Abnormalities of connected speech in semantic dementia vs Alzheimer's disease.

Background: Neurodegenerative syndromes are associated with varying degrees of language impairment. Connected speech analysis provides the most realistic measure of language function but its use has been restricted by operational constraints. Aims: In this prospective study we assessed the relative utility of a picture description task and a semi-structured interview in exposing the language decline in semantic dementia (SD) and typical mild Alzheimer's disease (AD), compared to each other and to healthy volunteers. Methods & Procedures: Our cohort comprised 16 patients with a clinical diagnosis of SD, 20 with mild typical AD, and 30 healthy participants. All speech samples were recorded during a face-to-face interview and were subjected to a well-controlled quantitative analysis. Outcomes & Results: Our results suggest that (a) the picture description task is more sensitive to abnormalities in the semantic measures, whereas the interviews are better at exposing morpho-syntactic deficits; (b) circumlocution is not, as has sometimes been claimed, a salient feature of speech in SD; (c) increases in the frequency of hesitation markers and of phonological and syntactic errors are prominent features of language decline in mild AD; and (d) a 150-word interview sample is adequate to provide a realistic reflection of language impairment. Conclusions: This study compared, for the first time, the two most commonly used methods of eliciting connected speech and documented many similarities in results from the two but also some salient differences in their sensitivity to specific aspects of language deficit. In addition, although replicating some aspects such as anomia, of what is already known about the language impairments in these two conditions, the study provides novel findings for both that could help clinicians in syndromic classification of anomic speech. [ABSTRACT FROM AUTHOR]