A new type of noun + verb compound formation is attested in Japanese which involves a sentential structure as its input and presumably takes place in the phonological component. This ‘postsyntactic’ compounding serves not only to weaken the strong lexicalist hypothesis but also to elucidate problems surrounding the delicate and controversial issue of where word formation processes take place. The discovery that morphological constraints that have hitherto been believed to characterize lexically formed words also apply to postsyntactic compounds leads us to dissociate these constraints from the lexicon and set them up as an independent system of general principles that constrain word formation processes in various components in a global manner. The proposed word formation model thus instantiates Chomsky's modular conception of grammatical organization.

In so-called ‘Bracketing Paradoxes’, it seems to be necessary to assign two incompatible constituent structures to a single expression. The paper outlines empirical and conceptual difficulties with three main current approaches and discusses in detail the problem of ‘paradoxical’ expressions relating to people, such as transformational grammarian, theoretical linguist, serial composer, and Southern Dane. It is shown that such expressions can be formed productively by means of a process of analogical back-formation defined strictly over existing entries in the permanent lexicon, a novel type of productive word formation process.

In the modern Aramaic dialect of the Jews of Azerbaijan, emphasis (pharyngealization) behaves much like vowel harmony, though it affects consonants and vowels alike. An autosegmental analysis of its distribution shows that emphatic spans are underlyingly marked as such, whereas plain spans must have no initial specification as to emphasis, and are eventually pronounced as non-emphatic by default. In mixed words, which are part plain and part emphatic, emphasis is underlyingly associated with a particular syllable; but in words which are emphatic throughout, the underlying mark of emphasis is floating—not associated with any particular segmental position.

This article develops an autosegmental account of tone and stress in Serbo-Croatian. The traditional four accents of Serbo-Croatian are decomposed into two independent subcomponents within the accentual system: tone and stress. While tone participates in lexical contrasts, the location of stress is predictable from that of tone, and in this respect Serbo-Croatian represents a previously unattested type of pitch-accent language. The analysis extends naturally to a range of intonation data, some of which are beyond the scope of earlier accounts.

In the syntactic process of causative formation in Oromo, the affixation of the causative morpheme is sensitive to initial grammatical relations: the number of causative morphemes corresponds to the number of logical subjects in the clause. Thus, transitive and unergative verbs can be distinguished from unaccusatives in causative constructions. A causative-intensive construction may also be formed via reduplication of this causative morpheme. However, not all predicates that appear to be causatives can be intensified in this way. We propose that these predicates (a restricted number of unaccusative verb stems) combine derivationally with the causative morpheme, and that the output of this derivation may not be intensified. Oromo, then, shows the distinct effects of similar morphological processes arising either in the lexicon or in the syntax.

This paper uses spontaneous and imitated speech data from 50 Mexican-American preschoolers to demonstrate that children learning Spanish as a first language learn rules for assigning stress, as opposed to simply memorizing stress on a word-by-word basis. From a developmental perspective, this finding is taken as an indication that children's propensity to hypothesize linguistic rules is so strong as to take effect even when rules are not needed to produce correct forms and are obscured by large numbers of exceptional forms. From a theoretical perspective, the acquisition data are used to support the metrical theory of stress advanced by Harris 1983 and Den Os & Kager 1986, and to suggest that future theories take into account degrees of irregularity among different irregular stress types.

We consider an anomalous but internally consistent set of WH-questions produced by a 3½-year-old child. They are of theoretical interest because they appear to violate constraints against the movement of constituents out of noun phrases, thus raising questions about the conditions of applicability of such constraints. After giving an account of the probable sources of these questions in several kinds of verbal interactions between the child and his father, we propose that discourse patterns may affect the acquisition of sentence grammar in unexpected ways, and may even allow the language learner to temporarily modify the domain of applicability of universal constraints.

The likelihood that substrate languages play a key role in pidgin/creole genesis is widely acknowledged; still, because so many instantiations of substratal influence involve phenomena that occur widely in the world's languages, the question arises, for a given pidgin or creole, whether the responsible factor is the substrate or linguistic universals. To separate the specifically substratal from the universal, the present study examines highly marked substratal input and its impact. Resumptive pronouns in relative clauses are considered in two varieties of pidginized Liberian English and in the relevant Niger-Congo languages. The evidence indicates that degree of homogeneity in substratal input bears crucially on the extent of substratal influence in a pidgin or creole.

This paper juxtaposes two well-known competing hypotheses about the structure of resultative and descriptive sentences in Chinese. According to the P[rimary] P[redication] hypothesis, the second verb in a ...V1...V2... sequence is treated as the main verb of the sentence; according to the S[econdary] P[redication] hypothesis, the second verb is treated as a complement to the first. Three major arguments for PP are reviewed and shown not to be cogent, and an alternative analysis is proposed for each of the issues raised: the distribution of A-not-A questions, negation, aspect marking, and the properties of certain complex causative constructions. The otherwise peculiar distribution of the negative and A-not-A morphemes turns out to follow from a principle requiring them to be lexically realized in INFL, and a D-Structure analysis of complex causative sentences consistent with the SP hypothesis explains the related facts in a desirable way. Additional arguments for SP are presented which turn on certain facts of anaphora, verb reduplication, scope of negation, and tone sandhi.

Through the detailed investigation of the syntax, semantics, and pragmatics of one grammatical construction, that containing the conjunction let alone, we explore the view that the realm of idiomaticity in a language includes a great deal that is productive, highly structured, and worthy of serious grammatical investigation. It is suggested that an explanatory model of grammar will include principles whereby a language can associate semantic and pragmatic interpretation principles with syntactic configurations larger and more complex than those definable by means of single phrase structure rules.

This study is an investigation of the structural relationships between spoken and written Nukulaelae Tuvaluan, a Polynesian language spoken in a restrictedly literate society in the Central Pacific. The results of a factor analysis of the frequency of co-occurrence of 42 linguistic features across a computerized corpus of naturalistic spoken and written texts show that three dimensions must be identified to account for variation between Nukulaelae registers: attitudinal vs. authoritative discourse; informational vs. interactional focus; and rhetorical manipulation vs. structural complexity. Contrary to claims advanced for English and tacitly for speaking and writing in general, spoken Tuvaluan is not necessarily more involved, less complex, and more context-dependent than written Tuvaluan. These characteristics are a function of the communicative norms at play in each register. The structural relationships of spoken and written language must be explained in terms of the social context of orality and literacy in different literacy traditions, rather than the cognitive demands of language production and comprehension in the spoken and written modes.

Research on ‘universals of union’, fueled by new data on causatives and by the recognition of non-causative unions, has taken an unexpected turn. For over ten years it presupposed a certain structure and sought LAWS SPECIFIC TO UNION. Remarkably, the empirical findings suggest the opposite: union has a structure never hypothesized for it before (one clause where TWO OR MORE PREDICATES occur successively)—and there is little to stipulate, since nearly all observed effects are entailed by broader laws of clause structure. Further, the new analysis delimits the class of unions more narrowly, and reconciles them with otherwise problematic rules in Italian, Quechua, Tamil, and Turkish.

In the verbal morphology of many languages of the world, certain elements define the semantic functions of particular arguments of the verb, e.g. the benefactive relation, the instrumental, and the locative. Often these elements are clearly adpositional in origin. In some of the New World languages examined here, the source of these elements, which we call RELATIONAL PREVERBS, is clearly the postposition. The hypothesis that relational preverbs are incorporated into the verb by means of a syntactic rule accounts for their observed properties in some languages, but not in others. An adequate picture of their nature emerges only when their typology is interpreted in both synchronic and diachronic perspective.

A central issue in the field of language processing concerns how grammatical theory and parsing are related. Evidence from processing breakdown (GARDEN PATH phenomena) reveals the conditions under which local ambiguity results in unprocessable sentences. These data provide evidence that the processor operates by admitting structure which maximally satisfies the principles of Government and Binding Theory locally at every point during a parse, and that the constraints on syntactic reanalysis during processing are also derived from grammatical theory. Alternative approaches to parsing are demonstrated to be incapable of accounting for the wide range of garden path effects.

Two peculiar properties of French voici/voilà are shown to be directly related: the fact that they are subjectless verbs follows from their not being marked for tense or subject agreement, although they are interpreted as present indicative. This in turn supports an analysis of the pro-drop phenomenon where empty category subjects are present at all levels in the structure of ‘subjectless’ sentences, albeit not an analysis which has ‘rich agreement’ as the major licensing factor for the empty category. Rather, it is absence of abstract case assignment that licenses the presence of a non-lexical category.

Bouchard 1988 offers a new theoretical analysis which addresses a problem raised by Morin 1985. However, his analysis is based on incorrect premises. Contrary to Bouchard's central claim, there exists at least one variety of French in which subjectless voilà has a person marker, namely the Marais-Vendéen dialect.