The development and validation of the Closed-set Mandarin Sentence (CMS) test

Area of Science:

• Audiology
• Linguistics
• Speech-Language Pathology

Background:

• Evaluating speech intelligibility in noise is crucial for understanding communication challenges.
• Existing sentence tests may not accurately reflect the phonemic distribution of target languages.
• Matrix-styled sentence tests offer a closed-set paradigm for speech intelligibility assessment.

Purpose of the Study:

• To develop and validate the Closed-set Mandarin Sentence (CMS) test for assessing Mandarin speech intelligibility in noise.
• To ensure CMS test materials represent the natural distribution of Mandarin phonemes, including vowels, consonants, and lexical tones.
• To create a large stimulus set for reliable and repeatable measurement of speech understanding in noise.

Main Methods:

• Developed the CMS test using 50 familiar Mandarin words across five categories (Name, Verb, Number, Color, Fruit) to generate 100,000 unique sentences.
• Collected normative data from 10 normal-hearing, adult Mandarin-speaking listeners using a closed-set paradigm.
• Measured speech reception thresholds (SRTs) in noise, with excellent test-retest reliability (Cronbach's alpha = 0.952).

Main Results:

• The CMS test demonstrated excellent test-retest reliability for word-in-sentence recognition.
• Mean SRTs were -7.9 dB (Run 1) and -8.1 dB (Run 2), with a mean standard deviation of 0.93 dB across runs.
• No significant difference was found between test runs, indicating high reliability despite random sentence generation.

Conclusions:

• The Closed-set Mandarin Sentence (CMS) test is a reliable tool for assessing Mandarin speech intelligibility in noise.
• The test provides a large, adaptable stimulus set suitable for repeated and dependable measurements.
• CMS facilitates accurate evaluation of speech understanding in noise for Mandarin speakers.