Including people in the automatic text adaptation process: empirical findings from students with intellectual...

Area of Science:

• Natural Language Processing
• Educational Technology
• Disability Studies

Background:

• Automatic text adaptation often fails to meet diverse reader needs.
• Students with intellectual disability (ID) face unique challenges in text comprehension.
• The TextAD project sought to develop tailored texts for this population.

Purpose of the Study:

• To explore the effectiveness of automatic text adaptations for students with ID.
• To refine adaptations at lexical, syntactic, and discourse levels.
• To assess comprehension, interest, perceived difficulty, and metacognitive insights.

Main Methods:

• Iterative testing of automatic text adaptations across three rounds.
• Assessment of reading comprehension and self-reported measures in 27 students with ID.
• Analysis of lexical, syntactic, and discourse level adaptations in expository texts.

Main Results:

• No significant overall improvement in reading comprehension with adapted texts.
• Promising outcomes observed with lexical and syntactic adaptations in later testing phases.
• Student heterogeneity impacted adaptation effectiveness; high self-reported comprehension but persistent conceptual difficulties noted.

Conclusions:

• Automatic text adaptation shows potential for students with ID, particularly at lexical and syntactic levels.
• Further research is needed to address group heterogeneity and conceptual understanding challenges.
• Insights provided on assessment strategies for developing tailored text adaptations for this audience.