LINGUIST List 13.1200

Wed May 1 2002

Review: Morphology/Syntax: The Verb in Turkish

Editor for this issue: Terence Langendoen <terrylinguistlist.org>

What follows is another discussion note contributed to our Book Discussion Forum. We expect these discussions to be informal and interactive; and the author of the book discussed is cordially invited to join in.

If you are interested in leading a book discussion, look for books announced on LINGUIST as "available for discussion." (This means that the publisher has sent us a review copy.) Then contact Simin Karimi at siminlinguistlist.org or Terry Langendoen at terrylinguistlist.org.

Directory

  • Lisa A Shannon, Taylan (2001) The Verb in Turkish

    Message 1: Taylan (2001) The Verb in Turkish

     Date: Mon, 29 Apr 2002 15:18:44 -0700
    From: Lisa A Shannon <lisashanjuno.com>
    Subject: Taylan (2001) The Verb in Turkish


    Taylan, Eser E., ed. (2001) The Verb in Turkish. John Benjamins Publishing Company, xvii+267pp, hardback ISBN 90-272-2765-9 (Eur), 1-58811-073-7 (US & Can), $77.00, Linguistik Aktuell/Linguistics Today 44

    Lisa Shannon, University of Arizona

    OVERVIEW The study of Turkish in the theoretical linguistics tradition is a relatively new endeavor, spanning only a few decades. Research on this language has expanded, though not nearly to the extent of the European languages. For this reason, when a collection of papers is published which are entirely dedicated to the morphosyntax and semantics of the Turkish verb, it is a welcome addition to the linguistics literature. In May of 1999 a workshop was held at Bogazici University in Istanbul, at which much of the research in this volume was first presented. Bogazici is an impressive university whose Department of Linguistics is beginning to attract well-deserved attention. The papers included in this collection, many of which are authored by Bogazici faculty, showcase a wide range of expertise and theoretical approaches to Turkish linguistics that either challenge or complement much of the work that has been done to date.

    Because of the limited characters allowed by this format, I will use the following convention when Turkish orthography is necessary (I have not used it with Turkish proper names because it looks too weird): C = voiceless palatal affricate (a lower-case c is the voiced counterpart in Turkish words) I = the undotted i (mid back vowel) G = the 'soft' g (unpronounced) S = voiceless alveolar fricative Note that Turkish affixes undergo alternations due to vowel harmony and consonant voicing, so there will be spelling variations in the examples.

    SUMMARY In the first paper, "Finite inflection in Turkish", Engin Sezer of Harvard University presents a formal account of the behavior of the Turkish inflectional complex within a Minimalist framework. In Sezer's analysis, the inflected verb is laid out in this way: V + Tense1 + Aux + Tense2 + Aux + Tense3 + Agr There are nine Tense/Aspect/Mood suffixes and three possible positions for them to appear, with most suffixes allowed only in Tense1 position. However, three of the suffixes, -(y)di [past], -(y)miS [evidential], and -(y)se [conditional], can appear in Tense2, and the latter two can also appear in Tense3. As seen in the verbal string, the Tense2 and 3 suffixes can only hosted by an auxiliary verb. For this reason, Sezer presumes them to be morphologically complex, and therefore syntactically and semantically distinct from Tense 1 affixes. He treats tense as a syntactic category such as a noun or verb, with lexical descriptive content, and believes that tense, aspect, and mood are not purely functional, but are semantic characteristics which contribute to the well-formedness of the inflected verb. Semantic differences in Turkish verbal complexes, Sezer argues, do not correlate with scope or relative position of the affixes, but rather with the morphosyntactic form, a claim he illustrates with various examples. Sezer proposes that Tense 1 affixes carry complement features such as [+V,-N,-F], while Tense 2 and 3 affixes carry [+V,-N,+F]. The feature F is a characteristic of either a true verb stem [-Functional] or the defective verb stem [+Functional]. Consequently, different tense affixes head different complements, and these differences are accountable by feature checking theory. This is further complicated by the fact that some tense suffixes take different agreement paradigms than others. The reasons for this are not fully understood, but it is widely believed that there are two classes of tense markers, the so- called 'true' tenses and the participial tenses (based on certain behaviors in some predicate expressions). By this account, Sezer motivates the selection of the appropriate agreement paradigm with the features [+/-Finite] and [+/-Nominal]. This works quite well until the appearance of the yes/no question clitic, which interacts with the two paradigms differently. For this Sezer proposes a clitic theory for Turkish in which some principles may be universal; others are necessarily language-specific.

    The second paper, "A note on mood, modality, tense and aspect affix in Turkish" by Guglielmo Cinque of the University of Venice, seeks to analyze the order of Turkish inflectional suffixes within the framework of his (1999) work "Adverbs and Functional Heads". According to Cinque's observations on numerous languages of the world, it appears that the relative order of free and bound grammatical morphemes of mood, modality, tense, aspect and voice are rigidly ordered. Moreover, within each of those categories is apparently a subhierarchy of distinct functional heads, also rigidly ordered, and which correspond to various classes of adverbs in a Spec-Head relation. Unlike Engin Sezer, who argues that (at least in Turkish) tense, aspect and mood have lexical content, Cinque maintains that these are purely functional categories that obey certain universal conditions. In this paper Cinque applies his hypothesis to Turkish, a language which at first glance might seem to provide a number of counterexamples since several of the morphemes can appear in different orders, even two places at once. But upon further analysis, the morphemes clearly have different grammatical characteristics in different configurations, which appear to match the specified mood, modality, tense, aspect and voice distinctions in a particular order, as predicted by Cinque's model.

    The next paper entitled "Periphrastic tense/aspect/mood", by Gerjan van Schaaik of Bogazici University, explores the nature of verbal complexes in which the auxiliary verb 'ol' is present in the string of morphemes. Many inflectional affixes have morphotactic constraints with respect to each other, and often 'ol' is employed as an intervening host for affixes that cannot attach directly to the verb base (e.g. bekliyor olacaklar '[they] will be awaiting'). The various configurations have semantic consequences as well, based on the temporal, aspectual and modal properties of the affixes with respect to 'ol' in periphrastic verbs. In this paper van Schaaik attempts to determine how the expression of tense and aspect in certain periphrastic constructions relates to the theoretical approach taken by Johanson (1994) and Dik (1997), which analyzes compositional forms but does not specifically treat, nor account for, the range of periphrastic types presented in this paper. van Schaaik looks specifically at the most frequently occurring periphrastic constructions, namely those of 'ol' + tense following '-iyor' [PROG], '-ecek' [FUT] and '-er/-mez' [AOR] (affirmative and negative forms). The data was obtained from a corpus of almost two million words scanned from electronic texts in a number of genres. After a thorough discussion of tense, aspect and mood, particularly as they relate to Turkish inflectional forms, van Schaaik then analyzes their expression within a Functional Grammar framework as proposed by Dik (1989) and Hengeveld (1989). The aspectual interpretation of some of these constructions cannot be decomposed, as van Schaaik seeks to demonstrate, and accounts for the periphrastic form as a whole by assuming certain operators to be applied at the level of predication (level 2 of the FG model) in a way that generates the proper suffixes.

    The fourth paper by Eser Erguvanli Taylan, also of Bogazici University, entitled "On the relation between temporal/aspectual adverbs and the verb form in Turkish", is a detailed analysis of the close interplay between Turkish tense and aspect. Since Turkish has no grammatical markers that only code aspect, this is accomplished in various ways by the configuration of tense/aspect/mood morphemes, the contribution of the verb and argument types, and by the inclusion of temporal/aspectual adverbs. Consequently it is difficult to isolate aspect as a distinct category in Turkish. Using Smith's (1997) situation types (activity, accomplishment, achievement, semelfactive and stative) as a launching point, Taylan begins her analysis using the features [+/-durative] and [+/-telic] to determine constraints on various adverbs in combination with tense/aspect inflections in simple clause structures. Among Taylan's suggested findings are: (1) morphology signals an opposition between the imperfective and non-imperfective aspects; (2) context, pragmatic factors, adverbials, and polarity contain the information necessary to specify the intended viewpoint aspect; (3) in some instances the feature [+/-control] was necessary to account for co- occurrence restrictions not handled by the durative/telic features.

    Departing from tense, aspect and mood is "The referential properties of the implicit arguments of impersonal passive constructions" by Mine Nakipoglu-Demiralp of Bogazici University. The class of intransitives that can undergo impersonal passivization (IP) is nearly universal in languages that allow it, and has been applied as a diagnostic to separate unaccusatives from unergatives since the formation of Perlmutter's (1978) Unaccusative Hypothesis. However, some verbs in Turkish do not behave in a predictable way. Nakipoglu-Demiralp seeks to uncover the semantic reasons for their variant behavior and provide a model that accounts for it. The nature of impersonal passives is such that their implicit arguments have different referential properties in different temporal domains. In the past tense, the sole argument of an intransitive takes a 1st person plural interpretation, whereas in the temporally neutral aorist, the sole argument takes an arbitrary interpretation. Therefore, (IP) can only be used as a diagnostic for unaccusative/unergative distinctions in the past tense. Nakipoglu- Demiralp concludes that a verb can only be unergative when a situation is interpreted to be internally instigated or experienced, and proposes a scalar distribution of intransitives on this basis. The result is a five point scale on which (1) is unergative, consistent across languages, (5) is unaccusative, also universal, and at the middle three points are verbs that show mixed behaviors cross-linguistically, but which align themselves somewhere toward one end or the other.

    The sixth paper is by Asli Goksel of Bogazici University & SOAS, University of London, entitled "The auxiliary verb 'ol' at the morphology-syntax interface". Here we revisit the auxiliary verb in one of its apparent functions as a semantically and syntactically inactive morphological buffer stem. As mentioned before, 'ol' can be employed as a host for affixes that cannot attach directly to the verb base. Sometimes it contributes to the interpretation of the clause, and sometimes it appears to be present only to meet well-formedness conditions, namely (1) a match between the suffix types and slots that host them, and (2) the size of a word and restrictions on concatenation of suffixes. In this paper, non-finite object relative clauses (ORCs) are compared and contrasted with matrix clauses to look at the various roles of the auxiliary verb in both structures and the morphological constraints on compound verbs. In ORCs 'ol' clearly behaves differently in conjunction with other elements such as temporal adverbs, negation, and conditionals than it does in matrix clauses. There is convincing evidence that 'ol' does not contribute to the semantic interpretation nor the syntactic structure within ORCs, and it may be the case in some main clauses as well. This seemingly empty morphology and its asymmetrical appearance has possible implications for syntactic theory.

    Next is a paper called "Functional projections and their subjects in Turkish clauses" by Jaklin Kornfilt of Syracuse University. In this work Kornfilt addresses certain non-finite argument and adjunct clauses to account for the variations in the case assignment of the clausal subjects (some subjects are marked with Genitive case while others are not). In Turkic languages subordinate clauses are generally non-finite by virtue of a nominalizing morpheme; however, some of these structures retain verbal properties of tense, aspect and mood. Thus, these types of clauses are thought to have a verbal functional projection which is dominated by a nominal projection. In order to motivate her claim that agreement assigns case to its subject, Kornfilt assumes the existence of a syntactic category 'Agr' as the head of the projection 'AgrP'. She states that in order for this case-assignment potential to be 'unlocked' by the Agr, it must be either theta-governed or co-indexed by a syntactic operator. Thus, only a marked Agr can assign case to its subject in a non-finite clause. Kornfilt next addresses certain adverbial clauses which have an overt subject but no agreement morphology, among which some types have no case marking while others are assigned Genitive case. To explain this, she proposes a default mechanism which assigns a phonologically null case-assignment to subjects of adjuncts with or without Agr morphology. The adjunct clauses whose subjects are co-indexed the syntactic operator get Genitive case; those that are not get the default (Nominative) case.

    The eighth paper in this collection is entitled "On 'small' clauses, other 'bare' verbal complements and feature checking in Turkish", by A. Sumru Ozsoy of Bogazici University. Ozsoy discusses Turkish ECM (exceptional case marking) constructions in which a subject of an embedded clause has Accusative case, yet its predicate bears agreement morphology. The agreement implies that the NP was the subject of the lower clause when Agr was checked. Two possible analyses appear to raise problems for the Minimalist Program: 1) DP raises to SpecAgrOP of higher clause, which violates Shortest Move Principle because it skips over SpecAgrSP, or 2) DP does not skip over SpecAgrSP, but lands there to license Agr on the lower verb, thereby violating Case Licensing (AgrSp assigns Nominative, not Accusative case). Ozsoy compares these constructions with other bare clauses which also have accusatively- marked subjects but do not have agreement on their predicates. These 'small' clauses inherently lack an AgrSP-projection; consequently SpecAgrOP is the closest case checker the DP can land in and no violations occur. Ozsoy proposes one possible explanation for the ECM problem: it is possible that Turkish distinguishes between strong and weak AgrSP. A weak AgrSp whose features are absorbed by a lexical head would not be able to check for case. Ozsoy alternatively suggests that perhaps case is not actually checked in a Spec-Head relation. None of these proposals is satisfactory, she says, but the issues raised by Turkish ECMs must be addressed by any theory of case checking.

    The final paper in the series departs from the analysis of predicate structures and instead focuses on the subject. Entitled "Turkish as a non pro-drop language" by Balkiz Ozturk of Harvard University, the theoretical assumption that Turkish is a pro-drop language is challenged. Others have argued that both the presence and absence of overt personal pronouns is pragmatically conditioned (Enc 1986, Erguvanli-Taylan 1986). Ozturk proposes that these are not subject pronouns, but topic pronouns, and that they are generated in a higher position in the C system. This then raises the question of what is the VP-internal subject. To this Ozturk claims that agreement is not a functional head, but a lexical subject head base-generated in SpecVP. Thus Turkish can be analyzed as a non-pro-drop language. She illustrates with examples from ECM constructions, adjunct clauses, and headless genitive phrases, evidence against agreement as a Spec-Head relation, and argues for removing AgrP from the inflectional domain of Turkish. Agr is reintroduced as a bundle of features assigned to certain heads, as earlier proposed by Kural (1993) and Chomsky (1993). Instead, a TopicP projection is proposed, at whose Spec position an overt pronoun could land.

    DISCUSSION A thorough discussion of each theoretical analysis presented in this collection is not possible in the remaining space of this review. The reader should note that there are many inconsistencies in both the observations and analyses among the authors, which only reinforces the obvious fact that none of these linguistic phenomena is entirely understood or accounted for under a unifying theory. The research presented in this volume offers a broad spectrum of current theoretical viewpoints on a number of interesting morphosyntactic and semantic issues in Turkish. For those who have not worked on this language, the papers provide adequate background from which to become acquainted with its properties and features. I personally commend editor Eser Erguvanli Taylan for the sequencing of the papers. Leading off with Engin Sezer's long and detailed work gives the reader an overview of the Turkish inflectional system, while offering a historical background rich with citations of classic works that are still influencing the research of the language today. From there, the following papers depart notably from Sezer's proposals and theoretical orientation and provide a balanced view of Turkish tense, aspect and mood morphology from several perspectives. Taylan divides the chapters into two parts: the first five deal with the properties of verbal inflection and the rest have implications for syntactic theory. For example, the last three papers raise important questions in terms of agreement as a Spec-Head relation. This is a collection that has been personally helpful to me and I am certain that others will find the research contained therein to be of considerable interest.

    One final comment about typos and inconsistencies in the text, not as criticism (I live in the same glass house): for the benefit of the readers, following the bibliography I have listed some of the more troublesome errata regarding numerical references to examples and end notes. If the publisher or authors would like more information concerning the less problematic errors I personally discovered, they are free to contact me. If it looks like I'm picking on anyone--take heart. It probably means I read your paper more thoroughly.

    REFERENCES - Chomsky, N. 1995. The Minimalist Program. Cambridge, Mass.: MIT Press. - Cinque, G. 1999. Adverbs and Functional Heads. Oxford: Oxford University Press. - Dik, Simon. 1997. The Theory of Functional Grammar. Part I: The Structure of the Clause. - Dik, Simon. 1989. The Theory of Functional Grammar. Part I: The Structure of the Clause. Dordrecht: Foris. - Hengeveld, K. 1988. Layers and operators in Functional Grammar. Journal of Linguistics 25 (1): 127-157 (=WPFG 27 (1988)). - Johanson, L. 1994. Turkeiturkische Aspektotempora. In Tense Systems in European Languages, R. Theiroff and J. Ballweg (eds.), 247-266. Tubingen: Max Niemeyer Verlag. - Kural, M. 1993. V-to(I-to)-C in Turkish. UCLA Occasional Papers in Linguistics, 11, F. Begnelli and M. Kural (eds.), 1-37. - Perlmutter, D. 1978. Impersonal Passives and the Unaccusative Hypothesis. In Proceedings of the Fourth Annual Meeting of the Berkeley Linguistics Society, 157-189. University of California, Berkeley. - Smith, C. 1997. The Parameter of Aspect. Dordrecht: Kluver.

    ERRATA - Sezer, Engin p. 20: Reference to example (29) below, should be (39). - van Schaaik p. 62: Reference to examples: aspectual forms (2)-(5) should be (2)- (6); modal forms (6)-(9) should be (7)-(9). p. 64, below table (10): Explanation of table states those under 5-6 express aspect; should read 3-6. p. 74: There are two examples (21); the first should be (20). p. 75: All in-text references to (29) should read (20). p. 86, Example (37): In gloss, first tense morpheme reads PRES2; should be PRES1 (aorist tense). - Nakipoglu-Demiralp p. 144 Table: In column 3 'doG'(be born) should be in column 4, in accordance with classification in discussion of example (12) on page 135. - Ozsoy p. 221: Reference to example (7) should be (6). p. 221, end of 2nd paragraph: Most appropriate place for end note 14 (which is not there). p. 222, subheading 3.2.3: End note 14 should be 15. p. 222, end of paragraph: End note 15 should be 16 and actually seems to refer to example (21). p. 226: End note 16 should be 17. p. 227: End note 17 should be 18. pp. 233-4, Notes: Note 8 discussion, which refers to (5a-c) and (5a-c), should refer to (6a-d).

    ABOUT THE REVIEWER Lisa Shannon is a recent graduate of Linguistics from the University of Arizona, soon to begin graduate studies at the University of New Mexico. Her primary interests are in typology, syntax, discourse, and L2 acquisition. She works mainly with Italian and Turkish.