Context-dependent categorisation of vowels: a mismatch negativity study of positional neutralisation

ABSTRACT

Studies exploring the mismatch negativity (MMN) response to speech sounds have identified neural activity associated with processing of phonologically distinctive information and language-specific perceptual categorisation. Yet little attention has been given to a basic fact of phonology, namely, that not all phoneme distinctions in a language are functional in all phonological contexts. The present ERP study explores a case in which the low-mid versus high-mid vowel distinction is limited to stressed syllables, resulting in category merger elsewhere – i.e. “positional neutralisation”. We provide evidence that the sensitivity of MMN generator processes to vowel distinctions parallels their position-dependent phonological status (functional versus neutralised). As an additional finding, the MMN peaked earlier for stressed than for unstressed distinctions, indicating that stress facilitates automatic auditory discrimination. The results fit neatly into models assuming that MMN reflects a mismatch between a deviant stimulus and an abstract representation of standards that omit phonologically non-distinctive information.

KEYWORDS:

• Neutralisation
• stress
• categorisation
• mismatch negativity
• speech perception

Acknowledgments

The authors would like to thank Ricardo Augusto de Souza and the UFMG Psycholinguistics Lab for laboratory facilities.

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

The data that support the findings of this study are available from the corresponding author (RRN) upon request.

ORCID

Daniel Márcio Rodrigues Silva http://orcid.org/0000-0001-7884-9205

Rui Rothe-Neves http://orcid.org/0000-0002-8896-8862

Notes

1 This results in a phonological reduction of the vowel inventory in prestressed position, but it involves no phonetic reduction such as centralization. Rather, prestressed mid vowels alternate between high-mid and low-mid in a dialect-specific and partially predictable way (Kenstowicz & Sandalo, 2016; Lee, 2013; Santana, 2018). In final post-stressed syllables, neutralization of the high:mid distinction further reduces the inventory to three vowels, which are phonetically reduced to [ɪ], [ɐ] and [ʊ] (Barbosa & Albano, 2004).

2 Note that, for each participant, condition, and latency window, averaging the so obtained data points across epochs would yield the exact same result as the mean amplitude calculated from the conventional difference wave between averaged ERPs.

Additional information

Funding

This work was supported by the National Council for Scientific and Technological Development (CNPq/Brazil) under grant number PQ 312277/2015-6 awarded to the second author and by a CNPq postdoctoral fellowship awarded to the first author.