Opportunistic processing of language

• This paper supports the idea of opportunism in language processing with new data.
• It is also an attempt to force a deeper reflection on the shape of a viable and useful theory of language.
• Language is seen as a complex, dynamical system of co-adapted traits.

This paper is an attempt to tackle the idea of opportunism in language processing seriously – and its implications for language theory if one is to avoid what Poeppel and Embick (2005) call “interdisciplinary cross-sterilization”, that is the failure of linguistics and psycholinguistics to communicate with each other. It is also an attempt to force a deeper reflection on 1) the shape of a viable and useful theory of language, and 2) the relation between (and respective place of) linguistics and experimental psycholinguistics in the study of language. Towards that, I review a number of psycholinguistic findings with a view to showing how routinely parsers opt for opportunistic (as opposed to ‘elegant’) wayouts from processing dilemmas. Most of the evidence reviewed involves research of a cross-linguistic type, the common thread being that different languages resort to different solutions to the same processing problems, even when a unitary solution to at least many of these problems would be computationally within easy reach. The main purpose of this review is to provide a quantitatively suggestive account of how massively opportunism works in setting processing biases. Based on it, I go on to suggest that grammars can only be psychologically viable if they incorporate a fairly large number of interacting constraints, a default ability to generate pieces of structure without a commitment to satisfy large-scale well-formedness conditions, and no strict, fixed ordering of operations. These observations are compatible with a view of language as a complex, dynamical system of co-adapted traits, a system containing a fairly large number of possible initial states and a fairly large number of functionally optimal (opportunistic) continuations of those states. This work assumes the merits of espousing psychological adequacy.