An Exploratory Validation Study of Computer-based Dictation Tests Utilizing TTS

Abstract

The present research explored the impact of different types of aural input recorded by native speakers (NS) and text-to-speech (TTS) on EFL learners’ performance in computer-based dictation tests. A set of computer-based dictation tests was developed. Participants in the study were 117 EFL learners, grouped into three proficiency levels. Results showed no significant effects of both input order and test types on the performance across the three levels. In contrast, there was a statistically significant difference among the three groups in dictation test scores. Tenuous interactions were detected between the test type and aural input regardless of learners’ proficiency levels even though they were not statistically significant. In the phrase dictation test, the types of input did not have an impact on the performance across the three levels. However, high school students were shown to be influenced by the types of input in the clause dictation test. The survey revealed that most respondents were in favor of dictation as a learning tool while noticing no significant differences between NS and TTS. The findings are expected to shed light on how to capitalize on potentials of TTS to develop an automated dictation test as an integrative testing and learning tool.