Abstract
The paper presented an analysis of the distribution of silent and filled pauses in relation to several structural and usage factors. The analysis is based on a subcorpus of narrative and descriptive discourse of 11 speakers with chronic aphasia (a total of 1118 conversational units).
Conversational units containing one or more disfluencies were extracted and annotated. Sequences of the type "preceding word (PW) - disfluency - target word (TW)" were annotated for TW retrieval success, their position within the unit and syntactic structure (phrase type, clause/phrase boundary or phrase-internal position).
PW and TW were coded for word class, frequency and transitional probabilities, based on a corpus of Czech subtitles. The analysis focused specifically on prepositional phrases produced by the participants.