Poster: When Alice has an accent: The effects of accented speech on attentional networks
Authors: Hernández, M., Ventura-Campos, N., Costa, A., Miró-Padilla, A., & Ávila, C.
The flow of languages across the world has drastically increased over the last years. As a result, it is not rare to interact with foreign individuals who speak our native language with an accent. This hampers the processing of the message at the perceptual level because the phonological properties of accented speech depart from those of the native listener. Indeed, prior fMRI studies have shown that neural activity in regions that process the acousticphonetic signal is affected by accented speech, either foreign (e.g. Adank et al., 2012, 2013; Callan et al., 2014) or dialectal (e.g. Bestelmeyer et al., 2015). However, the effects of accented-speech processing beyond the perceptual level remain unclear. In the present fMRI study we ask whether accented speech would also hamper the processing of the message at the attentional level.
To investigate how our attentional system deals with accented messages, we examined the neural networks involved in the perception of dialectal-accented speech using stimuli that are close to daily basis scenarios of speech perception: movie watch. We chose dialectalaccented
instead of foreign-accented speech to reduce the unintelligibility effects: foreign accents depart much more from the native one than dialectal accents (at the phonological level, and also in speakers' pitch and intonation contour), which reduces intelligibility to a greater extent. In other words, we sought to compromise comprehension as little as possible.
In the (1.5T) scanner, 30 natives of Standard Spanish (that of Madrid) watched 21 scenes from Alice in Wonderland (Burton, 2010), one third of which being dubbed into (a) the participants' native Spanish dialect (Standard Spanish, non-accented condition); (b) a different Spanish dialect (Mexican Spanish, accented condition); or (c) a completely unknown language (Dutch, baseline). We used the same Independent Component Analysis (ICA, Calhoun et al., 2001) procedures as in Costumero et al. (2015a,b) to identify the task-relevant networks. We performed a separate one-way ANOVA for each identified network, with the factor "language" including the beta-weights of the three conditions. In all networks reported below, there was a significant main effect of language (significant differences at the p < 0.05 FDR-corrected level; all Fs > 21.332). Pairwise comparisons showed that the beta-weight values of the accented condition were always in between the non-accented and the baseline ones (significant differences at the p < 0.05 Bonferroni-corrected level).
In line with prior literature, the results showed that (relative to non-accented) accented speech perception enhanced bilateral activity in the perceptual speech evaluation network
(including the inferior frontal cortex, insula, supramarginal gyrus, cerebellum, and putamen), which is involved in processes related to the evaluation of acoustic-phonetic stimuli (Callan et al., 2004; Pastor et al., 2008). More importantly for our purposes, the results also showed a
reduction of bilateral activity in the dorsal attentional network for the accented condition. This network's activity correlated negatively with participants' performance on a post-scanning questionnaire on dialog content. This suggests that (compared to non-accented) accented
speech leads to a less efficient attentional processing, even though it allows less mindwondering, as revealed by the non-native condition reducing bilateral activity in the default mode network. Finally, accented speech enhanced bilateral activity in the salience network, which is consistently recruited in attentionally demanding tasks (Eichele et al., 2008).
These results indicate that the effects of accented speech go indeed beyond the perceptual level: the acoustic-phonetic signal that does not match the native prototype increases attentional demands and reduces the attentional processing of the message.