Showing papers in "Natural Language and Linguistic Theory in 1988"

The role of agreement in clitic-doubled constructions

Abstract: We began with the premise that doubling with DO- and IO-CLs requires that both elements of the chain agree in features. With IOs, doubling is universal because the features of the IO argument are practically irrelevant (section 1.1). However, since DO-CLs are inherently specific, this feature is crucial for DO-doubling to be grammatical. Animacy is the second feature necessary in most contexts. However, as pointed out in section 1.3.1, doubling may operate even in the absence of the latter. The same matching of features required of doubling constructions is required under extraction (a subset of doubling structures), hence, no special mechanisms are needed to reject ill-formed sentences. What causes ungrammaticality in extractions from CL-D DOs is a clash between the specific referential CL and a nonspecific interpreted object argument, a violation of the Matching Principle. Since lexical partitives and unagreement phenomena are productive ways to signal the specificity of the DO, extractions at LF and in the syntax proper can produce well-formed sentences (section 2.2). Moreover, the Matching Principle follows directly from the unique indexing peculiar to chain coindexing and Spec-head agreement. These results have two main consequences: (a) the hypothesis that CLs are agreement morphemes which do not absorb Case becomes viable. This claim is supported by the doubling of inanimates (section 1.3.1) and even of some animates (17) in the absence of a, by facts about Case with IOs (section 1.3.2), and by weak cross-over effects (section 3.1) where CLs-as-agreement serve to identify the relevant empty category. (b) Since extractions of both CL-D IOs (section 2.1) and CL-D DOs (section 2.2) are possible, it follows that CLs are not theta-role absorbers, because the doubled constituent must be in an argument position. The moral is that the absence of extraction does not itself show that the double is in an A' position. Rather one must look deeper to discover whether independent principles in the language can account for lack of extraction in such contexts. For extensions of our main hypothesis to other Spanish dialects see Suner (1986). Finally, the assumption that datives in a CL-chain are NPs and not PPs (see Appendix) makes possible the generalization that contemporary Spanish CLs — whether direct or indirect — enter only into nominal chains, a generalization confirmed by their parallel behavior with respect to weak crossover and scope (sections 3.1–3.2). This conclusion brings the analysis of Spanish in line with those of French and Rumanian (Steriade 1980/81) where datives have been argued to be NPs, and it opens the door for the possibility that datives in all the Romance languages might be nominal in nature. Interestingly, datives are also NPs in Hebrew (Borer and Grodzinsky 1986).

On multiple questions and multiple wh fronting

Abstract: My goals in this article have been to describe two types of multiple-Wh-fronting languages and to integrate them into a general typology of Wh-movement and multiple-Wh-constructions. In the first part of the article I demonstrated that in spite of their superficial similarity, the East European languages which normally place all Wh-words at the beginning of the clause fall into two distinct groups with different structures for multiple-Wh-constructions. One group, which includes Bulgarian and Romanian, places all of the Wh-words of a multiple question in SpecCP at S-structure, while the second group, including Polish, Czech, and Serbo-Croatian, has only one Wh-word in SpecCP. I have referred to these two groups as the [+MFS] (Multiply Filled Specifier) and [−MFS] languages, respectively. The difference in S-structure accounts for systematic differences in extraction possibilities for multiple Wh-words, island effects, clitic position and occurrence of parentheticals and other material within the fronted Wh-word sequence, and strictness of word order in multiple-Wh-constructions in the two types of languages. The second part of the article attempts to locate the two types of multiple-Wh-fronting languages within the broader range of possible structures for multiple questions and to provide some account of this range. As a preliminary to this discussion, the two multiple-fronting structures are spelled out in more detail. For the [+MFS] languages I argue for a hierarchical structure within SpecCP, such that noninitial WHs are adjoined to the right of the specifier position, while the initial one is the head of SpecCP. For the [−MFS] languages, I argue for a structure in which each noninitial WH is adjoined to IP. Among other advantages, these structures make it possible to bring the word order differences between the two types of languages under a version of the split ECP, specifically a requirement that Wh-traces be locally bound at LF. Sections 2.4 and 2.5 present several factors which, taken together, account for much of the crosslinguistic variation in Wh-constructions. One of these, the Condition of SpecCP Adjunction, holds that languages differ in the levels at which they permit adjunction to SpecCP. Other factors considered are the level(s) at which Wh-movement applies, whether Comp can be intrinsically [+WH], and whether LF Wh-movement is from A or A' positions. Further work may well show some of these to be derivable from more basic parameters. In closing, I would like to touch briefly on the important issue of learnability. How does a child learning a multiple-Wh-fronting language come to know whether it is a [+MFS] or a [−MFS] language? More generally, how does any child know where its language fits in the range of types available for Wh-constructions, particularly multiple questions? Following Adams (1984), I assume that the unmarked value for the CSA is that it applies at all levels. The child thus needs positive evidence in the form of multiple-Wh-questions to conclude that a language allows adjunction to SpecCP at LF, and in the form of Wh-island violations or multiple extraction to conclude that it does so at S-structure. Evidence for SpecCP adjunction at PF could include multiple Wh-phrases preceding a complementizer, or perhaps second position clitics following multiple WHs, if cliticization is sensitive to PF. Without such evidence at each level, the language learner assumes adjunction to SpecCP is impossible. This part of the system is thus definitely learnable, and a child presented with evidence for SpecCP adjunction at all three levels would have all the information necessary to know he or she was learning a [+MFS] language. Similarly, positive evidence is available for the level at which Wh-movement operates. In the absence of overtly moved Wh-phrases the learner assumes no syntactic Wh-movement. In the absence of WH in situ the learner assumes no LF Wh-movement from A positions. And in the absence of a [−MFS] pattern (see below) the learner assumes no LF movement from A' positions. If both moved and in situ or [−MFS] pattern Wh-phrases are found, the learner concludes that Wh-movement applies at both levels. By a [−MFS] pattern I mean multiple Wh-fronting with no subject/object word order asymmetry, second position clitics and parentheticals following the first Wh-word, and no multiple extraction. Any evidence from clitics would be especially useful for a learner, since clitics are found in short, high-frequency sentence types. This [−MFS] pattern allows the child to conclude that he or she is learning a language with LF movement from A' positions, i.e. a [−MFS] language.

English suffixation is constrained only by selectional restrictions

Abstract: This article discusses English suffixation is constrained only by selectional restrictions.

Theta theory and the syntax of applicatives in Chichewa

Abstract: Abbreviations used in the glosses in this article are: A, gender agreement; ABS, absolutive; APPL, applied affix; ASP, mood morpheme; EX, exclusive; HAB, habitual tense; OP, object prefix; PASS, passive; PRES, present; RECIP, reciprocal; SP, subject prefix; SUF, suffix. Glosses of pronominal agreement consist of a number indicating person, a lower case letter indicating number (singular or plural), and an upper case letter indicating grammatical function (subject or object). Noun classes are not indicated in the glosses, but certain agreement relationships are shown by highlighting. The notation...*(B)...means that the structure is grammatical only if B is included;...(*B)...means that it is grammatical only if B is not included. Tone is not marked.

Template morphology and the direction of association

Abstract: I have argued in favor of allowing association to proceed from the edges inward in stems as well as affixes, and given an analysis of Classical Arabic verbal morphology as illustration. Since edge-in association is needed for reduplication, making use of the same principle for all association and relinquishing LR association entirely would be a desirable ambition. It remains for future research to establish the viability of this goal.

Proper government of antecedentless VP in English and Spanish

Abstract: It has been argued that null VP structures in English occur where they are properly governed by INFL. Proper government was shown to be satisfied by INFL where it is lexically headed and it fully θ-marks VP: that is, it both assigns a θ-role to VP and assigns a referential index to VP in the form of identification for a time — a value for the feature [PAST]. This latter requirement, which is not fully met when INFL is [-TENSE], accounts for the restricted distribution of null VP in infinitives. In clauses with auxiliaries, Head-Head Agreement allows the Minimality condition to be observed, and allows a non-sister of INFL (or of perfectivehave) to be θ-governed.

Templatic transfer in Arabic broken plurals

Abstract: In summary, we have shown how transfer can account for some cases of templatic morphology, beyond the reduplication examples discussed by Clements. John McCarthy observes that Modern Hebrew Pi'el and Hitpa'el could also be analyzed along these lines. See McCarthy (1984) and Bat-El (1986) for alternative analyses. We are not proposing that transfer should replace normal direct melodic linking. Rather, we are proposing that direct melodic linking cannot insightfully account for all cases of templatic morphology and must be supplemented by transfer. It should be noted that our use of transfer differs from that of reduplicative transfer. First, while a reduplicative affix undergoes linearization with respect to the stem, templatic transfer does not entail linearization. A second difference between templatic transfer and reduplicative transfer is that, in templatic transfer, association between skeleta is one-to-one, left-to-right. In reduplicative transfer, Clements proposes a different procedure where vowels associate before consonants. This latter method simply would not work in the case of broken plurals. A third potential differences between them is OVERRIDING. In templatic association, a directly associated melodic element overrides a transferred melodic element. In reduplication, it is unclear whether prespecificaion precludes association or overrides it as here. To maintain this analysis in light of these differences, it must be supposed that these differences correspond to parameters along which languages can vary. The strongest claim we could make is that these differences are not independent, but correspond to a single parameter. Another explanation for these differences might be that transfer simply does not occur in reduplication, and that its apparent effects there are due to other mechanisms, perhaps along the lines of those suggested in note 8 above. However, whatever the merits of reduplicative transfer, some sort of transfer must be available for templatic morphology.

Subjects and the θ-criterion

Abstract: In conclusion, I have proposed that subject is a role relevant to an extension of the θ-criterion, and that there are two structural positions to which it can potentially be assigned — NP of S (or NP), and object ofby. I have suggested that while some difficulties remain, certain properties of nominals and of passive sentences follow directly — in the former case, the incompatibility of a subject genitive with aby-NP; in the latter, the general inability of an implicit agent to serve as a controller. Whether the residual difficulties can be overcome and whether these proposals will prove to be consistent with a wider range of phenomena are questions for future research.

Complementizer-cliticization in Hebrew and the empty category principle

Abstract: I have tried to show that the complementizer se adjoins to or substitutes for an X0 category on its right. This process of syntactic movement vacates the head of the CP node. That is why se fails to trigger complementizer-trace phenomena. I have assumed in this paper the X-bar schemata proposed in Chomsky (1986). Specifically, I adopted the proposal that the X-bar convention carries over to the non-lexical projections of COMP and INFL. Under this view, S' is a projection of C(OMP), labelled CP and has the structure given in (20), with [SPEC/CP] being the landing-site of Whoperators and the head of CP, the node under which the complementizer is generated. Note, now that in a theory where COMP and S' are taken to be defective categories from the point of view of X-bar theory, the first because it does not project beyond its head, the second because it fails to have either a SPEC or a head, it is hard to motivate movement of a complementizer.13 If Wh-movement involves adjunction to COMP, as a number of theories have held, it is hard to see how movement of COMP itself, that is, the category to which Wh-elements adjoin, could be licensed. In the theory advanced in Chomsky (1986), movement of a complementizer is a sub-case of Head Movement, a perfectly licit and familiar sort of movement. The analysis proposed for the Hebrew complementizer se in this paper may thus be taken as an argument in favor of the structure in (20) above.

Causatives: From lexicon to syntax

Abstract: By exploring the correspondences ofhacer withordenar on the one hand, and withhacer itself in simplex structures, we have derived some of the most intriguing properties of Causative structures with respect to word-ordering, theta marking, the case marking processes without appealing to exceptional transformations or structures in order to develop our analysis. The apparently exceptional position of the Controller was derived naturally from the parameter which correlates Controller positions with subject positions and the observation that inverted Controllers coincide with inverted subjects in Romance languages. The definition of the Extended Governing Category, independently needed in other Long Clitic Movement environments, has helped account for the transparency of Causative complements. The ergative structures, yielding accusative case marking, have been explained in the light of an analysis independently needed forparecer; furthermore, ergatives cannot be embedded below dative controllers because this would require movement of PRO into the embedded subject position which is impossible because they lack INFL. Lack of INFL (which appears to be semantically linked to implicative verbs, a set which Causatives belong to), is responsible for most of the features that make Causatives appear as syntactically deviant. It accounts for the obligatory adjacency of the Causative complement and the matrix verb as well as for the obligatoriness of Long Clitic Movement; it explains the complementary distribution of Pattern A and Pattern B and the curious patterning which distinguishes passives embedded under Causatives from passives in inflected environments, bringing Causative complements closer to nominalizations. All in all, this analysis shows that a simpler image may be obtained — and a clearer one — if we consider Causative complements as a unique bridge between lexicon and syntax, the privileged component where INFL emerges.

Feature percolation, unary features, and the coordination of English NPs

Abstract: SGWW's adoption of unary features creates a range of problems The general problems affecting feature percolation can be cured by minor redefinitions of the relevant principles But if we wish to maintain SGWW's attractive position that the syntax of coordination depends on fully general principles which extend directly to cover person categories in NP coordination, while the distinctive behaviour of number in such coordination follows from properties of the phrases which introduce specific conjunctions, then something further is required A straightforward solution, which will have implications for the treatment of person, number and gender in other languages, is to retain orthodox features with more than one value and modify the scope of feature percolation by permitting the Head Feature Convention to hold for specific, probably unmarked, feature values This imports the intended effect of unary feature specification into the workings of feature percolation

Disjunctive ordering and French morphology

Abstract: A principle such as disjunctive ordering definitely has a considerable appeal and appears to play an important role in many descriptions. While its role in morphology has been seriously questioned by Janda and Sandoval (1984), Anderson (1986) shows conclusively that the cases that pose serious potential problems can be given alternative accounts which resolve the apparent contradiction. The case studies presented here, however, show that not all problems have been solved satisfactorily.