Background: Difficulty understanding dialogue is one of the most frequent causes of complaints for media broadcast companies. Despite most productions presenting dialogue at speech-to-noise ratios (SNRs) in line with guidance and standards, reported subtitle use has increased among listeners with and without a hearing impairment. Background music and effects in broadcast audio can introduce significant listening effort even when objective intelligibility is high. Effortful listening can result in reduced recall and disrupt language processing. The impact of extended periods of effortful listening on the comprehension of broadcast content and how this affects viewer engagement is not known.

Aims: To investigate the influence of background noise levels on listening effort and dialogue comprehension in broadcast content.

Methods: Thirty-one adult (18-36 years) participants were presented with fifteen, two-minute clips from a radio broadcast in the presence of background noise. The degree of acoustic challenge was modulated by varying the speech-to-noise loudness ratio within the range commonly found in broadcast media (4LU, 10LU, 16LU). Following each clip, self-reported listening effort was recorded through ratings of five dimensions based on the NASA-TLX (Audibility, Mental Demand, Effort, Frustration, Engagement). Participants then completed a recognition task to assess their comprehension of the spoken content by probing for different levels of memory representation (surface form, propositional, situation model). Response times for the recognition task were recorded as a behavioural measure of listening effort. Electrophysiological (EEG) recordings were also taken throughout stimulus presentation and recognition task although this presentation covers only the participant response measures.

Results: Analysis of participant responses showed an increase in reported listening difficulty corresponding to reduced response accuracy at higher background noise levels. Analysis of the recognition task revealed a significant effect of loudness difference conditions on memory of the surface form and at the propositional level (between 16-10LU and 10-4LU respectively). A significant reduction in discrimination of contextually incorrect information was also observed. Results indicate that different information processing components were reduced differentially as acoustic challenge increased.

Conclusions: Extended periods of effortful listening had a negative impact on comprehension and memory of spoken media material. SNRs within the range commonly found in broadcast media can result in significant listening effort. A reduction in recognition for the surface form and an increased reliance on propositional information was observed between the two most favourable SNRs, with a decrease in propositional recognition at the lowest SNR. This points to a greater reliance on information summarisation as increasing acoustic and cognitive demand reduced capacity for encoding more precise details, highlighting the need to ensure that background levels are set to reflect the degree of comprehension necessary to understand and enjoy spoken media content, even when intelligibility is at ceiling level.