Date: Fri, 11 Jun 2004 14:04:22 +0100 From: Gergana Popova Subject: Syntactic Structures and Morphological Information
EDITOR: Junghanns, Uwe; Szucsich, Luka TITLE: Syntactic Structures and Morphological Information SERIES: Interface Explorations 7 PUBLISHER: Mouton de Gruyter YEAR: 2003
Gergana Popova Department of Language and Linguistics, University of Essex
This volume is a collection of ten papers dealing with a range of morphosyntactic phenomena in a variety of languages with rich morphology and within a number of different frameworks. The volume starts with an introduction by the editors, which describes the background of the volume, gives a short overview of the importance of morphosyntactic issues in general and outlines the different approaches to their solutions. The editors also provide a summary of the content of each paper. In what follows, I give my own summary of the papers, and a brief overall discussion.
Tania Avgustinova's paper (''Metagrammar of systematic relations: a study with special reference to Slavic morphosyntax'', pp. 1-24) focuses in general terms on the systematic relations that hold between words in a sentence or in a construction, or between constructions (syntagmatics). The author builds an ontology of syntagmatics and integrates into it traditional notions like subcategorisation, modification, marking, etc. A classification is proposed, which allows to order these and other concepts in a multiple-inheritance hierarchy of the kind often used in the framework of Head Driven Phrase Structure Grammar. Syntagmatics is divided into structural (concerned with linguistic function) and observable (concerned with linguistic form). Structural syntagmatics is classified around the notions of ''centricity'', i.e. the idea that one element in a structure might be dominant over the others (centric structure) or that all elements may have equal status (acentric structure), and ''taxis'' or the idea that one element may be subordinated to another (hypotactic structure), or that no elements may enter into a subordinating relationship (paratactic structure). Combinations of these types are possible (i.e. we can have centric hypotaxis, acentric hypotaxis, centric parataxis, acentric parataxis). Subcategorisation, for example, inherits from the centric hypotaxis type (since this is a multiple inheritance hierarchy, subcategorisation might and does inherit from other types as well).
Similarly, an ontology is proposed for observable syntagmatics, or those relations that are encoded morphologically. At a high level of abstraction observable syntagmatics is subclassified into combinatorial (which would mostly be observable in languages with rich morphology) and alignment (more relevant to configurational languages like English). At the lower levels of the hierarchy more familiar notions are placed (or their equivalents), like agreement, valence, etc. The concepts discussed by the author are exemplified by data from Russian and Bulgarian. The ontology is meant to be universal, i.e. to serve the description of syntagmatic relationships in all languages, allowing to capture similarities and differences between them. Another question addressed is finding a theoretical justification of what relations may in principle hold between two items at the same time. This is also identified as an area for future research. The author points out that (apart from their theoretical importance) one application of the generalisations she makes is to provide an interlingua between various theories of morphology and syntax.
At the centre of Huba Bartos's article (''On-line morphology: The morphosyntax of Hungarian verbal inflection'', pp. 25-56) are scope effects in Hungarian, i.e. cases where affixes with an invariant order can give rise to (at least two) interpretations where the affixes scope differently over each other. The author makes three theoretical assumptions: (i) scope depends on c-command, so that if X c-commands Y, then X outscopes Y; (ii) within a language the order of functional projections is invariant; (iii) the Mirror Principle, proposed by Baker (1985) holds, i.e. there is an isomorphy from syntactic derivation to morphological derivation. What the author wants to achieve is an account that will be faithful to these assumptions, but also to the data (verb inflection for tense, mode and modality and, less centrally, the so called ''roll-up'' verb clusters in Hungarian) without appealing to covert movement and without resorting to a purely semantic account. Bartos assumes a model of morphosyntax which combines a derivational, minimalist syntax (Chomsky 1995, 1999, 2001) with a late-insertion model (Distributed Morphology-Halle and Marantz 1993, Marantz 1997). Although he assumes that syntax and morphology are separate modules with their own properties, they are conceived as working together, in a syntax-makes-a-step, then-morphology-makes-a-step fashion. Morphology tries to mirror syntax, but not vice versa (in this respect Bartos modifies the Mirror Principle). A morphosyntactic merger operation replaces similar operations in Distributed Morphology and Minimalism. In essence, morphosyntactic merger puts two morphological structurally adjacent entities together (a stem and an affix) and then deletes the structure, so that it is no longer available at later stages of the derivation. If there is no affix to instantiate a particular functional projection, then this projection is void of feature content and can be featurally fleshed out by other (lower) verbal inflectional properties. For example, if in the following structure
[MP [M TP [T ModP [Mod VP]]]]
Mood (M) is not instantiated by an affix, it can be fleshed out by Modality (Mod) by substitution. If this happens, Mod will c-command Tense (T), so a scopal interpretation will be available where modality outscopes tense. At the stage this substitution happens, however, in the morphological derivation the modality affix and subsequently the tense affix have already been added to the root. The structure has been deleted, so morphology cannot follow the syntactic substitution. Therefore, the different scope cannot be mirrored in morphology. If the substitution does not happen, and modality is featurally fleshed out in situ, then the scopal relations (tense outscopes modality) reflect the morphological order. This account explains why different scopal relations are not possible in structures where all three affixes (modality, tense and mood) are present.
The paper by Miriam Butt and Louisa Sadler (''Verbal morphology and agreement in Urdu'', pp. 57-100) builds an account of case and agreement phenomena in Urdu within the framework of LFG, i.e. within a lexicalist approach to morphosyntax (whereby syntax and morphology are different modules of grammar). The interface between morphology and syntax is seen as consisting of the exchange of syntactic functional information between them. The authors assume and defend a non-morpheme based model of morphology, where morphemes don't have lexical entries with form- meaning correspondences. In Urdu there seems to be a correspondence between case marking and agreement: the verb agrees with the subject or with the object, depending on which one is not marked for case. If both happen to be marked, the verb appears in the 3rd person singular masculine form. Butt and Sadler discuss the relation between case and agreement more broadly for South Asian languages and reach the conclusion that the larger empirical basis speaks against a correlation of case and agreement. The independence of the two phenomena is supported also by data from pro-drop. The authors show that case in Urdu (a non-configurational language) is not a straightforward realization of features, but rather case markers themselves determine the functional status of the nominals which they mark and contribute semantic information to the clause level. In this sense, case in Urdu is constructional (on the understanding of this term presented in (Nordlinger 1998)). Agreement in Urdu is modeled as a lexical constraint incorporated in the lexical entry (within f-structure) of the verb. The verbal lexical entry directly refers to functional information associated with the nominal, which allows accounting for the preference of Urdu verbs to agree with unmarked subjects/objects. Butt and Sadler address a number of problems a morpheme- based approach might run into given the data discussed in the paper and argue for a separation of morphology and syntax. They show that the data can adequately be described in a computational environment with a finite- state morphology model. Although the phonological spell-out remains opaque, the functional information output by the morphology is accessible to the syntax.
Gisella Ferraresi and Maria Goldbach (''Particles and sentence structure: a historical perspective'', pp. 101-127) see the interaction between syntax and semantics and/or phonology as a driving force behind language change (following methodological proposals in (Keenan 2001) and (Longobardi 2001)). In particular they discuss the gradual grammaticalisation of the Old French particle ''si'' which disappeared before it became an affix. In Old French the particle ''si'' is partially grammaticalised. It can function as a phonological host for object clitics and adverbial clitics, but can also be cliticised to another element. It can be phonologically independent and can even be an a sentence initial position, however its behaviour in poetry suggests that ''si'' is phonologically and metrically weak. The authors suggests that ''si'' heads its own phrase and present a syntactic generalisation that will account for its peculiar distribution, based on the split CP notion of (Rizzi 1997). The authors suggest that certain phonological changes (weakening of the first syllable of the clitic group) and the semantic bleaching of ''si'' led to the disappearance of this sentence particle in the 17th century. This loss is seen as related to the loss of V2 effects in Old French, as the functional projection headed by ''si'' was no longer recognised as a separate node in syntax by language learners.
Jaklin Kornfilt (''Subject case in Turkish nominalised clauses'', pp. 129-215) discusses case marking on the subject in Turkish syntactic nominalisations, defined as 'extended clausal projection[s] with some nominal functional layers that represent ''nominalisation''' (p. 139). There are two types of such nominalised clauses: factive/indicative (they have tense and some verbal properties, are CPs, and can act as embedded questions or relative clauses) and non-factive/subjunctive (they have no tense and can sometimes be pluralised or occur with demonstratives, and are homogeneously DPs and so cannot act as embedded questions or relative clauses). Kornfilt couches the analysis in the Principles and Parameters framework. The subject in Turkish nominalised clauses can carry a genuine subject case or default case. The genuine subject case is licensed by overt agreement morphology on the verb. This agreement morphology has to be licensed itself. This is the case if there is no categorial mismatch (i.e. nominal agreement is in a nominal extended projection, or verbal agreement is in a verbal extended projection). If there is a categorial mismatch, the agreement morphology can be further licensed by a primary thematic index or by a predicational index. In this respect the clauses that are in an argument position (and thus receive a primary thematic index) will behave differently from clauses that are in an adjunct position. Default case on the (overt) subject is the non-licensed case. The nature of the default case and constraints over it are discussed. Korfilt also looks at (overt) subjects which are not marked for case at all. The author reviews a wealth of data and presents a number of related analyses, both of Turkish and of similar phenomena in other languages.
Esther Rinke (''On the licensing of null subjects in Old French'', pp. 217-247) looks at the conditions for licensing null subjects in Old French. The author starts from the empirical observation that null subjects in Old French are restricted to certain structural environments: main clauses with an initial non-subject constituent and conjugational subordinate clauses which contain a preverbal topic (p. 217). The distribution of omitted referential subjects is explained on the one hand along the lines of Kato (1999), i.e. with the availability of a [+pronominal] agreement system. Following Kato (op cit.), Rinke assumes that in null subject languages the agreement morphemes have the grammatical status of a pronominal subject. This explanation, however, is not restrictive enough for the Old French data, so in addition the analysis utilises Rizzi's (1997) split CP system with different functional layers (ForceP, TopP, FocP and FinP. Rinke suggests that the licesing of null subjects depends on the realisation of Fin. This category is available in precisely the contexts where null subjects are possible.
Andrew Spencer (''Periphrastic paradigms in Bulgarian'' pp. 249-282) focuses on the paradigmatic organisation of periphrastic constructions like the English progressive ('is writing') or passive ('was written'). The author argues for a non-morpheme, construction based approach to morphology, within which the notion of paradigm has a central place. Periphrastic constructions are in paradigmatic opposition to inflected forms and therefore occupy cells in the paradigms of lexemes, just like inflected forms of lexemes. Although they serve as exponents of some morphosyntactic features, the individual items which enter into the construction are not themselves bearers of these features (thus neither 'was' not 'written' contributes the meaning 'passive'). Paradigms with periphrastic constructions have the same properties as other paradigms: they can be exhaustive (all and only the expected feature-value combinations are realised), underexhaustive (morphologically unmotivated gaps can be found), or superexhaustive (containing more forms that can be expected under the usual syntactic principles of compositionality), they exhibit syncretism, cumulation, zero exponence, etc. Properties of morphological paradigms in general and periphrastic paradigms in particular are demonstrated through data from English, Russian, Chukotko- Kamchatkan languages, but primarily through a detailed analysis and formalisation of the (Emphatic) Renarrated and Conclusive forms in Bulgarian. An extreme example of the central tenet of the article is a Bulgarian clausal construction which can be used as a partial exponent of morphosyntactic features. One of the central assumptions of the analysis is the separation of morphological and syntactic features, with the possibility of complex mappings between them.
Barbara Stiebels's empirical preoccupation in the article entitled ''Transparent, restricted and opaque affix orders'' (pp. 283-315) is the possibility for some affixes to have a fixed and in other cases variable order, which affix order may or may not correlate with the scopal interpretation of the meaning of the affixes. The author's goal is to ''provide a programmatic and semantically based overview of possible affix orders within the domain of diathesis morphology: which diathesis markers may be combined in principle and to which extent is the resulting morphological structure compositional...'' (p. 286). More specifically, the author aims to show that the Mirror Principle, as formulated in Baker (1985:375), does not account for the attested data. The semantic representation of the argument structure of verbs follows proposals within Lexical Decomposition Grammar (Joppen and Wunderlich 1995, Wunderlich 1997b, Stiebels 1999). Stiebels reformulates the Mirror Principle in the following way: ''The affix order must mirror semantic composition'' (p. 291), but assumes that this principle may be violated by higher ranking constraints (for example language specific constraints on linking) and demonstrates how this might work on a wealth of data from languages like Chichewa, Quechua, Kinyarwanda, etc.
Jochen Trommer's article is entitled ''Direction marking as agreement'' (pp. 317-339). The Algonquian language Menominee 1st/2nd person subjects and 3rd person objects are marked by the suffix -a, whereas 3rd person subjects and 1st/2nd person objects are marked by -eko. Traditionally this so called 'direction marking' has been explained with reference to animacy hierarchies.The author proposes to account for these and related data in terms of agreement, governed by universal markedness constraints (where constraints are violable and ranked, in the sense of Optimality Theory). The formal framework within which the analysis is formulated is Distributed Optimality (Trommer 2001). Syntactic derivations result in bundles of morphosyntactic features. Vocabulary items then associate a subset of these features with some form, i.e. they spell out the features (and are not allowed to insert features). Morphological constraints can refer only to small, word- like units, not to syntactic phrases. The so called PARSE constraints require that certain feature combinations be spelled out in the output form. PARSE constraints can have the following form: PARSE [P] A/B, which means that if a syntactic head contains A next to a B, then the spell out of A is to be preferred over B. There are also BLOCK constraints, which determine the number of affixes that are allowed to realise the same feature, and IMPOVERISH constraints which limit the spell- out of certain features or combinations thereof via affixes. These are universal but violable constraints, some of which are linked to feature hierarchies. A number of such constraints is proposed and ranked in different ways. The claim of the author is that the different rankings can correctly predict the different grammars of languages like Turkana, Dumi, Menominee, Quechua.
Ilse Zimmermann (''On the semantics of cases'', pp. 341-380) aims to identify a suitable theory for the semantic representation and licensing of cases. The data for the study are taken from Modern Russian. Structural cases are modeled via abstract semantico-syntactic features (+/- higher role /hr/ and +/- lower role (lr)) related to the hierarchy of arguments, which are mapped to the morphosyntactic realisation of case markings on complements. The case of adjuncts is discussed with reference to the instrumental case. The morphosyntactic case features and the respective case markings on adjuncts are mediated by semantic templates. To make sure that the right case markings surface on the arguments of verbs and nouns, Stiebels uses a number of ranked constraints (within Optimality and Correspondence Theory). The author also adopts a minimalist framework of sound-meaning correlation (Chomsky 1995), lexicalist conception of morphology (Stiebels and Wunderlich 1994; Wunderlich and Fabri 1995; Wunderlich 1997a?), and the differentiation between Semantic Form and Conceptual Structure of two- level semantics (Bierwisch 1983, 1987, 1997; Bierwisch and Schreuder 1992; and others).
DISCUSSION This collection of papers is a rich source of case studies for those interested in the interface between syntax, semantics and morphology. They illustrate different approaches to a number of fundamental morphological (and morphosyntactic) debates: are morphemes signs, in other words do they have lexical entries where form is correlated with meaning; is morphology an independent module, or is it essentially part of syntactic operations; is the functional information used by morphology the same as that used by syntax, or can it be different. Some of the papers contain an explicit formulation of their position with respect to these issues (for example Butt and Sadler; Spencer; Trommer), in others this is more implicit.
In all cases the papers illustrate a detailed analysis of a subset of morphosyntactic phenomena within the linguistic framework chosen. The wide range of approaches in itself presents the reader with a reward and a challenge. The reward is obvious. The challenge is in being able to grasp the workings of each of the frameworks. The papers vary in the extent to which they accommodate the reader who is not familiar with the adopted framework: some provide introductions for the uninitiated, some don't.
Inevitably, there is some relation between the data each paper discusses and the framework the authors choose to adopt. For example the treatment of scopal ambiguities is less obvious in a lexicalist, non-morpheme based framework. So readers might find themselves wondering how the data in one paper can be accounted for under the approach of another. This in itself is a positive outcome of reading this volume.
In sum, this collection of papers is a useful resource for anyone working or interested in the interface between morphology and syntax.
REFERENCES Baker, Mark (1985) The Mirror principle and morphosyntactic explanation. Linguistic Inquiry 16: 373-415.
Bierwisch, Manfred (1983) Semantische und konzeptuelle Repraesentation lexikalischer Einheiten. In: Ruzicka, Rudolf and Wolfgang Motsch (eds.) Untersuchungen zur Semantik. (Studia grammatica 22.) Berlin: Akademie- Verlag, pp. 61-99.
Bierwisch, Manfred (1987) Semantik der Garduierung. In: Bierwisch, Manfred and Ewald Lang (eds.) Grammatische und konzeptuelle Aspekte von Dimensionsadjektiven. (Studia grammatica 26/27.) Berlin: Akademie- Verlag, pp. 91-286.
Bierwisch, Manfred (1997) Lexical Information from a minimalist point of view. In: Wilder, Chris, Hans-Martin Gaertner and Manfred Bierwisch (eds.) The Role of Economy Principles in Linguistic Theory. (Studia grammatica 40.) Berlin: Akademie-Verlag, pp. 227-266.
Bierwisch, Manfred and Robert Schreuder (1992) From concepts to lexical items. Cognition 42: 23-60.
Chomsky, Noam (1995) The Minimalist Program. Cambridge, MA: The MIT Press.
Chomsky, Noam (1999) Derivation by Phase. Ms., MIT.
Halle, Morris and Alec Marantz (1993) Distributed Morphology and the Pieces of Inflection. In: Samuel Jay Keyser and Ken Hale (eds.) The View from the Building 20. Cambridge MA: The MIT Press, pp. 111-176.
Joppen, Sandra and Dieter Wunderlich (1995) Argument Linking in Basque. Lingua 97: 123-169.
Kato, Mari (1999) Strong and weak pronominals in the null subject parameter. Probus 11: 1-37.
Keenan, Edward L. (2001) Explaining the creation of reflexive pronouns in English. Ms., UCLA.
Longobardi, Giuseppe (2001) Formal syntax, diachronic minimalism, and etymology: the history of French 'chez'. Linguistic Inquiry 32.2: 275- 302.
Marantz, Alec (1997) No escape from syntax: Don't try morphological analysis in the privacy of your own lexicon. Upenn Working Papers in Linguistics 4.2: 201-226.
Rizzi, Luigi (1997) The fine structure of the left periphery. In: Liliane Haegeman (ed.) Elements of Grammar. Handbook of Generative Syntax. Dordrecht: Kluwer, pp. 281- 337.
Stiebels, Barbara (1999) Noun-verb symmetries in Nahuatl nominalisations. Natural Language and Linguistic Theory 17: 783-836.
Stiebels, Barbara and Dieter Wunderlich (1994) Morphology feeds syntax: The case of particle verbs. Linguistics: 32:913-968.
Trommer, Jochen (2001) Distributed Optimality. PhD Dissertation, University of Potsdam.
Wunderlich, Dieter (1997) CAUSE and the structure of verbs. Linguistic Inquiry 28: 27-68.
Wunderlich, Dieter and Ray Fabri (1995) Minimalist moprhology: An approach to inflection. Zeitschrift fuer Sprachwissenschaft 14.2: 236- 294.
ABOUT THE REVIEWER:
ABOUT THE REVIEWER Gergana Popova is a PhD student at the Department of Language and Linguistics, University of Essex, UK. She is currently researching aspect and Aktionsarten in Bulgarian. Her research interests are in morphology and lexical semantics.