Attenuation of the phonological mapping negativity (PMN) by stimulus repetition

• ¹ McMaster University, Canada
• ² Université de Montréal, Canada
• ³ University of Helsinki, Finland
• ⁴ Queen’s University, Canada

The PMN component is a relative increase in an event-related negative deflection 250 to 350 ms after stimulus onset. It is thought to reflect mapping of acoustic speech signals onto phonological representations. It is small, sometimes undetectable, when the beginnings of auditory words match strong phonological expectations. It is large when strong expectations of the first phoneme of a word are not met. The PMN differs from the MMN response in that it is not limited to oddball paradigms. It differs from the semantic N400 in that it can be elicited in conditions with no semantic task or meaningful stimulus material. It is, however, likely that the auditory N400 incorporates ongoing phonological analysis of words based on the same processes as the PMN. We investigated in two experiments whether the PMN, like the N400, is sensitive to repetition. We predicted that repeated presentation of the same phonological stimulus would result in an attenuation of the PMN even when the stimuli violated task-dependent expectations. For this purpose, we studied a task previously found to elicit a large PMN. Participants were asked to strip an auditory stimulus word of its first consonant (e.g., “/sti:p/ without /s/”) and press a button to indicate whether the pseudoword they heard matched the correct response (/ti:p/). In this task, a strong PMN is typically elicited to all target pseudowords that begin with a different phoneme (e.g. /mi:p/) from the correct response whereas the correct response elicits a minimal PMN. We presented three kinds of targets with equal probability: correct answers, incorrect answers and repeated incorrect answers. The repeated incorrect answers were always the same pseudoword. In Experiment 1, conducted in English, the incorrect answer /kel/ was repeated 120 times. Compared to the correct answer, there was a significant PMN for both incorrect and repeated incorrect targets. However, the magnitude of the PMN was smaller for the repeated incorrect targets (always /kel/). In Experiment 2, conducted in French, the repeated incorrect answer was varied between participants to control for possible effects of the acoustic idiosyncrasies of the target words. The results from Experiment 1 were replicated. In both experiments, the repeated incorrect answer also gave rise to a large posterior positivity, peaking after 600 ms from stimulus onset, possibly reflecting recognition memory for the repeated stimulus. We conclude that the PMN reflects a process that is modified by both task-related expectations and incidental phonological learning during task performance.