Developing a Real-Time Test to Investigate Conversational Speech Understanding.

Area of Science:

• Speech perception research
• Auditory processing
• Cognitive psychology

Background:

• Understanding continuous speech in real-world environments is challenging.
• Assessing speech comprehension in noisy conditions requires robust measures.
• Naturalistic dialogue provides a ecologically valid context for speech understanding research.

Purpose of the Study:

• To develop and validate a novel measure for real-time continuous speech understanding.
• To evaluate the measure's sensitivity to varying noise levels and semantic category characteristics.
• To assess the measure's ability to capture higher-level linguistic and discourse processing.

Main Methods:

• A category monitoring paradigm was used with five natural dialogue recordings.
• Participants (N=63 young adults) identified target words within semantic categories (family, media, season, temperature, travel).
• Testing occurred in three noise conditions: quiet, 0 dB, and -5 dB signal-to-noise ratio (SNR), measuring accuracy and response time.

Main Results:

• Speech understanding performance significantly differed across noise conditions, decreasing with lower signal-to-noise ratios (SNR).
• Response accuracy and speed were influenced by semantic category breadth and target word typicality.
• Lower hit rates and slower response times were observed for broader categories and less typical target words.

Conclusions:

• The developed measure effectively captures real-time speech understanding in varying noise levels.
• The measure demonstrates sensitivity to semantic category complexity and word typicality.
• The findings support the measure's utility in evaluating higher-level semantic, contextual, and discourse processing during natural speech comprehension.