The Effect of Perception-based training on Pronunciation of L2 Saudi Learners in EFL context

This study explored the relationship between perception and production in second
language acquisition and examined the effectiveness of perceptual phonetic training
with different models: native speaker and non-native speaker models for Saudi EFL
learners. A total of 46 participants were divided into three groups: one trained with
native English speaker recordings, another with proficient Saudi English (non-native)
recordings, and a control group that received no training. The training focused on
selected English segmental contrasts and was assessed through an AX
discrimination task and a production task, with production data transcribed and
aligned using WhisperX, an automatic speech recognition system. A pretest,
posttest, and delayed posttest design allowed for the measurement of both
immediate and retained effects of training, and learners’ reflections were also
gathered through surveys and interviews.
The results showed that perceptual training led to clear improvements in
perception for the experimental groups, whereas the control group’s performance
generally declined. In production, however, gains were less consistent and varied
depending on the contrast, with less improvement observed in difficult contrasts than
in easier ones. The training appeared to help learners notice phonetic differences
between L1 and L2 sounds, but statistical analyses revealed no significant effects of
group on either perception or production outcomes. Importantly, no meaningful
differences were found between the two experimental groups: native speaker model
and Saudi speaker model, suggesting that the use of native or non-native models in
perceptual training did not provide any particular advantage or disadvantage. Selfreported
changes in participant attitudes toward non-native speaker modes were
also observed in a post-hoc questionnaire and interviews, as a result of the training
intervention experience