Presentation
Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Lightning Talks
Iceberg or cut off – how adults who stutter articulate fluent-sounding utterances
Poster E72 in Poster Session E, Thursday, October 26, 10:15 am - 12:00 pm CEST, Espace Vieux-Port
Martin Sommer1, Andreas Leha2, Nicole Neef3, Susanne Dickhut1, Daniela Ponssen1, Annika Primassin4, Alexandra Korzeczek1, Arun Joseph5, Walter Paulus6, Jens Frahm5; 1Department of Neurology, University Medical Center Göttingen, Germany, 2Department of Medical Statistics, University Medical Center Göttingen, Germany, 3Department of Diagnostic and Interventional Neuroradiology, University Medical Center Göttingen, Göttingen, Germany, 4FH Münster University of Applied Sciences, Fachbereich Gesundheit, Münster, Germany, 5Biomedical NMR, Max-Planck-Institute for Multidisciplinary Sciences, Göttingen, Germany, 6Department of Neurology, LMU Hospital, LMU, Munich, Germany
It is unclear where fluent speech ends and where stuttering begins, i.e., whether the border between fluent speech and stuttering is a ‘gradual transition’ or an abrupt ‘all or none’ phenomenon. The gradual transition assumption would mean that stuttering is a continuous phenomenon audible only when reaching a certain threshold. This view has been illustrated as an iceberg, where the visible part is only a small fraction of the true phenomenon. By contrast, the ‘all or none’ view which we call ‘cutoff’ implies that stuttering events are categorically different from fluent speech. A distinction between these two cases is vital to facilitate the disorder's neurophysiological basis and instruct therapeutic strategies. We used real-time MRI (rtMRI) to characterize the speech movements of inner articulators in 15 adults who stutter (AWS) and 17 fluent speakers (FS). Participants uttered the nonword "natscheitideut" 15 times in a 3 T MRI scanner while recording rtMRI videos at 55 frames per second in a mid-sagittal plane. Only fluent productions were considered for analysis. We defined fluent by the absence of typical stuttering characteristics such as repetitions of sounds, sound prolongations, or blocks. We discarded productions with longer pauses or hesitations between two words or within a word if they led to a pause of more than 250 ms and recordings with pronunciation errors. A customized MATLAB toolkit was used for the extraction of line profiles from MRI videos to quantify lip, tongue, and velum movements. We used penalized flexible, functional regression for modeling the movement of the tip of the tongue, lips, and velum using the group as an independent predictor and the subject as a random effect. We assessed the group effect on the movement patterns. The patterns differed significantly between subjects, but only slightly so, and showed a high similarity across subjects. Hence, movement patterns were distinguishable but still very similar across subjects, and only slight differences between groups were observed. We further analyzed whether participants were grouped due to distinguishable movement patterns across the three articulators. Therefore, we calculated all pairwise distances between the profile lines for each participant. The hierarchical clustering of participants based on these distances resulted in 3 clusters. Cluster one included predominantly AWS (11 AWS and 3 FS), and cluster two (1 AWS, 4 FS) and cluster three (2 AWS, 10 FS) included predominantly FS. An additional principal component analysis detected the same clusters. Here, the distances of the tip of the tongue to the other articulators explained the variance in the first component. To summarize, differing articulatory patterns separated AWS from FS. This distinction was evident in AWS` fluent productions of the nonword "natscheitideut". Our observation suggests that the transition between fluent and stuttered speech is not abrupt but rather a gradual continuum.
Topic Areas: Speech Motor Control,