Slide Slam G3
An integrated approach to speech and silence: Combining linguistic, neuropsychological, and anatomical data
Georgia Angelopoulou1, Dimitrios Kasselimis1, Michel Rijntjes2, Marco Reisert3, Dimitrios Tsolakopoulos1, Georgios Papageorgiou1, Maria Varkanitsa4, Georgios Velonakis5, Efstratios Karavasilis5, Varvara Pantoleon5, Nikolaos Kelekis5, Dionysios Goutsos6, Cornelius Weiller2, Constantin Potagas1; 1Neuropsychology and Language Disorders Unit, Eginition Hospital, 1st Department of Neurology, National and Kapodistrian University of Athens, Greece, 2Department of Neurology and Clinical Neurosciences, Faculty of Medicine, University of Freiburg, Germany, 3Department of Medical Physics, Faculty of Medicine, University of Freiburg, Germany, 4Sargent College of Health & Rehabilitation Sciences, Boston University, Boston, 52nd Department of Radiology, General University Hospital “Attikon”, National and Kapodistrian University of Athens, Greece, 6Department of Linguistics, School of Philosophy, National and Kapodistrian University of Athens, Greece
This study investigated pause-related variables and speech/articulation rate (henceforth SR and AR) in two different speech genres, in healthy individuals. Our aim was threefold. First to investigate whether SR and AR, as well as silent pauses’ frequency and duration significantly differ in two narrative tasks. Then, to explore whether silent pauses’ frequency is related to performance in specific neuropsychological tests. Finally, to assess the possible predictive value of anatomical indices for silent pauses’ duration for each narrative task. Sixty-five neurotypicals (33 males), 25–65 years old, right-handed, monolingual Greek, were recruited. Speech elicitation tasks consisted of a personal story narration and a picture description. Speech samples were analyzed and AR, SR, silent pauses’ frequency and duration were calculated. All participants underwent a neuropsychological assessment, including language, verbal memory and short/working memory tasks. Neuropsychological scores derived from participants’ assessments were entered into a PCA with Promax rotation to reduce the number of neuropsychological scores. 3D-T1w and DTI images were acquired. T1s were analysed using Free surfer and DTI following the global tractography approach, DTI&Fibertools toolbox. GLM models indicated that articulation rate remains stable across narrative tasks, while speech rate is significantly increased in picture description [t(64)=-2.02,p=.0476], and pause frequency and duration are significantly increased in story narration [t(64)=3.814,p=.00031 and t(64)=2.926,p=.00475]. PCA analysis yielded a three-factor solution, interpreting 64.32% of variance in participants’ performance. The first (36.87%) included verbal learning tasks. The second (15.34%) contained verbal short-term/working memory tasks and the third (12.10%) included naming, receptive vocabulary, and semantic fluency. Exploratory regression models were conducted between factors’ individual values and total silent pauses’ frequency, pauses prior to nouns and prior to verbs, separately for each speech genre. The only significant model was for silent pauses prior to nouns from picture description [F(6,64)=3.258,p=.008], as it predicted 25% of the variance (R2=.252). Within this model, only the third factor appeared as a significant predictor (β=-.5231,p=.00086). Finally, separate linear mixed effects models were conducted for each speech genre including individual durations of silent pauses as dependent variables, cortical and white matter indices as factors and participants as random effects. The best fitting models for fractional anisotropy (FA) and tract volume included left arcuate, extreme capsule and middle longitudinal fasciculi, with demographics as covariates. Results for picture description indicated that FA [t(51)=-2.770,p=.00782] and volume [t(49)= 2.532, p =.0146] of the arcuate fasciculus can successfully predict pause duration. Extreme capsule’s volume could successfully predict pause duration [t(50)=3.238,p=.00214] for personal story. The best fitting model for picture description and cortical regions included the surface area of inferior parietal lobule, superior and middle temporal gyri, pars triangularis and opercularis and insula and revealed only pars opercularis as a significant predictor [t(46)=-2.228,p=.0308]. Pars triangularis had a significant effect [t(60)=-2.097,p=.04020] on silent pause duration for personal story. Overall, by combining linguistic, neuropsychological and anatomical data we argue that different narrative tasks may be associated with distinct aspects of cognition and brain anatomy in healthy individuals, and further support the notion that silent pauses’ may reflect underlying cognitive processes during speech flow.