Measuring intelligibility in spontaneous speech using syllables perceived as understood

Area of Science:

• Speech-language pathology
• Linguistics
• Audiology

Background:

• Speech intelligibility is crucial in speech-language pathology.
• Assessing intelligibility, especially in spontaneous speech, presents significant challenges.
• Existing transcription-based methods can be time-consuming.

Purpose of the Study:

• To investigate the validity and reliability of the syllables perceived as understood (SPU) method for assessing speech intelligibility.
• To evaluate SPU as a more time-efficient alternative to traditional transcription methods.
• To compare SPU with the gold standard syllables correctly understood (SCU) measure.

Main Methods:

• 20 adult listeners transcribed spontaneous speech from 16 children (14 with speech sound disorder).
• Intelligibility was measured using proportion of syllables perceived as understood (SPU) and proportion of syllables correctly understood (SCU).
• Validity was assessed by correlating SPU and SCU; reliability was analyzed using inter-listener intra-class correlation.

Main Results:

• A strong, statistically significant correlation was found between SPU and SCU, with SPU scores consistently higher.
• Inter-listener reliability for single SPU measures was moderate to low.
• Inter-listener reliability for average SPU measures was high.

Conclusions:

• The SPU method can be a valuable clinical and research tool for speech intelligibility assessment.
• Using the median from multiple listeners or comparing a single listener over time enhances SPU reliability.
• SPU offers a more time-efficient and valid alternative to rating scales, though SCU remains more reliable.