BackgroundNumerous studies have claimed that high variability phonetic training (HVPT) has significant efficacy in improving learners’ phonological decoding ability. Although mobile learning (ML) based HVPT program integrated with sound instructional design also demonstrated the effectiveness, its reliability and validity need to be further verified. AimsThis study evaluated the training's efficacy in improving Chinese learners' perception and production of English words by comparing learners' brain activation changes and network connectivity changes because of the training. SamplesParticipants were 40 undergraduate students. MethodsThis study developed a Wechat application to implement a self-regulated high-variability-phonetic-training (HVPT) for 20 university students; employed a modified phonics screening test (MPST) as a benchmark test to assess learners’ phonetic decoding ability before and after the training; designed a quasi-experiment (control/experimental design) for 40 participants; most importantly, this study used a functional near infrared spectroscopy (fNIRS) device to monitor brain activation and networks synchronisation for participants in both groups before and after training. ResultsThe results showed that learners' performance on 5-phoneme pseudowords was significantly improved. The training inhibited learners’ right fusiform gyrus. A significant correlation was observed between enhanced RMTG and increased testing scores. ConclusionThe transfer effect of Mobile MVPT on long pseudowords is significant, suggesting its efficacy in helping learners apply the phonetic patterns perceived from the training to segmentally analyse new difficult words.