Observer agreement on disfluency and stuttering

Groups of undergraduate and graduate student listeners identified the stutterings and disfluencies of eight adult male stutters during videotaped samples of their reading and speaking. Stuttering and disfluency loci were assigned to words or to intervals between words. The data indicated that stuttering and disfluency are not too reliable and unambiguous response classes and are not usually assigned to different, nonoverlapping behaviors. Furthermore, judgments of stuttering and disfluency were distributed similarly across words and intervals. For both undergraduate and graduate student listeners, there was relatively low unit-by-unit agreement among listeners and within the same listeners from one judgment session to another.