Latent Semantic Analysis Discriminates Children with Developmental Language Disorder (DLD) from Children with Typical...

Area of Science:

• Computational linguistics
• Developmental psychology
• Speech-language pathology

Background:

• Objective assessment of semantic-linguistic quality in narratives is challenging.
• Traditional language analyses often focus on form or word level, potentially missing semantic nuances.
• Developmental Language Disorder (DLD) affects language acquisition and impacts narrative skills.

Purpose of the Study:

• To evaluate the efficacy of a novel Semantic Language Impairment Index (SELIMI) in differentiating children with DLD from typically developing peers.
• To explore the potential of Latent Semantic Analysis (LSA) for objective, text-level narrative quality assessment.
• To determine if SELIMI can serve as a complementary tool to existing language assessment methods.

Main Methods:

• Spoken narratives were collected from 54 children with DLD and 54 age-matched controls.
• Latent Semantic Analysis (LSA) was employed to create semantic representations (vectors) of the narratives.
• A statistical model was trained using these semantic vectors to discriminate between the DLD and control groups.

Main Results:

• The SELIMI, derived from LSA, demonstrated a significant ability to distinguish between children with DLD and those with typical language development.
• The discriminative power of SELIMI was significantly better than chance.
• SELIMI showed potential as an objective measure of semantic-linguistic quality at the text level.

Conclusions:

• SELIMI, based on LSA, offers a promising objective method for assessing semantic-linguistic quality in children's narratives.
• This computational approach can potentially complement traditional, form-focused, or word-level language assessments.
• SELIMI may aid in the identification and understanding of language impairments in children.