Initial phoneme versus whole-word criterion to initiate pronunciation: Evidence based on response latency and initial phoneme duration

Undergraduates participated in 4 speeded naming experiments investigating 2 criteria to initiate articulation-initial phoneme (IP) or whole word (WW). These criteria make different response latency and TP duration predictions for words with regular versus irregular vowel pronunciations (e.g., "pump" vs. "pint"). The IP criterion predicts no latency differences but longer IP durations for irregulars, whereas the WW criterion predicts no IP duration differences but longer latencies for irregulars. The latencies and IP durations of words beginning with plosives are measured (a) indirectly by exploiting the conflation of latency and IP duration in the standard naming task and (b) directly by determining when closure begins and ends in the postvocalic naming task (participants say "uuhhh" until responding). Results support both criteria: Response latencies and IP durations are longer for irregular words compared with regular words.