How lingering representations of abandoned context words affect speech production

Four experiments tested whether and how initially planned but then abandoned speech can influence the production of a subsequent resumption. Participants named initial pictures, which were sometimes suddenly replaced by target pictures that were related in meaning or word form or were unrelated. They then had to stop and resume with the name of the target picture. Target picture naming latencies were measured separately for trials in which the initial speech was skipped, interrupted, or completed. Semantically related initial pictures helped the production of the target word, although the effect dissipated once the utterance of the initial picture name had been completed. In contrast, phonologically related initial pictures hindered the production of the target word, but only for trials in which the name of the initial picture had at least partly been uttered. This semantic facilitation and phonological interference did not depend on the time interval between the initial and target picture, which was either varied between 200 ms and 400 ms (Experiments 1–2) or was kept constant at 300 ms (Experiments 3–4). We discuss the implications of these results for models of speech self-monitoring and for models of problem-free word production.

► We used a picture naming task in which pictures were sometimes suddenly replaced. ► A semantically related context picture often helped the production of the target. ► A phonologically related context picture often hindered the production of the target. ► Initially planned but then abandoned speech influences a subsequent resumption.