Date of Award

December 2021

Degree Type


Degree Name

Doctor of Philosophy



First Advisor

Hanyong Park

Committee Members

Anne Pycha, Sandra Liliana Pucci, Jae Yung Song


This study aimed to investigate the role that phonetic convergence plays in the acquisition of L2 segments. In particular, it examined whether phonetic convergence towards native speakers could help Arabic-speaking second-language (L2) learners of English improve their pronunciation of four problematic English segments (/p, v, ɛ, oʊ/). To do so, the study went through several phases of experimental studies. Phonetic convergence was first explored in the productions of Arabic L2 learners towards five different English native model talkers in non-interactive setting. Five XAB perceptual similarity judgments and acoustic measurements of VOT, vowel duration, F0, and F1*F2 were used to evaluate phonetic convergence.Based mainly on perceptual measures of phonetic convergence, learners were divided evenly between two groups. C-group (convergence group) received phonetic production training from the model talkers to whom they showed the highest degree of phonetic convergence, while D-group (divergence group) received training from the model talkers they showed divergence from or the least convergence to. Training lasted three consecutive days with target segments (i.e., /p, v, ɛ, oʊ/) presented in nonsense words. They were trained using the shadowing technique that used low-variability training paradigm in which each learner received training from one native model talker. Native-speaker judgments on segmental intelligibility indicated both groups showed significant improvement on the post-test; however, no significant differences were found between groups in terms of the overall magnitude of this change. Perceived convergence in learners’ speech failed to explain the improvement. However, some patterns of acoustic convergence towards their trainers, regardless of group, predicted the overall segmental intelligibility gains. The findings suggested that the more trainees converged their vowel duration and formants to their trainers, the more their performance improved. At featural level, the study examined the relationship between the preexisting phonetic distance between the Arabic L2 learners of English and model talkers before the exposure and the degree of convergence. Results indicated that there was a direct relationship between how far Arabic L2 learners were from the native model talkers and the degree of convergence in all measured acoustic features. That is, the greater the baseline distance, the greater the degree of phonetic convergence was. However, such a relationship might be due to the metric used to assess phonetic convergence. The relationship between phonetic convergence measured by difference in distance (DID) and the absolute baseline distance is always biased due to the way they are calculated (Cohen Priva & Sanker, 2019; MacLeod, 2021). This study found shadowing to be an effective technique to promote segmental intelligibility among Arabic-speakers learning English as an L2. However, this effectiveness might be increased by trainees converging more to their trainers in vowel duration and vowel spectra or being similar to their trainers in this regard from the beginning.

Included in

Linguistics Commons