Linguistic processing in narrative production: Evidence from the errors and self-initiated repairs produced by normal and language-disordered school-age children

Date of Completion

January 1989


Language, Linguistics|Health Sciences, Speech Pathology|Psychology, General




This study presents an approach to narrative analysis which focuses on verbal errors, repetitions, and self-initiated repair behaviors as one way of probing the nature of monitoring during narrative production. A review of the literature shows a strong tradition for the analysis of hesitation phenomena, slips of the tongue, and speech errors as signals of linguistic processing. Conclusions are that these behaviors offer "windows" on the self-monitoring strategies subjects use during verbal production. Studies of verbal processing in language disordered populations have been primarily restricted to assessments of comprehension (rather than production) monitoring, and to studies based on older models of language which do not account for processing in extended units of discourse.^ A procedure, the Departure-Repair Analysis, was developed for the present research in order to mark behaviors indicative of verbal self-monitoring processes. Fourteen normal and 14 language-disordered children (CA: 8;6-12;6) viewed a film, retold the story of the film to a naive listener, and answered comprehension questions about the film. Each narrative was transcribed verbatim (including all repetitions, asides, and revisions). Groups were compared on measures of story comprehension, narrative length, mean utterance length, grammaticality, departure characteristics (type, acceptability), and repair characteristics (type, function, success, quality).^ The results indicated good comprehension of the film story by both groups. Both produced narratives which were similar in overall length. As expected, the language-disordered group demonstrated power grammaticality and produced utterances with shorter mean T-Unit lengths. Language-disordered subjects repaired agrammatic departures less frequently than the normal children. Both groups directed more repair attempts to text meaning than to language form. Group performance differences were apparent for the quality of the repair activity, specifically as it related to lexical and cohesive levels of text monitoring.^ The results supported use of the Departure-Repair Analysis procedure as a viable measure of verbal self-monitoring. Conclusions were that the language-disordered children demonstrated underlying metalinguistic competencies similar to normal peers, but they lacked the skills to use language in ways which adequately demonstrated those competencies. ^