Integrating Words into Grammatical Description: A Proposal

The relationship between words and grammatical constructions is of fundamental interest in many linguistic theories and descriptive traditions, whether from the perspective of the verb (in valence-based theories like valency grammar or HPSG), or from the perspective of constructions (in construction-based theories like Pattern Grammar and Construction Grammar).The latter has inspired a wealth of empirical research on the association between (slots in) constructions and the individual lexical items occurring in them (cf.Stefanowitsch and Gries 2009 for an overview).The results of this researchlists of words with their frequency in (or statistical association with) constructional slots -, have occasionally made their way into reference works (Francis et al. 1996;Hunston et al. 1998, and more recently Herbst and Hoffmann 2024), where they are intended as information about conventional usage.In the context of linguistic theorizing, these results, and relationships between constructions and specific words in general, have been fruitfully used to investigate constructional semantics (e.g.Levin 1993 and more recently work using the methods of collostructional analysis, cf.Stefanowitsch and Gries 2009, Stefanowitsch 2013 for overviews).Such relationships have not, however, been systematically integrated into syntactic descriptions of constructions, one reason being that there is no coherent descriptive convention for doing so.In this discussion note, I will attempt to sketch the foundation for such a convention, using the notational apparatus provided by set theory.This seems to be a natural choice, first, because the words occurring in a particular constructional slot (which I will refer to as collexemes) quite literally form sets of various kinds, and second, because the notational conventions in question are widely familiar and have proven useful for thinking about linguistic phenomena in many contexts.