Metacognition in second language speech perception and production

Abstract

Metacognition is the ability to monitor and control one's own cognition and behaviour. However, the role of metacognition in language remains poorly understood. Here we investigated metacognitive processing in non-native language perception and production, by asking participants to rate, on a trial-by-trial basis, their self-confidence in the accuracy in a phoneme identification and production task. The results revealed metacognitive ability in perception, as participants' confidence judgments aligned with the accuracy in the non-native speech discrimination task. In the production task, self-confidence did not align with a fine-grained precision measure of one’s own production - indexed by Mahalanobis distance to the target-vowel native space. However, self-confidence ratings predicted whether one’s production was within/outside the ‘native’ zone, suggesting that metacognitive monitoring in non-native language production operates on relatively coarse, yet meaningful sound level representations. While overall confidence ratings were similar and highly correlated between the perception and production tasks, there were no associations between the two language domains regarding the primary task performance or metacognitive ability. We discuss the ramifications of these findings for domain-generality/specificity in metacognitive processing in non-native language, and the unsettled debate on the relationship between language perception and production. Finally, we note future research directions that emerge from the present work.

Metacognition in Second Language Speech Perception and Production

“…However, it is not impossible that native Spanish speakers’ rounding of these vowels (revealed by the third formant, F3) differed from that implemented by native French speakers, and adding this cue in the accuracy measure would have increased the strength of the perception–production relationship (see Llompart & Reinisch, 2018, for work on the role of acoustic cues in L2 perception). To examine this possibility and the role of F3 in the production of this novel contrast, we ran a series of analyses (available in the project's OSF profile at https://osf.io/usrdw, see also Kartushina et al., 2022a), that revealed: (a) no difference in F3 between Spanish and native French groups, nor interaction between group and vowel, (b) no correlation between the F3‐distance to the norms (participants’ F3 and native French speakers’ F3) and the confidence rating, and (c) no correlation between the F3‐distance to the norms and participants’ identification accuracy for either vowel. These additional analyses suggested that, similar to native speakers of French, Spanish participants did not use F3, a roundness cue, to distinguish between the two French vowels and that the accuracy in F3 (alone) was not related to the participants’ contrast perception.…”

“…We used a two-tailed alpha level of .05 to establish significance for all statistical tests. Experimental data, stimuli, experimental software, and analysis scripts are available on IRIS (Kartushina et al, 2022a(Kartushina et al, , 2022b(Kartushina et al, , 2022c and via the OSF (https://osf.io/usrdw).…”

“…In each of these tasks, on each trial, the participants had first to either press a button with the correct label or read a displayed consonant–vowel word and then to indicate the degree of confidence in their answer. The original material can be accessed through the project's OSF page ( https://osf.io/usrdw) and via IRIS (see Kartushina et al., 2022c). [Colour figure can be viewed at wileyonlinelibrary.com]…”