Why do we sometimes… Hesitate?: A study investigating the variation in prevalence of disfluencies observed in the spontaneous speech of typical speakers.

Abstract

Disfluencies are a common phenomenon in everyday conversational speech, occurring approximately 6 times per 100 words. Previous studies attempting to explore what may affect the frequency of disfluencies in the spontaneous speech of typical speakers address factors such as age, gender and education. However, reliable trends between groups have been difficult to find, highlighting that there is not only huge variation between groups but also within. This study considers that the production of disfluencies such as silent pauses, filled pauses, repetitions, prolongations and repairs may be attributed to difficulties in the planning and formulation stages of spoken language production. Therefore, this study suggests that the variation may be as a result of both personal factors i.e rate and efficacy of an individual's language production processes, and discourse related factors i.e cognitive and psycholinguistic demands the speech task places on the speaker. This study investigates the variation in the prevalence of disfluencies observed in spontaneous speech (average rate and type of disfluencies observed in 2 speech tasks) and how they correlate with rate of lexical access variation. Two speech tasks were completed within this study, with each placing different demands on the speaker. This study found a significant difference between the rates of disfluencies produced in each condition. To determine participants' rate of lexical access 4 verbal fluency tasks were used: 2 semantic and 2 phonological fluency tasks. In this study a significant negative correlation is found between lexical access and the frequency of disfluencies in speech. Specifically, a strong significant negative correlation is established between lexical access abilities and rate of hesitations: suggesting that a reduced rate of lexical access will result in an increased frequency of hesitations. The results of this study indicate that lexical access abilities significantly contribute to the variation in disfluencies found in spontaneous speech. An attempt to explain this relationship is made within this study in relation to Levelt's model of language production. Additionally the results of this study suggest that the variation in cognitive and psycholinguistic demands in the initial planning stage of language production also contribute to the variation in disfluency rates.