Automatic Speech Recognition (ASR) Systems Applied to Pronunciation Assessment of L2 Spanish for Japanese Speakers

General-purpose automatic speech recognition (ASR) systems have improved in quality and are being used for pronunciation assessment. However, the assessment of isolated short utterances, such as words minimal pairs segmental approaches, remains an important challenge, even more so non-native speakers. In this work, we compare performance our own tailored ASR system (kASR) with one Google (gASR) Spanish pair produced by 33 native Japanese speakers a computer-assisted training (CAPT) scenario. Participants pre/post-test experiment spanning four weeks were split into three groups: experimental, in-classroom, placebo. The experimental group CAPT tool described paper, which specially designed autonomous training. A statistically significant improvement in-classroom groups was revealed, moderate correlation values between gASR kASR results obtained, addition to strong correlations post-test scores both application found at final stages use. These suggest that alternatives valid assessing tools, current configuration. Discussion on possible ways improve possibilities future research included.