An Entropy-Based Computer Model for the Measurement of Phonetic Similarity: Dyslalia Screening in Early School-Age Children

This paper presents a computer model for the assessment of the similarity between two sound patterns, to identify phoneme mispronunciations circumscribed by dyslalic disorders in early school-age children (6-10 year olds). From a linguistic standpoint, it is the phonetic tier that is mainly engaged...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Applied medical informatics Ročník 40; číslo 1/2; s. 15 - 23
Hlavní autoři: Mahmut, Emilian Erman, Della Ventura, Michele, Stoicu-Tivadar, Vasile
Médium: Journal Article
Jazyk:angličtina
Vydáno: Cluj-Napoca SRIMA Publishing House 01.06.2018
Iuliu Hatieganu University of Medicine and Pharmacy, Cluj-Napoca
Témata:
ISSN:1224-5593, 2067-7855
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:This paper presents a computer model for the assessment of the similarity between two sound patterns, to identify phoneme mispronunciations circumscribed by dyslalic disorders in early school-age children (6-10 year olds). From a linguistic standpoint, it is the phonetic tier that is mainly engaged in dyslalia. Unlike other speech disorders, which involve meaning-coding and decoding mechanisms (semantics), dyslalia lends itself more easily to mathematical analysis in the screening stage. The method is based on the analysis of the sound waves and on the quantification of the information carried by every single sound pattern, by calculating its entropy. It is an empirical methodology that provides results that may be analyzed. An experimental study was conducted according to the model and method presented on a sample of 30 subjects. The results are assessed and conclusions are issued. The representation using an isometric diagram accommodates a better interpretation of the results.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1224-5593
2067-7855