The presence of a foreign accent introduces lexical integration difficulties during late semantic processing

ABSTRACT

Previous research suggests that native listeners may be more tolerant to syntactic errors when they are produced in a foreign accent. However, studies investigating this topic within the semantic domain remain conflicting. The current study examined the effects of mispronunciations leading to semantic abnormality in foreign-accented speech. While their EEG was recorded, native speakers of Spanish listened to semantically correct and incorrect sentences produced by another native speaker and a native speaker of Chinese. The anomaly in the incorrect sentences was caused by a subtle mispronunciation (typical or atypical in Chinese-accented Spanish) during a critical word production. While initial-stage semantic processing yielded no accent-specific differences, late processing revealed a persistent N400-effect in the foreign-accent but not in the native-accent. These findings suggest that foreign-accented mispronunciations are more difficult to integrate than native-accented errors, regardless of their relative typicality. The distinction between syntactic and semantic processing of foreign-accented speech is discussed.

KEYWORDS:

• Error typicality
• event-related potentials
• N400
• non-native accent

Acknowledgements

The authors are grateful for the help of all research assistants at the BCBL who aided with data collection and the two anonymous Reviewers whose comments greatly improved the manuscript.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

A full stimuli list, as well as the offline questionnaire and the additional analysis and figures can be found online at: https://github.com/leahgosselin/Pelo-Perro. All data are available upon request.

Notes

1 For instance, semantic violations in past studies differed with respect to the position of the critical word in each sentence (fixed vs. variable; close to the sentence-onset vs. close to the end of the sentence). The absence of a common standard for EEG analyses should also be mentioned, as differences regarding time-windows, the number and distribution of electrodes chosen, as well as other criteria, may impact results.

2 Note that intelligibility, comprehension and error identification are not necessarily interrelated (see Derwing & Munro, 1997; Munro & Derwing, 1995). As such, it is not contradictory to report that respondents found the speakers similarly intelligible, but were still less successful at detecting errors in foreign-accented speech as compared to native speech.

3 See supplementary materials for an analysis including all 27 electrodes and the additional topographic factor of Longitude (frontal, central, parietal). Additional figures involving the topographic factor of Latitude, as well as individual conditions for the Block analysis, are included in the supplementary material.

Additional information

Funding

This international collaboration was made possible by a Globalink Research Award Abroad (MITACS), as well as a CGS-M scholarship (Social Sciences and Humanities Research Council of Canada) conferred to the first author. The research was supported by the Spanish Ministry of Economy and Competitiveness [Ministerio de Economía y Competitividad] (SEV-2015-490 to the BCBL; PSI2017-82941-P to C.D.M.), the Basque government [Eusko Jaurlaritza] (PIBA18_29 to C.D.M.), and the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement No 837228 to S.C. and No 819093 to C. D. M.), and by MINECO predoctoral grants from the Spanish goverment (BES-2016-078896 to E.N.B).