Abnormalities of connected speech in the non-semantic variants of primary progressive aphasia

Background: The importance of connected speech analysis in the diagnosis and further classification of primary progressive aphasia (PPA) is becoming apparent. However, methodological inconsistencies in elicitation and analysis of speech samples hinder comparison of different laboratories' results. The recent recommendations for further classification of PPA have characterised three main variants, but we have demonstrated that a considerable proportion of patients do not conform to any of these, as their language deficits extend beyond those of a single syndromic variant (mixed PPA). This report is an attempt to characterise the language profiles of different non-semantic variants of PPA using a combination of connected speech analysis and standard neuropsychological tests.Aims: Our aims were (a) to assess the relative efficacy of semi-structured interviews and picture description tasks in revealing the important aspects of language deficit in our patient groups, (b) to document the language profile of a mixed PPA group for the first time, (c) to compare the patterns of speech abnormality in non-fluent variant and mixed PPA, relative to each other and to normal control participants, and (d) to identify those features of spontaneous speech decline that are frequent and tangible enough to be detected during clinical consultations.Methods & Procedures: Connected speech samples obtained from picture description and semi-structured interviews were analysed quantitatively and the outcomes were contrasted between the elicitation methods and among the participant groups. Due to a very small number of cases fulfilling criteria for logopaenic PPA, the speech of these patients was scrutinised individually.Outcomes & Results: Pair-wise comparisons of the two speech elicitation methods at a group level revealed only a partial concordance and greater sensitivity of semi-structured interviews in detecting different aspects of speech abnormality. At a group level there was a significant impairment in the morpho-syntactic aspects of connected speech in both non-fluent and mixed PPA. A statistically significant difference from controls was observed in the semantic measures of discourse only in the mixed PPA group. Clinically useful features that differentiate both non-fluent and mixed PPA groups from healthy volunteers were decreased speech rate, shorter utterances, fewer complex grammatical structures, lack of spontaneity, more non-sentential speech units (elliptical and abandoned units), more hesitation markers, and, to a lesser extent, more phonological errors and editing breaks.Conclusions: The combination of the present results with those from our previous report on semantic dementia that used the same methodology offers a quantitative detailed assessment of connected speech across the spectrum of PPA