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

Abstract

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

When A Nonnative Accent Lets You Spot All the Errors: Examining the Syntactic Interlanguage Benefit

“…Recall that the speakers recorded both correct and incorrect versions of the sentences in real time. In line with similar past ERP research (Gosselin et al, 2021;Grey et al, 2019Grey et al, , 2020Xu et al, 2020;Caffarra & Martin, 2019;Grey & van Hell, 2017;Hanulíková et al, 2012), incorrect critical determiner phrases were not cross-spliced onto grammatically correct sentences to prioritize naturalistic, co-articulated speech. To ensure that the speakers had not unconsciously introduced prosodic "markers" of an upcoming error in the stimulus (e.g., differences in speech rate, pitch, pauses), each sentence was trimmed immediately before the target noun and presented to 30 additional native speakers of Spanish (M age = 23.8 years, SD = 4.7, 20 women, 10 men).…”

“…had a strong, detectable nonnative accent, but that they were as intelligible as the Spanish speakers. Though the informants were highly accurate in detecting syntactic errors overall, native-accented errors were identified at a higher rate than English-accented errors (see Gosselin et al, 2021;Grey & van Hell, 2017;Hanulíková et al, 2012, for similar results).…”

“…The informants tended to predict that an error was forthcoming in English-accented fragments (M = 46.37%, SD = 17.59%) much more often than in native Spanish-accented fragments (M = 7.88%, SD = 8.95%), F(1, 179) = 818.45, p < .001, regardless of the syntactic correctness of the complete sentence (English accent: ps > .090; native accent: ps > .53). This suggests that listeners implicitly expect an influx of errors when speakers are nonnative (see also Gosselin et al, 2021). However, prediction accuracy (i.e., accurately identifying a correct sentence as errorless or an incorrect sentence as containing an error before encountering the critical noun phrase) was below chance for both accents (see Table 3), indicating that the recordings did not give reliable prosodic cues for the eventual syntactic correctness of the sentence.…”

“…A limited number of studies have examined L1 listeners' processing of syntactic errors produced in nonnativeaccented speech (see also Gosselin, Martin, Navarra-Barindelli, & Caffarra, 2021;Sabo, 2021;Holt, Kung, & Demuth, 2018;Grey & van Hell, 2017;Hanulíková, Van Alphen, Van Goch, & Weber, 2012;etc., for the processing of nonnative-accented semantic errors). The results yielded by this line of research are relatively consistent: When grammatical errors are produced by nonnative speakers of a target language, L1 listeners exhibit a reduced (or absent) P600 effect as compared with errors produced by native speakers (Xu, Abdel Rahman, & Sommer, 2020;Caffarra & Martin, 2019;Hanulíková et al, 2012; see also Grey & van Hell, 2017, for similar results involving the Nref, an ERP component associated to pronoun resolution).…”

An ERP investigation of accented isolated single word processing

“…Hanulikova et al (2012) observed a similar N400 effect during listening to both native and foreign-accented spoken sentences with semantic violations (cf. Gosselin et al 2021). While Goslin et al (2012) found a reduced N400 effect in correctly spoken sentences during foreign accent listening as opposed to regional or native listening.…”