Among the numerous factors that influence pronunciation, many have argued that listening – in particular, listening discrimination, plays an important role.
Lee and Lyster (2016) explore this connect between how listening – namely, speech perception, influences speech production. This idea, known as the perception-first view, is well-supported by empirical studies, though it is not without some contention. Lee and Lyster in particular focus on speech perception training and its effect on phonological production (pronunciation). Reviewing a number of studies, the authors indicated that a common training element was corrective feedback. Their study presented below looks at the possible role corrective feedback (CF) may play in moving from accurate speech perception to accurate speech production.
Lee, A. H., & Lyster, R. (2016). Can corrective feedback on second language speech perception errors affect production accuracy?. Applied Psycholinguistics, 1-23. Retrieved from http://dx.doi.org/10.1017/S0142716416000254.
They conducted their research with 100 Korean learners of English. They divided them into five groups, each of which underwent speech perception training for eight sessions during two weeks through specially designed software. The training included listening to various words that represented words with trouble vowels for Koreans: /i/–/ɪ/ and /ɛ/–/æ/. They were able to listen to each word as many times as they wanted and then they had to choose the word orally represented. For example, they heard /ʃɪp/ and had to choose between “ship” and “sheep”. Based on their answer and the group they were in, they received the following corrective feedback:
|GROUP||INCORRECT ANSWER (CF)||CORRECT ANSWER|
|Target Group||“No, s/he said ‘ship’.”||Yes|
|Nontarget Group||“No, not ‘sheep’.”||Yes|
|Combination Group||No, s/he said ‘ship’ not ‘sheep’.”||Yes|
The participants were audiorecorded three times (pre, post, delayed post). They had to produce sentences that included the trained words, as well as some untrained words. Analysis of these recordings was done using native English speakers and acoustic analysis software.
Lee and Lyster found the following:
- Target Group:
- Production accuracy was significantly higher for trained words at both the post- and delayed posttest;
- Production accuracy for /ɛ/–/æ/ untrained words was higher at both posttests;
- Production accuracy for /i/–/ɪ/ untrained words was higher for the immediate posttest only.
- Nontarget Group:
- No significant changes for /i/–/ɪ/
- Production accuracy was higher for /ɛ/–/æ/ trained words at both postests
- Production accuracy was higher for /i/–/ɪ/ and /ɛ/–/æ/ untrained words at the immediate posttest
- Combination, Wrong, Control:
- No significant changes
Overall, they reinforced the idea in the relationship between speech perception and speech production, but CF type was a major factor. They found that providing target feedback (which is akin to a ‘recast’) is more effective than providing negative feedback (which is akin to prompts). That is, giving the target form in response to incorrect perception was better than simply telling them which sound was wrong.
How does this influence improved speech production? The researchers noticed that both target and nontarget groups would verbally respond to CF by trying to produce the correct utterances. The target groups did this more often than the nontarget group, and the other three groups, by the nature of the CF type, did not engage in this behavior. Therefore, speech perception alone is important, but “opportunities for noticing, awareness, and practice, in addition to CF” might be necessary (p. 18).