Presentation

Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Poster Slams

Activity in the Subthalamic Nucleus Indexes the Coupling of Articulation and Vocal Intensity

Poster E23 in Poster Session E, Saturday, October 8, 3:15 - 5:00 pm EDT, Millennium Hall

Christina Dastolfo-Hromack^1,2, Witek Lipski², Alan Bush^3,4, Jason Bohland², Robert S Turner², Scott Fraundorf², Lori L Holt⁵, Julie Fiez², Susan Shaiman², R Mark Richardson^3,4; ¹West Virginia University, ²University of Pittsburgh, ³Harvard Medical School, ⁴Massachusetts General Hospital, ⁵Carnegie Mellon University

Introduction: In speech therapy for Parkinson’s disease, patients are instructed to speak loudly, which induces improvements in articulation. This powerful therapeutic effect relies on phonetic-intensity encoding, defined as the theoretical process that links articulation and vocal intensity processing. Neurological correlates of phonetic-intensity encoding are unknown, leaving gaps in speech therapy mechanisms and speech production models. Objectives were to 1) Investigate the presence of phonetic-intensity encoding in the basal-ganglia cortical loop and 2) Evaluate the relationship between intensity encoding and established articulatory topography. Methods: 20 patients with Parkinson’s disease were tested during the awake portion of deep brain stimulation implantation surgery. Patients spoke three syllable non-words which were constructed from 4 consonants (/g, t, s, v/) and 3 vowels (/i, a, u/). Non-words were produced loudly and softly, via experimental manipulation. Local field potentials (LFPs) were recorded from the subthalamic nucleus (STN) and precentral gyrus (PreCG), as well as acoustic productions. Acoustic analyses were conducted to determine 1) vocal intensity variation between trials and 2) changes in articulatory precision with increasing vocal intensity. Articulatory precision was measured by obtaining the spectral centroid, second formant ratio, vowel duration and fricative duration, which were then correlated to vocal intensity. Neural power in gamma band (70-150 Hz) and theta band (4-8 Hz) were extracted from the LFPs using a Wavelet transformation. To address the first objective, mixed-effects models were fit in each region and frequency band combination. Neural power was predicted from phoneme, vocal intensity, electrode location and all possible interactions. To address the second objective, a functional analysis was completed to determine electrodes that responded to lip, tongue or larynx articulator movements in the PreCG. Data from these articulator electrodes were then included in a mixed-effects model predicting neural power from vocal intensity, phoneme identity, articulator, and all possible interactions. Results: All measures of articulatory precision significantly correlated with vocal intensity (Bonferroni-corrected), confirming behaviorally that phonetic-intensity encoding occurred. Consistent with prior literature, fricative duration negatively correlated with vocal intensity while vowel duration, second formant ratio and spectral centroid positively correlated with vocal intensity. Neurological evidence of phonetic-intensity encoding was found in a significant three-way interaction between vocal intensity, phoneme, and electrode location in STN theta band. A main effect for vocal intensity was found in the STN theta band. No evidence for phonetic-intensity encoding was found in the PreCG, but interaction effects for electrode location by vocal intensity and electrode location by phoneme were found in gamma band. Results from the articulator model in the PreCG showed an interaction effect between vocal intensity and articulator. As vocal intensity increased, gamma band power in lip and tongue electrodes increased while power in larynx electrodes decreased. Conclusions: Phonetic-intensity encoding occurs as vocal intensity-dependent increases in power at phoneme-specific locations in the STN, but not PreCG. Vocal intensity control may be influenced by articulator regions in the PreCG. Interactions between STN and PreCG effects may occur via gamma-theta coupling, requiring future research. Explicit links between articulation and intensity should be incorporated into speech motor control models.

Topic Areas: Speech Motor Control, Prosody