An Empirical Study Of The Effectiveness Of ASR-aided Pronunciation Training For EFL Students

As a crucial part in learning a foreign language,pronunciation can affect learners’ listening and speaking ability.A poor mastery of English pronunciation may largely dampen learners’ self-confidence and enthusiasm for further study.But the current situation of Chinese non-English majors’ pronunciation teaching cannot be too optimistic.Traditional phonetic courses aren’t attached great importance to because of oversized English classes,limited course hours and constraints from regular course arrangements.Also,students themselves may not be able to pay much attention to their pronunciation as most of them haven’t received systematic pronunciation training.Therefore,it is urgent and important for non-English majors to receive English pronunciation training in order to have proper and effective oral communication.From the last decades,speech recognition technology has been applied in different fields especially in foreign language teaching.Hence,related software or applications of ASR-assisted pronunciation learning and teaching also come into being.Nowadays,with the integration of Automatic Speech Recognition technology,these ASR-based software have showed their edges,especially providing learners with a relatively relaxed learning environment and timely feedback.An essential stage of English pronunciation training lies in mimicking what a native speaker sounds or at least avoiding misunderstanding by one’s pronunciation.And the process of mimicking contains two parts: listening and speaking.From the theoretical point of view,listening belongs to the speech input while speaking refers to speech output.To generate more speech outputs,the first task one must complete is taking in more speech inputs.As for this aspect,ASR-based software for pronunciation training can provide sufficient comprehensible speech inputs for learners in ways of sounds with images or video in and out of the class.Besides,one of the three functions of comprehensible output is noticing/triggering.It helps learners notice the pronunciation gap between a native speaker and themselves,and impels them to bridge the gap.With the aid of ASR technology,learners feel motivated to generate comprehensible speech outputs.If the speech output is problematic,learners may notice the difference.And if they cannot be triggered,ASR-based software as a monitoring role supply learners pronunciationfeedback and corrections.In practical teaching,few studies have been conducted on the effectiveness of adopting ASR-aided software into English pronunciation teaching.And a large quantity of studies focus on reading and writing from the Input and Output Hypotheses.However,these two hypotheses are equally suitable to pronunciation training.Therefore,more empirical researches are needed to prove whether ASR-aided pronunciation instruction can improve the learners’ English pronunciation.By selecting the ASR-based speech training software—Rosetta Stone,this study conducts an empirical research to verify whether English pronunciation training by using this software is effective in improving non-English Majors’ pronunciation in China.First of all 93 non-English freshmen,distributed into rank B after English proficiency test from Zhejiang University of Media and Communications,participate in this experiment,and are divided into the experimental group and the control group.Before the training both of the two groups are asked to complete background questionnaire and English pronunciation pretest by removing some disqualified samples and ensuring the two groups are at nearly equal level before the training.Then,the two groups receive an eight-week pronunciation training with the same teacher and teaching materials.Participants in the control group receive the traditional teacher-led pronunciation training,while students in the experimental group receive ASR-aided pronunciation training guided by the same teacher.After the training,the two groups are told to accomplish the post-test with the same contents as in the pretest.Next,the author adopts the statistical software SPSS 19.0,comparing the pretest scores and post-test scores in the two groups of data analysis and the differences of the two groups’ pronunciation achievement;discover the different performances of the two groups lies in six problematic phonomes: [w],[d?],[r],[θ],[e],[?].Then,participants in the experimental group are undergone the evaluation questionnaire concerning the effectiveness of ASR-aided pronunciation teaching approach and attitudes towards speech recognition software.Lastly,11 students in the experimental group accept the interview focusing on the students’ opinions for ASR-aided pronunciation training.The data results of the Paired Samples T-test show that,a significant difference between the experimental group and the control group in enhancing English pronunciation after two different approaches of English pronunciation training.However,results of Independent Samples T-test of the two groups in posttest scores demonstrate that the improvement of the experimental group in posttest is significantlyhigher than that of the control group in segmental proficiency.At the same time,according to the results of the evaluation questionnaire and interview,most of the participants in the experimental group hold a positive attitude towards Rosetta Stone as an aided learning tool to improve their English pronunciation level,and put forward the drawbacks and suggestions of this software for improvement.According to the research results and further discussion,it’s effective for ASR-aided pronunciation teaching as a complementing role of conventional teacher-led training to enable students to improve their pronunciation level in segmental proficiency.In the face of the current situation of English pronunciation teaching in colleges and universities in China,the combined application of the automatic speech recognition technology and the traditional pronunciation teaching approach will not only reduce the burden on teachers,but also provide a relatively relaxed learning environment and timely speech feedback for the learners,even beneficial for students practicing English pronunciation on their own.Therefore,this study proposes some suggestions and implications for computer-assisted English phonetics teaching.