Analysis of atypical prosodic patterns in the speech of people with Down syndrome. (August 2021)

• Record Type: Journal Article
• Authors: Corrales-Astorgano, Mario; Escudero-Mancebo, David; González-Ferreras, César; Cardeñoso Payo, Valentín; Martínez-Castilla, Pastora
• Is Part Of: Biomedical signal processing and control. Volume 69(2021)
• Publication Date: 2021-08
• Subjects: Down syndrome speech -- Prosody assessment -- PEPS-C -- Sequential feature selection; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2021.102913
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: Method for the analysis of atypical prosodic patterns of Down syndrome (DS) speakers. Automatic identification of differences between DS and typical speakers. Atypical speech related biosignal features in DS speakers. Implications for intervention to improve communication skills of speakers with DS. Abstract: The speech of people with Down syndrome (DS) shows prosodic features which are distinct from those observed in the oral productions of typically developing (TD) speakers. Although a different prosodic realization does not necessarily imply wrong expression of prosodic functions, atypical expression may hinder communication skills. The focus of this work is to ascertain whether this can be the case in individuals with DS. To do so, we analyze the acoustic features that better characterize the utterances of speakers with DS when expressing prosodic functions related to emotion, turn-end and phrasal chunking, comparing them with those used by TD speakers. An oral corpus of speech utterances has been recorded using the PEPS-C prosodic competence evaluation tool. We use automatic classifiers to prove that the prosodic features that better predict prosodic functions in TD speakers are less informative in speakers with DS. Although atypical features are observed in speakers with DS when producing prosodic functions, the intended prosodic function can be identified by listeners and, in most cases, the features correctly discriminate the function with analytical methods. However, a greater difference between the minimal pairs presented in the PEPS-C test is found for TD speakers in comparison with DS speakers. The proposed methodological approach provides, on the one hand, an identification of the set of features that distinguish the prosodic productions of DS and TD speakers and, on the other, a set of target features for therapy with speakers with DS.