You are viewing the SNL 2018 Archive Website. For the latest information, see the Current Website.

Poster A49, Thursday, August 16, 10:15 am – 12:00 pm, Room 2000AB

Automatic speech analysis technology yields reproducible dysprosodic markers in Primary progressive aphasia

Naomi Nevler1, Sharon Ash1, David J Irwin1, Molly Ungrady1, Mark Liberman1, Murray Grossman1;1University of Pennsylvania

Objective To identify and quantify specific disease biomarkers of prosody from the acoustic properties of speech in patients with primary progressive aphasia, utilizing automatic speech analysis techniques. Methods 59 digitized speech samples were collected from patients with primary progressive aphasia (PPA, non-fluent/agrammatic=15, semantic=21, logopenic=23; ages 50-85 years, 39% males) and 31 matched healthy controls (ages 54-89 years, 36% males). These audio samples were analyzed with a novel, automated speech analysis protocol that relies on automatic speech activity detection. Acoustic measurements of prosody were extracted and calculated, including fundamental frequency (f0) and speech and silent pause durations. We compared these acoustic features between groups and then examined their relationships with clinical tests, gray matter atrophy, and cerebrospinal fluid analytes. Results We found a narrowed f0 range in patients with non-fluent/agrammatic variant aphasia (mean 3.86±1.15 semitones) compared with healthy controls (6.06±1.95 semitones; p<0.001) and patients with semantic variant aphasia (6.12±1.77 semitones; p=0.001). Mean pause rate was significantly increased in the non-fluent/agrammatic group (mean 61.4±20.8 pauses per minute) and the logopenic group (58.7±16.4 pauses per minute) compared to controls (mean 32.24±9.75 ppm; p≤0.002 per contrast). Narrowed f0 range was associated with atrophy in the left inferior frontal cortex. Cerebrospinal level of phosphorylated-tau (p-Tau) was associated with an acoustic classifier combining f0 range and pause rate (r=0.58, p=0.007). Receiver Operating Characteristic analysis with this combined classifier distinguished non-fluent/agrammatic speakers from healthy controls (AUC=0.94) and from semantic variant patients (AUC=0.86). Conclusions Restricted f0 range and increased pause rate are characteristic dysprosodic markers of speech patterns in non-fluent/agrammatic PPA. These acoustic markers can be extracted automatically from the audio signal and are associated with left inferior frontal atrophy and cerebrospinal p-tau level.

Topic Area: Language Disorders