Measuring out reflexivity in secondary predication in English and Spanish

Abstract

This article provides a usage-based, bottom-up constructionist analysis la Goldberg (2006) of formally identical instances of secondary predication featuring find/encontrar (find) and a reflexive pronoun in the object slot in English and Spanish. Specifically, it shows that these two configurations can be aptly regarded as two different, though closely connected, constructions (i.e. learned form-function pairings), namely, the reflexive subjective-transitive construction and the self-descriptive subjective-transitive construction. At a higher level of resolution, it is argued that the reflexive subjective-transitive construction imposes an agentive, intentional construal on the event/state of affairs in question, while the self-descriptive subjective-transitive construction requires a non-volitional, non-agentive construal. Crucially, configurations of the reflexive subjective-transitive construction are shown to be closer to a two-participant event elaboration in which the entity encoded in the main clause and the reflexive is construed as a divided self (Haiman 1998) between an experiencer subject and an affected object. By contrast, instances of the self-descriptive subjective-transitive construction can be aptly considered to be functionally equivalent to one-participant events, and can thus often be paraphrased by means of (in-)transitive or intensive clauses. At an ever higher level of granularity, the continuum between reflexives (encoding two-participant events) and middles (encoding one-participant events) can also be observed vertically within the self-descriptive subjective-transitive construction. The crucial determinant in this respect is taken to be the inherent meaning and form properties of the object-related predicative phrase (i.e. the XPCOMP) and its transitivity properties (Hopper & Thompson 1980). Finally, corpus-based evidence is provided that although English and Spanish share a considerable number of morphosyntactic realizations of the XPCOMP, the inventory of such realizations is not fully symmetrical, thus lending further credence to Crofts (2003) contention that argument structure is not only construction-specific but also language-specific.