Temporal processing and phonotactic probability in spoken word recognition

This study examined the influence of phonotactic information on spoken language recognition in adults. Subjects performed a lexical decision task on 120 phonotactically legal bisyllablic CVCCVC nonsense words that varied in their phonotactic probability and primary stress placement. Subjects responded faster to nonsense words containing low probability initial syllables and slower to nonsense words containing high‐probability initial syllables. On the other hand, subjects responded faster to nonsense words containing high probability final syllables and slower to nonsense words containing low‐probability final syllables. Our results suggest that two phonotactically based processes may be used during word recognition. The first process is a winnowing of candidate words based on their phonotactic probability. The second process is a confirmatory process in which the lexical status of the sole candidate is confirmed based on the phonotactic probability of the item. [Work supported by NIH‐NIDCD.]