It has been widely assumed that linguistic heterogeneity reflects social heterogeneity, with differences in the use of linguistic variants corresponding to social groupings and sensitive to social self-presentation in terms of one or more prestige norms. This assumption is challenged by patterns of variation within the former fishing communities of East Sutherland, Scotland. In these Gaelic-speaking communities exceptionally homogeneous single-village populations show well established patterns of intravillage and intraspeaker variation that do not correlate with such familiar social factors as socioeconomic status, sex, social network, or style, and only in a limited number of cases with age. Absence of any locally workable prestige norm (the result of geographical isolation, low literacy, dialect aberrance, and minority-language enclavement) is considered as a factor in the absence of social weighting of village-internal variants. High-proficiency, Gaelic-dominant speakers participate fully in the variation, making the obsolescence of the dialect unlikely as an explanation. Reasons for inattention to individually patterned variation within small and homogeneous speech communities are considered.

This paper provides an analysis of the syntax of Straits Salish, according to which these languages lack a noun/verb contrast at the word level. Main clauses consist of an initial predicate, minimally containing a lexical root, a functional head where valence [± TRANSITIVE:] is marked, and possibly a pronominal suffix marking an internal argument. The predicate is followed by a second position clitic string of inflectional elements, the subject pronoun and tense. Determiner phrases are derived subordinate structures, adjuncts to the main clause. We present evidence against a copular verb analysis as further substantiation of the lack of a noun/verb distinction at the lexical level. We identify certain properties of quantified contexts in Straits Salish which provide important evidence for our analysis of argument structure.

Clark's 1987 Principle of Contrast (‘Every two forms contrast in meaning’) seems inconsistent with the widespread synonymy exhibited by inflectional affixes in languages with inflection classes. But if inflection class membership is counted as part of the ‘meaning’ of an inflectional affix under tight conditions—namely if it unambiguously identifies the inflection class of the lexemes to which it attaches—then inflectional affixation is found to comply with the Principle of Contrast after all. A version of the principle extends to covert gender systems too, although apparently not to overt gender systems. Far- reaching implications for morphological theory and grammatical change are suggested. Evidence is drawn in particular from English, German, Icelandic, Latin, Rumanian, Amo, Archi, Andi, Zulu, Georgian, and Afrikaans.