Banca de QUALIFICAÇÃO: GUSTAVO OLIVEIRA TOMAZ

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
STUDENT : GUSTAVO OLIVEIRA TOMAZ
DATE: 01/10/2025
TIME: 15:00
LOCAL: Plataforma Google Meet
TITLE:

Development and Validation of a Language Environment Analysis System for Children with Hearing Impairment

KEY WORDS:

Child Language; Hearing Loss; Evaluation Study; Parent-Child Relations; Technological Development and Innovation Projects.

PAGES: 48
BIG AREA: Ciências da Saúde
AREA: Fonoaudiologia
SUMMARY:

Family–child interaction is a key determinant of early development, particularly in the first years of life. The child’s linguistic environment shapes communicative and language outcomes and, in clinical populations such as children with hearing loss, is one of the strongest developmental predictors. In low- and middle-income countries, however, research on natural language environments remains scarce. Although advanced tools for capturing and analyzing such environments exist in high-income settings, there is an urgent need for affordable, user-friendly technologies tailored to the Brazilian context, particularly for integration into the Unified Health System (SUS). This work was structured into two studies. The first was a methodological development study aimed at creating a prototype system for language environment analysis (the ECO System) as a minimum viable product (MVP) and assessing its accuracy by comparing automated transcriptions and classifications of caregiver–child interactions in families of children with hearing loss with manual analyses conducted by a trained researcher using ELAN software. The second was a cross-sectional, exploratory validation study comparing the language and auditory environments of children with hearing loss and of typically hearing peers, using the ECO System against human transcription as the benchmark. In Study 1, recordings from three caregiver–child dyads (ages 3, 6, and 12 years) were transcribed manually in ELAN and automatically by ECO. Comparisons of word counts and grammatical categorization employed intraclass correlation coefficients (ICC) and Bland–Altman plots. Results showed high agreement (ICC = 0.95), with stronger accuracy for nouns and adjectives, whereas verbs and informal expressions showed greater divergence. Study 2 aims to analyze ECO transcriptions from 20 children with bilateral sensorineural hearing loss of varying degrees and 20 age-matched children with typical hearing and development. Expected outcomes include the consolidation of the ECO software’s first version for naturalistic language environment analysis and deeper insights into the role of this environment in shaping child language development.

COMMITTEE MEMBERS:
Presidente - 1717611 - JOSELI SOARES BRAZOROTTO
Interna - 2321749 - ANA MANHANI CACERES
Externa à Instituição - ADRIANE LIMA MORTARI MORET - USP