Still no evidence for audience design in syntax: Resumptive pronouns are not the exception

Abstract

Speakers often tailor their speech to the needs of their interlocutors to facilitate comprehension. While such audience design can be observed in the words speakers choose (e.g., using simpler words when talking to children) or the volume of their voice (louder in noisier environments), it has rarely been observed in what syntactic structures speakers use. Consequently, many researchers have concluded that syntactic audience design is impossible. However, there exists a parallel literature in which syntactic audience design is assumed to play a central role. Specifically, English resumptive pronouns (e.g., …pronouns that nobody knows why people saythem) are puzzling in that, despite being ungrammatical, they are regularly and reliably produced. To account for this, some theories hold that speakers produce resumptive pronouns to improve the acceptability of their utterances. If so, this would be an exceptional case of syntactic audience design. In three experiments, we test this hypothesis by eliciting sentences from speakers while manipulating acceptability. We consistently show no effect of acceptability on rates of resumptive pronoun production, despite successfully manipulating production rates with a comparably sized control manipulation. We conclude that resumptive pronouns are not in fact the result of audience design, and therefore do not constitute a challenge to the idea that syntactic audience design is impossible.