According to constructional approaches, grammar consists of an inventory of symbolic pairings of a syntactic form with an abstract meaning. Many of such so-called constructions can be perceived as having highly similar meanings: such pairs have been discussed under the name of alternations, especially in the domain of argument structure, for example the widely documented dative alternation (e.g. John gave Mary a book vs. John gave a book to Mary). This paper explores what status such pairs of constructions can be given in construction grammar, on the basis of a sorting task experiment.
Construction grammar traditionally recognizes generalizations of a common syntactic form over semantically similar sentences, but the status of higher-level generalizations of a common meaning over syntactically different forms is rarely discussed. In our study, we devised a sorting task that subjects could resolve by relying on generalizations of either of these two kinds. We find that subjects rely on alternation-based generalizations more often than purely constructional ones in their sorting behavior. We suggest these results show that generalizations of a common meaning between formally different constructions are plausible categories stored by speakers and should be given more attention in construction grammar research.