Many if not most language perception phenomena have listening and reading correlates in psycholinguistic experimentation. In the perception of filled pauses (e.g., uh/um in English, e-to/ano in Japanese), experimental work has thus far looked only at filled pauses in speech showing that hearers process incoming speech differently when there are filled pauses. The present research therefore was a pilot experiment which sought to examine whether analogous results could be obtained in reading.

The experiment used a self-paced reading task design with three syntactic configurations: subordinate clauses, coordinate clauses, and nouns of high contextual probability. The sentences were presented on scren in either a disfluent (with a filled pause) or fluent condition. The prediction was that the disfluent condition should generate processing difficulty and hence longer reading times.

Results (N=22 native English speakers) showed that longer reading times were obtained in the disfluent condition in the region of interest (immediately after the filled pause). The results do not quite reach statistical significance, however, using mixed-effects modeling with fluency as a fixed effect and users and stimuli as random effects.  Future work will seek to increase the number of participants to that comparable to other studies of similar design.