Optimizing a Vietnamese Sentence Repetition Task Using Item Response Theory

Area of Science:

• Linguistics
• Speech-Language Pathology
• Psychometrics

Background:

• Sentence repetition tasks are crucial for identifying developmental language disorder (DLD).
• Optimization of these tasks for clinical practice, particularly in diverse linguistic contexts like Vietnamese, is limited.
• Item response theory (IRT) offers a robust framework for task optimization.

Purpose of the Study:

• To optimize a Vietnamese sentence repetition task using IRT for both screening and full assessment of DLD.
• To evaluate the diagnostic utility and accuracy of the newly developed item sets.
• To provide clinically applicable tools for identifying DLD in Vietnamese children.

Main Methods:

• Expanded an existing 28-item task to 40 items for administration to 196 children (ages 4-6) in Vietnam.
• Participants included children with DLD (n=28), typical development (n=122), and at-risk (n=46).
• Applied IRT to select optimal item sets for assessment and screening, followed by discriminant function analysis for diagnostic accuracy.

Main Results:

• An optimal 28-item set (15 original items) demonstrated strong diagnostic accuracy for full assessment.
• A 14-item subset (seven original items) showed high accuracy for screening purposes.
• Discriminant function analysis yielded the strongest diagnostic values for both item sets.

Conclusions:

• Successfully optimized Vietnamese sentence repetition tasks for DLD assessment and screening in monolingual children.
• The optimized tasks offer practical utility for clinicians in identifying DLD.
• Future research should explore these tasks with older children and bilingual populations.