How to Compare Automatically Two Phonological Strings: Application to Intelligibility Measurement in the Case of Atypical Speech

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 1689–1694 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 1689 How to Compare Automatically Two Phonological Strings: Application to Intelligibility Measurement in the Case of Atypical Speech Ghio1 A., Lalain1 M., Giusti1 L., Fredouille2 C., Woisard3 V. (1) Aix-Marseille Univ, LPL, CNRS, Aix-en-Provence, France (2) LIA, Avignon University, France (3) UT2J, Octogone-Lordat, Toulouse University & Toulouse Hospital, France alain.ghio@lpl-aix.fr, muriel.lalain@lpl-aix.fr, corinne.fredouille@univ-avignon.fr, woisard.v@chu-toulouse.fr Abstract Atypical speech productions, regardless of their origins (accents, learning, pathology), need to be assessed with regard to "typical" or "expected" productions. Evaluation is necessarily based on comparisons between linguistic forms produced and linguistic forms expected. In the field of speech disorders, the intelligibility of a patient is evaluated in order to measure the functional impact of his/her pathology on his/her oral communication. The usual method is to transcribe orthographic linguistic forms perceived and to assign a global and imprecise rating based on their correctness or incorrect. To obtain a more precise evaluation of the production deviations, we propose a measurement method based on phonological transcriptions. An algorithm computes automatically and finely the distances between the phonological forms produced and expected from cost matrices based on the differences of features between phonemes. A first test of this method among a large population of healthy speakers and patients treated for cancer of the oral and pharyngeal cavities has proved its validity. Keywords: intelligibility, phonological features, clinical phonetics 1. Introduction We call "atypical speech" oral utterances deviating from a regular form in their pronunciation. If the notion of standard pronunciation remains a questionable notion when it touches on regional or sociologically marked forms of speech, it remains accepted in the case of learners of a foreign language (Kang et al., 2018) or in the case of motor- speech disorders or organic speech sound disorders. In functional assessment of patients with speech disorder, intelligibility is a key parameter in, for example, dysarthria (Kent, 1992), head and neck cancer (Meyer et al., 2004) or speech production after cochlear implantation. Several methods of speech perception assessment are available to measure the severity of speech production disorders. Kent (1992) defines intelligibility as “the degree to which the speaker's intended message is recovered by the listener”. This author defines also as “item identification” the perceptual objective measurement, which is usually the percentage of items that are accurately recognized by a listener. In such a context, the atypical speaker can be asked to read a list of words or phrases, and the examiner writes down what (s)he has understood; the transcription is compared against the original list, and a score is calculated as the percentage of correctly understood items. The transcription of the production as well as the target form are generally available in an orthographic form. However, if we are interested primarily in oral production and as this oral production is sometimes restricted to isolated words, it is ultimately more important to place the analysis at the …

Content truncated to 500 words