Master of Science
Health and Rehabilitation Sciences
Scollie, Susan D.
The Connected Speech Test (CST) assesses an individual’s ability to understand everyday contextualized running speech amidst competing background babble. To improve the validity and reliability of the test, an updated version was developed, by Saleh et al. (2020), with a General American accent to replace the Southern U.S. accent in the original CST. To evaluate normative performance on this updated test, forty native English-speaking adults with normal hearing were recruited. Multitalker babble was presented at a fixed level, and speech was presented at varying levels to participants in either a co-located or separated loudspeaker condition. At each SNR, participants were scored based on keywords correctly identified. Subjective speech recognition and listening effort were evaluated using a self-report scale. Results showed significantly better performance in the separated loudspeaker condition, indicating a spatial release from masking. For both conditions, performance and subjective speech recognition increased, while listening effort decreased with increasing SNRs.
Summary for Lay Audience
Listening to speech in noisy environments, such as a busy mall, is a challenging task, especially for individuals with hearing impairments. One popular hearing test used to assess listening performance is the Connected Speech Test (CST). Each passage contains several sentences about a particular topic. Participants must listen and repeat each sentence, and are scored based on correctly identified key words. This task is made challenging by simultaneously playing background babble to obscure the target speech. Listening to the target speaker while ignoring all other speakers requires mental effort (known as listening effort). This task emulates difficulties that listeners may encounter in their everyday life. Originally developed in Memphis in 1987, the CST contained a Southern accent, which was hypothesized to affect Canadian participants’ ability to understand the speech. Researchers have recently updated the test with the General North American accent, a more familiar accent for North Americans, to improve the participant’s ability to understand the speech. This project aims to assess how well adults with normal hearing perform on this updated test. Forty native English-speaking adults with normal hearing participated in this study. The background babble was presented at a fixed volume, while the volume of the speech was changed so that it was louder, the same level, or softer than the background babble. To assess whether having background noise coming from different directions will make it easier for the listener to separate different sounds and concentrate on the target speech, there were two conditions: (1) all sounds were played from the front loudspeaker, and (2) the target speech came from the front loudspeaker, while the background babble was played from other surrounding speakers. Participants were also asked to rate their listening effort. Results showed that participants performed better in the condition with speech and babble coming from different speakers. In general, performance scores increased while listening effort decreased when the speech was louder than the background noise. This experiment collected data to assess how normal listeners perform on the test. This data will serve as a foundation for all future studies involving the updated version of the CST.
Sung, Yu-Ying, "Normative Performance On The Connected Speech Test" (2023). Electronic Thesis and Dissertation Repository. 9791.
Available for download on Wednesday, December 31, 2025