Aplicabilidade da análise de grafos na fala espontânea de crianças com Transtorno de Linguagem

Abstract

Objective: To assess Lexical Diversity and connectivity in the spontaneous speech of children undergoing speech therapy, and compare the parameters considered in the different types of narratives while investigating whether there is an adequate change according to the configuration of the organization of the sequences. Methodology: This is a prospective cross-sectional observational study comprising eight children between 5 and 11 years old, Brazilian Portuguese native speakers, with preserved speech intelligibility and diagnosed with interventional language disorder for at least two years. The data is a collection of Spontaneous Speech samples based on seven logical-temporal sequences of the Mechanical, Behavioral and Intentional types. Using the software SpeechGraphs, we analyzed the speech transcriptions and extracted Lexical Diversity and Connectivity features. Results: Lexical diversity showed little variation in the number of nodes, while word count and edges had greater amplitude. As for connectivity, the largest connected component had less variation than the largest most strongly connected component. The Intentional sequences compared to the Mechanical ones obtained better results in all the parameters and compared to the Behavioral ones, they presented better answers for Word Count, Number of Edges and Largest Connected Component. Only the Behavioral sequences showed differences in the connectivity parameter for the Largest Most Strongly Connected Component when they were presented previously organized. Conclusion: The subjects showed greater lexical diversity and connectivity when narrating Intentional type sequences. There was no significant difference between the sequences previously organized and those requested to organize. The analysis is promising for Speech-Language Pathology and Audiology due to its objective assessment of language.