It is argued that a wide range of extraction and coördination phenomena in English can be accounted for by a simple extension of Categorial Grammar. The same extension will account for a similar range of related phenomena in Dutch, including certain notorious cases of ‘intersecting’ dependency among discontinuous constituents of Dutch infinitival complements. Some universal implications of the theory are considered, and the relation of such grammars to processors which carry out incremental semantic interpretation is discussed.

This paper offers a reply to Coopmans' 1984 critique of Hawkins 1979, 1980, 1982. Hawkins had attempted, inter alia, to make typological word order universals relevant to concerns of generative grammar. Coopmans denies this relevance. His critique raises fundamental issues about the nature of language universals and their explanation, and about research methodology. Some of these issues are taken up in this reply. It is argued that Coopmans' dismissal of the typological approach is neither well-founded nor very helpful in the current state of the art. The different methodologies have complementary strengths, and can be of considerable mutual benefit.

This paper describes a fairly rich area of complementation in Caribbean English Creole (CEC) within the broad framework of Chomsky's Revised Extended Standard Theory, and compares the description with other recent treatments. The area covered includes a number of complement types in which the CEC particle fi functions. It is argued that the key to understanding the syntax of the complement types lies partly in the variety, and sometimes ambivalence, of fi's functions. Earlier treatments are criticized for failing to clarify these functions, and for assuming too close a parallel between CEC and English complementation. In addition, it is suggested that the present analysis may provide a basis for explaining changes in the grammar of CEC complementation that result from the process of decreolization in continuum situations.

The conceptual domains REALIS and IRREALIS form an epistemic scale representing a speaker's subjective evaluation of the reality of a given situation. This scale accounts for grammatical phenomena such as conditionals, Japanese complementizer choice, and ‘evidential’ in Turkish, Bulgarian, and Macedonian. Conditionals are identifiable not by their syntactic forms, but by speaker attitudes within the IRREALIS division of the scale. Surprise is also an IRREALIS phenomenon, since it takes time for newly-learned information to be assimilated into one's established body of knowledge.

Conversational coherence is a coöperative enterprise in which speaker and hearer jointly negotiate (a) a focus of attention—a referent—and (b) a response which further selects what aspect(s) of that referent will be attended to. Because not all potential referents can be attended to simultaneously, discourse markers like well help speakers locate themselves and their utterances in the on-going construction of discourse. Analysis of everyday talk shows that well anchors a speaker in a system of conversational exchange when the options which a prior referent has opened for upcoming coherence are not fully met. Thus well is sensitive to the information structure of questions, answers, the underlying conditions of requests, and various participation shifts in talk.