This paper uses a corpus study to investigate the influence of text genre on the frequency and semantic interpretation of certain pattern/concept relations. In linking pattern/concept relations to text genre, this study identifies three types of dependency: weak dependency, where the relation appears in almost any kind of text; complete dependency, where it is strongly linked to a particular text or group of related texts; and dependency in terms of text genre. The particular examples that form the basis of the study are meronymic chez, which is found to have a significant dependency in didactic texts in the natural sciences; comme as a marker of hypernymy and co-hyponymy, which has a weaker, but observable dependency in technical and didactic genres; nominal anaphora involving hypernyms, where no consistent conclusions can be reached; and meronymic avec, where the significant factor is shown to be communicative objective rather than domain (subject matter). I discuss the relevance of such studies to Natural Language Processing, and indicate the potential for further research.
No takes yet. Share an insight, caveat, or question.
Anne Condamines (2008) studied this question.