Cet article réexamine le concept de réseau social utilisé en sociolinguistique variationniste en analysant le rôle que jouent divers domaines d’affiliation liés aux liens forts et aux liens faibles à l’intérieur et à l’extérieur de la communauté dans la variation morphosyntaxique dans les expressions en tête des propositions adverbiales du français acadien du nord-est du Nouveau-Brunswick (Canada). D’après les résultats d’analyses de type règle variable, le facteur réseau social est en mesure de prédire l’emploi des formes traditionnelles comme que, quand que et si que de manière plus stable que tout autre facteur extralinguistique, incluant la classe sociale. L’analyse des correspondances est utilisée pour opérationnaliser le facteur réseau social à partir des affiliations des sujets dans les domaines d’affiliation. Cette étude suggère que, dans l’explication de la variation sociolinguistique, les liens faibles méritent un examen plus approfondi et que les statistiques multivariées sont un outil des plus appropriés pour ce genre d’étude.

The position that languages require both coda and onset options for the syllabification of word-final consonants is adopted. The latter option is further divided into languages where final consonants are onsets of empty-headed syllables and those where final consonants are syllabified through onset-nuclear (ON) sharing. ON sharing is reserved for languages where final consonants display fortition (overt release): the nucleus hosts the release of the consonant. Empirical evidence from across populations demonstrates that ON sharing is unmarked. It is favoured among the outputs of first and second language learners and individuals with Specific Language Impairment. It is further argued that final onsets are optimal for parsing in end-state grammars, as they demarcate the right word-edge more effectively than codas. Among the two types of onsets, ON sharing is preferred: through the nuclear release, it is better able to host the range of contrasts that right-edge onsets display. The parsing argument serves to illustrate how ON sharing provides an advantage to end-state grammars, beyond being an emergent property from acquisition.

According to Talmy’s typology of “lexicalization patterns” for a motion event, English is the type of language that conflates Motion and Manner in the verb root, whereas Japanese is the type of language that conflates Motion and Path in the verb root. This article provides an explanation for why there is such a difference between English and Japanese within the framework of Hale and Keyser’s syntactic approach to argument structure. It extends Hale and Keyser’s approach to motion verbs and locational/directional Ps in English and Japanese and shows that given a “Lexical Relational Structure” of a motion event, the difference between English and Japanese derives from the fact that the former has a variety of directional Ps, allowing the insertion of a manner-of-motion verb into the Motion V, whereas the latter has a variety of directed motion verbs with the Path P incorporated into the Motion V. This suggests that Talmy’s lexicalization patterns are constrained by general syntactic principles.