Speakers
Description
Edward Schaefer will report on an L2 writing assessment study that investigates whether a combination of many-facet Rasch measurement (MFRM) and cluster analysis can statistically identify differing rater types among 40 native English-speaker raters of English essays written by Japanese university students. MFRM revealed significant severe or lenient bias among a subgroup of the raters, while cluster analysis could distinguish three significant rater types: a rhetorical features type, a linguistic features type, and a mechanics type. Implications of the study for L2 writing instruction and assessment will also be discussed. Heather Woodward will report on a mobile app she codeveloped to track and enhance spoken fluency, grounded in speech production theories (Levelt, 1993) and skill development (DeKeyser, 2007). This presentation reviews research on spoken fluency development, explores assessment methods, and presents findings from on the app’s syllable-counting ability.The study involved 84 first-year undergraduate students from a private university in Tokyo. Participants recorded one-minute monologues, and data was analyzed using Bland-Altman analysis to determine which speech-to-text method—pitch analysis, Apple’s speech-to-text, or WhisperX’s speech-to-text—best matched human syllable counts. WhisperX outperformed both pitch analysis and Apple’s in-house speech-to-text offering superior performance in analyzing speech rate.
Summary
This 60-minute forum features two presentations, one on the use of Many-facet Rasch Measurement together with cluster analysis to examine rater differences in the assessment of L2 writing, and the other on a mobile app designed to track and assess L2 learners' spoken fluency development.
| Teaching Context | College and university education |
|---|
