Staum Casasanto Colloquium

May 9
Harper 103
Stony Brook University
Rethinking acceptability judgments

Acceptability judgments are the building blocks of grammatical theories. The theorist’s job is to create the minimal set of grammatical constraints that will account for the observed acceptability distinctions. However, under many models of the competence/performance distinction, both processing constraints and grammatical constraints can rule things out of the set of acceptable utterances. Thus, some constructions (Unacceptable Grammatical items, such as triple center embeddings) elicit low judgments of acceptability that can be accounted for without grammatical constraints.

In principle, there could also be constructions that are judged to be fairly acceptable, and that sometimes occur, despite being ruled out by grammatical constraints (Langendoen and Bever, 1973). If they exist, such Acceptable Ungrammatical items raise another problem for theorists: High acceptability judgments and attestations in non-­‐disfluent speech do not necessarily constitute evidence for grammaticality. These Unacceptable Grammatical and Acceptable Ungrammatical constructions pose a fundamental problem: How do we turn acceptability judgments into evidence for grammatical constraints?

In the first part of this talk, I will present experiments showing that processing factors can mitigate the acceptability effects of grammatical constraint violations, resulting in Acceptable Ungrammatical items which, despite their ungrammaticality, are easier to understand than their grammatical alternatives. In the second part, I’ll present experiments investigating strategies to clarify the roles of processing constraints and grammatical constraints in producing acceptability judgments: superadditivity of constraint combinations, and correlations with individual working memory capacity. I’ll also present preliminary results of an experiment applying one of these strategies to the controversial case of island violations. This research represents first steps towards determining the extent to which acceptability contrasts are underconstrained by grammar, and toward solving the problems that Unacceptable Grammatical and Acceptable Ungrammatical constructions pose for linguists who use acceptability judgments to guide their theorizing.