Phrased: Language, Gender, and Sexuality, 335-345. New York: Oxford University Press. OUT!SPOKEN Styleguide: A Guide for the Media on Lesbian, Bisexual, Gay and HIV/AIDS Issues. (n.d. [circa 1994]) Toronto: OUT!SPOKEN. Peele, Gillian (1984) Revival & Reaction: The Right in Contemporary America. Oxford: Clarendon Press. Reddy, Vasu (2002) Perverts and sodomites: homophobia as hate speech in Africa. Southern African Linguistics and Applied Language Studies, 20: 163-175. Rubin, Gayle (1993) Thinking sex: Notes for a radical theory of the politics of sexuality. In Henry Abelove, Michèle Aina Barale and David M. Halperin (eds.), The Lesbian and Gay Studies Reader, pp. 3-44. New York: Routledge. Simpson, Paul (1993) Language, Ideology and Point of View. New York: Routledge. 144 DONNA L. LILLIAN

Smith, Barbara (1993) Homophobia: Why bring it up? In Henry Abelove, Michèle Aina Barale and David M. Halperin (eds.), The Lesbian and Gay Studies Reader, pp. 99-102. New York: Routledge. Sykes, Mary (1985) Discrimination in discourse. In Teun A. van Dijk (ed.), Handbook of Discourse Analysis, Vol. 4, pp. 83-102. London: Academic Press, Inc. van Dijk, Teun A. (1990) Discourse & Society: A new journal for a new research focus. Discourse & Society, 1(1): 5-16. —— (1994) Critical discourse analysis. Discourse & Society, 5(4): 435-436. Whillock, Rita Kirk (1995) The use of hate as a stratagem for achieving political and social goals. In Rita Kirk Whillock and David Slayden (eds.), Hate Speech, pp. 28-54. London: Sage.

Contact information:

Donna L. Lillian Department of English 2201 Bate Building East Carolina University Greenville, NC 27858-4353, U.S.A. e-mail: [email protected]

Francesca Masini

Multi-word Expressions between Syntax and the Lexicon: the Case of Italian Verb-particle Constructions*

Abstract

This paper focuses on multi-word expressions, i.e. lexical units larger than a word that can bear both idiomatic and compositional meanings. Multi-word expressions are a privileged domain for the study of the syntax-lexicon relationship, which is also the main concern of current theoretical approaches such as Construction Grammar. In particular, the paper provides a synchronic and diachronic analysis of a particular (and quite neglected) kind of multi-word expression, i.e. Italian verb-particle constructions. The synchronic part provides a syntactic and semantic analysis of these complex predicates: their syntactic and semantic behaviour will be taken as evidence of their structural cohesion. The second part deals with the development of verb-particle constructions in Italian and provides original diachronic data that will shed new light on the issue. Finally, we will draw some theoretical conclusions from the analyses and show how the findings seem to speak in favour of a constructional approach to verb- particle constructions and to multi-word expressions in general.

1. Introduction: multi-word expressions and the lexicon

This paper focuses on multi-word expressions1 (MWE), i.e. lexical units larger than a word that can bear both idiomatic and compositional meanings. MWEs have received more and more attention in recent years (cf. for instance Moon 1998 and Wray 2002). From a theoretical point of

* This work was carried out within the PhD programme of the Department of Linguistics, University Roma Tre. I am particularly grateful to Raffaele Simone for invaluable support and to Geert Booij, Claudio Iacobini and Alessandra Trecci for useful remarks and suggestions on an earlier draft of this paper. Besides, I would like to thank the audiences at the symposium “The lexicon: its status in the theory of language” (Turku, 18–19 November 2004) and at the seminars held at the University of Bologna (31 January 2005) and the Free University of Amsterdam (19 May 2005) for their comments. Last but not least, I would like to thank the two reviewers for considerably improving the form and the contents of the paper. All remaining errors are of course mine. 1 Here the term multi-word expression is used as a pre-theoretical label to include the range of phenomena that goes from collocations to fixed expressions.

SKY Journal of Linguistics 18 (2005), 145–173 146 FRANCESCA MASINI view, due to their double status as phrasal structures and lexical items (which are usually identified with words), MWEs are a privileged domain for the study of the relationship between syntax proper (intended as free combination of words) and the lexicon. This very issue is the main concern of current theoretical approaches such as Construction Grammar (cf. Fillmore, Kay and O’Connor 1988, Goldberg 1995, 2003), which claims that “constructions” (i.e. a form/meaning pairings) are the basic units of linguistic knowledge. In this framework, syntax and the lexicon are not strictly separated, but rather we have a continuum that goes from morphemes to general syntactic configurations (cf. Goldberg 2003). In this paper we will bring evidence in favour of a constructional view of language and, more specifically, we will claim that certain phrasal constructions may functionally correspond to morphologically complex words. In particular, we will apply this view to the case of Italian verb- particle constructions. The paper is divided into three main parts. The first part will provide a synchronic syntactic and semantic analysis of these complex verbs, in order to underline their peculiar structural and semantic behaviour. In the second part, we will discuss the question of the diachronic development of these verbs in the and provide new data that will shed new light on the issue. Finally, we will pass on to theoretical considerations and try to interpret the findings in a constructional perspective.

2. Italian verb-particle constructions

So-called verb-particle constructions are very common in and have inspired much work in recent years (cf. Dehé et al. eds. 2002, and Booij and van Marle eds. 2003).2 In Germanic languages, we can distinguish two main patterns:

ƒ ‘separable complex verbs’ in German and Dutch, which are formed by a verb and a separable prefix or preverb; the latter is separated from the verb under certain syntactic conditions, like for example verb second movement in main clauses (e.g. Dutch opzoeken lit. up search ‘to look up’: dat Jan de informatie opzoekt ‘that John looks up the

2 In Europe we can find a productive system of ‘separable complex verbs’ also in Ugro- such as Hungarian and Estonian (cf. Ackerman and Webelhuth 1998). MULTI-WORD EXPRESSIONS BETWEEN SYNTAX AND THE LEXICON 147

information’ vs. Jan zoekt de informatie op ‘John looks up the information’);3

ƒ ‘particle verbs’ or ‘phrasal verbs’ in English, Swedish, Norwegian, Icelandic and Danish, which are formed by a verb followed by a modifying particle (e.g. English eat up: John ate the cake up or John ate up the cake).4

These constructions do not, however, seem to be very popular among the major . Although we can find sporadic examples in French (e.g. jeter dans lit. throw in ‘to throw in’) or Spanish (e.g. echar abajo lit. put down ‘to demolish’), the pattern does not seem to be productive in these languages. In this general picture, Italian is an exception. Actually, Italian displays a quite productive and rich system of verb-particle constructions (VPC)5 with the structure in (1):

(1) [ [Verb] [Particle] ] where Particle = locative adverb6

Some examples follow in (2):

(2) [mettere]V [giù]ADV lit. put down ‘to put down’

[mandare]V [avanti]ADV (un’impresa) lit. send forward ‘to run (a business)’

[fare]V [fuori]ADV lit. make out ‘to kill’

However, despite their great popularity in present-day Italian, VPCs were first acknowledged as a widespread (and neglected) phenomenon less than ten years ago by Simone (1997).7 Indeed, linguists (and lexicographers) have devoted, for some reason, little attention to these constructions.

3 The data are from Blom (2005: 6), to which we refer for a comprehensive study of the phenomenon in Dutch. As for German, cf. Stiebels and Wunderlich (1994), Lüdeling (2001) and Müller (2002). 4 As regards English, the literature is vast. Cf. in particular Bolinger (1971), Fraser (1976), Dixon (1982) and, more recently, den Dikken (1995) and Dehé (2002). Besides, see Toivonen (2003) for Swedish and Jansen (2002) for Danish, as well as Haiden (2001) for an overview of the phenomenon in the various Germanic languages. 5 Though it is by far less rich and productive than in English, for example. 6 Here the term particle is used as a descriptive, atheoretical notion. 7 Cf. also Schwarze (1985) and Venier (1996). 148 FRANCESCA MASINI

Nevertheless, they seem to be all but a marginal phenomenon in present- day Italian: according to Iacobini (2003, to appear), most of the new verbs with locative meaning in Italian are VPCs. This led us to investigate them more closely and to look for the reasons behind their development.

3. The synchronic analysis

Italian VPCs display specific syntactic and semantic properties that testify to the high cohesion of the pattern. First of all, the combination of verb (V) and particle (P) is not as free as the sequences formed by verb plus prepositional phrase (V+PrepP) or verb plus adverbial phrase (V+AdvP). This means that the P does not form a separate constituent and that V and P form a complex predicate. Section 3.1 will illustrate this point. Secondly, the semantic behaviour of VPCs is rather varied and the metaphorical bleaching many of them underwent is further evidence of their cohesion. Section 3.2 will provide a semantic classification of VPCs.

3.1 The syntactic and morphological behaviour of Italian VPCs

In what follows, I shall present a set of syntactic criteria (3.1.1–3.1.3) for identifying Italian VPCs and distinguishing them from V+AdvPs and V+PrepPs.8 Section 3.1.4 will examine the nominalization of VPCs.

3.1.1 Insertion of material between V and P and object shift

Only light non-argumental constituents can be inserted between V and P. With transitive verbs (example 3), the direct object normally follows the whole VPC (3a), so object shift, like in (3b), is not possible.9 Light

8 The following criteria are partially based on Fraser (1976: 2–4) and Simone (1997). 9 This feature distinguishes Italian VPCs from English VPCs, which are well-known for their object shift properties, e.g. I looked the information up vs. I looked up the information (cf. e.g. Dehé 2002). Of course, this generalization might face some counterexamples:

(I) a. Metti su la borsa put.imperative up the bag ‘Put up the bag’ b. Metti la borsa su put.imperative the bag up ‘Put the bag up’

MULTI-WORD EXPRESSIONS BETWEEN SYNTAX AND THE LEXICON 149 constituents can be interposed between V and P, like subito ‘immediately’ in (3c), but heavy constituents (e.g. con accanimento ‘with tenacity’ in 3d) cannot. Note that order changing with PrepPs (4a,b) or AdvPs (5a,b) does not cause any grammaticality problems.10

(3) a. Luca ha lavato via la macchia Luca have.3sg wash.part.past away the stain ‘Luca removed the stain (by washing)’

b. *Luca ha lavato la macchia via c. Luca ha lavato subito via la macchia d. *Luca ha lavato con accanimento via la macchia

(4) a. Luca ha lavato la macchia con il sapone Luca have.3sg wash.part.past the stain with the soap ‘Luca washed the stain with the soap’

b. Luca ha lavato con il sapone la macchia Luca have.3sg wash.part.past with the soap the stain

(5) a. Luca ha lavato la macchia subito Luca have.3sg wash.part.past the stain immediately ‘Luca washed the stain immediately’

b. Luca ha lavato subito la macchia

However, this possibility seems to involve borderline cases in which the P can be interpreted both as a direction marker and as the goal of the motion, like in (I). In these cases, object shift seems to be a strategy to stress the latter interpretation, like in (Ib) (cf. also section 3.1.2). Of course, information structure has a role here (cf. e.g. Lambrecht 1994). In this respect, one of the reviewers noticed that la macchia ‘the stain’ is the focused information and thus needs to occur in final position, which is also supported by the fact that, adding dai jeans ‘from the jeans’ to (3b), the example becomes perfectly acceptable:

(II) Luca ha lavato la macchia via dai jeans ‘Luca removed the stain from the jeans’

However, la macchia is not necessarily the focused information in (3a). Further, and most importantly, there are reasons to think that via da ‘away from’ forms a separate construction. Italian actually seems to have a set of complex locative relators of different kinds formed by prepositions and adverbs (cf. e.g. Jansen 2004, who proposes to consider this set of constructions as a “syntactic space” à la Croft 2001). 10 Of course, there may be pragmatic differences between (4a) and (4b) on the one hand, and (5a) and (5b) on the other, but all of them are acceptable, whereas (3b) is not. 150 FRANCESCA MASINI

In our view, the limited separability of V and P testifies to the high cohesiveness of their relationship and to the non-independent nature of the particle. Of course, the adverbial constituent-like origin of the particle is still visible at certain levels and gives rise to borderline cases like those illustrated in note 9. This results in a scale of acceptability, which also applies to other tests, as we will see in the following sections.

3.1.2 Left dislocation and topicalization

Whereas PrepPs and AdvPs can normally undergo extraposition (cf. for example 9 and 10 below), particles cannot be freely dislocated with constructions such as è... che... ‘it is... that...’, as examples (6b), (7b) and (8b) show, and cannot be freely topicalized (cf. examples 6c, 7c and 8c).

(6) a. È andato dentro be.3sg go.part.past inside ‘He went in’ b. ??È dentro che è andato c. ??Dentro è andato

(7) a. Abbiamo messo su il caffè have.3pl put.part.past up/on the coffee ‘We put the coffee on’ b. *È su che abbiamo messo il caffè c. *Su abbiamo messo il caffè

(8) a. Maria manda avanti l’ azienda di famiglia Maria send.3sg forward the business of family ‘Maria runs the family business’ b. *È avanti che Maria manda l’azienda di famiglia c. *Avanti Maria manda l’azienda di famiglia

Examples (7) and (8) are rather clear and unproblematic. Of course, the metaphorical (7) and idiomatic (8) meaning of the VPCs involved plays a role here. The examples in (6), on the other hand, are borderline cases between VPCs and proper V+AdvP combinations. Consider for example (9) below.

MULTI-WORD EXPRESSIONS BETWEEN SYNTAX AND THE LEXICON 151

(9) a. È andato là dentro be.3sg go.part.past there inside ‘He went inside over there’

b. È là dentro che è andato c. Là dentro è andato

The presence of the deictic là ‘there’ tells us that dentro is here a true adverb, hence the whole AdvP là dentro ‘there inside’ can be either dislocated (9b) or topicalized (9c). Thus, whereas in (9) the adverbial phrase (là) dentro unambiguously indicates the concrete endpoint of the process of going, in (6) dentro can have a double interpretation: of course it indicates the path of the generic verb of motion andare ‘to go’ (cf. also section 3.2), but the transparent semantics of the combination and the ambiguous categorial nature of the particle dentro11 can favour the concrete endpoint interpretation, which is why (6b) and (6c) have a double question mark.12 In order to see the difference to a true adverbial use, compare example (10), in which dentro can normally be topicalized (10b).

(10) a. La casa dentro è pulita the house inside be.3sg clean ‘The house is clean inside’

b. Dentro la casa è pulita

Summing up, the above considerations further support the idea of a continuum of constructions and of a scale of acceptability. The extraction of the particle is influenced not only by proper semantic reasons, but also by the ambiguous categorial status of some particles.

11 Of course, the categorial ambiguity of Ps consists in their adverbial nature (cf. also section 3.1.1 and note 9). Moreover, most elements that function as particles have homonymous prepositional counterparts. The close diachronic connection between adverbs, prepositions and particles/prefixes/preverbs in Indo-European languages is pointed out, among others, by Lehmann (1995). 12 This explanation in terms of categorial interpretation is just one of the possible viewpoints. Actually, the data might also be interpreted in terms of the “referential value” of the elements involved. As one of the reviewers noted, là dentro in (9) has a much higher referential value than dentro in (6). Indeed, this is a consequence of the shift from a pure adverbial function, which is syntactically free, to a modifying function, which is syntactically more bound. 152 FRANCESCA MASINI

3.1.3 Coordination

The third and last syntactic test we will adopt to measure the cohesiveness of VPCs is coordination. As we can see, PrepPs (11a) and AdvPs (11b) can be freely coordinated, since they behave like separate constituents.

(11) a. Marco sta dietro a Giovanni e davanti a Sandra Marco stay.3sg behind to Giovanni and in front of to Sandra ‘Marco is behind Giovanni and in front of Sandra’

b. Marco ha pulito la casa dentro e fuori Marco have.3sg clean.part.past the house inside and ouside ‘Marco has cleaned the house inside and outside’

On the contrary, it seems that particles cannot be coordinated as freely as PrepPs and AdvPs, as the examples in (12) illustrate. However, (12) also shows that things are rarely clear-cut. While (12a) seems completely out, (12b) is somehow more acceptable, and (12c) almost fine.

(12) a. *Sara ha portato fuori la bici e poi su la spesa Sara have.3sg take.part.past out the bike and then up the shopping ‘Sara took out the bycicle and then up the shopping’

b. ??Sara ha portato fuori la bici e Luca dentro la spesa Sara have.3sg take.part.past out the bike and Luca in the shopping ‘Sara took out the bycicle and Luca up the shopping’

c. ?Sara ha messo dentro la bici e fuori la spazzatura Sara have.3sg put.part.past in the bike and out the trash ‘Sara put the bike in and the trash out’

In conclusion, the coordination test confirms the peculiar status of particles and the existence of a scale of acceptability. Before passing on to the semantic properties of VPCs, we will briefly discuss their behaviour in relation to morphological operations such as nominalization.

3.1.4 Nominalization

Simone (1997) claims that Italian VPCs can be nominalized only by means of a nominal infinitive. Actually, in cases such as (13), in which the verbal MULTI-WORD EXPRESSIONS BETWEEN SYNTAX AND THE LEXICON 153 element is followed by a PrepP, the V can normally turn into either a nominal infinitive (13b) or a deverbal nominal (corsa ‘run’) (13c).

(13) a. La gente è corsa fuori dallo stadio the people be.3sg run.part.past out from.the stadium ‘The people ran out of the stadium’

b. Il correre della gente fuori dallo stadio the run.inf of.the people out from.the stadium

c. La corsa della gente fuori dallo stadio the run of.the people out from.the stadium

In contrast, with VPCs (like in 14), we cannot convert the V to the corresponding noun (14b), but we have to nominalize the whole VPC by means of the nominal infinitive (14c). Also note that the limited separability of V and P (cf. section 3.1.1) is also true of : as showed in (14d), it is not possible to insert the phrase di Gianni between the infinitive and the particle.

(14) a. Gianni è corso via subito dopo la partita Gianni be.3sg run.part.past away immediately after the game ‘Gianni ran away immediately after the game’

b. *La corsa via di Gianni subito dopo la partita the run away of Gianni immediately after the game

c. Il correre via di Gianni subito dopo la partita the run.inf away of Gianni immediately after the game

d. *Il correre di Gianni via subito dopo la partita

We must add that the vast majority of the verbal bases of Italian VPCs (which are basically motion verbs) present only one type of nominalization based on the feminine past of the verb, which can be both regular (tenere – tenuta ‘to keep – keeping’, venire – venuta ‘to come – coming’) or irregular (mettere – messa ‘to put – putting’, correre – corsa ‘to run – running/run’).13 Hence, verbal bases already feature limited nominalizing

13 The irregular type is a legacy. However, there is a debate on the exact source of this form. For discussion cf. Gaeta (2002: 150). 154 FRANCESCA MASINI possibilities. However, when occurring within a VPC, verbs seem to lose even that possibility, as (14) shows.14 In our view, this illustrates once more the cohesion of the construction. In fact, the formation of the deverbal noun seems to be blocked when the verb enters a VPC. One possible reason could be that the verb, becoming part of a new complex lexeme, does not inherit part of the information contained in its lexical entry, including the link to the correspondent derivational form that was associated with the verb alone. Consequently, the VPC will look for an alternative way of nominalization, and nominal infinitive, due to its applicability to any verbal entry, seems to be the best possible candidate.15

3.2 A semantic classification of Italian VPCs

From a semantic point of view, Italian VPCs function as single lexical items. A sign of this semantic cohesion is the metaphorical bleaching which many of these constructions have undergone, as well as the presence of completely idiomatic forms. We can identify three main semantic classes of VPCs: a. intensification: the particle may intensify a piece of information (i.e. the path or direction of motion) that is already contained in the verb itself, see the examples in (15):

(15) uscire fuori lit. to exit out ‘to go out’ entrare dentro lit. to enter in ‘to go in’ scappare via lit. to escape away ‘to escape’ b. direction marking: when added to generic (especially manner) verbs of motion, the particle may function as a direction marker:

14 Of course, one might come across some counterexamples in language use, and actually, a limited set of nominalizations such as la venuta giù ‘the coming down’ or l’andata giù ‘the going down’ sound weird but not completely unacceptable. However, we carried out an informal Google search and we found no results for these and also other examples. 15 On the nominal infinitive cf. Simone (2004). It is worth noting that in English the most common ways of nominalizing particle verbs are also the more “neutral”, i.e. conversion or substantivized in –ing. MULTI-WORD EXPRESSIONS BETWEEN SYNTAX AND THE LEXICON 155

(16) mettere giù lit. put down ‘to put down’ tirare su lit. pull up ‘to pull up(wards)’ saltare dentro lit. jump inside ‘to jump in’ c. metaphorical meaning: the VPC may assume a metaphorical non- compositional meaning. Metaphorical VPCs often derive from correspondent locative VPCs (see i and ii, examples 17 and 18), though we can also find a few completely idiomatic cases (iii) that do not have a locative counterpart, like (19). Thus, we can identify three subgroups:

i. transparent metaphors: the relationship between the locative VPC and its metaphorical counterpart is transparent and well recognizable:

(17) buttare via lit. throw away ‘to throw away’ and ‘to squander/waste’

ii. opaque metaphors: the relationship between the two VPCs is more opaque, but still somehow recognizable:

(18) portare avanti lit. bring forward ‘to bring forth’ and ‘to run (a business)’

iii. completely idiomatic forms: the VPC has a completely idiomatic meaning:

(19) fare fuori lit. do out ‘to kill’

This threefold classification can be easily related to the notions of encoding and decoding idioms, originally found in Makkai (1972) and then adopted by Fillmore, Kay and O’Connor (1988). The former are (more or less interpretable) idioms that the speaker pronounces with no decoding effort (e.g. answer the door), while the latter are expressions that cannot be interpreted by the listener without previous knowledge of what it means (e.g. kick the bucket). Hence, all decoding idioms are also encoding idioms, but not vice versa. Going back to VPCs, it is quite clear that transparent metaphors are only encoding idioms and completely idiomatic forms are 156 FRANCESCA MASINI also decoding idioms, whereas opaque metaphors stay somewhere in- between these two extremes.16 Besides this general semantic classification that underlines both the primary function of particles as locative modifiers and the metaphorical bleaching to which VPCs can be subject, it would appear that some particles have developed a particular aspectual meaning. Consider the example in (3), repeated here as (20a), and the corresponding sentence without the particle (20b).

(20) a. Luca ha lavato via la macchia Luca have.3sg wash.part.past away the stain ‘Luca removed the stain (by washing)’

b. Luca ha lavato la macchia Luca have.3sg wash.part.past the stain ‘Luca washed the stain’

In (20a) the particle via seems to convey a meaning of accomplishment: the focus is on the result of the washing process. By contrast, in (20b) the focus is on the process itself. Therefore, while in the latter example we do not know if Luca succeeded in removing the stain or not, in (20a) it is clear that the stain is not there anymore. Iacobini and Masini (2005) have recently showed that some Italian Ps have in fact acquired an aspectual, or rather actional, meaning. The authors claim that the actional contribution of Ps is primarily related to the kind of path they express, so that we obtain an iconic relation between the spatial movement denoted by the P (and by the whole VPC) and the event structure. This of course stresses once more the close connection that exists between V and P.

3.3 Interim conclusions

The previous sections were meant to illustrate that Italian VPCs are not a simple juxtaposition of a verb and a locative adverb. In particular, the syntactic tests in 3.1 aimed at showing that V and P are not to be considered separate constituents, but rather parts of a unique verbal

16 In a similar way, we could link our classification and the encoding-decoding notions to the distinction in Nunberg, Sag and Wasow (1994: 491) between idiomatically combining expressions (e.g. take advantage), “whose meanings – while conventional – are distributed among their parts”, and purely idiomatic phrases (e.g. kick the bucket), “which do not contribute their meanings to their components”. MULTI-WORD EXPRESSIONS BETWEEN SYNTAX AND THE LEXICON 157 construction where the V functions as the head and the P as a modifier, or ‘satellite’ in Talmy’s (1985) terminology. In fact, VPCs display extraordinary syntactic and morphological properties that set them apart from canonical V+AvdPs and V+PrepPs, i.e. restrictions concerning the insertion of material between V and P (cf. 3.1.1), the extraction of P through dislocation or topicalization (cf. 3.1.2), the coordination of Ps (3.1.3) and finally the nominalization of VPCs (3.1.4). The scale of acceptability that seems to characterize the examples given in 3.1 is mainly due to the ambiguous categorial nature of particles, which are closely related to both prepositions and adverbs. Moreover, ambiguity and graduality also extend to the construction itself, since it seems to be but one particular instantiation of a more general pattern that involves a verb and a relational element, as also suggested by Jansen (2004). Syntactic cohesion is also mirrored at the semantic level, which by the way contributes in a crucial way to the determination of the syntactic behaviour of VPCs (cf. examples 6–8). Ps mostly combine with motion verbs and their primary function is to express (or stress) the path of the motion. However, VPCs have developed metaphorical meanings that are different from the basic locative ones, and which seem to be derived from them (see point c in the semantic classification in 3.2). Besides, Ps seem to have undergone a further development from locative to actional meanings (cf. Iacobini and Masini 2005), which is also true of English phrasal verbs (cf. Brinton 1988). Finally, the semantic cohesion of these verbal constructions is also testified to by the fact that some metaphorical VPCs may happen to have synthetic quasi-synonyms, like for example allevare (synthetic form) – tirare su (VPC) ‘to bring up’. The main difference between the two is one of register: tirare su is in fact the colloquial variant of allevare. These considerations lead us to think that VPCs are a well-established and widespread pattern in present-day Italian and that they cannot be considered a purely syntactic juxtaposition of two elements, but rather conventionalized constructions placed somewhere in-between syntax proper and the lexicon. In other words, they are phrasal lexical items. A similar conclusion is reached by Booij (2002a,b) with respect to ‘separable complex verbs’ (SCV) in Dutch. Booij’s contributions are part of the lengthy discussion in the literature about the (morphological vs. syntactic) nature of particle verbs in Germanic languages. Of course, the 158 FRANCESCA MASINI discussion refers to an architecture of the language faculty that is modular in nature. The basic points of the debate are as follows.17 On the one hand, the separability of these combinations has been interpreted as a sign of their syntactic nature and has led some scholars to look for structural solutions, like the Small Clause analysis (cf. e.g. den Dikken 1995). On the other hand, their semantic/aspectual properties and their possibility to function as a base for morphological operations like nominalization (cf. section 3.1.4) or the formation of adjectives (e.g. break in → break-in-able, from Miller 1993: 132), were taken by other scholars as evidence for their word-like status.18 Booij (2002a,b) claims that we can do justice to the phrase-like and word-like properties of SCVs only if we consider them as intermediate structures between phrases and words. According to Booij, the creation of SCVs in Dutch can be referred to as a case of “periphrastic word formation”. This mechanism is an alternative (phrasal) means to derivational morphology for the creation of new lexical items, just as periphrastic constructions can realize functional categories like tense or aspect (cf. also Spencer 2001). In order to account for the semi-productivity of some SCVs, Booij proposes to consider them as “constructional idioms”, i.e. schemas consisting of a specific particle (door in the example below) followed by an open slot to be filled with a verb (21a).19 The schema is also associated with a specific semantics (21b).

(21) a. [door [x]V]V' b. ‘to go on V-ing’

As Booij declares, his proposal is perfectly in line with the basic tenets of Construction Grammar. As is well known, constructional approaches advocate for a non-strict division between syntax, morphology and the lexicon, between grammar and idiosyncratic facts. In this framework, language entirely consists of constructions that differ in size, complexity, and productivity. Consequently, a constructional approach can easily

17 For details cf. the introduction in Dehé et al. eds. (2002). 18 Of course, the whole discussion originates from the erroneous assumptions that words coincide with lexical items and that everything which is phrasal in nature should be accounted for in purely syntactic terms. 19 For the notion of constructional idiom, cf. also Jackendoff (1997, 2002) and Goldberg (1996). MULTI-WORD EXPRESSIONS BETWEEN SYNTAX AND THE LEXICON 159 account for the existence of lexical items that are phrasal in nature. Moreover, it can account both for idiomatic phrasal lexical items and for semi-productive, partially lexically-fixed patterns that give rise to new items. In our view, a similar line of reasoning can also apply to Italian VPCs. For some reason, the general syntactic pattern [[Verb][Locative adverb]] has become a means for the formation of new complex predicates in Italian, i.e. has acquired a new (more lexical) function. If this is true, we might want to know the reasons for the emergence of this new function. This leads us straight into the next section, in which we will investigate the diachronic development of Italian VPCs. As we will see, the findings of the diachronic analysis will turn out to support the view just outlined.

4. The diachronic analysis

4.1 The development of Italian VPCs: comparing hypotheses

As already noted in section 2, Italian VPCs, contrary to their Germanic counterparts, are a quite neglected research topic, the most relevant contribution being Simone (1997). In his work, Simone raises the question of the relationship between Standard Italian and Italian dialects and claims that VPCs do not seem to be an original Tuscan heritage, but rather might have entered the Italian language through North Italian dialects, where these constructions seem to be widely attested.20 Therefore, according to this view, which we will call the contact hypothesis, the development of VPCs in Standard Italian is due not to an autonomous development but to a “loan” from dialectal varieties. Let’s discuss this hypothesis in more detail. A noticeable feature of VPCs is that they especially seem to be used in the spoken language, due to the high frequency of the elements involved and to the small size and great “cognitive pregnancy” of the particle (cf.

20 Documentation of the occurrence of VPCs in North Italian dialects can be found in Schwarze (1985), where it is noticed that native speakers of Milan and Veneto dialects tend to realize verbs of motion by means of analytic forms, and especially in Vicario (1997), which is completely devoted to the study of VPCs in the Friulian dialect. According to my intuition as a native speaker of the Romagnolo dialect (spoken in the eastern part of the Emilia-Romagna region), VPCs are well attested in this variety. 160 FRANCESCA MASINI

Jansen 2004).21 Jansen (2004) also notes that the establishment of Standard Italian as a spoken language goes back only to the middle of the 20th century (cf. De Mauro et al. 1993). Before then, people would mainly use dialect. Jansen therefore proposes that, during this process of establishment, features that were typical of the dialects passed on to regional varieties and from there finally entered the national standard variety. This thesis (which the author calls the diamesic hypothesis) is claimed to complete the contact hypothesis. In conclusion, the contact hypothesis claims that Italian VPCs, which are typical of spoken language, are a product of the influence of North Italian dialects and entered the Italian language together with the (late) establishment of a standard spoken variety. Even though this hypothesis is quite appealing, the high integration of VPCs in the linguistic system of Italian led us to look for possible internal reasons for their development. It seems to us that a second hypothesis can be formulated, which we will call the typological-structural hypothesis. According to this hypothesis, the emergence of VPCs is the result of an autonomous development internal to the Italian language, mainly due to both typological and structural factors. The former factors consist in the consequences of the well-known typological change from Latin to Italian and can be summarized as follows:

ƒ the passage from a modifier-head (MH) to a head-modifier (HM) order, which entailed:

o the word order change from Latin (SOV) to Italian (SVO); o a general tendency towards postmodification;

ƒ the loss of the Latin case system and the emergence of prepositions for the expression of syntactic relations, which entailed a higher degree of analyticity of the language.

As for structural factors, the emergence of VPCs appears to be related to the progressive decline of the Latin system of verbal prefixes, in particular

21 We must say that, even though it is generally acknowledged that VPCs are colloquial in style and are preferably used in the spoken language, there is, to the best of my knowledge, no specific study on the issue. MULTI-WORD EXPRESSIONS BETWEEN SYNTAX AND THE LEXICON 161 their locative meanings, as noticed by Iacobini (2003, to appear). Of course, this decline is accompanied by the gradual loss of transparency of Latin prefixed forms (cf. Vicario 1997: 129 and Tekavčić 1972: §948.3, §1345). This view seems to be supported by a number of different sources. First, Tekavčić (1972: §1951–1952) states that Latin prefixed verbs such as circumdặre ‘to surround’ lost their productivity in neo-Latin languages and were superseded by “V+Adv compounds” such as the Italian forms pensarci su lit. think.about up ‘to think about’, mettere addosso lit. put on ‘to put on’, venire incontro lit. come towards ‘to come greet’, mandar giù lit. send down ‘to swallow’, far fuori lit. make out ‘to kill’, which were to become “the main substitute of prefixed verbs in Latin” (cf. also Tekavčić 1972: §948.3).22 Second, the same kind of hypothesis is put forward by Vicario (1997) to account for the rise of VPCs in the Friulian dialect (spoken in the Friuli-Venezia Giulia region, in the north-eastern part of Italy). According to the author, the development of these verbs is mainly due to the typological change from Latin to Friulian, which prompted a change of verbal modifiers from prefixes to postverbal particles. Finally, we may note that the same sort of change from a prefixal system to a postverbal modification system also occurred in the passage from Old English to Middle English, when the old verbal prefixes gradually disappeared and postverbal particles emerged, giving rise to the phrasal verbs of contemporary English (cf. Brinton 1988). In conclusion, the typological-structural hypothesis claims that the development of VPCs in Italian is mainly due to structural changes that took place in the passage from Latin to Italian. The basic reason for the emergence of VPCs would lie in the loss of the locative values of Latin verbal prefixes (Iacobini 2003, to appear), which led Italian to find an alternative means to prefixation for the expression of directional meanings. At the same time, the emergence of a postverbal element functioning as a modifier was made possible by the typological change from MH to HM and the greater analyticity of the Italian language.

22 A similar statement can be found in Durante (1981: 66), who argues that, as the SVO word order took on, prefixed verbs (like for example subeo ‘to pass under’) were no longer in accordance with the new structure of the language and thus stopped being productive. That was why prefixed verbs were superseded by analytic forms like e.g. ire susu(m) for subire ‘to go from below upwards’. Further still, the postverbal particle solution seems to have been the only way out. As noticed by one of the reviewers, the MH → HM change might have entailed a passage prefix → suffix, but this was in fact blocked by the rich inflectional morphology of the verb, which is entirely suffixal. 162 FRANCESCA MASINI

4.2 The diachronic analysis: methodology and findings

In order to find evidence of an autonomous development of VPCs internal to the Italian language, we decided to look for VPCs in Ancient Italian texts. The corpus used is Dante’s opera omnia, which dates back to the end of the 13th century and the beginning of the 14th century. In particular, we made use of tools such as the Enciclopedia Dantesca (henceforth ED) and the OVI textual database.23 The first step was identifying the set of locative adverbs to be found in Dantesque texts. Next, we examined the single entries of the adverbs in the ED in order to find their verbal occurrences, and finally we checked the type and number of occurrences in the OVI database.24 The search gave the following results: all in all, we found 121 VPCs (with 216 token occurrences), with a total of 68 verbal bases and 21 particles. The particles occurring with the highest number of bases are giù ‘down’ (occurring with 19 verbs), su ‘up’ (18) and via ‘away’ (10), while the bases occurring with the highest number of particles are generic verbs of motion such as andare ‘to go’ (occurring with 12 particles), mettere ‘to put’ (7), venire ‘to come’ (7), trarre ‘to pull/grab’ (4) and portare ‘to take/bring’ (4). Overall, the picture is similar to contemporary Italian, in which the most frequent bases are either simple verbs of motion (such as andare ‘to go’, mettere ‘to put’), or verbs with a generic meaning (e.g. essere ‘to be’). The most frequent VPCs found in the corpus are motion verbs in which the P either functions as a direction marker or as an intensifier: andare suso/su lit. go up ‘to go up(wards), to turn up(wards)’ (10 occurrences), tòrre via lit. remove away ‘to destroy, to eliminate, to exclude’ (8), andare via lit. go away ‘to go away, to leave’ (8), cadere giuso/giù lit. fall down ‘to fall down, to drop, to vanish’ (7), tornare suso/su lit. come back up ‘to come back up(wards), to rise’ (5), scendere giù/giùe/giuso lit. descend down ‘to go down(wards), to fall’ (5), montare

23 Available on the website: http://www.lib.uchicago.edu/efts/ARTFL/projects/OVI/. 24 The identification of VPCs was conducted in accordance with two basic syntactic criteria: obligatory V+P order and no insertion of material between V and P. As for VPCs separated by very light constituents (such as adverbs or ), they were included only if there was at least another non-discontinuous example of the same VPC. MULTI-WORD EXPRESSIONS BETWEEN SYNTAX AND THE LEXICON 163 su lit. mount up ‘to mount up, to go upwards’ (5). These VPCs are exemplified in Table 1.25

VPCs Glosses and Examples translations andare suso/su go up(wards) Purgatorio IV 114 ‘to go up’ Or va tu sù che se’ valente! ‘Then you go on up who are so sturdy!’ tòrre via remove away Inferno X 91–92 ‘to destroy’ Ma fu’ io solo, là dove sofferto / fu per ciascun di tòrre via Fiorenza ‘[B]ut I was alone there where all agreed to make an end of Florence’ andare via go away Purgatorio XIV 124 ‘to go away’ Ma va via, Tosco, omai ‘But, Tuscan, go on your way now’ cadere fall down Inferno XXV 121 giuso/giù ‘to fall down’ l’un si levò e l’altro cadde giuso ‘the one rose upright and the other fell down’ tornare suso/su come back up Inferno XXI 46 ‘to rise’ Quel s’attuffò, e tornò sù convolto ‘The sinner sank under and rose again, rump up’ scendere descend down Paradiso I 136–138 giù/giùe/giuso ‘to fall’ Non dei più ammirar, se bene stimo, / lo tuo salir, se non come d’un rivo / se d’alto monte scende giuso ad imo ‘You should not wonder more at your rising, if I deem aright, than at a stream that falls from a mountain top to the base’ montare su mount up Purgatorio XVI 49 ‘to mount up’ Per montar sù dirittamente vai. ‘For mounting up you are going aright.’ Table 1. The most frequent VPCs found in Dantesque texts.26 From a semantic point of view, the situation resembles the present-day classification (cf. section 3.2). Besides the purely locative meaning

25 As one may notice from the translations just given, a number of VPCs proved to be polysemic. In Table 1, the meaning refers to the specific example in the right column. 26 The free translations of the examples are taken from The Divine Comedy, translated by Charles Singleton. 164 FRANCESCA MASINI expressed by the majority of VPCs (e.g. correre via lit. run away ‘to run away’ or gittare giù lit. throw down ‘to throw down’), we found completely idiomatic cases (such as disporre giù lit. place down ‘to leave aside’ or porre giù lit. put down ‘to give up’) and polysemous examples conveying both a concrete and a more metaphorical meaning (such as andare dietro, that can mean ‘to follow’ – in a concrete or abstract sense – and ‘to aim at’). In conclusion, our study showed that VPCs were not only present, but also well-established in Dante’s writings. However, even if Dante is regarded as the father of the Italian language, one might argue that a great part of his works is not in prose and, therefore, data might be affected by metrical reasons. In order to counter such doubts, we also conducted an informal search of Boccaccio’s texts, which are mostly in prose. The study revealed that VPCs are well represented also in Boccaccio’s prose, thus demonstrating that these constructions were a well-known resource in Ancient Italian texts. By way of illustration, we provide in (22) the list of VPCs found in the Proemio, the introduction and the first novella of the Decameron.

(22) andare attorno lit. go around ‘to wander around’ entrare dentro lit. enter in ‘to go in’ mandare fuori lit. send out ‘to oust’ togliere via lit. remove away ‘to eliminate’ uscire fuori lit. exit out ‘to go out’

In general, we can say that the findings of the diachronic study shed new light on the origin of Italian VPCs and speak against the contact hypothesis as illustrated in section 4.1: VPCs were already attested in Ancient Italian texts and, therefore, are indeed part of the Tuscan heritage of the Italian language. In our view, the findings also speak in favour of the typological- structural hypothesis: the presence of VPCs in Dante and Boccaccio supports the view of the autonomous development of these constructions in Italian. Further still, the high number of VPCs expressing locative meanings might be an indication of the need to compensate for the loss of Latin prefixed verbs with locative meaning. Given that VPCs date back to the origins of the Italian language, i.e. 13th century Tuscan, another interesting issue arises. Although nowadays VPCs are generally considered to be a widespread analytic resource in Standard Italian as well as in Northern and Central dialects, they seem to be MULTI-WORD EXPRESSIONS BETWEEN SYNTAX AND THE LEXICON 165 less productive in Southern dialects.27 If this proved to be true, one might think that VPCs emerged in the northern-central part of Italy due to the contact with Germanic populations and languages in the Middle Ages, a hypothesis that we might call Germanic hypothesis. For instance, Rohlfs (1969: §916) suggests that the item via ‘away’ in Italian constructions such as buttar via lit. throw away ‘to throw away’ or portar via lit. take away ‘to take away’, could be a calque from German. However, Durante (1981: 89) argues against this hypothesis and says that ire viā was already attested in Prisciano (5th –6th century AD).28 Also Vicario (1997), in his book on Friulian VPCs, takes this hypothesis into consideration, due to the close contact of the Friuli region with the German speaking area. However, he claims that the influence of neighbouring languages such as German, Slovene, Croat and partly Hungarian, might have played a role in the establishment of VPCs, but cannot be considered the main cause of their emergence; rather, their development is mainly due to the typological change from Latin to Friulian. We might add that the typology of particle verbs to be found in the Central European area is quite different from the one we find in Italian. As we already noticed in section 2, German and Dutch present the ‘separable complex verb’ pattern,29 whereas Italian VPCs are much more similar to the English type.30 Besides, as already noticed, the similarities with English go even further: both English and Italian feature the emergence of postverbal particles and the decline of verbal prefixes, and in both languages particles seem to have developed aspectual values out of locative ones. This similarity, which cannot of course be due to contact reasons, is better understood if we trace it to a more general explanation in structural and typological terms. In conclusion, we do not deny that the influence of Germanic languages might have played a role in the establishment of these

27 There are no specific studies supporting this assumption, but only intuitions I collected from some native speakers of the Southern varieties. A survey of the presence of these constructions in the various dialects of the Italian peninsula is still missing, and certainly desirable. 28 A search by means of the OVI database revealed that examples of VPCs with via and few other particles can also be found in 13th century southern Italian texts such as the sonnets by the Sicilian poets. However, these data are not completely reliable, as the texts are quite restricted in number and they are all sonnets, and, moreover, many poets of the Sicilian school were not from Sicily. 29 The same pattern occurs in Hungarian as well. 30 I owe this observation to Geert Booij (p. c.). 166 FRANCESCA MASINI constructions, since there is too little evidence to decide whether this hypothesis is right or not. Incidentally, if this hypothesis proved to be right, it would not necessarily be in stark contrast to the typological-structural hypothesis supported here, since there might have been a concurrence of circumstances. However, we think that at present the typological-structural hypothesis is the best answer and that the typological and structural reasons illustrated above take priority over areal factors.

5. Theoretical implications

In this section we will try to give a theoretical account of VPCs. We propose that VPCs (and MWE in general) can be optimally accounted for in a Construction Grammar (CxG) framework.31 In particular, we will try to interpret the findings of the synchronic and diachronic analyses in the light of some of the basic tenets of the CxG theory. Firts of all, as we anticipated in the introduction, CxG claims that the ‘construction’ is the basic notion of linguistic knowledge and can be defined as a form/meaning pairing that can differ in size, complexity and abstractness. This entails that constructions include words as well as more abstract patterns. The consequence is that “[t]he totality of our knowledge of language is captured by a network of constructions: a ‘construct-i-con’” (Goldberg 2003: 219). Of course, this is a non-modular view of language that implies a non-strict division between syntax, morphology and the lexicon. Given this, the presence of phrasal lexical items, intended as intermediate constructions with phrasal structure and lexical meaning and function, is not only allowed, but even predicted by CxG. Secondly, CxG is a full-coverage theory, which means that it does not posit any principled distinction between idiomatic and fully productive phenomena. Again, this accounts for both completely idiomatic VPCs (like e.g. far fuori lit. make out ‘to kill’), and for semi-productive cases, such as

31 It is well-known that, nowadays, Construction Grammar is a cover term that hosts many different, though basically similar, approaches. Apart from the traditional current headed by Charles Fillmore and his colleagues (Fillmore, Kay and O’Connor 1988, Kay and Fillmore 1999), we should at least mention Adele Goldberg’s approach (Goldberg 1995, 2003), which derived from the Fillmorean tradition but has been greatly influenced by Lakoff’s studies, Cognitive Grammar (Langacker 1987, 2003) and Radical Construction Grammar (Croft 2001). We refer to Croft and Cruse (2004) for a survey of the different constructional approaches and to Östman and Fried (eds.) (2005) for an overview of the possible theoretical extensions of CxG. Here we will adopt the terms Construction Grammar and constructional approach in their broader acception and we will refer to the basic tenets presented in Goldberg (2003). MULTI-WORD EXPRESSIONS BETWEEN SYNTAX AND THE LEXICON 167 verb-particle patterns in which the P position is lexically specified and the V position is an open slot to be filled with a limited set of verbs that fulfil certain conditions (cf. Booij’s proposal for Dutch in section 3.3, example 21). Let us consider for example the case in (23). Here the particle via ‘away’ conveys a constant meaning of ‘accomplishment’,32 which contributes to the general semantics of the construction (23c), while the V position remains open, though subject to some restrictions (23b). Some exemplifications are given in (23d).

(23) a. [ V [via]P]VPC b. V = agentive verb of removing c. ‘to remove by V’ d. raschiare via lit. scrape away ‘to scrape away/off’ lavare via lit. wash away ‘to remove by washing’ grattare via lit. scrape away ‘to scrape away/off’

Thirdly, according to CxG, the constructicon is not a mere list of constructions, but an organized network whose elements are linked by inheritance hierarchies. This view allows for the existence of families of constructions,33 in which similar though different patterns are at the same time linked to one another (in order to obtain the highest level of generalization) and separated (in order to grasp their more specific or even idiosyncratic properties). This view, together with the gradual vision of the grammar-lexicon distinction, enables us to explain the high number of cases in which it was necessary to introduce a scale of acceptability in section 3.1. In a constructional perspective, VPCs and V+AdvP/PrepP combinations are set along a continuum that is identified by two extremes: completely free combinations vs. completely fixed combinations. The historical relationship between particles and adverbs/prepositions gives grounds for this assumption. Moreover Italian, as many other languages, presents a set of complex relational elements mainly formed by prepositions and/or adverbs (e.g. via da lit. away from ‘away from’, dietro a lit. behind to ‘behind’), which seem to be related to proper Ps. Jansen (2004), following Croft (2001), proposes that they might identify a

32 Iacobini and Masini (2005) analyse these examples as a case of telicization of the verbal base by means of the particle. Cf. also section 3.2. 33 Cf. e.g. the account of resultatives in English as a family of constructions proposed by Goldberg and Jackendoff (2004). 168 FRANCESCA MASINI

“syntactic space” (cf. note 9), i.e., in our terms, a family of constructions.34 The exact determination of this network of constructions constitutes an interesting issue for future research. Overall, we can say that the synchronic properties of VPCs as outlined in the paper can be easily interpreted in constructional terms. However, also the diachronic findings illustrated in section 4 seem to support this view. CxG has a single level of representation and only one notational system. This means that all linguistic units, be they syntactic or lexical, are represented in the same way, i.e. as constructions or instantiations of constructions.35 Of course, such a unified model allows for a comparison and a cooperation of these units on a functional basis. The development of VPCs in Italian seems to speak for this unified view of the traditional modules of the grammar, in that it entails a cooperation between morphological and phrasal means in the realization of a particular function or category. In fact, if the typological-structural hypothesis is true, VPCs took over the expression of locative meanings in the verbal system, thus compensating for the loss of Latin verbal prefixes (cf. also Iacobini and Masini 2005). Finally, the diachronic analysis tells us something more about the possible motivations behind the emergence of new patterns or constructions. In our case, it seems that VPCs emerged under the drive of a functional motivation, i.e. to fill a gap in the system. Of course, this was made possible by a set of typological and structural conditions, such as those illustrated in 4.1. In our view, originally it was exactly that functional motivation that gave the V+P sequence the necessary cohesive force to become a construction. Afterwards, as it acquired more and more independence, the construction allowed for the emergence of new metaphorical meanings, thus becoming a productive source for the formation of new phrasal lexical items in Italian.

34 Cf. also Masini (2005) for a tentative overview of some verbal constructions related to proper VPCs. 35 Of course, this does not mean that there is no difference between the word cat and general syntactic patterns like ditransitives. As we know, constructions differ in terms of size, complexity and level of abstractness. The crucial point here is that the word cat and the ditransitive construction are only the extremes of a scale. Therefore it is no longer necessary, and not even auspicable, to posit a principled and clear-cut distinction between what belongs to one module and what belongs to another one. MULTI-WORD EXPRESSIONS BETWEEN SYNTAX AND THE LEXICON 169

6. Conclusions

In this paper we focused on a particular kind of MWE that has been scarcely studied in the literature, i.e. VPCs in Italian. The structural analysis aimed at illustrating the peculiar syntactic and morphological behaviour of VPCs with respect to free V+PrepP or V+AdvP combinations. The semantic analysis showed that VPCs have both a primary, locative meaning and a series of metaphorical, idiomatic and even aspectual meanings. The presence of more abstract meanings, mainly derived from the locative ones, confirms the high cohesion of the pattern. The structural and semantic properties of VPCs were then taken as evidence of their status as phrasal lexical items. The second part of the paper dealt with the reasons for the development of VPCs in Italian. We compared three hypotheses: the contact hypothesis, the typological-structural hypothesis and the Germanic hypothesis. The diachronic analysis showed that VPCs were already attested in Ancient Italian texts, which speaks against the contact hypothesis and in favour of the typological-structural hypothesis. As regards the Germanic hypothesis, there is circumstantial evidence that this is not the right track. However, further research is certainly needed to understand if and to what extent Germanic influence might have played a role in the establishment of VPCs. The typological-structural hypothesis, which we support, claims that the basic reason for the development of VPCs might be traced back to the loss of locative values of Latin verbal prefixes, together with typological factors such as the passage from a modifier-head (Latin) to a head-modifier (Italian) order and the increase in analyticity of Italian due to the loss of cases and the emergence of prepositions. In this view, VPCs are an analytic resource that emerged in the passage from Latin to Italian to fill a gap in the system. As one would expect, VCPs then acquired more abstract meanings due to their high cohesion. Finally, we tried to give a theoretical interpretation of the results of the synchronic and diachronic analyses and we showed how the picture outlined can be easily explained and accounted for in terms of a constructional theory. First, we showed how the presence of phrasal lexical items is perfectly in line with the tenets of Construction Grammar, since this theory advocates for a non-strict separation between syntax and the lexicon. Second, the existence of VPCs of different degrees of productivity is easily 170 FRANCESCA MASINI accounted for, as Construction Grammar is equally committed to both idiomatic and fully productive phenomena. Third, the fact that VPCs seem to be related to other similar constructions can be explained by referring to the notion of a family of constructions. Moreover, we showed that the development of Italian VPCs supports a constructional view of language, in that it implies a cooperation between morphological and phrasal means in the realization of the functions of the language. Besides, we briefly addressed the question of the motivations that cause the formation of new constructions and concluded that, in our case, the drive was a functional motivation, i.e. the need to fill a gap in the system. In conclusion, it seems to us that a constructional approach to language like the one adopted here is not only an optimal solution for VPCs, but can also be easily applied to multi-word phenomena in general.

References

Ackerman, Farrell & Webelhuth, Gert (1998) A Theory of Predicates. Stanford: CSLI Publications. Blom, Corrien (2005) Complex Predicates in Dutch. Synchrony and Diachrony. PhD Dissertation, Vrije Universiteit Amsterdam. Bolinger, Dwight (1971) The Phrasal Verb in English. Cambridge: Harvard University Press. Booij, Geert (2002a) Constructional Idioms, Morphology and the Dutch Lexicon. Journal of Germanic Linguistics 14 (4): 301–329. —— (2002b) Separable Complex Verbs in Dutch: A Case of Periphrastic Word Formation. In Nicole Dehé et al. (eds.), Verb-Particle Explorations, pp. 21–41. Berlin/New York: Mouton de Gruyter. Booij, G. and van Marle, J. (eds.) (2003) Yearbook of Morphology 2003. Dordrecht: Kluwer. Brinton, Laurel J. (1988) The Development of English Aspectual Systems. Aspectualizers and Post-Verbal Particles. Cambridge: Cambridge University Press. Croft, William (2001) Radical Construction Grammar. Syntactic Theory in Typological Perspective. Oxford: Oxford University Press. Croft, William & Cruse, Alan (2004) . Cambridge: Cambridge University Press. Dehé, Nicole (2002) Particle Verbs in English. Syntax, Information Structure and Intonation. Amsterdam/Philadelphia: John Benjamins. Dehé, N. et al. (eds.) (2002) Verb-Particle Explorations. Berlin/New York: Mouton de Gruyter. MULTI-WORD EXPRESSIONS BETWEEN SYNTAX AND THE LEXICON 171

De Mauro, Tullio et al. (1993) Lessico di frequenza dell’italiano parlato. Milano: ETAS libri. Decameron by Giovanni Boccaccio, edited by Vittore Branca (1991). Torino: Einaudi. Dikken, Marcel den (1995) Particles: On the Syntax of Verb-Particle, Triadic, and Causative Constructions. Oxford: Oxford University Press. Dixon, Robert (1982) The Grammar of English Phrasal Verbs. Australian Journal of Linguistics 2: 1–42. Durante, Marcello (1981) Dal latino all’italiano moderno. Saggio di storia linguistica e culturale. Bologna: Zanichelli. Enciclopedia dantesca, 6 voll. Treccani. Fraser, Bruce (1976) The Verb-Particle Combination in English. New York/San Francisco/London: Academic Press. Fillmore, Charles J., Kay, Paul & O’Connor, Mary Catherine (1988) Regularity and Idiomaticity in Grammatical Constructions: the Case of Let Alone. Language 64 (3): 501–538. Gaeta, Livio (2002) Quando i verbi compaiono come nomi. Un saggio di Morfologia Naturale. Milano: Franco Angeli. Goldberg, Adele (1995) Constructions. A Construction Grammar Approach to Argument Structure. Chicago/London: The University of Chicago Press. —— (1996) Making One’s Way Through the Data. In Masayoshi Shibatani & Sandra Thompson (eds.), Grammatical Constructions: Their Form and Meaning, pp. 29– 53. Oxford: Clarendon Press. —— (2003) Constructions: a new theoretical approach to language. Trends in Cognitive Studies 7(5): 219–224. Goldberg, Adele & Jackendoff, Ray (2004) The English Resultative as a Family of Constructions. Language 80 (3): 532–568. Haiden, Martin (2002) Verb Particle Constructions. Aston University: Ms. (available at http://www.univ-lille3.fr/silex/equipe/haiden/particle/case_117_vepa.htm). Iacobini, Claudio (2003) L’indicazione di valori locativi a partire da basi verbali, Seminar held at the University Roma Tre (Department of Linguistics), 13 November 2003. —— (to appear) I verbi italiani come base di derivazione prefissale. In Maria Grossmann & Anna Maria Thornton (eds.), La formazione delle parole. Proceedings of the XXXVII International Congress of the SLI. Roma: Bulzoni. Iacobini, Claudio & Masini, Francesca (2005) Phrasal and Morphological Complex Predicates in Italian: a Semantic Analysis. Paper presented at the 5th Mediterranean Morphology Meeting, Frejus, 15–18 September 2005. Jackendoff, Ray (1997) The Architecture of the Language Faculty. Cambridge: The MIT Press. —— (2002) Foundations of Language. Brain, Meaning, Grammar, Evolution. Oxford: Oxford University Press. Jansen, Hanne (2002) Spatialpartikler. Forstudier om brugen af præpositioner og lokative adverbier på italiensk og dansk. Ny forskning i grammatik. Fællespublikation 9: 121–140. 172 FRANCESCA MASINI

—— (2004) La “particella spaziale” e il suo combinarsi con verbi di movimento nell’italiano contemporaneo. In Paolo D’Achille (ed.), Generi, architetture e forme testuali, pp. 129–144. Firenze: Franco Cesati editore. Kay, Paul & Fillmore, Charles (1999) Grammatical Constructions and Linguistic Generalizations: the What’s X doing Y? Construction. Language 75 (1): 1–33. Lambrecht, Knud (1994) Information Structure and Sentence Form. Cambridge: Cambridge University Press. Langacker, Ronald (1987) Foundations of Cognitive Grammar, Vol. 1: Theoretical Prerequisites. Stanford: Stanford University Press. —— (2003) Construction Grammars: Cognitive, Radical, and less so. Plenary paper, ICLC 8, Logroño (Spain), 25 June 1003. Lehmann, Christian (1995) Thoughts on Grammaticalization. München-Newcastle: Lincom Europa. Lüdeling, Anke (2001) On Particle Verbs and Similar Constructions in German. Stanford: CSLI Publications. Makkai, Adam (1972) Idiom Structure in English. The Hague: Mouton. Masini, Francesca (2005) Phrasal Verbs in Italian: Synchrony and Diachrony. Seminar held at the Free University of Amsterdam, 19 May 2005. Miller, Gary (1993) Complex Verb Formation. Amsterdam/Philadelphia: John Benjamins. Moon, Rosamund (1998) Fixed Expressions and Idioms in English: A Corpus-based Approach. Oxford: Clarendon Press. Müller, Stefan (2002) Complex Predicates: Verbal Complexes, Resultative Constructions, and Particle Verbs in German. Stanford: CSLI Publications. Nunberg, Geoffrey, Ivan A. Sag & Thomas Wasow (1994) Idioms. Language 70: 491– 538. OVI textual database: http://www.lib.uchicago.edu/efts/ARTFL/projects/OVI/. Rohlfs, Gerhard (1969) Grammatica storica della lingua italiana e dei suoi dialetti, II Morfologia and III Sintassi e formazione delle parole. Torino: Einaudi. Schwarze, Christoph (1985) “Uscire” e “andare fuori”: struttura sintattica e semantica lessicale. In Annalisa Franchi de Bellis & Leonardo M. Savoia (eds.), Sintassi e morfologia della lingua italiana d’uso. Teorie e applicazioni descrittive, pp. 355– 371. Roma: Bulzoni. Simone, Raffaele (1997) Esistono verbi sintagmatici in italiano? In Tullio De Mauro & Vincenzo Lo Cascio (eds.), Lessico e grammatica. Teorie linguistiche e applicazioni lessicografiche, pp. 155–170. Roma: Bulzoni. —— (2004) L’infinito nominale nel discorso. In Paolo D’Achille (ed.), Generi, architetture e forme testuali, pp. 73–96. Firenze: Franco Cesati editore. Spencer, Andrew (2001) The Paradigm-based Model of Morphosyntax. Transactions of the Philological Society 99 (2): 279–313. Stiebels, Barbara & Wunderlich, Dieter (1994) Morphology feeds Syntax. Linguistics 32: 919–968. Talmy, Leonard (1985) Lexicalization Patterns: Semantic Structure in Lexical Forms. In Timothy Shopen (eds.), Language Typology and Syntactic Description. Volume MULTI-WORD EXPRESSIONS BETWEEN SYNTAX AND THE LEXICON 173

III. Grammatical Categories and the Lexicon, pp. 57–149. Cambridge: Cambridge University Press. Tekavčić, Pavao (1972) Grammatica storica dell’italiano. II Morfosintassi and III Lessico. Bologna: Il Mulino. The Divine Comedy by Dante Alighieri, translated by Charles Singleton (1970–1975). Princeton: Princeton University Press. Toivonen, Ida (2003) Non-Projecting Words. A Case Study of Swedish Particles. Dordrecht: Kluwer. Venier, Federica (1996) I verbi sintagmatici. In Rovere Blumenthal & Christoph Schwarze (eds.), Lexikalische Analyse Romanischer Sprachen, pp. 149–156. Tübingen: Max Niemeyer Verlag. Vicario, Federico (1997) I verbi analitici in friulano. Milano: Franco Angeli. Wray, Alison (2002) Formulaic Language and the Lexicon. Cambridge: Cambridge University Press. Östman, J.-O. and Fried, M. (eds.) (2005) Construction Grammars. Cognitive Grounding and Theoretical Extensions. Amsterdam/Philadelphia: John Benjamins.

Contact information:

Francesca Masini Università Roma Tre Dipartimento di Linguistica Via Ostiense 236 – 00146 Roma - Italy e-mail: [email protected]

Ana Ibáñez Moreno

A Semantic Analysis of Causative Active Accomplishment Verbs of Movement

Abstract

This paper provides an exhaustive account of the semantic features and the logical structure of causative active accomplishment verbs of movement. Following the approach of Componential Analysis, and described within the theoretical framework of Role and Reference Grammar, these verbs are arranged into three subgroups according to a number of semantic variables related to their argument structures, and above all, to the types of locative arguments they govern. Emphasis is given to such arguments, since they reflect the relevant verb class. Additionally, new logical structures that account for the differences that exist among these verbs are proposed.

1. Introduction1

Verbs of movement are interesting for semantic analysis, since they have an inherent spatial dimension, which constitutes, together with a temporal dimension, the starting point of human cognition. Everything we do is located in a concrete point of time and space, and can be linguistically encoded. For this purpose some languages use locative expressions, that is, prepositional phrases and adverbial phrases of space and time, and different classes of verbs. One important class are verbs of motion which are described in this paper within the framework of functional grammar, specifically Role and Reference Grammar (RRG), as developed by Van Valin & LaPolla (1997). The semantic side of this theory consists of an analysis of the Logical Structure2 (LS) of verbs. Verbs are classified

1 I would like to thank my colleague, Christopher Phipps, and the referees of this paper, for their valuable help in the process of revision of this paper. Funding for this paper has been provided by the research project “Elaboración de un metadiccionario sintáctico de ingles antiguo: predicados verbales primitivos y derivados”, whose main researcher is Francisco Javier Martín Arista. 2 Note that in Role and Reference Grammar, LS refers to the semantic argument structure of the verb, not to its syntactic structure.

SKY Journal of Linguistics 18 (2005), 175–196 176 ANA IBÁÑEZ MORENO according to their Aktionsart3, that is, the internal temporal composition of an event. In this sense, the verbs under study here are classified as causative movement verbs, also called “induced motion verbs” (Jolly 1991, 1993). Examples are given in (1):

(1) a. Mary took the book to the library b. Sarah, remove the key from the box, please

Only transitive verbs which imply an induced motion of the UNDERGOER have been included in the corpus of analysis. Thus, verbs such as push, which may be transitive or intransitive, have been discarded. After an analysis of 6,500 causative movement verb samples, extracted from the British National Corpus, the results show that these verbs can be organized into two main groups according to their Aktionsart: a) causative accomplishment verbs and b) causative active accomplishment verbs. For an illustration of each group, see (2) below:

(2) a. Mary puts the plate on the table (causative accomplishment) b. Mary carries the bag (from the supermarket through the parking lot) to the car (causative active accomplishment)

The differences are explained below. The latter group has not been paid much attention in RRG until now, especially in relation to the semantic field of movement. Therefore, this paper focuses on the analysis of the verbs in this group and demonstrates that they correspond to an independent mode of action, by means of an analysis of their LS and their semantic features. Furthermore, a deep examination of each suggests three distinct subgroups. This study follows the semantic approach of Componential Analysis,4 according to which lexical decomposition provides a description of the meaning components of words (in this case, of movement verbs). These meaning components then make it possible to divide verbs into groups and to deal with their argument structure. Furthermore, the view is defended that the different semantic classes of verbs reflect different syntactic as well

3 The notion of Aktionsart is adopted from Vendler (1967[1957]), and is used as a basic criterion for identifying argument structure and predicate relations, in line with Van Valin & LaPolla (1997) and other linguists of the RRG school. 4 There are a few studies in Componential Analysis, related to the interaction between semantics and syntax, which are worth reviewing: Pinker (1989), Gropen et al. (1991), Levin (1993), and Levin & Rappaport (1995) among others. A SEMANTIC ANALYSIS OF CAUSATIVE ACTIVE ACCOMPLISHMENT VERBS 177 as semantic argument structures, which explains the direct relation of the type of verb with the type of PPs and Adverbial Phrases it takes. Focusing now on the semantic representation of the verbs under analysis, the LS below is the only one given in RRG for causative accomplishment movement verbs:

(3) [do’ (x, Ø)] CAUSE [BECOME be-LOC’ (z, y)]

The elements in bold type followed by a prime (do’ and be-LOC’) are part of the metalanguage used in the decomposition. That is, they are not terms related to any particular language. Thus, the same representation is used for all languages. This means that we would apply the same representation for put (English) and for poner (Spanish): do’ and be-LOC’, the LOC’ element varying depending on the PP used: put on/poner en: be-at’, but put down/poner debajo de: be-down’. The elements in capital letters, ‘CAUSE’ and ‘BECOME’ are the modifiers of the predicate in the LS, and they perform specific functions: ‘CAUSE’ indicates that the states of affairs (hereafter SoAs) is induced, that is, not spontaneous, which means that the Aktionsart is causative. ‘BECOME’ indicates that we are dealing with an accomplishment, i.e., with an event or a process that has an ending. In other words, the verb that is represented is [+telic]. The LS given in Van Valin and LaPolla (1997 : 109) for causative active accomplishment verbs is outlined below:

(4) [do’ (x, Ø) CAUSE [do’ (x, [predicate1’ (x, (y))]) & BECOME predicate2’ (z, x) or (y)]

However, an examination of the verbs mentioned and of their interrelation with spatial items shows that this LS is insufficient. For this reason, in this paper a LS for causative active accomplishment movement verbs is provided, which distinguishes these verbs from causative non-active accomplishment movement verbs. Hence, RRG’s system of semantic representation is used, although it has still not accounted for all predicates’ LSs. The incompleteness of the system of lexical representation in RRG is admitted by Van Valin & LaPolla (1997: 156) themselves:

Many aspects of the meaning of a verb [the specific requirements that a verb imposes on one or more of its arguments] would be represented in a full decomposition, but given that no such representation exists at present, they will have to be stipulated for the time being. 178 ANA IBÁÑEZ MORENO

Nevertheless, this system provides a basis for the analysis of motion and the LSs used fit the functional requirements of the dynamic phenomenon of language, so I have chosen it as the theoretical basis of this study. What is more, as Van Valin and LaPolla (1997) declare, this theory possesses typological, and consequently universal, validity, as is demonstrated by the hundreds of languages that have been successfully described by it. This is essential to provide any theoretical study with systematicity. Nevertheless, in order to make the LSs better suited for accounting for the compositionality of Aktionsart, I have introduced some new semantic variables in line with the Lexico-Grammar Model (Faber & Mairal 1999, Mairal & Faber 2002), where LSs are dealt with in terms of their lexical templates.

2. Causative accomplishment movement verbs versus causative active accomplishment movement verbs

Causative verbs of motion can be of three types: causative active accomplishment verbs, like guide in John guided the tourists to the museum, causative accomplishment verbs, like put in Mary has put the book on the table, and causative activity verbs, like walk in The woman walked the dog in the park. In this piece of research I focus on the first two, which share the feature [+telic]. Both groups are similar in that they are accomplishments. This means that they are “temporally extended (not instantaneous) changes of state leading to a terminal point” (Van Valin & LaPolla 1997: 92). In their LS, ‘BECOME’ in both types of verbs indicates that they are accomplishments, and thus telic. Therefore, is an inherent feature of such verbs, and the third argument is the one which carries the telicity feature: a GOAL argument. This is one of the reasons why I defend the view that all these verbs should be considered to have three 5 arguments, although the third (LOCATIVE: GOAL) is not always expressed. Focusing on Aktionsart, the mode of action in each group differs: active accomplishments invoke a SoA that goes from the original point of the UNDERGOER to the endpoint. This is due to their feature of dynamicity

5 In RRG causative – active and non-active – accomplishment verbs are considered to have either two or three arguments, depending on whether the GOAL argument is overtly expressed or not. For a discussion on this matter, see Ibáñez Moreno and Ortigosa Pastor (2004). A SEMANTIC ANALYSIS OF CAUSATIVE ACTIVE ACCOMPLISHMENT VERBS 179

[+dynamic], characteristic of active predicates, which non-active accomplishments lack. These verbs allow for the occurrence of multiple locational prepositions. That is, PATH and SOURCE PPs can be specified. This is due to their inherent nature as derivations of activity predicates, which are atelic, and therefore provide the verb with a complex combination of temporal and spatial indeterminacy on one hand and an end-point on the other. Nevertheless, note that the only inherent and necessary PP to complete their LS is the GOAL PP. Thus, depending on whether the GOAL is specified or not, the resulting Aktionsart is an activity or an active accomplishment (further explained below). On the other hand, accomplishment verbs only invoke that SoA at the endpoint. They express the resulting state of a non-active process of change. A change is understood as extended in time, but it is not a change that is evoked by these verbs; only a result is. In fact, as stated above, accomplishment verbs lack the feature [+dynamic], so the referring scope of the accomplishment only points to the endpoint, in time and in space, as is seen in (1.b). A useful test to distinguish active accomplishments from non-active ones is presented in Van Valin & LaPolla (1997: 95, originally from Dowty 1979): if it is an active accomplishment, adverbs such as vigorously or actively can be added:

(5) a. John carried the bags actively and vigorously. b.? John installed the TV aerial actively and vigorously.

Note that these are manner adverbs that imply dynamicity. In (5a) we have an active accomplishment, carry, which therefore admits such adverbs. This is not true of the verb in (5.b), which denotes an accomplishment that does not derive from an activity verb. As a result, it does not admit these adverbs. Thus, in “John installed the TV aerial carefully” we also have a manner adverb, but in this case the clause is correct because carefully does not entail dynamicity. Such an adverb, as well as others of the same type such as gently, cannot occur with stative verbs, e.g. *“He carefully knew the answer”. That is, they do not posses the [+static] feature. However, (non-active) accomplishment verbs admit them because they are [-static]. The subdivision of [-static] predicates into [+dynamic] and [-dynamic] constitutes the starting point of this piece of research. Thus, carefully and gently are non-statives, but they do not evoke the dynamic internal process of the state of affairs. 180 ANA IBÁÑEZ MORENO

3. Causative active accomplishment movement verbs

These verbs take GOAL PPs, but they also admit PATH PPs. If such motion verbs are followed by a GOAL PP, they are called active accomplishments. If they are not followed by a GOAL PP they are called activity verbs, even if they are followed by a PATH PP or by no PP at all. This Aktionsart interpretation is not possible for causative (non-active) accomplishment movement verbs. These do not allow for such alternation, called activity- active accomplishment alternation, by means of which an atelic verb becomes telic (Dowty 1979, Levin 1993). The causative active accomplishment verbs analysed here have been extracted from the Lexicon of Contemporary English (1985):

(6) guide, lead, conduct, escort, accompany, show, direct, draw, tow, usher, carry, bear, bring, fetch, transport, deliver, ship, dispatch, despatch, take.

All the verbs of motion in that dictionary which correspond to the LS of causative active accomplishments have been selected. Therefore, this corpus is extensive enough for the study carried out here, although there may be more verbs which fit this same criterion. All these verbs can be arranged into further subfields in order to account for the slight semantic differences that they display, and which are reflected in their LSs. This is important in the sense that the decomposition of their meaning is what allows us to fully provide their LS. Thus, once analysed into their semantic components, these verbs can be subdivided into three coherent subgroups, which are outlined in (7) below:

(7) a. guide, lead, conduct, escort, accompany, show, usher, direct, draw, tow b. carry, bear, transport, ship, despatch/dispatch c. bring, fetch, deliver, take

The verbs in (7.c) are exceptional in one aspect: the PP they take is a SOURCE PP, as is seen in 3.3. In the case of causative accomplishment verbs of movement, GOAL PPs and SOURCE PPs can function as AAJs.6

6 Argument-adjuncts is the name given in Role and Reference Grammar to those expressions which stand in the middle between being arguments, that is, being essential for the LS of the clause, and being adjuncts, that is, being additional elements that modify the clause as a whole and that are additional elements, which do not affect the verbal LS. Locative arguments are included within this group, due to the fact that the A SEMANTIC ANALYSIS OF CAUSATIVE ACTIVE ACCOMPLISHMENT VERBS 181

PATH PPs are adjuncts, since their being present in the clause does not affect the LS of the clause. They provide additional information to the clause as a whole, as can be seen in John guided us through the town to our hotel. Thus, they are represented differently:

(8) a. Causative accomplishment: Sarah put the book on the table [do’ (Sararh, Ø)] CAUSE [BECOME be-at’ (book, table)]

b. Causative active accomplishment: John guided us through the town to our hotel (through.town’ [Clausal LS])

As can be observed, the PATH AAJ in this case falls out of the core structure of the clause, that is, it falls within the more general scope of the sentence. In this paper it is assumed that the verb has a maximum number of three semantic arguments, contrary to Jolly (1993: 285), who states that semantically all motion accomplishment verbs have a maximum valence of four. This implies that the PATH PP is considered as a potential argument. Indeed, she provides the following LS:

(9) Rita walked from the school through the park to the store [walk’ (Rita)])] CAUSE [[BECOME NOT be-at’ (school, Rita)] & BECOME be- via’ (park, Rita)] & BECOME be-at’ (store, Rita)]]

However, from my point of view PATH PPs are not arguments of the clause, but adjuncts, and so this must be reflected in the semantic representation. They must be located outside the core. This is consistent with the syntactic structure of the English language and with the semantic organization of verbs. This hypothesis is demonstrated by the fact that if in (8.b) we elide

locative word that introduces them modifies the Aktionsart and the meaning of the verb as a whole. That is, it is not the same to say Put the book on the table and Put the book down the table. In this paper argument-adjuncts are given the same semantic and syntactic status as arguments, since they are in fact essential for the LS. Thus, we cannot just say *Put the book. There is an argument lacking here. In the case of causative accomplishment verbs of movement, both GOAL PPs and SOURCE PPs are argument-adjuncts. PATH PPs are adjuncts, since their presence in the clause does not affect the LS. They provide additional information to the clause as a whole, as can be seen in John guided us through the town to our hotel. 182 ANA IBÁÑEZ MORENO

7 the PATH PP, the clause makes sense. Finally, note that the LS of the verb guide has not been given here, since it is described and explained in the following subsection.

3.1 Verbs of accompaniment

In this subsection I analyse the verbs in (7.a). They all have a common semantic parameter that distinguishes them from the rest: their central meaning is based on the action of accompanying, which implies that both the ACTOR and the UNDERGOER carry out the action of moving from one location to another together, because the ACTOR voluntarily goes with the UNDERGOER. This, in terms of lexical template variables, would be represented within the predicate LS as:

(10) [DO (x (go.with (y) (x))]

Here, the element DO indicates agency, that is, that the action has been carried out intentionally. This implies that the ACTOR is always an AGENT. However, we may find exceptions, as in “The map guided us to the castle”. However, in my view, the verb guide is here used in a metaphorical sense, so the argument the map is qualified with the features of agency and , even if ordinary maps are [-animate] and consequently cannot be 8 AGENTS. Thus, the ACTOR and the UNDERGOER have to be animate beings. Evidently, metaphorical senses can be given to any of them, so we can find examples such as the following one, extracted from the British National Corpus:

(11) AMY 30 In the main they draw attention to the changes of attitude and behaviour over the years.

Draw is widely used in this non-literal sense. In fact, draw attention to has become a fixed expression, in such a way that attention always collocates with draw in that sense. Although this question does not concern us here,

7 The view held here, nonetheless, deviates from the view of one of the reviewers of this paper, who was in favor of Jolly’s (1993: 285) representation of semantic structures and arguments. 8 There are, nonetheless, counterarguments to this idea, as one of the reviewers of this paper has proposed. To mention one, a map could guide without being physically present (i.e. from memory). A SEMANTIC ANALYSIS OF CAUSATIVE ACTIVE ACCOMPLISHMENT VERBS 183 idioms and metaphors have the same LS as literal meanings. We focus on the latter since, as the example shows, figurative uses only affect the semantic restrictions of thematic relations. Thus, the NP attention in the example does not refer to an animate being, and the PP to the changes of attitude and behaviour over the years does not refer to a location. Both the THEME and the LOCATIVE arguments are expressed metaphorically through an orientational metaphor. However, figurative senses have to be represented as literal senses in order to reach some degree of systematicity. To continue with the LS in (10), go.with’ has been applied as a substitute for the general predicate do’. Thus, go.with’ is my proposal for the decomposition of the meaning of verbs of accompaniment. Let us now consider another example:

(12) B77 456 Friendly staff guide casual visitors to their first encounter with LOGO.

This is a prototypical example of causative active accomplishment 9 movement verbs. It shows the three maximally possible arguments: ACTOR, UNDERGOER and LOCATION, if we focus on macrorole assignment (ACTOR, UNDERGOER) and primitive abstract predicates (LOCATION), and AGENT, THEME and GOAL, if we specify their correspondent microroles. We can also observe how the semantic restrictions of such predicates operate on the thematic roles of the arguments: the AGENT and the THEME roles are selected so that both contain the semantic component [+animate]. The semantic LS of such verbs is presented in (13):

(13) DO (x, [do’ (x, Ø)] CAUSE [do’(y,[go’(y)]) & BECOME be-at’ (z, y)]

First, we have the activity part: DO (x, [do’ (x, Ø)]. The use of ‘DO (x,…)’ indicates that the ACTOR is an AGENT. That is, it is used to represent verbs with lexicalized agency, which can never be used to express an action carried out unintentionally. There are, nonetheless, cases in which guide takes a non-agentive ACTOR, such as in The signs guided visitors to the viewing area. In such cases, we would have to omit ‘DO’. However, since here I deal with the prototypical sense of guide and of verbs of accompaniment in general, I have included it in the general LS. Second, we find a problem that needs to be solved: the LS as a whole, as presented in (13), does not specify the type of action that is carried out –

9 The concept of prototypicality applied here is based on Taylor (1989). 184 ANA IBÁÑEZ MORENO that is, it does not make a distinction between the different kinds of active accomplishment verbs of movement. With such an LS, one cannot extract any differences from any of the verbs in (7). This means that it is incomplete. Thus, in order to specify the type of action, the following can be used:

(14) a. DO (x, [do’ (x, [guide’ (x, y)])) b. DO (x, [do’ (x, [go.with’ (x, y)]))

The LS in (14.a) is not abstract enough, since guide is not a primitive verb. Unfortunately, RRG does not provide the lexical decomposition of all verbs, but it constitutes an excellent starting point for developing it. This has been the way followed by the Lexico-Grammar Model, which is still being developed. In this paper a provisional representation such as the one in (14.a) will be used for the verbs under study. Nonetheless, I have already proposed an alternative representation, rendered in (14.b), for verbs of accompaniment, but the other subgroups remain to be decomposed. Hence, since my intention is to explain the LS that is applicable to all causative active accomplishment verbs, I will use predicates such as guide’. Guide has two arguments syntactically and three semantically. Its complete LS is presented in (15):

(15) DO (x, [do’ (x, [guide’ (x, (y)])) CAUSE [do’(y [go’(y)]) & BECOME be-at’ (z, y)]

Since the first part of this LS has already been explained above, I now focus on the second part, which is headed by the causative element ‘CAUSE’. As one can note, there are two LSs combined by the element ‘&’. This is not the case in non-active accomplishments, since only one process is taking place. In this case, the [+dynamic] feature of the verb, characteristic of activity predicates, calls for including the LS [do’(z,[go’(z)], which indicates the fact that active accomplishments also contain this feature. This is reasonable, since active accomplishments derive from activity predicates. Besides, such a feature is combined with the feature [+telic], which activity verbs lack, but which is present in all accomplishments of movement, whether active or not: [BECOME be-at’ (z, y)]. Let us focus on another example, where we have a PATH PP, and the GOAL AAJ is left unspecified: A SEMANTIC ANALYSIS OF CAUSATIVE ACTIVE ACCOMPLISHMENT VERBS 185

(16) ABC 112 Dolphins have suddenly appeared to help fight off a shark attack, and they frequently guide boats through storms or treacherous waters.

In this example we find dolphins as AGENT, boats as THEME, and the GOAL is left unexpressed. The PATH PP is through storms or treacherous waters, and there is no GOAL PP, which is, as has been explained, the only type of PP that can function as an AAJ. This raises an important issue: these verbs, which are inherently active, when appearing without a GOAL PP – that is, either with no directional PP or with a LOCATION PP which is not of the GOAL type – turn out to be interpreted as atelic, since the feature of telicity is expressed through the GOAL argument. Thus, they are not active accomplishment verbs any more, but just activity verbs, as is the case with the example above. Any of the other types of PP perform an additional role of modifiers, and appear in the periphery in the constituent projection if they modify the whole clause. In this case, they modify the second argument (UNDERGOER), so their scope is narrower. This is the most frequent case with directional PPs, while temporal PPs tend to modify the whole clause, perhaps because time is a more abstract concept, and the more abstract a concept is the more scope it has in a clause. The PP through storms or treacherous waters is not regarded as a PATH PP by all authors. That is, a preposition like through could make a sentence telic as a complement of guide for Tenny (1999), who has pointed out that telicity depends on measuring out, which may be different with different verbs. According to this proposal, with a predicate like ‘guide X through Y’ the sentence is telic if X gets guided all the way through Y, which is the measure in this case. Thus, if the boats in the example above manage to get out of storms or treacherous waters, the proposition is telic. However, I do not agree with this view, because it completely ascribes the interpretation of the sentence to the context. In other words, the fact that the PP through storms or treacherous waters is telic or not depends on the context in which such sentence is uttered. This view of semantic representations is too broad, since it subordinates sentences to the multiple interpretations they can have according to their context. As a result, the minimum requirements of systematicity needed in order to explain and describe linguistic phenomena are lacking. In my view, we must not neglect all these possible interpretations, but we must try to categorize lexical items in terms of prototypicality, in order to systematize and simplify the process of semantic analysis. Thus, the prototypical role of the preposition through is PATH, even if there may be sentences in which, 186 ANA IBÁÑEZ MORENO depending on the context, speakers may use it as a GOAL preposition. I would propose that through is used with a less prototypical meaning in such cases.

3.2 Verbs of transporting

In this subsection I deal with the group of causative active accomplishment verbs illustrated above in (7.b). The verbs in (7.c) and (7.a) share similar features, but they have distinctive properties. Most features that characterize these verbs come from their Aktionsart, as has been discussed in the previous subsection. Therefore, only the aspects that distinguish them from the other verbs are considered here. The main difference is their core meaning, which in this case is related to the action of removing items from their original location and taking them to another location. These verbs present the same semantic parameters as verbs of accompaniment, according to which the AGENT moves the THEME. However, in this case the second argument of the verbal predicate is restricted to refer to inanimate objects, as least prototypically. Hence, sentences such as the one below are not ungrammatical:

(17) KS8 921 They are highways for dead souls moving into paradise and often carry emigrants deep into the heart of a new country.

In this sentence, the second argument has the semantic feature [+human]. As with draw above, this is not a central feature of such verbs. Thus, the UNDERGOER of the verbs in this group is prototypically [+object], therefore [-animate]. Other features are less prototypical. In order to prove this, we have carried out a study of the frequency of occurrence with which each type of UNDERGOER occurs in the sentences of our corpus, and this is the result obtained:

VERB OF TRANSPORTING (CARRY) animate: 16.7% inanimate: 83.3% VERB OF ACCOMPANIMENT (LEAD) animate: 82.9% inanimate: 17.1%

Table 1. Frequency rates with which carry and lead take animate or inanimate objects.

A SEMANTIC ANALYSIS OF CAUSATIVE ACTIVE ACCOMPLISHMENT VERBS 187

As can be observed, there are clear tendencies in these verbs. Apart from this predicate restriction, there are no other differences between these two groups of verbs, so we now turn to the following ones.

3.3 Special verbs of transporting

This group of verbs consists of bring and fetch, take and deliver. These verbs differ from the rest of transporting verbs for a number of reasons. First, I deal with bring and fetch. Bring and fetch also encode an induced change of location of a THEME by an AGENT, as do the other transporting verbs, but the orientation of the path through which the action is carried out is different. With these verbs, the AGENT goes from the front to the back, where the RECIPIENT is situated. In order to grasp this difference, consider the following representations:

Figure 1. Representation of verbs of transporting Note that in all these figures, the AGENT controls the THEME throughout the PATH. This fact is represented by the unbroken line of the arrows. In terms of SoAs, it can be observed in these schemes that the unmarked orientation of the action starts in the position of the AGENT and goes from the AGENT to the GOAL, while in the case of bring and fetch this process is differently oriented, as the arrow shows. Another important difference is that the GOAL argument can also be the RECIPIENT, which means that this participant can show the semantic features [+animate] or [- animate], while this is not allowed for the other verbs. Thus, it is possible to say Bring that book to me, while it is not correct to say *Transport this package to me. Further evidence is found if we change the order of the THEME and the GOAL: 188 ANA IBÁÑEZ MORENO

(18) Bring me that book *Transport me that book

Thus, the alternation in bring and fetch is of the RECIPIENT/GOAL argument kind, which implies that they allow for two different Aktionsart interpretations, and depending on which one is being expressed they will have a different LS. Let us see how this alternation works:

Figure 2. Alternation of verbs like bring

Note first that the SOURCE can coincide with the AGENT, which is the case with bring, as can be seen in the following examples in (19), in which an illustration of figure 2 is presented:

(19) a. KBW 15163 You bring him to me and then b. KBW 18406 and then we’ll get on the bus about half past four and bring him to your house.

In (19.a), bring is used as a transfer verb, and therefore me is the RECIPIENT, which stands for an animate being and which receives something or, as in this case, someone. In (19.b), on the other hand, bring is used as a motion verb, and so your house is the GOAL argument, which is similar to the recipient but with the [- animate] feature. Thus, in the case of these verbs, as well as in that of the following two, there is an alternation which depends on the feature [+/- animate] of the third participant role. This variation affects the verbal LS, as is shown in (20):

(20) a. LS for (17.a): [do’ (you, Ø)] CAUSE [BECOME have’ (me, him)] b. LS for (17.b): [do’ (we, Ø)] CAUSE [BECOME be-at’ (our house, him)]

Pinker (1989) also supports this alternation. He has suggested that ‘have’ is a primitive verb and that there is a difference between BECOME be-at’ A SEMANTIC ANALYSIS OF CAUSATIVE ACTIVE ACCOMPLISHMENT VERBS 189 and BECOME have’. On the other hand, one could argue (like Jackendoff 1991, for example) that this is only a difference between spatial and uses of the verb. In any case, in my view, what cannot be denied is that there is a difference between both interpretations, and that this difference should be reflected in the semantic representation of the verb. The orientation of fetch is different from that of bring, as can be seen by comparing the illustrations in figure 1. Fetch implies a double way orientation, so that the AGENT starts the action of movement in the same position as the RECIPIENT/GOAL, goes along the path until it arrives at the THEME, and then comes back to the RECIPIENT/GOAL. The RECIPIENT can coincide with the AGENT, as in (21.a), or not, as in (21.b):

(21) a. A74 178 She looks a bit cold, so I go and fetch the blanket off my bed and wrap it round her shoulders. b. Go and fetch me a candle/Go and fetch a candle for me

In (21.b) the second argument is the RECIPIENT and the LOCATION does not need to be specified.10 In this case, the LS for fetch is the same as the one given for bring above, and it must be concluded that it is possible to combine all the types of participant roles in one sentence, as is shown below:

(22) (You) go and fetch me a candle from the kitchen

In this case, none of these participants – the RECIPIENT me and the SOURCE from the kitchen – seem to be essential for the correct understanding of the mode of action. However, for an adequate establishment of the verbal LS, the following question arises: if the LS in (23.a) below is applied, then me is predicative and therefore peripheral. That is, it does not function as a

10 This discussion is problematic if we focus on the use of the structure for + NP, which is a benefactive argument, which in this case functions as a RECIPIENT. In Jolly (1993: 301) this structure is represented semantically by the purposive LS. In Van Valin & LaPolla (1997: 382-384), a discussion is brought up on this idea. It is stated that such a LS is predominant over the thematic relation labels when it comes to accounting for the differential behavior of NPs, in the sense that a benefactive thematic relation does not fully account for all the uses of for. In this paper, nonetheless, I am concerned with verbal LSs and their thematic relations. I do not intend to provide a general LS for all the components of the sentences rendered as examples, due to the scope of this piece of research. Thus, for our purposes here, just the thematic relation of for that concerns the verb fetch is shown and developed. 190 ANA IBÁÑEZ MORENO verbal argument. On the other hand, if the LS in (23.b) is used, then from the kitchen is the peripheral, non-core element:11

(23) a. [do’ (you, Ø) CAUSE [BECOME NOT be-at’ (kitchen, candle)]] b. [do’ (you, Ø) CAUSE [BECOME have’ (me, candle)]]

Both LSs are correct, but only one of them can be applied at a time, since the LS only admits three arguments maximum, which in this case would be you, candle, and either me or kitchen. This is coherent with the mode of action (i.e. Aktionsart) which encodes a SoA. In a SoA we have a maximum of three main elements that affect its composition, and the rest of the elements modify it as a whole. One may argue that these two LSs can be combined into one, as is done in Van Valin & LaPolla (1997 : 109) through the use of the element ‘&’, meaning ‘then’. The example provided by them is repeated below:

(24) Tom took the knife from the prisoner [do’ (Tom, Ø)] CAUSE [BECOME NOT have’ (prisoner, knife) & BECOME have’ (Tom, knife)]

In this case, this complex LS can be applied because we have three arguments: Tom, knife and prisoner, which is the maximum number allowed in the LS of movement verbs, as already stated. In the case of (22), we have more than three arguments, so we have to choose between the two potential AAJs (me or kitchen) and leave the unselected one as an Adjunct, that is, as a phrase providing additional information to the clause. The hypothesis that only three arguments are allowed in the LS is in accord with the verbal Aktionsart; that is, it is faithful to the representation and categorisation of SoAs. Thus, we cannot combine (22) in one LS through the element ‘&’. This possibility is only available for sentences with three maximally possible arguments. Two LSs in (23) correspond to bring. In fact, they can be applied to take and to deliver too. The question is, therefore, which LS represents one

11 In relation to footnote 10 about for with the verb fetch, it must be noted that one of the reviewers of this paper was of the opinion that the benefactive argument formed by for plus a NP is represented semantically by the purposive LS. Hence, according to him/her (23.b) is incorrect; it should be (23.a) plus the purposive representation. However, I still defend the view that both LSs in (23) are equally valid and independent, and that the AAJs they represent have the same level of semantic validity, so this must be reflected in the LS of the predicate. A SEMANTIC ANALYSIS OF CAUSATIVE ACTIVE ACCOMPLISHMENT VERBS 191

of these verbs when both the SOURCE and the RECIPIENT participant roles appear in the same clause. In this case, since no logical explanation seems to be to the point – due to the impossibility of explaining why one of these roles is more essential to the clause than the other – a pragmatic point of view is adopted in order to find the correct answer. From this point of view, it could be stated that the closer the argument is to the verbal predicate the more important for the clause it is. Thus, if the nominal phrase (hereafter NP) me appears before the PP from the kitchen, we may suppose that such a NP is given primacy, and the LS in (23.b) will be the adequate one. In the same way, if the PP from the kitchen is placed before the PP to me – the RECIPIENT takes oblique case when it is located farther away from the verb than the theme – then the LS in (23.a) will be accurate:

(25) a. LS in (22.a): You fetch the candle from the kitchen for me. b. LS in (22.b): You fetch me the candle from the kitchen.

Off and from are SOURCE prepositions that form SOURCE AAJs, whose LSs show the following abstract predicate:

(26) …CAUSE [BECOME NOT be-LOC’ (y, x)]

This SOURCE PP is used because the action is oriented in the opposite direction to GOAL PPs from the point of view of the speaker. In clauses with this LS, the speaker is situated in the point where the action finishes, and therefore the point of departure is specified. In those verbs with which the GOAL PP is expressed, the hearers and the speaker are, on the contrary, at the point where the action starts, so the final destination has to be expressed. Take, deliver, bring and fetch are interesting verbs because they present semantic alternations, not only in the same way as the rest, but also in that they are polysemous verbs. They have been included in this group in their sense of ‘transporting’, since they have the same features that have just been presented. However, they also have the opposite meaning of ‘extracting’, so they should also be included within the group of verbs which have this central meaning, which are non-active accomplishment verbs. Thus, they correspond to two different senses, and therefore to two different semantic representations. Let us see an illustration of how take works in both senses:

192 ANA IBÁÑEZ MORENO

(27) a. ARK 2070 Now, take me to the office’; Horowitz ordered the guard. DO (you, [do’ (you, [take’ (you, (me)])) CAUSE [do’ (me [go’(me)]) & BECOME be-at’ (office, me)] b. Mary took the book from Peter. [do’ (Mary, Ø)] CAUSE [BECOME NOT have’ (Peter, book) & BECOME have’ (Mary, book)]

The question remains of what happens if a SOURCE PP is added to the clause in (25.a), or if the GOAL PP is omitted. In that case, we would be dealing with sentences like the following ones:

(28) a. Now, take me from the station to the office b. Now, take me from the station

The result is that the SOURCE PP has a different role in (28.a) than in (28.b). In (28.a) it is just a modifier, as is the case with any PP that is not a GOAL PP with active accomplishment verbs. The GOAL PP is more central in the sentence than the SOURCE PP, which cannot function as an AAJ unless it comes with the verbal predicate alone. If this is the case, as in (28.b), take changes its meaning and Aktionsart, and it becomes a non-active accomplishment verb. In that case, only one PP, either SOURCE or GOAL, can function as an AAJ with the verbs of this list, and it can only appear alone, with no other directional PP. These are the only verbs that admit this alternation. Deliver constitutes an outstanding case, since it has the same features as take, in the sense that it is a two-way verb (polysemous), and at the same time it behaves like the verbs in the previous subsection, bring and fetch. Let us see some examples:

(29) a. CFF 557 And yet the author of so many adulatory sermons preached before James I cries out in these private prayers, `Deliver me from making Gods of Kings!’ (You) deliver me from + [ subordinate clause] [do’ (you, Ø) CAUSE [BECOME NOT be-at’ ([subordinate clause], me)]]

b. EEB 873 Please deliver the goods to our Manchester office. (You) deliver the goods to our Manchester office [do’ (you, [deliver’ (you, (goods))]) CAUSE [do’(goods [go’(goods)]) & BECOME be-at’ (Manchester office, goods)]

A SEMANTIC ANALYSIS OF CAUSATIVE ACTIVE ACCOMPLISHMENT VERBS 193

c. CRM 8206 Typical stony asteroids, however, deliver the bulk of their energy near an altitude of 9km, and this is consistent with observations. Asteroids deliver their energy near 9km high [do’ (asteroids, [deliver’ (asteroids, (energy))]) CAUSE [do’(energy [go’(energy)]) & BECOME be-near’ (9kms high, energy)]

As can be seen in (29.a), deliver can have the LS of an accomplishment verb, with the feature ‘BECOME NOT’ in it, which explains why it allows a SOURCE PP to appear alone in the clause, as an AAJ. In that case, similarly to take, deliver is related to the sense of extracting, and the orientation of the process is opposite to that in (29.b) and (29.c), where deliver has the same meaning and Aktionsart as the rest of active accomplishment movement verbs. What is more, deliver shows further features that relate it to the rest of such verbs. Consider the sentences below:

(30) G4X 1436 Who then deliver that team brief, in the same way to all members of staff. GV2 2241 It sounds as though Maurin was paying him to keep Barbara there and she was paying him to run errands --; deliver the note to you and the photograph to Nice Matin --; and to turn a blind eye when she went out.’;

In these cases the third argument qualifies as a RECIPIENT, and it remains to say, in relation to the LS that corresponds to this Aktionsart, that atypical verbs of transporting do not behave like active accomplishment verbs, but like accomplishment verbs. In order to prove this, let us try adding the specification of the PATH:

(31) *He delivered the note through the park to you.

In this case, deliver is not a motion verb, but a verb of transfer of possession – the LS of transfer verbs is outlined in (23.b). In that mode of action there is a telicity implied and no dynamicity is possible, since the verb presents facts at the end-point of the process.

4. Final remarks

In this study I have presented and analysed English verbs of causative active accomplishment movement within the RRG framework, and I have arranged them into a typology by taking into account a number of semantic criteria related to their argument structure and, more specifically, to the 194 ANA IBÁÑEZ MORENO kind of locative AAJs they take. In this respect, these verbs have been divided into three subtypes, which have been described in terms of their semantic features and LSs. I have made an attempt to formalize some verbs of movement within the RRG notation, so that my analysis shows the possibilities of the RRG approach to account for expressions of motion. I have proposed an LS that accounts for how active accomplishment verbs of movement differ from (non-active) accomplishment ones. The proposed LSs of both groups of verbs are repeated below:

(32) a. Causative accomplishment movement verbs [do’ (x, Ø)] CAUSE [BECOME be-LOC’ (z, y)] b. Causative active accomplishment movement verbs [do’ (x, [predicate’ (x, (y)])) CAUSE [do’(y [go’(y)]) & BECOME be-at’ (z, y)] b1. Causative active accomplishment movement verbs (verbs of accompaniment) DO (x, [do’ (x, [predicate’ (x, (y)])) CAUSE [do’(y [go’(y)]) & BECOME be-at’ (z, y)]

The LS in (33.a) is the only one put forward by Van Valin and LaPolla (1997) and others working within the RRG framework (cf. Jolly 1991) in order to account for the compositionality of induced movement. I have proposed a LS as in (33.b) to account for all those verbs that cannot be represented by (33.a), that is, causative accomplishment verbs of movement that contain the [+dynamic] feature due to their being derived from activity verbs. Non-active causative accomplishments derive from stative verbs, so they lack dynamicity. As a result, the LS in (33.a) is insufficient to provide all their semantic components and the argument structure of active verbs. As regards the LS in (33.b1), it differs from the more general one in (33.b) because it contains the element ‘DO (x,[…])’, which encodes agency. Although it is true that many active accomplishment verbs of movement take AGENTS as their first argument, I prefer not to include this element as a “must” in their LS, since there are many instances in which that is not the case. That is, the ACTOR is not necessarily an AGENT. A close study of all these verbs shows the importance that their arguments play, specially the third one expressing location. This locative argument is treated as an AAJ in RRG. The use of this term is helpful in the sense that it permits us to distinguish those clausal constructions which affect the mode of action of the verb from those that do not. It is important A SEMANTIC ANALYSIS OF CAUSATIVE ACTIVE ACCOMPLISHMENT VERBS 195 to remark that this study reflects how the LS of the verb determines the clausal structure and the Aktionsart interpretation. This implies that for an adequate study of locational expressions, verbs of movement have to be granted primary importance, and vice versa. Finally, due to the limitations of this piece of research, the LSs of the verbs presented are not yet fully decomposed, so further research will be devoted to this question.

References

Tenny, C. and Pustejovsky, J. (eds.) (1999) Events as grammatical objects: the converging perspectives of and syntax. Stanford (California): CSLI. Dowty, David (1979) Word Meaning and Montague Grammar. Dordrecht: Reidel. Faber, Pamela & Mairal, Ricardo (1999) Constructing a Lexicon of English Verbs. Berlin: Mouton de Gruyter. Gropen, Jess, Pinker, Steven, Hollander, Michelle & Goldberg, Richard, (1991) Affectedness and Direct Objects: The Role of Lexical Semantics in the Acquisition of Verb Argument Structure. Cognition 41: 153–195 Ibáñez Moreno, Ana & Ortigosa Pastor, Ana (2004) A Semantic Typology of Causative Accomplishment Movement Verbs and their Argument-adjuncts in Role and Reference Grammar. Atlantis 26.2: 35–49. Ibáñez Moreno, Ana (2004) A Semantic Analysis of Locative Expressions: The problem of Intransitivity in Prepositions. Taiwan Journal of Linguistics 2.2: 27– 45. Jackendoff, Ray (1991) Semantic structures. Cambridge, Massachusetts: MIT Press. Jolly, Julia (1991) Prepositional Analysis within the Framework of Role and Reference Grammar. New Cork: Peter Lang. —— (1993) Preposition assignment in English. In Robert D. Van Valin (ed.) Advances in Role and Reference Grammar, pp. 275–310. Amsterdam/Philadelphia: John Benjamins. Levin, B. and Pinker, S. (eds.) (1991) Lexical and Conceptual semantics. Oxford: Blackwell. Levin, Beth (1993) English Verb Classes and Alternations: A Preliminary Investigation. Chicago: Chicago University Press. Levin, Beth & Rappaport Hovav, Maika (1995) Unaccusativity: at the Syntax-lexical Semantics Interface. Cambridge: MIT Press. MacArthur, Tom (1985) Lexicon of Contemporary English. England: Longman. Mairal, Ricardo &Faber, Pamela (2002) Functional Grammar and Lexical Templates. In Ricardo Mairal Usón & María Jesús Pérez Quintero (eds.) New Perspectives on Predicate Argument Structure in Functional Grammar, pp. 39–94. Berlin and New York: Mouton de Gruyter. Pinker, Steven (1989) Learnability and Cognition. Cambridge, M.A: MIT Press. 196 ANA IBÁÑEZ MORENO

Mairal Usón, R. and Pérez Quintero, M. J. (eds.) (2002) New Perspectives on Predicate Argument Structure in Functional Grammar. Berlin and New York: Mouton de Gruyter. Van Valin, Robert. D. (ed.) (1993) Advances in Role and Reference Grammar. Amsterdam and Philadelphia: John Benjamins. Van Valin, Robert D. & LaPolla, Randy (1997) Syntax: Structure, Meaning and Function. Cambridge: Cambridge University Press. Vendler, Zed (1967 [1957]) Linguistics in Philosophy. Ithaca: Cornell University Press.

Contact information:

Ana Ibáñez Moreno University of La Rioja Departamento de Filologías Modernas San José de Calasanz s/n 26004, Logroño (La Rioja) Spain e-mail: [email protected] Maarit Niemelä

Voiced Direct Reported Speech in Conversational Storytelling: Sequential Patterns of Stance Taking

Abstract

The paper discusses how voiced direct reported speech (DRS) can display a speaker stance in talk-in-interaction. The analysis concentrates on the sequential organization of matching voiced DRS and the types of actions that this practice performs within three storytelling sequences. Two of the extracts exemplify voiced DRS sequences in displaying a shared stance, whereas one extract illustrates a contrary case of recipient- initiated voiced DRS in displaying a disaligning stance. The analysis of three examples indicates that a voiced DRS utterance that displays a shared stance appears to be a preferred response to voiced DRS. A sequence of subsequent matching voiced DRS utterances therefore appears to be an orderly phenomenon in interaction and a sequentially relevant practice of stance taking.

1. Introduction

In this paper I examine the role of voiced direct reported speech (DRS) as an interactional practice of stance taking.1 Voicing (Couper-Kuhlen 1998) here refers to the speaker’s attempt to produce a type of voice quality that mimics the voice quality of the original speaker or attempts to replicate a certain affective way of speaking (e.g. angry, whiney, exaggeratedly polite, dumb-sounding or stilted). This paper is part of a current research interest in the linguistics of affect in conversation.2 DRS has been widely studied

1 The present study builds upon my MA thesis (Niemelä 2003) and has been much revised based on the observations, comments and feedback from Elise Kärkkäinen, Jane Stuart-Smith, Tiina Keisanen, Mirka Rauniomaa and Pentti Haddington and three anonymous reviewers, to all of whom I owe my deepest gratitude. 2 Research on affect in conversation is being/has been conducted by, for example, Ochs 1989 (affect) , Sandlund 2004 (emotions), Local & Walker in the project “Phonetic and interactional features of attitude in everyday conversation” (http://www- users.york.ac.uk/~lang4/emotion-proposal.html) and Sorjonen et al. in the project “Language and Social Action” which focuses on the affiliative and disaffiliative

SKY Journal of Linguistics 18 (2005), 197–221 198 MAARIT NIEMELÄ within the field of linguistics; however, it has not been investigated in more detail as a linguistic resource of displaying affect in talk-in-interaction.3 DRS is a recurrent and potentially an affectively charged interactional resource and therefore an appropriate object of study within the framework of stance taking. My aim is to examine and describe the sequential environment in which voiced DRS appears in storytelling. Furthermore I will expand on the interactional functions of this device and its role as a resource in the participants’ co-construction of stance. In what follows I will suggest that the use of voiced DRS not only does stance taking but more importantly can induce similar conduct in other conversationalists; thus, other participants subsequently recycle DRS with matching voicing in the following turns. Voiced DRS is recycled by the recipient in naturally occurring talk-in-interaction, which can be interpreted as conversationalists’ strive for solidarity and a shared stance. Recent studies have also established that stance taking is an interactional rather than an individual process (cf. DuBois 2003a, Haddington 2004, 2005, Keisanen under review, Kärkkäinen 2003, Rauniomaa under review). As Kärkkäinen et al. (2004: 49) suggest, “stance taking is an activity that is essentially dialogic, interactive and intersubjective in nature: it is oriented to by co-participants who frequently engage in jointly constructing and negotiating their stances across turns”. In DuBois’ (2000b) understanding, speakers frequently construct a stance by building on, modifying, and aligning or disaligning with the immediately co-present stance of a dialogic partner. The present study deals with DRS as one practice of stance taking, which will be analyzed as it occurs in naturally occurring conversational interaction. DRS can be characterized as a common feature in conversational storytelling. Several studies have pointed out that DRS is often used at the climax of stories (see Drew 1998, Clift 2000, Golato 2000). According to Mayes (1990: 326), DRS in narrative is used to perform the task of highlighting and dramatizing the key elements. Thus, the locus of DRS in a piece of conversation, especially in storytelling, emphasizes its function as what Clark and Gerric (1990: 764) call ‘highlighting demonstration’ that depicts the original conversation. Holt (2000: 435) also states that people activities across languages and institutional settings (http://www.uta.fi/laitokset/sosio/ project/affiliation/index.htm). 3 Sandlund’s (2004) study on emotion in institutional academic talk and Besnier’s (1993) study on reported speech and affect on Nuculaelea Atoll deal with this phenomenon. VOICED DIRECT REPORTED SPEECH IN CONVERSATIONAL STORYTELLING 199 often use DRS when they tell amusing stories. DRS can be used as a punch line in a humorous story or at the peak of a complaint sequence (see Drew 1998, Haakana forthcoming). Moreover, reported speech has a dualistic function in talk-in- interaction. According to Besnier (1993: 161) reported speech, on one hand, is “the representation of linguistic actions” and, on the other hand, “commentaries about these actions”. Due to the dualistic nature of DRS, investigating it provides direct access to stance taking processes. Narrators not only report what has been said but also use DRS to do something in this interaction here and now: the teller chooses what to present and how, and through these choices constructs a specific representation of the original situation and seeks a specific response from the recipient. This paper shows that DRS is often not only designed to convey affective meaning but often responded to by recipients it by producing a similar DRS utterance. Mathis and Yule (1994: 63) refer to this as echoing an attitude, “which appears to function as a means of both linguistic and psychological convergence”. It therefore appears legitimate to say that the elicited response is an act of stance taking. I look at the sequential organization of matching voiced DRS and the types of actions that this practice performs within three storytelling sequences. The analysis of conversation sequences in this study will be based on what Sacks (1984: 25) refers to as “actual occurrences in their actual sequence”, which displays the interactional relevance of voiced DRS in naturally occurring talk-in-interaction. Hence I approach voiced DRS mainly within the framework of Conversation Analysis (see Goodwin 1991, Hutchby & Wooffit 1998), but I will also adopt the point of view of Interactional Linguistics (Ford et al. 2003, Selting & Couper-Kuhlen 2001), which aims at describing recurrent linguistic resources as means of accomplishing interactional functions. This paper is organised as follows. Chapter 2 will deal with the data used in the present study in more detail. Chapters 3, 4 and 5 move on to present the analyses of the use of DRS in interactional stance taking. Chapters 3 and 4 exemplify voiced DRS sequences in displaying a shared stance, whereas chapter 5 introduces a contrary case of recipient-initiated voiced DRS in displaying a disaligning stance. 200 MAARIT NIEMELÄ

2. Data

My data consists of recordings of naturally-occurring conversational interactions.4 I will first look at two extracts5 taken from the Santa Barbara Corpus of Spoken American English (SBCSAE). The first extract (59 sec) comes from a tape called ‘Raging Bureaucracy’. This 20-minute- conversation was recorded in Santa Fe, New Mexico, in September 1990. The primary conversationalists on the tape are three sisters, who are all in their twenties. The second extract (43 sec) is taken from a tape called ‘Cuz’. The conversationalists are Alina and Lenore, two cousins in their mid-thirties. The lively 30-minute-conversation was recorded in Los Angeles, California, in November 1988. I will then move on to analyse a third extract (47 sec) taken from a tape recorded in Glasgow in 2004. The two informants on the tape are Carrie and Stephen, who are close friends in their late twenties and early thirties, and who at the time of the recording had not seen each other in a while. The 45-minute-conversation was recorded in Carrie’s flat.

3. Voiced DRS in collaborative storytelling: displaying a shared stance

Example 16 presents a sequence of DRS utterances with matching voicing in negotiating a shared stance, thus emphasizing the fundamental role of responses and participation in joint stance taking. In the following example Carolyn, Sharon and Kathy discuss the reliability and competence of substitute teachers (tutors). Carolyn categorizes them into passive ones, who do not seem to care about teaching at all, and active ones, who have just started working as substitute teachers and are therefore still enthusiastic but not very experienced.

4 In my data there are speakers of both American English and Scottish English. There may be differences between the conversation practices of speakers of different varieties of English; however, my assumption is that there are no major differences in the use of voiced DRS. Thus, my intention is not to evaluate only speakers of American and Scottish English but to draw conclusions of the use of DRS among English-speaking comversationalists in general. 5 For full transcription key, see Appendix I 6 Examples (1) and (2) are originally transcribed by using the so-called Discourse Transcription (DT) conventions (Du Bois et al. 1993) which has been used to transcribe all the SBCSAE tapes. VOICED DIRECT REPORTED SPEECH IN CONVERSATIONAL STORYTELLING 201

Example (1): Square root of pi (Raging Bureaucracy 345.58)

1 CAROLYN: you have to tell (.) students, 2 never get_ 3 (0.3) 4 CAROLYN: give a tutor anything. 5 (0.5) 6 CAROLYN: anything. 7 (.) ever. 8 (0.2) 9 CAROLYN: .hh not even homework. 10 they wouldn’t let us give homework to hh the high school 11 (.)tut[ors ]. 12 SHARON: [(I know)]. 13 CAROLYN: because they just, 14 they lose it. 15 SHARON: (.).hhh 16 CAROLYN: they don’t care. 17 (0.2) 18 SHARON: yeah. 19 (0.2) 20 SHARON: w[ell ], 21 CAROLYN: [°they do]n’t give a shit about anything°. 22 (0.2) 23 SHARON: I know. 24 CAROLYN: hhh[((noise))] 25 SHARON: [I I ], 26 (.)I hadn’t thought of that at th- at that point ( ). 27 @°let me out of he[:re° ]/@.((anxious)) 28 CAROLYN: [((clearing throat))un]less they’ve just 29 begun, ((1.9 paper rustling in the background)) 30 CAROLYN: subbing,= 31 SHARON: =well this [woman ], 32 CAROLYN: [because th]en they’re like, 33 SHARON: hhhh 34 1→ CAROLYN: .hhh @ what are we gonna learn today. ((perky)) 35 SHARON: uh [hu ] 36 1→ CAROLYN: [.hhh see] I’m originally a (.) PE teacher, 37 [but ], 38 SHARON: [uh huh] huh huh huh huh [hu: ] 39 KATHY: [ah hah hah hah hah hah hah ] 40 1→ CAROLYN: [I guess I can teach ma[th /@]?= 41 SHARON: [uh 42 huh huh huh huh hu:] 43 CAROLYN: =[you know, 44 they] come in with that attitude, 45 and they go, 46 1→ (.)((thump)) @↑I’ve always↑ wanted to teach math. 47 SHARON: uh huh 48 1→ CAROLYN: .hh now, ((perky)) 49 1→ .hh [what are we on/@]? 50 2→ SHARON: [@it’s gonna be ] (.) [great/@]. ((perky)) 51 CAROLYN: [eh heh ] 52 .hh you know, 202 MAARIT NIEMELÄ

53 and you’re saying, 54 (.)uh, 55 CAROLYN: we’re o::n, 56 ((thump)) 57 (0.3) 58 CAROLYN: uh:::, 59 (0.2) 60 KATHY: eh heh heh= 61 CAROLYN: =? 62 [uh, 63 KATHY: [eh heh heh heh heh] 64 CAROLYN: and they look at ] you and they go, 65 (1.9) 66 3→ CAROLYN: @the what/@? ((dumb-sounding)) 67 SHARON: (.)e heh heh heh heh heh heh heh 68 CAROLYN: °so they don’t know what the hell they’re doing/°. 69 3→ [@uhm ], 70 4→ SHARON: [@pi=]? ((dumb-sounding)) 71 3→ CAROLYN: .hhh [why don’t we go out and run some laps/@]. 72 4→ SHARON: [I didn’t bring any pie with me today ]. 73 CAROLYN: you [know, 74 KATHY: [eh heh heh] 75 CAROLYN: they]_ 76 they don’t know what they’[re doing ]? 77 4→ SHARON: [I don’t think] we can have pie 78 ‘til lunch, kids/@.= ((dumb-sounding)) 79 KATHY: =eh heh [heh heh] 80 CAROLYN: [yeah. 81 se]e? 82 it’s like, 83 (0.6) 84 CAROLYN: distorted. 85 (0.5) 86 CAROLYN: °they don’t know what they’re doing/°.

Carolyn produces an explicit evaluation of the passive substitute teachers in her ‘own voice’, her natural speaking voice (i.e. relatively low in pitch, evenly paced and moderate in loudness) on line 21 they don’t give a shit about anything. She moves on to introduce the category of the active substitute teachers unless they’ve just begun subbing, because then they’re like on lines 28, 30 and 32. Those lines provide a setup for DRS: she escalates her affective opinion by introducing the first voiced DRS utterances on the following lines 34–40 when she utters what are we gonna learn today, see I’m originally a PE [Physical education] teacher, but I guess I can teach math. The overall pitch of the utterance is notably high, which is a crucial element in depicting this exaggerated perky and enthusiastic voicing. To achieve such voicing, Carolyn also modifies other prosodic parameters: the tempo is faster, the rhythm more ‘jerky’, and the volume increased. She also produces a tighter articulation by producing VOICED DIRECT REPORTED SPEECH IN CONVERSATIONAL STORYTELLING 203 intense velars, alveolar stops and plosives compared to her prior turns. Voicing is enhanced with a slightly inquisitive tone of voice, which, conjoined with the phonetic design and semantic content, creates the impression that the reported speaker is enthusiastic but not very bright. Carolyn then elaborates on the affective frame of the voiced DRS utterance and the entire sequence, by explicitly topicalizing the stance displayed by the previous animation of the substitute teacher’s attitude, they come in with that attitude on line 44. After this she engages in a further enactment of the active substitute teacher by continuing the use of the same voicing in the next DRS utterance on lines 46–49, I’ve always wanted to teach math. now, what are we on. The following fundamental frequency analyses7 are presented to give support to the auditory analysis of the data. Carolyn’s utterances on lines 52–61 (see Figure 1) represent a sample of her ‘own voice’, that is, her natural speaking voice. We can see that her pitch varies between 181 Hz and 237 Hz, making the pitch span 4.6 semitones. However, when she uses voicing in a DRS utterance on line 46 (see Figure 2), the fundamental frequency shows that the overall pitch is considerably higher, between 271 and 666 Hz and the pitch span thus is 15.5 semitones. Pitch alone is not the variable feature for doing DRS, as high pitch can certainly be part of the speaker’s ‘own voice’ as well. But clustered together with changes in tempo, volume and rhythm, speaker’s use of high pitch achieves the desired voice quality (i.e. perky and enthusiastic voicing of a substitute teacher).

7 In Figure 1–4, the topline and the baseline of the Praat pitch tracts represent the highest point and the lowest point of the speaker’s actual pitch range that has been calculated by measuring a representative sample of approximately two minutes of speech. The logarithmic mark (the horizontal line) across the tract represents the median of the speaker’s range. 204 MAARIT NIEMELÄ

668

500

300

200

137

You know and you're sayiuhng we're o:::n (pause) square root of pi

0 5.7951 Time (s) Figure 1. F0, example (1), lines 52–55, 61

668

500

300

200

137

I've always wanted to teach ma::th

0 1.43383 Time (s) Figure 2. F0, example (1), line 46

Further, Sharon responds to Carolyn’s lengthy DRS sequences by producing a matching second DRS utterance when she utters it’s gonna be great on line 50. Voiced DRS utterances are often produced in overlap with other participants’ turns, which complicates the measurement of fundamental frequency. Here Sharon’s utterance overlaps with Carolyn’s utterance on line 49, what are we on?, which makes it impossible to draw a pitch curve of Sharon’s voiced DRS. However, based on an auditory analysis one can argue that Sharon’s voicing follows along the lines of Carolyn’s voicing on line 46–49. In fact, her voicing is significantly similar to the voicing that Carolyn uses to report the speech of an enthusiastic VOICED DIRECT REPORTED SPEECH IN CONVERSATIONAL STORYTELLING 205 substitute teacher. It depicts vigour and liveliness as her pitch rises to a slightly higher level compared to her normal voice. Also the tempo of her speech is faster and she utters the sentence in a fast and snappy manner. Sharon thus engages in what Szczepek (2001) calls “prosodic matching of voice quality”, in which co-conversationalists imitate each other by producing matching voice quality patterns and thus link their own talk to the talk of the preceding speaker. Sharon’s voiced DRS emerges from and ultimately rests upon the stance displayed in the prior talk, thus complementing it with the second sample of the enthusiastic substitute teacher in a sequence of matching voiced DRS utterances. The phenomenon described above is repeated soon afterwards in the conversation. By contributing a third and a fourth voiced DRS sequence of the same type, the conversationalists continue to participate in negotiating and displaying a shared stance on the issue of substitute teachers. Carolyn produces an imaginary dialog between herself and the generic substitute teacher when she utters the square root of pi on line 61, in answer to the substitute teacher’s question on line 49, what are we on. When she produces her ‘own speech’ she uses more or less her normal voice; in other words, she does not apply any particular voicing that would depict an affective animation. She further continues the dialog on line 66 by producing a question the what? as a response to the DRS on lines 55–61, where she herself was the reported speaker. The voiced DRS sequence is designed to invite laughter: as Haakana (2002: 217) suggests, alterations in voicing and voice pattern can be used in such a way. The co- conversationalists indeed respond by laughing on lines 67, 74 and 79, thus signalling that they understand and, at least seemingly, agree with the current speaker. Previous research (e.g. Jefferson 1984, Glenn 1989) shows that, with the exception of troubles-telling, recipients’ laughter displays affiliation and a shared stance with the prior speaker. On line 68 Carolyn concludes the dialog by summing up her stance explicitly in words, so they don’t know what the hell they’re doing. This evaluation is nevertheless followed by further enactment of the substitute teacher’s bewilderment, or a voiced DRS utterance by Carolyn on lines 69 and 71, uh, why don’t we go out and run some laps?. Sharon now also responds to Carolyn’s preceding evaluation sequence on lines 70 and 72, when she produces a fourth subsequent DRS utterance pi? I didn’t bring any pie with me today. The semantic content of her utterance ridicules the generic substitute teacher: it displays a scenario in which the substitute teacher cannot tell the difference between the square root of pi and lunch 206 MAARIT NIEMELÄ pie, for instance. Carolyn’s utterance on the other hand displays the substitute teacher as someone who is unfamiliar with the notion of square root of pi and attempts to hide it by rather taking the children out to run laps. Both of these utterances of reported speech are produced by applying similar voicing to that in the earlier DRS utterances of this extract. They can be treated as extensions to Carolyn’s previously enacted dialog, and the fact that they are produced simultaneously in almost the same way supports the claim that voiced DRS is produced as a response to a certain kind of stance. Next, after her further enactment of the substitute teacher’s erratic behavior, Carolyn repeats her earlier explicit evaluation by stating that they don’t know what they’re doing on line 76. At the same time, Sharon produces one more DRS utterance on lines 77–78, I don’t think we can have pie till lunch, kids, which is a continuation of her own DRS on lines 70 and 72 and of the voicing introduced in the previous utterances. The semantic content and voicing of Sharon’s DRS utterance create a humorous impact which seems to indicate that it is designed to invite laughter from the recipients as well as to participate in the stance taking process. But possibly because it overlapped with Carolyn’s DRS, it was lost and was not adequately reacted to by the other conversationalists. Sharon’s utterance only receives a reaction from Kathy, who appears to be laughing at both rivals, the overlapping utterances by Carolyn and Sharon, at the same time. Therefore, Sharon’s utterance does not get individual attention from the recipients. Sharon then reintroduces a display of her stance by reformulating and repeating her voiced DRS utterance, thus providing the participants with a further possibility to acknowledge the utterance (Rauniomaa in progress). Sharon’s turn, which is marked here as part of the fourth DRS sequence, is one more subsequent reported speech utterance that contributes to the establishment of a shared stance between participants. All Sharon’s voiced DRS utterances are what Mathis and Yule (1994: 74) call ‘zero quotatives’,8 i.e. they are not preceded by any quotatives (e.g. ‘she said’ or ‘she goes’) and there is thus no attributed speaker. Carolyn’s previous turns provide the relevance for Sharon’s DRS. Kathy responds with laughter on line 79 and Carolyn utters yeah, see? on lines 80–81, which can be seen as a display of affirmation that the

8 The term ‘free-standing quotation’ introduced by Clark and Gerric (1990: 772) also refers to the phenomenon of reported speech that has no initial markers such as she said or she goes, the only indicator of DRS thus being prosody. VOICED DIRECT REPORTED SPEECH IN CONVERSATIONAL STORYTELLING 207 conversationalists have successfully participated in acting out and have now reached a mutual stance. She produces a follow-up to the series of matching voiced DRS sequences by yet again explicating the stance in it’s like, distorted, they don’t know what they’re doing on lines 82–86, which brings the overall sequence to an end. Carolyn thus sums up the shared stance that was formulated by using voiced DRS. In this first example, Carolyn is the primary storyteller whereas Sharon acts as a co-teller who produces prosodically similar extensions to the current speaker’s reported speech utterances. Carolyn sets up the context for the upcoming sequence of (matching) voiced DRS utterances and finally it is Carolyn who also produces a concluding follow-up. Thus, Carolyn and Sharon collaboratively construct a shared stance towards substitute teachers.

4. Voiced DRS in second stories: resonating with prior stance

In the previous example, the recipient contribution is in line with the projected storyline. The storyteller and the recipient only momentarily discontinue the telling to evaluate aspects of the story by producing matching voiced DRS sequences, and thus establish a shared stance. The following example 2 provides a similar practice in terms of collaborative stance taking. Nevertheless, the sequential environment of the use of voiced DRS is different from that in the first example: here the first story prompts a second story (Sacks 1992: 765, Tainio 1996: 16, Routarinne 1997: 152, Norrick 2000: 112, Arminen 2004: 320) from the initial recipient. I apply Du Bois’ (2003b) notion of resonance, or the process of activating “potential affinity across instances of dialogic language use” throughout the analysis of voiced DRS in the second story. Example 2 thus provides a similar practice to that of the first example in terms of collaborative stance taking, i.e. the recipient’s use of voiced DRS in the second story sustains and contributes to a stance displayed in the first story. In example 2 Alina first tells about an older guy and a young chick who are a couple. Lenore then tells a second story about a tape she has made on which guys talk about younger women. Both stories ultimately deal with the relationship between an older man and a younger woman, or the attitudes of men towards younger women. 208 MAARIT NIEMELÄ

Example (2): Real pill (Cuz 200.73)

1 ALINA: the friend that was there with them, 2 is this older guy with this young chick. 3 (1.2) 4 ALINA: @and she was like a real pill, ((whiny)) 5 you know, 6 LENORE: [hm mh mh mh mh mh mh ((nasal sound laughter)) 7 gh gh gh] ((glottal laughter)) 8 ALINA: [she’s sitting there, 9 with this hai:r pulled back, 10 in a little pony tai::l]. 11 LENORE: [HHHH ] 12 ALINA: [and she’s like] sitting the:re and/@, 13 .hh he said, 14 I would have been here, 15 >but @she was so late. ((annoyed)) 16 and getting her any place on time/@<, 17 she’s going, 18 1→ .hh @well, ((whiny)) 19 1→ I had to get rea::dy::/@. 20 .. [I don’t know why:. 21 LENORE: [hm mh mh mh mh mh ]ah [hah hah hah hah hah] 22 ALINA: [.skuh hm mh mh] 23 .hhh nothing was gonna help her. 24 .hhh no makeup, 25 no nothing. 26 1→ cause she’s the little @gir=l, ((girly)) 27 1→ and he’s the older man, 28 1→ and [he’s taking care of me /@]. 29 LENORE: [HHHH ] 30 ALINA: .nff 31 (0.5) 32 LENORE: ... g[od ], 33 ALINA: [tsk] 34 LENORE: it sounds like the tape I made last night, 35 and these guys s-(.) [start] talking about, 36 ALINA: [.nff ] 37 LENORE: (.) you know, 38 some, 39 2→ .hh @you know, ((sappy)) 40 2→ h she’s so vul[nerable /@. 41 ALINA: [((drinking sound))] 42 LENORE: .hh she’s probably like] twenty-six, 43 2→ @and she looks thirteen, ((sappy)) 44 2→ and it[’s just so], 45 ALINA: [.nhhh ] 46 2→ LENORE: .hh it’s like/@, 47 .hh >what did<_ 48 they said something like, 49 2→ .hh @[you know what], ((sappy)) 50 ALINA: [.nff ] 51 2→ LENORE: .. it’s the butterfly, 52 2→ £you can’[t catch/@_ 53 ALINA: [AH HAH HAH HAH] VOICED DIRECT REPORTED SPEECH IN CONVERSATIONAL STORYTELLING 209

54 LENORE: .hh] and I’m l(h)ike, 55 I’m t(h)rying not(h) to vomit, 56 [>lis(h)tening< to this/£. 57 ALINA: [£give me= br=eak. 58 LENORE: ah hah hah] 59 ALINA: give me a break /£]. 60 .hhh 61 LENORE: .hh[h] 62 ALINA: [that] stupid little b::itch. 63 she just married [d:addy to take care of her ]. 64 LENORE: [.hh ((swallowing sound)) hh]

It is noteworthy that Alina starts implementing voicing already prior to the actual DRS utterance on lines 4–12 when she describes the younger woman to Lenore. Alina applies the whiny voicing outside the DRS sequence and thus animates her evaluative description and gives it further potency by prosodically enhancing her negative stance towards the protagonist. The rises and falls of the intonation pattern are strongly exaggerated. The implications of using such an exaggerated voicing just prior to a DRS utterance could be that the change in voice quality not only emphasizes the message that the speaker is setting forth but also prepares the ground for the following DRS. Additionally, the boundaries of voicing are not clear cut: voicing is not restricted to DRS but appears to be a phenomenon that is more diffuse in nature. She also evaluates the younger woman verbally on lines 2 and 4 by using the terms young chick and a real pill9 that both present her in a negative light. Alina’s evaluative utterances as a whole display the upcoming reported speaker as someone with unwanted characteristics, as a perky, overzealous, and rather nonsensical girl. On lines 18–19 Alina uses DRS when she utters well I had to get ready (pitch track illustrated in Figure 3). She applies voicing that imitates a little girl’s voice, which is high in pitch and also significantly more whiney, breathy and theatrical than Alina’s own voice. The following fundamental frequency curve represents Alina’s utterance on lines 1–2 (F0 in Figure 3) the friend that was there with them was this older guy with this young chick. This utterance can be seen to represent a sample of Alina’s natural speaking voice.

9A disagreeable or tiresome person. 210 MAARIT NIEMELÄ

513

300

200

100

80

The friend that was there with them was this older guy with this young chick

0 2.86766 Time (s) Figure 3. F0, example (2), lines 1–2

513

300

200

100

80

I had to get rea:: dy:

0 1.52671 Time (s) Figure 4. F0(4), example (2), line 19 The pitch curve shows that Alina’s utterance is situated approximately between 125 and 206 Hz, making the pitch span 8.6 semitones, whereas the frequency of the voice fundamental of her DRS on lines 18–19, I had to get ready (see Figure 4), is considerably higher, between 140 and 413 Hz with a significantly wider pitch span of 18.7 semitones. Additionally, the difference in pitch movement between F0 in Figure 3 and F0 in Figure 4 is phenomenal: the pitch in 4 shows more dramatic and noticeable rises and falls. In contrast to example 1, Lenore replies to Alina’s story on line 34 by producing a topic marker that acts as a preface to her second story It sounds like the tape I made last night. The second story here is a response, as Jefferson (1978: 220) suggests, to something that in the ongoing VOICED DIRECT REPORTED SPEECH IN CONVERSATIONAL STORYTELLING 211 conversation reminds the participant of a story.10 Lenore’s story turns out to be somewhat similar to Alina’s and concerns men and their attitudes towards younger women. However, Lenore’s story, which also contains extensively animated reported speech utterances, provides a different angle to the subject in that it seems to ridicule the men rather than the younger women. Lenore thus recycles the practice that Alina introduced in the preceding storytelling sequence by applying voiced DRS on lines 39–52 in her second story. It must be noted that Lenore does not recycle a prosodically matching voicing to Alina’s voicing but rather a voicing that performs the same action, namely ridiculing the reported speaker. On lines 35–38 Lenore indicates that she is about to describe what the men were talking about by stating and these guys [start] talking about, you know, some, but on lines 39–40 she already switches into voiced DRS and states you know, she’s so vulnerable. She applies voicing that is more breathy and gentle than her own voice. The tempo is also slower and there are pauses that create a dreamy and sappy tone of voice, which is full of emotion. By switching from regular talk to voiced DRS, Lenore is able to display, on one hand, how the reported speaker presumably characterized the girl as someone who is vulnerable and, on the other hand, her own stance towards the reported speaker by adding a melodramatic and ridiculing voicing to the reporting. It can be argued that Lenore’s sudden shift into voiced DRS produces the desired results with a minimum expenditure of energy, time, and words. Previous research (Holt 1996, Mayes 1990) shows that DRS is an extremely effective and economical way of reporting someone else’s speech and conveying affective information. But by adding voicing, Lenore here not only significantly adds to the efficiency of DRS as a stance taking practice, but actually constructs the negative implication which is not conveyed by the actual words spoken. It is the voicing that turns an otherwise straightforward DRS utterance around and creates a rather negative affective stance towards the reported speaker. In the present example, Lenore does not provide Alina with descriptive information about her stance towards the two guys on her tape but uses only voiced DRS to display her affective stance towards them. Lenore’s telling is contingent (Ford 2002) with Alina’s previous telling:

10 Jefferson (1978) describes the process of launching into storytelling from ordinary turn-by-turn talk; however, my data shows that more or less the same procedure applies to launching into second stories as well. 212 MAARIT NIEMELÄ

Alina has just used a similar practice in her previous story, which provides the relevance for Lenore’s use of voiced DRS to display her stance. Lenore produces further matching voiced DRS to quote the men on lines 43–46 and 49–52 by uttering and she looks thirteen and it’s just so, it’s like […] you know what, it’s the butterfly you can’t catch. The sappy and affectionate voicing further continues the prosodic design of her previous voiced DRS. She then concludes by reporting with a laughing voice her own rather negative reaction to the two men on lines 54–56, and I’m like, I’m trying not to vomit, listening to this (to the men). Alina responds to Lenore’s evaluative account of her own feelings by producing a follow-up, give me break, give me a break. that stupid little bitch. she just married daddy to take care of her on line 57–63, which sums up the sequence of voiced DRS utterances. Alina thus adds to her previous comments on the young chick by producing a strong evaluation that stupid little bitch on line 62, which enhances her already harsh opinion of her. Both stories display similar stances towards the older guys; however, they seem to have slightly different approaches to the matter at hand. In Alina’s first story, the negative attitude is directed essentially towards the young chick rather than the older guy, whereas Lenore’s story portrays the older guys as sappy and sentimental i.e. in a negative way. Nevertheless Alina’s final explication on line 63, she just married daddy to take care of her, takes a more or less negative stance towards the older men as well, i.e. she implies that the young woman just marries an older guy so that he would provide for her, which portrays the older man as somewhat gullible. In her follow-up, Alina thus finally wraps up the two stories and a shared stance towards the older men and younger women. Previous research (Houtkoop & Mazeland 1985, Ford 2002) shows that participants identify stories and project their ending, for example, based on recognizable story components. In the present example, Lenore exploits the participants’ shared knowledge of the identifiable story components to compose a second story, which resonates structurally with the first story. Both storytellers first produce a preface followed by a telling sequence which includes a series of evaluative voiced DRS utterances. Both voiced DRS sequences are also displays of stance towards the same subject: the relationship between (older) men and younger women. Both stories also resonate in that they apply prosodically designed voicing to display stance towards the subject matter at hand. VOICED DIRECT REPORTED SPEECH IN CONVERSATIONAL STORYTELLING 213

5. Responding to recipient-initiated voiced DRS: agreeing with a disaligning stance

The following extract provides a contrasting example of the use of voiced DRS in storytelling: it differs from previous examples in that it is the recipient who acts as the initiator of DRS, and thus the sequential environment but also the function are different. Moreover, this particular example exemplifies the stance-relevant function of voiced DRS in displaying a disaligning stance. Despite the conflicting stance expressed by the recipient, the initial storyteller responds by producing a matching voiced DRS utterance. Most importantly, it manifests the orderliness of subsequent matching voiced DRS utterances which are collaboratively produced by co-participants. In example 3, Carrie tells Stephen a story about an occasion when she visited a bar which advertises half priced Bloody Mary cocktails. Carrie was told by the bartender that the bar has run out of tomato juice, which is an essential part of the drink. Carrie then suggests that the bartender get some tomato juice from the grocery store next door. When her suggestion is refused she offers to take the money and go to the store herself to get some tomato juice. Carrie’s friend Ronald is the third character in her story.

Example (3): You’re a pushy bastard (Glasgow tape)

1 CARRIE: #a(h):n/#, 2 he was going, 3 (.) we:ll, 4 (0.1) 5 CARRIE: we’re really busy_= 6 =I was going, 7 no honestly, 8 it’ll take you . 9 (0.7) 10 CARRIE: and he went, 11 @oh, (.) ((discouraged)) 12 well, 13 (0.3) 14 CARRIE: [you know I can’t really/@]_= 15 STEPHEN: [°you’re a nighmare/° ]?, 16 CARRIE: =and I was like, 17 if y- I was your manager, 18 (0.3) 19 CARRIE: you would be ↓there ↓now. 20 STEPHEN: buying up [( )]. 21 CARRIE: [(go) you ha]ve to go and get stuff, 22 if you run out, 23 you’ve got a shop nearby, 214 MAARIT NIEMELÄ

24 (.) .hhh he just went @a:lright/@,= 25 =and he went off, 26 and when he came back, 27 (he goes), 28 @eh I don’t think we’re going to be able to do it. 29 .hh what else can I offer you/@. 30 .hh @no. 31 give me the money, 32 (0.3) 33 CARRIE: and I will go and get it/@. 34 and he was like, 35 (1.0) 36 CARRIE: @ehh .hh ((nervous)) 37 oh, 38 ehh he/@. 39 CARRIE: (.) and he was just like, 40 it ↑didn’t even seem to affect him. 41 (.)I was getting really annoyed?, 42 and Ronald’s just going, 43 (0.2) 44 STEPHEN: ↑you’re a [pushy ] bastard?, 45 CARRIE: [you’re like,] 46 ↑yeah↑ well_ 47 (.)↑fuckin’ right though. 48 1→ STEPHEN: @there’s a sign outside [saying Bloody Marys], ((angry)) 49 CARRIE: [eh he he ] 50 51 1→ STEPHEN: (.)I [WANT ONE ]/@! 52 2→ CARRIE: [@ and you don’t have] tomato juice/@. ((angry)) 53 (0.2) 54 Ronald’s going, 55 you’re like The Loog.

In this example, the positioning of the series of DRS utterances is yet again different compared to the previous two examples. Here voiced DRS does not involve the actual story line or the climax of the story. Instead, Stephen interrupts Carrie’s storytelling 0.2 seconds after her utterance on line 42, and Ronald’s just going, when he displays his understanding of the story so far and produces an explicit assessment of Carrie’s behaviour by stating you’re a pushy bastard on line 44. Stephen thus provides a setup for the following voiced DRS utterance, which is directed against the protagonist, namely the teller of the story. Stephen’s DRS utterance serves as an example of what Goodwin (1997:78) calls byplay, in which the evaluation provided by the recipient turns aside from the actual storyline but nevertheless “serves to delineate the principal conversational activity in progress”. Carrie stops the telling after Stephen’s assessment and defends her action by uttering yeah well, fucking right though on lines 46–47. She thus treats Stephen’s comment as criticism towards her actions. Stephen then VOICED DIRECT REPORTED SPEECH IN CONVERSATIONAL STORYTELLING 215 produces a voiced DRS utterance, there’s a sign outside saying Bloody Marys, I want one on lines 48 and 51, which falls into Tannen’s (1989: 110) category of dialogue constructed by the listener. He fabricates a direct order which the protagonist could have produced in situ. He applies a snappish and exaggeratedly demanding voicing, which gives the impression of an exceedingly demanding customer and thus displays the protagonist as a demanding and tantrum-prone individual, i.e. indeed a pushy bastard. Nevertheless his voicing is over-exaggerated to such an extent that it creates a comical effect; at least the recipient, Carrie, responds with laughter, which partly overlaps with the utterance. Stephen’s voiced DRS utterance does not actually act as a direct order but rather functions as an account for the previous assessment, you’re a pushy bastard, and as an animated representation of Stephen’s stance towards Carrie’s behaviour. The most striking aspect of this extract is that despite the conflicting stances of the participants the initial storyteller responds to the recipient’s disaligning stance by producing a matching voiced DRS. On line 52 Carrie produces an overlapping collaborative completion to Stephen’s voiced DRS. She utters and you don’t have tomato juice, which is prosodically and semantically designed to act as a continuation and provides an alternative ending to the first half of Stephen’s voiced DRS, there’s a sign outside saying Bloody Marys.

600

300

200

100

70

50

There's a sign outside sying bloody marys

a:nd you don't have tomato juice

0 4.10993 Time (s) Figure 5. F0, example (3), lines 48, 52

F0 in Figure 5 shows the overall pitch pattern of the collaborative completion, which seems to comply with the earlier findings on the 216 MAARIT NIEMELÄ phonetic design of collaborative completions, i.e. they are designed so that they follow the overall pitch of the prior talk and thus appear as continuation of that talk (see Couper-Kuhlen 1996, Szczepek 2000, Local 2005). Carrie’s completion displays understanding of what has just been suggested by Stephen. It can be argued that Carrie nevertheless does not see eye to eye with Stephen in terms of a shared stance because she immediately resumes storytelling after her prosodically matching collaborative completion. Local (ibid: 10) suggests that often in collaborative constructions “not only is the incoming speaker bringing the turn to a syntactic and prosodic completion, they are potentially bringing the activity which the turn-in-progress was undertaking to a completion.” Carrie here participates in bringing the disaligning activity, initiated by Stephen (you’re a pushy bastard), to a completion and immediately resumes storytelling on line 54, Ronald’s going, by partially repeating what she had said on line 42, and Ronald’s just going, before the recipient- initiated DRS sequence began. The sequence initiated by Stephen is treated by both participants as a separate disaligning evaluation sequence which is not expanded any further. Participants nonetheless engage in stance taking by recycling each others voicing, which seems to indicate that responding to voiced DRS with subsequent voiced DRS can be more or less an ordered phenomenon in interaction. Furthermore, it accomplishes a display of alignment and a shared stance between participants.

6. Conclusion

Voiced DRS is a sequentially relevant interactive practice of stance taking in conversational storytelling. As we have seen, conversationalists can recycle each other’s voicing in conversation, which can be seen as indicative of a shared stance and part of a wider stance taking activity between co-participants. By way of recycling matching voicing in DRS, participants thus engage in the reciprocal activity of building a shared stance turn by turn. Matching voiced DRS can appear in various sequential environments in conversational storytelling. Chapter 3 introduces an example of a sequence of matching voiced DRS utterances within a single story. The initial storyteller provides a setup for the voiced DRS sequence, which is followed by a series of voiced DRS utterances by her. The recipient then produces a subsequent series of voiced DRS utterances, which in turn VOICED DIRECT REPORTED SPEECH IN CONVERSATIONAL STORYTELLING 217 sometimes overlap with further voiced DRS utterances from the initial teller, who finally wraps up the sequence by producing a follow-up. In chapter 4 the recipient’s use of voiced DRS in the second story builds upon and contributes to a stance displayed in the first story. The teller of the first story sets up a context for voiced DRS and again produces a series of voiced DRS utterances. The teller of the second story then produces a similar series of voiced DRS utterances, which resonate with the prior stance by way of applying a voicing that performs the same overall action as in the first story. Finally, a concluding follow-up is produced by the teller of the first story. Chapter 5 introduces the stance-relevant function of voiced DRS in displaying a disaligning stance. Unlike in the first two examples, the voiced DRS is not in line with the projected story line, but rather interferes with the story line of the initial teller. The initial storyteller nonetheless responds by producing a matching voiced DRS utterance, which momentarily displays a shared stance and thus allows her to resume the telling. This seems to indicate that a voiced DRS utterance that displays a shared stance is in fact a preferred response to a first voiced DRS. This suggests that the occurrence of subsequent matching voiced DRS utterances is an orderly phenomenon in interaction. It therefore appears that, in the analyzed examples, matching voiced DRS is a sequentially relevant practice in stance taking. However, a larger collection of matching voiced DRS within the storytelling structure and a detailed look at the non-occurrence of matching voiced DRS is needed in order to establish how strong an interactional order it represents.

Appendix: Transcription key

For the purpose of this paper, I have compiled a transcription key from Tainio (1997) and Gardner (2001) which chiefly follows the notation that is based on the Jeffersonian system. All examples have been transcribed into intonation units. No capital letters have been used except in and first person singular personal pronouns.

. terminal contour: falling ? terminal contour: strongly rising ?, terminal contour: slightly rising

; continuative contour : slight fall 218 MAARIT NIEMELÄ yes_ continuative contour: level pitch , continuative contour: slight rise ! strongly animated tone, pitch movement in any direction

[ ] overlapping = latching ( . ) micropause (less than 0.2 sec) (2.0) length of pause in approximate seconds ye:s stretching of sound ye- truncated word yes contrastive stress or emphasis YES increased volume °yes/° decreased volume hhh audible breath .hhh audible in-breath ↑ A marked upward shift in pitch, high pitch has sudden onset but gradually wanders down ↓ A marked downward shift in pitch, low pitch has sudden onset but gradually wanders up ye(hh)s within-speech aspiration, laughter ((cough)) untranscribable sounds, transcriber’s comments or description of voice quality e.g. ((angry)) (yes) uncertain hearing, transciber’s best guess ( ) uncertain hearing -yes, ¯yes higher, lower than surrounding speech £yes/£ smiley voice #yes/# creaky voice >yes< faster than surrounding speech slower than surrounding speech @yes/@ change in voice quality, usually reported speech

References

Arminen, Ilkka (2004) Second stories: the salience of interpersonal communication for mutual help in Alcoholics Anonymous. 36: 319–347. Besnier, Niko (1993) Reported speech and affect on Nukulaelae atoll. In Hill, Jane H. and Judith T. Irvine (eds.), Responsibility and Evidence in Oral Discourse, pp. 161–181. Cambridge: Cambridge University Press. VOICED DIRECT REPORTED SPEECH IN CONVERSATIONAL STORYTELLING 219

Clark, Herbert H. & Richard J. Gerrig (1990) Quotations as demonstrations. Language 66: 764–805. Clift, Rebecca (2000) Stance-taking in reported speech. Essex Research Reports in Linguistics, vol. 32. University of Essex. Couper-Kuhlen, Elizabeth (1996) The prosody of repetition: on quoting and mimicry. In Couper-Kuhlen, E. & M. Selting (eds.), Prosody in conversation: interactional studies, pp. 366–405. Cambridge: Cambridge University Press. —— (1998) Coherent Voicing. On Prosody in Conversational Reported Speech. Interaction and Linguistic Structures, InLiSt, No. 1. http://inlist.uni- konstanz.de/issues/1/index.htm [1999, March 30]. Drew, Paul (1998) Complaints About Transgressions and Misconduct. Research on Language and Social Interaction 31(3&4): 295–325. Du Bois, John, S. Schuetze-Coburn, S. Cumming & D. Paolino (1993) An Outline of Discourse Transcription. In Edwards, J & M. Lampert (eds.), Talking data: Transcription and coding in discourse research. pp. 45–87. Hillsdale, NJ: Lawrence Erlbaum. Du Bois, John (2000a) Santa Barbara Corpus of Spoken American English, Part 1. Six hours of digitized audio and transcriptions on 3 CD-ROMs. Linguistic Data Consortium, University of Pennsylvania. —— (2000b) Taking a stance: Constituting the stance differential in dialogic interaction. Paper given at the conference of the American Anthropological Association, San Francisco, November 2000. —— (2003a) Stance and Consequence in Interaction. Paper given in Langnet conference in Oulu University, Finland, September 2003. —— (2003b) The Dialogic Moment. Paper given in a lecture series on “Stance and Dialogic Syntax in Interaction. University of Oulu, Department of English, September 2003. Ford, Cecilia (2002) Contingency and units in interaction. Discourse Studies 6(1): 27– 52. Ford, Cecilia, Barbara Fox & Sandra Thompson (2003) Social interaction and grammar. In Tomasello, M. (ed.), The New Psychology of Language: Cognitive and Functional Approaches to Language Structure, Vol. 2, pp. 119–143. Mahwah, NJ: Erlbaum. Gardner, Rod (2001) When listeners talk: Response tokens and recipient stance. Amsterdam: Benjamins. Glenn, Phillip (1989) Initiating shared laughter in multi-party conversations. Western Journal of Speech Communication 53: 127–149. Golato, Andrea (2000) An innovative German quotative for reporting embodied actions: Und ich so / und er so ‘and I’m like / and he’s like’. Journal of Pragmatics 32: 29–54. Goodwin, Charles. & John Heritage (1991) Conversation Analysis. Annual Review of Anthropology 19: 283–307. Goodwin, Marjorie (1997) By-Play: Negotiating Evaluation in Storytelling. In G. R. Guy, C. Feagin, D. Schriffin and J. Baugh (eds.) Towards a Social Science of 220 MAARIT NIEMELÄ

Language: Papers in Honor of William Labov 2: Social Interaction and Discourse Structures, pp. 77–102. Amsterdam/Philadelphia: John Benjamins. Haakana, Markku (2002) Laughter in medical interaction: From quantification to analysis, and back. Journal of 6(2): 207–235. —— (forthcoming) On the interactional differences between ‘saying’ and ‘thinking’: Reported thoughts in complaint stories. In: Elizabet Holt & Rebecca Clift (eds.): Voicing – Reported speech and footing in conversation. Haddington, Pentti (2004) Stance taking in news interviews. SKY Journal of Linguistics 17: 101–142. —— (2005) The Intersubjectivity of Stance Taking in Talk-in-interaction. Oulu, Finland: Oulu University Press. Hautkoop, Henneke & Harrie Mazeland (1985) Turns and Discourse Units in Everyday Conversation, Journal of Pragmatics 9: 595–619. Holt, Elisabeth (1996) Reporting on Talk: The use of direct reported speech in conversation. Research on Language and Social Interaction 29(3): 219–245. —— (2000) Reporting and Reacting: Concurrent Responses to Reported Speech. Research on Language and Social Interaction 33(4): 425–454. Hutchby, Ian & Robin Wooffitt (1998) Conversation Analysis. Principles, practices and applications. Cambridge: Polity Press. Jefferson, Gail (1978) Sequential aspects of storytelling in interaction. In Schenkein, J. (ed.), Studies in the organization of conversational interaction, pp. 219–248. New York: Academic. —— (1984) On the organization of laughter in talk about troubles. In J.M. Atkinson and J.C. Heritage (eds.) Structures of social action: Studies in conversation analysis, pp. 346–369. Cambridge, UK: Cambridge University Press. Keisanen, Tiina (under review) Patterns of stance taking: Challenging with a negative interrogative in spoken American English. In Englebretson, R. (ed.) Stancetaking in discourse: the intersubjectivity of interaction. [Papers from the Tenth Biennial Rice Linguistics Symposium] Kärkkäinen, Elise (2003) ’Is she vicious or dense?’ Dialogic practices of stance taking in conversation. In Toshihide Nakayama, Tsuyoshi Ono and Hongyin Tao (eds.), Santa Barbara Papers in Linguistics 12. Recent Studies in Empirical Approaches to Language. pp. 47–65. University of California, Santa Barbara. Kärkkäinen, Elise, Tiina Keisanen, Mirka Rauniomaa, Maarit Niemelä & Pentti Haddington (2004) Interactional practices and linguistic resources of stance taking in spoken English. In Jan-Ola Östman (ed.) PIC Bulletin 7. University of Helsinki: 45–48. Local, John (2005) On the interactional and phonetic design of collaborative completions. In Hardcastle, W. and J. Beck (eds.), A : a Festschrift for John Laver. Mathis, Terri & George Yule (1994) Zero quotatives. 18: 63–76. Mayes, Patricia (1990) Quotation in Spoken English. Studies in Language 14: 325–363. Niemelä, Maarit (2003) Stance taking in direct reported speech. Unpublished MA thesis. Oulu: University of Oulu. VOICED DIRECT REPORTED SPEECH IN CONVERSATIONAL STORYTELLING 221

Norrick, Neil (2000) Conversational Narrative: Storytelling in everyday talk. Amsterdam/Philadelphia: John Benjamins Publishing Company. Ochs, Elinor (1989) (ed.) The pragmatics of affect, Special issue of Text 9(1). Rauniomaa, Mirka (in progress) Repetition, recovery and stance taking in spoken English and Finnish. University of Oulu. Doctoral Thesis. —— (under review) Stance markers in spoken Finnish: minun mielestä and minusta in assessments. In Englebretson, R. (ed.) Stancetaking in discourse: the intersubjectivity of interaction. [Papers from the Tenth Biennial Rice Linguistics Symposium]. Routarinne, Sara (1997) Kertomuksen rakentaminen [Story Structure]. In Tainio, L. (ed.) Keskusteluanalyysin perusteet [Introduction to Conversation Analysis], pp. 138–155. Tampere: Vastapaino. Sacks, Harvey (1984) Notes on methodology. In Atkinson, J. & J. Heritage (eds.), Structures of social action: Studies in conversation analysis, pp. 21–27. Cambridge, England: Cambridge University Press. —— (1992) (ed. Jefferson, G.) Lectures on Conversation. Oxford: Blackwell. Sandlund, Erica (2004) Feeling by Doing: The Social Organization of Everyday Emotions in Academic Talk-in-Interaction. Doctoral dissertation, Karlstad University: Karlstad University Studies. Selting, Margaret & Elisabeth Couper-Kuhlen (2001) (eds.), Studies in interactional linguistics. Amsterdam: Benjamins. Szczepek, Beatrize (2000) Formal Aspects of Collaborative Productions in English Conversation. In Interaction and Linguistic Structures, InList, No. 17, http://inlist.uni-konstanz.de/issues/17/index.htm [2000 June]. —— (2001) Prosodic Orientation in Spoken Interaction. In Interaction and Linguistic Structures, InLiSt, No. 27, http://inlist.uni-konstanz.de/issues/27/index.htm [2001 November]. Tainio, Liisa (1996) Arkikertomukset kulttuurisen tietämyksen välittäjinä. [Conversational storytelling as mediator of cultural knowledge] Naistutkimus 9(2): 11–29. —— (1997) Keskusteluanalyysin perusteet [Introduction to Conversation Analysis]. Tampere: Vastapaino. Tannen, Deborah (1989) Talking voices: Repetition, dialogue, and imagery in conversational discourse. Cambridge: Cambridge University Press.

Contact information:

Maarit Niemelä Department of English P.O. Box 1000 90014 OULUN YLIOPISTO Finland +3588553 3426 e-mail: [email protected]

Timothy Osborne

Coherence: A Analysis

Abstract

Bech’s seminal distinction between coherent and incoherent constructions in German does not allow for those structures that display the traits of both coherence and incoherence simultaneously, e.g. the third construction. Hence his original two-way distinction has given way to a three-way distinction. This paper proposes that a three- way division is indeed more accurate. The approach presented here distinguishes between coherent, incoherent, and pseudoincoherent constituents. Nine diagnostics are utilized to support this three-way division, i.e. extraposition, intraposition, bare infinitive fronting, infinitival fronting, scrambling, pied-piping, position of negation, scope of negation, and gapping. The analysis is couched in a dependency grammar framework and formalized in terms of the two features ±s (scrambling) and ±p (predicate). Three principles of word order are formulated in terms of these features: the Scrambling Principle, the Predicate Serialization Principle, and the Predicate Weight Principle. An introduction to dependency grammar is provided.

1. Coherence

Most studies of nonfinite verb constructions in German build on, or modify, Bech’s (1955) seminal work on coherence fields. Bech discerns between coherent and incoherent constructions. He groups verbs according to the type of construction in which they appear.

(1) a. Thomas braucht die Zeitung nicht zu lesen. Thomas needs the paper not to read ‘Thomas doesn’t need to read the paper.’

b. *weil Thomas nicht braucht, die Zeitung zu lesen ‘because Thomas doesn’t need to read the paper’

c. weil die Zeitung Thomas nicht zu lesen braucht

SKY Journal of Linguistics (18) 2005, 223–286 224 TIMOTHY OSBORNE

(2) a. Thomas fordert mich auf, die Zeitung zu lesen. Thomas encourages me up the paper to read ‘Thomas is encouraging me to read the paper.’

b. weil Thomas mich auffordert, die Zeitung zu lesen ‘because Thomas encourages me to read the paper’

c. *weil Thomas mich die Zeitung auffordert zu lesen

(3) a. Thomas versucht, die Zeitung zu lesen. Thomas tries the paper to read ‘Thomas tries to read the paper.’

b. weil Thomas versucht, die Zeitung zu lesen ‘because Thomas tries to read the paper’

c. weil Thomas die Zeitung versucht zu lesen

The examples (1a–c) illustrate that nicht brauchen ‘not need’ constructs coherently. According to Bech (1955: 75), a central trait of coherent constructions is that the infinitival phrase may not be extraposed as in (1b). Sentence (1c) demonstrates another trait of coherent constructions observed by Bech (1955: 61f., 74f.): they allow the discontinuities of overlapping verb fields. The examples (2a–c) illustrate that auffordern ‘encourage’ can construct incoherently: the zu-infinitive phrase die Zeitung zu lesen ‘to read the paper’ can be extraposed, as in (2a– b), and the complement of the zu-infinitive, i.e. die Zeitung ‘the paper’, may not appear in a position resulting in a discontinuity, as shown in (2c). Bech’s theory in terms of coherence fields is challenged by example (3c), though. His analysis assumes two construction types, i.e. coherent vs. incoherent, whereby many verbs have the option to construct coherently or incoherently. This delineation of phenomena implies that a construction can be either coherent or incoherent, but it does not allow for a construction to demonstrate the traits of both coherence and incoherence simultaneously. Example (3c) is hence problematic because it simultaneously demonstrates the discontinuity of a coherent construction and the extraposition of an incoherent construction. This difficulty with Bech’s work is widely acknowledged. Discussions of the problem – e.g. Besten and Rutten (1989), Kiss (1995: 109ff.), Hinrichs and Nakazawa COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 225

(1998), Müller (1998: 189ff.), Wöllstein-Leisten (2001: 8), Rambow (2003: 232), Reis and Sternefeld (2004: 488ff.) – focus on the discontinuity of the third construction, which is illustrated in (3c). Typical instances of the third construction have the object of the embedded zu-infinitive appearing in the Mittelfeld ‘middle field’, whereas the zu-infinitive itself is extraposed. Due to the difficulties posed by the third construction to Bech’s theory, a three-way distinction is now acknowledged, cf. Wöllstein-Leisten (2001: 67). The need for this three-way distinction is further illustrated in the following contrast.

(2) d. *Aufgefordert zu lesen hat mich Thomas die Zeitung. e. *Zu lesen aufgefordert hat mich Thomas die Zeitung. f. *Zu lesen hat mich Thomas die Zeitung aufgefordert.

(3) d. Versucht zu lesen hat Thomas die Zeitung e. Zu lesen versucht hat Thomas die Zeitung. f. Zu lesen hat Thomas die Zeitung versucht.

Auffordern and versuchen both allow extraposition, as illustrated in (2a) and (3a). In this regard, they are similar. Examples (2d–f, 3d–f) illustrate, however, that auffordern disallows certain discontinuities, whereas versuchen allows these discontinuities. In order to allow for this contrast, a three-way distinction is necessary. This paper presents a dependency grammar analysis of coherence in German. In place of Bech’s two-way distinction, a three-way distinction is adopted. The approach distinguishes between coherent, incoherent, and pseudoincoherent constituents. This three-way division is formalized in terms of two features that appear on the root words of the constituents involved.1 The feature ±s (scrambling) addresses the extent to which the rising of scrambling is allowed, and the feature ±p (predicate) addresses the positional restrictions on that constituent. These features help establish three core principles of word order in German: the Scrambling Principle, the Predicate Serialization Principle, and the Predicate Weight Principle. The topological model as it is commonly understood results from the interaction of these three principles with the V2 principle. To my knowledge, a dependency grammar theory of coherence is lacking.

1 The term “root” is defined in section 3.1. The root of a constituent is the highest word in that constituent. 226 TIMOTHY OSBORNE

2. Confusion in the literature

The manner in which ‘coherence’ is discussed and employed in the literature varies greatly. The terminology is inconsistent and at times contradictory. The confusion is at least in part due to the fact that Bech himself was not clear about the criteria that must be fulfilled in order for a construction to qualify as coherent or incoherent – see Kvam (1979), Stechow (1984), and Grewendorf (1988: 267, 274ff.) in this regard. One can distinguish between two opposing stances. The one stance defines ‘coherence’ in terms of a single criterion, namely the position of the infinitival phrase in relation to the verb that immediately governs it. If the infinitival phrase precedes its governor, e.g. it is center-embedded, one is dealing with a coherent construction. If the infinitival phrase follows its governor, e.g. it appears in the Nachfeld ‘after field,’ one is dealing with an incoherent construction.

(4) a. Er fährt zu arbeiten fort. - Coherent construction he drives to work further ‘He continues to work.’

b. Er fährt fort zu arbeiten. - Incoherent construction

The only thing that counts is the serial position of the infinitival phrase in relation to the right bracket. Following Eisenberg (1999: 495), this understanding of coherence shall be referred to as the topology stance. The second possibility understands coherence in terms of , whereby numerous criteria, not just one, are employed to distinguish coherent from incoherent constituents, e.g. extraposition, intraposition, bare infinitive fronting, infinitival fronting, scrambling, pied- piping, the position of negation, the scope of negation, gapping, etc. A verb like fortfahren ‘continue’ subcategorizes for an incoherent zu-infinitive, and this zu-infinitive is incoherent regardless of whether it precedes or follows the right bracket:

(4’) a. Er fährt zu arbeiten fort. - Incoherent constituent b. Er fährt fort zu arbeiten. - Incoherent constituent

This understanding of coherence shall be referred to as the subcategorization stance. COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 227

As Eisenberg (1999: 495) notes, the manner in which the terminology on coherence is employed in the literature quite often confuses these two stances. To illustrate this fact, the discussion here shall consider three ways in which linguists employ the terminology on coherence. The literature distinguishes between

1. Coherent vs. incoherent constructions, 2. Coherent vs. incoherent verbs, and/or 3. Coherent vs. incoherent infinitives.

Even though these terms are mixed and matched in various ways, it is fair to acknowledge these three areas. Each of these oppositions shall be considered in turn. Coherent vs. incoherent constructions: For the most part, the manner in which the terms coherent vs. incoherent construction are employed is consistent with the topology stance, e.g. Kvam (1982), Stechow (1990), Kiss (1994, 1995), Hinrichs and Nakazawa (1998), S. Müller (2002). That is, a coherent construction is defined solely in terms of the position of the infinitival phrase with respect to its governor. If the infinitival phrase is center-embedded, a coherent construction obtains. If it is extraposed, an incoherent construction obtains. None of the other criteria used to determine coherence are needed. Coherent vs. incoherent verbs: The use of the terms coherent vs. incoherent verb is less clear – see Stechow (1990), Eisenberg (1999: 353), De Kuthy and Meurers (2001), Rambow (2003). A coherent verb obligatorily constructs coherently, i.e. it appears only in coherent constructions. An incoherent verb in contrast, like fortfahren in (4), has the option to construct coherently or incoherently, i.e. it can appear in a coherent or incoherent construction. The grouping of verbs according to the type of construction in which they appear is an attempt to acknowledge subcategorization traits. Hence the use of these terms takes a significant step in the direction of the subcategorization stance. The problem with these terms, though, is that those verbs that are labeled “incoherent” actually have the option to appear in coherent or incoherent constructions. Thus it can occur that an incoherent verb constructs coherently, which is a contradiction in terms.2 The only way around this contradiction would be to

2 De Kuthy & Meurers (2001) provide a good example of this contradiction in terms. They acknowledge obligatorily coherent (e.g. scheinen ‘seem’, pflegen ‘usually do’, 228 TIMOTHY OSBORNE posit two lexical entries for each of the verbs. A verb like fortfahren would have an entry as a coherent verb and an entry as an incoherent verb. To increase the number of lexical entries in this manner would be to mistakenly situate an aspect of coherence in the lexicon that belongs in the syntax. Coherent vs. incoherent infinitives: Some linguists apply the terms coherent and incoherent to the embedded verbs rather than to the matrix verbs, e.g. Fanselow (1989), G. Müller (1998: 17ff.), Sabel (1999). This use of the terminology is very close to the subcategorization stance. An infinitival phrase is deemed to be coherent or incoherent regardless of whether it is center-embedded or extraposed. The zu-infinitive zu arbeiten in (4), for instance, is incoherent in both (4a) and (4b). The linguists who use the terminology in this manner generate confusion, though, when they also employ the terms coherent vs. incoherent construction. An incoherent zu-infinitive can appear in a coherent construction as in (4a), again a contradiction in terms. The dependency grammar approach developed below is not confronted with these difficulties. Its use of the terminology is entirely consistent with the subcategorization stance. In fact, it is not possible for dependency grammar to acknowledge the topology stance. Dependency structures cannot, namely, be understood in terms of constructions. According to Wells (1947), a construction is a sentence or constituent that can be broken down into two or more immediate constituents. Thus only terminal constituents fail to qualify as constructions. An immediate constituent analysis of this sort is not applicable to dependency structures. A non-terminal constituent in dependency grammar cannot be understood in terms of the immediate constituents of which it consists, but rather it can only be broken down by separating off the constituents lower in the hierarchy. This point becomes evident in the next section where some fundamentals of dependency grammar are presented. With the inability of dependency grammar to produce immediate constituent analyses, the possibility to use the term construction is absent. Therefore the terminology that is most closely associated with the topology stance, i.e. coherent vs. incoherent construction, should not be employed in

bekommen ‘receive’), optionally coherent (e.g. versuchen ‘try’, wagen ‘dare’, hoffen ‘hope’), and obligatorily incoherent verbs (e.g. auffordern ‘encourage’, begehren ‘desire’, fortfahren ‘continue’). Even though the latter are obligatorily incoherent, they have the option to appear in coherent constructions, e.g. fortfahren in (4a). COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 229 a dependency grammar theory of coherence to begin with. This situation is beneficial since it reduces the source of confusion.3

3. Dependency grammar

Since dependency-based grammars are seldom in comparison with constituency-based ones, an introduction to the framework is warranted. This introduction is accomplished here by contrasting dependency with constituency. Comparisons of the two – e.g. Baumgärtner (1970), Anderson (1979: 92ff.), Hudson (1980), Matthews (1981 Ch. 4), Engel (1982: 27ff.), Mel'čuk (1988: 12ff.), Siewierska (1988: 142ff.), Jung (1995: 15ff.), Eroms (2000: 75ff.), Hudson (2000: 20ff.), Tarvainen (2000: 11ff.) – emphasize that constituency is a part-whole relation, whereas dependency is a strict mother-daughter relation. The part-whole relation is a result of syntactic units combining with each other to form greater units. The mother-daughter relation, in contrast, is a result of syntactic units attaching to each other, the result being a greater unit. The distinction is best understood in terms of the tree structures that each approach generates:

3 An anonymous reviewer comments that the topology vs. subcategorization dichotomy established here is not warranted and that both views of coherence are necessary. The reviewer writes: “Ich möchte behaupten, dass unterschiedliche infinitivregierende Elemente für Infinitive mit unterschiedlichen topologischen Eigenschaften subkategorisert sind, d.h. ich halte beide Begriffe für gleich empirisch und deskriptiv notwendig.“

‘I would maintain that different governors of inifinitives subcategorize for infinitives having different topological characteristics, i.e. I think both concepts are equally empirically and descriptively necessary.’

I do not disagree with this statement; using topology to describe the behavior of coherent/incoherent constituents is indeed quite useful. The statement does not, however, address the core definition of coherence. The problem is that if you define coherence in terms of topology, then certain verbs require two lexical entries (e.g. fortfahren): one entry has the verb appearing in a coherent construction and the other has it appearing in an incoherent construction. In so doing, you have missed a generalization by situating coherence in the lexicon. However, if you define coherence in terms of subcategorization as done here, then you have considerably reduced the size of the lexicon by situating coherence in the syntax. Stated another way, I am arguing that topology should be used to describe the subcategorization traits of verbs, not vice versa. 230 TIMOTHY OSBORNE

The constituency tree (5a) illustrates the manner in which syntactic units combine to create greater units. The word sie X2 combines with the word bleibt X3 to create the sentence X1. The dependency tree (5b) shows that dependency has the words attaching to each other. The daughter word sie X1 attaches to its mother word bleibt X2, the result being a sentence. Notice that there are two words in the string but three nodes in the constituency structure (5a). In contrast, there are two words in the string and exactly two nodes in the dependency structure (5b). The difference between constituency and dependency can indeed be understood in terms of the word-to-node ratio. Consider the following principles of tree construction:

I. a. One word per node, and b. One node per word. II. One head per node, and III. One root node per structure.

Dependency and constituency alike generally adhere to principles Ia, II and III. They differ, however, with respect to Ib. The projections of constituency structures necessitate the presence of ‘higher nodes’, e.g. X1 in (5a). The presence of higher nodes means that the number of nodes in the structure always outnumbers – by at least one – the number of words in the string. The absence of such projections in dependency structures, in contrast, results in a situation where no higher nodes are present, meaning the number of nodes in the structure is the same as the number of words in the string. Dependency can hence be understood as a one-to-one relation, whereas constituency is a one-to-more-than-one relation. The one-to-one relation of dependency results in syntactic structures that generally contain half the number of nodes and edges as the corresponding constituency structures. Despite this paucity of structure however, dependency inherently conveys information that constituency does not. COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 231

In4the dependency tree (6a), it is visible that Studenten is the head of die, that werden is the head of die Studenten and die konkurrierenden Syntaxtheorien studieren, that studieren is the head of die konkurrierenden Syntaxtheorien, and that Syntaxtheorien is the head of die and konkurrierenden. In the constituency tree (6b), in contrast, the head-dependent relation is not visible. It is not apparent for instance whether die or Studenten is the head of die Studenten, or whether konkurrierenden or Syntaxtheorien is the head of konkurrierenden Syntaxtheorien, etc. In order to convey this information, constituency syntax must include the category labels of the phrase markers; the status of the various projection levels – i.e. minimal, intermediate, maximal – must be visible on the node labels. Dependency structures, in contrast, can be further reduced by collapsing the word-node distinction altogether. The words are positioned directly in the hierarchy in the following manner:

4 This paper is consistent insofar as the node labels in the dependency trees always have the linear index of the word. 232 TIMOTHY OSBORNE

The question arises as to whether an approach is possible that combines aspects of constituency and dependency. In other words, is a hybrid system possible? I believe the telescope principle of Brody’s (1998, 2003) Mirror Theory does just that, i.e. it results in a hybrid system. The telescope principle collapses the minimal, intermediate, and maximal projections of a lexical item into a single node. The following trees are adapted from Brody (2003: 252f.):

The structure (7b) results when the various projection levels of the lexical items in (7a) are each collapsed into a single projection. That is, Infl-Infl'- InflP, v-v'-vP, and V-V'-VP are all collapsed to the single nodes Infl, v, and V, respectively. The tree (7b) qualifies as a hybrid dependency- constituency tree because it has one non-terminal node occupied but the others unoccupied by lexical items. In this regard, compare the pure dependency tree (6a) and the pure constituency tree (6b) with the hybrid tree (7b). Pure constituency structures such as (6b) have only the terminal nodes (X4, X5, X6, X10, X12, X13, X9) occupied by lexical items; the nonterminal nodes remain unoccupied. Pure dependency structures such as (6a), in contrast, have all nodes, regardless if they are terminal or nonterminal, occupied by lexical items. In this respect, (7b) is indeed a hybrid structure since it has one non-terminal node (v) occupied, and the COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 233 other two non-terminal nodes unoccupied (Infl and V) by lexical items. The dependency grammar utilized in this paper is based on pure dependency structures, not on hybrid structures. The reason for mentioning the hybrid system here has been to further illuminate the dependency vs. constituency distinction. See Matthews (1981: Ch. 4) for further discussion on the possibility of hybrid systems. The dependency structures produced in this paper are consistent in relevant respects with a long and established tradition, e.g. Tesnière (1959/69), Hays (1964), Robinson (1970), Kunze (1975), Matthews (1981), Engel (1982), Miller (1985), Mel’čuk (1988), Schubert (1988), Starosta (1988), Lobin (1993), Pickering and Barry (1993), Jung (1995), Heringer (1996), Groß (1999), Eroms (2000), Kahane (2000), Tarvainen (2000), Hudson (1984, 1990, 2000, 2003), Osborne (2003, 2005a), etc. There are a number of areas where these linguists are almost unanimous in their views. For instance, in surface syntax the subject is a dependent of the finite verb,5 the object is a dependent of the infinitival verb (assuming one is present), the infinitival verb is the daughter of the finite verb, the full verb is the daughter of the , etc. One prominent point of disagreement is the status of determiners, i.e. NP vs. DP.6 Since this area does not bear directly on a theory of coherence, it will not be addressed in this paper. NP is assumed. A major strength of dependency grammar is its structural minimalism. This strength is evident when the need arises to demonstrate graphically the structures and principles of syntax. Dependency trees are easily produced and can hence be utilized often to illustrate the principles of syntax under investigation. The numerous dependency trees in this paper bear witness to this point.

3.1 Constituents, roots, heads, dependents, mothers, daughters, and governors

The following terminology shall be used to describe the surface structural relations that obtain between the units of syntax in dependency structures.

5 Whether or not the subject is understood as a dependent of the nonfinite verb that then obligatorily rises shall not enter into the discussion. The important point for the purpose of this paper is that the subject is always a dependent of the finite verb in surface syntax. 6 Another possibility is that determiner and noun are interdependent. Eroms’ (1988) Doppelkopf ‘dual head’ analysis pursues this possibility. 234 TIMOTHY OSBORNE

Constituent: A word plus all the words that that word dominates. Root: The one word in a given constituent that is not dominated by any other word in that constituent. Head: The one word that immediately dominates a given constituent. Dependent: A constituent that is immediately dominated by a given word. Mother: The one word that immediately dominates a given word. Daughter: A word that a given word immediately dominates. Governor: The one word that licenses the appearance of a given word, constituent, or dependent.

The one-to-one relation (nodes to words) inherent in dependency structures makes it possible to collapse the node-word distinction entirely, as illustrated in (6b') above. Accordingly, these definitions refer to “words” only, whereby “word” is the same as “node”. The paragraphs below discuss these seven units using the following dependency structure. The nature of the dashed dependency edge connecting X1 to X2 is discussed below in section 3.2.

Constituents: It is not common for dependency grammars to view syntactic structure in terms of constituents. This aspect of dependency grammar is, though, just a matter of terminology. Dependency grammarians use various terms to denote the syntactic unit defined in the definition. Tesnière (1959/69: 14) calls the unit a nœud ‘node’; Kunze (1975: 13) names it a vollständiger Teilbaum ‘complete partial tree’; Hays (1964: 520) and Mel'čuk (1988: 14) call it a subtree; Groß (1999: 69) and Eroms (2000:86ff.) call it a phrase. Pickering and Barry (1993: 865) use the term full-constituent. Hudson (1984: 92) and Siewierska (1988: 142) use the term constituent. This paper follows Hudson and Siewierska in this regard. There are hence seven constituents in (8): was X1, die X3, die Dozenten X3 X4, den X5, den Studenten X5 X6, den Studenten sagen X5 COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 235

X6 X7, and den Studenten sagen sollen X5 X6 X7 X8. Note that the single words hätten X2, Dozenten X4, Studenten X6, sagen X7, and sollen X8 do not each alone qualify as a constituent since they dominate other nodes. Due to this aspect of dependency hierarchies, dependency constituent structure differs a lot from constituency constituent structure. Roots vs. heads: When focusing on a given constituent, it is advantageous to distinguish between the one word that is supreme in that constituent, i.e. the root of that constituent, and the one word appearing outside of that constituent to which that constituent is connected, i.e. the head of that constituent. Consider the constituent die Dozenten in (8): Dozenten is its root and hätten is its head. Consider the constituent den Studenten sagen: sagen is its root and sollen is its head. The root-head distinction is not possible in constituency grammar, which can only acknowledge heads. Dependents, mothers, and daughters: At times it is necessary to distinguish between the word(s) that a given word immediately dominates, i.e. its daughter(s), and the constituent(s) that that word immediately dominates, i.e. its dependent(s). Consider for instance the word hätten in (8): its daughters are was, Dozenten, and sollen, whereas its dependents are was, die Dozenten, and den Studenten sagen sollen. Consider next the word sagen: it has one daughter, i.e. Studenten, and one dependent, i.e. den Studenten. If one looks up the hierarchy from a given node, then the one node that immediately dominates that node is its mother, e.g. focusing on sagen, sollen is its mother. A word may have more than one daughter, but never more than one mother. Heads vs. governors: The head-governor distinction is necessary to address the various discontinuities that occur in the grammar. The term head refers to an aspect of surface configurations. The term governor, in contrast, refers to an aspect of subcategorization. In most cases, the head and the governor of a dependent are one and the same word. When discontinuities occur however, the assumption is that the relevant constituent has taken on a higher word as its head. That is, it has taken on a word as its head that is not its governor. In (8) for instance, the head of was is hätten, but the governor of was is sagen. Sagen qualifies as the governor of was because sagen subcategorizes for an (accusative) object, meaning it is sagen that licenses the appearance of was. The head-governor distinction has precedents in the dependency grammar literature. Where the current system employs the term head, Bröker (2000: 253) uses the term “linear governor”, and Hudson (2000: 32) the term “surface parent”. And where 236 TIMOTHY OSBORNE the current system employs the term governor, Bröker (2000: 253) uses the term “syntactic governor”, and Hudson (2000: 32) the term “extra parent”. It should be apparent that the head-governor distinction is in general the means by which dependency grammar addresses discontinuities of every sort.

3.2 Projectivity and Rising

Dependency grammar defines discontinuities in terms of crossing lines. If a projection line crosses a dependency edge, then a discontinuity is present. Discontinuities are hence called “projectivity violations” – see Mel'čuk (1988: 35ff.), Heringer (1996: 243ff.), and Eroms (2000: 311ff.).

The position of seinen Vorschlag in each case with respect to its governor abgelehnt results in crossing lines, i.e. the dependency edge connecting seinen Vorschlag to abgelehnt crosses two projection lines. Thus the structures (9a–b) each contain a projectivity violation. Addressing such discontinuities is of course a major goal of dependency grammar. In this respect though, it is worth noting that dependency structures involve fewer discontinuities than the corresponding constituency structures. This is so because dependency structures are usually flatter than constituency structures – see Starosta (1988:106), Heringer (1996: 27f.), Hudson (2000: 22). For instance, compare tree (6a), four levels, with tree (6b), six levels. Despite the flatter structures, discontinuities of the sort illustrated in (9a–b) are a common occurrence in German. The V2 principle of German seems to allow more projectivity violations than the subject-verb principle of English.7

7 An anonymous reviewer raises the question whether dependency grammar acknowledges German as essentially a V-last language as many linguists assume. This COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 237

The strategy employed here to address projectivity violations was mentioned in the previous section. The assumption is that rising occurs, i.e. a dependent rises and attaches to a word above its governor.8 The result of rising is that the risen constituent’s head is no longer its governor. Constituents that have risen are, as mentioned above, indicated via a dashed dependency edge. The rising approach hence assumes the following structures for (9a–b):

In each case, seinen Vorschlag rises in such a manner that the projectivity violation is overcome. The ‘rising’ idea to be employed here has precedents in the dependency grammar literature: Bröker (2000) sees the relevant constituent “lifting”, Duchier and Dubesmann (2001) choose the term “climbing”, and Gerdes and Kahane (2001) opt for “emancipation”. Although there are differences in the approaches of these linguists, the underlying idea is the same: to avoid a discontinuity (projectivity violation), a flattening of structure occurs. The discussion of nonfinite verb complexes below seeks to identify when rising is and is not possible. In so doing, the basics of a theory of discontinuities in dependency grammar are established. The concept of rising just introduced receives empirical support from at least two areas. The first is the behavior of negation. The following sentence is ambiguous:

issue is inapplicable to the current system, for the current approach is like most dependency grammars insofar as it is monostratal. Monostratal grammars do not entertain derivational concepts of verb movement. If one does nevertheless opt for a derivational dependency-based approach though, then verb movement is understood much differently than in constituency grammars. Since it is always the root of the matrix clause, the finite verb itself never moves alone, but rather its dependents move and shift around it. 8 The term ’rising’ is used metaphorically. It does not necessitate that the current theory be understood as derivational. Stating that a constituent “rises” should be understood as synonymous with the statement that it “has exercised the option to attach to a node that dominates its governor”. 238 TIMOTHY OSBORNE

(10) a. Du darfst nicht essen. you may not eat

This sentence can mean either ‘You are not allowed to eat’ or ‘You are allowed to not eat’. The latter meaning requires the negation to be emphasized. Dependency grammar captures the ambiguity as follows:

The negation nicht can attach to darfst as a post-dependent or to essen as a pre-dependent. Note that if topicalization of the infinitive occurs, the sentence is no longer ambiguous; only the first meaning obtains:

(10) b. Essen darfst du nicht. ‘You are not allowed to eat.’

The ambiguity disappears because the negation no longer has the option to attach to the infinitival verb; it must attach to the finite verb. Next, consider what happens when an object is added:

(11) a. Du darfst das nicht essen. you may that not eat ‘You are not allowed to eat that.’ or ‘You are allowed to not eat that.’

b. Das essen darfst du nicht. ‘You are not allowed to eat that.’

Now the argument in favor of rising is seen in the ambiguity of (11a). On the first reading of (11a), the object must have risen. If it were not capable of rising, one would expect only the second reading to be available because nicht would be prevented from attaching to darfst by the dependency connecting das to essen.

COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 239

Since the first reading is not only possible but actually the preferred reading, we have proof that das can rise.

The9second type of evidence supporting the rising approach occurs with instances of the long passive. A widely acknowledged fact about certain verbs is that a curious case alternation can obtain – see Stechow (1990: 189ff.), S. Müller (2002: 94), Haider (2003). The following data is from Haider (2003: 97):

(12) a. dass DEN Wagen zu reparieren versucht wurde - that the car to repair tried was ‘that one tried to fix the car’

b. dass DER Wagen zu reparieren versucht wurde -

(13) a. dass uns DEN Erfolg auszukosten erlaubt wurde - Accusative case that us the success to.enjoy allowed was ‘that we were allowed to enjoy our success’

b. dass uns DER Erfolg auszukosten erlaubt wurde - Nominative case

In these examples, both nominative and accusative case is possible. At other times however, only the accusative is possible:

9 There is a third analysis of (11) that is also valid. It has das rising to attach to darfst and nicht as a dependent of essen. This analysis would mean the same thing as (11a'''). 240 TIMOTHY OSBORNE

(12) c. Es wurde versucht, den / *der Wagen zu reparieren. It was tried the the car to repair ‘One tried to repair the car.’

(13) c. Uns wurde erlaubt, den/* der Erfolg auszukosten. us was allowed the the success to.enjoy ‘We were allowed to enjoy our success.’

And yet at other times, only the nominative is possible:

(12) d. Zu reparieren versucht wurde *den/der Wagen nicht. (Haider 2003:97) to repair tried was the the car not ‘One did not try to have the car repaired.’

(13) d. Auszukosten erlaubt wurde *den/der Erfolg nicht. (Haider 2003:97) to.enjoy allowed was the the success not ‘It was not allowed to enjoy one’s success.’

The possibility of rising helps explain these data. The position of the NPs in (12a–b, 13a–b) allows for optional rising. When rising is absent, the accusative is necessary; when it is present, the nominative obtains: 10

10 Since zu and its infinitive behave as a single word in every way in German, unlike in English, the two are granted just a single node throughout the trees in this paper. COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 241

When rising is not a possibility due to the position of the infinitival phrase, the accusative is obligatory:11

11 A rising analysis of (12c) and (13c) is actually conceivable: both the object NP and the zu-infinitive would attach as post-dependents to the finite verb. The Predicate Weight Principle, however, which is presented in section 5.2, prevents this analysis. The -s object NP would be illicitly following the +s past participle as a co-sister. 242 TIMOTHY OSBORNE

And when only an analysis is possible where rising has occurred, the nominative is obligatory:

If one were to adopt a dependency grammar approach that rejected rising and in its place allowed projectivity violations, these data would be difficult to explain. The concept of rising just established is what makes the dependency grammar theory of coherence presented below possible.

3.3 Predicate chains

The three-way distinction between coherent, incoherent, and pseudoincoherent constituents mentioned in the introduction shall be grounded on the ‘predicate’ concept. The term predicate has various meanings depending on the context. In the semantic sense, i.e. in Predicate Calculus, the predicate is the central relational meaning of an utterance that relates the arguments of that utterance to each other. In the syntactic sense, in contrast, there are COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 243 basically two distinct meanings of the term. The first sees the predicate as that which is predicated of the subject. This understanding of the concept stems from antiquity and is prominent in traditional grammars. It has everything that is not the subject of a sentence qualifying as the predicate of that sentence. Most modern constituency grammars have this binary division at their cores. In Transformational Grammar for instance, the binarity is seen in the first rewrite rule of the base, i.e. S → NP VP. The other meaning takes those words in the syntax of an utterance as the predicate that correspond to the central relational meaning in the semantics of that sentence. This second meaning tends to acknowledge a three-way distinction: predicates, arguments, and adjuncts. It is interesting to note that the former meaning of the term predicate, i.e. in terms of binarity, dominates in grammars of the English language. Trask (1997: 174), for instance, provides only this definition of predicate in his dictionary of linguistics terminology. Even Quirk, Greenbaum, Leech, and Svartvik’s (1985: 79, 1118, 1398) and Huddleston and Pullum’s (2002: 25, 44, 50) comprehensive grammars of the English language prefer this understanding of the concept. In grammars of the German language in contrast, the second meaning – i.e. in terms of predicates, arguments, and adjuncts – is more prominent, e.g. Engel (1982:124ff.), Lühr (1993: 97ff.), Duden (1995: 605ff.), ZHS (1997: 659ff.), Hentschel and Weydt (2003: 338ff.). In view of the differences in the syntax of English and German, it is perhaps not surprising that the grammars of English prefer the former meaning, and that those of the German language prefer the latter. English as a strict subject-verb language lends itself to analyses in terms of binary divisions, whereas German as a V2 language is less accessible using such approaches. This paper adopts the second understanding of ‘predicate.’ The predicate of a sentence is taken to be the word(s) that correspond(s) to the central relational meaning in the semantics of that sentence. This understanding is currently prominent in the LFG and HPSG frameworks, e.g. Kathol (1998, 2000), Ackerman and Webelhuth (1998), Webelhuth and Ackerman (1999), Müller (2002). Grimshaw’s (1991) ‘extended projection’ concept is the same basic idea – see Schmid and Vogel (2004: fn. 7). Analytic verb complexes are a good orientation point in this regard. The main and auxiliary verbs of an analytic verb complex in one language that correspond to a synthetic verb form in another language qualify together as the predicate, e.g. hat...gesehen ‘has seen’, wird...gesehen haben ‘will have 244 TIMOTHY OSBORNE seen’, ist gesehen worden ‘has been seen’, wird gesehen worden sein ‘will have been seen’, etc. In dependency grammar, the verbs of an analytic verb complex always form a chain – see O’Grady (1998) and Osborne (2005a). A chain is a word combination that is continuous on the vertical axis.

The words of the predicates are in italics. Each of these predicates is a chain because it is top-down continuous. For instance, werden in (14d) immediately dominates sein, sein immediately dominates worden, and worden immediately dominates gesehen. Note that hat and gesehen in (14a) and wird and gesehen haben in (14b) are not left-to-right continuous, i.e. they are discontinuous on the horizontal axis because uns in each case intervenes. The fact that the words of predicates are always continuous on the vertical axis is important. It means that the analysis of predicates can refer to the relative position of a predicate element within the predicate chain. The importance of this point will become evident in sections 5.1 and 5.2, where the Scrambling Principle and the Predicate Serialization Principle are presented. COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 245

3.4 Coherence defined in dependency grammar

The discussion has now reached the point where the specific terminology to be employed for the dependency grammar analysis of coherence can be presented. The three-way distinction mentioned in the introduction is expressed here in terms of words.

Coherent word: a word that is part of the matrix predicate chain. Incoherent word: a word that is not part of the matrix predicate chain. Pseudoincoherent word: a word that is not actually part of the matrix predicate chain but that behaves as if it were in certain ways.

And given these definitions, the following definitions are straightforward:

Coherent constituent: A constituent the root of which is part of the matrix predicate chain. Incoherent constituent: A constituent the root of which is not part of the matrix predicate chain. Pseudoincoherent constituent: A constituent the root of which is not actually part of the matrix predicate chain but that behaves as if it were in certain ways.

These definitions shall be illustrated with the help of the following trees. The predicate chains are in italics.

246 TIMOTHY OSBORNE

According12to the definitions, die Kinder unterhalten and die Kinder unterhalten wollen are coherent constituents in (15). Wir, die, and die Kinder are incoherent constituents in (15). In (16), die Kinder versuchen zu unterhalten is a coherent constituent, and wir, die, and die Kinder are again incoherent constituents. The interesting constituent in (16) is zu unterhalten; it is pseudoincoherent. This point will become clear as the discussion continues. There are numerous types of coherent constituents. Their roots are bare infinitives, zu-infinitives, non-adjectival , and predicative elements of various sorts. Concerning coherent and pseudoincoherent constituents, the discussion below focuses on zu-infinitives in accordance with Bech’s analysis of nonfinite verb complexes. Zu-infinitives can, namely, be the roots of coherent, incoherent, and pseudoincoherent constituents. The following verbs among others subcategorize for coherent zu-infinitives:

bekommen ‘receive’, bleiben ‘stay’, nicht brauchen ‘not need’, drohen ‘threaten’, gedenken ‘think of’, haben ‘have’, sein ‘be’, versprechen ‘promise’, wissen ‘know’13

12 An alternative analysis of (16) has die Kinder rising to attach to werden, not to versuchen. Sentence (i) demonstrates that the analysis shown is plausible, and (ii) shows that the alternative analysis is also plausible: (i) Die Kinder versuchen zu unterhalten werden wir (schon). (ii) Versuchen zu unterhalten werden wir die Kinder. 13 According to Prinzhorn (1990: 200), the type of dependents that drohen and versprechen take depends on the subject. With non-agent subjects, these verbs take coherent constituents. With agent subjects, they take (pseudo)incoherent constituents. COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 247

The following list is a sample of the verbs that subcategorize for incoherent zu-infinitives:

andeuten ‘indicate’, anflehen ‘beseech’, anspornen ‘spur’, antreiben ‘encourage’, anweisen ‘instruct’, auffordern ‘ask’, beauftragen ‘commission’, bedauern ‘regret’, befähigen ‘empower’, begehren ‘desire’, beneiden ‘envy’, bereuen ‘regret’, bestechen ‘bribe’, bewegen ‘move’, bitten ‘ask’, drängen ‘press’, einladen ‘invite’, erinnern ‘remember’, ermutigen ‘embolden’, ersuchen ‘request’, fortfahren ‘continue’, hindern ‘hinder’, locken ‘lure’, motivieren ‘motivate’, nötigen ‘compel’, überreden ‘convince’, überzeugen ‘convince’, veranlassen ‘arrange’, verführen ‘lead astray’, verhindern ‘prevent’, verlangen ‘demand’, verleiten ‘lead astray’, verlocken ‘entice’, verpflichten ‘oblige’, vorschlagen ‘suggest’, verzichten ‘do without’, zurückhalten ‘hold back’, zögern ‘hesitate’, zwingen ‘force’

And the following list is a sample of the verbs that subcategorize for incoherent constituents, but that allow these constituents to be pseudoincoherent.

anfangen ‘start’, beabsichtigen ‘intend’, beginnen ‘begin’, beschließen ‘decide’, drohen ‘threaten’, erlauben ‘allow’, erwägen ‘consider’, fürchten ‘fear’, gestatten ‘allow’, hoffen ‘hope’, lernen ‘learn’, lieben ‘love’, meinen ‘think’, planen ‘plan’, streben ‘strive’, überlassen ‘let have’, vergessen ‘forget’, vermögen ‘can’, verstehen ‘understand’, versprechen ‘promise’, versuchen ‘try’, vorhaben ‘plan’, wagen ‘dare’, wähnen ‘think’, wünschen ‘wish’

These groupings have been compiled based on the examples appearing in the literature – see Stechow (1984), Grewendorf (1988: Ch. 12), Prinzhorn (1990: 202), Grewendorf and Sabel (1994), Hinrichs and Nakazawa (1998: 125), ZHS (1998: 2198), Meurers (1999), Sabel (1999: 420), De Kuthy and Meurers (2001), Wöllstein-Leisten (2001: 58). There are some widely acknowledged tendencies concerning the syntax and semantics of these groupings. Subject control verbs can take pseudoincoherent zu-infinitives, whereas object control verbs tend to take incoherent zu-infinitives. At times there is disagreement about the classification of a given verb and grammaticality judgments vary – in this area, see the comments of Fanselow (1989: n. 6) and Hinrichs and Nakazawa (1994: 13, 1998: 125). The difficulties in classifying many verbs suggest the distinction is probably gradient.14

14 A good example of the difficulty to classify various verbs is seen in De Kuthy and Meurers (2001: 155, 158f.) and Hinrichs and Nakazawa (1998: 125). The former 248 TIMOTHY OSBORNE

Syntactic factors also influence the type of infinitival dependent that a verb subcategorizes for. Hinrichs and Nakazawa (1998: 125) and Sabel (1999: 422f.) note that accusative object control verbs necessarily subcategorize for incoherent zu-infinitives. In addition, Kvam (1982: 340) and Wöllstein-Leisten (2001: 60) observe that the appearance of a correlative element with the matrix verb also forces incoherence.

(17) a. Sie hat uns verpflichtet, unsere Sachen zu organisieren. She has us required our things to organize ‘She required us to organize our things.’

b. *Sie hat uns unsere Sachen verpflichtet zu organisieren.

(18) a. Sie wird es vergessen, mich anzurufen. She will it forget me to.call ‘She will forget to call me.’

b. Sie wird mich vergessen anzurufen.

c. *Sie wird es mich vergessen anzurufen.

The analysis of examples such as these will become clear below. For now one should note that the appearance of an accusative object or correlative element with the matrix verb prevents the third construction. These syntactic factors influencing coherence can be utilized to reduce the difficulties associated with varying grammaticality judgments. Accordingly, this paper sticks mainly to accusative object control verbs to demonstrate incoherence. Finally, note that the definitions above are not limited in application to infinitival verbs. This means that all constituents – be their roots verbs, nouns, prepositions, adverbs, etc. – can be classified in terms of coherence. This point is important because it enables the notion of coherence to be extended to all words and constituents, regardless of their syntactic category. Indeed, section 6 demonstrates that the theory of coherence can shed light on the behavior of certain discontinuous noun phrases.

produce more than one example illustrating that empfehlen constructs incoherently, whereas the latter list empfehlen as a verb that allows the third construction. COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 249

4. Diagnostics for coherence

The validity of the three-way distinction – coherent vs. incoherent vs. pseudoincoherent constituent – is established in the following sections. Nine diagnostics are used:

1. Extraposition 2. Intraposition 3. Bare infinitive fronting 4. Infinitival fronting 5. Scrambling 6. Pied-piping 7. Position of negation 8. Scope of negation 9. Gapping

Coherent constituents behave much differently than incoherent constituents, whereas pseudoincoherent constituents demonstrate traits of both coherent and incoherent constituents simultaneously.

4.1 Extraposition

Perhaps the easiest and most widely employed test for coherence is extraposition, e.g. Bech (1955: 84), Kvam (1982: 337ff.), Askedal (1983: 182), Fanselow (1989: 3f.), Prinzhorn (2000: 201), Stechow (1990: 148), Reape (1994: 166ff.), Kiss (1995: 30), Meurers (1999: 20), Müller (2002: 42f.). If an infinitival phrase can appear to the right of its governor, then a (pseudo)incoherent constituent is present. In the examples here and further below, the relevant constituent is italicized and its status is given on the right. When the object NP of the zu-infinitive is not italicized with its governor, it means that object NP has risen.

(19) a. *weil niemand hat ein Lied gesungen - Coherent because no.one has a song sung ‘because no one sang a song’

b. weil niemand ein Lied versuchte, zu singen - Pseudoincoherent because no.one a song tried to sing ‘because no one tried to sing a song’

c. weil niemand mich zwang, ein Lied zu singen - Incoherent because no.one me forced a song to sing 250 TIMOTHY OSBORNE

Coherent constituents cannot be extraposed. Thus ein Lied gesungen in (19a) is a coherent constituent. Incoherent constituents can, in contrast, be extraposed. Hence the constituents ein Lied zu singen in (19c) is an incoherent constituent. Example (19b) has zu singen as a pseudoincoherent constituent: the zu-infinitive itself is extraposed, but its object has risen to appear in the Mittelfeld.

4.2 Intraposition

A less widely employed, but quite convenient, test for coherence is intraposition, e.g. Fanselow (1989: 3), G. Müller (1998: 24), De Kuthy and Meurers (2001: 155f.). If a constituent can be separated from its governor in the Mittelfeld, then it is incoherent.

(20) a. *weil das Lied gesungen niemand hat - Coherent because the song sung no.one has ‘because no one sang the song’

b. weil das Lied zu singen niemand versuchte15 - Incoherent because the song to sing no.one tried ‘because no one tried to sing the song’

c. weil das Lied zu singen niemand mich zwang16 - Incoherent because the song to sing no.one me forced ‘because no one forced me to sing the song’

The governor of the italicized constituent in each case is the finite verb. Since the subject intervenes between the two, intraposition obtains. Coherent constituents may not be separated from their governors in this manner, as illustrated in (20a).

15 Similar examples from the literature: (i) dass das Buch zu lesen keiner versucht hat (G. Müller 1998: 24) (ii) Er wird das Pferd zu verkaufen noch heute versuchen. (De Kuthy & Meurers 2001) 16 Similar example from the literature: (i) Er wird das Pferd zu verkaufen ihr noch heute empfehlen. (De Kuthy & Meurers 2001) COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 251

4.3 Bare infinitive fronting

A widely observed, but poorly understood, trait of predicate chains is that an intermediate link of a predicate chain may not be fronted alone. These intermediate links are usually bare infinitives, hence the term bare infinitive fronting. This peculiarity of predicate chains is sometimes used as a test for coherence, e.g. Kiss (1995: 31), De Kuthy and Meurers (2001), Müller (2002: 44).

(21) a. *Wollen wird niemand das Lied singen17 - Coherent

b. Versuchen wird niemand das Lied zu singen18 - Incoherent/pseudoincoherent19

c. Zwingen wird niemand mich das Lied zu singen - Incoherent

Data like (21a) occur frequently in the literature, e.g. Engel (1982: 225), Olszok (1983: 109), Fanselow (1987: 93), Grewendorf (1988: 300f.), Nerbonne (1994: 118, 136ff.), ZHS (1997: 1623ff), Bouma and Van Nord (1998: 62f.), Kathol (1998: 230f., 2000: 205), Meurers (1999: 245), Webelhuth and Ackerman (1999). The ungrammaticality of (21a) obtains because wollen is an intermediate link in the predicate chain wird-wollen- singen. Versuchen in (21b) and zwingen in (21c), in contrast, are not intermediate links in their respective predicate chains, but rather they are the terminal links; they can hence be fronted alone. This situation demonstrates that das Lied zu singen in (21b) and in (21c) are (pseudo)incoherent constituents. Section 5.2 has more to say about this aspect of predicate chains.

4.4 Infinitival verb fronting

A test similar to bare infinitive fronting is infinitival fronting. An infinitival verb that subcategorizes for an object complement may not be fronted alone without its complement if it is the root of an incoherent constituent.

17 Similar example from the literature: (i) *Gewesen ist er auf seine Kinder stolz. (De Kuthy & Meurers 2001) 18 Similar examples from the literature: (i) Zwar vermochten mich seine Ergebnisse nicht zu befriedigen. (Kiss 1995: 27) (ii) Versuchen wird er, das Pferd zu verkaufen. (S. Müller 2002: 44) 19 The example allows two analyses, i.e. one in terms of incoherence where the object NP has not risen and one in terms of pseudoincoherence where the object NP has risen. 252 TIMOTHY OSBORNE

Fanselow (1989: 4), Grewendorf and Sabel (1994: 265), Kiss (1994: 96), G. Müller (1998: 18) and De Kuthy and Meurers (2001) use this aspect of infinitival verb phrases as a test for coherence.

(22) a. Gesungen hat niemand das Lied. - Coherent

b. Zu singen hat niemand das Lied versucht.20 - Pseudoincoherent

c. *Zu singen hat niemand mich das Lied gezwungen.21 - Incoherent

Das Lied in each case is the complement of the fronted infinitival verb. When this fronted verb is incoherent, ungrammaticality obtains.

4.5 Pied-piping

Pied-piping in relative clauses is a widely employed diagnostic for coherence, e.g. Bech (1955: 79), Kvam (1982: 338, 350), Askedal (1983: 182f.), Grewendorf (1988: 271), Stechow (1990: 148), Grewendorf and Sabel (1994: 275f.), Kiss (1994: 71,1995:30f.), Hinrichs and Nakazawa (1998: 120f.), ZHS (1998: 2191), Meurers (1999: 21), Müller (2002: 42, 70). If the can pied-pipe its governor, an incoherent constituent is present:

(23) a. *das Lied, das gesungen niemand hat - Coherent the song that sung no.one has ‘the song that no one sang’

b. das Lied, das zu singen niemand versucht hat - Incoherent the song that to sing no.one tried has ‘the song that no one tried to sing’

c. das Lied, das zu singen niemand mich gezwungen hat - Incoherent the song that to sing no.one me forced has ‘the song that no one forced me to sing’

20 Similar examples from the literature: (i) Zu füttern versucht hat den Hund keiner. (Grewendorf & Sabel 1994: 265) (ii) Zu verkaufen versuchte er das Pferd. (De Kuthy & Meurers 2001) 21 Similar examples from the literature: (i) *Zu stören aufgefordert hat Max mich ihn. (Fanselow 1989: 4) (ii) *Zu füttern gezögert hat den Hund keiner. (Grewendorf & Sabel 1994: 265) (iii) *Zu lesen hat das Buch keiner abgelehnt. (G. Müller 1998: 18) COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 253

The ungrammaticality in (23a) obtains because the relative pronoun may not pied-pipe its governor if that governor is coherent. If that governor is incoherent though, as in (23b–c), then pied-piping is fine.

4.6 Scrambling

Scrambling is also commonly used to identify coherence, Bech (1955: 61f., 74f.), Fanselow (1989: 3), Prinzhorn (2000: 203), Grewendorf and Sabel (1994: 264f.), Reape (1994: 171ff.), ZHS (1998: 2192), Meurers (1999: 21), De Kuthy and Meurers (2001: 154f.), Wöllstein-Leisten (2001:12), Müller (2002: 40f.).

(24) a. weil das Lied niemand gesungen hat - Coherent

b. weil das Lied niemand zu singen versucht hat22 - Pseudoincoherent

c. *weil das Lied niemand mich zu singen gezwungen hat23 - Incoherent

The object complement of an incoherent infinitival verb may not be scrambled out from under that infinitival verb, as illustrated in (24c). When the infinitival verb is coherent or pseudoincoherent however, as in (24a) and (24b), the result is fine.

4.7 Position of negation

The position of the negation in a sentence can be used as a diagnostic for coherence, e.g. S. Müller (2002: 41f.). The negation may not split left- predicate chains.

(25) a. *weil Thomas das Lied gesungen nicht hat - Coherent because Thomas the song sung not has ‘because Thomas has not sung the song’

22 Similar examples from the literature: (i) dass das Buch keiner zu lesen versucht hat (G. Müller 1998: 17) (ii) dass das Pferd keiner zu verkaufen versucht hat (De Kuthy & Meurers 2001) (iii) weil es ihm jemand zu lesen versprochen hat (S. Müller 2002: 40) 23 Similar examples from the literature: (i) ?*dass das Buch keiner zu lesen abgelehnt hat (G. Müller 1998: 17) (ii) *Noch heute wird es der Mann zu verkaufen empfehlen. (De Kuthy & Meurers 2001) 254 TIMOTHY OSBORNE

b. weil Thomas das Lied zu singen nicht versucht hat - Incoherent because Thomas the song to sing not tried has ‘because Thomas did not try to sing the song’

c. weil Thomas mich das Lied zu singen nicht gezwungen hat- Incoherent because Thomas me the song to sing not forced has ‘because Thomas did not force me to sing the song’

Nicht in (25a) separates gesungen from hat; this situation results in ungrammaticality because gesungen and hat are links of a single predicate chain. Based on this observation, zu singen in (24b) and in (25c) is not part of the matrix predicate chains and is hence incoherent.

4.8 Scope of negation and cohesion

A similar test for coherence using negation places the negation immediately in front of all the verbs – see Kiss (1994: 74f.), S. Müller (2002: 40). If the matrix main verb cannot be negated in doing so, then one has incoherence.

(26) a. weil Thomas das Lied nicht gesungen hat - Coherent

b. weil Thomas das Lied nicht zu singen versucht hat - Pseudoincoherent

c. (*)weil Thomas mich das Lied nicht zu singen gezwungen hat - Incoherent

Sentence (26b) is actually ambiguous, although the reading indicated by the italics has the matrix verb versucht, not the embedded verb zu singen, being negated. Sentence (26c) is allowed on the reading where zu singen is negated. On the other reading however, i.e. where gezwungen is negated, it is disallowed. The scope of cohesion is the same test. Cohesion obtains when a negation and an indefinite expression combine into a single word, e.g. nicht ein = kein, nicht jemand = niemand, nicht etwas = nichts, etc. Cohesion is a widely employed diagnostic for coherence, e.g. Grewendorf (1988: 270f.), Fanselow (1989: 4), Stechow (1990: 147).

(27) a. weil Thomas kein Lied gesungen hat - Coherent because Thomas no song sung has ‘because Thomas did not sing a song’ COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 255

b. weil Thomas kein Lied zu singen versucht hat - Pseudoincoherent because Thomas no song to sing tried has ‘because Thomas did not try to sing a song’

c. (*)weil Thomas mich kein Lied zu singen gezwungen hat - Incoherent because Thomas me no song to sing forced has ‘because Thomas did not force me to sing a song’

The same result obtains in (27) as in (26). Kein negates the matrix verb gesungen in (27a). On the reading indicated by the italics, kein negates the matrix verb versucht in (27b). In (27c) however, kein cannot negate the matrix verb gezwungen, but rather it is limited to negating the embedded verb zu singen.

4.9 Gapping

A test for coherence that I have not encountered in the literature is gapping. The gap of a gapped conjunct may not ‘cut into’ an incoherent constituent. It may, however, cut into a (pseudoin)coherent constituent:

(28) a. weil sie ihn besucht hat, und er sie. - Coherent because she him visited has and he her ‘because she visited him, and he her’

b. weil sie ihn zu besuchen versucht hat, und er sie.-Pseudoincoherent because she him to visit tried has and he her ‘because she tried to visit him, and him her’

c. *weil sie ihn zu besuchen vorgeschlagen hat, und er sie.- Incoherent because she him to visit suggested has and him her ‘because she suggested to visit him, and him her’

The relevant constituents are in italics. When this constituent is coherent or pseudoincoherent, the embedded infinitival verb can be ‘gapped’, e.g. besucht in (28a) and zu besuchen in (28b). When that infinitival verb is the root of an incoherent constituent however, the embedded infinitival verb may not be gapped, e.g. zu besuchen in (28c). 256 TIMOTHY OSBORNE

4.10 Summary of diagnostics

To summarize the results of the diagnostics, the discussion takes a closer look at pseudoincoherent constituents. Pseudoincoherence obtains when the object of what would be an incoherent zu-infinitive under normal circumstances has risen. The following examples illustrate: 24

Sentence (31) illustrates the extraposition associated with incoherent constituents – zu spielen appears to the right of versprochen – as well as the scrambling associated with coherent constituents – das Lied is nonadjacent to its governor zu spielen. Sentence (32) shows the scrambling of coherent constituents – es is non-adjacent to its governor zu spielen – as well as the bare infinitive fronting possible of incoherent constituents – wagen is in the initial position without its dependent es zu spielen. In other words, each of the examples (31–32) displays the traits of a coherent and an incoherent constituent simultaneously. The term pseudoincoherent constituent is suited to capture this unique behavior.

24 The tree shows das Lied as attaching to versprochen. An alternative analysis has it attaching to hatte. I have chosen the analysis shown due to the acceptability of Das Lied versprochen zu spielen hatte er wahrscheinlich. See footnote 12. COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 257

5. Two features

Two features are necessary to capture the three-way distinction: ±s (scrambling) and ±p (predicate). The feature ±s is analogous to the ±LEX feature employed in HPSG accounts, e.g. Hinrichs and Nakazawa (1994), Nerbonne (1994), Meurers (1999), De Kuthy and Meurers (2001). It is associated with the possibility of scrambling. A +s word is transparent for the rising of scrambling, whereas a -s word is a barrier to the rising of scrambling. “Transparent for rising” means that the dependents of that node have the option to rise. The feature ±p is associated with the ability of a constituent to be extraposed and to occur with bare infinitive fronting. A +p word is subject to more positional restrictions than a -p word. An overview of these features is as follows:

coherent word +s +p pseudoincoherent word +s -p incoherent word -s -p25

Coherent words have the feature combination +s +p, which means they are transparent for the rising of scrambling and adhere to strict positional restrictions. Incoherent words are -s -p, which means they are barriers to the rising of scrambling and do not adhere to strict positional restrictions.

25 Notice that the feature combination -s +p is not included in the table. I believe that this combination is actually manifest on certain words, for instance on some prepositions and predicate nouns. Unlike English, German disallows preposition stranding. This fact is witness to the inherent trait of prepositions as -s words in German. Thus if a preposition is included in the predicate, it will have the feature values -s +p. The preposition of a predicate PP illustrates: (i) a. Alles soll unter dem Bett sein. everything should under the bed be ‘Everything should be under the bed.’ b. *Unter soll das dem Bett sein. - Fronting/scrambling b. *Alles soll sein unter dem Bett. - Extraposition c. *weil unter dem Bett alles sein soll - Intraposition e. *weil dem Bett alles unter sein soll - Scrambling f. das Bett, unter dem alles sein soll - Pied-piping (obligatory) g. *Alles soll unter dem Bett nicht sein. - Position of negation These data are just as they should be if unter is -s +p. The non-possibility of scrambling is consistent with -s. The non-possibility of fronting, extraposition, and intraposition, and the necessity of pied-piping as well as the position of negation are all consistent with +p. 258 TIMOTHY OSBORNE

Since pseudoincoherent words demonstrate the traits of both other types of words, they have the combination +s -p. The distribution of these features is determined in part by generalizations across broad syntactic categories. The rule of coherence – see Bech (1955: 68) and Stechow (1990: 150) – states that first and third status supina, i.e. bare infinitives and predicate participles, necessarily construct coherently. In the current system, this means they have the feature combination +s +p. The features of second status supina however, i.e. complement zu-infinitives, are, in contrast, in part determined by the subcategorization traits of their governors.26 Most of these governors subcategorize for a -s -p zu-infinitive, e.g. ablehnen ‘reject’, auffordern ‘encourage’, zwingen ‘force’. Some, however, subcategorize for +s +p zu-infinitives, e.g. nicht brauchen ‘not need’, sein ‘be’, wissen ‘know’. And some – e.g. beginnen ‘begin’, versuchen ‘try’, wagen ‘dare’ – have the option to subcategorize for -s -p or +s -p zu-infinitives. The following sections examine these features. The feature ±s is behind the Scrambling Principle, and the feature ±p is behind the Predicate Serialization Principle. Both features are behind the Predicate Weight Principle.

5.1 The Scrambling Principle

The ability of a constituent to take part in scrambling is expressed as the Scrambling Principle:

Scrambling Principle: If a word has the feature +s, then its dependents can rise to non-fronting (as well as fronting) positions.

An important thing to acknowledge about the +s feature is that in many instances the dependents of a +s word have the option to rise. In other instances however, rising is forced due to the need to overcome a

26 Note that subject and adjunct zu-infinitives never allow scrambling – see Askedal (1983): (i) a. Die Zeitung zu lesen verlangt viel Zeit. the newspaper to read demands much time ‘Reading the newspaper requires much time.’ b. *Zu lesen verlangt die Zeitung viel Zeit. (ii) a. Er wird kommen, uns zu helfen. he will come us to help ‘He will come help us.’ b. *Er wird uns kommen zu helfen. COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 259 projectivity violation. When the rising and non-rising analyses are both possible, more than one structure is possible. Sentence (33) is such a case; the analysis shown here is the non-rising one: 27

Mich and nach Japan can be scrambled because geflogen is a +s category. In each of (33a–f), the position of mich and/or nach Japan with respect to its/their governor geflogen apparently qualifies as a discontinuity (projectivity violation). It is in such cases that the rising of scrambling is necessary, i.e. mich and/or nach Japan has/have taken on a node above its/their governor as its/their head.

27 The unacceptability of (33b) is due to the relative ‘weight’ of the dependents involved. A heavier dependent cannot rise if its lighter sister dependent does not also rise. In the case of (33b), the heavier nach Japan cannot rise if the lighter mich does not also rise. The V2 principle, however, prevents mich from rising. This curious aspect of fronted nonfinite verbs has been noticed by others, e.g. Heidolph et al. (198: 721), Fanselow (1987: 94f.), Hinrichs and Nakazawa (1994: 94), Nerbonne (1994: 112f.), ZHS (1997: 1634), Eroms (2000: 368), Müller (2002: 95f.). More discussion on this further below. 260 TIMOTHY OSBORNE

Given the understanding of rising expressed by the Scrambling Principle, the V2 principle of matrix declarative clauses is easily formulated. One assumes that a single dependent of the root word, and only a single dependent, is allowed to precede the root. In this manner, the V2 principle easily predicts the ungrammaticality of the following sentences among others:

(33) g. *Nach Japan Karsten hat mich geflogen. h. *Mich Karsten hat nach Japan geflogen. i. *Geflogen Karsten hat mich nach Japan.

Since the subject is obligatorily a dependent of the finite verb, these sentences each have two pre-root dependents, a situation that V2 disallows. Returning to the three-way distinction, incoherent words are barriers to the rising of scrambling. In the following examples, gedrängt itself is a +s category, but it assigns -s to zu sagen: COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 261

28 29

Examples (34b–f) are each disallowed because alles has been scrambled, i.e. it has risen to a non-fronting position. Note that the rising of Tanja in (34d, f) is not the source of the ungrammaticality since gedrängt is a +s word. Since pseudoincoherent words are +s like coherent words, they, unlike incoherent words, are transparent for scrambling. In the following examples, versucht is as a predicate participle a +s category that has the option to assign +s or -s to zu lesen. When rising has occurred, versucht has assigned +s to zu lesen:

28 Tanja and alles zu sagen are both shown as dependents of gedrängt. Another possibility is that Tanja is viewed as rising to attach to hat. I have chosen the analysis due to the acceptability of Tanja gedrängt alles zu sagen hat man schon. 29 Tanja and alles are both shown as dependents of gedrängt. Another possibility would be to have both as dependents of hat. Since either way alles must rise, the Scrambling Principle is violated on both analyses. 262 TIMOTHY OSBORNE

There30is an aspect of weight that is essential for the analysis of scrambling. Example (36) appeared above as (33) – see footnote 27.

The a-sentences show unmarked order. The b-sentences violate the principle of ‘weight’: nonstressed definite pronouns are ‘lighter’ than full noun phrases and PPs, meaning the pronoun must precede the noun phrase. The c-sentences demonstrate that the heavier dependent without the lighter

30 Den Brief is shown as a daughter of versucht due to the acceptability of Den Brief versucht zu lesen hat nur einer. In this regard, note the ungrammaticality of *Den Brief versucht hat nur einer zu lesen. This permutation is disallowed because den Brief must attach to versucht; it cannot attach to hat due to the V2 principle. As a dependent of versucht, it has not risen. The Scrambling Principle allows dependents to rise, meaning the head of the risen dependent must dominate the governor of that dependent. COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 263 dependent can be fronted together with the nonfinite verb. And the d- sentences demonstrate that the lighter dependent without the heavier dependent may not be fronted together with the nonfinite verb. This curious aspect of fronted infinitives is acknowledged elsewhere, e.g. Heidolph et al. (1981: 721), Fanselow (1987: 94f.), Hinrichs and Nakazawa (1994: 94), Nerbonne (1994: 112f.), ZHS (1997: 1634), Eroms (2000: 368), Wöllstein-Leisten (2001: 106ff., 218ff.), Müller (2002: 95f.). The current system accounts for these cases by referencing rising and the relative weight of the risen dependent(s). Among the dependents of a governor, a heavier dependent may not rise if a lighter one does not also rise. This aspect of rising is added to the Scrambling Principle as part (ii).

Scrambling Principle (i) If a word has the feature +s, then its dependents can rise to non-fronting (as well as fronting) positions. (ii) A dependent may not rise to a non-fronting position if its governor has a lighter dependent that does not also rise.

Examine the trees of (36d, 37d):

Example (36d') violates part (ii) of the Scrambling Principle because the heavier nach Japan does, but the lighter mich does not, rise. Similarly, (37d') violates part (ii) of the Scrambling Principle because the heavier dem Kind does, but the lighter es does not, rise. Additional support for part (ii) of the Scrambling Principle is found with the third construction. The heavier constituent may not scrambling rise to the Mittelfeld unless the lighter constituent also rises:

(38) a. Karsten hat mich vesucht, nach Japan zu fliegen. Karsten has me tried to Japan to fly ‘Karsten tried to fly me to Japan.’

b. ??Karsten hat nach Japan versucht, mich zu fliegen.

c. Karsten hat mich nach Japan versucht zu fliegen. 264 TIMOTHY OSBORNE

(39) a . Er wird es versuchen, dem Kind zu schenken. He will it try the child to give ‘He will try to give it to the child.’

b. *Er wird dem Kind versuchen, es zu schenken.

c . Er wird es dem Kind ve rsuc h en zu schenken.

(40) a . Er dürfte ihr beginnen, viele sehr schöne und teure Blumen zu senden. He may her begin many very nice and expensive flowers to send ‘He will probably begin sending her many very nice and expensive flowers.’

b. *Er dürfte viele sehr schöne und teure Blumen beginnen, ihr zu senden.

c. Er dürfte ihr viele sehr schöne und teure Blumen beginnen zu senden.

Fronted infinitivals taking pseudoincoherent zu-infinitives provide another source of support for part (ii) of the Scrambling Principle.

(41) a. Euch versuchen alle Geheimnisse mitzuteilen dürfen wir nicht. you try all secrets to.tell may we not ‘We are not allowed to try to tell you all the secrets.’

b. *Alle Geheimnisse versuchen euch mitzuteilen dürfen wir nicht.

(42) a. Ihr wagen etw as Negatives zu sagen sollte man nicht. her risk something negative to say should one not ‘One should not dare to say something negative to her.’

b. *Etwas Negatives wagen ihr zu sagen sollte man nicht.

The a-sentences have the lighter constituent rising to attach as pre- depen d ents to the bare infinitives, whereby the heavier constituent remains a dependent of the zu-infinitive. In the b-sentences, the opposite situation obtains .31

31 The following cases discussed by Fanselow (1987: 94f.) challenge part (ii) of the Scrambling Principle: (i) a. Sie hat das Brot in ihrer blauen Schürze gebacken. b. *?In ihrer blauen Schürze gebacken hat sie das Brot. (ii) a. Sie hat das Brot besoffen gebacken. b. *?Besoffen gebacken hat sie das Brot. COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 265

To conclude this section, a word of caution is necessary about those instances of rising that do and do not qualify as scrambling. As Grewendorf and Sabel (1994: 293), G. Müller (1998: 18, fn.17) observe, fronting rising (e.g. topicalization, w-fronting, relative pronoun fronting) is much more liberal than scrambling rising. In this regard, the Scrambling Principle is intended to address only those instances of rising that traditionally fall under the rubric of scrambling. It says nothing about fronting and extraposition discontinuities. The discussion returns to aspects of weight in section 5.3.

5.2 The Predicate Serialization Principle

The extraposition and bare infinitive fronting data from sections 3.1.1 and 3.1.3 demonstrate that with respect to positional restrictions, coherent constituents behave much differently than incoherent and pseudoincoherent constituents. Coherent constituents may not be extraposed over their predicate governors, nor may the governor of a coherent constituent be fronted without its predicate daughter. Osborne (2005b) presents a principle of serialization that predicts these cases. He calls it the Predicate Serialization Principle. This principle is expressed here in terms of the feature ±p (=predicate):

Predicate Serialization Principle: If X and Y are +p words, X is Y’s governor, and X is not a matrix root, then Y must precede X. (Adapted from Osborne (2005b))

The Predicate Serialization Principle is a far-reaching principle of word order in German. It predicts the ungrammaticality of the b- and c-sentences in the following cases:32, 33

These examples appear to contradict part (ii) of the Scrambling Principle. The a- sentences indicate that in ihrer blauen Schürze in (i) and besoffen in (ii) can be heavier than das Brot. This situation predicts that (ib) and (iib) should be acceptable. Since they are bad, part (ii) of the Scrambling Principle is challenged. The answer to this challenge is with a theory of adjunct placement. If one assumes that the adjuncts in question are sentence adjuncts, as opposed to verb adjuncts, and must therefore attach to the roots of the predicate chains, and not to their lowest links, then (ib) and (iib) violate the V2 principle; the ungrammaticality is thus expected. 32 Osborne (2005b) emphasizes that there are two exceptions to the Predicate Serialization Principle that have received a lot of attention in the literature. These exceptions occur with the auxiliary flip associated with the Oberfeld ‘upper field’ – see 266 TIMOTHY OSBORNE

(43) a. Sie hat schlafen dürfen. - Perfect active with modal verb she has sleep be.allowed ‘She was allowed to sleep.’

b. *Sie hat dürfen schlafen.

c. *Dürfen hat sie schlafen.

(44) a . Das Fahrrad wird repariert werde n. - Future passive the bicycle will repaired be ‘The bicycle will be repaired.’

b. *Das Fahrrad wird werden repariert.

c. *Werden wird das Fahrrad repariert.

(45) a . Er ist gesehen worden. - Perfect passive he is seen been ‘He has been seen.’

b. *Er ist worden gesehen.

c. *Worden ist er gesehen.

(46) a . Sie hat alles geschenkt bekommen. - Bekommen passive she has all given received ‘She received everything as a gift.’

b. *Sie hat alles bekommen geschenkt.

c. *Bekommen hat sie alles geschenkt.

(47) a . Die Wohnung kann wohl renoviert sein. - Statal passive the apartment can certainly renovated be ‘The apartment can certainly have been renovated.’

Hinrichs and Nakazawa (1989, 1994). These constructions generally involve the double infinitive associated with the Ersatzinfinitiv ‘replacement infinitive’ (IPP - infinitivus pro participio). The discussion here shall not examine these cases. 33 According to Schmid and Vogel (2004), many Swiss dialects prefer the verb ordering Aux-Modal-Main, e.g. hat dürfen schlafen for (43). Hence the Predicate Serialization Principle is invalid for these dialects. COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 267

b . *Die Wohnung kann wohl sein renoviert.

c. *Sein kann die Wohnung wohl renoviert.

(48) a. Sie ist stolz gewesen. - Perfect with copular verb she is proud been ‘She has been proud.’

b. *Sie ist gewesen stolz.

c. *Gewesen ist sie stolz.

(49) a. Der Hund wird dein bester Freund sein. - Future with copular verb the dog will your best friend be ‘The dog will be your best friend.’

b . *Der Hund wird sein dein bester Freund.

c. *Sein wird der Hund dein bester Freund.

(50) a. Sie scheint müde zu sein. - Modality verb with copular verb she seems tired to be ‘She seems to be tired.’

b . *Sie scheint zu sein müde.

c. *Zu sein scheint sie müde.

(51) a. Niemand will ihn anrufen. - Separable prefixes no.one wants him to.call ‘No one wants to call him.’

b . *Niemand will ihn rufen an.

c. *Rufen will niemand ihn an.

(52) a. Sie konnte uns spielen hören. - Aci structures she could us play hear ‘She could hear us playing.’

b . *Sie konnte uns hören spielen.

c. *Hören konnte sie uns spielen.

268 TIMOTHY OSBORNE

(53) a. Sie muss ihn gehen lassen. - lassen+infinitive structures she must him go let ‘She must let him go.’

b . *Sie muss ihn lassen gehen.

c. *Lassen muss sie ihn gehen.

(54) a. Wir müssen morgen einkaufen gehen. - Gehen plus infinitive we must tomorrow shopping go ‘We have to go shopping tomorrow.’

b . *Wir müssen morgen gehen einkaufen.

c. *Gehen müssen wir morgen einkaufen.

(55) a. Sie hat gut ausgesehen. - Subject predicates she has good looked ‘She looked good.’

b . *Sie hat ausgesehen gut.

c. ??Ausgesehen hat sie gut.

(56) a. Das Wetter hat uns traurig gemacht. - Object predicates the weather has us sad made ‘The weather has made us sad.’

b. *Das Wetter hat uns gemacht traurig.

c. *Gemacht hat uns das Wetter traurig.

(57) a. Sie will das in Anspruch nehmen. - Function verbs she wants that in claim take ‘She wants to lay claim to that.’ b . *Sie will das nehmen in Anspruch. c. *Nehmen will sie das in Anspruch.

(58) a. Sie wird Auto fahren. - Function nouns She will car drive ‘She will drive.’

b. *Sie wird fahren Auto.

c. *Fahren wird sie Auto. COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 269

(59) a. Wir wollen noch einen pfeifen. - Idiomatic expressions we want still one whistle ‘We want to drink another one.’

b. *Wir wollen pfeifen noch einen.

c. *Pfeifen wollen wir noch einen. (Disallowed on idiomatic reading)

This list can serve as a tentative inventory of predicate types. The ungrammaticality of the c-sentences is widely acknowledged, but poorly understood, e.g. Engel (1982: 225), Olszok (1983: 109), Fanselow (1987: 93), Grewendorf (1988: 300f.), Nerbonne (1994: 118, 136ff.), ZHS (1997: 1623ff), Bouma and Van Nord (1998: 62f.), Kathol (1998: 230f., 2000: 205), Meurers (1999: 245), Webelhuth and Ackerman (1999), De Kuthy and Meurers (2001: 164f), and Müller (2002: 44f., 72, 82, 97f.). The strength of the Predicate Serialization Principle is that it provides a unified explanation of both the b- and c-sentences. The governor of a predicate word may not precede its predicate daughter unless that governor is the matrix root. A non-matrix-root +p word must follow its +p daughter. The trees of (43a–c) illustrate the effect of the +p feature. Since the root words are always part of the matrix predicate, they also carry the +p feature.

The bare infinitives dürfen and schlafen are inherently +p. Dürfen in each case is the governor of schlafen and is not the matrix root. Therefore the Predicate Serialization Principle dictates that ungrammaticality results if schlafen follows its governor dürfen. In contrast, grammaticality obtains if schlafen precedes dürfen; this is true also regardless of whether rising occurs.

270 TIMOTHY OSBORNE

The ungrammaticality illustrated in (60b–c) does not occur if the relevant words are -p. This fact is illustrated first with the incoherent constituent zu gehen in (61a-f):

34

In these cases, gezwungen is +p and assigns -p to zu gehen. The result is that the Predicate Serialization Principle does not restrict the relative order of gezwungen and zu gehen.

34 Zu gehen in (61b) is shown as a post-dependent of gezwungen. Another possibility is that zu gehen actually rises to attach to hat. I have chosen the former analysis due to the acceptability of (61f). COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 271

Since potentially pseudoincoherent constituents behave like incoherent constituents with respect to extraposition and bare infinitive fronting, they too are -p constituents.

35

In these cases, beginnen is +p and assigns -p to zu lachen. Since zu lachen is -p, its position with respect to its governor beginnen is not restricted.

5.3 The Predicate Weight Principle

The Predicate Serialization Principle requires all links of a predicate chain below the matrix root to be left-branching. There are, however, two additional types of data examined in the literature that are of interest in this area. The first is intraposition:

(63) a. *dass das Pferd verkaufen keiner will. - Coherent that the horse sell nobody wants ‘that nobody wants to see the horse’

35 Similar to the situation with gezwungen in (61c), the rising in (62c) is possible because beginnen is part of the matrix predicate, i.e. it is +s. 272 TIMOTHY OSBORNE

b. dass das Pferd zu verkaufen keiner beabsichtigt - Incoherent that the horse to sell nobody intends ‘that nobody intends to sell the horse’

c. dass das Pferd zu verkaufen keiner begehrt - Incoherent that the horse to sell nobody desires ‘that nobody desires to sell the horse’

Each of these sentences obeys the Predic ate S e ri a li z at i on Principle. The consti tu ent das Pferd zu verkaufen in (63b) and (63c) is a -p constituent, so its position cannot violate the Predicate Serialization Principle. The consti tu ent das Pferd verkaufen i n (63a), h oweve r, is a +p constituent, which means it must precede its governor will, which it does. Why then is (63a) ungrammatical? Nothing established so far predicts this ungrammaticality. The second case that is of in terest in this area is evident in the follow i ng data:

(64) a. *dass verkaufen das Pferd keiner will - Coherent that sell the horse nobody wants ‘that nobody wants to sell the horse’

b. *dass zu verkaufen das Pferd keiner beabsichtigt - Pseudoincoherent

c. *dass zu verkaufen das Pferd keiner begehrt - Incoherent

Data like (64a–c) are discussed in the literature, e.g. Fanselow (1989: 3), Grewendorf and Sabel (1994: 266, 284ff., 294), Reape (1994: 192ff.), Müller (1998: 214ff.), Grewendorf and Sabel (1999: 42f.), Sabel (1999: 432f.) . The uncont rov ersial assu m ption that N Ps always attach as pre- depen d ents to their infinitival governors – never as post-dependents – means that das Pferd in (64a–c) has undergone scrambling rising. The Scrambling Principle thus predicts the ungrammatic a li ty of (64c), since in this case zu verkaufen is a -s word. It does not, however, predict the ungrammaticality of (64a–b), since verkaufen a n d zu verkaufen are +s words. What is responsible for the ungrammaticality of (64a–b)? Nothing established so far predicts this ungrammaticality. To answer these questions, the current approach points to an additional principle of word order that is independent from the Scrambling Principle and Predicate Serialization Principle. This additional principle considers the relative ‘weight’ of the constituents involved. The basic COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 273 observation is that +p words outweigh -p words, and furthermore, +s words outweigh -s words. Consider the following data in this regard:

Standard assumptions explain the ungrammaticality of the b- and c-examples by referring to the sentence frame. NP arguments do not appear in the right bracket or Nachfeld, but rather they are restricted to appearing in the Vorfeld or Mittelfeld. Given the current dependency grammar approach, the pertinent insight in this regard is that the +s +p infinitival verb angerufen is part of the predicate chain hat...angerufen. The nominals Johanna and alle are, in contrast, -s -p dependents of that predicate chain. Apparently when equi-level constituents appear on the same side of their head, +s constituents outweigh -s constituents, and +p constituents outweigh -p constituents. This insight is formalized here as the Predicate Weight Principle: 274 TIMOTHY OSBORNE

Predicate Weight Principle: A +s constituent generally outweighs its -s co-sisters, and a +p constituent generally outweighs its -p co-sisters.

Saying that one constituent is ‘heavier’ than another is only pertinent if the two are sister constituents that appear on the same side of their head. The term co-sisters is used to denote this relation. A heavier constituent appears to the right of its ‘lighter’ co-sisters. In (65) for instance, Johanna and alle angerufen are co-sisters. Since the root of alle angerufen is part of the predicate chain, it is a +s +p constituent. Johanna, in contrast, is -s -p constituent. Therefore the Predicate Weight Principle requires alle angerufen to appear to the right of Johanna. Consider (65b) next: there alle has risen so that Johanna, angerufen, and alle are co-sisters. Ungrammaticality obtains because the lighter -s -p argument alle appears to the right of the heavier +s +p predicate angerufen. This reasoning explains the data (65c) and (66a–c) as well. It is easily possible to survey the structures where the Predicate Weight Principle plays a role:

(67) a. Er will die Kinder verwöhnen. - Post-root domain of declarative he wants the children spoil V2 clauses ‘He wants to spoil the children’

b. *Er will verwöhnen die Kinder.

(68) a . Wessen Hut haben die Kinder gefunden? - Post-root domain of whose hat have the children found interrogative V2 clauses ‘Whose hat did the children find?’

b. *Wessen Hut haben gefunden die Kinder?

(69) a. Darf er ein Eis bekommen? - Post-root domain of May he an icecream receive interrogative V1 clauses ‘Is he allowed an icecream?’

b. *Da rf er bekommen ein Eis?

(70) a. Lass die Leute schlafen. - Post-root domain of let the people sle ep imperative V1 clauses ‘Let the people sleep.’

b . *Lass schlafen die Leute.

COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 275

(71) a. dass alle geschlafen haben - Pre-root domain of that all sleep have subordinate V-last clauses ‘that everyone slept’

b. *dass geschlafen alle haben

The finite verb is the relevant root in each of these examples. The Predicate Weight Principle has no bearing on t h e pre-root domain of declarative and interro g ative matrix clauses, since the pre-root domain of such clauses contains only one or no constituents. In other words, the Predicate Weight Princ iple can have no influence on those constituents that have no co-sisters. It does, however, influence greatly the order of constituents in the pre-root domain of subordinate V-last clauses, as illustrated in (71). The data (67–71) illustrate the broad types of clause structure where the Predicate Weight Principle is important. It is also possible to test the principle with respect to predicate types. Consider the tentative inventory of predicate chain types produced in (44–60). That entire list shall not be reproduced here, but rather just a few of the more interesting predicate types for illustration:

(72) a. Naturlich sind die Kinder stolz. - Subject predicate adjectives of.course are the children proud ‘Of course the children are proud.’

b. *Natürlich sind stolz die Kinder.

(73) a. Das Wetter macht die Schüler traurig. - Object predicate adjectives the weather makes the pupils sad ‘The weather is making the pupils sad.’

b . *Das Wetter macht traurig die Schüler.

(74) a. Sie holt niemanden ab. - Separable prefix structures she picks no.one up ‘She is not picking anyone up.’

b . *Sie holt ab niemanden.

(75) a. Wir hörten die Arbeiter arbeiten. - Aci structures we heard the workers work ‘We h eard the workers working.’

276 TIMOTHY OSBORNE

b. *Wir hörten arbeiten die Arbeiter.

(76) a. Er bringt seine Sachen in Ordnung. - Function verb structures he brings his things in order ‘He is bringing his things in order.’

b. *Er bringt in Ordnung seine Sachen.

(77) a. Jetzt hören alle Radio. - Function noun structures now listen all radio ‘Everyone is listening to the radio now.’

b. *Jetzt hören Radio alle.

In each case, the two constituents that are switched are co-sisters. The a- sentences observe the Predicate Weight Principle; they have the predicate constituent following the non-predicate constituent. The opposite situation obtains in the b-sentences, hence the ungrammaticality. The discussion can now return to examples (63–64) from the beginning of this section. The dependency structure of (63a) is produced here:

Das Pferd verkaufen and keiner are co-sisters. The Predicate Weight Principle is violated because the +s +p das Pferd verkaufen does not follow its -s -p co-sister keiner. The analysis of (64a) is similar:

COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 277

Since the relevant co-sisters are all -p, only the feature ±s is important. The violation occurs because the +s verkaufen does not follow both of its lighter -s co-sisters das Pferd and keiner. The analysis of (63b) is similar because zu verkaufen is a +s pseudoincoherent word. The Predicate Weight Principle predicts the ungrammaticality in many other cases as well. For instance, it predicts the ungrammaticality of the following b-examples:

(78) a. dass den Hund zu füttern zweifellos keiner versuchte that the dog to feed doubtless no.one tried ‘that no one tried to feed the dog’

b. *dass den Hund zweifellos zu füttern keiner versuchte

(79) a. weil uns anzurufen die Kinder mehrmals wagten because us to.call the kids many.times dared ‘because the kids dared to call us many times’

b. *weil uns die Kinder anzurufen mehrmals wagten

(80) a. obwohl den Text niemand zu übersetzen beginnt although the text no.one to translate begins ‘although no one is beginning to translate the test’

b. *obwohl den Text niemand zu übersetzen heute beginnt although the text no.one to translate today begins ‘although no one will begin to translate the text today’

Examples (78a, 79a) are fine because the zu-infinitives can be -s words, so rising is not forced. The positions of the object NPs in (78b, 79b), in contrast, necessitate an analysis in terms of rising, which means that zu füttern and anzurufen must be +s pseudoincoherent words. As +s words, 278 TIMOTHY OSBORNE they outweigh their -s co-sisters keiner and mehrmals, respectively, and must therefore appear to the right of them. Examples (80a–b) are particularly telling. Zu übersetzen in both cases is a +s word; thus it outweighs its co-sisters and must appear to their right. In (80a), zu übersetzen does indeed appear to the right of all its co-sisters, but in (80b), its co-sister heute illicitly follows it. The presentation of the Predicate Weight Principle here has been very brief. Unfortunately there is not enough space in this paper to explore many of the intricacies of, and exceptions to, the principle. A comprehensive investigation of the principle must wait for another day.

6. Extending the analysis to NPs

A particular strength of the current approach is that it can be extended to other constituents beyond infinitival phrases. For instance, it is applicable to NPs. Data like the following are discussed by De Kuthy and Meurers (2001):

(81) Keiner hat ein Buch über diese Theorie gelesen. Nobody has a book about this theory read ‘Nobody read a book about this theory.’ a. Ein Buch über diese Theorie hat keiner gelesen. b. Keiner hat über diese Theorie ein Buch gelesen. c. Ein Buch hat keiner über diese Theorie gelesen. d. Gelesen hat keiner über diese Theorie ein Buch. e. Hat keiner über diese Theorie ein Buch gelesen? etc.

(82) Keiner hat das Buch über diese Theorie gelesen. Nobody has the book about this theory read ‘Nobody read the book about this theory.’ a. Das Buch über diese Theorie hat keiner gelesen. b. *Keiner hat über diese Theorie das Buch gelesen. c. *Das Buch hat keiner über diese Theorie gelesen. d. *Gelesen hat kein er über diese Theorie das Buch. e. *Hat keiner über diese Theorie das Buch gelesen? etc.

Simi lar data could easily be produced for subordinate clauses as well. The contrast in acceptability between (81b–e) and (82b–e) is due the specificity effect. Indefinite complement nouns can be transparent for scrambling rising as in (81b–e). If the noun in question is sufficiently specified however, then it becomes opaque for scrambling rising as in (82b–e). In COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 279 this case, the contrast is induced by the presence vs. absence of the specifying definite article das. It should be apparent that the noun Buch in (81) is behaving essentially like a pseudoincoherent word; it has the feature combination +s -p. It is thus transparent for scrambling rising but at the same time not subject to the strict positional restrictions placed on predicate words. Consider that many of the tests employed above to identify coherent, pseudoincoherent, and incoherent constituents support the analysis of Buch in (81) as a potential pseudoincoherent word. These tests can be adapted to NPs as shown here:

(83) a. Sie hat ein Buch über diese Theorie im Bus gelesen. - Intraposition she has a book about this theory in.the bus read ‘She read a book about this theory in the bus.’

b. Lesen wird sie ein Buch über diese Theorie. - Infinitive fronting

c. Gelesen hat sie ein Buch über diese Theorie. - Infinitival fronting

d . Sie wird über diese Theorie ein Buch lesen. - Scrambling

e. Sie hat kein Buch über diese Theorie gelesen. - Scope of negation she has no book about this theory read ‘She hasn’t read a book about this theory.’

f. Sie liest ein Buch über díese Theorie, und er über jéne. - Gapping she reads a book about this theory and he about that ‘She is reading a book about this theory, and he/him about that one.’

The i n ability of NPs to attach as post-dependents to their infinitival governors and the Predicate Weight Principle combine to r ender the extrap o sitio n test in th is case inapplicable, e.g. *Sie hat gelesen ein Buch über diese Theorie.36 Pied-piping is also inapplicable to NPs, e.g. *die

36 Based on the following data, we know that zu-infinitives can, but NPs cannot, attach to their infinitival governors as post-dependents: (i) a. Zu essen versuchen werden wir schon. to eat try will we certainly ‘We will certainly to eat.’ b. Versuchen zu essen werden wir schon.

280 TIMOTHY OSBORNE

Theorie, das Buch über die ich gelesen habe.37 And concerning the position of negation test, it is rendered inapplicable by the necessity that cohesion obtain. The other six tests, however, show that Buch is indeed behaving like a pseudoincoherent word. With respect to the overall theory of coherence developed in this paper, the data in (81–83) are important. They demonstrate that the current theory extends beyond the infinitival verbs most closely associated with theories of coherence.

7. A problem

Aci structures pose a problem for the current theory. Data like the following are discussed by Grewendorf (1988: 283ff.), Prinzhorn (1990: 204ff.), Reape (1994).

(84) a. Natürlich wird Daniel die Kinder die Pizza essen lassen. of.course will Daniel the children the pizza eat allow ‘Of course Daniel will let the children eat the pizza.’

b. *Natürlich wird Daniel die Pizza die Kinder essen lassen.

c. *Natürlich wird die Pizza Daniel die Kinder essen lassen.

The relevant constituent is in italics in (84a). It is easy to show that die Pizza essen behaves as a coherent constituent with respect to extraposition, intraposition, bare infinitive fronting, infinitival fronting, pied-piping, and the position and scope of negation. With respect to scrambling though, problems arise. The theory predicts the scrambling in (84b–c) to be possible. The fact that these sentences are bad, however, indicates that there is some particular aspect of aci-structures that needs to be identified

(ii) a. Pizza essen werden wir schon. pizza eat will we certainly ‘We will certainly eat pizza.’ b. *Essen Pizza werden wir schon. With the V2 principle in mind, sentences (ia-b) demonstrate that zu-infinitives have the option to attach as pre- or post-dependents to their infinitival governors. Sentences (iia- b) demonstrate, in contrast, that NPs must attach as pre-dependents to their infinitival governors. 37 English, however, allows such NP pied-piping, e.g. the theory, the book about which I’ve read. COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 281 and explored. At present, I do not have an explanation for these data. This matter remains to be worked out at a later date.

8. Conclusion

The dependency grammar analysis of coherence presented in this paper began with the assumption that a three-way distinction is warranted: coherent vs. incoherent vs. pseudoincoherent constituents. The approach view s coherence primarily as a principle of subcategorization, not as a matter of topology. The three-way distinction was formalized in terms of two features, ±s (scrambling) and ±p (predicate). The various combinations of these three features are as follows:

coherent word +s +p pseudoincoherent word +s -p incoherent word -s -p

Referring to these two features, three principles of word order were presented. These three principles are repeated here:

Scrambling Principle (i) If a word has the feature +s, then its dependents can rise to non-fronting (as well as fronting) positions. (ii) A dependent may not rise to a non-fronting position if its governor has a lighter dependent that does not also rise. Predicate Serialization Principle: If X and Y are +p words, X is Y’s governor, and X is not a matrix root, then Y must precede X. Predicate Weight Principle: A +s constituent generally outweighs its -s co-sisters, and a +p constituent generally outweighs its -p co-sisters.

If one adds the V2 principle to this list, then one has the basic components together that result in the topological model. The Vorfeld and left bracket of the model are due to the V2 principle. The Mittelfeld and right bracket are due to the Predicate Serialization Principle and Predicate Weight Principle. And the multiple orderings of constituents in the various fields are possible due to the Scrambling Principle. One final comment concerning the choice of a dependency-based framework is warranted. While a similar constituency-based approach to coherence is conceivable, such an approach would lack the elegance of the current system. The success of the dependency-based approach is largely due to its understanding of chains. While it is possible to produce a 282 TIMOTHY OSBORNE constituency-based definition of “chain”, such a definition would be quite laborious – see Osborne (2005a). Furthermore, if the constituency-based approach actually succeeded in making the same accurate predictions as the current dependency-based approach, the latter would still be preferable due to its structural minimalism. The dependency-based approach accomplishes a lot with very little.

References

Ackerman, Farrell & Webelhuth, Gert (1998) A Theory of Predicates. CSLI Lecture Notes 76. Stanford: CSLI Publications. Anderson, John (1979) On : Prolegomena to a Theory of Grammatical Relations. London: Croom Helm. Askedal, John Ole (1983) Kohärenz und Inkohärenz in Deutschen Infinitfügungen: Vorschlag zur begrifflichen Klärung. Lingua 59: 177–196. Baumgärtner, Klaus (1970) Konstituenz und Dependenz. In Hugo Steger (ed.), Vorschläge für eine strukturale Grammatik des Deutschen, pp. 52–77. Bech, Gunnar (1955) Studien über das deutsche Verbum infinitum. Det Kongelige Danske Videnskabernes Selskab, Historisk-filologiske Meddelelser 35, 2. Copenhagen. Besten, Hans den & Rutten, Jean (1989) On verb raising, extraposition and free word order in Dutch. In Dang Jaspers et al. (eds.), Sentential Complementation and the Lexicon: Studies in Honor of Wim de Geest, pp. 41–56. Dordrecht: Foris. Bouma, Gosse & van Noord, Gertjan (1998) Word order constraints on verb clusters in German and Dutch. In Erhard Hinrichs, Andreas Kathol, & Tsuneko Nakazawa (eds.), pp. 43–72. San Diego: Academic Press. Brody, Michael (1998) Mirror Theory: syntactic representation in perfect syntax. Linguistic Inquiry 31(1): 29–56. —— (2003) Towards an Elegant Syntax. London: Routledge. Bröker, Norbert. (2000) Unordered and non-projective dependency grammars. In Sylvain Kahane (ed.), pp. 79–111. Paris: Hermes Sciences Publications. De Kuthy, Kordula & Meurers, Walt (2001) On partial constituent fronting in German. Journal of Comparative Germanic Linguistics 3(3): 143–205. Duchier, Dennis & Debusmann, Rolf (2001) Topology dependency trees: a constraint based account of linear precedence. ACL 2001. Duden (1995) Grammatik der deutschen Gegenwartssprache. Mannheim: Dudenverlag. Eisenberg, Peter (1999) Grundriß der deutchen Grammatik: Band 2: Der Satz. Stuttgart: Metzler. Engel, Ulrich. (1982) Syntax der deutschen Gegenwartsprache. 2nd abridged ed. Berlin: Schmidt. Eroms, Hans-Werner (1988) Der Artikel im Deutschen und seine dependenzgrammatische Darstellung. Sprachwissenschaft 13, 257–308. —— (2000) Syntax der deutschen Sprache. Berlin: Walter de Gruyter. COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 283

Fanselow, Gisbert (1987) Konfigurationalität: Untersuchungen zur Universalgrammatik am Beispiel des Deutschen. Studien zur deutschen Grammatik 29. Tübingen: Narr. —— (1989) Coherent infinitives in German: Restructuring vs. IP complementation. In Christa Bhatt, Elizabeth Löbel & Claudia Schmidt (eds.), Syntactic Phrase Structure Phenomena in Noun Phrases and Sentences, pp. 1–16. Amsterdam: Benjamins. Gerdes, Kim & Kahane, Sylvaine (2001) Word order in German: a formal dependency grammar using a topology model. ACL 2001. Grewendorf, Günther (1988) Aspekte der deutschen Syntax: Eine Rektions-Bindungs- Analyse. Tübingen: Narr. Grewendorf, Günther & Sabel, Joachim (1994) Long scrambling and incorporation. Linguistic Inquiry 25(2): 263–308. —— (1999) Scrambling in German and Japanese. Natural Language and Linguistic Theory 17(1): 1–65. Grewendorf, G. & Sternefeld, W. (eds.), Scrambling and Barriers. Amsterdam: Benjamins. Grimshaw, Jane (1991) Extended projections. Ms. Brandeis University. Groß, Thomas (1999) Theoretical Foundations of Dependency Syntax. München: Iudicium. Haider, Hubert (2003) V-clustering and clause union – causes and effects. In Pieter Seuren & Gerard Kempen (eds.), pp. 91–126. Amsterdam: Benjamins. Hays, David (1964) Dependency theory: a and some observations. Language 40: 511–525. Heidolph, Karl, Flämig, Walter & Motsch, Wolfgang (1981) Gründzüge einer deutschen Grammatik. Berlin: Akademie Verlag. Hentschel, Elke & Weydt, Harald (2003) Handbuch der deutschen Grammatik. 3rd completely revised edition. Berlin: Walter de Gruyter. Heringer, Hans Jürgen (1996). Deutsche Syntax Dependentiell. Tübingen: Stauffenberg. Hinrichs, Erhard & Nakazawa, Tsuneko (1989) Flipped out: AUX in German. Chicago Linguistic Society 25: 193–202. —— (1994) Linearizing AUXs in German verbal complexes. In John Nerbonne, Klaus Netter & Carl Pollard (eds.), pp. 11–38. Stanford: CSLI Publications. —— (1998) Third Construction and VP extraposition in German: an HPSG analysis. In Erhard Hinrichs, Andreas Kathol, & Tsuneko Nakazawa (eds.), pp. 115–157. London: Academic Press. Hinrichs, E., Kathol, A. and Nakazawa, T. (eds.), Complex Predicates in Nonderivational Syntax. Syntax and Semantics 30. San Diego: Academic Press. Huddleston, Rodney & Pullum, Geoffrey (2002) The Cambridge Grammar of the English Language. Cambridge University Press. Hudson, Richard (1980) Constituency and dependency. Linguistics 18(3/ 4): 179–198. —— (1984) Word Grammar. Oxford: Basil Blackwell. —— (1990) An English Word Grammar. Oxford: Basil Blackwell. —— (2000) Discontinuity. In Sylvaine Kahane (ed.), pp. 15–56. Paris: Hermes Sciences Publications. 284 TIMOTHY OSBORNE

Jung, Wha-Young (1995) Syntaktische Relationen im Rahmen der Dependenzgrammatik. Hamburg: Buske. Kahane, Sylvain (ed.) (2000) Les grammaires de dépendance (Dependency Grammars). Traitement automatique des langues 41. Paris: Hermes. Kathol, Andreas (1998) Constituency and serialization of verbal complexes. In Erhard Hinrichs, Andreas Kathol, & Tsuneko Nakazawa (eds.), pp. 221–270. San Diego: Academic Press. —— (2000) Linear Syntax. New York: Oxford University Press. Kiss, Tibor (1994) Obligatory coherence: the structure of German modal verb constructions. In John Nerbonne, Klaus Netter & Carl Pollard (eds.), Stanford: CSLI Publications. —— (1995) Infinitive Komplementation: neue Studien zum deutschen Verbum infinitum. Tübingen: Niemeyer. Kunze, Jürgen (1975) Abhängigkeitsgrammatik. Studia Grammatica XII. Berlin: Akademie-Verlag. Kvam, Sigmund (1979) Diskontinuierliche Anordnung von eingebetteten Infinitivphrasen im Deutschen. Deutsche Sprache 7: 315–325. —— (1982) Die Einbettung deutscher und norwegischer Infinitivkonstruktionen: Eine kontrastive Studie zur Topologie der Infinitivkonstruktionen im Deutschen und Norwegischen. Deutsche Sprache 10: 333–365. Lobin, Henning (1993) Koordinationssyntax als prozedurales Phänomen. Studien zur deutschen Grammatik 46. Tübingen: Narr. Lühr, Rosemarie (1993) Neuhochdeutsch: Eine Einführung in die Sprachwissenschaft. 4th unchanged edition. München: Fink. Matthews, Peter (1981) Syntax. Cambridge University Press. Mel'čuk, Igor. (1988) Dependency Syntax: Theory and Practice. Albany: State University of New York Press. Meurers, Walt. (1999) Lexical Generalizations in the Syntax of German Non-Finite Constructions. Doctoral Thesis. Universität Tübingen. Miller, J. (1985) Semantics and Syntax: Parallels and Connections. New York: Cambridge University Press. Müller, Gereon (1998) Incomplete Category Fronting: A Derivational Approach to Remnant Movement in German. Dordrecht: Kluwer. Müller, Stefan (2002) Complex Predicates: Verbal Complexes, Resultative Constructions, and Particle Verbs in German. Stanford: CSLI Publications. Nerbonne, John (1994) Partial verb phrases and spurious ambiguities. In John Nerbonne, Klaus Netter & Carl Pollard (eds.), pp. 109–150. Stanford: CSLI Publications. Nerbonne, J., Netter, K. and Pollard, C. (eds.), German in Head-Driven . Stanford: CSLI Publications. O’Grady, William (1998) The syntax of idioms. Natural Language and Linguistic Theory 16: 79–312. Olszok, Klaus (1983) Infinite Formen im Vorfeld. In Klaus Olszok & Edit Weuster, Zur Worstellungsproblematik im Deutschen, pp. 89–169. Tübingen: Narr. COHERENCE: A DEPENDENCY GRAMMAR ANALYSIS 285

Osborne, Timothy (2003) The Left Elbow Constraint. Studia Linguistica 57(3): 233– 257. —— (2005a) Beyond the constituent: a dependency grammar analysis of chains. Folia Linguistica XXXIX/3–4. —— (2005b) The Predicate Serialization Principle of German. Ms. Pennsylvania State University. Pickering, Martin & Barry, Guy (1993) Dependency categorial grammar and coordination. Linguistics 31: 855–902. Prinzhorn, Martin (1990) Head movement and scrambling domains. In Gunther Grewendorf and Wolfgang Sternefeld (eds.), Scrambling and Barriers, pp. 199– 215. Amsterdam: Benjamins. Quirk, Randolph, Greenbaum, Sydney, Leech, Geoffrey & Svartvik, Jan (1985). A Comprehensive Grammar of the English Language. London: Longman. Reape, Mike (1994) Domain union and word order variation in German. In John Nerbonne, Klaus Netter & Carl Pollard (eds.), pp. 152–197. Stanford: CSLI Publications. Rambow, Owen (2003) Coherent constructions in German: lexicon or syntax. In Pieter Seuren & Gerard Kempen (eds.), pp. 223–242. Amsterdam: Benjamins. Reis, Marga & Sternefeld, Wolfgang (2004). Review article: Susanna Wurmbrand Infinitives, Restructuring and Clause Structure. Linguistics 42(2): 469–508. Robinson, Jane (1970) Dependency structures and transformational rules. Language 46: 259–285. Sabel, Joachim (1999). Coherent infinitives in German, Polish, and Spanish. Folia Linguistica 33(3–4): 419–440. Schmid, Tanja & Vogel, Rolf (2004). Dialectal variation in German 3-verb clusters. Journal of Comparative Germanic Linguistics 7: 235–274. Schubert, Klaus (1988) Metataxis: Contrastive Dependency Syntax for . Dordrecht: Foris. Seuren, P. & Kempen, G. (eds.), Verb Constructions in German and Dutch. Amsterdam: Benjamins. Siewierska, Anna. (1988) Word Order Rules. London: Croom Helm. Starosta, Stanley (1988) The Case for Lexicase: An Outline of Lexicase Grammatical Theory. London: Pinter Publishers. Stechow, Armin von (1984). “Gunnar Bech’s Government and Binding Theory”. Linguistics 22, 225–241. —— (1990). Status government and coherence in German. In Günther Grewendorf and Wolfgang Sternefeld (eds.), pp. 143–195. Amsterdam: Benjamins. Tarvainen, Kalevi (2000) Einführung in die Dependenzgrammatik. Germanistische Linguistik 35. Tübingen: Niemeyer. Tesnière, Lucien (1959) Éléments de syntaxe structurale. Paris: Klincksieck. —— (1969) Éléments de syntaxe structurale. Deuxième édition revue et corrigée. Paris: Klincksieck. Trask, Robert (1997) A Student’s Dictionary of Language and Linguistics. London: Arnold. 286 TIMOTHY OSBORNE

Webelhuth, Gert & Ackermann, Farrel (1999) A Lexical-Functional analysis of predicate topicalization in German. The American Journal of Germanic Linguistics and Literature 11, 1: 1–61. Wells, Rulon (1 947) Immediate Constituents. Language 23: 81–117. Wöllstein-Leisten, Angelika (2001) Die Syntax der dritten Konstruktion: Eine repräsentationelle Analyse zur Monosententialität von ‘zu’-Infinitiven im Deutschen. Tübingen: Stauffenberg. Wunderlich, Dieter (1980) Diskontinuierliche Infinitivkonstruktionen im Deutschen: Anmerkungen zu einem Aufsatz von Sigmund Kvam. Deutsche Sprache 8: 145– 150. (ZHS) Zifonun, Gisela, Hoffmann, Ludger, Strecker, Bruno et al. (1997) Grammatik der deutschen Sprache. Berlin. v. 2. Berlin: Walter de Gruyter.

Contact information:

Timothy Osborne Department of Germanic and and Literatures Pennsylvania State University 311 Burrowes University Park, PA, 16802 USA e-mail: [email protected] Michael T. Putnam

An Anti-Local Account of Why Scrambled Datives in German Can’t Bind Anaphors∗

Abstract

This article provides the sketch of a novel analysis of the binding properties of scrambled DPs in German along purely syntactic terms. In contrast to accusative DPs, dative objects that scramble outside of the vP cannot enter into licit binding relations with anaphora. Based on Hornstein’s (2001) assertion that reflexives and reciprocals are generated in the derivation rather than existing as independent lexical items in the Numeration, I adopt Grohmann’s (2000, 2003) Anti- hypothesis (lower-bound restriction on XP-movement) to illustrate that dative objects are incapable of generating anaphora in the derivation due to the natural composition of vP.

1. Introduction

Perhaps one of the most problematic issues connected with middle field scrambling in German is the fact that dative objects that scramble out of the vP cannot enter into new anaphoric binding relationships whereas accusative objects can (cf. Grewendorf 1984, 1988).

(1) …dass Fritz die Gästei gestern einanderi t vorgestellt hat. …that Fritz.NOM the guests.ACC yesterday each introduced has ‘…that Fritz introduced the guests to each other yesterday.’

(2) *…dass Fritz den Gästeni gestern einanderi t vorgestellt hat. …that Fritz.NOM the guests.DAT yesterday each other introduced has ‘…that Fritz introduced one another to the guests yesterday.’

∗ This article has benefited from discussions and criticisms from the following individuals: Acrisio Pires, Kleanthes Grohmann, Sam Epstein, Tom Stroik and Takae Tsujioka as well as three anonymous reviewers. I would also like to thank Kleanthes Grohmann and Jörg Meindl for their assistance with the German data in this article. All shortcomings and errors of this work remain my own.

SKY Journal of Linguistics 18 (2005), 287–309 288 MICHAEL T. PUTNAM

(3) …weil der Hansh der Mariam sichh/*m t auf dem Photo zeigte …since the John.NOM the Mary.DAT self in the picture showed ‘…since John showed Mary himself/*herself in the picture.’

Example (1) demonstrates that the accusative object die Gäste ‘the guests’ can and does create a new binding relation with the anaphor einander ‘each other’. Assuming that the predicate adverbial adjunct gestern ‘yesterday’ is adjoined to vP, it is evident that some form of displacement outside of the vP has occurred. Conversely, the dative objects in (2) and (3) above show that the indirect objects den Gästen ‘the guests’ and der Maria ‘Mary’ cannot exist in a binding relation with the sich ‘self’. The exact Case of sich ‘self’ in German is difficult to determine, provided that both the dative and accusative reflexives share the same morpheme with no distinctions. These results are unexpected and puzzling given recent theoretical treatments of scrambling in West Germanic, especially given the fact that scrambled datives can exist in a co-referential binding relationship with full DPs.

(4) …dass der Jörg jedem [seinen Vater] gestern [t gezeigt hat]. …that the Jörg.NOM everyone.DAT his father.ACC yesterday shown has ‘…that Jörg has shown his(x) father to everyone(x) yesterday.’

Although scrambling the dative indirect object over the direct object can create a well-formed operator variable in German (4), it cannot create the anaphor reciprocal binding relation in (2) above.1 The true nature of this restriction has escaped a sufficient explanation in minimalist syntax to date. If scrambling is the result of an overt movement process, it is most likely not linked to a morphological property of the attracting head. Scrambled DPs often show “rich” Case morphology, but there are languages with a rich Case system without scrambling (Icelandic), and languages with scrambling but lacking rich Case distinctions (Bulgarian) (cf. Alexiadou &

1 McGinnis (1999) points out the contrast in the behavior of dative objects with regard to their binding properties, i.e., that a licit operator-variable relation can be established by moving a dative indirect object over an accusative object whereas forming binding relation with an anaphoric reciprocal is not isomorphic only to German. A similar contrast arises when an object scrambles over the subject in Hindi, Japanese, and Georgian. WHY SCRAMBLED DATIVES IN GERMAN CAN’T BIND ANAPHORS 289

Fanselow 2002).2 Even within the West Germanic family of languages, dialects such as Swiss German that exhibit a four-Case distinction, do not license scrambling (cf. Kiparsky 1997). The lack of connection between the ability of a language to license scrambling and robust morphological Case is also attested in diachronic analyses of historical Germanic languages. Allen (1995:417) notes that almost a century after the loss of m-Case inflections word order freedom persisted from Old English into the Early Middle English period, i.e. for almost 200 years. In a similar light, Putnam (2006) observes the opposite pattern of a loss of scrambling preceding paradigmatic m-Case leveling in Pennsylvania German. Regardless of whether one wishes to interpret Case as an LF-interpretable phenomenon (Bobaljik 1995, Richards 2005) or remove the licensing of Case from the narrow syntax altogether (McFadden 2004, Bobaljik 2005), Case does not function as the impetus to this problem. Prescribing either a government- like account for Case assignment in the syntax or advocating movement to a functional projection cannot explain why dative objects cannot bind. Movement to a hierarchically superior specifier position will not disrupt any relevant c-command relations necessary to license Case. Likewise, scrambling viewed as adjunction to a higher position has been shown to not prevent anaphoric binding for accusative objects (Grewendorf & Sabel 1994, 1999, Mahajan 1990, 1994, Webelhuth 1989, 1992, 1995). As originally noted by Lenerz (1977), middle field scrambling in German conforms to discourse and semantic requirements. Meinunger (1995, 2000), among others, classifies definite XPs that scramble from their base-position in the VP as discourse themas (i.e., discourse-old information). In contrast, definite objects that remain in the VP are read as discourse rhemas (i.e., discourse-new information) and are stressed. Indefinites exhibit the opposite syntactic behavior in comparison to their definite counterparts; whereas definite objects predominantly scramble and are defocused in the middle field, only indefinites that can be interpreted as generic and not existential may undergo displacement to the middle field (cf. Heim 1982, Diesing 1990, 1992). Although an appropriate answer as to why scrambled dative objects cannot create new anaphoric binding relations seems to fall outside the borders of orthodox minimalist syntax, there is a way to explain this

2 Although similar to scrambling, Object Shift in Icelandic displays unambiguous A- movement characteristics, whereas scrambling exhibits both A- and A’-movement properties. 290 MICHAEL T. PUTNAM phenomenon through Grohmann’s (2000, 2003) Anti-Locality Hypothesis. Following Hornstein (2001), it is conceivable that reflexivization is generated through the course of syntactic derivation rather than existing as unique lexical items in the lexicon. Based on the predominant default base- argument order of the vP in German for full DPs (SUBJ >> IO >> DO), this article adopts both Hornstein’s (2001) approach that reflexives are generated in the syntax and Grohmann’s (2000, 2003) lower-bound restriction on XPs demanding that they must undergo a minimal distance of traversal (i.e., Anti-Locality). The remainder of the article adheres to the following schematic. In Section 2 I outline the basic tenets of Prolific Domains and Anti-Locality. Section 3 applies Anti-Locality to account for data from German involving anaphoric binding relations. Section 4 critiques McGinnis’ (1999) Lethal Ambiguity as a theoretical construct that can account for the anaphoric binding restriction exhibited by scrambled dative objects. In Section 5 I address residual issues/data that potentially pose a threat to the analysis put forth in this article. Section 6 serves as a summary of arguments and proposes avenues of future research.

2. Prolific Domains and Anti-Locality

2.1 Prolific Domains

The relative cyclic units within the framework of orthodox minimalism are called phases, which are assumed to be the verb (vP) and complementizer (CP) in the their strongest form and tense phases (TP) in the weakest. At these points where propositional units are completed, material at both strong phases (vP, CP) are regarded to be independent at the interfaces. Recent challenges to the concept of strong phases, however, have proven these constructions to not be optimal tools. For example, Epstein and Seely (2002) discuss at length why the specification of vP and CP as phases is potentially problematic. Chomsky (2000, 2001, 2004) attempts to motivate vP and CP as phases on the grounds that they are relatively independent at the interface levels. Although it is unclear exactly what Chomsky means by “relatively independent” in this context, Epstein and Seely (2002:78) raise the following key question: “How can we know that they are relatively independent at the interface if Spell Out applies before the interface is reached, and without access to interface properties?” Epstein and Seely (2002:78) comment further that “…it is a potential architectural paradox to WHY SCRAMBLED DATIVES IN GERMAN CAN’T BIND ANAPHORS 291 hypothesize that vP and CP are spelled out cyclically, internal to the narrow syntax by virtue of them having the property being later relatively independent at the interface.”3 The ideal derivational system based on contextual information rather than mere ontological, structural commitments should include not only CP and vP as chief benchmarks of systematic evaluation, but also TP. Based on Chomsky’s (1986) earlier conceptions of clause structure, Platzack (1986) and Grohmann (2000, 2003) elucidate the natural tripartite division of labor exerted by natural clauses, with the vP layer responsible for thematic relations, the TP layer licensing agreement and the CP-layer (cf. Rizzi 1997) available for discourse properties. Grohmann’s Prolific Domains (2000, 2003) function as the ideal units of evaluation for the construction of syntactic structure. Derivation qua Prolific Domains not only achieves a natural functional tripartite structure of clauses, but also establishes a tripartite structure of layers relevant to contextual information: thematic context (in terms of vP/VP or separate v/V-projections), agreement context (vis-à-vis split Infl: AspP, AgrP, TP etc.), and discourse context (viz. Split-CP: TopP, FocP, CP, etc.). The individual layers responsible for thematic, agreement and discourse contextual relations are referred to as Prolific Domains (Π∆):

(5) Prolific Domain (Grohmann 2003: 78) Let a Prolific Domain Π∆ be a contextually defined part of CHL: (i) Each Π∆ spells out its context information and (ii) Spell Out feeds the PF and LF interface levels

The generation of Prolific Domains and their existence in the CHL is chiefly motivated due to their intimate connection with contextual information and its direct relation with phrasal hierarchy. We can understand the notion of context information as follows:

(6) Context Information (Grohmann 2003: 78) Context Information is determined by context values: (i) |Θ| ranges over thematic relations; (ii) |Φ| ranges over agreement properties; (iii) |Ω| ranges over discourse information.

3 For further criticisms of phases see Grohmann & Boeckx (2004), Legate (2002), Epstein & Seely (2002), Bošković (2002), Matushansky (2004), Hiraiwa (2003), Bouchard (2002), Putnam (2006). 292 MICHAEL T. PUTNAM

Grohmann employs the mnemonics Θ, Φ and Ω to be context values that are part of the information contained within each head. In order to stay consistent with previous work dealing within this theoretical framework, I, too, will adopt these symbols in this work. Accordingly, V and v, for example, come equipped with the context value |Θ|, and Top or C with the context |Ω|. This information is relevant in two respects: first, it groups various projections into a single Π∆; second, each Π∆ contextually identified in this way ships the information applicable for the specific context to the interfaces. This convention identifies a number of heads, and by extension projections, as part of the same context (Grohmann 2003: 78). In connection with introducing the context variables in (6), we can rename the phrasal units to conform to their interpretable identity; accordingly, the vP-layer is properly known as the Θ-domain, the IP-layer as the Φ-domain and lastly the CP-layer as the Ω-domain. Prolific Domains depart from orthodox minimalist desiderata in the stipulation that these contextually- valued metric units serve as the hierarchical units of evaluation that are cyclically assessed and shipped off to Spell-Out.

2.2 Anti-Locality

A chief motive in implementing Prolific Domains in the CHL is its inherent ability to evaluate the displacement of items in the syntax in regards to their locality status. Locality is conceived as a restriction on the maximum distance that an element may move in the derivation in one cyclic step. Locality conditions are imposed on all relevant dependencies (i.e. those created by Move/Attract) and delimit the upper-bound of a well-formed dependency between two positions created by movement. Grohmann (2000, 2003) argues in favor of a lower-bound restriction on movement requiring displacement operations to travel a minimal distance that he labels Anti-Locality. Under this guise, Anti-Locality is violated when a particular movement chain is deemed to not move “far enough.” Prolific Domains serve as the ideal contextually defined units to evaluate Locality and Anti-Locality.4

4 Abels (2003) also discusses the ungrammaticality of movement from a complement- to-specifier movement chain involving Merged items of the same head that he also labels Anti-Locality. The reason for discarding this version of Anti-Locality from the current discussion is the fact that it is not sensitive to Prolific Domains. For an analysis of how both treatments of Anti-Locality can be integrated in the narrow syntax see Putnam (2006: Chapter 3). WHY SCRAMBLED DATIVES IN GERMAN CAN’T BIND ANAPHORS 293

(7) Anti-Locality (Grohmann 2000, 2003) [α∆ XP Yo … [α∆ … XP … ]] (anti-local movement) {α = any Prolific Domain}

Movement within any given Prolific Domain would violate Anti-Locality. In the event that anti-local movement occurs in the derivation, Grohmann (2000, 2003) proposes a repair mechanism that is implemented by the CHL to prevent the derivation from crashing. Since the goal of the MP is to refine a system to include the least amount of exceptions and principles to achieve successful derivations, Grohmann reformulates the concept of “XP uniqueness” in regards to anti-locality effects through the Condition on Domain Exclusivity (CDE).

(8) Condition on Domain Exclusivity (CDE) (Grohmann 2003: 80) For a given Prolific Domain Π∆, an Object O in the phrase-marker must receive an exclusive interpretation at the interfaces, unless duplicity of O yields a drastic effect on the output of that Π∆.

The object O refers to any element of the phrase-marker within a Π∆, i.e. all substantive and functional heads and projections.5 Pending a more detailed discussion of interface interpretation, the notion understood here regarding O is that O signals its presence to the interfaces in terms of a set of instructions that LF and PF pass on to the respective interfaces (Grohmann 2003: 81). The “drastic effect on the output” correlates to the PF-matrix of O at the PF-interface. In most cases, the higher copy will spell out at PF. This makes sense, given that higher copies of an object O will possess one less uninterpretable feature than its lower copies. If, however, a multiple occurrence of O involves two phonetically distinct copies of O, the CDE is not violated. Therefore, the CDE can be upheld if movement within a Π∆ takes place. The structural representation in (9) below illustrates the effects of Anti-Locality on movement chains qua the CDE.

5 This lower-bound restriction on a minimal required distance of traversal is unique only to phrasal constituents and does not apply to minimal projections (Xo). Moving a head Xo to adjoin to the next highest head Yo results in the creation of a complex head [Xo- Yo]-Yo rather than substitution and the building of new syntactic structures. 294 MICHAEL T. PUTNAM

(9) YP 2 (X3) xP 2 (X2) x’ 2 xo XP Anti-Locality 2 o Violation X (X1)

The movement chain (X1, X2) above occurs within the maximal projection xP. Assuming that xP represents a contextually defined Prolific Domain, this intra-domain movement chain (X1, X2) violates Anti-Locality. The higher link in the movement chain (X2, X3) targets a superior domain and therefore averts any potential violation of Anti-Locality. An anonymous reviewer claims that the application of Prolific Domains (vP, TP, CP) as modular cyclic units that are shipped to the interfaces upon their completion is irrelevant for the current hypothesis. Rather, a phase-based notion of cyclic evaluation and Anti-Locality is all that is needed: as long as scrambling takes place from its base position to a multiple specifier of vP, i.e., to the edge of the phase, the derivation should proceed without crashing. What the reviewer fails to take into account is the distinct advantage Prolific Domains affords us in reducing the taxing complexity a phase-based system places on the human faculty of language. A view of syntactic composition along the lines of Prolific Domains eliminates the “escape hatch” necessitated in a phase-base system reminiscent of earlier stages in generative syntax (cf. Chomsky 1986). The infinite proliferation of phase edges can also be eliminated by Prolific Domains, thus regaining the distinction between unique specifiers and adjuncts that is currently not possible in a phase-based approach. Consequently, if the scrambled XP were to move to a multiple specifier of vP, according to the Anti-Locality Hypothesis it would inflict a lower- bound movement violation. As shown in example (1), if the scrambled DP die Gäste ‘the guests’ moves to structurally higher position than the predicate adverbial adjunct gestern ‘yesterday’, CDE effects forcing the phonological realization of both elements of the chain will not take place. Assuming that predicate adverbials are adjoined to vP in German, the scrambled DP must leave the contextually-defined domain for thematic WHY SCRAMBLED DATIVES IN GERMAN CAN’T BIND ANAPHORS 295 properties (vP). An intricate discussion of the motivation of scrambling in West Germanic falls outside the scope of this article; however, one thing is certain: scrambling is not driven by theta-considerations.6 If scrambling is not ubiquitous A-movement, the exact contextual composition of TP as an agreement domain requires reconsideration. Putting that argument aside, our concern is centered less on the exact properties of TP but rather the fact that scrambling is not movement to the “edge” of vP but rather to a functional projection outside of the thematic layer. The structures in (10a) and (10b) below illustrate the fundamental differences between scrambling in a phased-based account of syntax and one employing Prolific Domains respectively.7

(10) a. vP b. XP 2 2 DPi vP DPi vP 2 2 adjunct vP adjunct vP 2 2 Subj v’ Subj v’ 2 2 ti v ti v

Focusing our attention on the lower link of the movement chain, at this juncture of the derivation two outcomes are possible. Due to the CDE, which will force both the higher (X2) and lower copy (X1) of the first link in this multiple movement chain to be spelled out as two separate phonological entities at the PF-interface, the system can either accept the appearance of each individual object or the CDE will subsequently cause the system the crash. It may seem intuitive to posit that violations of Anti- Locality will unambiguously cause the system to crash. There are however instances in which the spelling out of both copies of the movement chain (X1, X2) is required to generate fully grammatical structure interpretable at the interfaces. Hornstein (2001) puts forward the argument that anaphora

6 For a detailed account of the motivation of scrambling in West Germanic within a Prolific Domains-based system, see Putnam (2006). 7 Further evidence supporting the hypothesis that TP is indeed separately interpreted at the interfaces and a relevant domain delimiting the minimal distance of traversal XPs may go in the derivation can be found in Grohmann’s (2003: Ch. 2) discussion and explanation of left-dislocated topic constructions. 296 MICHAEL T. PUTNAM are generated in the run of the derivation and are not present in the Numeration. This generalization is neatly captured and explained within the framework of Anti-Locality and the CDE. Example (11) represents a sample application of the CDE generating anaphora.8

(11) a. * Sara likes. o b. #[vP Sara v [VP likes (Sara)]] [AG] ([TH])

Based on the assumption that vP is the layer in which thematic contextual information is both processed and accounted for (hence, Θ∆), we know that any XP-movement within this Prolific Domain is illegal due to the fact that it violates the Anti-Locality Hypothesis, unless it is followed by Copy Spell Out of the lower XP qua the CDE. The only way to salvage (11b) is through the application of the CDE.

(12) [Θ∆ Sara v [VP likes Sara  herself]

In the derivation (12), the diacritic  represents the spelling-out of the lower copy of at PF. Sara doesn’t move far enough (i.e. outside of the Π∆ it is generated in, in this case Θ∆), which is a clear violation of the Anti-Locality Hypothesis. Here we see that it is possible, and in this particular case, expected, that we have two PF-matrices within the same Π∆. This example suggests that this approach is indeed conceivable, in that it correctly predicts the possible ways of expressing reflexivization qua the Anti-Locality Hypothesis (cf. Hornstein 2001). Movement to another position within the Θ-domain would imply the existence of two distinct θ- roles. This is reminiscent of Lidz’s (2001) Condition R constructed to account for the binding differences between anaphors and logophors.

(13) Condition R (Lidz 2001: 14) λx [P(x,x)] Ù ( θ1 = θ2) semantics theta-grid

The condition states that if a predicate is semantically reflexive, then it must be lexically reflexive. Similarly, if a predicate is lexically reflexive,

8 The distinction between reflexives and reciprocals is not clearly spelled out in either Hornstein’s or Grohmann’s approach. Furthermore, it is unclear how the derivation makes the distinction between reciprocal and reflexive anti-local copies in a movement chain. I leave this issue open to future inquiry. WHY SCRAMBLED DATIVES IN GERMAN CAN’T BIND ANAPHORS 297 then it must be semantically reflexive as well. The desired result, namely, that movement within a given Prolific Domain will force the lower copy of the moved item to receive prominence at PF (qua the CDE), is achieved.

3. Why Datives Can’t Bind Anaphors

Focusing on the internal structure of the vP-layer/Θ-domain, similar to Larson (1988), I adopt a version of the light verb approach and take the structure of the thematic layer to consist of “shells” – in particular VP (hosting the lexical verb), usually dominated by vP (or the light verb). An issue of much debate, however, is the position of the direct (THEME) and indirect objects (GOAL) within the vP, or Θ∆. The composition of the argument structure of ditransitive verbs theoretically need not occur in a set order given that the application of Merge is free and that vP is designated at the contextual domains responsible for thematic properties. Cross- linguistically, however, the general order of indirect (GOAL) objects being merged into a structural higher position than direct (THEME) objects is widely maintained (cf. Pylkkänen 2002, Grohmann 2003, Miyagawa and Tsujioka 2004, among others). If we assume that Θ-features are the features that motivate Merge, it is possible to view the internal structure of vP/ Θ∆ as “unfixed” regarding the THEME >> GOAL ordering. In other words, the verb – being the head of the VP – possesses a Θ-feature to be assigned to the object that merges with it, thus projecting further as VP. Under such assumptions, it would make little difference if the direct object or the indirect object were to merge as [CompVP]. Miyagawa and Tsujioka (2004) provide evidence counter to the idea that the order of objects in the vP is “unfixed”, maintain that the THEME >> GOAL ordering is the result of optional scrambling. As will be demonstrated below, the ordering of GOAL >> THEME on double object constructions is crucial to the hypothesis put forward in this article, therefore the internal structure of vP/ Θ∆ is illustrated below in (14):

298 MICHAEL T. PUTNAM

o (14) a. [vP Ext-Arg V-v [VP GOAL tv THEME]] o b. [vP Ext-Arg V-v [VP tv GOAL]] c. vP 3 EA/SU v’ 3 o v VP 3 GO/IO V’ 3 o V 6 … TH/DO …

According to the structural outline of the vP-shell/Θ-domain in (14c) above, arguments Merge into the thematic layer of the narrow syntax in the following positions: [CompVP], [SpecVP] or [SpecvP]. Based on Hornstein’s assumption within the framework of Anti-Locality as conceived by Grohmann, in order for reflexives to be generated in the derivation, movement within the vP-layer must take place in one way or another. In this section the true nature of the problem regarding why scrambled indirect objects cannot bind anaphors is revealed: indirect objects (GOAL) are inaccessible to any potentially empty nodes within the vP-shell to generate reflexives. The sentences in (15) and (18) contrast the ability of displaced direct (THEME) and indirect (GOAL) objects and their ability (or lack thereof) of entering into licit binding relations with reflexives.

(15) …dass der Fritz die Gästei einander ti vorgestellt hat. WHY SCRAMBLED DATIVES IN GERMAN CAN’T BIND ANAPHORS 299

(16) CP 2 C’ 2 dass TP 2 der Fritz T’ 2 AgrOP T 2 die Gäste vP 2 tFritz v’ 2 VP hat 2 tdie Gäste V’ 2 tdie Gäste vorgestellt

The movement chain involving the direct object (THEME) die Gäste ‘the guests’ violates Anti-Locality by moving from [CompVP] to [SpecVP]. Crucially, there is only one DP that scrambles possessing two distinct θ- roles, not two separate DPs that are extracted from the Numeration. Movement of a maximal projection within a contextually-defined domain responsible for thematic properties violates Anti-Locality and will be interpreted by the CHL as movement to a Θ-position, resulting in both the lower and upper copy of the displaced XP being exclusively interpreted as independent objects at the interfaces. What makes this step in the derivation legal is the generated empty position [SpecVP]: The fact that no indirect object (GOAL) or its copy occupies this position makes it a desired landing site the subject die Gäste ‘the guests’. This operation then invokes CDE effects that in turn generate the reflexive anaphor einander ‘each other’.

(17) [Θ∆ die Gäste v [VP die Gäste  einander]

Indirect objects (GOAL), however, are not afforded the same structural opportunities to generate reflexives as are their direct object (THEME) 300 MICHAEL T. PUTNAM counterparts. Consider the ungrammatical sentence in (18) and its subsequent derivational history in (19).

(18) *…dass der Fritz den Gästeni einander ti vorgestellt hat. ...that the Fritz.NOM the guests.DAT each other introduced has ‘…that Fritz introduced each other to the guests.’

(19) *CP 2 C’ 2 dass TP 2 der Fritz T’ 2 AgrOP T 2 den Gästen vP 2 tFritz v’ 2 VP hat 2 tden Gästen V’ 2 vorgestellt

If indirect objects (GOAL) are base-generated (or adjoined) to [SpecVP], there does not exist any possibility for the indirect object (GOAL) to land in an open specifier in the vP-shell. The lower copy of the subject der Fritz ‘the Fritz’ remains in [SpecvP], thus thwarting any attempt of the indirect object (GOAL) of generating a reflexive anaphor by means of the CDE. Equipped with the knowledge of how reflexives come into existence in the narrow syntax, this theoretical framework can also account for how subjects generate anaphors. Consider sentence (20).

(20) Die Gästei rasieren sichi den Bart. The guests.NOM shave self the beard.ACC ‘The guests shave their beards.’

WHY SCRAMBLED DATIVES IN GERMAN CAN’T BIND ANAPHORS 301

(21) TP 2 die Gäste T’ 2 vP rasieren 2 tdie Gäste v’ 2 tdie Gäste trasieren 2 den Bart trasieren

The subject – which is usually base-generated in the canonical [SpecvP] position – is Merged into the derivation [SpecVP]. The rationale behind this unusual approach is as follows: the DP die Gäste ‘the guests’ bears two distinct θ-roles, i.e., GOAL/BENEFACTOR and AGENT. In order to exhaust both of them, the DP die Gäste ‘the guests’ originates in [SpecVP] to check its first θ-role (GOAL/BENEFACTOR) and then moves to [SpecvP] to discharge AGENT. The movement violates Anti-Locality and forces the lower copy in [SpecVP] to receive its own phonetic representation at PF as the anaphor sich ‘self’.

(22) [Θ∆ die Gäste v [VP die Gäste  sich]

Once again, indirect objects (GOAL) have nowhere to move within the vP- shell/Θ-domain, thus eliminating the possibility of generating reflexives qua Anti-Locality and the CDE.

(23) *Die Freunde haben den Gästeni sichi den Bart rasiert. The friends.NOM have the guests.DAT self the beard.ACC shaved ‘The friends shaved themselves the beards of the guests.’

4. Lethal Ambiguity

The theoretical approach put forward here has a distinct advantage over McGinnis’ (1999) approach to the contrastive binding behavior of scrambled dative indirect objects. McGinnis proposes that anaphoric dependencies cannot be established between two specifiers of the same head, a constraint that she labels Lethal Ambiguity.

302 MICHAEL T. PUTNAM

(24) Lethal Ambiguity (McGinnis 1999:2) An anaphoric dependency cannot be established between two specifiers of the same head.

Although McGinnis comes close in her analysis to the proposal put forward here, her hypothesis faces shortcomings. First, the motivation for an anaphor such as sich ‘self’ in German to move to a multiple specifier position at the edge of a phase is not specified. Movement to the edge of a phase is to ensure that constituents are visible candidates to participate in operations accessible to the higher phase (in this case, CP). As elucidated earlier in this work, the Case of reflexives in German is rather difficult to ascertain; however, if Case is checked vP-internally, the only potential motivation to move to [SpecvP] would be an EPP feature. The existence of multiple specifiers destroys the contrast between specifiers and adjuncts. Movement solely to eliminate an EPP feature is illogical, provided that EPP is not interpreted at either the LF or PF-interface (see Epstein, Seely and Pires 2005, among others). Consequently, even if we accept multiple specifiers as licit constructions, it is unclear how Lethal Ambiguity can determine whether or not a scrambled DP can bind a lower anaphor. Following McGinnis’ proposition of Lethal Ambiguity, it should be the case that not only are dative objects unable to bind anaphors but also accusative objects should not be able to perform such a task in this model. If all multiple specifiers are considered to be equidistant from a given head, accusatives would also violate Lethal Ambiguity. To illustrate this crucial point, I turn once again to the initial data set at the beginning of this article, repeated below as (25) and (26) for the sake of the reader.

(25) …dass Fritz die Gästei gestern einanderi t vorgestellt hat. …that Fritz.NOM the guests.ACC yesterday each other introduced has ‘…that Fritz introduced the guests to each other yesterday.’

WHY SCRAMBLED DATIVES IN GERMAN CAN’T BIND ANAPHORS 303

(26) *…dass Fritz den Gästeni gestern einanderi t vorgestellt hat. …that Fritz.NOM the guests.DAT yesterday each other introduced has ‘…that Fritz introduced one another to the guests yesterday.’

Utilizing multiple specifiers, the structures in below in (27a) and (27b) – representing (25) and (26) respectively – illustrate that Lethal Ambiguity cannot account for the ungrammaticality of (26). The two structures are indistinguishable from one another, thus demonstrating the inability of Lethal Ambiguity in explaining these data.

(27) a. vP b. *vP 2 2 DPACC vP DPDAT vP 2 2 adjunct vP adjunct vP 2 2 anaphor … anaphor …

This raises an important question as to how Case is assigned within this framework. Agreeing with McFadden (2004), I maintain the distinction between DP licensing and the assignment of morphological Case. With that said, what is important is that both the scrambled DP and the bound anaphor share the same Case. Under the assumption that reflexives and reciprocals are generated in the derivation as described in this article, this will happen by default based on the fact that the anaphor is a lower copy of the displaced XP with which it is bound. Perhaps the argument can be made that if dative is a structural Case in German, it is non-transferable to the lower copy of the same DP. Admittedly, the assignment of Case is a very important part of the system; however, it bears little relevance to the current argument. I leave this for future consideration. Adopting the approach to anaphora and reciprocal binding relations outlined in this study obviates the shortcoming presented by McGinnis’ proposal of Lethal Ambiguity. Movement to the “edge” is no longer needed in a system utilizing Prolific Domains. As established in the introduction of this work, agreement features are not responsible for licensing scrambling, therefore it is highly unlikely that A-movement of any kind is involved in middle field scrambling.9 Lastly, by allowing reflexives and reciprocals to

9 Constituents scrambled to the middle field in German can be iterative; both crossing and nesting paths are possible constructions. This is a crushing blow to any theory that interprets middle-field scrambling as A-movement based on Chomsky’s (1995) 304 MICHAEL T. PUTNAM be generated in the derivation qua Anti-Locality and CDE effects, the theory need not develop a fixed canonical base-generated position for anahora.

5. Residual Issues

This article proposes at the very least a sketch of a novel syntactic solution to the troublesome data dealing with why scrambled dative objects cannot bind anaphora. This solution stands in contrast to the pseudo-semantic accounts that currently abound in the literature. One reviewer proposed the following data set as potential evidence again a purely syntactic account of this phenomenon.

(28) … weil Hansj [Mariai ohne was an] sichi/j auf dem Photo zeigte. since Hans Maria.DAT without something on refl .ACC on the photo showed

(29) … weil Hansj [Mariai [sichi/j ohne was an] auf dem since Hans Maria.DAT refl.ACC without something on on the Photo zeigte. Photo showed ‘…since Hans showed Mary a picture of himself/herself without something on.’

The argument put forward is that the adjuncts such as auf dem Photo ‘in the photo’ as opposed to im Spiegel ‘in the mirror’ change the anaphoric reference dramatically, hence virtually excluding a generalization in only syntactic terms. The generation of the reflexive sich ‘self’ in both (28) and (29) above can be easily accounted for under the approach developed here. To generate the reflexive sich ‘self’, either DP (Hans, Maria) would merge into the derivation bearing the θ-role THEME. Movement would then take place either to [SpecVP] or [SpecvP] in order to discharge the second θ-role on the DP and generate the reflexive. The two possible derivational histories of sich ‘self’ in (28) are schematized below in (30a) and (30b).

(30) a. [Θ∆ Hansj [Maria …] Hans  sichj ...] b. [Θ∆ Hans [Mariai [Maria  sichi…]]

assertion that movement from a position responsible for Case to a higher position responsible for the same grammatical purpose is illegal. WHY SCRAMBLED DATIVES IN GERMAN CAN’T BIND ANAPHORS 305

Anti-Locality offers a suitable solution to these supposed problematic data, and as a result issues a feasible syntactic explanation to this movement type.

6. Conclusion

In this article I have illustrated why and how displaced dative indirect objects cannot enter into licit anaphoric binding relations. First, based on Hornstein’s (2001) hypothesis that reflexives and reciprocals are generated in the syntax accompanied by Grohmann’s (2000, 2003) introduction of a lower-bound movement constraint on XP-displacement (e.g., Anti- Locality), it is predicted that objects that are base-generated in [SpecVP] are not afforded the opportunity of moving within the vP-layer. The failure to violate Anti-Locality results in the inability for indirect objects to create anaphora. Second, whereas indirect objects that are base-generated cannot move to [SpecvP] because they are blocked from such a move by the presence of the lower copy of the subject, subjects that are base generated in [SpecVP] can and must move to [SpecvP] to ensure that both θ-roles are properly discharged. Third, based on the hypothesis that passive VPs lack the functional vo determining voice and Case, it stands to reason that reflexives should not be able to appear in passive sentences. The lack of the functional vo also means the lack of [SpecvP] within the Θ-domain, i.e., the specifier where the θ-role AGENT is assigned and deactivated. These facts are borne out in the data below.

(31) Frederiki hat sichi ein Haus gekauft. (Wurmbrand 2002: 111) Frederic has self a house bought ‘Frederic bought himself a house.’

(32) Ein Hausi wurde (*sichi) gekauft. A house was (*self) bought ‘A house was bought (*oneself).’ 306 MICHAEL T. PUTNAM

(33) TP 2 Ein Haus T’ 2 VP wurde 2 V’ 2 tein Haus gekauft

The lower copy of ein Haus ‘a house’ cannot simultaneously be the THEME and AGENT of this collocation unit. Admittedly, this article presents a sketch of a solution that should be extended to empirical data from typologically diverse languages. From a theoretical standpoint, although generating anaphora in the derivation qua Anti-Locality and the CDE is now a well-attested phenomenon established by Grohmann (2000, 2003), it remains to be seen if the same processes exist for the generation of logophors by means of the same constraints and procedures. Lidz (2001) defines logophors (e.g., near-reflexives: λx [P(x,f(x))]) to function as semantically distinct elements, although the antecedent and the anaphor in such relations can be the same entity in the world. From a preliminary view, the generation of logophors through the similar processes afforded to anaphors under Anti-Locality is problematic due to the fact we would have to argue for an incomplete or errant Copy Spell-Out procedure. These topics remain open for further inquiry. In sum, this article solves an issue that has plagued the theory for quite some time: scrambled dative objects cannot bind anaphors, because they cannot create them.

References

Abels, Klaus (2003) Successive Cyclicity, Anti-Locality, and Adposition Stranding. Ph.D. dissertation, University of Connecticut. Alexiadou, Artemis & Fanselow, Gisbert (2002) On the Correlation Between Morphology and Syntax: The Case of V-to-I. In C. Jan-Wouter Zwart and Werner Abraham (eds.), Studies in Comparative Germanic Syntax: Proceedings from the 15th Workshop on Comparative Germanic Syntax, pp. 219–244. Amsterdam: Benjamins. Allen, Cynthia (1995) Case Marking and Reanalysis. Oxford: Claredon Press. Bobaljik, Jonathan (2005) Where’s phi? Ms. University of Conneticut. WHY SCRAMBLED DATIVES IN GERMAN CAN’T BIND ANAPHORS 307

—— (1995) Morphosyntax: The Syntax of Verbal Inflection. Ph.D. dissertation, MIT. Bošković, Željko (2002) A-Movement and the EPP. Syntax 5:167–215. Bouchard, Denis (2002) Adjectives, Number, and Interfaces. Why Languages Vary. London: . Chomsky, Noam (1986) Barriers. Boston: MIT Press. —— (1995) The . Boston: MIT Press. —— (2000) Minimalist Inquiries: The Framework. In Roger Martin, David Michaels & Juan Uriagereka (eds.), Step by Step: Essays on Minimalist Syntax in Honor of Howard Lasnik, pp.89–155. Boston: MIT Press. —— (2001) Derivation by Phase. In Michael Kenstowicz (ed.), Ken Hale: A Life in Language, pp.1–52. Boston: MIT Press. —— (2004) Beyond Explanatory Adequacy. In Adriana Belletti (ed.), Structures and Beyond: The Cartography of Syntactic Structures 3, pp. 104–131. Oxford: Oxford University Press. Diesing, Molly (1990) The Syntactic Roots of Semantic Partition. Ph.D. dissertation, University of Massachusetts-Amherst. —— (1992) Indefinites. Boston: MIT Press. Epstein, Samuel D. & Seely, T. Daniel (2002) Rule Applications as Cycles in a Level- Free Syntax. In Samuel D. Epstein & T. Daniel Seely (eds.), Derivation and Explanation in the Minimalist Program, pp. 65–89. Oxford: Blackwell. Epstein, Samuel D., T. Daniel Seely & Acrisio Pires (2005) EPP in T: More Controversial Subjects. Syntax 8:65–80. Grewendorf, Günther (1984) Reflexivierungsregeln im Deutschen. Deutsche Sprache 1: 14–30. —— (1988) Aspekte der deutschen Syntax. Tübingen: Narr. Grewendorf, Günther & Sabel, Joachim (1994) Long Scrambling and Incorporation. Linguistic Inquiry 25:263–308. —— (1999) Scrambling in German and Japanese: Adjunction versus Multiple Specifiers. Natural Language and Linguistic Theory 17:1–65. Grohmann, Kleanthes (2000) Prolific Peripheries: A Radical View from the Left. Ph.D. dissertation, University of Maryland. —— (2003) Prolific Domains: On the Anti-Locality Movement Dependencies. Amsterdam: Benjamins. Grohmann, Kleanthes & Boeckx, Cedric (2004) Putting Phases in Perspective. Ms. University of Cyprus and Harvard University. Heim, Irene (1982) The Semantics of Definite and Indefinite Noun Phrases. Ph.D. dissertation, University of Masschusetts-Amherst. Hiraiwa, Ken (2003) Cyclic Locality. Ms. Cambridge: MIT. Hornstein, Norbert (2001) Move! A Minimalist Theory of Construal. Oxford: Blackwell. Kiparsky, Paul (1997) The Rise of Positional Licensing. In Ans van Kemenade & Nigel Vicent (eds.), Parameters of Morphosyntactic Change, pp.460–494. Cambridge: Cambrige University Press. Larson, Richard (1988) On the Double-Object Construction. Linguistic Inquiry 19:355– 391. 308 MICHAEL T. PUTNAM

Legate, Julie (2002) Some Interface Properties of the Phase. Linguistic Inquiry 34:506– 516. Lenerz, Jürgen (1977) Zur Abfolge nominaler Satzglieder im Deutschen. Tübingen: Narr. Lidz, Jeffrey (2001) Condition R. Linguistic Inquiry 32:123–140. Mahajan, Anoop (1990) The A/A-bar Distinction and Movement Theory. Ph.D dissertation, MIT. —— (1994) Against the Relevance of Subjacency at LF. Linguistic Inquiry 25.1:171– 179. Matushansky, Ora (2004) Going Through a Phase. In Martha McGinnis & Norvin Richards (eds.), Proceedings of the MIT Workshop on EPP and Phases, to appear. Boston: MITWPL. McFadden, Thomas (2004) The Position of Morphological Case in the Derivation: A Study on the Syntax-Morphology Interface. Ph.D. dissertation, University of Pennsylvania. McGinnis, Martha (1999) A-Scrambling Exists! In Michelle Minnick and Na-Rae Han (eds.), Penn Working Papers in Linguistics. Meinunger, André (1995) Some Remarks on Argument Ordering in German: an Endorsement for a Universal Hierarchy. FAS Working Papers 4:83–98. —— (2000) Syntactic Aspects of Topic and Comment. Amsterdam: Bejamins. Miyagawa, Shigeru & Takae Tsujioka (2004) Argument Structure and Ditransitive Verbs in Japanese. Journal of East Asian Linguistics 13: 1–38. Platzack, Christer (1986) Comp, Infl, and Germanic Word Order. In Lars Helland & Kirsti Koch Christensen (eds.), Topics in Scandinavian Syntax, pp. 185–234. Dordrecht: Kluwer. Putnam, Michael (2006) Prolific Scrambling: A Radical Approach to Middle Field Scrambling in West Germanic. Ph.D. dissertation: University of Kansas. Pylkkänen, Liina (2002) Introducing Arguments. Ph.D. dissertation: MIT. Richards, Marc D. (2005) Object Shift and Scrambling in North and West Germanic: A Case Study in Symmetrical Syntax. Ph.D. dissertation: Robinson College. Rizzi, Luigi (1997) The Fine Structure of the Left Periphery. In Liliane Haegeman (ed.), Elements of Grammar, pp. 281–337. Dordrecht: Kluwer. Webelhuth, Gert (1989) Syntactic Saturation Phenomena and the Modern Germanic Languages. Ph.D. dissertation, University of Massachusetts-Amherst. —— (1992) Principles and Parameters of Syntactic Saturation. Oxford: Oxford University Press. —— (1995) X-bar Theory and Case Theory. In Gert Webelhuth (ed.), Government and Binding Theory and the Minimalist Program, pp.15–96. Oxford: Blackwell. Wurmbrand, Susi (2002) Syntactic versus Semantic Control. In C. Jan-Wouter Zwart and Werner Abraham (eds.), Studies in Comparative Germanic Syntax: Proceedings from the 15th Workshop on Comparative Germanic Syntax, pp. 93– 130. Amsterdam: Benjamins. WHY SCRAMBLED DATIVES IN GERMAN CAN’T BIND ANAPHORS 309

Contact information:

Michael T. Putnam University of Michigan Department of German Languages and Literatures 3110 MLB 812 E. Washington St. Ann Arbor, Michigan 48105 e-mail: [email protected]

Jari Sivonen

An Exercise in Cognitive Lexical Semantics: The Case of the Finnish Motion Verb Kiertää

Abstract

This paper explores the different usages of Finnish motion verb kiertää within the theoretical framework of Cognitive Grammar. First, the different objective motion process types profiled by kiertää (as in the sentence Lentokone kiertää kentän yllä ’The plane is circling above the field’) are analyzed, and then, the different abstract and subjective motion types of kiertää based on the objective motion usage (for example, Tie kiertää pihaan ’The road bends into the yard’) are discussed. The paper shows that the different process types of kiertää are not random but semantically well motivated. The paper emphasizes that the (too often neglected) analysis of Path provides an interesting and rewarding viewpoint to the lexical semantics of verbs of motion. The paper also attests that Cognitive Grammar provides solid tools for describing and analyzing the lexical semantics of verbs of motion.

1. Introduction

In this paper, I will discuss the semantics of the Finnish verb of motion kiertää1 within the theoretical framework of Cognitive Grammar. My goal is twofold. According to Seuren (2001: 236), Cognitive Grammar is particularly useful in lexical semantics, and my first goal is to test the relevance of this argument by analyzing one polysemous verb of motion. My second aim is more practical. Although semantic studies of verbs of motion and typologies based on these studies have been quite popular,

1 According to the web-based MOT 98 dictionary (provided by Kielikone Oy), kiertää can be translated into English, depending on the context, by the verbs bypass, circle, circulate, circumvent, encircle, evade, fudge, get (a)round, go (a)round, hand (a)round, orbit, revolve, rove, skirt, surround or tour.

SKY Journal of Linguistics 18 (2005), 311–340 312 JARI SIVONEN the emphasis of the analysis has usually been on the manner of motion or on whether the motion expressed by the verb is goal-oriented (for example, arrive) or source-oriented (for example, leave). However, regardless of the importance of the path in the motion event, the path itself has usually gained less attention and remained almost unexplored in the lexical semantics of verbs of motion. Thus, it can be claimed that there is an evident demand for an analysis of the path. My second aim here is to fill in this gap in our knowledge. The semantic analysis of kiertää is especially useful in the pursuit of these two goals for the following reasons. i) Salience of verbs of motion

Verbs of motion are considered the most prototypical representatives of the entire verb category (Miller–Johnson-Laird 1976: 527, 529–530, 547), and the field of motion verbs has been characterized as an essential area of the lexicon (Diller 1991: 241). Bearing this in mind, it is fruitful to test the capacity of the lexical analysis of any semantic theory by using verbs of motion: a theory should be able to give a sufficient description of precisely these highly essential lexemes in order to serve as an adequate framework. ii) Polysemy

According to Leino (1994b: 12), Cognitive Grammar has been frequently used to analyze polysemy. Since kiertää is a highly polysemous motion verb that profiles a number of distinctive process types, Cognitive Grammar should be able to provide a good analysis of the different senses of this single verb lexeme. Kiertää denotes objective, abstract and subjective motion. In this paper, I will illustrate the semantic motivations of the different meanings in the different semantic domains,2 and in this sense, I will also test the competence of Cognitive Grammar in such analytic description.

2 This study is purely qualitative in nature; for a quantitative description of kiertää, see Sivonen 2005. AN EXERCISE IN COGNITIVE LEXICAL SEMANTICS 313 iii) Indirectness as the marked type of motion

When referring to concrete motion, kiertää encodes the Trajector's path (the mover's route in Cognitive Grammar terminology) as indirect when viewed from above in a two-dimensional space. By “indirect” I mean a path that has one or more explicit lateral movements with regard to the main direction of movement. For example, an English verb outflank, when referring to concrete motion, profiles a path that involves marked lateral movement compared to straight, rectilinear progression. It is probably a generally accepted fact that the great majority of verbs of motion in any language do not particularly specify the shape of the path at all (e.g. go, come, run, walk etc.), but even in these processes, the path is normally understood as being direct. Thus, it can be argued that the canonical, unmarked way of conceiving of a motion event involves encoding the path as direct rather than indirect. Keeping this in mind, it is worth investigating what kind of semantic features are emphasized when the path is, unexpectedly, indirect. In this study, I take one verb expressing indirect path, contrast its path to the direct route and consider how this path differs from the direct counterpart. It is no surprise that, in Modern Standard Finnish, there are several other verbs of indirect path that overlap kiertää in many respects. For example, the verbs kaartaa ’bend’ and mutkitella ’meander’ share many features with kiertää, but as pointed out in Cognitive Grammar (Langacker 1988: 10–11), the verbs are not semantically identical, i.e. “synonymous”, even though they may be used as functional counterparts in certain contexts. The choice of kiertää from among the others alternatives is justified since it is probably the semantically most complicated verb encoding indirect path in Modern Standard Finnish and therefore a good way to illustrate indirect path. I this paper, I will show what the indirect paths expressed by kiertää are actually like (or how they are conceived to appear) by taking a closer look at them. iv) Semantic features related to indirect path

In addition to indirect path, the verbs of indirect path have other semantic features that specify their processes. For example, the verb kiertää profiles the moving entity (the Trajector) and the location where, or with respect to which, the movement takes place in its processes (the Landmark). Since 314 JARI SIVONEN indirectness can be considered the marked type of path, it is worthwhile to find out what other kinds of semantic contributions are related to it. In this study, I will also shed some light on this issue by analyzing the verb kiertää. The examples I used in the analysis came from the database Language Bank of Finland, which includes 2 699 tokens of kiertää and almost 180 million tokens altogether. My data consists of written (modern) Standard Finnish mostly taken from several newspapers of the 1990’s by statistically reliable measures. Occasionally, however, I use my own examples for illustrative reasons.

2. Some basic tools of Cognitive Grammar

Cognitive Grammar is a wide-ranging linguistic theory developed mainly by Ronald W. Langacker3 (for example, 1987, 1988, 1991 and 1999), which emphasizes the role of meaning in linguistic analysis. Cognitive Grammar aims to be a psychologically realistic theory, and it views language as closely related to other human cognitive phenomena. Cognitive Grammar deals with mental representations, while many other modern semantic theories use more formal notation. Cognitive Grammar emphasizes the bond between syntax and semantics and aims to integrate different dimensions of linguistic structure. (Langacker 1987: 42; Leino 1993: 54.) In this paper, I will make use of some basic concepts of Cognitive Grammar, which are introduced below. In Cognitive Grammar, the Trajector refers to the most salient entity in the relational predication (predication is a concept used in Cognitive Grammar to refer to the semantic description of a linguistic unit), one of which is a process expressed by a verb of motion. The Landmark refers to the second most salient participant in the relational predication. (Langacker 1987: 217.) In a process expressed by a motion verb, the Trajector is prototypically represented by the syntactic subject of the sentence, whereas the Landmark is expressed by the syntactic object (Langacker 1990: 33) or by a valence adverbial (Leino 1993: 85−86; Leino and Onikki 1992: 36).

3 Verbs of motion within a cognitive framework are also discussed by Leonard Talmy (for example 1985 and 2000). AN EXERCISE IN COGNITIVE LEXICAL SEMANTICS 315

As I see it, the path expressed by a verb is highly relevant in the lexical semantics of verbs of motion. According to Talmy (1975: 181), the path can be defined as “the respect in which one object is considered as moving or located to another object” in a motion situation. In other words, the path can be understood as a route on which the Trajector is moving in, into or with respect to the Landmark. I also make some use of Basic Conceptual Relations. These refer to the relation between the Trajector and the Landmark of a relational process. Depending on the amount of space between the Trajector and the Landmark, at least five different Basic Conceptual Relations can be distinguished: SEPARATION, INCLUSION, IDENTITY, CONTACT and ASSOCIATION. In SEPARATION, there is some space between the location of the Trajector and the Landmark, whereas in INCLUSION, the Trajector is situated inside the Landmark. In CONTACT, the Trajector and the Landmark have one common intersection, and IDENTITY refers to a situation where the Trajector and the Landmark are identical with respect to their comparable parts. ASSOCIATION is a conceptual relation where, besides the Trajector and the Landmark, there is a third, usually functional, participant involved in the interaction of the state of affairs. (Langacker 1987: 225−230; Lakoff 1987: 420.) The processes expressed by verbs of motion are usually divided into three subtypes in Cognitive Grammar. These three motion types include objective, abstract and subjective motion. Objective motion refers to a state of affairs where some entity changes its location in the external world (for example, A plane was circling above the field). Thus, objective motion refers to a type of motion that is intuitively understood as motion, and it can also be called “concrete motion”. In abstract motion, verbs of motion are used, but no actual moving entity can be observed in the state of affairs referred to by the sentence. This kind of motion verb usage is often called “metaphorical” or “figurative”. For example, a sentence A nasty rumor was circulating in the office may be characterized as a abstract motion. In this case, the content of the rumor is conceived of as moving from one speaker to another even though there is no visible moving entity involved. It is precisely the subjective or virtual conceptualization of movement that motivates the usage of motion verb in this kind of occurrence. Subjective motion (also “fictive motion”, “virtual motion”) can be considered a subtype of abstract motion, but since it denotes a 316 JARI SIVONEN distinguishable type, it is reasonable to consider subjective motion a unique type of motion. In subjective motion, as in the sentence The fence zigzags from the door to the field, the verb of motion is used to refer to the physical shape of an external entity. Subjective motion is based on a dynamic, abstract sort of interaction between the conceptualizing observer and the location referred to. In the process of subjective motion, the observer mentally “follows” the shape of the Trajector’s referent from one end to the other and conceptualizes its shape as a motion event. For example, in the subjective motion of kiertää, the conceptualizing observer follows mentally the shape of the Landmark’s referent and finds its shape similar to the path of a Trajector in the objective motion usage of kiertää. Thus, the similarity of the Trajector’s path in the objective motion compared to the shape of the Trajector’s referent motivates the use of the same verb lexeme. (Different types of motion are discussed in detail by Langacker 1987: 170−173, 1990: 157−160 and Talmy 2000: 99–175.)

3. Cognitive Lexical Semantic of kiertää

According to the Finnish Etymological Dictionary, the verb kiertää has the same root as the adjective kiero ’indirect, wrong, twisted’ (SSA 1992: s.v. kiertää). These semantic features are also involved in the processes expressed by kiertää. The verb kiertää includes a derivative affix tä, which refers to factivity, meaning the making of what the stem expresses (Hakulinen 1979: 288). In the process profiled by kiertää, this can be understood as a making of a certain type of twisted path. Before going into details, it is important to clarify my starting-point for lexical verb semantics. I presume that a polysemous verb itself is the key element to this kind of . The verb has an ability to profile a limited set of process types (Langacker 1999: 23, 271), and one of these is normally elaborated in the actual usage of the verb. Thus, neither tense, Aktionsart, the transitivity–intransitivity dichotomy, the verb’s argument structure (e.g. its argument’s cases, word classes, etc.) nor any other formal feature will explain exhaustively any particular meaning of the verb. On the contrary, I believe things are the other way around. A verb lexeme profiles a process type or, in a polysemous case, a number of process types that have several specifications regarding suitable argument AN EXERCISE IN COGNITIVE LEXICAL SEMANTICS 317 structure, Aktionsart, transitivity–intransitivity, etc.4 Consider the example sentence 1 (my own).

(1) Nykyisin Pekka kiertää baareja. Nowadays Pekka-nom circle5-3.sg bar-pl-part ’Nowadays Pekka never goes into bars, avoids bars.’ ’Nowadays Pekka keeps going from one bar to another.’

The sentence 1 has two explicit interpretations, and only what we know about Pekka can tell us whether he never goes into bars or perhaps goes there too often. In other words, there is no formal marker, such as a morpheme, which could determine the reading, but the interpretation is done on the basis of encyclopedic knowledge. Naturally, not all process types profiled by kiertää are similar to example 1, and some of them can be distinguished by formal features. The transitivity–intransitivity dichotomy, the case and the category of the Landmark (locative adverbial or object) in addition to the encyclopedic knowledge implicit in the verb’s arguments characterize some process types profiled by kiertää.

3.1 Objective motion

In the objective motion profiled by kiertää, four different processes can be distinguished based on the path’s shape and the nature of the Conceptual Relation between the Trajector and the Landmark. Below, I discuss these in detail.

3.1.1 ’Moving inside the Landmark’

The basic process type in the objective motion of kiertää is one where the Trajector is moving inside the Landmark (2 and 3). These processes are inherently imperfective, and the conceptual relation between the Trajector and the Landmark is INCLUSION during these processes. There are two

4 However, this is a difficult theoretical dilemma that cannot possibly be answered in this paper. Basically, it is a question of whether to choose a verb lexeme or a larger construction to be the starting point of the linguistic description. 5 In the glosses, I will use systematically the English verb circle for Finnish kiertää because I assume its stem to be the closest to the Finnish stem kier. In the translations, I will try to interpret the particular sense of kiertää into English as well as possible by using different English verbs. 318 JARI SIVONEN subtypes of these processes, intransitive and transitive. In the intransitive type, the Landmark is represented by a syntactic adverbial, usually in internal local cases, as the inessive in sentence 2 (Lapissa). The Landmark of a transitive process is an object (3: Lappia), which shows the normal variation of the Finnish object case between accusative (actually expressed by nominative or genitive) and partitive.

(2) Kesällä kierrämme Lapissa. summer-adess circle-1.pl Lapland-iness ’During the summer, we travel around Lapland.’

(3) Talvella kierrämme Lappia. winter-adess circle-1.pl Lapland-part ’During the winter, we tour in Lapland.’

It is important to notice that the encyclopedic knowledge implicit in the Landmark plays an important role when defining the process type in question. For example, in sentence 3, what we know about the object Lappia (Lapland-part ’in Lapland’) defines the usage of kiertää in this context and makes it different from the process type described later in section 3.1.2. (compare to example 10). This also illustrates why mere formal criteria are not sufficient to define the process type of kiertää. But this is also true the other way around: you cannot take a certain Path or process type expressed by kiertää and then define the formal syntactic context in which it must occur. One basic assumption in Cognitive Grammar is that different syntactic constructions are not semantically identical, although they can be used as functional counterparts (Langacker 1988: 10−11). Bearing this in mind, it can be argued that the examples 2 and 3 both mean that the Trajector moves in the Landmark through an indirect path of a certain shape, but since the Landmark is represented by an adverbial modifier in the former sentence and by an object in the latter, the sentences are likely to involve at least some semantic difference. A non-controversial explanation of this semantic difference would be to say that, in the transitive sentence, the Landmark is conceived more as a target of the motion action similarly to a prototypical transitive sentence (for example A boy reads a book). Thus, the transitive construction of the motion verb kiertää is schematically similar to the prototypical transitive verb’s process. In this sense, even with a motion verb, the object represents AN EXERCISE IN COGNITIVE LEXICAL SEMANTICS 319 the Landmark as some sort of target of the action expressed by the verb, whereas the locative adverbial profiles the Landmark more explicitly as a location. In the transitive construction, the object’s case variation can therefore be used to express the state of motion as either imperfective with the partial object case partitive or perfective with the total object case (genitive-accusative or nominative-accusative). However, it seems to me that this formal variation of the Landmark’s case may have another semantic contribution as well. My suggestion is that the way the path is conceptualized plays an important role in this semantic disparity. Prototypically, in this type of intransitive use of kiertää (2), the path is conceived of as inherently uniform in quality, whereas the transitive construction (3) places some salient stages, certain intermediate stopping points into the path, which are highlighted. Figure 1 demonstrates these two different interpretations. The Trajector's path in the (prototypically) intransitive construction (a) has no dots, while the salient stopping points are marked with black dots in the (prototypically) transitive construction (b). In the figures, the circle stands for the Trajector, the square for the Landmark and the arrow line for the path. 320 JARI SIVONEN

LM

TR

LM

TR

Figures 1a and 1b. The path expressed by kiertää in the intransitive construction (a) and in the transitive construction (b) when the Trajector is moving inside the Landmark.

The existence of such intermediate stopping points can be supported by a simple syntactic test, a kind of minimal pair. For example, the sentence Kotisairaanhoitajan työssä saa kiertää kaupunkia ~ ?kaupungilla (visiting nurse-gen work-iness get-3.sg circle-1.inf city-part ~ city-adess) ’As a visiting nurse one gets to circulate in the city’, where the Landmark represented by an object (in a : kaupunkia) is fully acceptable, while its counterpart with a syntactic adverbial (in an : kaupungilla) may sound odd to a native speaker. This is due to the fact that the transitive construction of kiertää implies the presence of stopping points, while the intransitive sentence indicates their absence, which makes the intransitive construction semantically defective with regard to the speaker’s encyclopedic knowledge of a visiting nurse’s job description. AN EXERCISE IN COGNITIVE LEXICAL SEMANTICS 321

However, the presence or absence of stopping points in the Trajector’s path does not depend merely on the syntactic type of the sentence and the Landmark's representative. It should be emphasized that syntactic means can be used both to create salient stages, e.g. stopping points, into the path of an intransitive construction and to eliminate them from the transitive construction. In the intransitive example in sentence 4, the idea of stopping points is due to a plural affix in the Landmark, which brings about the impression of intermediate stopping points. In the transitive sentence 5, on the other hand, the impression of an inherently uninterrupted path with no stopping points is based on the adverbial phrase idiom ristiin rastiin ’through the length and breadth of’.

(4) Ryhmä kiertää Oulun yläasteilla. team-nom circle-3.sg Oulu-gen junior high school-pl-adess ’The team visits junior high schools of Oulu.’

(5) Kierrämme maailmaa ristiin rastiin. circle-1.pl world-part cross-illat tick-illat ’We travel through the length and breadth of the world.’

Interestingly enough, Finnish and English seem to have different ways of encoding salient stages into the path. Whereas Finnish uses case variation of the Landmark, English uses lexical means. According to the New Oxford Dictionary of English (s.v. roam, tour), for example, the verb roam means ’move about or travel aimlessly or unsystematically, especially over a wide area - - - without stopping - - - without dwelling on anything in particular’, and the verb tour is explained as ’to make a tour (= a journey for pleasure in which several places are visited / to view or inspect something)’. To my knowledge, there are no corresponding motion verbs in Standard Finnish, but a similar semantic difference can be implied by the case variation of the Landmark of the verb kiertää as shown here. An interesting addition to the semantics of the verb kiertää in its objective motion usage is provided by transitive constructions with a total, nominative-accusative object in plural (6 and 7). 322 JARI SIVONEN

(6) Kierrämme pohjoisen kunnat circle-1.pl northern-gen municipality-pl-nom-acc kertomassa lainsäädännöstä. tell-3.inf-iness legislation-elat ’We will visit all the northern municipalities telling about the legislation.’

(7) Yhdessä kierrettiin lääkärit ja hoidot. together circle-pass-imperf doctor-pl-nom-acc and treatment-pl-nom-acc ’Together we tried all doctors and treatments.’

The sentences 6 and 7 suggest that all entities included in the Landmark are in the scope of the predication, e.g. all municipalities (6) or doctors (7) were visited. This type of motion verb usage is parallel to the “normal” transitive verb use, where the total object in plural nominative-accusative implies that all of the object’s referents are encompassed in the process expressed by the verb, as in the sentence Pekka luki kirjat (Pekka read-3.sg- imperf book-pl-nom-acc ’Pekka read all the books’), the object (kirjat) case nominative-accusative implies that he read all the books that were available in that particular situation (Hakulinen−Karlsson 1979: 185; Heinämäki 1994: 225). The most interesting cases, however, are the ones like sentence 7, where, because there are countless real-world doctors and treatments that simply cannot all be visited, this way of conceptualizing the path (and the state of affairs) is actually subjective. The motivation for such subjectively conceptualized use of a path with salient stages is probably pragmatic: it provides a possibility to exaggerate6 and to create a dramatic, appraising impression (see Yli-Vakkuri 1986: 259−261). If a person visits all doctors and tries all treatments, it makes the point more effectively than visiting some doctors and using some treatments, which was probably what happened. In transitive constructions of kiertää, not only is the path saliently elaborated but the Trajector also has some semantic specifications. In transitive sentences, the Trajector appears to have planned in more detail the path in advance. This is most clearly shown in sentences that refer to future action, as in examples 8 and 9 (number 8 is an authentic example taken from newspaper Kaleva 21.8.2001), even though reference to the future is not necessary.

6 Vilkuna (1992: 51) even calls comparable subjective case usage “legal exaggeration”. AN EXERCISE IN COGNITIVE LEXICAL SEMANTICS 323

(8) Lomallaan presidentti Putin kiertää Karjalaa. holiday-adess-3.sg-px president-nom Putin circle-3.sg Karelia-part ’During his holidays, president Putin will tour around Karelia.’

(9) Lomallaan presidentti Putin kiertää Karjalassa. holiday-adess-3.sg-px president-nom Putin circle-3.sg Karelia-iness ’During his holidays, president Putin will tour around in Karelia.’

In example 8, where the Landmark is represented by an object (partitive case), the path is conceptualized as well-thought-out beforehand compared to the expression with an adverbial (). The intransitive sentence 9 may even sound strange to the ears of a native speaker. This is due to the fact what we know about the president’s holiday journeys: they are well-planned and scheduled beforehand, but the intransitive construction of kiertää implies the opposite, which may cause some confusion to the reader of the sentence.

3.1.2 Circumferential Path

In the second objective motion process type expressed by kiertää, the Trajector moves on a path of a special shape. The sentences 10 and 11 are examples of this process type.

(10) Kierrän kaappia ja tutkin sitä. circle-1.sg closet-part and inspect-1.sg it-part ’I go around the closet and inspect it.’

(11) Lapsi kiertää puun ympärillä. child-nom circle-3.sg tree-gen around-adess ’A child is running around the tree.’

A path of this kind can be characterized as circumferential. Svorou (1994: 152) defines circumferential path as an instance where the Landmark is treated as a spherical object, either because of its shape or because of the path that is followed by a moving entity that starts from a specific point and moves along the boundaries of the Landmark till it reaches the same starting point. The Landmark of this process may refer to an artifact represented by a syntactic object, around which the Trajector moves (10). It is also possible 324 JARI SIVONEN that the Landmark is expressed by a valence adverbial construction, which typically consists of a noun in genitive accompanied by a p-position ympärillä ’around’, which profiles the path and process type in question (11). Thus, this process type can be expressed both by transitive and intransitive constructions, and the form of the Landmark varies accordingly. The encyclopedic knowledge associated with the Landmark has a vital role in the determination of the process type and, thereby, in the interpretation of the whole sentence (compare to sentence 3). It must be emphasized that the shape of the path profiled by kiertää in this process type may be similar to the shape profiled by a verb in the first group (illustrated in section 3.1.1). The semantic difference between these two process types lies in the conceptual basic relation between the Trajector and the Landmark. When the Trajector is moving inside the Landmark, the relation can be characterized as INCLUSION, but in the process type illustrated here, this relation is rather SEPARATION or CONTACT. The circumferential usage of kiertää can be divided into several subtypes. First of all, the Trajector can move around the Landmark, as we saw in the examples 10 and 11. If we go into details, this subtype can profile three shapes of path with differential syntactic representations and semantic readings. The example sentences 12 and 13 illustrate these types.

(12) Kirkko kierrettiin pyhää church-nom-acc circle-pass-imperf holy-part vettä heittäen. water-part throw-2.inf-instr ’The church was circled once by people throwing holy water.’

(13) Kirkkoa kierrettiin pyhää vettä heittäen. church-part circle-pass-imperf holy-part water-part throw-2.inf-instr ’There were people going round and round the church and throwing holy water.’ ’There were people in the process of going around the church and throwing holy water.’

If the Trajector moves only one round (see Figure 2a), this is expressed by a nominative object (12). The sentence has a bounded, perfective reading. If the object is in the partitive case (13), there are two possible readings: the Trajector either moves several circles (Figure 2b) or is halfway through the first round, but not yet finished with it (Figure 2c: the dotted line stands for the unfinished part of the path). The latter two cases have unbounded, imperfective readings. Figure 2 illustrates the different paths. AN EXERCISE IN COGNITIVE LEXICAL SEMANTICS 325

TR LM

TR LM

TR

LM

Figures 2a, 2b and 2c. Circumferential use of kiertää. Perfective (a), imperfective ’many rounds’ (b) and imperfective ’part of one round’ (c) path.

326 JARI SIVONEN

In the examples 14 and 15, the Trajector is not moving around an entity, but rather along a manmade track. Because the shape of the path is, nevertheless, similar to the previous ones, these paths can also be characterized as circumferential paths. It is typical of these paths that the starting and ending points of the track are at the same real-world location, as in the track of a stadium. However, kiertää can also be used in the same sense to refer to a situation where the starting point is not even close to the ending location. This is typical of paths representing nature trails (14) and certain sport events, such as the routes in a car race (15).

(14) Kuhasalon luontopolku on kierretty Kuhasalo-gen nature trail-nom-acc is-3.sg circle-pass-1.partic-nom kymmenen kertaa. ten-nom time-part ’The Kuhasalo nature trail has been walked ten times.’

(15) Grönholm kiersi 3 kilometrin radan Grönholm-nom cirle-3.sg-imperf 3 kilometer-gen track-gen-acc toiseksi nopeimmin. second-transl fast-sup-cl ’Grönholm raced the 3 kilometer track the second fastest.’

The motivation for this kind of use of kiertää is probably the way the Trajector's state is conceived of. The similarity to a track that starts and ends at the same location lies in the fact that, in both cases, the Trajector is no longer “on the path” when it has stopped moving. For example, when the car in a race crosses the finish line, the race is over, and the car in a way returns to its initial condition or state of affairs, even though there is no actual physical entry to the starting point at the end of the motion process. In Cognitive Grammar, this may be characterized as a change from one cognitive dominion, a “conceptual region to which a particular reference point affords direct access” (discussed in detail in Langacker 1993), to another.

3.1.3 ’Moving to avoid the Landmark’

The sentences 16 and 17 are examples of the third objective motion process type expressed by kiertää. AN EXERCISE IN COGNITIVE LEXICAL SEMANTICS 327

(16) Kierrä haaskat, koska circle-2.sg-imperat carcass-pl-nom-acc since karhu on todennäköisesti lähellä. bear-nom is-3.sg probably nearby ’Avoid the carcasses, since the bear is probably nearby.’

(17) Kierrämme aidan ja astelemme ovelle. circle-1.pl fence-gen-acc and pace-1.pl door-allat ’We will go past the fence and walk towards the door.’

In this process type, the Trajector tries to avoid CONTACT with the Landmark. Therefore, the Landmark is conceptualized as a BARRIER, an entity to be avoided (for BARRIER, see for example Rudzka-Ostyn (1988: 523)). The BARRIER is located in the Trajector's original direction of movement, and in order to avoid CONTACT with it, the Trajector needs to take some steps to the side. It is worth noticing that kiertää profiles only lateral movement of avoidance, but does not specify whether the Trajector passes the Landmark on the right or the left hand side, nor does it specify how far from the Landmark the Trajector moves. Having passed the BARRIER, the Trajector usually returns to its original route, but this part of the motion is not profiled in the process of kiertää, but rather by another motion verb, as the verb astella ’pace’ in the example 17 shows. The dotted line in Figure 3 stands for the Trajector’s initial path. 328 JARI SIVONEN

TR

LM

Figure 3. The process of kiertää where the Trajector is avoiding CONTACT with the Landmark. This process type is transitive, and the Landmark is represented by a syntactic object, which has normal case variation that can be used to express aspectual variation. In example 16, the information about a bear being close in the latter sentence guides the interpretation of the first sentence: the verb kiertää has here the meaning of avoidance of the entity expressed in the Landmark. The encyclopedic knowledge associated with the Landmark may also have a crucial effect when deciding about the process type. Thus, what is generally known about fences in example 17 (they cannot be walked through) guides the interpretation of this sentence and makes it different from, for example, the sentences 6 and 15.

3.1.4 ’Moving into the Landmark’

The final, fourth, objective motion process type profiled by kiertää refers to a motion where the Trajector moves into the location expressed by the Landmark (18). However, kiertää does not specify precisely the entrance point of the Landmark, but only the fact that the Trajector goes into the Landmark. Figure 4 illustrates the path.

(18) Kierrän lammen päätyyn. circle-1.sg pond-gen end-illat ’I will walk around to the end of the pond.’

AN EXERCISE IN COGNITIVE LEXICAL SEMANTICS 329

LM TR

Figure 4. The Trajector's path expressed by kiertää in a process of entering the Landmark. This process type is intransitive by nature. From the point of view of conceptual relations, it can be said that, in this inherently bounded, perfective process, the Trajector moves from SEPARATION to INCLUSION (or sometimes CONTACT) with respect to the Landmark. This is why the Landmark is represented in this process type by an adverbial modifier in an illative or , which are called “movement towards” cases by Karlsson (1987: 99). In the state of affairs referred to by kiertää in this process type, there is a BARRIER involved that prevents the Trajector from proceeding along a straight route. However, the BARRIER is not necessarily shown explicitly in the sentence but its presence can be inferred. In the example sentence 18, the BARRIER is represented by a lake, which cannot be walked through but has to be bypassed.

3.2 Abstract motion

It is a generally accepted view that the direction of semantic change is usually from concrete to abstract (Leino 1993: 153). In this respect, kiertää makes no exception. On the contrary, all abstract usages of kiertää can be seen as semantic extensions of some type of objective motion. In other words, the figurative usage of kiertää is semantically motivated by the concrete usage of the verb. In the following Figure 5, I give a general view of the polysemy of kiertää. The figure shows a schematic presentation of the objective motion process types expressed by kiertää discussed earlier in this paper. 330 JARI SIVONEN

Moreover, the figure illustrates the semantic motivations of the different abstract senses of kiertää that will be discussed in detail below. In other words, Figure 5 shows how the abstract usage of kiertää is based the objective senses of the verb. The arrow stands for the assumed semantic extension and points to the direction of the semantic motivation of the different senses of kiertää.

AN EXERCISE IN COGNITIVE LEXICAL SEMANTICS 331

3.2.1 ’Avoiding to do something’

In the first abstract motion process type of kiertää, the Trajector is avoiding to do something. The avoidable thing can be, for instance, the need to obey the law or some regulation (19) or a certain topic in a speech act (20). This process type is transitive, and the Landmark is represented by an object, which shows normal aspectual case variation. The reader’s encyclopedic knowledge of the Landmark guides the interpretation of the sentence. Thus, for example, in the sentences 19 and 20, the objects määräys ’regulation’ and totuus ’truth’ determine the process type in question. Since they are abstract entities, the motion must be abstract rather than objective, and, moreover, the specific process type in question, ’Avoiding to do something’, is due to the information related to the Landmarks.

(19) Määräystä kierrettiin eri tavoin. regulation-part circle-pass-imperf different way-instr ’The regulation was evaded in different ways.’

(20) Kiertäisin totuutta, jos sanoisin näin. circle-1.sg-cond truth-part if say-1.sg-cond this ’I would be fudging the truth if I said this.’

It is easy to see that such figurative usage of kiertää is based on the objective motion where the Trajector tries to avoid CONTACT with the Landmark (3.1.3.). The Landmark is here conceived of as a BARRIER, and the action taken by the Trajector is to mentally avoid CONTACT with it, just as the Trajector avoids touching the Landmark concretely in the corresponding object motion usage.

3.2.2 ’Circumstantial chance’

The other abstract motion processes profiled by kiertää can be called circumstantial chance (even though they involve many kinds of usage), and they can be divided into two relatively distinct subtypes. Both transitive and intransitive sentences occur in these processes, and the Landmark can therefore be represented by an adverbial or an object. The adverbial can be a compounding construction, which consists of a pair of adverbials in different local cases expressing the starting and ending points or some 332 JARI SIVONEN intermediate phase of the virtual movement process (21 and 22), or it can be in internal local case (inessive or adessive, see example 24).

(21) Puheenjohtajuus kiertää maalta toiselle. chairmanship-nom circle-3.sg country-ablat other-allat ’Chairmanship rotates from one country to another.’

(22) Näyttely kiertää kivikaudesta muovirahaan. exhibition-nom cirle-3.sg Stone Age-elat plastic money-illat ’The exhibition consists of items from the Stone Age to plastic (money).’

It is mostly the encyclopedic knowledge associated with Trajector’s representative, which determines the process type. In this process type, the Trajector refers to some abstract entity, whose virtual location is conceived of as moving. In the first case, the Trajector, in a sense, abstractly moves, acts or appears iteratively in the location expressed by the Landmark. One elaboration of this process type is example 21, where the chairmanship of a certain organization is rotated from one country to another. The second type refers to a process where the Trajector’s profile is conceptualized as changing steadily without any clear intermediate stopping points; in the state of affairs referred to by sentence 22, the items of the exhibition are from different eras. The semantic difference between these two process types lies in the existence of salient stopping points. Example 21 profiles a process where the holder of chairmanship is a certain country for some time, followed by another country. Because these countries are located all over Europe in the real world, the changing (similarly to the change of the Trajector’s location of the in objective motion) of the host country is conceptualized as a movement along a path from one salient stage to another.7 Thus, the host countries are conceptualized as the salient stages in the path expressed by kiertää (compare to example 3), which motivates the usage of this particular verb in this context. In sentence 22, however, there are no salient intermediate points profiled in the path. The sentence is metonymical in nature, and can be interpreted to mean that the exhibits (rather than the exhibition itself) are from different eras. The variation of these exhibits is

7 Because this example does not actually refer to the shape or form of the real-world Trajector, but rather denotes the quality of chairmanship, I categorize it as an example of abstract rather than subjective motion.

AN EXERCISE IN COGNITIVE LEXICAL SEMANTICS 333 conceived of as subjective movement: the items are conceptualized as moving entities and the implicit visitor as immobile. In reality, however, things happen the other way around. What about the semantic difference in such examples as 23 and 24, which also profile the abstract motion of kiertää.

(23) Ajatukseni kiertävät sinua. thought-pl-nom-px circle-3.pl you-part ’My thoughts are encircling you.’

(24) Ajatukseni kiertävät sinussa. thought-pl-nom-px circle-3.pl you-iness ’My thoughts are encircling you with great intensity.’

If similar variation between the partitive and the inessive cases in the Landmark causes the implication of planning in the objective motion of kiertää (see examples 8 and 9), what is its influence on abstract motion (23 and 24)? The answer probably lies in the intensity of the process. In 23, the partitive Landmark profiles an iterative thinking process and seems to imply the possibility of thinking about something else every now and then, whereas in 24, the inessive case Landmark guides the reader to perceive the state of affairs as more intensive. This semantic phenomenon is based on the fact that the Finnish grammatical cases, such as partitive, profile an entity, whereas the local cases, such as inessive, profile an atemporal relation (see Leino 1994a: 209−210, 213−214). In the abstract motion usage of kiertää, this means that partitive creates the impression that thinking has a target, but this target does not necessarily need to be attained all the time. Inessive in the Landmark, on the contrary, profiles a setting of abstract motion and therefore implies incessant thinking of the target. An additional interesting point is the fact that, if the Landmark is expressed by a total object case (accusative) in this sentence, Ajatukseni kiertävät sinut (thought-pl-nom-px circle-3.pl you-acc), the interpretation is ’My thoughts are bypassing you completely’, ’I am not giving you any thought at all’.

3.3 Subjective motion

The subjective motion usage of kiertää is based on a mental conceptualization where the different paths of objective motion play an 334 JARI SIVONEN important role. Even though subjective conceptualization can also be involved in other kinds of usage of kiertää (as examples 21 and 22), I have categorized as subjective motion only the cases where the motion verb is used to refer to the Trajector’s concrete occurrence, its shape in the real world. The prototypical cases are shown in the sentences 25 and 26.

(25) Maratonreitti kiersi järven rantaa. marathon route-nom circle-3.sg-imperf lake-gen shore-part ’The marathon route followed the lake shore.’

(26) Kauniit käytävät kiertävät pihan. beautiful-pl-nom pathway-pl-nom circle-3.pl yard-gen-acc ’The yard is circled by beautiful pathways.’

The subjective motion usage of kiertää can be expressed by an intransitive sentence or by a transitive sentence. In a transitive sentence, the case of the Landmark’s representative, syntactic object, can be either partitive (26), when the aspect is imperfective, or genitive-accusative (27) or nominative- accusative, which makes the sentence perfective. Example 25 shows the most typical subjective motion use of kiertää. In such processes, the physical shape of the Trajector’s (maratonreitti ’marathon route’) referent is composed of the actual shape of the referent of the Landmark (järven ranta ’lake shore’). In other words, the marathon route, when observed from above, follows the path of an entity that is moving more or less close to the shoreline of the lake (see example 13). The interpretation of subjective motion is mostly due to the encyclopedic knowledge associated with the Trajector; in example 25, the subject maratonreitti ’marathon route’ brings about the reading of some elongated entity that is situated in the external world. Example 26 profiles an interesting subjective motion process, where the Trajector surrounds the Landmark. This kind of subjective motion has its roots in the circumferential objective movement where the Trajector moves once round the Landmark (see example 12). This is expressed by the genitive-accusative case of the Landmark (pihan). The examples 25 and 26 show how object case variation reveals the nature of the mental conceptualization process in subjective motion usage of this kind. Huumo (2002: 549) argues that, in subjective motion, the restrictive object (nominative-accusative and genitive-accusative; 26) implies that mental conceptualization reaches the end of the object’s AN EXERCISE IN COGNITIVE LEXICAL SEMANTICS 335 referent. In the partial object construction (partitive; 25), on the other hand, it can be claimed that only some part of the object’s referent is under mental contemplation. The last example of the subjective motion usage of kiertää is the sentence 27.

(27) Kylätie kiertää tilan. village road-nom circle-3.sg farm-gen-acc ’The village road bypasses the farm.’

The subjective motion in example 27 is based on the objective motion process where the Trajector is avoiding CONTACT with the Landmark conceived of as a BARRIER (16 and 17). Thus, the path of the Trajector bypassing the Landmark in the corresponding objective motion is similar to a village road, in example 27, which motivates the use of kiertää in this sense.

4. Problematic cases

Even though the data seem to fall quite neatly into the process types presented here, the categories I have created are prototypical by nature, and with a large amount of material, it is obvious that some cases are not easy to classify. It was shown earlier in this paper that many process types profiled by kiertää are inherently ambiguous in the sense that they can be interpreted in two or, occasionally, even in three ways. The sentence 28 has three different readings.

(28) Aion kiertää Kainuuta. intend-1.sg circle-1.inf Kainuu-part ’I intend to tour in Kainuu.’ ’I intend to avoid Kainuu.’ ’I intend to travel along the borderline of the Kainuu area.’

The semantic differences in example 28 can be described by using Conceptual Basic Relations. In the first interpretation of example 28, kiertää profiles a process where the Trajector moves in the Landmark (’I intend to tour in Kainuu’). In this case, the Conceptual Relation between the Trajector and the Landmark is INCLUSION. The second reading is that the Trajector is avoiding the area of Kainuu (’I intend to avoid Kainuu’), 336 JARI SIVONEN and in this process, the relation can be called SEPARATION. In the third possibility, the Trajector is moving along the borderline of the Kainuu area (’I intend to travel along the borderline of the Kainuu area’), and this is best characterized as CONTACT. The ambiguity of sentences like 28 is due to the fact that the sentence represents elaborations of three different process types profiled by kiertää, and there is no formal criterion to disambiguate the sentence. Thus, only a larger context or encyclopedic knowledge would indicate the intended reading. Similar examples of kiertää supporting the view of encyclopedic semantics are found in the sentences Matti kiersi Lontoota (Matti-nom circle-3.sg.-imperf London-part) ’Matti was touring London’, and Matti kiersi pöytää (Matti-nom circle-3sg.-imperf table-part) ’Matti was going round the table’. In both sentences, the Landmark of the process is represented by a partitive case object. However, the semantic difference is obvious: the former profiles a process where the Trajector is moving inside the Landmark’s referent, whereas the latter refers to motion outside the Landmark. I would argue that it is encyclopedic knowledge, such as knowledge about the size of the Landmark compared to the size of the Trajector, associated with the Landmark (and the Trajector), that determines the interpretations of such ambiguity. For example, the sentence Muurahainen kiersi pöytää (ant-nom circle-3.sg-imperf table-part) has two readings: ’An ant was circling around the table’ and ’An ant was circling on the table’. This is due to the fact that ants are such small creatures that they can quite easily be moving either beside or on a table, whereas a person, in this context, would usually be moving only around the table with no ambiguity involved. Some classification problems are also due to metonymy. A typical case is sentence 29. The example 29 can be understood as subjective motion if the Trajector’s representative vaellus ’hike’ is understood as a route of the hike, which is probably the most predictable reading. However, it seems to me that the Trajector can also be seen as referring to the people actually on the hike, e.g. moving along the route. If this interpretation, regardless of its peripheral probability, is accepted, sentence 29 is also an example of the objective motion of kiertää. AN EXERCISE IN COGNITIVE LEXICAL SEMANTICS 337

(29) Vaellus kiertää Kuusamossa. hike-nom circle-3.sg Kuusamo-iness ’The trail of the hike circles in the Kuusamo area.’ ’The people on hike are circling in Kuusamo.

5. Concluding comments

At the beginning of this paper, I set myself two goals. The first was to evaluate the functionality of Cognitive Grammar in the field of lexical semantics. There are many different competing theories in the field of lexical semantics, which makes the position of Cognitive Grammar an issue. I find the following two aspects of Cognitive Grammar, illustrated in this paper, the most powerful support to the theory’s capacity in lexical semantics, especially when describing verbs of motion. First, Cognitive Grammar seems to be able to cover adequately all important aspects of a motion event (e.g. the moving entity, the location of movement, etc.). Most of all, the shape of the path cannot be ignored when dealing with verbs of motion that particularly specify the path, and solid tools to concentrate on this aspect of a motion event are provided by Cognitive Grammar. The Cognitive Grammar’s practice to use figures to illustrate linguistic meaning is particularly useful in the semantics of verbs of motion. Second, the division of the motion event into three types, objective, abstract and subjective motion, seems intuitively reasonable and emphasizes the semantic motivations behind the different usages of a polysemous motion verb lexeme. As a whole, Cognitive Grammar places a particular emphasis on the senses of a polysemous verb that are often considered peripheral and therefore left out of linguistic analysis. It may even be said that, in this sense, “figurative” or “metaphorical” usage of motion verbs is actually not secondary at all but rather an equally expressive manner of speech. One detail to argue for Cognitive Grammar is the fact that the way subjective motion is handled in Cognitive Grammar is intuitively plausible and makes sense. I hope that my paper has been able to attain my second goal (concerning the importance of analyzing path) by illustrating that an effort to focus on the path, which has been rare so far, can certainly make a contribution to the lexical semantics of verbs of motion. Since I have concentrated here only on kiertää, the other Finnish verbs encoding indirect 338 JARI SIVONEN path are likely to shed more light on the question of what kind of things are salient when expressing indirect path. As I see it, there is no doubt that the path itself would also deserve to be studied more closely in other languages.

Appendix: Abbreviations used in the glosses ablat = ablative acc = accusative adess = adessive allat = allative cl = clitic cond = conditional elat = elative gen = genitive illat = illative imperat = imperative imperf = imperfect = iness = inessive inf = infinitive instr = instructive nom = nominative part = partitive partic = participle pass = passive pl = plural px = possessive suffix sg = singular sup = superlative transl = translative

References

Diller, Anne-Marie (1991) The conceptual structure of linguistic action verbs in Bahasa Indonesia. Cognitive Linguistics 2-3: 225-245. Hakulinen, Auli & Karlsson, Fred (1979) Nykysuomen lauseoppia. [Syntax of modern Finnish]. Helsinki: SKS. Hakulinen, Lauri (1979) Suomen kielen rakenne ja kehitys [Structure and development of Finnish]. Neljäs, korjattu ja lisätty painos. Helsinki: Otava. Heinämäki, Orvokki (1994) Aspect as boundedness in Finnish. In Carl Bache et alia (eds.), Tense, Aspect and Action. Empirical and Theoretical Contributions to Language Typology, pp. 207-233. Berlin: Mouton de Gruyter. Huumo, Tuomas (2002) Mikä kielen mielestä on tuloksellista? Polysemian silta resultatiivisuudesta kvasiresultatiivisuuteen [What language sees as resultative? The bridge of polysemy from resultativity to quasiresultativy]. In Ilona Herlin et alia (eds.), Äidinkielen merkitykset [The meanings of ], pp. 534- 555. Helsinki: SKS. Karlsson, Fred (1987) . Second edition (translated by Andrew Chesterman). Porvoo: WSOY. Lakoff, George (1987) Women, Fire, and Dangerous Things. What Categories Reveal about the Mind? Chicago: The University of Chicago Press. AN EXERCISE IN COGNITIVE LEXICAL SEMANTICS 339

Langacker, Ronald W. (1987) Foundations of Cognitive Grammar I. Theoretical Prerequisites. Stanford: Stanford University Press. —— (1988) An Overview of Cognitive Grammar. In Brygida Rudzka-Ostyn (ed.). Topics in Cognitive Linguistics, pp. 3-48. Amsterdam: John Benjamins Publishing Company. —— (1990) Subjectification. Cognitive Linguistics 1: 1. 5-38. —— (1991): Foundations of Cognitive Grammar II: Descriptive Application. Stanford: Stanford University Press. —— (1993): Reference-point Constructions. Cognitive Linguistics 4: 1. 1-38. —— (1999): Grammar and Conceptualization. Cognitive Linguistics Research 14. Berlin: Mouton de Gruyter. Leino, Pentti (1993): Polysemia - kielen moniselitteisyys [Polysemy – The ambiguity of language]. Kieli 7. Helsinki: Helsingin yliopiston suomen kielen laitos. —— (1994a) Kognitiivinen esitys suomen sijajärjestelmästä [Cognitive description of the Finnish case system]. In Kieli ja sen kieliopit. Opetuksen suuntaviivoja [Language and its grammars. Guidelines for teaching], pp. 208-214. Helsinki: EDITA. —— (1994b) Morfeemien polysemia [The polysemy of morphemes]. In Pentti Leino & Tiina Onikki (eds.), Näkökulmia polysemiaan [Viewpoints to polysemy], pp. 11– 35. Helsinki: Helsingin yliopiston suomen kielen laitos. Leino, Pentti & Onikki, Tiina (1992) Comments on Cognitive Grammar. In Maria Vilkuna (ed.), SKY 1992, pp. 35-51. Helsinki: Suomen kielitieteellinen yhdistys. Miller, George A. & Johnson-Laird, Philip N. (1976) Language and Perception. Cambridge: Cambridge University Press. The New Oxford Dictionary of English Language (1998) Oxford: Clarendon Press. Rudzka-Ostyn, Brygida (1988) Semantic Extensions into the Domain of Verbal Communication. In Brygida Rudzka-Ostyn (ed.), Topics in Cognitive Linguistics, pp. 507-553. Amsterdam: John Benjamins Publishing Company. Seuren, Pieter A. M. (2001) The Cognitive Dimension in Language Study. In Folia Linguistica. Acta Societatis Linguisticae Europaeae. XXXV/3-4: 209-242. Sivonen, Jari 2005: Mutkia matkassa. Nykysuomen epäsuoraa reittiä ilmaisevien verbien kognitiivista semantiikkaa. [Slight hitches. Cognitive semantics of verbs denoting Indirect Path in Modern Finnish]. Helsinki: SKS. SKS(T) = Suomalaisen Kirjallisuuden Seura(n toimituksia) [Publications of The Finnish Literature Society] SSA = Suomen sanojen alkuperä Suomen sanojen alkuperä (1992-2000) [The Origin of Finnish Words]. Etymologinen sanakirja I-III. Helsinki: Kotimaisten kielten tutkimuskeskus. Svorou, Soteria (1994) The Grammar of Space. Typological Studies in Language 25. Amsterdam/Philadelphia: John Benjamins Publishing Company. Talmy, Leonard (1975) Semantics and Syntax of Motion. In John P. Kimball (ed.), Syntax and Semantics 4, pp. 181-238. New York: Academic Press. —— (1985) Lexicalization patterns: semantic structure in lexical forms. In Timothy Shopen (ed.), Language typology and syntactic description III. Grammatical categories and the lexicon, pp. 57-149. Cambridge: Cambridge University Press. 340 JARI SIVONEN

—— (2000): Toward a Cognitive Semantics. Volume I: Concept Structuring Systems. Cambridge: The MIT Press. TYSYKLJ = Turun yliopiston suomalaisen ja yleisen kielitieteen laitoksen julkaisuja [Publications of the Department of Finnish and General Linguistics of the University of Turku] Vilkuna, Maria (1992) Referenssi ja määräisyys suomenkielisten tekstien tulkinnassa [Reference and in the interpretation of Finnish texts]. Helsinki: SKS. Yli-Vakkuri, Valma (1986) Suomen kieliopillisten muotojen toissijainen käyttö [Secondary usage of grammatical forms in Finnish]. Turku: TYSYKLJ.

Contact information:

Jari Sivonen Department of Finnish, Information Studies and Logopedics P.O. Box 1000 90014 University of Oulu Finland e-mail: [email protected]

Remarks and Replies

Fred Karlsson

Nature and Methodology of Grammar Writing

1. Introduction

Esa Itkonen (2003) embarks on no lesser a task than to construct a philosophy of linguistics. This has been a major concern of his for decades (e.g. Itkonen 1972, 1974, 1978). Now, his intention is to “explicate the concept language, understood as the logical prerequisite for studying various aspects of particular languages” (2003: 13; emphasis original). His central claim is that grammars composed in all cultures and all historical periods are remarkably uniform and that “this is the fact” (ibid.; emphasis original) that philosophers of linguistics must explain before doing anything else. In developing his argument, Itkonen voices strong views on many other important topics such as ontology (What levels should one postulate in grammar description? What kinds of entities populate those levels?) and methodology (What are the proper practices for the grammarian to acquire knowledge about grammatical constructions?). My purpose is to demonstrate that Itkonen’s argumentation is unclear, inconsistent, and even false. The main reasons are that he confuses grammatical theories and descriptions of individual languages, that he misrepresents the manifold nature and methodology of grammar writing (there are different types of grammars, with different objectives and different principles and methods of composition), and, especially, that he fails to appreciate the important function of corpus observation.

SKY Journal of Linguistics 18 (2005), 341–356 342 FRED KARLSSON

2. Grammatical theories and grammars of individual languages

As typical grammars Itkonen (2003: 13) lists ten descriptive grammars of ‘exotic’ languages such as Diyari, Hua, Rapanui, Wari’, West Greenlandic, Yagua, and Yoruba. He then defines his agenda in relation to these grammars:

(1) “If the first task of philosophy of linguistics is to explain the uniform nature of grammars, the second — and related — task is to explain the activity of the grammarian. What does he do in fact (as opposed to what he should do according to this or that pre-conceived philosophy of science)? And why is it possible for him to do what he does in fact?” (Itkonen 2003: 14; emphases original)

This statement is clear enough in its context. My reading of it is that Itkonen promises an in-depth analysis of the methodology of descriptive grammar writing and the practices of descriptive grammarians, perhaps including also comprehensive reference grammars and even mundane traditional school grammars. But, surprisingly, on the 200 pages after the initial mention Itkonen never returns to those grammars of Diyari, Hua, Rapanui etc., the very methodological uniformity of which he set out to explain. Rather, he climbs one step up the ladder of abstraction and starts analyzing the methodological stance of certain grammatical theories, especially generative-transformational grammar. Of course, ‘theoretical grammar’ is a valid notion that is more or less coextensive with ‘grammatical theory’ but the clarity and cogency of Itkonen’s arguments suffer badly from this lumping together of descriptive grammars and theoretical grammar.1 Note, by the way, as for (1), that it is not natural to claim that Chomsky would be a grammarian — he is a theoretical linguist, or perhaps a theoretical grammarian. An archetypal grammarian would be Curme (e.g. 1931). ‘Ordinary working grammarians’ write grammars of individual languages, they do not design grammatical theories.

1 In his earliest (1972) publication on these matters Itkonen claimed without qualification that (all of) synchronic linguistics is non-empirical (from the classical positivistic point of view). In (1974, 1978) he exempted psycho- and sociolinguistics from the original claim. REMARKS AND REPLIES 343

There is in (1) a hint that Itkonen would see theoretical grammarians and writers of grammars for individual languages as performing on the same level, viz. when he asks which philosophy of science a grammarian entertains. A realistic view of the relationships between metatheory, linguistic theorizing (design of grammatical theories), and description of individual grammars is that there are various (partly overlapping and more or less explicit) grammatical theories (‘traditional school grammar’, principles and parameters grammar, and so on) which provide overall frameworks among which writers of grammars for individual languages pick their favorite theory — or perhaps even an eclectic combination of several theories. A grammar of an individual language, once written, is a model implementing the theory picked by the grammarian. Metatheory is a third level which as its proper objects has the grammatical theories (all commitments included) rather than individual grammars which are models of their language in the respective theory. An (‘ordinary working’) grammarian is not the first person that should be asked what philosophy of science he entertains because he is entitled to apply the concepts, methods etc. provided by the theory he picked. But it is all-important to ask this question of theoretical grammarians and, of course, ordinary grammar writers too benefit from metatheoretical insights. These are obscurities that the reader of Itkonen (2003) faces. I shall take seriously the questions posed in (1): What is grammar writing and by what methods do grammar writers go about their task? My answers are quite different from those of Itkonen which I argue are mistaken. In answering these questions one must keep in mind that there are different types of grammars. I shall consider the three common types already mentioned: (i) descriptive (field) grammars of little described or hitherto undescribed exotic languages, typically written by field grammarians who do not have a practical command of the language when they go about their task; (ii) voluminous reference grammars for languages with many speakers and established writing conventions, typically written by native speakers; and (iii) elementary grammars such as traditional school grammars, written either by native speakers or foreigners. 344 FRED KARLSSON

3. Why “atheoretical rule-sentences” are not needed in grammar writing

Itkonen (2003: 15, 20) makes a distinction between atheoretical rule- sentences and empirical hypotheses. Atheoretical rule-sentences such as (2) describe language norms (henceforth: L-norms) and constitute the input to grammatical description whereas empirical hypotheses like (3) describe (assumed) regularities:

(2) “In English, the definite article (i.e. the) precedes the noun (e.g. man).” (3) “All ravens are black.” (Itkonen 2003: 15)

According to Itkonen, the difference between (2) and (3) is that (3) is falsified by spatiotemporal occurrences (non-black ravens), whereas (2) “is not, and cannot be, falsified”, e.g. by the utterance of the sentence *Man the came in, because this sentence is incorrect (ungrammatical). Itkonen emphasizes that rule-sentences are not to be equated with grammar rules proper (henceforth G-rules). Itkonen does not state what the G-rule corresponding to rule-sentence (2) would be like, but a reasonable guess is something to the effect of (4):

(4) In English, the definite article (i.e. the) precedes the noun (e.g. man).

Itkonen (2003: 20) concedes that G-rules, in contradistinction to rule- sentences, are falsifiable. But rule-sentence (2) and G-rule (4) are in fact identical. How then can one be falsifiable but not the other? If we are to believe Itkonen, grammars are systematizations of rule- sentences which describe L-norms. As rule-sentences (but not e.g. real sentences) are assumed to be the basic data of grammatical description, one would expect Itkonen to provide a succinct statement of what the full conceptual apparatus and terminology of atheoretical rule-sentences is. Likewise, given the considerable complexity of natural language morphosyntax, one would expect several detailed examples of various types of grammar-related rule-sentences (henceforth abbreviated GRS), i.e. rule-sentences describing grammatical regularities that do not concern just the minutiae of individual lexemes but extend over classes of morphemes and structural configurations. But (2) is the only example of a GRS treated with more than passing mention in Itkonen’s collected works. For the sake REMARKS AND REPLIES 345 of completeness, I note that Itkonen (2003: 16–17; emphases original) does give two examples of rule-sentences of a more concrete type:

(5) “It is correct to say I confided in him and incorrect to say I confided him.” (6) “It is correct to say I am upset and incorrect to say You am upset.”

(5, 6) are indeed atheoretical. They are formulated in everyday language using common words, without any theoretical commitments. If Itkonen would systematically implement his claim that rule-sentences are atheoretical, GRS (2) should just like (5, 6) be spelt out using only common everyday words and the theoretical terms “definite”, “article”, and “noun” should be disposed of:

(7) It is correct to say the man and incorrect to say man the. (8) It is correct to say the aardvark and incorrect to say aardvark the. (Etc., for every noun in the language.)

But this is idle repetition. Native speakers of English surely acquire (the equivalent of) (4) as a generalization very early and do not have to memorize self-evident trivia like (7, 8) for every noun they encounter.2 There are two reasons why Itkonen wants to call the rule-sentences atheoretical. First, they are supposed to describe the elementary correctness notions of ordinary linguistically naive speakers who are unfamiliar with the concepts of grammars and grammatical theories. Second, as rule- sentences are supposed to be distinct from G-rules, they must not contain theoretical terms. But GRS (2) does not comply with this criterion, as testified by the theoretical concepts “definite”, “article”, and “noun”. Itkonen commits the methodological fallacy of inconsistency when claiming that rule-sentences are atheoretical while simultaneously allowing (2) to contain theoretical terms. Furthermore, Itkonen leaves his philosophy inexplicit when not elaborating what concepts are in use for describing GRS’s and how these concepts relate to G-rules proper. The suspicion arises that the conceptual apparatus needed for GRS’s is coextensive with whatever concepts he would need for G-rules. Furthermore, note that Itkonen’s grammatical

2 Itkonen (2003: 114) remarks in passing that linguistic knowledge is only derivatively about particular sentences. This might suggest that he would not subscribe to the construal of (1) as (6, 7, ..., n). But how then should (1) be made truly atheoretical? 346 FRED KARLSSON ontology contains two levels of description where other frameworks normally fare with one. Itkonen’s G-rules “systematize” rule-sentences which in turn “describe” L-norms, as opposed to mainstream grammar ontology where G-rules are descriptions of sentences (utterances), understood as manifestations of L-norms. “Entia non sunt multiplicanda praeter necessitatem”: by Occam’s razor the level of rule-sentences must be eliminated. GRS’s are not needed in the first place because all descriptive work needed is done by properly conceived G-rules. Grammarians do not need one theory for G-rules, and either a separate replica of it or another theory for rule-sentences. Note that this sobering step of elimination does not imply rejecting the important function of correctness notions, rightly emphasized by Itkonen. Typical grammars of any of the three types under consideration do indeed describe correct sentences such as The man snored to the exclusion of incorrect sentences such as *Man the snored. In grammatical description it is enough to let the correctness notions of the grammarian function as a corrective filter on the data. If a clearly abnormal (= ungrammatical) sentence (e.g. *Man the snored) would come under consideration for relevance to (or inclusion in) the grammar being written (e.g., because this strange sentence occurred in a corpus), of course the grammarian is entitled to use his correctness notions (grammatical intuitions) to exclude such material. But this is a self-evident principle in the data formation of any empirical branch of study. Abnormal data are not considered and they do not require postulation of a new descriptive level in the overall theory, as we shall see next.

4. Normative filtering in data-formation: grammar writing goes with ornithology

Contrary to what Itkonen claims, normative filtering is indeed used in basic data-formation also in the sciences. As for the description of raven morphology and appearance (3), it is clear that ornithologists describe normal prototypical ravens, to the exclusion of abnormal individuals, just as grammarians describe normal prototypical (= grammatical) sentences (utterances). Thus, an albino raven would not count as evidence falsifying (3) for the very reason that it is abnormal. A competent ornithologist has practical experience and theoretical more or less normative knowledge of this which REMARKS AND REPLIES 347 he applies when deciding what raven individuals count as sufficiently typical and thereby as having potential bearing on the descriptive generalizations. Albino, three-legged, wingless, mashed etc. raven exemplars are excluded from primary consideration — on good grounds. Of course, the study of abnormal exemplars might be important in itself but it has little bearing on the construal of the prototype.3 Thus, Itkonen’s rendering of the research practice of ornithology is profoundly mistaken when he claims (2003: 15) that (3) “is falsified – in principle – by spatiotemporal occurrences, namely non-black ravens”. As always in the empirical sciences, observation is permeated with theory when the researcher decides on the relevant instances and this is true in both grammar writing and ornithology. Albino ravens on the ornithologist’s desk and ungrammatical sentences in the grammarian’s corpus have the same status. Grammar writing and ornithology are methodologically highly similar. (This is not meant to deny the existence of many obvious differences between grammar writing and ornithology, e.g. the fact that sentences are produced by rational goal-directed actions.)

5. Are “atheoretical rule-sentences” unfalsifiable?

Itkonen (2003: 15–29) insists that rule-sentences like (2) are unfalsifiable and claims to have proven this by adducing the string *Man the came in which is ungrammatical and therefore (rightly) dismissed. Itkonen maintains that we do not even know what a falsifying instance would be like, and asserts that rule-sentences are known to be true and therefore conceptually impossible to falsify. Let us next take rule-sentence (2) at face value and scrutinize it to find out whether these claims are true. Consider (9), a fully grammatical sentence (produced by myself in 1976):

(9) Look at that man the tall one over there.

Here the article the occurs after the noun man. There is only one conclusion: (9) falsifies (2), given the current formulation (2) has. (2) does not explicitly require that the noun in question (e.g. man) be the head of the NP to which it and the determiner jointly belong. In fact, (2) in its present

3 Mikael Fortelius, professor of evolutionary paleontology at the University of Helsinki, confirms (p.c.) that my rendering of the research practice of ornithologists and the role of normativity in it is accurate. 348 FRED KARLSSON form does not even explicitly require that N and DET belong to the same NP. This is the ultimate reason why we can squarely claim that (9) falsifies (2): (9) is correct and in it the determiner the follows the noun man. This is precisely what (2) is about. The correct sentence (9) has another devastating consequence. It falsifies the latter clause of (7), or more generally, the attempts to reformulate presumed atheoretical GRS’s on the level of individual words. In face of the new evidence (9), (7), if entertained, should be reformulated:

(7’) It is correct to say the man and correct to say man the.

But statements like (7’) are totally uninformative. It is obvious that this is not a fruitful track to pursue. Of course, the benevolent reader ‘understands’ that Itkonen has ‘meant’ that the relevant definite article should be a premodifier of its target noun and not a premodifier of the noun after the target noun, as in (9). But in scholarship all such details should be explicitly spelt out. An appropriate framework cannot rely on hidden assumptions concerning the benevolence of the reader. Thus, (2) as it stands is not an appropriate description of what speakers of English know about the co-occurrence of predeterminers and nouns in noun phrases. The upshot of our analysis is that nothing short of a detailed theoretically couched description of the English NP can do the proper job of spelling out exactly what the relations between predeterminers and head nouns are. But this is ordinary grammatical description of English, yielding G-rules formulated with defined theoretical concepts: head, modification, relative clause, embedding, constituent order, and so on. Itkonen’s problem is that he has no non-theoretical vocabulary at his disposal to talk about generalizations in atheoretical rule-sentences. He has no means of generalizing (2) so as to make it a full-blown atheoretical rule- sentence catering for all the relevant details of DET + N relationships in NPs. Only after having made this atheoretical rule-sentence explicit in its totality, or having postulated many more low-level rule-sentences without use of theoretical terms (but cf. the problems disclosed in connection with (7’)), could he embark upon the grammatical description of the phenomenon. Again, this seems to be an idle ceremony. But eliminating the separate level of atheoretical rule-sentences solves all these spurious problems. The descriptive grammarian has all the requisite concepts at her REMARKS AND REPLIES 349 immediate disposal in her favorite theory of grammar. She is free to directly apply these concepts to the description of the structure of (correct) sentences. The presumed atheoretical GRS’s thus constitute a pseudoproblem due to the lack of requisite vocabulary. There are no atheoretical terms for expressing generalizations on the level of abstraction needed. A famous dictum concerning conceptual dead-ends now comes to mind: “Wovon man nicht reden kann, darüber muß man schweigen”. The pseudoproblem vanishes once we put the (theoretical) G-rules to work. Their task is precisely to describe sentences conforming to L-norms. Here is an even more compelling piece of evidence than (9) of how falsifiable Itkonen’s GRS’s are:

(10) Third Man, The (http://www.imdb.com/title/tt0041959/, Nov. 14, 2004)

This is a (correct) spatiotemporal instance (from a movie database) which falsifies not only (2) but also such prospective improvements of (2) which still would state that DET never occurs after its N-head. That is, there are text types such as lists and registers where postmodifying DET occurs. Surely (10) is highly text-type specific and even marginal but it cannot be judged ungrammatical. A comprehensive description of the syntax of English determiners must take facts like (10) into account, however marginal and insignificant they may seem. Note that (10) carries more force than (9) because (10) has a genuinely postposed determiner still immediately dominated by its normal NP-head, whereas (9) concerns the relation between an NP-head and the first word (= determiner) of its appositive postmodifier. Thus, in contradistinction to Itkonen’s claims, (2) cannot be a conceptual truth because it has here been empirically falsified by invoking new (correct) observational evidence. The formulation of (2) could be improved upon in face of this adverse evidence but only if the requisite theoretical terms were available.

6. Is observation unnecessary in grammar writing?

The most erroneous claim by Itkonen (2003) is the following one (11), launched without any evidence e.g. in the form of references to real grammars (cf. 1) where the recipe (11) would have been followed: 350 FRED KARLSSON

(11) “As far as philosophy of linguistics is concerned, it is immediately evident that the (...) irrelevance of spatiotemporal evidence (...) demonstrates the non- empirical nature of grammatical descriptions. The irrelevance of space and time entails the irrelevance of the act of knowledge that applies to spatiotemporal occurrences, i.e. observation, and indicates the need for some other act of knowledge, namely intuition (...).” (Itkonen 2003: 23; original emphases)

So, Itkonen’s claim here is that observation is irrelevant in grammatical description. But what is the real practice of working grammarians (which Itkonen promised to analyze but never did)? First consider in-depth reference grammars. Here are some representative excerpts from the preface to Curme’s classical A Grammar of the English Language: Syntax (1931, 616 pages):

(12) “The purpose of this volume is to present a systematic and rather full outline of English syntax based upon actual usage. (...) The author owes much (...) to (...) the grammars of Jespersen, Poutsma, Kruisinga, Gustav Krüger, and Wendt (...). The author has learned much from the keen observations of these foreign scholars, who have sharp eyes for the peculiarities of our language. He has also made extensive use of the quotations gathered by them and the many other foreign workers in this field. In the same way he has availed himself of the materials gathered by English-speaking scholars. This book could not have been made without the aid of these great stores of fact. But to get a clear, independent view of present usage (...) the author found it necessary to read widely for himself, in older English and in the present period, in British literature and, especially, in American literature, which has not been studied so generally as it deserves. Almost the entire important literature of the early part of the modern English period has been read, in critical editions where such have appeared. (...) In the best literature of his own time the author has read so extensively that he feels that his findings have independent value. With his eyes constantly upon present usage, he has read a large number recent novels, dramas, lectures, orations, speeches, letters, essays, histories, scientific treatises, poems, etc., from all parts of the English-speaking territory. (...) In the novel and the drama, however, we find the irregular beat of changeful life, varying widely in different provinces and social strata, and, moreover, often disturbed by the exciting influences of pressing events, changing moods, and passionate feeling. (...) Loose colloquial English, as often described in this book, is frequently as appropriate as a loose-fitting garment in moments of relaxation. (...) In this book also the grammar of the common people is treated. (...) Diligent use has been made of every possible means to secure an accurate, reliable insight into existing conditions in all the different grades of English speech, both as to actual fixed usage of today and as to present tendencies. (...) Where British and American REMARKS AND REPLIES 351

English go different ways, both are described. (...) Present tendencies point to the possible ultimate loss of several valuable forms (...). is not a body of set, unchangeable rules (...) but to be constantly used and adapted to our needs.” (Curme 1931:v–xi; emphases added)

Curme eloquently explains why a grammarian simply must use the method of observation. No single person can master all varieties of a language, in all their multidimensionality and heterogeneity: the wealth of text types, the different styles of spoken and written language, all regional varieties, the individual properties of all words, etc. Curme uses thousands of examples collected by personal observation, e.g. on p. 290 alone he has quotations from Milton, Winthrop, Tarkington, Lowell, Lee, Holmes, Marlowe, Hobbes, Bradford, Churchill, and Mather. Of course, grammarians also use the method of intuition. Through the ages they have (rightly) used their intuitions for the clear cases, i.e. for constructed example sentences which are so self-evidently correct that it would be an idle ceremony to require full empirical documentation by external observation of where and when a native speaker in natural context produced them. Curme has hundreds of such examples, e.g. on p. 355 He is writing; Columbus proved that the world is round; There he comes; He loves his mother tenderly; etc. for which the ‘sources’ (obviously Curme himself) are (rightly) lacking. A few more examples from recent reference grammars are in order to establish our (rather self-evident) conclusion that Itkonen is fundamentally mistaken when claiming that observation is irrelevant in grammar writing. Here are hands-on descriptions of the methodology used in Hakulinen et al.’s Iso suomen kielioppi (“Comprehensive Finnish Grammar”, 2004, 1697 pp.), Huddleston and Pullum’s The Cambridge Grammar of the English Language (2002, 1842 pp.), and Teleman, Hellberg and Andersson’s Svenska Akademiens grammatik (“The Swedish Academy Grammar”, 1999), the largest reference grammar so far written for a single language (four volumes, 2745 pp.):

(13) “Between the corpus and the grammar there are of course the grammarians with their intuitions. (...) The majority of the examples are from genuine sources or speech situations. The made-up examples are a minority; they are used especially when brief and simplified examples are needed. (...) The basic source has been the Parole text corpus with its amplifications, totaling some 36.2 million words (...). The basic corpus of spoken language is the conversation archives of the Finnish Department at the University of Helsinki (...) which contains face-to-face and 352 FRED KARLSSON

telephone conversations. (...) during the final stages we also made corpus searches on the Internet.” (Hakulinen & al. 2004: 20, 30; my translation)

(14) “(...) we make frequent use of genuinely attested examples (often shortened or otherwise modified in ways not relevant to the point at issue), (...) it is significantly easier to obtain access to suitable large collections, or corpora, of written data in a conveniently archived and readily searchable form than it is for speech.” (Huddleston and Pullum 2002: 12)

(15) “The examples are largely authentic and have been drawn i.a. from the collection of machine-readable texts established at the Language Bank at the Department of Linguistic Computing, presently belonging to the Department of Swedish at the University of Gothenburg. (...) For describing the grammar of spoken Swedish the project has benefitted from the Speech Bank at the Department of Nordic Languages, Lund University. (...) When the grammar of a language is being investigated, the keys to understanding language competence and the various processes of language use are provided above all by spoken utterances and written texts, along with the correctness intuitions and meaning interpretations of grammarians and language users.” (Teleman, Hellberg and Andersson 1999: 6, 16; my translation)

That is: first and foremost corpus observation, of course guided by correctness notions, but also supplanted with intuition-based construction of simplified clear cases. This methodology is at the heart of all large reference grammars I am aware of. As for more elementary grammars, their very nature predisposes them to contain basic structures exemplified by straightforward clear cases. It is natural that such grammarians use the method of intuition more frequently than authors of comprehensive reference grammars. A pertinent example is my own Finnish: An Essential Grammar (Karlsson 2004) where the vast majority of examples are constructed (but not all, cf. p.242). This book is an intermediate-level reference grammar for grown-ups learning a foreign language, on a grammatically much less advanced and detailed level than (12–15) which are intended mainly for native professionals. Introductory basic-level school grammars might even fare with constructed examples alone. A typical example on my shelf is Miettinen’s (1955) introductory English grammar for Finnish elementary schools. Competent basic-level grammarians do not need extensive corpus observation, but neither do authors of basic-level textbooks on zoology use microscopes nor dissection methods when presenting elementary facts (e.g. on ravens) for schoolchildren. REMARKS AND REPLIES 353

We now turn to the third type of grammars, descriptive grammars based on fieldwork, i.e. the very type of grammars the methodological practice of which Itkonen promised to explicate (1). Is observation really a totally unnecessary method in this enterprise? Surely not. A descriptive field grammar is developed through complex interaction between the linguist and her informants where observation and intuition both have their roles but naturalistic observations predominate. The results of our brief examination are unequivocal. Disregarding the simplest types of school grammars, every ambitious grammar is based on a mix of (i) observation and (ii) intuition-based construction of pertinent examples. Itkonen’s total rejection (11) of observation as a knowledge source in grammar writing profoundly misrepresents established methodology. For completeness, it must be noted that Itkonen (2003: 33–36) contains a brief chapter titled “The limits of certainty” where he says that certainty (about atheoretic rule-sentences) is confined to the ‘core area’ of language (which, however, he leaves undefined). But he concedes that data is always uncertain in connection with variation, e.g. frequencies of occurrence investigated in the context of geographical or social dialects or language change. The data known with certainty coincide with the clear cases. But having made this appropriate distinction, Itkonen fails to draw any methodological conclusion from it. Does he still stick to the wholesale rejection of observation in (11)? If so, what is the method for investigating variation because, as admitted by Itkonen himself, intuition will not do this job even if it is potent of handling the clear cases? Or is this simultaneous condemnation and need of observational methods just another instance of inconsistency in Itkonen’s , similar to his postulation of atheoretical rule-sentences containing theoretical concepts? If, on the other hand, Itkonen ‘really means’ (without stating it) that intuition is enough to handle the clear cases whereas observation (contrary to the claim (11)) is normally called upon in instances of less-than-clear variation, this would be an adequate rendering of what serious grammar writing is about. But this is no news, as witnessed by the theory, practice, and explicit statements of Curme and countless other grammarians (12–15). One more interpretation is at hand: Itkonen’s manifold dilemma is due to a confusion of linguistic theory with the writing of grammars for individual languages. This suspicion arises in view of the following citation: “There is a genuine chasm between what linguistic (i.e. 354 FRED KARLSSON grammatical) descriptive practice is and what it, in the name of some ‘empiricist’ philosophy of science, is thought to be” (Itkonen 2003: 26). The context here is such that Itkonen on the preceding pages has demonstrated that Chomsky’s (1957) example sentences are intuition- based, not collected by observation. But, as pointed out above, Chomsky is not a typical (average working) grammarian. He surely works on the superordinate level of linguistic theory. Chomsky (1957) was not concerned with writing a grammar of English and it is unreasonable to evaluate his 39 example sentences in Syntactic Structures from this viewpoint. Furthermore, whatever criticisms may be leveled at Chomsky’s methodology and metatheoretical self-conception, they have little bearing upon the methodology and self-conceptions of real grammarians.

7. Are inadequate grammars due to the“lack of attention” of grammarians?

In Itkonen’s (2003: 40) wiew, the inadequacies of grammatical descriptions as for lack of coverage are only due to ”lack of attention and/or insight” on the part of the grammarian who has not been sufficiently diligent and inquisitive in tapping his intuitions. But, as so eloquently demonstrated by Curme (12), this position is untenable. It is a simple fact born out by everyday observation that no individual can exhaustively know the details of a language across all dialects, registers, styles, diachronic developments etc. Mere declaration does not suffice to make such ignorance equivalent to “lack of attention and/or insight” which presupposes the existence of unconscious full perfect knowledge. Making intuitions all-important rather leads to another type of problem, the temptation to overstep the confines of the clear cases and to invent overly complex sentences for playing important roles in grammatical description and theoretical debates. The generative tradition abounds with such examples. To pick a classical one, Chomsky and Miller (1963: 286–7) claimed full grammaticality for (16):

(16) Anyone who feels that if so-many more students whom we haven’t actually admitted are sitting in on the course than ones we have that the room had to be changed, then probably auditors will have to be excluded, is likely to agree that the curriculum needs revision.

REMARKS AND REPLIES 355

Clearly, the purported grammaticality of (16) is more controversial than that of the (also invented sentence) Homeros fell asleep. There is considerable irony in the fact that Esa Itkonen lines up with Noam Chomsky as an equally strong believer in the overriding primacy of intuition-based methodology. But this is not how ambitious real grammars are written. Nor are all grammars in all cultures remarkably uniform.

References

Chomsky, Noam (1957) Syntactic Structures. The Hague: Mouton. Chomsky, Noam & Miller, George A. (1963) Introduction to the formal analysis of natural languages. In R. D. Luce, R. R Bush. & E. Galanter (eds.) Handbook of mathematical psychology. II, pp. 269–321. New York: John Wiley and Sons. Curme, George O. (1931) A Grammar of the English Language: Syntax. Boston etc.: D. C. Heath and Company. Hakulinen, Auli, Vilkuna, Maria, Korhonen, Riitta, Koivisto, Vesa, Heinonen, Tarja Riitta & Alho, Irja (2004) Iso suomen kielioppi. Helsinki: Suomalaisen Kirjallisuuden Seura. Huddleston, Rodney & Pullum, Geoffrey K. (2002) The Cambridge Grammar of the English Language. Cambridge: Cambridge University Press. Itkonen, Esa (1972) Onko synkroninen kielitiede luonteeltaan empiiristä? Virittäjä 76: 1–14. —— (1974) Linguistics and Metascience. Studia Philosophica Turkuensia II. Kokemäki: Risteen kirjapaino. —— (1976) Linguistics and Empiricalness: Answers to Criticisms. Helsinki: Publications of the Department of General Linguistics, University of Helsinki. 4. —— (1978) Grammatical Theory and Metascience: A Critical Investigation into the Methodological and Philosophical Foundations of ‘Autonomous’ Linguistics. Amsterdam: Benjamins. —— (2003) What is Language? A Study in the Philosophy of Linguistics. Turku: Turun yliopisto, Publications in General Linguistics. 8. Karlsson, Fred (2004 [1982]) Finnish: An Essential Grammar. London and New York: Routledge. Miettinen, Eino (1955) Englannin kielioppi. Yhdeksäs muuttamaton painos. Helsinki: Otava. Teleman, Ulf, Hellberg, Staffan & Andersson, Erik (1999) Svenska Akademiens grammatik. 1. Inledning. Register. Stockholm: Svenska Akademien. 356 FRED KARLSSON

Contact information:

Fred Karlsson Department of General Linguistics P.O. Box 9 FI-00014 University of Helsinki, Finland [email protected] www.ling.helsinki.fi/~fkarlsso Esa Itkonen

Concerning the Synthesis between Intuition-based Study of Norms and Observation-based Study of Corpora

Motto 1: “The central issue here concerns the role of normativity in linguistic data. I do not think that the importance of this concept has yet been grasped in current . As long as this continues to be the case, no adequate understanding of the metascientific status of linguistics can, in my opinion, be reached” (Itkonen 1978: vi).

Motto 2: “I state as a desideratum of any adequate methodology of empirical linguistics that it should provide causal models of linguistic behavior. One type of model will be described in what follows. Attempts at axiomatization are commendable, but nevertheless of secondary importance only” (Itkonen 1980a: 350).

Motto 3: “The relation between linguistic intuition and linguistic corpus is certainly central to the theory of linguistics. It is only the more surprising that this relation has never been represented in an explicit and self-consistent way. (...) In this paper I intend to present an explicit and self-consistent account of what is at issue here. I intend to show what, precisely, is the relation of sociolinguistics to grammar, or of the quantitative linguistic analysis to the qualitative one. The results are directly generalizable to other human or social sciences as well” (Itkonen 1977a: 239).

Motto 4: “La vérité, l’âpre vérité” (Danton, quoted as the motto of Stendhal’s Le rouge et le noir).

1. General background

Ferdinand de Saussure conceptualized any given language as consisting of two distinct aspects, namely langue and parole. Langue was interpreted as a social institution which, like any other institution, is a system of norms (or rules or conventions). Parole was taken to be the actual behavior which conforms to, and occasionally deviates from, the norms of the institution. By definition, norms cannot be observed but only intuited. Actual linguistic behavior can of

SKY Journal of Linguistics 18 (2005), 357–377 358 ESA ITKONEN course be observed, but since it is an instance of normative behavior, this type of observation is not just a simple matter of sense-impression, but must always contain an element of intuition. De Saussure’s overall conception has been widely accepted, with occasional terminological variation. For instance, Trubetzkoy’s (1958 [1939]) respective terms for langue and parole were Sprachgebilde and Sprechakt (cf. Itkonen 2003a: 149–150), while Hjelmslev (1963 [1943]) used the terms ‘system’ and ‘process’ (or ‘text’) (cf. Itkonen 1968: 456). It goes without saying that a language qua system of norms is something of an idealization:

Any realistic conception of grammar allows for linguistic variability, and is therefore susceptible of further development in the direction of sociology. Of course, such a conception forces one to give up some of the formal elegance which had been acquired by adapting linguistic descriptions to the model of formal languages. But this is as it should be because formal languages admit neither of variability nor of (spontaneous) change, whereas natural languages do. To be sure, it may be a practical necessity to resort to the use of artificially homogeneous forms of language, as long as modes of description which would be theoretically more justified are not available. In particular, it seems that an exact specification of several logical properties of natural languages requires, for the present, that variability be artificially eliminated. (Itkonen 1974: 318)

The actual descriptive practice of ordinary working grammarians shows beyond any doubt that the variation-eliminating idealization of a more or less homogeneous langue is indeed a practical necessity. This was clearly realized e.g. by Bloomfield (1933: 37):

In no respect are the activities of a group as rigidly standardized as in the forms of language. Large groups of people make up all their utterances out of the same stock of lexical forms and grammatical constructions. A linguistic observer therefore can describe the speech-habits of a community without resorting to statistics” (emphasis added; quoted in Itkonen 2003a: 37).

The correctness of Bloomfield’s claim can be — and has been — demonstrated quite concretely. In recent years, based on corresponding grammars, I have been both investigating and teaching the following ten non- European languages: Diyari, Hindi, Hua, Rapanui, Swahili, Tamil, Wari’, West Greenlandic, Yagua, Yoruba. This is how I summarize the lesson that I have drawn from my efforts:

With increasing urgency, it is being claimed nowadays that “everything in language is dynamic, emergent, and variable”. When I see or hear people making this claim, I REMARKS AND REPLIES 359

ask myself whether they have ever read the grammar of any language and, if so, how much they have understood of what they have read. If everything in language is variable, then the only valid type of linguistic description must be statistical in character. The fact is, however, that while grammars provide the primary way of describing languages, they in general use no statistics at all. This elementary truth is well confirmed also by the grammars utilized in this book. (To be sure, Payne & Payne 1990 counts as a partial exception.) The variation in actual (linguistic) behavior can only be described by statistics. If grammarians do not use statistics and, by implication, do not describe actual (linguistic) behavior, what is, then, the subject matter of their (non-statistical, categorical) description? It is the structure, or more generally, the system, of the language in question. It follows that structure is primarily existent, and only secondarily ‘emergent’. This point has also been forcefully argued by Givón (1995: 175–176), who emphasizes the importance of “taking structure seriously”. Structure/system is in turn identical with Saussure’s langue (with the qualification that some parts of the langue may be non-categorical, not in the sense of ‘statistical’, but in the sense of ‘gradual’ or ‘continuum-like’). Thus, in my opinion, the fashionable criticism of the langue – parole distinction is based on a misunderstanding. This misunderstanding has been made possible by a related one, namely the view that this distinction was a conceptual innovation. But, in fact, Saussure merely gave a systematic expression to a practice that had always existed (and will always exist). Every grammarian describes langue (and not parole, or actual linguistic behavior). This is true of Pānini, Tolkaappiyanaar, Sībawaihi, Apollonius Dyscolus, Varro, Thomas of Erfurt, Arnold & Lancelot, and so on (for extensive documentation and discussion, see Itkonen 1991 and 2000). It is also true of all the grammars utilized in this book. In a complete description of any language, there is room both for categorical description and for statistical description. For some 30 years, I have been exploring the precise relation between these two types of description. However, they are asymmetric in the sense that, as shown by the history of linguistics, there can be categorical description without statistical description, but not vice versa. (Itkonen 2005a: 3–4)

What is, then, the relation between intuition (about norms) and observation (of a corpus of utterances)? How is the synthesis between the two to be achieved? In my published work, Itkonen (1978) and (1983) are devoted to langue and parole, respectively. From among hundreds and hundreds of pages, it is not easy to make a succinct and illuminating selection. Perhaps the following excerpts from Itkonen (1980a), which was written already in 1977, will be helpful here:

In this context linguists’ capacity for self-contradiction seems almost unlimited (...) E.g. Lieb (1976: 198) and Wunderlich (1976b: 81) claim that a grammatical description must absolutely be based on a corpus; and yet they themselves not once 360 ESA ITKONEN

make use of a corpus in their own published work. (...) Unlike the linguists mentioned so far, Labov has actually investigated real corpora, and therefore he can recommend the use of observation at least without being guilty of any obvious self- contradiction. (...) More recently he has come to explicitly entertain the view that in Chomskyan ‘clear cases’, where the results of intuition, observation, and experimentation coincide, or can safely be assumed to coincide, intuition suffices all alone, and insistence on the use of observation/experimentation results from misunderstanding (Labov 1975: 7–14). This view is, as it were, extensionally identical with mine; yet I find it unsatisfactory because it fails to indicate the precise relation between intuition and observation, or what it means for the two to ‘coincide’. (...) So far, we have been dealing with a straightforward dichotomy between self- invented sample sentences and a set of actual utterances, i.e. a corpus. The study of the latter type of data is quantitative in the sense that it must take relative frequencies of (different variants exemplifying) grammatical categories into account; moreover, in experimental-psycholinguistic research the data may be quantifiable also in the sense of containing several degrees of correctness or acceptability. The former type of data is that investigated by Saussure, Bloomfield, Harris, Chomsky, and Montague, among others. It is non-quantitative, i.e. qualitative, in a twofold sense. First, it has nothing to do with relative frequencies. Second, the data is discrete (= categorical, two-valued): these are the ‘clear cases’, which are (known to be) definitely correct, and are contrasted with all other, less than clear cases. Consequently, observation is connected to the quantitative analysis of actual utterances whereas intuition is connected to the qualitative analysis of conceptual possibilities, i.e. either correct or less than correct sentences which may or may not be exemplified by actual utterances. (...) (p. 336–337; emphasis added).

[There follows on p. 338 a discussion of Ross-type ‘squishes’, i.e. gradient or continuum-like phenomena.]

Rules [or norms] are not spatiotemporal entities and therefore cannot be observed but only intuited. The notion of correctness is inseparable from the notion of rule. Consequently, when one is observing a correct utterance, one’s observation (of space and time) is in fact subordinated to one’s intuition (of rule). This is the general relation, in linguistics, between intuition and observation (cf. also Friedman 1975). It parallels the general relationship between a rule and the actions conforming or failing to conform to it: the former is a conceptual precondition of, or a priori vis-à-vis, each of the latter. (p. 339)

(…)

Moreover, our world happens to be such that, as even Labov is willing to admit, there are ‘clear cases’ about which we do possess a rather secure knowledge. The world might have been otherwise; but it is not. (p. 340)

(…) REMARKS AND REPLIES 361

[Labov] states explicitly that sociolinguistic data is not described as such but is, rather, processed in accordance with certain “universal editing rules”; after the editing, “the proportion of truly ungrammatical and ill-formed sentences falls to less than two percent” (Labov 1972: 203). Now, it is clear that Labov edits his original data and evaluates the edited data as either correct or as “truly ungrammatical and ill- formed” on the basis of his intuitive knowledge about the rules of language. He cannot be relying just on observation, because what he is doing is precisely to evaluate his observations as either correct or incorrect. (...) The preceding argument involves an apparent difficulty. The learning of rules starts from observation, but in the course of this learning process there occurs a ‘leap’ from observing actual occurrences to (intuitively) grasping the rule, which subsequently serves as a criterion for evaluating what is observed. Because the gap between factuality and normativity can be neither bridged nor eliminated, as we know from philosophy, one has to leap over it. (p. 341–342)

(…)

To think that the concept of ‘correct sentence’ emerges as a result of experimentation is to commit the fallacy analogous to thinking that the concept of ‘centimetre’ results from measuring the height of a person and from noting that he is, e.g., 185 centimeters tall” (p. 344).

Up to now, we have reached the conclusion that the (normative) subject matter of grammatical analysis is the conceptual precondition of the (spatiotemporal) subject matter of those subdisciplines that investigate actual linguistic behavior, namely , sociolinguistics, and diachronic linguistics. Absolute black-or-white dichotomies are impossible in language (cf. Itkonen 1978: 108–109).Therefore there must be some sort of mediation between (linguistic) normativity and (linguistic) spatiotemporality. This is how I envision it:

Linguistic change, or more generally linguistic variation, and extraordinary use of language are, then, the two cases where atheoretical linguistic knowledge is less than certain, or where the social control of such knowledge is less than absolute. The possibility of spontaneous change is a necessary precondition for the continuous functioning of language, and distinguishes natural language from such artificial normative systems as formal logic or the game of chess. Moreover, linguistic change represents the exact point at which linguistic normativity and linguistic spatiotemporality contact each other, or merge into each other. For my general conception of science, such a point is of absolutely crucial importance since it provides the natural link between the empirical sciences and conceptual analyses in the widest sense. (Itkonen 1978: 153; emphasis added). 362 ESA ITKONEN

Grammatical analysis of ‘clear cases’ is a non-empirical undertaking comparable to philosophical or logical analysis, and it needs to be complemented by empirical considerations insofar as linguistic variation and extraordinary or non-standard uses of language (both of which are subsumable under ‘less-than-clear cases’) are also taken into account. The other linguistic subdisciplines are straightforwardly empirical undertakings (with an ineliminable normative component, to be sure). What is the methodological status of empirical subdisciplines like sociolinguistics or psycholinguistics? As indicated in my second motto, their overriding goal is to construct causal models for linguistic behavior. Correlational models, based on Boudon (1974), have been proposed for sociolinguistics in Itkonen (1977b), (1980a: 349–363), (1983: 260–278), (2003b: Chap. XVI). At the same time, I have defined such crucial notions as statistical causation and statistical explanation. I have also explored the relation between statistical behavior and free will (see, in particular, Itkonen 1983: 92–95 and 2003b: 192–194). It is a sad fact that the mainstream sociolinguistics, in Finland and elsewhere, has never shown any interest for these theoretical questions. Both postulational (or ‘analytic’) models and synthetic models have been proposed for psycholinguistics in Itkonen (1983: 278–313). The notion of synthetic model has been borrowed, with modifications, from Diesing (1972). It is the greatest shortcoming of the philosophy of the socio- psychological sciences that it has been unable to grasp the existence of synthetic models with sufficient clarity. As far as diachronic linguistics is concerned, it is legitimate to speak of a ‘causal model’ only in the general sense of referring to the bipartite (‘causal’) mechanism consisting of (psychological) innovation and (social) acceptance. It has been described, e.g., in Itkonen (1982), (1983: 201–211), (1984), (2002a). Regardless of the more specific features that characterize causal description in sociolinguistics, psycholinguistics, and diachronic linguistics, it is the basic tenet of Itkonen (1983) that a common denominator can be discerned in all these empirical subdisciplines, as summarized in the notion of rational explanation. My overall conception of the methodology of linguistics has not changed since 1983. Nevertheless, three subsequent developments deserve to be mentioned. First, additional support has been given by the history of linguistics: the logically prior subdiscipline of grammatical analysis (or ‘autonomous linguistics’) has everywhere been the first to develop, as REMARKS AND REPLIES 363 predicted; and, as predicted, the history of grammatical analysis has been everywhere similar to the history of philosophy and logic, and dissimilar from the history of the natural sciences (cf. Itkonen 1991, 2000, 2001a). Second, although the methodology of typological linguistics was already discussed in Itkonen (1983: 211–219), it has become the object of sustained analysis only in Itkonen (2003a: Appendix 5, pp. 172–199) and (2005a: Chap. XI). Third, the role of analogy in the causation of linguistic behavior is focused upon in Itkonen (2005b). — My overall conception was summarized in my plenary talk at the section ‘Philosophy of Linguistics’ of the 11th International Congress of Logic, Methodology, and Philosophy of Science, published as Itkonen (2002b). A couple of years ago, urged by my friends Chris Sinha and Jordan Zlatev, I decided to write a book-length compendium of my methodological views, a plan that resulted in Itkonen (2003a). This is how I characterize the motivation for, and the nature of, this book:

In my earlier publications I have tried to give a rather comprehensive answer to the question ‘What is language?’ In the present book I give a more condensed answer to the same question. To put it bluntly, this book offers a rapid, intensive course designed to enable beginners to master the essentials of the philosophy of linguistics. If you understand what follows, then you can move on to read the hard stuff, like Itkonen (1978) and (1983), which, even if sold out, can be unearthed at any decent library. (2003a: 10)

2. Intuition vs. observation in grammar-writing

Karlsson (2005) — henceforth to be abbreviated as ‘K-2005’ — criticizes what he considers to be the defects of Itkonen (2003a). He starts by deploring the fact that I give no account of the ‘ten non-European languages’, not realizing that these constitute the subject matter of an entirely different book, i.e. Itkonen (2005a). Itkonen (2003a) cannot be criticized for not being Itkonen (2005a). According to K-2005, the “most erroneous claim” of Itkonen (2003a) is made in a passage from p. 23, where I claim that the “twofold irrelevance of spatio-temporal evidence, mentioned above, demonstrates the non-empirical nature of grammatical description. The irrelevance of space and time entails the irrelevance of the act of knowledge that applies to spatiotemporal 364 ESA ITKONEN occurrences, i.e. observation, and indicates the need for some other act of knowledge, namely intuition.” This claim is qualified on p. 34, as the answer to ‘standard objection xii- b)’:

Data is always uncertain in connection with variation: there is no reliable intuition about frequencies of occurrence, investigated in the context of geographical or social dialects or of linguistic change. The same applies to extraordinary use of language. (...) The data known with [intuitive] certainty coincides with so-called clear cases, and the data not known with certainty covers the domain of less-than-clear cases. The boundary between the two is necessarily vague; “but to deny a distinction because of its vagueness is, of course, a semantic naiveté of the first order.” (Pap 1958: 401, n. 18).

When we compare these quotations from Itkonen (2003a) with some of the quotations given earlier, we see that, over the years, my position on this issue has remained exactly the same: in grammatical description, intuition is enough always, and only, in the clear cases; elsewhere observation is needed. Hence, K-2005 is wrong to say that, in my opinion, observation is not needed at all. What, exactly, is the nature of the “twofold irrelevance of space and time” that is characteristic of the clear cases? It can be explicated as follows:

A (sentence) grammar has to account for (or ‘generate’) all and only correct sentences of a language; but on the one hand, there are an indefinite number of correct sentences which never have been or will be uttered (i.e. exemplifications of which have never occurred or will never occur in space and time) and which must nevertheless be accounted for by the grammar; and, on the other hand, there is an indefinite number of incorrect sentences which have been or will be uttered, i.e. which must not be accounted for by the grammar (in spite of the fact that corresponding utterances have occurred or will occur in space and time). Therefore space and time are irrelevant to the grammar, namely in the twofold sense that we have here the combinations ‘correct & not spatio-temporal’ and ‘incorrect & spatio- temporal’. (Notice what is not claimed here: it is not denied that language- acquisition is based on observation of speech, i.e. utterances; nor is it denied that other types of linguistic research may or must concentrate on spatio-temporal utterances.) (Itkonen 2003a: 22)

The distinctness of the concepts ‘correct sentence’ and ‘spatio-temporal utterance’ follows from the general truth that ‘ought’ can be neither reduced to nor derived from ‘is’ (op. cit., Chap. 24). This truth also entails that there has to be a ‘leap’ from observing utterances to (intuitively) understanding the REMARKS AND REPLIES 365 norms they exemplify (p. 39, 65–66). It is not my fault if K-2005 does not understand these basic truths. It has become clear up to now that, as far as the data of grammar-writing is concerned, my position can be summed up by the following two statements: “In the clear cases it is fully satisfactory from the methodological point of view to use sentences the correctness of which has been established on the basis of intuition” and “Attested examples are needed when the expressions get more and more unusual and complicated: it is in such cases that intuition is the most unreliable”. Interestingly, these statements are to be found in Hakulinen & Karlsson (1979: 25, 63). To repeat an expression from Itkonen (1980a: 337), their position is “extensionally identical” with mine. But the ‘intension’ (i.e. interpretation) is all wrong, for the following four reasons. First, the role of observation is hugely exaggerated at the cost of intuition, to the point of nearly complete self-misunderstanding. For instance, in Chapter 2 of Hakulinen & Karlsson (1979) there are 83 example sentences (including those that occur in the running text), all of which have been invented by the authors on the basis of their linguistic intuition. In Chapter 12, there are 369 example sentences, only one of which (given on p. 269) has perhaps been observed to occur. Hence, in light of Hakulinen & Karlsson (1979), observation is not needed in grammar-writing at all. For my part, I reject this conclusion. Second, because of the de facto over-reliance on intuition, categorical statements are made that corpus data could easily have shown to be false (cf. Pajunen 2001: 375-382). Third, and inversely to the previous point, when attested examples are used in fact, there is often no genuine reason for doing so. For instance, in the footnote on p. 77 the authors proudly proclaim that the following sentence has actually been attested: “Tämä kesäloma oli ensimmäiseni moneen vuoteen.” But any fluent speaker of Finnish knows perfectly well the correctness of this sentence without any need of resorting to some corpus. So what is going on here? We clearly are dealing with a phenomenon that I have called “inessential use of a corpus: the grammarian knows that a given construction, e.g. ‘x differs from y’ is correct, but he refuses to illustrate it by means of an invented example and scans written texts or recorded speech until he comes upon a suitable example, e.g. “it differs from Standard English”. (Such a fragment of a sentence can be found in Labov 1969: 715.) This kind of procedure is based on the mistaken idea that the methodology of natural science must at any cost be imported into grammar” (Itkonen 1977a: 239). 366 ESA ITKONEN

Let it be added that there is also an experimental counterpart to the inessential use of a corpus:

Greenbaum & Quirk (1970: 18) note that there is simply no point in setting up experiments where there is “no reason to believe that we would have much less than 100 per cent acceptance”. In such a case, experimentation is “a slightly absurd exercise, with the results a foregone conclusion” (Wason & Johnson-Laird 1972: 78). To me it seems clear enough that people who wish either to eliminate or to ‘justify’ intuition even in the clear cases are engaged precisely in such ‘absurd exercises’ (Itkonen 1980a: 344).

Fourth, when over-emphasis on observation is combined with an inessential use of corpora, the end result is that intuition comes to be seen as more or less irrelevant and negligible. From the philosophical point of view, this is a disaster. The normative-conceptual core of linguistics — of any type of linguistics — is forgotten, and linguistics is taken to be just one natural science among others. This conclusion is indeed drawn by K-2005. As much is evident from his claim that, from the methodological point of view, non-prototypical animals (e.g. albino ravens) are exactly on a par with incorrect sentences. Notice that there is no reason why we could not speak e.g. of non-prototypical galaxies, with the consequence that any methodological difference between linguistics and astronomy disappears. But this is wrong. As argued in Itkonen (1974: 172–183), (1978: 198–219), and (2003a: 49–64), linguistics is a typical human science insofar as it contains two levels of knowledge, norm, and action: there are norms at the level of research objects, i.e. norms of language, and there are norms at the level of research, i.e. norms of linguistics; moreover, the linguist must come to understand the first-mentioned norms and hence to identify him-/herself with the research objects in the sense of learning to act in the same way. In these crucial respects, any natural science is dissimilar. Those who argue that grammar-writing (not to speak of other linguistic subdisciplines) is just one type of natural science, must of course deny the two-level nature of linguistics. I call this position ‘positivism’. Itkonen (1974) was criticized by Dahl (1975 [1980]) from the positivist point of view. This criticism was answered in Itkonen (1976b: 41–46) and (1980b). Today K- 2005 follows in Dahl’s (1975 [1980]) footsteps. One cannot help wondering what makes people endorse positivism, given that it is so obviously false. The following answer has been suggested: “The impulse to use Darwin [as a model to be imitated] is just as misguided as was the impulse to use Newton. What is the nature of this impulse? It is, first, to REMARKS AND REPLIES 367 feel hopelessly inferior to the representatives of the ‘hard sciences’ and, second, to think that one can get rid of this harrowing feeling by blindly imitating one’s betters (or those whom one considers such), come what may” (Itkonen 2003a: 198). Taken in by his own ‘observationalist’ rhetoric, K-2005 is convinced that the value of a grammar depends on how many examples from reputable authors it contains. For K-2005, the apotheosis is reached when a grammarian brings it about to have on a single page — on a single page! — “quotations from Milton, Winthrop, Tarkington, Lowell, Lee, Holmes, Marlowe, Hobbes, Bradford, Churchill, and Mather”. Grammars that strive after this ideal are called “ambitious” by K-2005. This is an unfortunate choice of words. A moment’s reflection is enough to show that grammars of this type do not qualify as ambitious, but as tedious. It is only too common that when people run out of ideas, they start filling their pages with more and more examples, in the hope that mind-numbing accumulation and cataloguing of data would somehow be accepted as a surrogate for theoretical insight. As an advocate of tedious grammars, K-2005 is bound to endorse the current vogue in linguistics which can perhaps be summed up in the slogan “Data is everything, theory is nothing!” It is my duty to firmly state that the attitude manifested in this stultifying slogan is inimical to any even halfway adequate notion of general linguistics. We have here a dramatic reversal of Paavo Ravila’s slogan, enunciated in 1944, “No theory, no facts” (cf. Itkonen 2004: 320). At this point, it is good to repeat that there are many important theoretical questions that can be answered only by statistical analysis of large corpora. For instance, Pajunen (2001) contains a judicious combination of intuition- based qualitative analysis and corpus-based quantitative analysis, fully in the spirit of Itkonen (1977a) and (1980a), even if — as is generally the case in — no explicit causal models are given. Just like any other grammars, grammars of ‘exotic’ languages are primarily, and often exclusively, based on intuition, but this time it is — at least in the early stage — the intuition of one or more informants, rather than the intuition of the linguist, i.e. of the person who is to become the author of the grammar in question. In other words, grammars of ‘exotic’ languages tend to be based on elicitation. This process is characterized by Haiman (1980: xi) as follows:

Kamani Kutane, my coeval and neighbour in Sara Village, has been my major informant both at Lufa and, in correspondence, in Canberra and Winnipeg over seven years. (...) I will always remember Kamani for his thought experiments: given a 368 ESA ITKONEN

minimally contrasting pair of sentences, he would construct elaborate background stories which would be appropriate for only one of these sentences. Eventually I would understand one of these, and we could move on. It was by means of such continued thought experiments that he was able to make clear to me that most mysterious of all Hua forms, the in -gasi’ [discussed in Itkonen 2005a: 92– 93].

The fact that grammars of ‘exotic’ languages are based on intuition-cum- elicitation is of course fully compatible with texts being used for illustration: “Running texts are given, at the end of the chapter, only from five basically unwritten languages, i.e. Diyari, Hua, Rapanui, Wari’, and Yagua, to show that — contrary to a wide-spread misconception — structural complexity is not a characteristic feature of written language only” (Itkonen 2005a: 9). It is clear that elicitation by means of thought experiments, as described by Haiman, goes much beyond K-2005’s ideal of grammar-writing, i.e. passively observing what has once been uttered. As for my own notion of grammar-writing, it is not adopted from the generativist tradition (as K-2005 suggests), but from the typological- functional tradition, as applied in Itkonen (2001a),with more than 900 examples from some 40 languages, and more recently in Itkonen (2005a), with ten case studies of non-European languages. The rise and fall of generativism in Finland has been described in Itkonen (1999).

3. The ‘atheoretical vs. theoretical’ distinction

K-2005 finds fault with my notion of ‘atheoretical rule-sentence’ (as opposed to ‘theoretical rule of grammar’). The objections are often expressed in a rather cumbersome way, and therefore the reader may wish to consult Dahl’s (1975 [1980]) more transparent formulations, answered — to repeat — in Itkonen (1976b) and (1980b). What is at issue here is the distinction, adumbrated already by the ‘two- level’ character of linguistics, between basic statements (about primary data) and theoretical generalizations. These two types of linguistic entities are strongly dissimilar, in the following sense. From the scientific point of view, the ‘rules of grammar’ (as constituents of theoretical-linguistic descriptions) are interesting because they are supposed to reveal the non-obvious structure- cum-function of language. By contrast, rule-sentences possess absolutely no interest because of their utterly obvious or trivial nature. From the metascientific or philosophical point of view, rule-sentences possess an REMARKS AND REPLIES 369 enormous interest, because they reveal the ultimately normative nature of linguistic data and thus show that, contrary to an obstinate misconception, linguistics is not a natural science. By contrast, rules of grammar — being hypothetical in character, just like theoretical descriptions in each and every academic discipline, including philosophy and formal logic (cf. Itkonen 1978: Chaps. X–XI) — possess no special interest. Over the years, I have given many examples of rule-sentences, but in general I have preferred this one: ‘In English the definite article (i.e. the) precedes the noun (e.g. man)’. This preference of mine is so well known that, for instance, in Edinburgh, April 1987, Geoffrey Sampson humorously suggested that I may have founded a whole new discipline called ‘the-ology’. Now, in what way can rule-sentences reveal the normative nature of linguistic data? — by showing that, unlike universal statements of the natural-science variety, they cannot be falsified by single spatiotemporal occurrences. For instance, our rule-sentence cannot be falsified by an utterance like ‘Man the came in’, because man the is incorrect; and it cannot be falsified by an utterance like ‘The man came in’, because the man is correct. This is the basic argument for normativity. Upon hearing it, people feel a nearly irresistible urge to disagree. Over the last 30 years or so, I have heard many objections, in fact, exactly 12 in number, and, labelled as the ‘standard objections’, they are first formulated and then answered in Itkonen (2003a: 18–21, 32–35). For reasons of space, they cannot be discussed here. But before you believe to have found the perfect objection against the basic argument for normativity, I advise you to check whether your objection is among the 12 ‘standard objections’. — Let us now at least have some quotations from Itkonen (1978) to support and further elucidate what was said above:

The hypothesis ‘All pieces of metal expand when heated’ is falsified if we find a piece of metal that does not expand when heated. On the other hand, a sentence referring to a rule is not falsified simply because there occur (what looks like) counter-instances to it. Consider the sentence ‘In poker a full house beats a flush’. This sentence it not falsified by the fact that in one particular game a player with a flush takes in the pot even though someone else is holding a full house. Such a performance is incorrect, whereas the rule-sentence is about correct performances only. It is a remarkable fact that behavior violating a rule does not falsify the corresponding rule-sentence. The reason is that what one does has no direct relation (although it certainly has some relation) to what one ought to do; and a rule-sentence is precisely about what one ought to do. In other words, a rule-sentence is about possible (correct) actions which ought to be done, and not about factual actions, 370 ESA ITKONEN

whether correct or incorrect, which are done. (Of course, a rule would cease to exist — in any strong sense of ‘existence’ — if correct actions conforming to it were no longer done as a matter of fact; but this is a different question.) Now since counter- instances are simply irrelevant, we cannot even specify the circumstances in which our rule-sentence could be taken to be falsified. But this means that rule-sentences do not satisfy the most basic requirement imposed on empirical hypotheses and theories, according to which “criteria of refutation have to be laid down beforehand: it must be agreed which observable situations, if actually observed, mean that the theory is refuted” (Popper 1963: 38) (p. 157).

[There follows a demonstration that sentences describing rules of a game and those describing rules of a language behave methodologically in the same way.]

As an example of a rule of language, the rule about the English definite article is unnecessarily abstract or general. For my purposes, it is sufficient to divide this rule up into a set of lower-level rules which determine for each particular noun in English that, when the noun is correlated with the definite article, the article precedes, and does not follow, it. The corresponding rule-sentences are of the type ‘In English, the definite article precedes the word man’, ‘In English the definite article precedes the word woman’, etc. It is clear that such sentences refer to genuine rules, and not to particular spatiotemporal events, because they refer to word-types (or word- concepts), and not to any particular one from among those potentially infinitely many spatiotemporal word-tokens which may exemplify any given word-type. (p. 160– 161)

(…)

I am, however, not concerned here with establishing the boundary between atheoretical and theoretical (...) Accordingly I am concerned with establishing the existence of those absolutely clear cases that may be taken as the two extremes of the continuum leading from atheoretical to theoretical: At one end, we have rule- sentences like ‘In English the definite article precedes the word man’ or ‘In English the plural of boy is boys’; at the other end, we have grammatical hypotheses like the ‘subject raising transformation’ or the ‘A-over-A principle’ (cf. 9.5 below). There can be, and are, falsificatory counter-instances to such hypotheses, for the obvious reason that neither their scope nor their truth is known with certainty (cf. 5.3–4 above). (p. 161–162)

(…)

It is an essential requirement for rule-sentences that they be absolutely trivial: sentences about language whose truth or falsity is not known immediately and REMARKS AND REPLIES 371

beyond the possibility of doubt are ex definitione not rule-sentences. Therefore there can be no scientific interest in stating rule-sentences of a given language. However, there is a considerable metascientific interest in merely realizing that there are such sentences, in view of the fact that, because of their non-empirical nature, they flatly contradict the claim that linguistics (in the sense of ‘grammar’) is an empirical science. (...) When I direct attention to rules, I am not reintroducing any new entities. All that I offer is reinterpretation of some well-known facts. This can be made more precise as follows. Each science must have its own set of basic statements, i.e. statements dealing with the simplest aspect of that region of reality with which the science in question is concerned. The basic statements of natural science are about particular spatiotemporal occurrences (cf. above). What do the basic statements of grammar look like? Bach (1974: 61–63) and Leech (1974: 84–90) give examples of, respectively, morpho-syntactic and semantic basic statements. For instance:

‘The past tense of play is played. The past tense of sing is not singed but sang.’ (...) ‘I am an orphan’ is synonymous with ‘I am a child and have no father or mother’. (...)

Notice that those ‘basic statements of grammar, i.e. rule-sentences, which contain theoretical terms like [‘past tense’] or ‘synonymous’ are not immediately understandable to a layman. To this extent, then, they do not just express prescientific or atheoretical everyday knowledge. However, it is obvious that they could be reformulated in purely atheoretical terms. Besides, the meaning of those theoretical terms that occur in the rule-sentences here considered can be easily taught to everybody. (p. 166–168)

All these claims and arguments have been made many times in my various publications, and they are repeated also in Itkonen (2003a), sometimes verbatim. K-2005 finds (atheoretical) rule-sentences superfluous because they are allegedly identical with (theoretical) rules of grammar. As shown by the quotations just given, this is nonsense. In a rare moment of lucidity, to be sure, K-2005 admits that rule-sentences may serve the purpose of clarifying the notion of correctness (i.e. normativity). This is, exactly, their sole purpose. But then K-2005 takes this admission back. There is no correctness, there is nothing to clarify. Why? — because there is, after all, no difference between empirical hypotheses like ‘All ravens are black’ and rule-sentences. Why is there no such difference? — because, just like a rule-sentence cannot be falsified by incorrect utterances, this empirical hypothesis cannot be falsified by albino ravens, i.e. ravens that are non-prototypical. This is a monumental confusion. Of course the empirical hypothesis is falsified by albino (= non-black) ravens. Just look at the empirical hypothesis as it is formulated here before your very eyes. It is not about protypical ravens, it is 372 ESA ITKONEN about all ravens. It claims that all ravens, without exception, are black. Therefore if, and when, we find an entity which is a raven and yet is not black (for instance, an albino raven), then the hypothesis is falsified. If the hypothesis had been formulated differently, i.e. as ‘All prototypical ravens are black’, and if we have reason to consider albino ravens as non- prototypical, then — and only then — the hypothesis would not be falsified by albino ravens. — If you do not understand this argument on the first reading, I advise you to read it again.

4. Dealing with the ‘standard objections’

Among my 12 standard objections, there are an objection — i.e. standard objection v) — against the use of a theoretical term like ‘definite article’ and an objection — i.e. standard objection iii) — pointing out that the definite article may occasionally also follow the noun. This is how I first formulate and then answer these standard objections in Itkonen (2003a):

v) “Since the rule-sentence A [= ‘In English, the definite article (i.e. the) precedes the noun (e.g. man)’] contains theoretical terms/concepts like ‘definite article’ and ‘noun’, it is not analogous to a simple generalization like B [= ‘All ravens are black’], which is expressed in ordinary language.” This is true, as such, and this is why we illustrated the terms in question with the words the and man. These illustrations do make A analogous to B” (p. 19).

(…)

iii) “It is just false to say that the definite article (always) precedes the noun; just think of an expression like Ivan the Terrible.” This objection is disingenuous. The one who makes it understands perfectly well the meaning of A, but pretends that he does not. (“If I were a robot which reads the sentence A, I would not know how the sentence A is meant to be taken; maybe I am a robot; therefore the meaning of A is not clear.”) This person might want to falsify B by painting a raven white. (“The meaning of B is not clear, i.e. it was not specified in what sense ‘black’ should be taken, and this thing here is a non-black raven, is it not?”) (p. 18–19).

K-2005 repeats standard objections v) and iii), but without acknowledging that they are taken from me and that I have already answered them. This is bizarre behavior. To round off the picture, and to show that I have left no stone unturned, I next give the complete answer to standard objection iii): REMARKS AND REPLIES 373

“Having established the existence of clear cases qua counter-examples to any universalist pretensions of variationism, (...) I must add that the behavioral basis for the clear cases is not as unequivocal or ‘clear’ as that which it is the basis for. That is, clear cases are idealizations. Lyons (1977: 568–569) enumerates the three most important types of idealization at hand, namely ‘regularization’, ‘standardization’, and ‘decontextualization’. These may be clarified, respectively, as follows. First, obvious mistakes, due say to drunkenness, must be discarded. Second, idiosyncratic elements like inside jokes must be discarded. Third, sample sentences must be imagined to be uttered in a standard or neutral context. It must be emphasized that all these idealizations are so natural that they might almost be said to be made instinctively. They already underlay the grammatical theories of Ancient Greece and India. Perhaps more importantly, they have always underlain the practice of language teaching, and have thus proved their worth (cf. Itkonen 1978: 149). Therefore it would in my opinion be nonsensical to try to undo these idealizations, which does not of course mean that they should or could not be analyzed. The existence of the above-mentioned quasi-instinctive idealizations has certain interesting implications concerning how my atheoretical norm-sentences [= rule-sentences] (cf. 3.2) ought to be formulated. I have claimed that just as (the clear cases of) norms are known with certainty, so is the truth of the corresponding norm-sentences. To this it has been objected that a norm-sentence like ‘In English the definite article precedes the noun’ has several prima facie counter-examples, e.g. ‘Ivan the Terrible’ or ‘Look at that man, the fat one over there’. Until such cases are explicitly accounted for, the norm- sentence can presumably neither be nor be known to be true. But accounting for them amounts to nothing less than giving an exhaustive description of the English determiners as they occur in all possible contexts. Such a task is in turn inseparable from writing a theoretical description of the entire English language. But no theoretical grammar can be known to be true. Therefore our norm-sentence cannot be known to be true either. — This is an interesting argument, but it must contain a fallacy somewhere, because its conclusion is self-evidently false. The fallacy consists in ignoring the workings of our quasi-instinctive idealizations. When I utter my norm-sentence to someone who knows English (as well as the rudiments of the linguistic terminology), he immediately understands which norm I am referring to, and he knows the sentence to be true. A computer could not perform the same feat, because it could not grasp the norm until it had been told all the prima facie exceptions to it. A normal speaker of English, by contrast, grasps the norm on the strength of his capacity to idealize, part of which is the capacity to disregard the prima facie exceptions. It is this aspect of the clear cases which Lyons captures with his previously quoted, apparently circular statement that he intends to concentrate on utterances whose pretheoretical status with respect to correctness can reasonably be assumed to be determinable. When formulating our norm-sentences, we must have some minimal trust in the intellectual capacities of our audience. In general our trust is justified. Therefore I could just as well have formulated my norm-sentence as ‘the man is right, and man the is wrong’, and could have let the audience infer, or rather recognize, the intended norm (cf. Itkonen 1978: 325–326, n. 90, and 167–168). The only members of my audience who would betray my trust are professional linguists. 374 ESA ITKONEN

Urged by their professional interest, and therefore acting in a computer-like fashion, they would pretend not to understand the norm which they do understand in fact, and would ask to be explicitly told all the prima facie exceptions. Although I intend to give an atheoretical (or pretheoretical) description, and they think they are asking me to give what I intend to give, they would in fact be asking me to give a theoretical description. This is the fallacy in their argument. (Itkonen 1983: 262– 264)

At the outset of this paper I said that langue, qua idealization, is a practical necessity. But to the extent that it turns out to be one of our ‘instinctive idealizations’ of the type discussed in the preceding quotation, it is not just a practical, but also a conceptual necessity:

Consider an analogy: Is the painting of Mona Lisa ‘nothing but’ patches of paint (= variation, parole), or is it the picture of a woman (= discreteness, langue)? Of course it is the latter. Certainly there is variation, but to some extent we as speakers do not experience it; we might be said to be so constrained as to ignore it. It would be a big mistake to disregard this fact, or the way that speakers in reality conceptualize language. The distinction between a conglomerate of patches of paint and a painting is analogous to the distinction between and phonology. Is everything just phonetics? Is there no phonology? Or should phonology be discovered again? (Itkonen 2003a: 35)

K-2005 finds it inconsistent that I pay an equal amount of attention to intuition-based non-causal analysis, as evidenced by Itkonen (1978), and to observation-based causal analysis, as evidenced by Itkonen (1983). The fallacy that K-2005 is here guilty of may be clarified with the aid of the following analogy. A building is a whole which has such parts as the walls, the doors, the windows, and the roof. It is not inconsistent to speak both about the walls and about the roof. Anybody should understand this, because the example is familiar and simple. Now, linguistics is a whole which has such parts as grammatical theory, sociolinguistics, psycholinguistics, and diachronic linguistics. Why does K-2005 not understand that it is not inconsistent to speak both about grammatical theory and about sociolinguistics, for instance? The reason must be that, for K-2005, this realm of phenomena is complex and unfamiliar. Data-collecting leaves one ill- prepared for the complexities of theoretical thinking. Is it possible to reconstruct the inference that has led K-2005 to this fallacious conclusion? I think so. Let A and B stand for grammatical theory and sociolinguistics, respectively. Now, the K-2005 type inference goes as follows: “Itkonen says that there is A and B; but B is not A, which means that REMARKS AND REPLIES 375

B is not-A; but to claim that there is A and not-A is the same thing as to assert the contradiction ‘p & ~p’; therefore Itkonen’s overall view of linguistics is inconsistent.” If I am not mistaken, this is K-2005’s lasting — and only — contribution to the methodology of linguistics: to show, with unique clarity, how not to think.

References

Bach, Emmon (1974) An introduction to transformational grammars. New York: Holt. Bloomfield, Leonard (1933) Language. New York: Holt. Boudon, Raymond (1974) The logic of sociological explanation. Penguin Books. Friedman, H.R. (1975) The ontic status of linguistic entities. Foundations of Language 13/1: 73–94. Dahl, Östen (1975) Is linguistics empirical? A critique of Esa Itkonen’s Linguistics and Metascience. Gothenburg Papers in Theoretical Linguistics. [Reprinted in Perry (ed.), pp. 133–145.] Diesing, Paul (1972) Patterns of discovery in the social sciences. London: Routledge. Givón, T. (1995) Functionalism and grammar. Amsterdam: Benjamins. Greenbaum, Sidney & Quirk, Randolph (1970) Elicitation experiments in English. London: Longman. Haiman, John (1980) Hua: A Papuan language of the Eastern Highlands of New Guinea. Amsterdam: Benjamins. Hakulinen, Auli & Karlsson, Fred (1979) Suomen kielen lauseoppia. Suomalaisen Kirjallisuuden Seura. Hjelmslev, Louis (1963 [1943]) Prolegomena to a theory of language. Madison: Wisconsin University Press. Itkonen, Esa (1968) Zur Charakterisierung der Glossematik. Neuphilologische Mitteilungen LXIX/3: 452–471. —— (1974) Linguistics and metascience. Studia Philosophica Turkuensia II. Risteen kirjapaino. —— (1976a) Was für eine Wissenschaft ist die Linguistik eigentlich? In Wunderlich (ed.), pp. 56–76. —— (1976b) Linguistics and empiricalness: Answers to criticisms. Publications of the Department of General Linguistics at the University of Helsinki, No. 4. —— (1977a) The relation between grammar and sociolinguistics. Forum Linguisticum I/3: 238–254. —— (1977b) Causal models in linguistics. Publications of the Linguistic Association of Finland: 25–35. —— (1978). Grammatical theory and metascience: A critical investigation into the philosophical and methodological foundations of ‘autonomous’ linguistics. Amsterdam: Benjamins. —— (1980a) Qualitative vs. quantitative analysis in linguistics. In Perry (ed.), pp. 334– 366. 376 ESA ITKONEN

—— (1980b) Reply to Dahl. In Perry (ed.), pp. 146–151. —— (1982) Short-term and long-term teleology in linguistic change. In Peter Maher et al. (eds.), Papers from the 3rd international conference on . Amsterdam: Benjamins, pp. 85–118. —— (1983) Causality in linguistic theory: A critical investigation into the philosophical and methodological foundations of ‘non-autonomous’ linguistics. London: Croom Helm. —— (1984) On the ‘rationalist’ conception of linguistic change. Diachronica I/2: 203– 216. —— (1991) Universal history of linguistics: India, China, Arabia, Europe. Amsterdam: Benjamins. —— (1999) Myytti Suomen kielitieteen ‘suuresta murroksesta’. Kielitieteen kääntöpuoli. University of Turku: Publications in General Linguistics 2, pp. 47–61. —— (2000) Tolkaappiyam: The basic work of Ancient Tamil language and culture. SKY Journal of Linguistics 13: 75–99. —— (2001a) The relation of non-Western traditions to linguistic typology. In Martin Haspelmath et al. (eds.), Language typology and language universals: An international handbook. Berlin: de Gruyter, pp. 275–279. —— (2001b) Maailman kielten erilaisuus ja samuus (2. uudistettu ja laajennettu painos). University of Turku: Publications in General Linguistics 4. —— (2002a) Grammaticalization as an analogue of hypothetico-deductive thinking. In Ilse Wischer & Gabriele Diewald (eds.), New reflections on grammaticalization. Amsterdam: Benjamins, pp. 413–422. —— (2002b) Linguistics as a system of distinct types of ontology-cum-methodology. In Peter Gärdenfors et al. (eds.), In the scope of logic, methodology, and philosophy of science. Volume two of the 11th international congress of logic, methodology, and philosophy of science, Cracow, August 1999. Dordrecht: Reidel, pp. 431–441. —— (2003a) What is language? A study in the philosophy of linguistics. University of Turku: Publications in General Linguistics 8. —— (2003b) Methods of formalization beside and inside both autonomous and non- autonomous linguistics. University of Turku: Publications in General Linguistics 6. —— (2004) Paavo Ravila yleisen kielitieteen edustajana. Journal de la Société Finno- Ougrienne 90: 319–329. —— (2005a) Ten non-European languages: An aid to the typologist. University of Turku: Publications in General Linguistics 9. —— (2005b) Analogy as structure and process: Approaches in linguistics, cognitive psychology, and philosophy of science. Amsterdam: Benjamins. Karlsson, Fred (2005). Nature and methodology of grammar writing. SKY Journal of Linguistics 18. Labov, William (1969) Contraction, deletion, and the inherent variability of the English copula. Language 45: 715–762. —— (1972) Sociolinguistic patterns. Philadelphia: University of Pennsylvania Press. —— (1975) What is a linguistic fact? Ghent: The Peter de Ridder Press. Leech,Geoffrey (1974) Semantics. Penguin Books. REMARKS AND REPLIES 377

Lieb, Hans-Heinrich (1976) Kommentare zu Kanngiesser, Ballmer und Itkonen. In Wunderlich (ed.), pp. 183–199. Lyons, John (1977) Semantics. London: Cambridge University Press. Pajunen, Anneli (2001) Argumenttirakenne: Asiaintilojen luokitus ja verbien käyttäytyminen suomen kielessä. Suomalaisen Kirjallisuuden Seura. Pap, Arthur (1958) Semantics and necessary truth: An inquiry into the foundations of analytic philosophy. New Haven: Yale University Press. Payne, Doris L. & Payne, Thomas E. (1990) Yagua. In D.C. Derbyshire & G.K. Pullum (eds.), Handbook of Amazonian languages, vol. 2. Berlin: Mouton de Gruyter. Perry, Thomas A. (ed.) (1980) Evidence and argumentation in linguistics. Berlin: de Gruyter. Popper, Karl (1963) Conjectures and refutations. London: Routledge. Trubetzkoy, N.S. (1958 [1939]) Grundzüge der Phonologie. Göttingen: Vandenhoeck & Ruprecht. Wason, P. & Johnson-Laird, P. (1972) Psychology of reasoning. Cambridge, MA: Harvard University Press. Wunderlich, Dieter (ed.) (1976a) Wissenschaftstheorie der Linguistik. Frankfurt a/M: Athenäum. —— (1976b) Kommentar zu Itkonen [1976a]. In Wunderlich (ed.), pp. 80–81.

Contact information:

Esa Itkonen Dept. of General Linguistics Henrikinkatu 4a FI-20014 University of Turku Finland e-mail: [email protected]

The debate will continue in SKY JoL 19 (2006).

The editors

Book Reviews

Aikhenvald, Alexandra (2004) Evidentiality (Oxford Linguistics). Oxford: Oxford University Press. Pp. xxvii, 452.

Reviewed by Heiko Narrog

1. Introduction

Evidentiality is a grammatical category arguably absent from English and many other European languages. In fact, according to the author of the volume reviewed here, although every language disposes of lexical and other means to convey the source of information, as a grammatical category it probably exists only in about a quarter of the languages of the world. In linguistics evidentiality has been recognized since Boas, but only recently it has come to the attention of a larger number of linguists. This can be attributed mostly to the collected volumes on the topic by Chafe and Nichols (1986), Johanson and Utas (2000) and Aikhenvald and Dixon (2003), as well as a special issue of the Journal of Pragmatics in 2001, and also a handful of frequently cited papers like the ones by Malone (1988) and Willett (1988). However, there can be little doubt that the present monograph by Aikhenvald marks a major advance in the study of evidentiality, both in terms of general recognition for the category, and in terms of knowledge and discussion about it. Of course, this is the first high profile monograph on the topic at all, but the scope and systematicity of Aikhenvald’s approach would benefit any topic, even well-studied ones. Aikhenvald works typologically and examined over 500 languages for grammatical evidentiality. She does so under a wide range of viewpoints, including semantic extension, interaction with other categories, grammaticalization, and use in discourse. In general, her results are convincing, but the information on evidentiality in individual languages appears not to be always equally accurate. Section 2 is devoted to a broad overview of her work. In section 3, I give an evaluation and point out some potential problems.

SKY Journal of Linguistics 18 (2005), 379–388 380 HEIKO NARROG

2. Summary of contents

The book consists of 12 chapters. 10 contents chapters are framed by an introduction and a summary. A definition of the category is given right in the beginning. “Evidentiality is a linguistic category whose primary meaning is source of information […] To be considered as an evidential, a morpheme has to have ‘source of information’ as its core meaning; that is, the unmarked, or default interpretation” (p.3). Aikhenvald is particularly intent on establishing evidentiality as a grammatical category in its own right, and preventing it from being included in other categories, especially modality, as has been done frequently before. For Aikhenvald, “marking data source and concomitant categories is ‘not a function of truth or falsity’. The truth value of an utterance is not affected by an evidential” (p. 4). Her approach to the investigation is rigorously inductive and empirical, in the spirit of Bloomfield (“The only useful generalisations about language are inductive generalisations”; p. 4), and based on Dixon’s Basic Linguistic Theory. Chapter 2 gives an overview over evidentials worldwide. Aikhenvald proposes to distinguish systems with two, three, four, and five or more choices of evidential marking. Those with only two choices are referred to as “small” systems. They are divided into five types, namely (1) firsthand and non-firsthand, (2) non-firsthand versus ‘everything else’, (3) reported (or ‘hearsay’) versus ‘everything else’, (4) sensory evidence and reported (or ‘hearsay’), and (5) auditory (acquired through hearing) versus ‘everything else’ (p. 25). The semantic domain covered by each of the two terms complement each other. According to Aikhenvald, types (1) to (3) are quite common worldwide while (4) and (5) are rather rare. In fact, (5) has been found only in a single language so far (Euchee). These small systems are particularly characteristic of evidential systems in Eurasia (e.g. Finno-Ugric or Caucasian languages). A map of distributions of evidential systems worldwide on p. 303 shows that Eurasia (including part of South Asia) and large chunks of the Americas are the only regions in the world with large contiguous areas of languages with evidential systems (isolated instances can be found in other regions such as Australia and New Guinea). Similarly to the two-term systems, systems with three or four choices are divided into types as well. When it comes to systems with five or more choices (e.g. Tuyuca and Central Pomo), each attested system seems to be unique. Aikhenvald makes a valid effort here to discover patterns and bring order into complex facts. For the reader, the borderline between different BOOK REVIEWS 381 systems often appears to be vague. Particularly, the less terms there are in a system, the broader and more difficult it becomes to delimit the terms. The author is aware of this problem and addresses it in a subsection devoted to “analytic difficulties” (2.1.2.). Chapter 3 deals with the encoding of evidentiality. Evidentials may be typically expressed inflectionally, with clitics, or with particles, but, as the author states, there are hardly any morphological limitations on their expression. Aikhenvald seems to demand, however, that in a single language the evidentials form a coherent morphosyntactic category in order to acknowledge the existence of an evidential “system” in a language. Some languages have evidentials which are “scattered” among various morphosyntactic classes (e.g. Japanese, West Greenlandic). Systems in these languages are seen by the author “somewhat problematic and thus only marginally relevant for the present study” (p. 81). A large part of the chapter deals with markedness in evidential systems, and, as a special problem, with evidentially neutral forms and omissions of evidentials. In chapter 4, evidential extensions of non-evidential categories are discussed. Quite often, a language does not have grammaticalized evidentiality, but one or more of its other grammatical categories may have evidential connotations. Thus, in many Caucasian languages, including Georgian, the perfect, entailing a meaning component of resultativity, has evidential connotations. Cross-linguistically, evidential extensions can be found also with passives, nominalizations, and complementations, among others. Also, all languages have some tools to convey reported speech, and the borderlines between reported speech and reported evidentiality may not always be clear. Chapter 5, discussing the meaning of evidentials, including also their extensions into other categories, approaches from the opposite direction, and chapter 6 follows up with extension into mirativity. Mirativity as a category of “speaker’s ‘unprepared mind’, unexpected new information, and concomitant surprise” (p. 195), is a category saliently related to evidentiality. Aikhenvald argues that despite the fact that in many languages mirativity exists only as an extension of evidential meaning, these are separate categories. The justification comes from languages which express mirativity as the extension of another category (e.g. Semelai; p. 210), or have independent expression of mirativity (e.g. Kham, p. 211). Chapter 7 brings up the topic of evidential and person. In brief, for many evidentials what matters is the distinction between first person and non-first person. Non-firsthand, non-visual, inferential, and reported evidentials often presuppose use with non-first person participants. The 382 HEIKO NARROG choice of an evidential that goes counter to this rule may trigger so-called ‘first-person effects’, including the implication of irony, surprise, or lack of intentionality or controllability. In some languages, like Eastern Pomo, evidentials may be a means of implicit person marking, since they presuppose the involvement of a certain person. (p. 235). Chapter 8 deals with the interaction of evidentiality with categories other than person. These include clause type, negation, modality, and tense and aspect. Most conspicuous are the interactions with clause type and tense. Usually, if a language has different evidential choices depending on clause type, it is declarative main clauses where the most choices are available, while in dependent clauses or imperatives the options are generally restricted. One may talk here of different subsystems of evidentiality, depending on clause type (e.g. p. 255). With respect to tense, many languages have more evidential choices available in the past than in other tenses, or, they may even have evidential marking only in the past. Chapter 9 discusses the origin of evidentials. Like most other grammatical categories, the most common source of evidentials is the grammaticalization of lexical categories, first and foremost verbs. Typical examples are verbs of speech which grammaticalize into reported or quotative markers. This often goes hand in hand with a reanalysis of the complement clause of such a verb as a main clause (p. 272). Evidentials can also be grammaticalized from deictic and locative markers or evidentiality strategies (chapter 4), among others. Chapter 10 is one of the most impressive chapters of the book, dealing with evidentiality in discourse and in the lexicon. The passages on how evidentials are linked to certain types of discourse, or speech registers, and can be manipulated by speakers for their rhetoric purposes are fascinating to read. With respect to the lexicon, there is salient interaction between evidentials and certain verb classes. One case that is mentioned throughout the book is that of verbs of internal states, including emotions, desires, physical states etc.. In many languages, one needs to use an “indirect” evidential, for instance an inferred evidential, when referring to the internal states of someone else than the speaker. The chapter ends with a flow chart that shows how “correct” evidentials are chosen, depending on a variety of factors including discourse and verb class (p. 331). Chapter 11, the last contents chapter, expands into further issues, many of which already emerged in the relationship between evidentials and discourse, namely the relationships between evidentials, cognition and culture. In the first place, this chapter brings home the point of how important evidentials are in BOOK REVIEWS 383 languages which have them as full-fledged systems, and how much speakers are aware of their importance. If speakers of a language with evidentials communicate in a language that doesn’t have them, they often try to make up for them with lexical means, which leads to awkward results. For them, lack of evidentials may mean lack of precision or even lack of truthfulness (p. 333-338). Chapter 12 provides a concise summary of the book. It is followed by a “fieldworker’s guide” with a catalogue of questions, which should be very welcome by any field linguists working on a language with evidentiality. In fact, as Aikhenvald states, “most languages with large evidentiality systems are spoken by small groups of indigenous peoples” (p. 334), and given the ongoing catastrophic decline of languages spoken by small communities world-wide, high-quality documentation of languages, including evidentiality systems, and by means of a solid descriptive framework like Dixon and Aikhenvald’s, is an urgent task.

3. Critical evaluation

As was already stated in the introduction, this is a highly important book which takes the field a step forward. Not only is this the first systematic monographic account of evidentiality as a category, the description is accomplished in admirable systematicity and comprehensiveness. From meaning range to formal expression to grammaticalization, one gets the feeling that the book provides the (preliminary) answer to every question that naturally emerges when dealing with this category. When it comes to accuracy in detail, however, this book may be somewhat controversial. I will mention here just a few points that I noticed and that I think may point to issues where closer investigation is desirable. A thread running throughout the book is that Aikhenvald’s data are most convincing, and her account is most compelling, when she draws on the data from languages of which she has first-hand knowledge. These are most importantly Tariana (see Aikhenvald 2003) and Estonian. There are also a number of other languages with which the author either appears to be well acquainted (e.g. Tucano, Quechua, varieties of Portuguese and Spanish in South America), or for which she can rely on data that she obtained directly from specialists on the language (e.g. Jarawara/Dixon or Qiang/LaPolla). In the case of many other languages, the data of which are taken from grammars or papers, the description seems to stand on more shaky feet. This is a methodological problem for language typology in 384 HEIKO NARROG general, and one with which anyone who has experience of working typologically can only sympathize. Perhaps one ought to simply admire the number of languages Aikhenvald manages to cover with first-hand or near first-hand data. On the other hand, I should say that I found the treatment of those languages I happen to have some first-hand knowledge of myself a little problematic, and this inevitably made me wonder about the reliability of data from all the other languages that do not belong to the author’s main sources. The first case in point is German. If it makes sense at all to claim that German has grammaticalized evidentiality (Aikhenvald believes that it doesn’t), the prime evidence would certainly be the evidential (‘reported’ or ‘quotative’) constructions of the modal verbs sollen and wollen. Sollen is brought up with the following example on p. 150:

(1) Er soll sich das Bein gebrochen haben ‘Apparently he has broken his leg’

The translation here is misguided. Instead of apparently, sollen should have been rendered with allegedly in English. Sollen + perfect is a construction that unambiguously marks the reported, or, as the common term in the literature goes, “quotative” (cf. Diewald 1999, Letnes 1997). Consider, in contrast, the explanation that is given by Aikhenvald (p. 150): “In German, the modal verb sollen ‘must’ may indicate that the speaker is reporting the information from someone else or that they inferred it or assumed it […] The English must may have similar meaning.” As stated above, the meaning in this construction is clearly reported and not inferred or assumed. English must, at least in the standard varieties, has no comparable use. In fact, if anything, the inferential use of English must corresponds to the inferential use of German müssen (cf. Mortelmans 2001), rather than sollen. Apart from the problematic interpretation of these data from a specific language, the question arises here where grammaticalized evidentiality begins. For Aikhenvald, the criterion is whether a “morpheme” has evidentiality as its primary meaning. In the case of German, the evidential meaning of wollen and sollen is certainly not primary for the morphemes. However, there are two constructions, namely sollen and wollen in present tense form and with perfect main verbs that are specialized on evidentiality. Taking into account that there is a growing recognition of the importance of constructions (versus morphemes “in isolation”) in linguistics, I think BOOK REVIEWS 385 that the existence of these specialized grammatical constructions is a good ground for claiming that German has grammaticalized evidentiality. On the other hand, it would certainly be an overstatement to assert that German has an evidential “system.” Few would doubt that Japanese has grammaticalized evidentiality. However, in the system presented in this book, it is one of the languages with “scattered” means of encoding modality. Therefore, in the author’s view, Japanese does not have evidentiality as a unitary grammatical category, and systems like that of Japanese are viewed as “only marginally relevant for [this] study of evidentiality.” Accordingly, the Japanese evidentials are labeled as mere “so-called” evidentials (Aikhenvald 2004: 81). Again, as Japanese is a fairly well-documented language like German, I feel that a closer investigation of the facts should have been possible (the author appears to base herself exclusively on Aoki 1986). This, then, might have led to a quite different, and arguably more accurate, interpretation. Japanese has four morphemes in the verbal complex which are widely recognized as being evidential. These are the nominal-adjectival suffix – soo for apparent imminence, the adjectival particle rasii for reported and inference, the particle soo for reported, and the (almost) synonymous noun yoo and nominal-adjectival particle mitai for inference based on appearance. Superficially, they are diverse indeed. But in terms of morphosyntactic combination, four of them, namely rasii, soo, yoo and mitai basically occupy the same structural slot in the verbal complex. Morphemes in the verbal complex, can, on the basis of their distribution, be classified into three classes, namely A inflections, B derivational suffixes, and C particles. These are combined according to the formula Verb+Br+A+(Cr+Br+A) r. (“r” signals recursivity). Everything except one inflection (A) is optional. Despite their different morphological makeup rasii, soo, yoo and mitai basically participate in slot C, and –soo in slot B. The combinations of morphemes are only partially restricted morphologically, but most simply by what makes sense. The particle soo can indeed appear further outside in recursive combination than other C morphemes, but not as the result of any morphological restrictions, or because it would occupy a different slot in a template, but simply to the extent that it makes sense. While all other evidentials in Japanese combine some kind of evidence with inference, soo is purely reported without any inferential connotations. It is understandable if evidentials which combine evidence with inference are labeled as “so-called” evidentials, although Aikhenvald recognizes inferentials as regular members of evidentials 386 HEIKO NARROG systems. I find it highly debatable from any perspective, however, if even a pure reported suffix is demoted to the status of a “so-called” evidential. From a broader point of view, the case of Japanese seems to raise the question of whether it is justified to exclude systems with markers that have divergent morphological properties as marginal. As shown here, they may appear to be mixed only from a specific theoretical viewpoint, or in a specific description. Also, no clear borderline is given between so-called “scattered” evidentiality (p. 80-82) on the one hand, and evidentiality subsystems (p. 82-87). In the viewpoint of the present book, it is apparently only the latter which merit closer investigation. Another point in which the author is perhaps not entirely successful is the discussion of the relationship between evidentiality and modality. Aikhenvald quite vigorously maintains that these are different categories, but I feel that the arguments presented in the book will not be sufficient to really lay this issue to rest. Apparently, the central argument is that “marking information source as a grammatical category does not imply any reference to validity or reliability of knowledge or information” (p. 5). If this turns out to be true, I find this a strong argument indeed, but as it stands now, it is simply a claim, and perhaps a book chapter of its own or a stand-alone paper would be necessary in order to back up this claim in more detail. Other arguments appear to be less convincing, or of purely rhetorical nature. Thus, for instance, it doesn’t really matter that evidentials have been grouped with epistemics from the point of view of European scholars who are not familiar with evidentiality from their own languages (p. 7). The only thing that counts is if this view is actually correct or not. The fact that evidentials may be compatible with irrealis or other mood and modality marking (e.g. p. 68) tells little as well. In many if not most languages, markers of different modalities are compatible with each other. Even in many Germanic languages deontic and epistemic modals may occur in the same clause (usually the epistemic modal takes the deontic one in its scope), and even epistemic markers in different slots may co-occur with each other. Subjunctive marking and modals may also co-occur frequently. Therefore, nothing emerges from this that would speak against regarding evidentiality as one type of modality on a par with epistemic modality, irrealis, or deontic modality. On the other hand, as the author remarks herself, in other languages, evidentiality and mood/modality are mutually exclusive or occupy the same slot in the morphological template (e.g. p. 13, 68). As I mentioned above, it may be questioned whether the author does not place too much weight on morphosyntactic criteria when BOOK REVIEWS 387 identifying grammatical evidentiality in languages. If morphosyntactic criteria are applied it should be done consistently, which might then, questionably, lead to the exclusion of markers which are in a paradigmatic relationship with epistemic or other modal markers. The typologically more profitable way, however, I suggest, is to view the category rigorously semantically, taking into account that morphosyntax is always the product of historical vagaries, and hardly ever consistent across languages, or even within one language. This would lead to a broader view of evidentiality in language. It deserves mentioning in this connection that recently a fruitful line of research of evidentiality in languages such as English, Dutch, German and Spanish has developed in functional and cognitive linguistics. The author apparently disapproves of the use of the term “evidentiality” in these languages, but with a few exceptions she cautiously avoids to comment on and (potentially) confront this type of research in any detail. Other theoretical issues that would have benefited from more discussion are the question of obligatoriness of the category, which is far from trivial if unmarked forms are also regarded as part of the system, and the distinction of “scattered” evidentiality and evidentiality sub-system, mentioned above. In general, it would be good to have a coherent picture about what is understood by “sub-system.” The book is generously equipped with indexes (language index, author index, subject index). One, however, is sorely missing, namely a list of those 500 languages which the author based her study on, and, most importantly, her (preliminary) judgment of which of these languages do have grammatical evidentiality, and if so, which system (chapter 2) they adhere to. The book would gain considerably in transparency if such a list were provided. Aikhenvald’s hypotheses would become more testable, and I sincerely hope that a second edition or a paperback edition will come with it. The prose of the volume is very dense and repetitious. Therefore it is suitable as a reference work rather than as a book to be read from the beginning to the end. In fact, the repetitions should be quite helpful if the book is used as a reference because the reader can be sure that an issue is mentioned at every point where it is relevant. The book is carefully edited and comes in a high-quality printing and binding. Overall, I can only repeat from the introduction that this is a book that represents a considerable advance in research on evidentiality. There is little doubt that it will become a reference on the topic for years to come. The author, on the other hand, has firmly established herself with this 388 HEIKO NARROG publication as the number one expert on evidentiality. Claims about particular languages or particular theoretical claims may turn out to be disputable, or may even be shown to be mistaken. In this respect, the book is reminiscent of genealogical-typological work of Greenberg. Painted in bold strokes, specific claims have been contested, often vigorously, and sometimes justifiedly, by specialists on particular languages or language areas. This doesn’t lessen the overall significance of the work. Aikhenvald has opened the floor for discussion, and everyone with an interest in this area can only appreciate this.

References

Aikhenvald, Alexandra Y. (2003) A Grammar of Tariana, from Northwest Amazonia. Cambridge: Cambridge University Press. Aikhenvald, Alexandra Y. and R.M.W. Dixon (eds.) (2003) Studies in Evidentiality. Amsterdam: Benjamins. Aoki, Haruo (1986) Evidentials in Japanese. In: Chafe, Wallace L. and Joanna Nichols (eds.), Evidentiality: The Linguistic Coding of Epistemology, pp. 223-238. Norwood, NJ: Ablex. Diewald, Gabriele (1999) Die Modalverben im Deutschen. Grammatikalisierung und Polyfunktionalität. Tübingen: Niemeyer. Johanson, Lars and Bo Utas (2000) Evidentials: Turkic, Iranian and Neighbouring Languages. Berlin: Mouton de Gruyter. Letnes, Ole (1997) Sollen als Indikator für Redewiedergabe. Debus, Friedhelm and Oddleif Leirbukt (eds.) Studien zu Deutsch als Fremdsprache III, 119-134. Hildesheim: Olms. Malone, Terrell (1988) The origin and development of Tuyuca evidentials. International Journal of American Linguistics 54/2, 119-140. Mortelmans, Tanja (2001) On the ‘evidential’ nature of the ‘epistemic’ use of the German modals müssen and sollen. Belgian Journal of Linguistics 14, 131-148. Willett, Thomas (1988) A cross-linguistic survey of the grammaticalization of evidentiality. Studies in Language 12/1, 51-97.

Contact information:

Heiko Narrog GSICS Tohoku University Aoba-ku, Kawauchi 41 Sendai 980-8576 JAPAN e-mail: [email protected] Etymologie, Entlehnungen und Entwicklungen. Festschrift für Jorma Koivulehto zum 70. Geburtstag. Herausgegeben von Irma Hyvärinen, Petri Kallio und Jarmo Korhonen. (Mémoires de la Société Néophilologique de Helsinki LXIII). Helsinki: Société Néophilologique. 2004. Pp. xvi, 489.

Reviewed by Marc Pierce

Jorma Koivulehto, professor emeritus of Germanic philology at the University of Helsinki, who is best-known for his work on loan relations between Indo-European and Uralic, especially between Germanic and Finnish (a number of his papers on the topic are reprinted in Koivulehto 1999), celebrated his seventieth birthday in 2004. The volume under consideration here is the Festschrift presented to him on that occasion. It contains 35 contributions by scholars from ten countries, all in English or German, and arranged in alphabetical order, as well as a brief biographical introduction and a list of Professor Koivulehto’s publications. As an exhaustive discussion of all the contributions is precluded by space limitations, I shall instead briefly outline the contents of the volume and then comment more extensively on some of the papers I found most stimulating. As one could guess, both from the scholarly interests of the Jubilar and the title of the Festschrift, a number of the papers deal with loan relationships (mainly Indo-European loans in Uralic). Albrecht Greule tackles the question of “Entlehnte ‘Wasserwörter’ in den ostseefinnischen Sprachen und die frühgermanische Hydronymie,” while Axel Holvoet looks at a Uralic construction borrowed into Indo-European in “Eine modale Konstruktion ostseefinnischer Herkunft im Lettischen.” More recent loan relationships are the subject of Wolfgang Mieder’s “Der frühe Vogel und die goldene Morgenstunde. Zu einer deutschen Sprichwortentlehnung aus dem Angloamerikanischen” and Norbert Richard Wolf’s “Anglizismen (bzw. Fremdwörter) im Sprachgebrauch. Eine Wortmeldung in einer immer wiederkehrenden Debatte.” Also, Santeri Palviainen examines “Die germanischen n-Stämme und ihre Reflexe in den ostseefinnischen Lehnwörtern.” Other contributions address etymological issues. Dagmar Neuendorff looks at “Etymologien in Rusticani und deutschen Predigten Bertholds von Regensburg. Struktur und Funktion.” Arend Quak examines the question of “Der altfriesische huslotha”, a type of tax, and a word attested since the tenth century AD. Peter Schrijver discusses “The of English

SKY Journal of Linguistics 18 (2005), 389–392 390 MARC PIERCE weapon, German Waffe, and the Indo-European root *Hwep,” while Patrick V. Stiles addresses “Place adverbs and the development of Proto-Germanic long *ê1 in early West Germanic.” Finally, Frederik Kortlandt looks at “Indo-Uralic consonant gradation,” and argues against the proposal of Koivulehto and Vennemann (1996) and Posti (1953) that consonant gradation in Finnic is the result of Germanic influence, in favor of the position that it originated in Proto-Uralic. A number of other topics are also addressed. In a historigraphical paper, Marja-Leena Piitulainen looks at the types of articles published in Neuphilologische Mitteilungen (the journal of the Société Néophilologique) between 1899 and 1901 and 1999 and 2001 in “Zu den Textsorten in den ‘Neuphilologischen Mitteilungen’ 1899-1901 und 1999-2001.” Kari Keinästö discusses Arthurian matters in “... er macht yn allen groß hercze. Über einige Herz-Phraseologismen im deutschen Prosa-Lancelot.” Outi Merisalo turns to more worldly issues in “Translations and politics in fifteenth-century Florence: Jacopo di Poggio Bracciolini and Domenico da Brisighella.” In my view, several contributions stand out above the rest. Raimo Anttila offers a paper with a title punning on the Jubilar’s name, entitled “Sap from the birch grove -- sulaa mahla koivulehdosta”, in which he suggests that Finnish sula ‘melted, unfrozen’, which is normally taken as an inherited Uralic word, could actually be an Indo-European loan word (there are parallel forms in Baltic and Indo-Iranian, like Sanskrit súrâ ‘kind of beer’ and Avestan hurâ ‘fermented mare’s milk’). Anttila thoroughly reviews the Uralic and Indo-European evidence, suggests that the phonological and semantic uniformity exhibited by the Uralic forms indicates that the word is a borrowing, and further argues that any rejection of the proposed loan relationship would be due to ‘semantic scruples’ (41). Joseph Salmons addresses a topic that has received increased attention in recent years, namely the problem of “How (non) Indo-European is the Germanic lexicon? ... And what does that mean?” While one often reads in Germanic handbooks that about one-third of the Germanic vocabulary is of non-Indo-European origin, other influential Germanic linguists have argued that the Germanic lexicon is more conservative in this area than such scholars would have us believe (cf. Mailhammer 2003 and Polomé 1989 for discussion of these differing viewpoints). Salmons offers a characteristically enlightening review of the problem, and ultimately suggests that the possibility of Indo-European-internal developments has not received enough attention, bolstering this claim with the support of a BOOK REVIEWS 391 number of ‘aberrant’ recent developments in colloquial American English, including front rounded vowels and prenasalization (both of which are admittedly confined to a small handful of cases). Michael Schulte’s contribution, “Zwei Miszellen zum Transfer und zur Adaption von Präfixstrukturen: Germanisch-Finnisch und Niederdeutsch-Norwegisch,” turns to the question of how prefixes are adapted into new languages. This is demonstrated through an examination of Germanic prefixes in Finnish and Low German prefixes in Norwegian. Schulte notes that a number of verbs loaned from Early Germanic into Finnish or Sami lack prefixes, presumably because of phonotactic considerations. In contrast to this, Norwegian seems happy to retain prefixes in borrowed verbs, although accentual patterns may be adjusted to fit Norwegian conditions. Finally, Theo Vennemann’s “Phol, Balder, and the Birth of Germanic” is another of Vennemann’s adventurous articles on the development of Germanic (many of which are reprinted in Vennemann 2003). Here Vennemann argues that Phol and Balder, the god names recorded in the beginning of the Second Meersburg Charm, are borrowings from Semitic, and are in fact a doublet, two manifestations of the same loan name (Semitic Ba’al and its later form Baldir). Vennemann further suggests that these borrowings took place ‘during a time of extensive Punic activities to the North’ (449). This claim has implications for the chronology of the Germanic Consonant Shift(s), as Phol must have been borrowed before the First Consonant Shift, but Balder after it; Phol also seems to show the effects of the Second Consonant Shift, although Vennemann attributes this to a scribe’s attempt to render a Low German original into High German (445). While I am not completely convinced by Vennemann’s arguments, his article is certainly thought-provoking and will hopefully spur further discussion of his theories. The volume itself is well and sturdily bound. Typographical errors are vanishingly rare, but there are a number of pages in my review copy where the printing is blurred. In sum, this book is a mixed bag, like all Festschriften, but the papers are generally of high quality, and some of them are absolutely first-rate. It is a worthy offering to a distinguished scholar.

392 MARC PIERCE

References

Koivulehto, Jorma (1999) Verba mutuata. Edited by Klaus Ph. Ruppel. Helsinki: Suomalais-Ugrilainen Seura. Koivulehto, Jorma and Theo Vennemann (1996) Der finnische Stufenwechsel und das Vernersche Gesetz. Beiträge zur Geschichte der deutschen Sprache und Literatur 118: 163-182. Mailhammer, Robert (2003) The strong verbs of Germanic: Morphology and etymological situation. Paper presented at the XVIth International Conference on Historical Linguistics, Copenhagen. Polomé, Edgar C. (1989) Substrate lexicon in Germanic. North-Western Language Evolution 14: 53-73. Posti, Lauri (1953) From Pre-Finnic to Late Proto-Finnic: Studies on the development of the consonant system. Finnisch-Ugrische Forschungen 31: 1-91. Vennemann, Theo (2003) Europa Vasconica -- Europa Semitica. Edited by Patrizia Noel Aziz Hanna. Berlin: de Gruyter.

Contact information:

Marc Pierce Department of Germanic Studies University of Texas at Austin 1 University Station C3300/ EPS 3.102 Austin, TX 78712-0304 e-mail: [email protected] Kaisa Häkkinen (2004) Nykysuomen etymologinen sanakirja. Juva: WSOY. Pp. 1633.

Reviewed by Ante Aikio

Etymological research has a long and fruitful history in Finland, and Finnish may in fact be the most thoroughly etymologically studied language outside the Indo-European language family. Several etymological dictionaries of Finnish have been published, most notably the multi-volume Suomen kielen etymologinen sanakirja (‘the etymological dictionary of Finnish’) (SKES; 1955–1981) and its successor Suomen sanojen alkuperä (‘the origin of Finnish words’) (SSA; 1992–2000). These works are primarily directed at the scholarly community, and until recently Kaisa Häkkinen’s brief Etymologinen sanakirja (1987) has been the only reference written with the non-expert reader in mind. The work under review, Nykysuomen etymologinen sanakirja (‘the etymological dictionary of modern Finnish’, henceforth NES) by the same author, is an updated and greatly expanded version of this earlier work, which makes it the third comprehensive etymological dictionary of Finnish. NES is a single hardcover volume of 1633 pages. Due to the size of the dictionary it would have been a good idea to publish it in two volumes, which would have made it both less clumsy to use and more resistant to wear. One expects a reference work of this kind to last in heavy use, but the binding of NES regrettably does not promise the book a long life; the back of the review copy got torn through wear already after a year of use. The word articles cover 1532 pages, in addition to which there is a brief introduction (p. 6–16), explanations of the most common special characters (p. 17), references (p. 1551–1570), an index of inventors of neologisms (p. 1571–1574), and an index of words discussed in other word articles but lacking one of their own (p. 1575–1633). On the inside of both the front and the back covers there is a map showing the distribution of the Finno-Ugric (Uralic) languages, a useful inclusion for readers unacquainted with the language family. Certain details of the map could have been improved, though: the speaking area of Inari Saami is missing, and the Saami and Samoyed languages have been indicated with single symbols despite of their deep divergence. On the first skim NES leaves a very favorable impression. It is pleasing to note that Häkkinen has disengaged herself from the Finno- Ugrianist jargonism characteristic of the earlier etymological dictionaries

SKY Journal of Linguistics 18 (2005), 393–405 394 ANTE AIKIO of Finnish. The word articles are written in a clear and informative style, without long word lists, sentence fragments or heavy use of abbreviations. This is a standard practice in good etymological dictionaries elsewhere in the world (e.g. Kluge 1995; Magnússon 1989), and it is praiseworthy that Häkkinen has brought the tradition to Finland.

1. The selection and citing of material

As the title of NES tells, the lexical material chosen in the dictionary represents the vocabulary of ‘modern Finnish’. The number of word articles in the dictionary is somewhat over 6000, which Häkkinen says to accord with Erkki Itkonen’s estimate of the number of underived word- stems in Finnish – a figure which does not include young loanwords and so-called ‘learned words’. Even so, the 6000 word articles include also new loans and internationalisms as well as derivatives and compounds, whereas the bulk of dialectal and obsolete vocabulary has been left out. The selection of material thus differs substantially from the previous etymological dictionaries. The inclusion of recent loans and internationalisms is an excellent choice, as they, too, form an integral part of the ‘modern Finnish’ lexicon. However, this good idea seems to have been carried out in a somewhat arbitrary way: for instance, abstrakti ‘abstract’, adressi ‘address, petition’ and aforsmi ‘aphorism’ are included, whereas absoluuttinen ‘absolute’, adoptio ‘adoption’, aerobic ‘aerobics’, agentti ‘agent’ and aggressio ‘aggression’ are not. It is not easy to see what kind of criteria have been applied here, and in any case a substantial part of this type of vocabulary is missing. Moreover, some entirely common words have slipped out as well, such as aavikko ‘desert’, ahtaa ‘to cram, stuff’ (this missing word article is even referred to on p. 27), katkarapu ‘shrimp’, kiihtyä ‘to accelerate; to get upset’, no ‘well (discourse particle)’, nääntyä ‘to starve’, sietää ‘to endure’, siika ‘whitefish’, siili ‘hedgehog’, siima ‘line (for fishing)’, siipi ‘wing’, vala ‘oath’ and väsyä ‘to get tired’. Also helmikuu ‘February’ and huhtikuu ‘April’ are missing, even though tammikuu ‘January’ and maaliskuu ‘March’ are included. Such shortcomings suggest that the book was perhaps too hastily edited. Some remarks can also be made on the citing of linguistic material. Häkkinen has made a few amendments to the standard Finno-Ugric phonological transcription, including the introduction of the grapheme <õ>, which stands both for a reduced vowel /é/ or /é@/ (e.g. in Mari and Khanty), BOOK REVIEWS 395 and a mid central unrounded vowel /e1/ (e.g. in Komi and Udmurt), as well as the replacement of /iº/ (a high central unrounded vowel) with the letter . While the idea of getting rid of unnecessary diacritics is in itself sensible, the justification given for these changes is somewhat curious: Häkkinen states that she has “used as simple an as possible, which guides also the reader unfamiliar with special characters to read the word approximately correctly” (p. 12).1 One can wonder whether this is a relevant or even a possible goal; and in any case õ is just another special character to the non-expert, who will no doubt also find it hard to associate y with a central unrounded vowel, as the same letter stands for a front rounded vowel in the Finnish orthography. Even so, Häkkinen’s changes to the standard transcription are small and cause no difficulty for specialists. It is more regrettable that there are occasional errors in the cited forms, a good example of which is provided by the Saami data. NES applies the standard spelling of North Saami instead of the complicated Nielsen’s orthography common in earlier etymological references, which would be a good choice, were it not that the forms include numerous translitteration mistakes. Examples include guoddu pro guottu ‘tree stump’ (p. 351), gaehpid pro geahppat ‘light (not heavy)’ (p. 400), leaggji pro leadji ‘heap’ (p. 652), njuoskas pro njuoskkas ‘wet’ (p. 768), čavgŋi pro čávgŋi ‘beam in a sod hut’ (p. 1132), sagŋat pro sakŋat ‘to thaw’ (p. 1223) and dárpmi pro dárbmi ‘energy, vigor’ (p. 1279); this list is far from exhaustive. Such mistakes are not confined to Saami – for example, Komi t'śeltny and Udmurt t'śeltyny ‘to throw, etc.’ (p. 1210) should read t'śõltny and t'śõltyny (= c`e1ltniº and c`e1ltiººniº in standard transcription), Hungarian hájt ‘to throw’ (p. 323) should read hajít, and the word turo ‘ski’ (p. 1203) is not Nenets but Enets instead.

2. The principles of etymology

The introduction of NES provides a lot of information useful to non-expert readers: Häkkinen briefly discusses the principles and criteria of etymological research, the development of the Finnish literary language, as well as the stratification of the Finnish lexicon and the most important sources of loanwords. Of particular interest here are the theoretical and methodological issues touched upon, as they also have a bearing on the content of the word articles. In Häkkinen’s view etymology operates with

1 This quote, as well as the ones that follow, are my translations from Finnish. 396 ANTE AIKIO three basic criteria: 1) the distribution, 2) the structure and phonological shape, and 3) the meaning of the word. As for the hierarchy of these criteria, she writes that “especially when evaluating the true age of the oldest words, the central criterion is the distribution. As a rule of thumb one may regard that the wider the word is attested in the cognate languages of Finnish, the older the word in question is.” (p. 7; cf. 1990: 24–67) Two remarks are in order here. First, distribution as such actually reveals little of the age of a word, because loanwords can also show extensive distributions. Instead, it is phonological criteria that are used to establish the age of words: old inherited words show regular sound correspondence between related languages whereas loanwords display different correspondences. Second, ‘distribution’ cannot really count as an etymological criterion because it is a research result instead: one cannot know the distribution of any word before it has been established through etymological study. Admittedly, this is partly acknowledged by Häkkinen as well, as she writes that “the criteria concerning phonological and morphological structure are needed when it is determined which words belong together etymologically...” (p. 7) and that “[the distribution] cannot be considered decisive if it seems to be at odds with the phonological criteria” (p. 8). Even so, the discussion on the application of phonological criteria is a bit confusing: Häkkinen states that one cannot reliably determine the age of loanwords on the basis of their phonological shape alone as unassimilated loans such as farao ‘pharaoh’, graniitti ‘granite’ and strutsi ‘ostrich’ already occurred in the earliest Finnish literary language (p. 9). Because etymology usually operates on a timescale of millennia rather than centuries, farao and the like count as young words regardless of whether they were present in “old” literary Finnish or not. What is more, the reader is also told that such words as kinkku ‘ham’, penkki ‘bench’, sielu ‘soul’ and ralli ‘rally’ may look like “autochthonous” words and could not be deduced as borrowings on the basis of their phonological shape alone (p. 15), even though it is textbook knowledge of Finnish historical phonology that the earlier language stages had no unstressed rounded vowels, geminate sonorants or three-consontant clusters (see e.g. Häkkinen 2002: 64–69). BOOK REVIEWS 397

3. Loanwords

As the bulk of Finnish vocabulary consists of loanwords, it is natural to deal with them first. Recent loanwords and internationalisms are treated well in NES, and a few word articles also provide welcome background information. For instance, the reader looking up bussi ‘bus’ is certainly interested in knowing that buses were introduced as a means of public transport in Helsinki in the 1920s, and prior to this bussi was used as a slang word for streetcars, which were originally horse-drawn. Occasionally, though, one would have hoped for a more exact wording. For instance, of iglu ‘igloo’ NES states that “originally the word derives from the Inuit or Eskimo language word igdlu.” Eskimo is not a single language, and Inuit is a subbranch of the Eskimo group, which also includes the Yupik languages spoken in Alaska and on the Chukchi Peninsula. In the various individual Eskimo languages this word appears in quite different forms such as iγlu, illu, ittiq, əŋluq, lu, etc. The exact source of iglu seems to be the form iγlu attested both in North Greenlandic and in certain Inuit languages of Alaska and Canada (Fortescue, Jacobson & Kaplan 1994: 112). Older loanwords from Germanic, Baltic, Aryan etc. are also generally treated well in NES. Often a reconstruction of the loan original is provided, as well as a few examples of its reflexes in the attested languages. However, in some cases the treatment could have been improved by presenting more information on the phonological development of the Finnic forms as well. Finno-Ugric reconstructions are routinely given in NES for old inherited words, but not in the case of old loans from Baltic and Germanic. For example, the connection between Finnish halla ‘frost’ and Lithuanian šalnà ‘frost; rime’ would no doubt seem more evident to the non-expert reader if he was told that the Finnish form has developed from Pre-Finnic *šalna via the regular changes *š > h and *ln > ll. In general loanwords receive a sound treatment in NES. The earlier dictionary SSA tends to be somewhat overcritical toward loan explanations in general, and it is pleasing to note that several convincing loan that SSA classifies as uncertain are accepted in NES; these include, for instance, the Germanic etymologies of Finn. nahka ‘skin, hide’, paasi ‘stone slab’, rasia ‘box’, runko ‘trunk’, tuppi ‘sheath’, uskoa ‘to believe’, vihjata ‘to hint’ and vihko ‘bundle, sheaf; notebook’. In a couple of cases, though, perhaps too cautious an attitude has been adopted in NES as well. For instance, there is hardly any reason to doubt that Finnish vyyhti 398 ANTE AIKIO

‘skein’ (~ Votic vehsi id.) < *vepti is a borrowing from Proto-Germanic *wefti- ‘weft’ (> English weft, Swedish väft). This etymology is considered “probable” in SSA, whereas according NES the word only “possibly” derives from Germanic. Besides, the attested English form is misspelled as *veft in NES.

4. Finno-Ugric inheritance

While the bulk of Finnish vocabulary consists of loans and other innovations, a few hundred of its core lexical items are inherited from earlier Finno-Ugric proto-language stages. Such old vocabulary has for long been a central object of etymological study, and the general picture of shared lexical material in the Finno-Ugric family seems quite clear by now, even though new etymological equations are still occasionally discovered. Even so, more important advances in this subfield have recently been reached through the critical reassessment of the corpora of old etymologies rather than the discovery of new ones. During the past few decades the understanding of Finno-Ugric historical phonology has become more precise especially due to the studies by Janhunen (1981) and Sammallahti (1988). At the same time the number of Finno-Ugric etymons that are considered reliably reconstructed has decreased rather than increased, as many comparisons that were considered well-founded in earlier research have turned out to involve irregularities after all. How are these developments reflected in NES? Häkkinen’s position can be characterized as rather traditionalistic: even though she states that “the earlier etymologies have been critically gone through and... amended when needed” (p. 6), in reality NES often just repeats what is said in SSA or Uralisches etymologisches Wörterbuch (UEW; 1988). Also the majority of proto-language reconstructions given for Finno-Ugric words are taken directly from UEW, which is a bad choice because there has been notable progress in phonological reconstruction since the publication of this dictionary. For instance, Janhunen (1981) and Sammallahti (1988) have convincingly argued that a consonant *x as well as a high back unrounded vowel *ï must be reconstructed for Proto-Uralic, both of which are missing from UEW’s reconstructions. In fact, UEW also ignores much of the etymological research conducted after the 1960s, and because of this the dictionary was regrettably outdated already at the time of its publication in 1988. This makes it a poorly suited primary source for Finno-Ugric etymologies, and BOOK REVIEWS 399 now many outdated comparisons continue their life in NES. Let us give a couple of examples:

• Finn. savi ‘clay’ is equated with Erzya Mordvin śovoń, Mari šun, Komi śun, Mansi šul' and Khanty sŏwĭ (all with the same meaning ‘clay’), and the Finno- Ugric proto-form is reconstructed as *śawe. The sound correspondences between these words are completely irregular, and the presented reconstruction has no basis. Only the Mordvin, Mari and Komi forms are doubtlessly etymologically linked, and even in their case borrowing in one direction or the other is possible.

• Finn. vaalea ‘light-colored’ is equated with Udmurt val' ‘shine, shimmer; shining’, Komi vol'alny ‘to shine, glitter’ and Mansi wol'gi ‘to shine’, and the Finno-Ugric proto-form of these words is reconstructed as *wal'V. The vowel correspondences are entirely irregular, and the same applies to the consonant correspondence l ~ l'. According to the present view there are no reliable etymologies supporting the reconstruction of a phoneme *l' in Finno-Ugric (Sammallahti 1988: 491).

Comparisons which are demonstrably wrong are especially common in the case of Samoyed languages. The following serve as examples:

• Tundra Nenets ngæwey° ‘brain’ is presented as an uncertain cognate of Finn. aivot ‘brain’. The word is actually a derivative of ngæwa ‘head’, which is cognate with Finnish oiva ‘good, splendid’ (cf. s.v. aivot, oiva).

• Nganasan mou ‘earth’ is presented as the cognate of Finn. maa ‘earth, land, ground’. However, it derives from Proto-Samoyed *məjå ‘earth, land’ (Janhunen 1977: 85), which is actually the cognate of Finn. muta ‘mud’ < *muδ'a (Aikio 2002: 22–23).

• Selkup mäkte and Kamas mekte ‘tussock’ are given as cognates of Finn. mätäs id., and these are claimed to derive from Proto-Uralic *mäkte. This equation is phonologically unacceptable, because Proto-Uralic *k has regularly disappeared in Proto-Samoyed adjacent to obstruents (*t, *c, *s, *ś): one would expect *mäkte to have developed into Selkup *mäte etc. (Janhunen 1981: 251).

One could have avoided repeating such mistakes of earlier etymological dictionaries by making more extensive use of newer primary studies. References to Janhunen (1981) and Sammallahti (1988), both fundamental works in Finno-Ugric historical phonology, are surprisingly rare in NES. In fact, as Janhunen and Sammallahti have already to a significant extent done the task of critically going through the earlier Finno-Ugric etymologies and 400 ANTE AIKIO amending them as needed, it is not easy to see why Häkkinen has chosen to ignore these results and to start over with the same task. Even though several outdated Finno-Ugric etymologies are included, references to many newer word comparisons are missing. NES does not mention, for instance, that according to Janhunen (1981) Finn. mukaan, mukana ‘along with’ are fossilized case forms of the Proto-Uralic noun *muka ‘back’, Finn. anoppi ‘mother-in-law’ is originally a compound consisting of Proto-Uralic *ïna ‘mother-in-law’ and *ïppi ‘father-in-law’, and Finn. kantaa ‘to carry’ is a causative derivative of Proto-Uralic *kani- ‘to go away’. More recent etymological equations unmentioned in NES include those presented for Finn. muta ‘mud’ (Abondolo 1996: 28–29), lähi- ‘near’ (Helimski 1999), sivu ‘side’ (Kulonen 2001), juoda ‘to drink’, kaivaa ‘to dig’, kääriä ‘to wrap’, kätkeä ‘to hide’, lämmin ‘warm’, nisä ‘teat’, sota ‘war’, salama ‘lightning’, seistä ‘to stand’ and täysi ‘full’ (Aikio 2002).

5. Sound-symbolic vocabulary

The origin of sound-symbolic vocabulary (a.k.a. ‘descriptive words’ in Finland) has always been one of the hardest nuts to crack in etymology. Irregular phonological variation, semantic instability and frequent associations and contaminations with other words make studies in this lexical domain notoriously difficult. Hence, it is not surprising that etymologists have often been content with just labeling such words as “descriptive” without analyzing them in detail. Recently, however, sound- symbolic vocabulary has received increasing attention in Finnish etymological studies. Compared to the recently published SSA, where numerous words are characterized as ‘descriptive’ or ‘onomatopoetic’ without further argumentation, NES has taken a step forward. Since such ‘descriptive’ etymologies are often based on loose criteria and subjective impressions (see e.g. Koivulehto 2001; Aikio 2001; Nikkilä 2002; Mikone 2002: 33), it is excellent that Häkkinen has adopted a somewhat more cautious attitude: a number of explanations of this kind are treated with suspicion or even left unmentioned in NES, including those SSA proposes for kaapata ‘to seize, snatch’, kumpu ‘hill, hillock’, nuttu ‘jacket’, riippua ‘to hang’, tuoksua ‘to smell good’, tuppi ‘sheath’ and vihko ‘bundle, sheaf; notebook’. Even so, there would still be room for improvement, as NES is not completely free from dubious descriptive etymologies either. For instance, BOOK REVIEWS 401 it is not easy to see why sarastaa ‘to shimmer, dawn’ should be “seen most likely as an autochthonous descriptive word”, how päre ‘shingle, chip’ could be “a derivative of a word-stem imitating a buzzing or a creaking noise”, or on what grounds hiipiä ‘to sneak’ is interpreted as “a descriptive formation that describes the dragging noise caused by sneaking” (isn’t the purpose of sneaking to avoid making any noise?). Even a couple of new suggestions of this type are put forward, a surprising example of which is the explanation proposed for Finn. murhe ‘sorrow’ and murehtia ‘to grieve’: “According to a simple but entirely possible explanation the verb murehtia would be a derivative of the same sound-imitative root as murista [‘to growl’]... A person can express sorrow with lamentations that resemble growling.” As with the assessment of proposed Finno-Ugric etymologies, one gets the impression that something is left half done. It is obviously an improvement that a number of problematic descriptive suggestions have been left out, but it is inconsistent at the same time to advocate other equally implausible explanations. It would seem justified to categorically reject all descriptive etymologies as long as they are not based on any explicitly defined and justified criteria. Loose guesswork will not advance the research, and more sophisticated methods should instead be developed for the etymological analysis of sound-symbolic and expressive vocabulary. Only recently there have been some real advances in this field; one can mention Nikkilä’s (1998; 1999) studies on ‘suffix replacement’ and ‘sporadic consonant alterations’ in Finnish, as well as Jarva’s (2003) monograph which sheds much light on expressivization of loanwords.

6. Words of unknown origin

In every language there remains a body of words of whose origins can never be explained, regardless of how thoroughly etymological studies are conducted. In light of this, it is a bit curious that the previous etymological dictionaries of Finnish have been reluctant to admit that the origin of any Finnish word is unknown. To avoid this, two strategies have usually been employed: either the lexical data is listed without stating anything of the etymological background, or “descriptive” origin is proposed (see above). In contrast to this, it is an improvement that NES often explicitly states that the word is of unknown etymology, as in the case of anastaa ‘to deprive, rob’, hullu ‘crazy, mad’, impi ‘virgin (poetic)’, iskeä ‘to strike’, loppu ‘end’, mekko ‘dress’, neuvo ‘advice’, nuotio ‘campfire’, outo ‘strange’, 402 ANTE AIKIO pettää ‘to deceive, betray’, as well as in many others. This principle has not been applied entirely consistently, though: there are many times more words of unknown origin in Finnish than are classified as such in NES. What, then, is said about the origin of the rest of this vocabulary? This time a reduction in the number of words of unknown origin has been achieved by applying the concept of ‘autochthonous words’ (“omaperäiset sanat”): words lacking an etymology are frequently classified into an ‘autochthonous’ lexical stratum. To give an example, Finnish kynä ‘pen; feather’ has cognates throughout the Finnic languages, but the further origin of the word is obscure. Hence, Häkkinen writes that “due to the lack of other explanations the word kynä is in any case considered an old autochthonous word.” There are innumerable similar word articles in NES.2 The problem with these numerous “autochthonous” words is that the concept itself is quite obscure. In NES one cannot find a clue how the term is supposed to be understood, but the following passage in Häkkinen’s earlier handbook on Finnish etymology sheds some light on the issue:

“One can maintain the working hypothesis that all words are ancient and autochthonous, until proved to the contrary. Applying all the basic criteria of etymological research, i.e. phonological form, meaning and distribution, one must then weigh whether there is reason to start suspecting another explanation.” (Häkkinen 1990: 88)

First, one must note that this “working hypothesis” is applied inconsistently in NES: some words are said to be of unknown origin whereas others are classified as old and autochthonous exactly because their origin is unknown. For instance, cognates of Finn. liha ‘meat’ are found only in Finnic languages, and hence according to NES the narrow distribution suggests that it is a loanword, even though no loan original has been

2 Further examples of supposedly “autochthonous” words include, for instance, aita ‘fence’, ampua ‘to shoot’, anoa ‘to beg’, haamu ‘ghost’, haapa ‘aspen’, haljeta ‘to split’, harva ‘sparse’, henki ‘breath; life’, hetki ‘moment’, hioa ‘to grind’, hopea ‘silver’, huhmar ‘mortar’, huuhtoa ‘to rinse’, häntä ‘tail’, jauhaa ‘to grind’, julma ‘cruel’, kanki ‘handspike’, kastaa ‘to dip’, katketa ‘to break in two’, kyntää ‘to plough’, lisä ‘addition’, läkähtyä ‘to stifle’, matka ‘trip’, mela ‘paddle’, painaa ‘to press’, polkea ‘to trample’, räppänä ‘smoke hole’, salmi ‘sound, strait’, sarja ‘series’, selkeä ‘clear’, siivota ‘to clean up’, sitkeä ‘tough’, sormi ‘finger’, sääski ‘mosquito’, tasa- ‘level, equal’, terä ‘blade’, tähti ‘star’, udella ‘to pry’, ura ‘groove’, vaimo ‘wife’, vanha ‘old’, viskata ‘to throw, fling’, and voida ‘can, to be able to’. Even this list is far from exhaustive. BOOK REVIEWS 403 discovered. Because the distribution of kynä ‘pen; feather’ is equally narrow, one can only wonder why it is not suspected as a loanword as well. Second, it is paradoxical to assume that the lack of an etymology would form an etymological explanation in itself. If there is no information about the origin of a given word, then its origin simply remains unknown; the term ‘autochthonous word’ seems to be a mere circumlocution for the lack of etymology. The easiest way out of this confusion would be to abandon the misleading term altogether; the constant references to ‘autochthonous’ origin are only likely to blur the readers’ understanding of how words can originate.

7. In conclusion

Even though critical remarks have been made above, this should not be understood to imply that NES is a failed work. On the contrary, the great majority of the word articles in NES are free of problems and reflect sound scholarship, and the dictionary presents a wealth of information on Finnish etymology in a clear and accessible form. This is already a substantial achievement in the popularization of Finno-Ugric , and it is certainly appropriate that this work was nominated as a candidate for the respected Finnish Tieto-Finlandia prize, which is awarded annually for the best non-fiction book of the year. On the other hand, an etymological dictionary cannot be reviewed as a purely popular work even if it has been authored primarily with the non- expert reader in mind. True, an etymological dictionary does not necessarily need to present many new research results, but it is still an extensive scholarly contribution to examine the entire body of relevant publications in order to judge which of the proposed etymologies can be considered plausible. One must note that the problems of NES – such as the misleading references to ‘autochthonous words’ or the less convincing ‘descriptive’ etymologies – seem to stem more from the traditions of Finnish etymology in general than from the author’s own ideas in particular. Hence one would have hoped that a somewhat more critical attitude had been adopted toward the explanations presented in earlier references. Regardless of its flaws, one can readily recommend NES as a good reference for both linguists and non-experts interested in the origins of Finnish words. Even Finno-Ugric specialists may find it handy, especially due to the references given under each word article. As this work will 404 ANTE AIKIO certainly find many readers over the next decades, one must hope that the next edition will be published with a more hard-wearing binding.

References

Abondolo, Daniel (1996) Vowel Rotation in Uralic – Obug[r]ocentric evidence. School of Slavonic and East European Studies Occasional Papers 31. London. Aikio, Ante (2001) Miten kuvaannollisuus selittää sanoja? Tieteessä tapahtuu 4 / 2001: 61–63. —— (2002) New and Old Samoyed Etymologies. Finnisch-ugrische Forschungen 56: 9–57. Fortescue, Michael, Jacobson, Steven & Kaplan, Lawrence (1994) Comparative Eskimo Dictionary with Aleut Cognates. Fairbanks: Alaska Native Language Center. Helimski, Eugen (1999) Ural. *läs “bei, in der Nähe” im Samojedischen (und im Ungarischen?) Finnisch-Ugrische Mitteilungen 21/22: 77–81. Häkkinen, Kaisa (1987) Etymologinen sanakirja. Porvoo: WSOY. —— (1990) Mistä sanat tulevat. Suomalaista etymologiaa. Helsinki: Suomalaisen kirjallisuuden seura. —— (2002) Suomen kielen äänne- ja muotorakenteen historiallista taustaa. 3rd, revised edition. Turku: Turun yliopisto, Suomalaisen ja yleisen kielitieteen laitos. Janhunen, Juha (1977) Samojedischer Wortschatz: gemeinsamojedische Etymologien. Castrenianumin toimitteita 17. Helsinki. —— (1981) Uralilaisen kantakielen sanastosta. Journal de la Société Finno-Ougrienne 77: 219–274. Jarva, Vesa (2003) Venäläisperäisyys ja ekspressiivisyys suomen murteiden sanastossa. Jyväskylä Studies in Humanities 5. Jyväskylä. Kluge, Friedrich (1995) Etymologisches Wörterbuch der Deutschen Sprache. Ed. by Elmar Seebold. 23rd, extended edtion. Berlin / New York: Walter de Gruyter. Koivulehto, Jorma (2001) Etymologie und Lehnwortforschung: ein Überblick um 2000. Finnisch-Ugrische Forschungen 56: 42–78. Kulonen, Ulla-Maija (2001) Eräiden ruumiinosien nimeämisen motivaatiosta. In Sándon Maticsák, Gábor Zaicz & Tuomo Lahdelma (eds.), Ünnepi könyv Keresztes László tiszteletére, pp. 293–301. Folia Uralica Debreceniensia 8. Magnússon, Ásgeir Blöndal (1989) Íslensk orðsifjabók. Reykjavík: Orðabók Háskólans. Mikone, Eve (2002) Deskriptiiviset sanat – määritelmät, muoto ja merkitys. Helsinki: Suomalaisen kirjallisuuden seura. Nikkilä, Osmo (1998) nop-ea → nop-sa, sel-keä → sel-vä: Suffiksinvaihto ja suomen sanojen etymologiointi. In Urho Määttä & Klaus Laalo (eds.), Kirjoituksia muoto- ja merkitysopista. Folia Fennistica & Linguistica 21, pp. 77–101. Tampere: Tampereen yliopiston suomen kielen ja yleisen kielitieteen laitos. —— (1999) Sporadischer Konsonantenwechsel im Ostseefinnischen. Finnisch- Ugrische Mitteilungen 21/22: 129–160. —— (2002) Suomen sanojen alkuperät päätökseen. [A review of SSA III.] Virittäjä 106: 124–136. BOOK REVIEWS 405

Sammallahti, Pekka (1988) Historical Phonology of the . In Denis Sinor (ed.), The Uralic Languages. Description, History and Foreign Influences, pp. 478–554. Leiden: Brill. SKES = Suomen kielen etymologinen sanakirja. Lexica Societatis Fenno-Ugricae XII. Helsinki: Suomalais-ugrilainen seura. 1955–1981. SSA = Suomen sanojen alkuperä. Etymologinen sanakirja. Helsinki: Kotimaisten kielten tutkimuskeskus / Suomalaisen kirjallisuuden seura. 1992–2000. UEW = Rédei, Károly (1988–1991) Uralisches Etymologisches Wörterbuch. Budapest: Akadémiai Kiadó.

Contact information:

Ante Aikio Giellagas Institute for Saami Studies P.O. Box 1000 90014 University of Oulu Finland e-mail: [email protected]

Anne-Marie Simon-Vandenbergen, Miriam Taverniers and Louise J. Ravelli (2003) Grammatical Metaphor: Views from systemic functional linguistics. Amsterdam / Philadephia: John Benjamins. Pp. 453.

Reviewed by Asunción Villamil

Since the 1980s, lexical metaphor has received a great deal of attention in linguistics, especially since the breakthrough of Lakoff and Johnson’s Metaphors we live by (1980). Metaphor was shown to be ubiquitous and not only a feature of crafted literary language. The interest in metaphor spread from cognitive linguistics to functional trends, resulting in Halliday’s proposal of the notion of grammatical metaphor. While other schools focus on lexical aspects, Systemic Functional Linguistics focuses on the grammatical dimension and socio-functional aspects. The result of this interest is a rich body of research to which this volume contributes. The book is a compilation of articles devoted to the study of grammatical metaphor from the perspective of Systemic Functional Linguistics (SFL). It represents the contribution of several distinguished scholars to present the current state of grammatical metaphor theory and its more relevant fields for further research, as well as the history and definitions of grammatical metaphor and problematic issues in the study of lexico-grammatical metaphor. It consists of a preface by J.R. Martin (pp. 1–3), an introductory chapter by one of the editors, and five main parts: Grammatical metaphor: Clarification and application (Part I., p. 35), Development of metaphor in children (Part II, p. 149), Interpersonal metaphor: Enactment and positioning (Part III, p. 221), ‘Metaphor’ in grammar and in other modes of meaning (Part IV, p. 309) and Metaphor in metalinguistic perspectives (Part V, p. 367). The book opens with an introductory chapter by Miriam Taverniers (“Grammatical metaphor in SFL: A historiography of the introduction and initial study of the concept”, pp. 5–33) that offers a historiographic perspective on the origins and early developments of the term ‘grammatical metaphor’. In a very concise and clear way the author includes a description of the term, the different types of metaphors and ‘congruent’ and ‘non-congruent’ realizations, all of them key concepts for approaching grammatical metaphor from the point of view of SFL. This provides a valuable help for those readers who are not familiar with the basic tenets of grammatical metaphor theory, while it increases their eagerness to know more about recent definitions and the evolution of the theory.

SKY Journal of Linguistics 18 (2005), 407–416 408 ASUNCIÓN VILLAMIL

The next articles address the questions raised in the previous chapter by offering a deeper analysis of the concept of grammatical metaphor together with some of its applications. It is centred on nominalization, considered by Halliday the typical and most abundant example of grammatical metaphor (Halliday 1994:352). The first article by Louise J. Ravelli (“Renewal of connection: Integrating theory and practice in an understanding of grammatical metaphor”, pp. 37–64) gives an insight into the improvements which the theory of grammatical metaphor has received from the practical use of the concept, such as the inclusion of new types of metaphors. She provides some practical and revealing applications of metaphor in the teaching of academic writing and explores possible future developments. The revealing examples are the most remarkable contribution of the article. The need to re-evaluate theoretical postulates is addressed by Liesbet Heyvaert in her chapter “Nominalization as grammatical metaphor: On the need for a radically systemic and metafunctional approach” (pp. 65–99). The article presents the concepts of ‘enation’ (the relationship between structures which have the same configuration) and ‘agnation’ (the relationship of a nominalization and its non-nominal equivalent), the latter already used in SFL. The author argues that a combined used of both terms would be a source of improvement for the theory of nominalization. The convincing theoretical discussion is exemplified through nominalizations construed by -ing and –er suffixes. Moreover, she advocates a study of the interpersonal component, since the focus has so far been placed on ideational metaphors. The article by Jorge Arús Hita (“Ambiguity in grammatical metaphor: One more reason why the distinction transitive/ergative pays off”, pp. 101–126) is also devoted to nominalization but it accounts for the differences in the nominalization of ergative and transitive clauses, with a special focus on ambiguous clauses. Ambiguity (transitive) and vagueness (ergative) are shown to be key factors in these two types of nominalizations, among other features. This hypothesis is again put to the test of a corpus analysis, in this case examples gathered from newspaper headlines. Apart from the theoretical relevance of the paper, the application of these findings and the analysis of examples to second language learning and teaching are undeniable. If the previous article defended the importance of agnates, this study illustrates a practical application of this

BOOK REVIEWS 409 notion (a nominalization from an ergative clause is not equivalent to that from a transitive sentence). Nominalization is also discussed in David Banks’ article (“The evolution of grammatical metaphor in scientific writing”, pp. 127–147), which explores the origins of this device in scientific writing. Despite being a common feature of scientific style nowadays, it has not always been so pervasive, as becomes evident from the contrasts of articles from different periods since the beginnings of modern science in the 17th century. Apart from variation in time, scientific writing also involved typological variety. Different scientific branches, namely, biological and physical sciences, present dissimilar characteristics in their use of nominalization, especially at their beginning, due to their different pace of development. It is also taken into account how the context may have influenced the differences in the increased use of nominalizations between various branches of science. In all cases, scientific style displays an evolution towards an increased used of nominalizations. The second part of the volume takes a radically different standpoint and concentrates on the acquisition of metaphor, from preschool children to early adolescence. The first article of this series (“The use of a metaphorical mode of meaning in early ”, by Clare Painter, pp. 151–167) is a longitudinal study of the evolution of ideational metaphor in two children. Children are reported to use lexical metaphor but also grammatical metaphor, as in the use of nominal expressions instead of a whole sentence. At an early stage of development the use of metaphor is restricted to one-word structures and the only metaphors appearing are dead metaphors copied from adult language. The earliest metaphor reported is the use of postmodifiers within the nominal group. Further examples are reported in metalanguage and abstract domains such as time. The examples show that children at very early stages (preschool years) already explore the meaning potential of a limited but existing class of grammatical metaphors. This is pushed further by exposure to written language. Although quite a number of examples are introduced as illustration, it would have been interesting to present them more exhaustively and in a more structured way for clarification. Preschool children are also the target of Jane Torr and Alyson Simpson (“The emergence of grammatical metaphor: Literacy-oriented expressions in the everyday speech of young children”, pp. 169–183). Their

410 ASUNCIÓN VILLAMIL longitudinal study of five children examines the expressions used by children in their exchanges with caregivers. The data is composed of spontaneous spoken language at home and is restricted to interpersonal metaphor, since no productive forms of ideational metaphor are found before five years. Metaphors are organized according to the age of the children and the result is that interpersonal metaphors become more elaborated with time, as systems of mood and transitivity expand, and are especially developed in the preschool years, particularly when children ask for goods and services. This development is related to literacy development: if children’s ability to employ these linguistic resources (which are literacy-oriented expressions) is fostered, they build on a path to a successful academic development. Beverly Derewianka discusses another longitudinal study of two children (“Grammatical metaphor in the transition to adolescence”, pp. 185–219). The research is based on the analysis of written production, with some spoken examples, in contrast to the two previous articles which were limited to oral production. Drawing on previous taxonomies, an analysis of the written production is carried out, from precursors of grammatical metaphor, like devices to play with interstratal relationships, to protometaphors and finally examples akin to those of the adult system. The author provides a detailed analysis and classification of examples, in addition to a very careful and systematic organization and discussion of data. The classification allows the reader to see which types of metaphors are developed earlier. Different motivations are suggested for the use of the metaphor in each case. The earliest examples are explained in terms of ‘trailer strategies’, where the child ventures beyond his current ability to test the potential of the system. A dramatic increase in the use of metaphor is found to take place around age nine, also due to the influence of context, especially teacher encouragement as well as the nature of the writing tasks that the child is required to master at school. It is also revealed that although grammatical metaphor is typical from written mode, much of the experimentation is carried out at the spoken level, where there is not such a strong commitment as in the written text. Leaving aside metaphor development, the next part aims at interpersonal metaphor and defends the view that metaphorical devices constitute a continuum. The first chapter by Anne-Marie Simon- Vandenbergen (“Lexical metaphor and interpersonal meaning”, pp. 223– 255) surveys the contrast between grammatical and lexical metaphor.

BOOK REVIEWS 411

Lexical metaphor has not been given too much emphasis in Systemic Functional Linguistics, as it does get in cognitive linguistics, but Simon- Vanderbergen claims that the dichotomy between both is artificial, since metaphors are used in larger configurations which include both lexical and grammatical shifts. To illustrate her point she analyses thoroughly a selection of metaphorical expressions of verbal processes from the Longman Dictionary of Contemporary English (1978, 1987). The connection of these grammatical metaphors with lexical aspects is highlighted before passing on to examine the advantages of that integrated analysis. She categorises the different types of interpersonal meanings that they convey (judgement, appreciation, graduation, etc). Apart from the dictionary data, she looks more closely at the actual use of the expressions barge into (a conversation), fabricate (a story), plunge into (a description) and dish out (advice). This analysis supports the thesis that the motivating force behind the creation of lexicogrammatical metaphors is primarily the speaker’s attitude towards the judgement of the speech event, that is, an interpersonal motivation. She concludes that lexicogrammatical metaphor can be conveniently studied in Systemic Functional Linguistics, since what were traditionally regarded as lexical metaphors can be related to basic conceptualizations of verbal processes. The discovery of further examples of interpersonal metaphors is the objective of “The elided participant: Presenting an uncommonsense view of the researcher’s role” (pp. 257–278), by Geoff Thompson. Thompson aims at pointing out strategies for eliding world participants (participants assumed to participate in the physical or mental event or state) in the clause. The participant under investigation is limited to that acting as “researcher”. Drawing from a corpus of university books and academic papers he explores the following strategies: nominalization, passivization, elision of the Senser, indexing the participant and metonymy. Discussion and conclusion sections are compared but no differences are visible. Motivations for elisions are argued to be complex, including, among others, conforming to research style conventions, impersonalising a sad event or offering controversial views. Thompson also discusses the process that the metaphor undergoes in order to be recognized by the reader and the status of the strategies commented as grammatical metaphor, as some of them are fundamentally different. He is inclined to consider them examples of grammatical metaphor but he emphasizes that the concept of metaphor has

412 ASUNCIÓN VILLAMIL fuzzy borders, so that some of these strategies might belong to the periphery cases. The purpose of the following article (“Imperative readings of grammatical metaphor: A study of congruency in the imperative” by Inger Lassen, pp. 279–308) is to discuss the nature of the imperative in grammatical metaphor theory. The imperative has often been presented as a congruent realization of instructions, as well as warnings, recommendations and other speech acts. However, the author questions whether the imperative is always congruent. In order to answer this question, she first reviews and contrasts exhaustively the meanings of lexical and grammatical metaphor by applying Halliday’s theory, alongside indirect speech acts and literal utterances as presented in Searle’s Speech Act Theory. An experiment is conducted to test the status of the imperative, selecting some non-specialist native speakers to react to some extracts from professional texts, providing co-operative and non-co-operative responses, without the aid of context or background knowledge. The author analyses the ways in which the co-operative principle is flouted in the responses as well as the difficulties in suggesting non-co-operative replies; the latter are taken to be congruent since there is coincidence of form and function. The results of this sample, although limited, lead the author to look upon the imperative on a cline of metaphoricity, with lexical metaphor at the one end and grammatical metaphor at the other. The imperative is therefore claimed to be used both for metaphorical and congruent purposes. Part IV enters a riskier and more challenging field, extending metaphor to modes of meaning other than grammar. The first suggestion by Robert Veltman is the extension of the analogy of grammatical metaphor to phonological metaphor (“Phonological metaphor”, pp. 311–335). Abundant theoretical arguments are presented concerning the nature of semiotic systems, sounds and intonation, focus, attitude, and arbitrariness, in order to argue that sounds can perform metaphorical realizations, including interpersonal and textual meanings. Furthermore, Veltman introduces a divergence from Halliday’s approach to the treatment of interpersonal metaphor: Halliday argues that interpersonal metaphor is implicitly concise and inexplicit in contrast with the explicit and transparent congruent form; however, Veltman argues that metaphorical realizations are more concise and inexplicit. Evidence is drawn from the analysis of the intonation patterns of a selection of BBC programmes. The recorded utterances are argued to be agnate with a set of (hypothetical)

BOOK REVIEWS 413 more complex utterances with an unmarked intonation pattern. The metaphorical recorded utterance is simpler than the possible agnates, in that it expresses far more economically the attitudinal and propositional information transmitted. The following article is the first to introduce the relationship between metaphor and semiosis, which constitutes the main topic of the last articles of the volume. This last part is concerned with the most abstract field of analysis so far, focusing on . Kay O’Halloran (“Intersemiosis in mathematics and science: Grammatical metaphor and semiotic metaphor”, pp. 337–365) discusses the interaction between mathematical symbolism and visual display and the development of grammatical metaphor in scientific writing. Like David Banks’ article, O’Halloran takes several examples from early research papers by Newton and Descartes, among others. From the basis of a Hallidayan description of scientific language metaphor, he compares the potential of language to configure processes and participants in comparison to mathematical symbolism. Mathematical symbolism is shown to express configurations of participants and processes as well as change and covariation; these areas being conveyed by symbolism, language evolved to provide causal explanations. Visual representation also comes into play, as it allows for the occurrence of semiotic metaphor, with a mapping from the congruent visual representation to linguistic construal. Thus, the author argues that the co-evolution of these semiotic resources - linguistic, symbolic and visual - leads to a shift in language from the concrete, in terms of entities and cause-effect explanations, to metaphorical language. Patrick Goethals deals with the nature of language and how different trends in linguistics regard communication in “The conduit metaphor and the analysis of meaning: Peircean semiotics, cognitive grammar and systemic functional grammar” (pp. 369–389). He examines how Peircean semiotics, Cognitive Grammar (CG) and Systemic Functional Grammar (SFG) tackle the CONDUIT METAPHOR and its view of communication, which suggests that linguistic meaning is restricted to building up a message that is transferred in a context which does not relate to representational meaning. The article explores the way how these different paradigms include the interactive and performative aspect of meaning which is ignored by the CONDUIT METAPHOR. After a basic description of each school and a lucid and clearly-structured comparison between

414 ASUNCIÓN VILLAMIL them, the author concludes that the three of them avoid the restrictions of the CONDUIT METAPHOR by including the interactive component, even though they present several discrepancies in their approach. The most basic difference lies in the fact that Peircean semiotics makes the most exhaustive categorization of ways of relating the utterance to the context, distinguishing tense and illocutionary force. This distinction is based on the contrast between indexical and symbolic signs: indexical signs identify the illocutionary force or the performative value of an utterance and indicate that there is a speech act taking place, operating on the performative level, whereas symbolic sings (as tense) build up propositional content and therefore work on the content level. In contrast, CG and SFG subsume them under the same category, within the grounding of the utterance and the mood component, respectively. Goethals defends the contrast between these two types of grounding and suggests that the differences between the three schools stem from different concerns about meaning. Systemic Functional Linguistics (SFL) and cognitivism are again contrasted in the following article (“Grammatical metaphor as a cognitive construct”, pp. 391–415). Randalm Holme looks at the possibility of combining systemic and cognitive accounts of metaphor through the example of metaphors that express cause-and-effect. Both descriptions give different emphasis to the description of the same phenomenon: while in SFL the focus lies on the structures that arise from interaction and social context, Cognitive Linguistics studies how properties of cognition affect those structures. In this sense they can be regarded as complementary, inasmuch as they provide different perspectives of analysis: more general and related to cognition vs. more concrete and related to the context of communication. In a search for a path common to both schools, it is suggested that the SFL notion of congruency can be regarded as a metalinguistic device to explain metaphorical meaning and not as an example of a natural use of language, since metaphor is argued to be natural to the mind. Moreover, the author proposes that SFL can cover some weaknesses of cognitive analysis, for instance how a given conceptual metaphor comes to take a given textual form. SFL concepts of genre and register can show how context may affect the preference of one metaphor over another. The last article by Robin Melrose (“‘Having things both ways’: Grammatical metaphor in a systemic-functional model of language”, pp. 417–442) discusses nominalization and its dual nature. It takes as a starting

BOOK REVIEWS 415 point Halliday’s description of its features as field- or wave-like: nominalization in discourse functions as a thing but it is also a textual strategy (wave-like). This function as a textual strategy is extended by Melrose to further purposes: nominalization is claimed to be a powerful device to condense information about the context of situation and of culture. For example, nominalization can be used as a way of showing that you have mastered a discipline or that the reader and writer belong to the same social group. In order to account for these extended functions, SFL views on semantics, genre and ideology are taken into consideration. Abundant theoretical evidence and references are presented to support these claims, together with practical analyses of some extracts from a piece of academic writing. The author stresses the fact that nominalization is an open and dynamic device, between the topological and typological poles. After summarizing the main points of the papers of which the volume consists, let us now turn to some concluding evaluative remarks. The volume will prove of special interest for researchers within the field of metaphor studies, either from systemic linguistics or other trends, as it offers a wide and updated view of the area. Not only does it offer a review of past research in metaphor, but it also opens a way for new and stimulating paths of investigation, especially in the last sections of the volume. The book is notable not just because it provides a rich amount of theoretical background, but also because of the excellent corpus studies which serve as the perfect counterpart for more theoretical approaches. Another significant aspect is the possibilities of communication and cooperation with other frameworks which are discovered throughout the book. This is especially relevant in the case of cognitive accounts (cf. Holme 2003: 391–415 or Simon-Vandenbergen 2003: 223–255). However, most of the articles are restricted to the application of the systemic approach to metaphor, which can prove satisfying but less innovative than the inclusion of other trends. Similarly, nominalization and ideational metaphors are the subject of many of the papers, which demonstrates the weight of early proposals by Halliday. The reader may found the originality of those articles that depart from these central topics more appealing. Especially rewarding are those articles which, like Veltman’s (2003: 311- 335), bring about controversial modifications of some theoretical aspects. Furthermore, some of the introductions of the articles are somewhat repetitive, as most of them make references to the same theoretical

416 ASUNCIÓN VILLAMIL background, but it should be remembered that this is due to the independent nature of each of the articles, common in compilations. The above critical remarks are not meant to question in any way the overall quality of the book. On the whole, this work represents a valuable contribution to the advance of the research in grammatical metaphor and a comprehensive overview of the field. It is an extremely worthy attempt at condensing in a unitary compilation the large amount of research that has been and is currently being carried out in metaphor studies.

References

Halliday, M.A.K. (1994) An Introduction to Functional Grammar. 2nd Edition. London: Edward Arnold. Lakoff, G. and M. Johnson (1980) Metaphors we live by. Chicago / London: The University of Chicago Press.

Contact information:

Asunción Villamil Touriño Escuela Oficial de Idiomas de Las Rozas Camino del Caño 2, 28230 Madrid. Spain e-mail: [email protected]

Anna Vogel (2004) Swedish Dimensional Adjectives. Stockholm: Almqvist & Wiksell International. (Illustrated by Agnes Stenqvist.) Pp. iv, 378.

Reviewed by Heli Tissari

This book deserves attention for several reasons, the main one being that it is a systematic and thorough account of Swedish dimensional adjectives, including hög, låg, bred, vid, tjock, smal, tunn, trång, djup, grund, lång, and kort, but there are several subsidiary reasons. This is a book for people interested in cognitive linguistics, prototypes, embodiment, and spatial conceptualization, as well as for people interested in combining corpus methodology with dictionaries and elicitation tests. Vogel herself also points to the relevance of her work for the study of antonymy (pp. 11, 359 et passim). What makes the book even more attractive to read is that it contains 35 apt illustrations of the issues discussed. Towards the end of her book, Vogel says that she has “wanted to pay … attention to interesting cases that were relatively infrequent” (p. 356). In my view, this statement characterizes her dissertation very well. It is full of small details which make it sometimes slow to assimilate, indeed puzzling or as if requiring a counterargument, but also charming like a well- renovated and lovingly decorated Swedish country house with many cupboards, small windows, niches and secret little doors. One must also be aware that the house overlooks a broad landscape, since the topic is pertinent to general theorizing within cognitive linguistics. The index (pp. 374–377) reveals a number of thematic threads which hold the book together. These comprise containment, direction, front (for functional reasons, as regards dealing with everyday objects), horizontality, orientation, path, perspective, surface, verticality, and vision. What also comes through in the text itself is grabbing, holding, moving, mending, and using objects and tools, as well as moving in changeable environments and regarding these and their consistent parts (such as houses, trees, streets or rivers) from various points of view. The focus is on non-metaphorical uses of the words studied (pp. 36–37). Vogel’s data come from Språkbanken (the Bank of Swedish corpus), an elicitation test, and five major Swedish dictionaries (pp. 33–44). She also emphasizes her native speaker intuition (p. 44), while admitting that not all native speakers agree on all issues (see, e.g., p. 327). Sometimes they do not even have an opinion (p. 306). It would hardly be reasonable for a single person to analyse more corpus data for one thesis, considering that the two Språkbanken sections she has used for her searches total more

SKY Journal of Linguistics 18 (2005), 417–422 418 HELI TISSARI than 9 million words (p. 33). It puzzled me a little that she suggested a “great amount of spatial, non-metaphorical uses in fiction” (p. 355). I would have liked to see this explained and evidenced in more detail as against other genres or text types. It is difficult not to be somewhat critical of the elicitation test, since Vogel herself admits that she would have liked to develop her first version of the test but did not have the time (p. 356). However, if there is a shortcoming in this respect, it is at least partially overcome by her own introspection and observations on spoken language and everyday life (p. 33, cf., e.g., pp. 182, 184, 188, 191). At this point it might be mentioned that humour enlivens the discussion of these and other issues throughout the book. The book presents us with everyday experiences ranging from eggs and soup to fields of wheat and ground frost (pp. 161, 205, 226, 277, 323, 325). It deals with pokers and ceiling boards as well as towels and jewellery (pp. 142–143, 147–148, 174–175, 223, 318–319). Of the natural elements, water and snow are favourite topics. Her summary on waves is (p. 350):

…vågor ‘waves’ can be described as höga ‘high’, but hardly as låga ‘low’, as djupa ‘deep’ but hardly as grunda ‘shallow’, as långa ‘long’, as korta ‘short’, as tunna ‘thin’ but not as tjocka ‘thick’.

Discussions of furniture and buildings continue throughout the book. Among other things, Vogel considers the orientation of the human body with respect to washing machines (pp. 135–138). It is difficult to imagine how to come closer to the motto formulated by Rohrer (2001: 78): “Cognitive linguistics can and should be vitally engaged with pragmatic problems.” The discussion of each Swedish adjective is preceded by an overview of research on related adjectives in other languages, including English, French, German, Polish, Russian, Spanish, Yucatec Maya, and some other, more sporadic examples such as Latin and Finnish. A difficulty concerning the discussion is that it combines studies based on several different theoretical assumptions. The title of Vogel’s book connects it with Bierwisch and Lang’s Dimensional Adjectives (1989), but they actually represent a rather different school of thought, intending what might be called a “highly specific system of logical relations” (Bierwisch 1989: 2), which they capture in tree diagrams and complex formulae (see, e.g., Bierwisch 1967: 19–21, 32–33, Lang 1989: 288–289, 373–375). The degree of mathematical BOOK REVIEWS 419 precision in their work is higher than that achieved by Vogel. While we must keep in mind that such precision was not her aim, simply considering her theoretical premises, she could perhaps still have attempted to summarize some of her results in a manner akin to theirs. A simple example is Bierwisch’s scheme for “different dimensions of certain objects” (1967: 14–15), in which he lists the German words used for the three dimensions of cars, doors, and streets. As regards theoretical premises, Vogel herself is avowedly open- minded, coming to the conclusion that studies representing various traditions yield useful and often similar results. She is also of the opinion that the Swedish dimensional adjectives behave much like their German and English counterparts (p. 353). She herself considers her adjectives as forming prototypical categories, favouring Langacker’s (1987: 369–408) network model e.g. on the assumption that “there are relations not only between the central member and the remaining members, but also between separate non-central members” of the network (p. 357). She arrives at final descriptions of adjectives in the following style (p. 242):

The prototypical use of trång ‘narrow’ combines the adjective with a noun referring to a path, passage or container for a human being. Less prototypical uses involve objects that are surfaces perceived as containers and passages, for which trång ‘narrow’ invokes only one dimension, as is the case with trång dörr ‘narrow door’ as opposed to trång öppning ‘narrow opening’. When only one dimension is invoked, trång ‘narrow’ and smal ‘narrow/thin’ overlap.

Trång is considered a simple category, since it has only one prototypical use, whereas the related adjectives smal ‘narrow/thin’ and tunn ‘thin’ are regarded as complex categories with more than one centre (p. 241). Developing the concept of semantic networks does not appear to be very high on Vogel’s agenda (cf. Evans 2005). Rather, she concerns herself with showing that dimensional adjectives do not belong to antonyms par excellence (as implied by Ljung 1974, although less directly, as Vogel, p. 359, seems to claim) and “do not form a neat patchwork” (p. 359). While the mere length of her discussion suggests that a “neat patchwork” account is not the whole story, she might have summarized her findings numerically as well in order to prove the point. She does include simple statistics all along, but finishes with almost apologising for not using “statistical analysis” (p. 356). This kind of hedging characterizes a number of passages of her thesis, but is often quite unnecessary and could either be entirely left out or supplemented by a bolder formulation. 420 HELI TISSARI

Vogel’s denial of the “neat patchwork” account also agrees with previous studies and is thus by no means revolutionary. Consider Bierwisch’s (1967: 31) statement that “not all the (+Pol) space adjectives have simple and unique antonymous counterparts” or the title of Lang’s (1989: 334–335) section “Designation gaps”. With respect to embodiment, Vogel favours Sinha’s ideas, being especially fond of the criticism he directs at the idea of the human body as a prototypical container (Sinha & Jensen de López 2000: 22). Vogel points out, for example, that the concept of fatness/bigness is not understood simply through having experienced one’s own body as heavy (and containing plenty of substance), but also through not being able to do things, such as buttoning up one’s jeans, through “feeling that [one’s] clothes are too tight”, or, even more importantly, associating other people’s bodies with fatness/bigness (pp. 198–199). By the way, in this passage as well as at some other points, it seemed to me that Vogel could probably have expressed herself even better in Swedish. It is a pity that the formulation of some of her ideas may suffer because she has directed her thesis to a larger audience than the Swedish-speaking, in itself is a very reasonable goal. Vogel repeatedly refers to Clark (1973), which is a good reference because it shows that many of the ideas proposed by cognitive linguistics from the 1980s onwards have been clearly expressed and discussed in research papers even earlier. Clark’s article (1973) is a somewhat superficial but concise and to-the-point summary of many issues concerning embodiment. These include a person’s “cognitive framework” (“man’s [sic] biological endowment”), the “three natural reference planes” (the left-right axis, the front-back axis, and the ground level), and the “canonical encounter”, which occurs face-to-face, rather than “side to side, or back to back” (Clark 1973: 28, 34–35). Vogel’s thesis demonstrates that she can analyse complex data productively. She clearly delights in pondering various spatial combinations and the linguistic expressions used to convey them, and is also quite interested in the theoretical aspects of these issues. She is nevertheless somewhat shy about making theoretical claims and developing them towards innovative generalizations. However, I suggest that other people might find her thesis a fascinating read even because of this – the book may be a source of inspiration for thinking about the nature of cognition and embodiment and for developing cognitive linguistics towards a better understanding and formulation of many issues concerning space. A BOOK REVIEWS 421 topic that is quite relevant here is spatial metaphors, which she does not discuss, but which have intrigued many other researchers. From a theoretical point of view, it would be very interesting to read a similarly detailed treatment of the behaviour of spatial metaphors and to see how they relate to their non-metaphorical counterparts. In my last quote from her book, Vogel comments on the process of collecting references for her thesis (p. 354): “Hopefully this has taught me to keep an open mind on ideas from unconventional sources.” I would like to finish by suggesting that if she wishes to develop her research along similar spatial lines, she might also consider whether and how the human body is partitioned (in relationship to other objects described) in terms of dimensional adjectives (cf. Dudis 2004), and how designers fit artefacts to people, more or less successfully (Norman 1993, 1998/2000). In the homeland of Ikea, such a project could have not only pragmatic but also quite practical consequences.

References

Bierwisch, Manfred (1967) Some semantic universals of German adjectivals. Foundations of Language 3: 1–36. Bierwisch, M. & Lang, E. (1989) Dimensional Adjectives: Grammatical Structure and Conceptual Interpretation. Berlin, New York, London etc.: Springer-Verlag. Clark, Herbert H. (1973) Space, time, semantics, and the child. In Timothy E. Moore (ed.) Cognitive Development and the Acquisition of Language pp. 27–63. New York, London etc.: Academic Press. Dudis, Paul G. (2004) Body-partitioning and real-space blends. Cognitive Linguistics 15–2: 223–238. Evans, Vyvyan (2005) The meaning of time: polysemy, the lexicon and conceptual structure. Journal of Linguistics 41: 33–75. Lang, Ewald (1989) The semantics of dimensional designation of spatial objects. In Manfred Bierwisch and Ewald Lang (eds.), pp. 263–417. Langacker, Ronald (1987) Foundations of Cognitive Grammar. Volume I: Theoretical Prerequisites. Stanford: Stanford University Press. Ljung, Magnus (1974) Some remarks on antonymy. Language 50–1: 74–88. Norman, Donald A. (1993) Things That Make Us Smart. Reading (Mass.): Addison Wesley. —— (1998/2000) The Design of Everyday Things. London: The MIT Press. Rohrer, Tim (2001) Pragmatism, ideology and embodiment: William James and the philosophical foundations of cognitive linguistics. In René Dirven, Bruce Hawkins & Esra Sandikcioglu (eds.), Language and Ideology. Volume I: Theoretical Cognitive Approaches, pp. 49–81. Amsterdam: John Benjamins. 422 HELI TISSARI

Sinha, Chris & Kristine Jensen de López (2000) Language, culture and the embodiment of spatial cognition. Cognitive Linguistics 11–1/2: 17–41.

Contact information:

Heli Tissari Helsinki Collegium for Advanced Studies P.O.Box 4 (Fabianinkatu 24) FIN-00014 University of Helsinki Finland e-mail: [email protected]

Top View