Lexical activity in speech processing: Evidence from pause detection

This study explores the sensitivity of a new paradigm—pause detection—to lexical activity during spoken word processing. Short pauses [p] inserted in five-syllable-long spoken sequences were detected more slowly after words (e.g., camper[p]ton·dee·lo) than after spliced nonwords (e.g., goomper[p]ton·dee·lo). This effect was substantial for one- and two-syllable initial sequences, and gradually decreased with longer stimuli. Lexical inhibition was also found to depend on the uniqueness point of the stimuli: Late-unique stimuli caused pause detection to be delayed. Finally, pauses were detected more slowly after high-activity than low-activity nonwords (based on the size of their initial cohort). It is argued that pause detection latencies can be used as a gauge of lexical activity in speech sequences and, in particular, at word boundaries. The value of the pause detection paradigm is discussed by reference to other paradigms and their ability to tap, on-line, lexical activation and decay.