Assessing listening effort by measuring short-term memory storage and processing of speech in noise

Abstract

Objective

The aim of this study was to investigate the effect of working memory capacity (WMC) and updating ability (UA) on listening effort measured using a new test, the Auditory Inference Span Test (AIST), as an objective measure of listening effort.

Design

The AIST is based on Swedish five-word sentences and taps into three memory load levels (MLLs). It was administered in stationary speech-shaped noise at −2, −4, and −6 dB signal-to-noise ratio (SNR). WMC was assessed using the reading span test, and UA was assessed using the letter memory test. We also collected data on speech-in-noise performance and subjectively rated listening effort at the three SNRs.

Study sample

Thirty-nine participants with normal hearing thresholds (≤20 dB HL for 250 to 4000 Hz) took part in the study.

Results

AIST performance decreased with increasing MLL and was related to WMC and UA. Participants with high WMC performed better than those with low WMC at more favorable SNRs. Participants with high UA performed better than participants with low UA at the intermediate MLL, which made particular demands on the UA. Neither speech recognition scores nor subjectively rated listening effort was associated with AIST performance or either of the cognitive variables.

Conclusion

AIST taps into cognitive functions necessary for understanding speech in noise. However, in its current form AIST may be too cognitively taxing to successfully measure graded listening effort in participants with lower cognitive capacity.

Keywords:

• Listening effort
• Speech-in-noise
• Cognition
• Working memory
• Updating

Acknowledgements

This work was supported by the Oticon Foundation. The authors would like to thank Jerker Rönnberg for comments on AIST design.