Event Abstract

Back to Event

Measuring prosodic deficits in oral discourse by speakers with fluent aphasia

Tan Lee1, Anthony Pak Hin Kong2* and Wang-Kong Lam1
  • 1 The Chinese University of Hong Kong, Department of Electronic Engineering, Hong Kong, SAR China
  • 2 University of Central Florida, Department of Communication Sciences and Disorders, United States

Introduction Prosody refers to the part of phonology that includes speech rhythm, stress, and intonation (Gandour, 1998). Proper intonation is crucial for expressing one’s emotion and linguistic meaning. Most studies examining prosodic deficits have been primarily focused on Broca’s aphasia (e.g., Danly & Shapiro, 1982). The current study proposed a computer-assisted method for systematic and objective examination of intonation patterns in aphasic oral discourse. The speech materials were in Hong Kong Cantonese, which is known of being rich in tones. Since surface F0 contour is the result of complicated interplay between sentence-level intonation and syllable-level lexical tones, the challenge is on how to extract meaningful representations of intonation from acoustic signals. Methods Four individuals with fluent aphasia (two anomic and two transcortical sensory) and four gender-, age-, and education-matched controls participated. Based on the Cantonese AphasiaBank protocol (Kong, Law, & Lee, 2009), narrative samples and corresponding audio recordings were collected using discourse tasks of personal monologue, picture and sequential description, and story-telling. There were eight recordings for each subject. Each oral discourse was divided into sentences by manual inspection of the orthographic transcription and the respective acoustic signal. A sentence was defined as a sequence of words that in principle covers a complete thought. However, it was common in spontaneous oral discourse, especially in aphasia, that some of the sentences did not end with a completed expression but switched to a new topic. Occasionally, an obvious interjection was observed during an attempt of restarting a statement. Phoneme-level automatic time alignment was performed on each audio recording using hidden Markov model (HMM) based forced alignment technique (Lee, Kong, Chan, & Wang, 2013). F0 was estimated from the acoustic signal at intervals of 0.01 second by applying three different algorithms, including PEFAC (Gonzales & Brookes, 2011), RAPT (Talkin, 1995), Voicebox (Brookes, 1997), and YIN (de Cheveigné & Kawahara, 2002), and combining their results. For each syllable, the F0 values of all voiced frames were considered. The median value was used as the representative F0 level of the syllable. This was motivated by our interest in intonation-related variation of F0 level over a sentence. Since Cantonese is rich in tones, the surface F0 measurement depends greatly on the tone identities. To address this problem, a novel method of tone normalization that attempts to produce a tone-independent F0 level for each syllable was applied (Li, 2003). Using the normalized F0 values, a sentence-level intonation curve was derived by performing linear regression. A positive slope of the intonation curve implied a trend of increasing pitch. Descriptive statistics on the slopes of intonation curves were conducted for both groups. Results Sentences in the aphasic group were significantly shorter than those by controls. Aphasic speech also contained significantly more sentences with rising intonation; this difference was more noticeable for shorter sentences. Concerning the effect of discourse task on speech prosody, story-telling discourses by speakers with aphasia tended to contain more rising intonation than descriptive discourses. Such a difference was absent in controls (see Table 1).

References

Brookes, M. (1997). Voicebox: Speech processing toolbox for matlab. Software, available [Mar. 2011] from www.ee.ic.ac.uk/hp/staff/dmb/voicebox/voicebox.html
Danly, M. & Shapiro, B. (1982). Speech prosody in Broca’s aphasia. Brain and Language, 16, 171-190.
de Cheveigné, A. & Kawahara, H. (2002). YIN, a fundamental frequency estimator for speech and music, Journal of the Acoustical Society of America 111 (4), 1917-1930.
Gandour, J. (1998). Phonetics and Phonology. In B. Stemmer & H. A. Whiteaker, (Eds.), Handbook of Neuro-linguistics (pp. 208-219). New York, NY: Academic Press.
Gonzalez, S. & Brookes, M. (2011). A pitch estimation filter robust to high levels of noise (PEFAC), Paper presented at the Proc. EUSIPCO, Barcelona, Spain.
Kong, A. P. H., Law, S. P., & Lee, A. S. Y. (2009). The construction of a corpus of Cantonese-aphasic-discourse: a preliminary report. Poster presented at the American Speech-Language-Hearing-Association Convention, New Orleans, LA, USA.
Lee, T., Kong, A. P. H., Chan, V., & Wang, H. P. (2013). Analysis of auto-aligned and auto-segmented oral discourse by speakers with aphasia: A preliminary study on the acoustic parameter of duration. Paper presented at the Academy of Aphasia 2013 Annual Meeting, Lucerne, Switzerland.
Li, Y. (2003). Prosody analysis and modeling for Cantonese text-to-speech. Unpublished MPhil Thesis. Hong Kong: The Chinese University of Hong Kong.
Talkin, D. (1995) A Robust Algorithm for Pitch Tracking (RAPT). In W.B. Kleijnm & K.K. Paliwal, (Eds.), Speech coding and synthesis (pp. 495-518). New York, NY: Elsevier.

Keywords: Prosodic deficits, Aphasia, discourse, intonation, Cantonese

Conference: Academy of Aphasia 53rd Annual Meeting, Tucson, United States, 18 Oct - 20 Oct, 2015.

Presentation Type: Poster

Topic: Not student first author

Citation: Lee T, Kong A and Lam W (2015). Measuring prosodic deficits in oral discourse by speakers with fluent aphasia. Front. Psychol. Conference Abstract: Academy of Aphasia 53rd Annual Meeting. doi: 10.3389/conf.fpsyg.2015.65.00047

Copyright: The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers. They are made available through the Frontiers publishing platform as a service to conference organizers and presenters.

The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated.

Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed.

For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions.

Received: 22 Apr 2015; Published Online: 24 Sep 2015.

* Correspondence: Dr. Anthony Pak Hin Kong, University of Central Florida, Department of Communication Sciences and Disorders, Orlando, United States, akong@hku.hk