This book presents a new theory of grammatical categories - the Universal Spine Hypothesis - and reinforces generative notions of Universal Grammar while accommodating insights from linguistic typology.
EDITORS: Erteschick-Shir, Nomi ; Rapoport, Tova TITLE: The Syntax of Aspect SUBTITLE: Deriving Thematic and Aspectual Interpretation SERIES: Oxford Studies in Theoretical Linguistics PUBLISHER:Oxford University Press YEAR: 2005
Brenda Laca, Sciences du Langage, Université Paris 8 and CNRS-UMR 7023
This carefully edited volume contains 13 contributions distributed in three parts (I: From Lexical Roots to Syntax, II: Event Structure and Feature Projections, and III: Lexical Restrictions on Syntax), preceded by a short preface, biographical notes on the contributors, an overview containing the abstracts of the contributions, and an introduction by the editors. The volume is completed by a single list of references, and two very useful indexes of names and terms.
As indicated by the sub-title, the topic of the volume is not ''outer'' aspect, but what used to be called Aktionsart and its relation to thematic structure and argument realization. The contributions, most of them by leading scholars in the field, explore the morphology-syntax-semantics articulation in the VP-domain, in which lexicalist and constructionist approaches have alternated for decades, and give quite a balanced picture of the current state of the discussion.
In 'Aspect and the Syntax of Argument Structure', Ken Hale and Samuel Jay Kayser further develop their influential approach to l-syntax, in which argument structure is held to be determined by the lexical syntactic configuration a lexical item appears in. Lexical verbs are syntactic configurations containing a root (R) and categorial hosts, as for instance V or P. The relation between the host and the root is one of complementation. Roots may require or preclude a specifier for the [V+R] projection, thus accounting for the presence or absence of an ''internal argument''. Transitivization and the middle construction are assumed to depend on dyadic projections containing both a complement and a specifier. The original incorporation analysis of denominal verbs is rejected in favor of an analysis by conflation/selection: a verb like 'dance' does not incorporate the element 'dance' as a complement to V, but it rather selects for a class of complements (the hyponyms of 'dance') which is restricted enough to license an empty category in the complement position. An important contrast, somewhat opaquely labelled the proximate vs. obviative contrast, is shown to account for the possibility of intransitive alternants and of the middle construction. Some verbs (like 'splash', 'dent', 'anger') contain a ''manner feature'' linked to the internal argument, in others (like 'smear', 'kick', 'love') this ''manner feature'' is linked to the external argument. As a consequence, constructions that eliminate the external argument are impossible with the latter. The authors explore in detail the issue of stativity and its relationship to categories and to argument structure, and provide analyses for psychological verbs which can be stative or eventive, such as 'fear', 'love', 'respect' and for verbs taking measure predicates as ''complements'', such as 'cost', 'weigh', which are analysed as copulas. Their conclusion is that argument structure is widely independent from aspectual type, but that aspectual type, in particular stativity, may be associated with the particular type of structures expressing central - as opposed to terminal – coincidence.
Heidi Harley applies the l-syntactic approach to the analysis of the different Aktionsart properties of denominal verbs. The l-syntax approach allows her to search for parallels between the factors determining the Aktionsart of VP, such as the presence of a bounded Incremental Theme argument, and the factors determining the Aktionsart of denominal verbs, which in her analysis may result from incorporation to v of the measuring-out argument held to be responsible for telicity. She shows that the telicity of 'foal', by contrast to the atelicity of 'sweat', derives from the bounded/non-bounded status of the nominal roots which are the complement of v in l-syntax. Verbs with event-denoting roots, as 'dance' or 'hop', are assumed to differ from verbs with thing-denoting roots, because in their case, no accomplishment, but only activity or semelfactive interpretations are available for the vP. The semantic conjectures explaining why this should be so (that things take up space, whereas events can be pointlike or arbitrarily extended, and that accomplishments necessarily comprise two events) could bear some elaboration. Harley correctly points to the shortcomings of the Affectedness Condition as a criterion for the existence of an Incremental Theme, and proposes instead that the lack of accomplishment readings for verbs like 'kick' or 'push' derives from the fact that the direct object is not a complement of v, but a complement of the root. Since only the root, i.e. the sister to v, can determine the event-object homomorphism, an object that checks Accusative Case but is not a sister to v, like the object of 'push the cart' or 'kick the wall' does not influence Aktionsart properties. Change-of-state verbs, which are usually deadjectival, are analysed as embedding a small clause whose predicate is the adjectival root. It is not the nominal argument (the subject of the small clause in l-syntax) that determines Aktionsart, but the Small Clause itself, in as far as ''the degree to which some state is true of the Theme'' is at stake. The nominal argument can, however, indirectly contribute to (a)-telicity by determining the boundedness or unboundedness of the Small Clause itself. Again, the semantic argumentation could and should have been made more precise. Telicity contrasts in location/locatum verbs are shown to depend both from the nature of the locatum (with saddle giving rise to boundedness and paint giving rise to unboundedness) and from the nature of the prepositional object. By contrast, boundeness of the nominal root does not play any role in verbs like 'hammer' or 'brush'. This is accounted for by a mechanism of Manner Incorporation, by which the noun surfacing as a verb is incorporated from an adjunct position expressing manner, and it is speculated that telic interpretations of manner of movement verbs, as well as verbs of creation, consumption and destruction, also rely on this mechanim.
The approach Nomi Erteschik-Shir and Tova Rapoport apply in their treatment of 'Path predicates', though inspired by the l-syntax theory, differs from it in the fact that it assumes that a verbal lexical entry contains a verbal head and one or more meaning components with a categorial specification, such as MANNER, STATE or LOCATION, determining its aspectual class and the theta-role of its subject. Meaning components, which form a very small universal set, might project or not, but they are always interpreted. Path predicates, a superclass comprising verbs of gradual change in state or location ('cool', 'advance'), of manner of progression ('march'), and iteratives ('flash', 'beep'), share two properties: the possibility of atelicity and the possibility of transitivization. What accounts for the behavior of the class is, according to the authors, that their elements describe ''a sequential (incremental or iterated) change'', and thus instantiate verbal plurality. This analysis is a very interesting contribution to a topic which has attracted considerable attention from the semantics side. However, equating event plurality with path structures, as the authors do, does not seem particularly convincing in the case of iterative, ''on-off'' readings. The same holds for the conjecture that free transitivization depends from the plural status of the predicate, inasmuch as the latter licenses an interpretation of the extra (causative) subject as a ''controller of the extent of the path''. In 'Tense, Person, and Transitivity', Jacqueline Guéron develops her hypothesis of a correlation between syntactic domains (phases) and semantic interpretation, assuming a spatial construal in VP/vP and a temporal construal for TP/CP, which is argued for on the basis of an array of very suggestive case analyses. A single feature,[ +/-EXT(ended)], determining if something has an extension (in space or time) or is point-like, is operative in both phases. Its value at VP-level decides both aspectual type and argument structure. For instance, a [–EXT]- VP does not select for a subject in the specifier of VP, and the presence or absence of this argument is responsible for auxiliary selection in French (with 'avoir' associated to the presence of an external argument and 'être' with its absence). A far-reaching consequence of the analysis is that holistic theta-roles, such as Agent, are replaced by different elementary functions at VP-level (Manipulator) and at TP-level (Trigger, T-controller, Experiencer). This accounts for the differences between the middle construction in English (which requires a Manipulator) and in French (which requires a Trigger or a T-controller). The article also offers a reinterpretation of the Vendlerian classes in terms of spatial Figure-Ground configuration, and splits the domain of transitivity in spatial transitivity (transitivity in the vP domain) and temporal/personal transitivity (transitivity in the TP/CP domain), a difference which is exploited in the analysis of possessive datives and of reflexive structures.
In 'Complex Aspectual Structure in Hindi/Urdu', Miriam Butt and Gillian Ramchand propose a decompositional analysis of light-verb constructions in terms of event structure. After showing how they differ from bona fide auxiliary constructions, they distinguish three morphosyntactic and semantic types which are related to the possibilities of Merging verbs at v- or V-level or as part of a Result Phrase. Whereas auxiliary constructions are clearly monoclausal, light verbs constructions are heterogeneous in this respect, with complex predicates of the 'result' and of the 'let'-type behaving as a single predicational unit with regard to anaphora, control, and agreement phenomena, and one further type exhibiting clearly biclausal behavior. Complex predicate constructions thus occupy an intermediate position between auxiliary and complementation structures: they comprise two lexical verbs which nonetheless build a single event description. Two binary relations between events, which yield single complex events, offer the semantic basis for complex predicates. The first is causation, and the second is telic augmentation. The potential threefold sub-event structure of a complex event can be mapped onto three syntactic projections, which may license each its own subject. The causing event corresponds to vP, the process in question corresponds to VP, and the result state corresponds to a Result Phrase. While complex event structures may be expressed by a single lexical item, light verbs permit overt instantiation of little v (the causing event leading to the event described by the main verb), or of V (the process leading to the result described by the main verb). This article offers very valuable insights into the notoriously difficult problem of ''periphrastic'' verbal constructions, and an unexceptionable methodology for their analysis, which can not only prove fruitful for other constructions in other languages, but also decisivily further our understanding of grammaticization processes in this domain.
'The Aspect of Agency' belongs, according to Edith Doron, to non-temporal event structure, which is what Semitic templatic verbal morphology mainly encodes. The contrasts between three templates in the active voice (the simple, the intensive and the causative one) show that they express finer-grained distinctions in the domain of Agency: intensive verbs either reclassify an external argument as an Actor or add an external argument to unaccusative predicates, whereas causative verbs invariably add a new external argument, the Cause. Actors presuppose animacy, whereas Causes do not. They are captured in the syntax by means of distinct agency heads, the first acting as a modifier of the root, and the second merging with a fully constructed verb to introduce a further argument. The two different case patterns found in the causativization of transitives, in which the Causee can appear either as an oblique or as an accusative object, are shown to depend on the type of subject of the corresponding simple verbs, with locative/experiencer subjects surfacing as obliques and affected subjects of psych-verbs surfacing as accusatives.
The question of the structural difference between Agents and Causes is taken up by Lisa Travis in 'Agents and Causes in Malagasy and Tagalog'. She assumes that Causes are realized lower than Agents, in an Asp-projection she locates below vP. The analysis is based on the hypothesis that certain heads are deleted – or have zero expression – under certain conditions inTagalog: for instance, causative pag- is not overtly realized when its Specifier position is filled, with the result that causatives from intransitives and causatives from lexical causatives have the same form. Telicity is assumed to be directly linked to agency. Telicity markers either add a Cause argument to unaccusative or adjectival predicates, or turn the Agent of a transitive structure into a Cause. The patterns of morpheme deletion lead to distinguish two types of structures for psych-verbs, with the Cause corresponding to the object of an emotion verb or to the Experiencer of a perception/cognition verb.
Carlota Smith discusses the respective adequacy of semantically based versus surface-structure interpreting approaches in 'Event Structure and Morphosyntax in Navajo'. She describes the Navajo verb word as comprising three levels: the verb theme, formed by the root/stem with classifiers and some thematic elements, the verb base, selecting one stem out of a set of them and combining it with prefixes conveying adverbial and thematic concepts, and finally the verb word, augmented by pronominal and conjugational prefixes. Linear order of the prefixes, however, follows templates which do not correspond to hierarchical ordering. Information conveyed by a verb is assumed to be universally organized in tiers comprising event structure (temporal structure or situation type), qualia structure, and argument structure. Morphology affecting event structure may correspond in Navajo to different Verb Lexeme Categories of a single root, to certain ''sub-aspectual'' prefixes, or to certain Modes. Combinations of prefixes do not seem to show consistent scopal direction, since both left-to-right and right-to-left interpretations seem possible. This poses a problem for l-syntax- inspired analyses of the Navajo verb word, since neither the hypothesized underlying structure nor the surface structure are able to account for semantic scope. By contrast, it is argued that generating the surface structure by simple phrase structure rules and head-movement mechanisms, and interpreting it by means of DRT-rules can give a more satisfactory account.
In 'Constructions, Lexical Semantics, and the Correspondence Principle: Accounting for Generalizations and Subregularities in the Realization of Arguments', Adele Goldberg applies a construction grammar approach to the analysis of different types of counterexamples to the principle according to which sub-events in an event-structure template must be identified by an overt phrasal argument (Argument Realization Principle). These counterexamples involve unexpressed Themes with verbs of change of location and lack of the Patient with causative verbs. They are analysed as the result of de-emphasizing a theme or a patient in a particular context. The Implicit Theme Construction appears with verbs of emission, ingestion, and contribution, and permits to recover the identity of the theme by inference. The Deprofiled Object Construction permits the omission of arguments that are non-focal and non-topical. Since it is difficult to come up with tight, independent criteria for the factors that play a central explanatory role (emphasis, importance, ''heaviness'', etc), the account does not seem free of circularity.
The omission or optionality of direct objects is treated also by Anita Mittwoch in 'Unspecified Objects in Episodic and Habitual Sentences'. In episodic sentences, objects can be omitted with verbs specifiying a manner component, and having fairly circumscribed selection restrictions. In habitual sentences, a much larger range of verbs permit object omission, and the objects are interpreted as a bare plural object would be. Pluractionality is assumed to play a role in the licensing of object omission, and missing objects in habitual sentences are shown to be typically backgrounded and thus to belong to the restriction of the generic sentence. For the analysis of the latter cases, a phonological null pro-NP bearing only the features +PLURAL and ± HUMAN is suggested.
In 'Resultatives Under the 'Event-Argument Homomorphism Model of Telicity', Stephen Wechsler develops the hypothesis that resultatives involve an abstract 'path' argument corresponding to a degree scale. The analysis addresses three problems arising in the resultative construction: lexical variation constraining which secondary predicates are acceptable with which verbs, the distribution of adjectival and prepositional resultatives, and the apparent need for a deep direct object. Instead of assuming, as usual, that the resultative predicate accounts for the telicity of the construction by supplying a result state, the author hypothesizes that the resultative expresses a property scale along which the affected object changes by degrees. Closed-scale adjectives (those having end-points), as well as non-gradable adjectives, provide suitable bounds for events and are clearly preferred in the resultative construction. Open-scale adjectives, however, occasionally appear in ECM-resultatives, in which the subject of the resultative is not an argument of the main verb and in which the ''fusion'' between the two sub-events is less tight.
In 'Change-of-State Verbs: Implications for Theories of Argument Projection', Malka Rappaport Hovav and Beth Levin explore argument alternations against the background of two hypotheses: that argument projection is aspectually driven, and that argument expression is not lexically determined. They argue against the extreme versions of both hypotheses, and their conjunction, by showing that the relevant facets of verb meaning that are responsible for the uniformity of argument expression with the class of Change-of-State verbs do not correlate with well-known aspectual notions such as telicity or punctuality. On the other hand, verbs that do share a common aspectual characteristic, the presence of an incremental theme, widely differ in their argument expression possibilities. Traditional aspectual classes and classes distinguished by argument expression possibilities do not coincide.
The volume will be of great service to scholars working in the domains of verb syntax, verb classification and thematic structure, and can constitute a welcome basis for graduate courses in these domains. It not only offers a representative selection of the work currently being carried out in the dominant approaches in the field, but it is also rich in detailed analyses and subtle empirical findings that could inspire further research. The thematic unity of the volume is remarkable. Several subjects are treated from different perspectives in the contributions (such as the problem of covert or ommitted objects, the possibility of more fine-grained analyses for traditional thematic roles, the basis for telicity/atelicity alternations, the more than plausible actual independence of temporal event structure (aspect) from argument structure), and this is adequately acknowledged by the numerous cross-references in the volume.
Since the volume is representative of the state of the art in the field, the main shortcomings that can be pointed out do not concern the book in particular, but rather the state of the field itself. The accrued integration of semantic considerations in syntax has undoubtedly been fruitful. But the danger pertaining to such an integration is that it might go hand in hand with a slackening of the standards of stringent argumentation in both disciplines. This danger is overcome to varying degrees in the different contributions.
ABOUT THE REVIEWER:
ABOUT THE REVIEWER
The reviewer has worked in the fields of derivational morphology , on the semantics of determiners, and on aspect and tense, in particular in the Romance languages. She has published several articles on aspectual periphrases, eventuality modification, and time-relational aspect, and is currently working on a book on the grammar of aspect in Romance. She has coordinated the TP-/VP- subproject of the typological research program "Sentence architecture", in collaboration with P. Cabredo-Hofherr (Fédération TUL – CNRS FRE 2559, 2002-2005), with whom she is currently coordinating the program "Distributive dependencies: nominal and verbal plurality" (Fédération TUL – CNRS FRE 2559, 2006-2008).