Effects of continuous shadowing practice on L2 overall proficiency improvement

L2 shadowing (simultaneous oral reproduction) tasks are highly cognitively demanding, because learners have to use almost all types of linguistic knowledge and skills such as phonology, syntax, semantics and so forth, while decoding auditory input, understanding message and reproducing sounds with appropriate prosodic planning concurrently. As a result, shadowing tasks are expected to not only improve listening and speaking skills but also accelerate language processing and thus improve overall proficiency of L2 learners. Significantly high correlation between L2 shadowing performances and overall proficiency has been reported in previous studies (Luo et al., 2009). However, little research on the cause-and-effect relationship between the two factors has been conducted to date. This study attempted to empirically clarify effects of continuous shadowing practice on overall proficiency. A total of 44 Japanese learners of English took the standardized proficiency test TOEIC and recorded their shadowed utterances to assess their baseline performances. Then all the participants were involved in three-month continuous shadowing practice, where they simultaneously orally reproduced a variety of 200- to 300-word passages five times per week. After this practice, they took TOEIC again and recorded their shadowed utterances of the same passages used in the pre-recording. Their utterances were assessed by cutting-edge speech technology named DNN-GOP (Deep Neural Network-Goodness of Pronunciation) (Hinton et al., 2012; Yue et al., 2017). Statistical analysis revealed significant improvement between pre- and post-TOEIC scores (t(43)=4.19, p< .001, d=.63, 95%CI [ 0.31, 0.95]) as well as between pre- and post-DNN-GOP scores (t(43)=3.40, p=.001，d=.51, 95%CI [ 0.01, 0.05]). These results show that continuous shadowing practice can improve overall L2 proficiency as well as oral reproduction performances.