Visual salience modulates structure choice in relative clause production.

The role of visual salience on utterance form was investigated in a picture description study. Participants heard spoken questions about animate or inanimate entities in a picture and produced a relative clause in response. Visual properties of the scenes affected production choices such that less salient inanimate entities tended to yield longer initiation latencies and to be described with passive relative clauses more than visually salient inanimates. We suggest that the participants' question-answering task can change as a function of visual salience of entities in the picture. Less salient entities require a longer visual search of the scene, which causes the speaker to notice or attend more to the non-target competitors in the picture. As a result, it becomes more important in answering the question for the speaker to contrast the target item with a salient competitor. This effect is different from other effects of visual salience, which tend to find that more salient entities take more prominent grammatical roles in the sentence. We interpret this discrepancy as evidence that visual salience does not have a single effect on sentence production, but rather its effect is modulated by task and linguistic context.