Parallel Corpus Analysis of Text and Audio Comprehension to Evaluate Readability Formula Effectiveness: Quantitative...

Area of Science:

• Health Communication
• Medical Informatics
• Human-Computer Interaction

Background:

• Health literacy is crucial for patient outcomes and healthcare system efficiency.
• Audio-based health information delivery is under-explored despite increasing use of digital health tools.
• Traditional readability formulas inadequately assess health text complexity, especially for audio formats.

Purpose of the Study:

• To investigate features influencing medical information comprehension and retention in text versus audio formats.
• To evaluate existing readability formulas' correlation with perceived and actual difficulty of health information across modalities.

Main Methods:

• Developed a parallel corpus of health information in text and audio formats from various sources (BMJ, WebMD, etc.).
• Recruited 487 participants to read or listen to health texts, assessing perceived difficulty, comprehension (MCQ, T/F), and retention (free recall).
• Analyzed syntactic, semantic, domain-specific features, and readability formulas for their relationship with information difficulty.

Main Results:

• Text formats were perceived as easier and yielded higher comprehension and retention than audio formats across all sources.
• Linguistic features like content word frequency and verb usage impacted comprehension and retention, while noun/adjective usage increased difficulty.
• Readability formulas showed limited correlation with actual comprehension or retention, but correlated with perceived text difficulty.

Conclusions:

• Text is more effective for complex health information; audio may suffice for simpler content.
• Specific linguistic features significantly influence information processing irrespective of modality.
• Existing readability formulas are insufficient for assessing the true difficulty of health information, necessitating modality-specific and content-aware approaches.