Revisiting category-change processes: towards a constructionist typology

Category change, broadly defined as the shift from one word class to another, is an important mechanism of language innovation, yet it is usually not given much consideration. Since the 1980s, language change and innovation have essentially been accounted for by processes of grammaticalization and lexicalization (cf. Lehmann 1995 [1982], Brinton & Traugott 2005), with category change often seen as a mere side-effect of these processes. However, in spite of their considerable merits, recent studies have pointed out major shortcomings of these approaches and have started to admit the relevance of constructions and context to language and language change. The fact that Construction Grammar, a linguistic framework that considers language as a complex network of ‘constructions’ (i.e. form-function pairings), has recently gained a lot of interest (cf. Goldberg 1995, Croft 2001, Hoffmann & Trousdale 2013) should be interpreted in this context. Against this background, I will revisit the definition of (lexical) category and category change. The focus will be on category change processes with a lexical item as output in Germanic and Romance languages. I will set out different criteria that underlie the distinction between the main processes of category change in the European languages (derivation, conversion, transposition and reanalysis). At the same time, the benefits of a Construction Grammar approach to category change will be demonstrated. More particularly, I adopt the view that a lexical category can be seen as a matrix of formal (morphosyntactic) and functional-semantic features. As a consequence, lexical categories can be understood as abstract instances of ‘constructions’ and category change will be shown to be closely connected to the process of ‘constructionalization’, i.e. the creation of new form-meaning pairings (Traugott & Trousdale 2013). Since the shift to another category in general involves prominent changes at the formal level because the item has to conform to the morphological and syntactic properties (in the sense of complement types or inflectional properties, for instance) of the new word-class it belongs to, and this is logically reflected at the functional-semantic level too, it mostly results in the creation of a new construction. It will, however, also be shown that constructionalization is not involved in all types of recategorization. Furthermore, I will advance that the constructionist approach offers the advantage of accounting for the variety of input categories (ranging from morphemes to multi-word units) as well as for some problematic characteristics related to certain types of category change, such as context-sensitivity, counterdirectionality and defectiveness