Do Speakers Build the Categories Linguists Postulate? A Usage-Based Exploration

Linguists are naturally inclined to seek maximally general categories for the description of linguistic phenomena, e.g. the present tense or the reflexive voice. It has been taken for granted that speakers use the same categories in their daily experience with language. A few studies have indicated, however, that speakers might not be able to build some general constructions that linguists postulate (see e.g. Dąbrowska 2008a; Perek 2015). If we would like for our descriptions to reflect the linguistic knowledge of native speakers, we need to empirically investigate the cognitive reality of the categories we develop. The main aim of this thesis was to investigate whether speakers build the categories linguists postulate and if so, how general these categories are. A number of corpus and experimental studies were conducted for Polish prefixed verbs and reflexive verbs, which explored categories of different levels of generality. The results of the studies suggest that speakers might build some general categories (e.g. the one for the Polish marker siebie), while they might not be able to build others (e.g. the ones for the different senses of the verbal prefix po-). These differences can be explained by the frequency with which the constructions occur as well as the nature of their typical contexts. The above result underscores the importance of empirically veryfing the categories linguists postulate. Linguists must not tacitly assume that their linguistic descriptions are cognitively real because it cannot be assessed a priori whether speakers use them or not. Since speakers might not be able to construct for categories that are established in linguistics, such as verbal prefixes, some other ‘traditional’ linguistic categories might need revisiting and empirical verification