Showing papers in "Natural Language and Linguistic Theory in 1993"

Alternatives to the iambic-trochaic law

Abstract: This article seeks to develop alternatives to recent theories (Hayes 1985, 1987, 1991; McCarthy and Prince 1986, 1990; Prince 1991) that explain quantitative asymmetries between iambic and trochaic systems on the basis of a rhythmiciambic-trochaic law. The theory proposed here derives such asymmetries from two different rhythmic factors: moraic prominence-relations internal to heavy syllables, and avoidance of clash and lapse in sequences of moras. Firstly, it argues for a distinction between parsing feet and surface feet. Parsing feet draw from a symmetric foot inventory, based on two parameters:stressable element (mora vs. syllable), andheadedness (iambic vs. trochaic). That is, the basic foot inventory is no longer governed by the iambic-trochaic law. Secondly, parsing feet are mapped into surface feet by rules that impose quantitative changes, or by stray adjunction, induced by strict prosodic layering. Thirdly, a rhythmic sub-theory defines filters that rule out clashes and lapses in sequences of moras or syllables, depending on the type of stressable element. It explains iambic-trochaic asymmetries with respect to lengthening and shortening, and an asymmetry with respect to directionality of iambic parsing. Evidence will be presented from Tubatulabal, Yidiny, Araucanian, Chugach Alutiiq Yupik, Cayuga, Latin, and English.

Movement of Wh-phrases in Japanese

Abstract: In this paper, I argue that a certain case of scrambling of Wh-phrases counts as syntactic Wh-movement in Japanese: Specifically, I argue that long-distance scrambling of a Wh-phrase to a clause headed by a [+ WH] COMP behaves exactly like Wh-movement. I defend this hypothesis by showing (i) that such movement lacks LF ‘undoing’ effects, a quite surprising fact considering that scrambling can be freely undone in LF, and (ii) that it exhibits the Superiority effects. This observation provides support for Mahajan's (1990) proposal that long-distance scrambling, unlike local scrambling, is uniformly A′-movement. Further, I show, on the basis of LF undoing effects, that multiple Wh-movement is possible in Japanese, and argue that this fact supports Kuroda's (1988) ‘multiple SPEC’ analysis for the phrase structure of Japanese.

Subject-extraction, negation and the antiagreement effect

Abstract: This paper deals with the relationship between subject agreement and extracted subjects. In some languages (local) extraction of the subject triggers the Anti-Agreement Effect (AAE), whereby the verb cannot agree with the extracted subject; instead, the verb has an invariable (third person singular) form. It is argued that the AAE is a strategy used by some Null Subject languages, in particular those which locally move theirwh-subjects in the syntax, to avoid the licensing of a resumptive pro in the closest subject position. This strategy is necessary because a resumptive pro in this position would be accessible to (A′-) binding by the movedwh-subject, in violation of an A′-disjointness requirement on the distribution of pronomonal elements. It is argued, following Aoun and Li (1989), that the latter incorporates a Minimality effect, necessary to account for the fact that the presence of negation helps undo the AAE, obligatorily in some languages and optionally in others.

Wh-movement and specificity

Abstract: This paper claims that the possibilities of movement of Wh-operators, as well as the scope possibilities of Wh-operators in situ, are constrained by a filter on interpretation requiring that an operator Op i having scope over Op j and binding a variable in the scope of Op j be specific. The operation of the filter is demonstrated in the case of extraction out of a Wh-island, in multiple questions, and in the case of nonWh-operators, as well. The analysis of the role of specificity in multiple questions is embedded in a theory of multiple questions which analyzes the wider scope Wh-phrase of a multiple question as a universal quantifier.

The balance between syntax and morphology: Dutch particles and resultatives

Abstract: This paper focuses on Dutch verb-particle constructions and verb-resultative constructions. On the one hand, Dutch particles and resultatives share some properties; for instance, they mutually exclude each other. On the other hand, they show contrastive behavior with respect to, for example, movement. The similarities can be captured if to some extent, the two constructions receive the same analysis. It is argued that both particles and resultatives are base generated in a position adjoined to the verb. The differences between the constructions follow from the assumption that resultatives are adjoined to the verb at D-structure, while particles are adjoined to the verb in the morphological component. This analysis has several consequences for the syntaxmorphology interface: (i) there has to be a separate morphological component, (ii) the relation between this component and syntax is determined by generalizing metarules, and (ii) morphological structures are visible to syntactic principles such as the proposed constraint on the complexity of heads.

Nimboran position class morphology

Abstract: The verbal morphemes in the Papuan language Nimboran are rigidly ordered; moreover, morphemes with identical ordering properties are in complementary distribution. This suggests that verbal morphemes belong to position classes, each permitting at most one member to surface. Certain morphemes belong simultaneously to more than one position class, with corresponding blocking of all morphemes in the relevant classes. A striking generalization is that position classes blocked in this joint fashion must be contiguous. The problem is that linear order and blocking diagnose two incompatible orderings for the position classes. The solution rests in reinterpreting verbalpositions as levels in a fixed morphological hierarchy; we resolve the ordering paradox by exploiting the distinction between dominance and precedence available in a hierarchical structure. This paper adduces new support for the theory of level-ordering and offers a formal theory of position class morphology, a well-known phenomenon which deserves attention in morphological theory. It also covers a large corpus of data — and certain phenomena — not previously discussed in the generative literature.

On the independence and interdependence of syntactic and morphological properties: English aspectual come and go

Abstract: The English verbscome andgo in the ‘aspectual’ use exhibit a curious restriction: they may not appear in any overtly inflected form. We argue that this inflectional restriction is due to syntactic and thematic considerations which are independent of word formation, supporting the claim that the basic properties of inflectional morphology are to be analyzed by making reference to the syntax proper. Our analysis has implications for several other topics, including the structure of INFL, the mechanisms involved in affix lowering, the ban on unbound affixes, the LF properties of primary vs. secondary ϑ-role assigners, and the role played by pleonasticdo.

Symmetric and asymmetric passives

Abstract: Double object constructions manifest three different passive patterns. In Kinyarwanda, Norwegian, and Swedish, either object can passivize (symmetric passive) while in English, Fula, and Chicheŵa, only one object can passivize (asymmetric passive). However, there are two types of asymmetric passives. In the English type, the accusative object with the highest thematic role must passivize and transitive impersonal passives are impossible, regardless of whether the language allows intransitive impersonal passives. In the Chicheŵa type, only the object associated with the Case-assigning morpheme adjacent to the passive morpheme can passivize.

Comitative coordination: A case study in group formation

Abstract: This paper argues that in Russian a (singular) NP can combine with a comitative PP to form a complex plural NP, and that this NP denotes a group in the sense of Landman (1989). A single-headed GPSG analysis of the construction is proposed and argued for, and the implications of the analysis for number agreement are discussed. The semantic properties of the construction (and its counterpart in Polish) are subsequently detailed and are compared with those of ‘ordinary’ NP coordination; the preliminary conclusion is that the construction differs both in denotation and in conventional meaning from NP coordination.

Cyclicity and stress in Moses-Columbia Salish (Nxa'amxcin)

Abstract: The very complex system of stress-assignment in Moses-Columbia Salish (Nxa'amxcin), an Interior Salish language that has not yet been discussed in the generative literature, is accounted for in a simple and principled manner within the Halle and Vergnaud (1987a, b) metrical framework by assuming that two rules of stress assignment interact with morphological properties ofcyclicity, accent, andextrametricality. The basic properties of the system are typologically similar to those found in Indo-European languages such as Sanskrit and Russian. Moses-Columbian distinguishes primarily two classes of suffixes: dominant suffixes which trigger cyclic stress assignment and recessive suffixes which are assigned stress by a noncyclic application of a stress rule. Which class a suffix belongs to cannot be predicted from its morphological category nor from its position in a word. Moses-Columbian thus provides support for the hypothesis that cyclicity is a diacritic property which must be specified for each affix.

The range of null operators: Evidence from clefting

Abstract: In this paper, I examine the assumption that a typology of lexical categories defines what may surface in the focus position of cleft sentences. Assuming the analysis of the copula in Heggie (1988) and the structure of clefts in Chomsky (1977) and Barss (1984), I argue that it is a constraint on the null operator in cleft constructions (the Null Operator Generalisation, or NOG) which largely determines the grammaticality of cleft sentences. The one domain in which this observation does not hold is a certain class of cleft sentences which I have labelled METALINGUISTIC. The two resultant types of clefts are distinct, however, with respect to a number of interpretative properties, providing evidence for the positing of separate syntactic derivations. The NOG is demonstrated to make correct predictions for the grammaticality of non-metalinguistic clefts. The NOG is then generalized to include other base-generated empty elements, as illustrated in the analysis of predicate clitics in French.

On the role of conceptual structure in argument selection: A reply to Emonds

Abstract: Emonds (1991) claims that many phenomena that I attribute to the effects of conceptual structure on syntax in Jackendoff (1987a) are more properly accounted for in terms of syntax alone. The present paper shows that Emonds has misconstrued the role intended for conceptual structure in the interface between language and cognition. It then specifies precisely where Emonds's approach to argument selection differs technically from mine, and shows that the technical differences favor my approach as further developed in Jackendoff (1990). Finally, an improved account of the semantic selection of oblique arguments is proposed, but it is shown that still not all syntactic argument selection can be eliminated from lexical entries of verbs.

On null operator structures

Abstract: This paper presents a unified account of null operator structures, based on the hypothesis that the categorial status of empty categories (ECs) in A′ positions is determined by free assignment of the features [±a(naphor)] and [±p(ronominal)], essentially as suggested by Brody (1984) for ECs in A-positions. In so-called predicational structures, the null operator is ungoverned and can only surface as [+p]. As expected, it can be locally A-bound or discourse-licensed, just like argument PRO orpro. Null operators in adjunct clauses with parasitic gaps are governed, and can surface as [−a, −p] or [−a, +p]. The properties of these structures are only compatible with the assignment of the features [−a, −p]. This suggests that even though no general principle blocks [−a, +p] as a governed null operator, there are no licensing conditions for this category.

On case and the passive morpheme

Abstract: Some recent analyses of the passive, such as Baker (1988) and Baker, Johnson, and Roberts (1989), claim that the passive morpheme is an argument and that the twin properties of ϑ-role absorption and Case absorption may be attributed to its argument status. Through an examination of passive constructions in several languages, we find that such analyses are unable to account for the ways in which Case absorption sometimes fails to apply. More generally, the facts examined provide important counterexamples to Burzio's generalization and show that there is greater diversity in the Case properties of passive morphemes than was previously thought.

A bound pronoun in modern Greek

Abstract: Modern Greekidhios has different properties depending on its grammatical function. Non-Subjectidhios must have a binder but it does not seem to obey the locality restrictions characteristic of anaphors (Iatridou 1986). Furthermore, it cannot occur in embedded questions or relative clauses thought it can be found in noun complement constructions and adjuncts. In addition it licenses only sloppy identity interpretations under ellipsis and cannot tolerate split antecedents. Subjectidhios differs in every respect. It need not be bound, it can occur in embedded questions and relative clauses. It does not require a sloppy identity interpretation under ellipsis and it can take split antecedents. We show that these properties ofidhios follow from the assumption non-subjectidhios is an A′-bound pronoun. We derive the properties of subjectidhios by showing that it cannot be a bound pronoun as it is subject to an A′-disjointness constraint typical of bound pronouns.

Skeletal versus moraic representations in Slovak

Abstract: It is argued that compensatory vowel vocalization and the rule of Depalatalization in Slovak crucially require reference to a representation at the skeletal tier. In the former case, a consonant deletes, and simultaneously a floating melodic segment is vocalized as a short vowel. In the latter case, adacency must be established between consonants across a floating segment. Both of these facts can be straightforwardly accounted for in the X-skeletal theory but not in the moraic framework, regardless of the variant of the moraic theory.

On a type of argument-island in French

Abstract: The syntactic relation of ‘sister to a theta-marking head’ plays a crucial role in recent accounts of long-distancewh-dependencies: when this relation holds for a given phrase, it does not act as a syntactic island but allows long distance dependencies. The present study examines French verbal arguments preceded bya andde, which appear to contradict this. These internal arguments are apparent sisters to their verb in the majority of cases; nonetheless, they constitute islands forwh-extraction out of them. I argue that the islandhood of these arguments is a result of their syntactic status as sister to V′ rather than V0, which in turn stems from their being mapped from the lexicon onto the syntax by means of non-canonical correspondence rules. Drawing on Jackendoff's (1990) Conceptual Structure, I propose that non-canonical mapping rules cannot create relationships of direct selection, in Cinque's (1990) sense. The syntax ofa andde complements thus confirms both the importance of the structural notion of ‘sister to a theta-marking head’ and the crucial role of non-canonical rules in mapping lexical structure onto syntactic structure.