Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Lightning Talks

Connected Speech Profiles in Mild Cognitive Impairment are Related to the Severity of Cognitive Changes

There is a Poster PDF for this presentation, but you must be a current member or registered to attend SNL 2023 to view it. Please go to your Account Home page to register.

Poster B55 in Poster Session B, Tuesday, October 24, 3:30 - 5:15 pm CEST, Espace Vieux-Port

Sophie Pellerin1,2, Christophe Bedetti1,2, Simona Maria Brambati1,2,3; 1Université de Montréal, 2Centre de recherche de l'Institut Universitaire de Gériatrie de Montréal, 3Centre de recherche du Centre intégré universitaire de santé et de services sociaux du Nord-de-l’Île-de-Montréal, Montréal, Canada

Individuals with MCI show heterogeneous cognitive deficits (e.g., memory, visuo-spatial and executive functioning). They also have widespread changes in the following Connected Speech (CS) domains: speech production (e.g., more word fragments), fluency (e.g., more pauses), lexical (e.g., greater proportion of pronouns versus nouns), semantics (e.g., fewer Information Content Units (ICUs)), syntax (reduced syntactic complexity), and pragmatics (e.g., reduced coherence). There also exists variability across studies in the CS characteristics that best distinguish controls and individuals with MCI. Our aim was to find CS profiles in MCI and if these profiles are accompanied by specific cognitive patterns and demographic characteristics. To do so, data from 56 controls and 126 individuals with MCI (60-89 years old) were drawn from the COMPASS-ND study of the Canadian Consortium of Neurodegeneration in Aging. All participants completed the Cookie Theft Picture Description Task, and neuropsychological tests (e.g., BORB, WAIS-III Vocabulary, Stroop) in English. Linguistic features covering the following domains: speech production, fluency, lexical, semantics, syntax, and pragmatics were extracted from MCI’s and controls’ CS transcriptions using Natural Language Processing techniques. To reduce dimensionality, K-means or Two-Step Cluster analyses (depending on the variable types in a category) were carried out for each linguistic characteristic category with the MCI participants’ data. A Two-Step Cluster Analysis was carried out using the previously identified linguistic characteristics without predefining the number of wanted clusters, and a two-cluster solution was obtained. Both clusters were compared to one another and to the control group in terms of linguistic, cognitive, and demographic characteristics (e.g., age, education, number of spoken languages) using one-way ANOVAs and Bonferroni tests. Cluster 2 (17.5% of MCI participants) was more educated than Cluster 1 (82.5%). Compared to controls and Cluster 2, Cluster 1 produced shorter CS samples, had more semantic (e.g., fewer ICUs mentioned), and syntactic (e.g., fewer subordinate clauses) difficulties, and slightly more pragmatic (e.g., reduced local coherence) and lexical difficulties (e.g., higher word frequency). Cluster 2 showed more fluency difficulties (more pauses, word and idea repetitions) than controls and Cluster 1. Also, Cluster 1 had a lower performance than controls and Cluster 2 in the following cognitive domains: visuospatial, language/semantic memory, visual attention, and inhibition. Cluster 2 performed similarly to controls in all cognitive domains, except for some portions of the Verbal Fluency task (more repetitions and set-loss errors). Moreover, both clusters performed more poorly than controls, but similarly to one another on the episodic memory tasks. These results show that there exists more than one CS profile in MCI and highlight the fact that CS analysis can help detect individuals with MCI with more severe cognitive changes. Cluster 1 could potentially be in later stages and/or more at risk of progressing to AD than Cluster 2. Furthermore, Cluster 1’s more severe changes in CS are coherent with the greater severity of their cognitive decline. Cluster 2’s fluency changes could be the result of word-finding/lexical access difficulties. Considering their higher education level, the less severe CS changes observed in Cluster 2 could be attributed to greater cognitive reserve.

Topic Areas: Disorders: Acquired, Language Production

SNL Account Login

Forgot Password?
Create an Account