Mismatch negativity is not always modulated by lexicality

Abstract

One factor that is commonly thought to influence MMN amplitude is lexicality; multiple studies have shown that real-word deviants elicit larger MMNs than pseudoword deviants. Here, however, we report data from two experiments challenging this assumption. In the first experiment (N = 48), real-word deviants did not elicit more negative MMNs than pseudoword deviants; the acoustic difference between standard and deviant was identical across these comparisons. In this experiment, the pseudoword deviant [pʰa˨˩] differed from a real-word [pʰa˧˥] in tone only; therefore, to test the possibility that the lexicality effect is real but is restricted to pseudowords that differ from real words by at least one segment, we ran a second experiment which included different items and participants, and also included a control comparison in which the pseudoword ([tsʰei˨˩]) differs from all real words by at least one segment (there is no existing Mandarin morpheme pronounced [tsʰei] in any tone). In the second experiment (N = 36), both types of pseudowords failed to elicit less negative MMNs than real words. These findings, together with other recent studies failing to show lexicality effects in MMN, challenge the assumption that wordhood reliably influences MMN amplitude.