New paper in the Journal of the Phonetic Society of Japan
For the past few years, I have been fortunate to join a project by Prof. Kikuo Maekawa on the crosslinguistic study of fillers. The project has mainly focused on acoustic-phonetic features of Japanese fillers, but he also included --for comparative purposes—some investigation of English and Chinese filled pauses. As a culmination of the project, we have submitted portions of our work to the Journal of the Phonetic Society of Japan in a special issue focusing on the topic of filled pauses.
The collection includes one paper by me focusing on the comparison of form and temporal characteristics of filled pauses in first language (L1) Japanese and second language (L2) English. My interest was in looking at how native Japanese speakers's use of filled pauses in their second language compares to that of their first across L2 proficiency levels. In order to do this, I used the Crosslinguistic Corpus of Hesitation Phenomena as a resource because it provides L1 and L2 speech data from each speaker. Thus, a direct comparison can be made between L1 and L2 speech on an intra-speaker basis. This difference is important as most studies do not do this: They compare different groups of speakers. So, the data is unique in this regard.
The key findings that I report are that Japanese speakers of English use filled pauses with a vowel sound that is higher (lower F1) than native English speaker averages (from previous studies), but, they strongly prefer uh to um. This is especially true for low L2 proficiency speakers who almost never used um. This trend, however, actually reflects a similar trend in L1 Japanese (though more strongly). High L2 proficiency speakers use a vowel that is closer to the schwa [ə] used in native English uh/um, but is still higher.
A curious result arose when looking at delay (i.e., duration of filled pause plus any adjacent silence). Low L2 proficiency speakers show a longer delay with uh than with um (though note that there are a very small number of the latter cases), while high L2 proficiency speakers show the opposite pattern with longer delays with um than with uh. If Clark and Fox Tree are correct, then high L2 proficiency speakers are following the native English norm, while low proficiency speakers are against the norm. But if there is to be any difference between high and low proficiency speakers, this is the direction one would predict: Low proficiency speakers have not yet managed to learn and use the difference between uh and um.
[Note: This post was written in September, 2020. However, in order to preserve the chronology of the blog, it has been dated to reflect when the described events actually took place.]