The development of Cantonese matrix sentence test

Abstract

Objective: To develop the Cantonese Matrix Sentence Test as a sentence-in-noise test for Cantonese native speakers in accordance with the international recommended procedure. Design: The development of the Cantonese matrix sentence test was divided into three separate experiments: (1) synthesis of speech test material (2) optimisation of speech material (3) evaluation on the test-specific properties. Initially, a 50-word Cantonese base matrix was established and combined into sentences. Individual word-specific speech reception thresholds (SRTs) and slopes were measured for all 500 word realisations. Homogenisation of the speech material was then conducted by adjusting the presentation level of the word realisations. The maximum level adjustment was limited to ± 3 dB. Afterwards, the optimised matrix test was evaluated on the properties of speech material, the effect of training, test-list equivalence, and establishing normative data for normal-hearing Cantonese native speakers. A total of 48 normal-hearing native Cantonese speakers participated in this study. Results: SRT measurement with adaptive procedure resulted in a reference speech reception threshold of -9.7 ± 0.7 dB SNR using open-set response format. Fixed SNR measurements suggested a test-specific speech intelligibility function slope of 15.5 ± 0.7 % per dB. Seventeen 10-sentences base test lists were confirmed to be equivalent with respect to speech intelligibility. A training effect was not observed after the first two measurements with 20-sentences lists. Conclusion: The Cantonese matrix sentence test was developed to measure sentence intelligibility in quiet and in noise and yields results comparable to matrix tests in other languages.