Analysing speech problems in a longitudinal case study of logopenic variant PPA



A. Hilger; G. Ramsberger; P. Gilley; L. Menn;A. Kong


Authors: contact us about adding a copy of your work at STARS@ucf.edu

Abbreviated Journal Title



Primary progressive aphasia; Logopenic; Speech sound errors; Analysis of; connected speech; Assessment; Diagnosis; Rater perception; PRIMARY-PROGRESSIVE-APHASIA; DISCOURSE; DEMENTIA; DECLINE; Clinical Neurology


Background: Primary progressive aphasia (PPA) is a neurodegenerative form of dementia in which gradually worsening language impairments are the prominent feature in the initial stages. PPA is commonly differentiated into three variants: nonfluent agrammatic (PPA-NVF), semantic (PPA-SV), and logopenic (PPA-LV).Aims: This article provides a longitudinal description of changes in picture description produced by a woman with PPA-LV, introduces a reliable new measure that captures those changes, and relates the measured changes to raters' perceptions of changes in discourse quality.Method & Procedures: Seven oral descriptions of the Boston Diagnostic Aphasia Examination (BDAE) Cookie Theft picture were digitally recorded over the course of 27 months and later transcribed. Transcriptions were analysed using a new adaptation of the Linguistic Communication Measure (LCM) and the Linguistic Communication Measure-Revised Cantonese (LCM-RC) designed to be sensitive to the features of PPA-LV. We have named this third form the LCM, the Linguistic Communication Measure--Speech Sounds (LCM-SS). Audio recordings of the seven picture descriptions plus three produced by typical speakers of similar age were rated for goodness by 15 raters.Outcomes & Results: Goodness ratings of the participants' speech samples decreased steadily over the 27 months. Although our previous measures of discourse quality (LCM, LCMC-RC) appeared to work well for capturing many of the speakers with vascular aphasia, they failed to capture the nature of this participant's decline: Her lexical access slowed over time, but did not become more error-prone, and morphosyntactic components did not worsen, with errors remaining low to almost absent. However, speech sound errors and repetitions increased steadily over the 27 months. The new measure, LCM-SS, succeeded in capturing this pattern of decline: Several of the LCM-SS measures were highly correlated to ratings of goodness, and two of the LCM-SS indices (sound errors and grammatical errors) accounted for 98% of the variance in the goodness ratings.Conclusions: Over the course of 27 months, the most significant change in this participant's Cookie Theft descriptions was the steady increase in sound errors, in the context of decreased efficiency in lexical retrieval and relatively stable grammatical form. This pattern was also highly related to listeners' perceptions of the quality of discourse. Neither of the previous versions of the LCM captured this debilitating increase in sound errors, but adding the index of sound errors to those previous versions resulted in an analysis method that was sensitive to the linguistic features exhibited by this participant with PPA-LV.

Journal Title






Publication Date


Document Type




First Page


Last Page


WOS Identifier