Edited by John M. Levis, Tracey M. Derwing and Murray J. Munro
[Benjamins Current Topics 121] 2022
► pp. 125–151
This study investigates how Mandarin and Slavic language speakers’ comprehensibility, accentedness, and fluency ratings, as assigned by experienced teacher-raters and novice raters, align with discrete linguistic measures, and raters’ accounts of influences on their scoring. In addition to examining mean ratings in relation to rater experience and speaker first language background, we correlated ratings with segmental, prosodic, and temporal measures. Introspective reports were segmented, coded, enumerated, and submitted to loglinear analysis to elucidate influences on ratings. Results showed that ratings were strongly correlated with prosodic goodness and moderately correlated with segmental errors, implying the importance of both segmentals and prosody in L2 speech ratings. Experienced teacherraters provided lengthier reports than novice raters, producing more comments for all coded categories where an error was identified except for pausing (a disfluency marker). This may be because novice raters observed little else about the speech or struggled to pinpoint or articulate other features.