Poster E50, Saturday, August 18, 3:00 – 4:45 pm, Room 2000AB

Cortical entrainment of continuous speech envelope is preserved in non-fluent variant PPA

Heather Dial1, Benjamin Zinszer1, Bharath Chandrasekaran1, Maya Henry1;1University of Texas at Austin

Introduction Primary progressive aphasia (PPA) is a disorder characterized by a gradual loss of speech and language functions resulting from neurodegenerative disease (Mesulam, 1982). In the nonfluent/agrammatic variant (nfvPPA), the primary features are apraxia of speech and/or agrammatic production, with comprehension deficits observed in a subset of patients (Gorno-Tempini et al., 2011). Recent evidence suggests low-level auditory processing impairments are also present in nfvPPA (Grube et al., 2016). Despite evidence of impairments in auditory processing and language comprehension, there is a paucity of research examining the underlying basis for connected speech comprehension deficits in these patients. In the current study, we examined continuous speech processing in nfvPPA using behavioral and neurophysiological measures. We related EEG signals to a continuous speech stream using temporal response function (TRF) modeling, a recent innovation in EEG analysis that directly compares continuously varying aspects of a stimulus with fluctuations in a participant’s EEG signal (DiLiberto et al., 2015). We used amplitude envelope as the stimulus model, which prior research has related to speech comprehension (e.g., Ding & Simon, 2014). Recent research demonstrated TRF model changes with language impairments, such as dyslexia (DiLiberto et al., 2018). Therefore, this approach has the potential to inform our understanding of the mechanisms supporting comprehension of continuous speech in nfvPPA. Method Five participants with a nfvPPA diagnosis (71-78 years old, M = 75; Mini-Mental State Exam scores ≥ 22), consistent with current diagnostic criteria, and 14 neurally-healthy, age-matched controls (61-77 years old, M = 71) participated in the current study. Participants listened to 15 one-minute segments of the audiobook Alice’s Adventures in Wonderland, and answered two multiple-choice questions after each segment. While participants listened to the story, EEG responses were continuously collected using a 32-channel electrode cap. We compared the amplitude envelope (absolute value of Hilbert-transformed stimulus) to EEG responses using the TRF model. EEG data predicted by the TRF were compared to observed EEG data via Pearson’s correlation, providing a measure of model fit (r). We predicted that nfvPPA participants would (a) demonstrate impaired comprehension of continuous speech, and (b) diverge from healthy age-matched controls in the relation between the EEG and amplitude envelope, reflecting impaired low-level auditory processing. Results Data from individuals with nfvPPA were compared to controls using Crawford and Howell’s (1998) modified t-test. For comprehension accuracy, 3/5 participants scored significantly lower than controls (p’s < .01), indicating impaired comprehension of continuous speech. In contrast, r values relating the envelope to EEG data were not significantly different from controls for nfvPPA participants, suggesting relatively intact low-level auditory processing. Conclusion Overall, we demonstrate intact processing of envelope cues, a feature shown to contribute to speech comprehension, suggesting that language comprehension impairments in nfvPPA are caused by higher-level linguistic processing impairments. One potential reason for the difference between the current study and Grube et al. (2016) is that EEG responses were analyzed independently of behavioral responses, eliminating confounds associated with response decision-making. In ongoing research we plan to examine higher-level linguistic models (e.g., Broderick et al., 2018) in nfvPPA.

Topic Area: Language Disorders