XIAONA ZHOU (THE EDUCATION UNIVERSITY OF HONG KONG)

EXPLORING THE EFFECTS OF PERCEPTUAL TRAINING WITH AI-GENERATED VOICES ON PRODUCTION OF ENGLISH LINKING : METHODOLOGIES AND TEACHING APPROACHES

Over the past decades, there has been a growing consensus among researchers about positive effects of high variability phonetic training (HVPT) on perception and production of English sounds. However, studies investigating effects of HVPT on suprasegmental aspect are rarely found. Linking, as one of suprasegmental features frequently used by native speakers in natural speech, can be seen as a key indicator of fluency, yet it poses challenges to Chinese learners to acquire. Therefore, the proposed study attempts to integrate high variability into linking training. Considering the expanding use of AI technology in pronunciation training, AI-generated voices using text-to-speech (TTS) technology will be incorporated. The purpose of the proposed study is two-fold: 1) to fill the research gaps concerning the effect of perceptual training on production of English consonant-vowel (CV) linking, and 2) to compare the effects of perceptual training using voices from humans and AI. A total of 15 participants, who are non-English major university students from Hong Kong with intermediate English proficiency, will be involved. They will be divided into three groups: two experimental groups receiving TTS-enhanced and traditional high variability perceptual training respectively, and one control group receiving no training. They will be tested at pre- and post-test on their production of English CV linking elicited from passage reading-aloud tasks. Their performance will be analyzed perceptually and acoustically. It is expected that the proposed study could shed light on integration of high variability training and TTS technology with suprasegmental perceptual training specifically for CV linking.

EdD student