Working memory is important for online language processing during conversation. We use it to maintain relevant information, to inhibit or ignore irrelevant information, and to attend to conversation selectively. Working memory helps us to keep track of and actively participate in conversation, including taking turns and following the gist. This paper examines the Ease of Language Understanding model (i.e., the ELU model, Rönnberg, 2003; Rönnberg et al., 2008) in light of new behavioral and neural findings concerning the role of working memory capacity (WMC) in uni-modal and bimodal language processing. The new ELU model is a meaning prediction system that depends on phonological and semantic interactions in rapid implicit and slower explicit processing mechanisms that both depend on WMC albeit in different ways. It is based on findings that address the relationship between WMC and (a) early attention processes in listening to speech, (b) signal processing in hearing aids and its effects on short-term memory, (c) inhibition of speech maskers and its effect on episodic long-term memory, (d) the effects of hearing impairment on episodic and semantic long-term memory, and finally, (e) listening effort. New predictions and clinical implications are outlined. Comparisons with other WMC and speech perception models are made.

The peak dilation amplitude, peak latency, and mean pupil dilation systematically increase with decreasing speech intelligibility. These results support that listening effort, as indicated by the pupil response, increases with decreasing speech intelligibility. This study indicates that pupillometry can be used to examine how listeners reach a certain performance level. Application of this technique to study listening effort can yield valuable insight into the processing resources required across listening conditions and into the factors related to interindividual differences in speech perception in noise.

The pupil response systematically increased with decreasing speech intelligibility. Ageing and hearing loss were related to less release from effort when increasing the intelligibility of speech in noise. In difficult listening conditions, these factors may induce cognitive overload relatively early or they may be associated with relatively shallow speech processing. More research is needed to elucidate the underlying mechanisms explaining these results. Better TRTs and larger word vocabulary were related to higher mental processing load across speech intelligibility levels. This indicates that utilizing linguistic ability to improve speech perception is associated with increased listening load.

To maintain similar intelligibility levels, participants needed more mental effort for speech perception in the presence of a single-talker masker and then with the two other types of maskers. This suggests an additive interfering effect of speech information from the single-talker masker. The dissociation between these performance and mental effort measures underlines the importance of including measurements of pupil dilation as an independent index of mental effort during speech processing in different types of noisy environments and at different intelligibility levels.

The pupil response to speech masked by interfering speech was assessed across an intelligibility range from 0% to 99% correct. In total, 37 participants aged between 18 and 36 years and with normal hearing were included. Pupil dilation was largest at intermediate intelligibility levels, smaller at high intelligibility, and slightly smaller at very difficult levels. Participants who reported that they often gave up listening at low intelligibility levels had smaller pupil dilations in these conditions. Participants who were good at reading masked text had relatively large pupil dilation when intelligibility was low. We conclude that the pupil response is sensitive to processing load, and possibly reflects cognitive overload in difficult conditions. It seems affected by methodological aspects and individual abilities, but does not reflect subjective ratings.

The measurement of cognitive resource allocation during listening, or listening effort, provides valuable insight in the factors influencing auditory processing. In recent years, many studies inside and outside the field of hearing science have measured the pupil response evoked by auditory stimuli. The aim of the current review was to provide an exhaustive overview of these studies. The 146 studies included in this review originated from multiple domains, including hearing science and linguistics, but the review also covers research into motivation, memory, and emotion. The present review provides a unique overview of these studies and is organized according to the components of the Framework for Understanding Effortful Listening. A summary table presents the sample characteristics, an outline of the study design, stimuli, the pupil parameters analyzed, and the main findings of each study. The results indicate that the pupil response is sensitive to various task manipulations as well as interindividual differences. Many of the findings have been replicated. Frequent interactions between the independent factors affecting the pupil response have been reported, which indicates complex processes underlying cognitive resource allocation. This complexity should be taken into account in future studies that should focus more on interindividual differences, also including older participants. This review facilitates the careful design of new studies by indicating the factors that should be controlled for. In conclusion, measuring the pupil dilation response to auditory stimuli has been demonstrated to be sensitive method applicable to numerous research questions. The sensitivity of the measure calls for carefully designed stimuli.

Perceptual load and cognitive load can be separately manipulated and dissociated in their effects on speech understanding in noise. The Ease of Language Understanding model assumes a theoretical position where perceptual task characteristics interact with the individual's implicit capacities to extract the phonological elements of speech. Phonological precision and speed of lexical access are important determinants for listening in adverse conditions. If there are mismatches between the phonological elements perceived and phonological representations in long-term memory, explicit working memory (WM)-related capacities will be continually invoked to reconstruct and infer the contents of the ongoing discourse. Whether this induces a high cognitive load or not will in turn depend on the individual's storage and processing capacities in WM. Data suggest that modulated noise maskers may serve as triggers for speech maskers and therefore induce a WM, explicit mode of processing. Individuals with high WM capacity benefit more than low WM-capacity individuals from fast amplitude compression at low or negative input speech-to-noise ratios. The general conclusion is that there is an overarching interaction between the focal purpose of processing in the primary listening task and the extent to which a secondary, distracting task taps into these processes.

Previous research has reported effects of masker type and signal-to-noise ratio (SNR) on listening effort, as indicated by the peak pupil dilation (PPD) relative to baseline during speech recognition. At about 50% correct sentence recognition performance, increasing SNRs generally results in declining PPDs, indicating reduced effort. However, the decline in PPD over SNRs has been observed to be less pronounced for hearing-impaired (HI) compared to normal-hearing (NH) listeners. The presence of a competing talker during speech recognition generally resulted in larger PPDs as compared to the presence of a fluctuating or stationary background noise. The aim of the present study was to examine the interplay between hearing-status, a broad range of SNRs corresponding to sentence recognition performance varying from 0 to 100% correct, and different masker types (stationary noise and single-talker masker) on the PPD during speech perception. Twenty-five HI and 32 age-matched NH participants listened to sentences across a broad range of SNRs, masked with speech from a single talker (-25 dB to +15 dB SNR) or with stationary noise (-12 dB to +16 dB). Correct sentence recognition scores and pupil responses were recorded during stimulus presentation. With a stationary masker, NH listeners show maximum PPD across a relatively narrow range of low SNRs, while HI listeners show relatively large PPD across a wide range of ecological SNRs. With the single-talker masker, maximum PPD was observed in the mid-range of SNRs around 50% correct sentence recognition performance, while smaller PPDs were observed at lower and higher SNRs. Mixed-model ANOVAs revealed significant interactions between hearing-status and SNR on the PPD for both masker types. Our data show a different pattern of PPDs across SNRs between groups, which indicates that listening and the allocation of effort during listening in daily life environments may be different for NH and HI listeners.