Cognitive explanations, distributional evidence, and diachrony

Cognitively oriented approaches to the study of linguistic structure standardly use synchronic distributional evidence to make assumptions about the psychological mechanisms and types of mental representation that lead speakers to create particular constructions. Yet, synchronic distributional patterns are the result of specific diachronic processes, and these often do not provide any evidence for the psychological mechanisms or types of mental representation that can be postulated to account for individual patterns on synchronic grounds. The paper illustrates this with regard to several patterns pertaining to alignment systems and prototype effects in part of speech categories. Diachronic evidence about the development of these patterns in several languages from different families suggests that assumptions about particular psychological mechanisms or types of mental representation cannot be inferred directly from synchronic distributional evidence, and should be investigated independently.