Comparative analysis of spoken and written versions of a narrative demonstrates (1) that features which have been identified as characterizing oral discourse are also found in written discourse, and (2) that the written short story combines syntactic complexity expected in writing with features which create involvement expected in speaking. Quintessentially literary devices (repetition of sounds and words, syntactic structures, and rhythm) are shared by written literary language and ordinary spontaneous conversation because both are typified by subjective knowing and by focus on interpersonal involvement. In contrast, expository prose and content-focused oral genres, such as lectures and instructions, may be typified by objective knowing and by focus on content.

Alternatives are discussed for a number of assumptions fundamental to established linguistic theories. These alternatives lead to a coherent view of linguistic structure that treats grammar as a symbolic phenomenon, and emphasizes the importance of analys-ability to grammatical structure. A descriptive framework called space grammar is outlined, in particular its approach to semantic structure and grammatical valence relations. The concepts are applied first to nominal structure and then to the English passive construction. It is claimed that be, by, and the participial morpheme are all meaningful, and that passives are not derived from an active clausal source.

The Western Muskogean languages Chickasaw and Choctaw have a subject/oblique opposition in nominal case-marking and syntactic relations, despite their ‘active’ system of pronominal agreement. In both languages, case assignment may reflect the operation of productive relation-changing rules. These facts call into question the classification of Choctaw by R. Van Valin and W. Foley (e.g. 1980) as ‘role-dominated’. Restrictions on NP density (the number of nominal arguments per clause) may account for some of the syntactic differences between the more familiar ‘reference-dominated’ languages and languages of the Western Muskogean type.

Two sorts of analyses have been proposed for the constituent structure of the infinitival complements found in sentences like John tried to leave and John appeared to leave. One analysis maintains that such ‘bare’ infinitives are sentential, the superficially absent complementizer and subject being represented by lexically empty categories. The alternative holds that they are a type of verb phrase complement, represented as they appear superficially, lacking both complementizer and subject. The present paper argues in support of the former view, and more generally that all infinitives are clausal. A specific analysis is presented, with a range of arguments bearing on the alternatives considered.

‘Ad-Verbs’ modify (i.e. ‘change') verbs. Adverbs internal to a VP are Ad-Verbs modifying the verb that heads that VP. VP-external adverbs may be interpreted as Ad-Verbs only with respect to some higher verb (e.g. an auxiliary or a ‘verb' in the translation into an interpreted logic). Basic Ad-Sentences also exist; but neither they nor Ad-Verbs are adequately translated as functional operators applying to independently evaluated arguments. Adverbs typically translate into expressions like variable-binding operators which introduce the ‘variables’ which they ‘bind’. Adverbs signal re-evaluation of the expressions on which they operate, helping to ‘build’ logical form.