音声マッピング陰性（PMN）は音韻特異的ではない：話者音声の不一致に対する感度

In the electrophysiological time course of speech processing, the Phonological Mapping Negativity (PMN) is a negative-going ERP component thought to specifically index pre-lexical phonological mapping. The response is often used in research investigating the cognitive mechanisms that underlie speech processing, but more recently has been employed in clinical research seeking to identify the specific deficits that contribute to a host of speech and language impairments. The PMN has long been assumed to be sensitive only to phonological manipulations. Here, we test that assumption in a classic phoneme blending task, comparing responses to phonological mismatches and speaker voice mismatches. Participants were instructed to blend nonword primes segmented into onset and rhyme (e.g., /l/ + /ɛk/), which were followed by targets that matched or mismatched based on either phonological expectancy (e.g., match: /lɛk/; mismatch: /wɛk/) or speaker voice expectancy (match: same voice; mismatch: different voice). Analysis of the 150-400 ms post-target epoch revealed the predicted negative ERP responses to phoneme mismatches than phoneme match trials, an effect typically characterised as a PMN response. Crucially however, analysis of speaker trials revealed a similar pattern of results in response to speaker mismatch vs speaker match. Our results are discussed in light of ongoing debates about the functional sensitivity and characterization of the PMN, alongside strong recommendations against the use of the PMN as a biomarker for phonological deficits until further research is done into the authenticity of this response.