A longitudinal study of the development of fluency of novice Japanese learners: Analysis using objective measures

Saori M Houston, Purdue University


Fluency has been studied extensively in ESL and EFL mainly to determine what spoken features are characteristics of fluent speech by comparing students who participated in study abroad programs with those who did not. These studies were mainly done with advanced learners of English as a second or foreign language, and there have not been many studies conducted with novice-level learners of foreign languages. Japanese fluency studies are especially in the minority. It is necessary to investigate the characteristics of fluency in Japanese novice-level learners since Japanese shares very little in common with English. This study investigated the developmental changes in fluency in Japanese foreign language learners (JFL) over the course of one semester using objective measures. Research questions are 1) which objective measures change in relationship to changes in L2 general proficiency throughout a semester, and how do they change, 2) Which objective measures correlate to subjective rating obtained from Japanese instructors? The participants were 30 students enrolled in Japanese 101. The objective measures were obtained by annotating audio samples with Praat and Syllable Nuclei and by parsing the annotations and calculating measures with Fluency Calculator (Fukada, Hirotani & Matsumoto 2015). The audio data was collected at the beginning and end of the semester with the same set of tasks. Objective measures taken were speed, quantity of speech, pause related measures and several measures concerned with repairs. Accuracy was also measured by the number of AS-units with or without errors. The results for the first research question suggested that the speed of speech showed steady development from very early stages in the students’ language learning process. Silent pause measures indicated that leaners became able to pause at grammatical junctures as the semester went on, but the overall pause ratio did not seem to decrease between the collection points. In addition, it was found that the two tasks used in this study generated very different results. It is not clear which task was better able to gain access to the learners’ true fluency, and this should be further investigated in future studies. Correlation coefficients were calculated to see the relationship between subjective measures and objective measures in order to answer research question 2. The results indicated that speed related measures showed high correlation values indicating that they could be good measures to predict oral proficiency. Mean run length also showed steady correlations to subjective scoring at both the first and second collection points. Pause related measures showed quite different correlation values from the first to second collection points. There were some measures that changed between the collection points, so it will be necessary to see how the relationship between oral proficiency and the objective measures may change with a wider variety of learners in future studies.




Fukada, Purdue University.

Subject Area


Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server