Subjective Evaluation of Communicative Effort for Younger and Older Adults in Interactive Tasks with Energetic and Informational Masking

Valerie Hazan, Outi Tuomainen, Linda TaschenbergerInterspeech 2019

TL;DR

This study investigates how different types of background noise (energetic vs. informational masking) affect the perceived effort and concentration of younger and older adults during interactive communication tasks. The findings suggest that while informational masking is generally more disruptive, age doesn't significantly impact subjective ratings of effort or the ability to ignore noise in these tasks, challenging the notion that older adults are universally more affected by such conditions in real-world communication.

What Problem Does This Paper Address?

Communicating effectively in noisy environments is a significant challenge, particularly for older adults. While traditional speech perception tests in noise are common, they often don't reflect the complexities of real-time, interactive communication. This paper addresses the gap between laboratory-based speech perception measures and the subjective experience of communication effort in noisy, interactive settings. Specifically, it explores how two distinct types of auditory interference – energetic masking (EM), like white noise, and informational masking (IM), like competing speech – differentially impact younger and older adults during a communicative task. Understanding these differences is crucial for developing more effective assistive technologies and communication strategies for aging populations.

Methods

The researchers employed a novel approach using the DiapixUK task, a "spot-the-difference" picture game where pairs of participants must communicate to find discrepancies. This task was conducted under four conditions: quiet, energetic masking (speech-shaped noise), and two types of informational masking (semantically related and unrelated competing speech). Participants were seated in separate, acoustically shielded rooms and communicated via headsets.

To capture subjective experiences, participants completed questionnaires after each task, rating their perceived effort, concentration, and the ease of ignoring background noise on a 10-point Likert scale. These ratings were collected separately for the "Talker A" (leading the interaction) and "Talker B" (more passive) roles to differentiate between speaker and listener effort.

In addition to the interactive task, participants underwent several background tests:

  • CCRM test: Measured signal-to-noise ratio (SNR) thresholds in both babble and speech-shaped noise to assess basic speech-in-noise perception.
  • BKB sentence test: Evaluated sentence intelligibility in a babble masker.
  • Expressive vocabulary (WISC): Assessed verbal knowledge.
  • Letter-number sequencing (WAIS): Measured working memory.
  • Speech, Spatial and Qualities of Hearing Scale (SSQ): A standardized questionnaire to assess general hearing difficulties.

Datasets

The study utilized the DiapixUK picture sets for the interactive communication task. No new datasets were introduced; rather, existing standardized tests (BKB, WISC, WAIS, SSQ) were used for background measures.

Key Insights and Ideas

  • Informational Masking is More Disruptive: As expected, informational masking (competing speech) led to higher ratings of concentration and interference compared to energetic masking (speech-shaped noise) during the communicative task.
  • Age Doesn't Always Mean More Subjective Difficulty: Contrary to some expectations, older adults did not report significantly higher levels of perceived effort or difficulty ignoring noise in the interactive task compared to younger adults. This suggests that subjective experience in real communication might be more resilient to age-related changes than pure perception tests imply.
  • The "Lead Role" Matters: Participants taking the lead in the conversation (Talker A) consistently reported higher effort and concentration than those in the passive role (Talker B). This highlights that speech production and the cognitive demands of leading an interaction are significant contributors to perceived effort, a factor often overlooked in listener-focused studies.
  • Semantically Unrelated IM is More Effortful: Interestingly, the semantically unrelated informational masking (IMUR) was rated as more effortful than the semantically related (IMRE) condition. This suggests that the presence of irrelevant speech might impose a greater cognitive load, perhaps due to the need to actively filter it out, rather than any potential benefit from related keywords.
  • Subjective vs. Objective Measures: Correlations between subjective ratings from the Diapix task and the SSQ questionnaire were generally weak, indicating that these different assessment methods capture distinct aspects of communication experience, especially for individuals with normal hearing.

Key Results

  • Effort Ratings: The quiet condition was rated as least effortful. Both energetic masking (SPSN) and semantically related informational masking (IMRE) were rated similarly, while semantically unrelated informational masking (IMUR) was rated as most effortful. Participants in the "lead" role (Talker A) reported more effort than in the passive role (Talker B).
  • Concentration Ratings: Similar to effort, quiet required the least concentration. Informational masking conditions (IMRE and IMUR) required more concentration than energetic masking (SPSN). An interesting age-by-talker role interaction was observed, where middle-aged adults (OMA) reported less concentration in the lead role compared to the passive role, a reversal of the trend seen in younger (YA) and older adults (OA).
  • Interference Ratings: Energetic masking (SPSN) was rated as easier to ignore than both informational masking conditions (IMRE and IMUR). Again, the lead role (Talker A) reported greater interference than the passive role (Talker B).
  • Background Tests: Older adults (OMA and OA) showed poorer performance on the CCRM test with babble noise (suggesting greater impact from informational masking) compared to younger adults. Older adults also scored lower on letter-number sequencing (working memory). However, older groups had better expressive vocabulary.
  • Predictors for Older Adults: For older adults, better expressive vocabulary and poorer performance on the CCRM with speech-shaped noise (EM) predicted higher perceived difficulty in both IM and EM conditions during the Diapix task.

Limitations and Future Work

  • Sample Size for Age Groups: While the study included three age groups, the number of participants in each group (YA=20, OMA=13, OA=18) is relatively small, particularly for the OMA group. The finding that better expressive vocabulary predicted higher IM difficulty in older adults, for instance, was based on a small sample and requires further replication.
  • SSQ Correlation: The weak correlation between Diapix ratings and the SSQ questionnaire suggests that the SSQ might be less sensitive to the nuances of communication difficulty in individuals with normal hearing, as it is designed for those with hearing loss. Future work could explore more targeted subjective measures.
  • Focus on Normal Hearing: The study focused on participants with normal hearing thresholds. Future research could investigate how these findings extend to individuals with varying degrees of hearing impairment, where the impact of masking might be more pronounced.
  • Task Specificity: The Diapix task, while interactive, is still a controlled experiment. Further research could explore these effects in more ecologically valid, spontaneous communication scenarios.
  • Exploration of IMUR vs. IMRE: The finding that IMUR was more effortful than IMRE warrants further investigation into the specific cognitive mechanisms underlying this difference.

Summary updated

BibTeX
@inproceedings{hazan2019subjective,
  author = {Valerie Hazan and Outi Tuomainen and Linda Taschenberger},
  title = {{Subjective Evaluation of Communicative Effort for Younger and Older Adults in Interactive Tasks with Energetic and Informational Masking}},
  booktitle = {Interspeech},
  year = {2019},
}
Figure