P27Session 1 (Thursday 12 January 2023, 15:30-17:30)Nonnative-accented speech and masking noise increase listening effort for native and nonnative listeners: A dual-task approach
Background: Nonnative-accented speech presents a form of signal-intrinsic interference (Mattys et al., 2012, doi:10.1080/01690965.2012.705006) which can lead to decreased intelligibility and processing speed (Bent & Bradlow, 2003, doi:10.1121/1.1603234; Munro & Derwing, 1995, doi:10.1111/j.1467-1770.1995.tb00963.x). These processing difficulties may in part be due to increased listening effort even when nonnative-accented speech is highly intelligible (McLaughlin & Van Engen, 2020, doi:10.1121/10.0000718; Van Engen & Peelle, 2014, doi:10.3389/fnhum.2014.00577). Background noise (energetic masking) also affects speech processing and requires listeners to recruit extra cognitive resources to successfully recognize speech (Pichora-Fuller et al., 2016, doi:10.1097/AUD.0000000000000312). The adverse effects may be additive when processing accented speech in noisy environments (Van Wijngaarden et al., 2002). The present study explores the effect of nonnative-accented speech and energetic masking on listening effort, separately and in combination, for native and nonnative listeners using a dual-task paradigm (Gagné et al., 2017, doi:10.1177/2331216516687287) along with subjective measures of listening effort (NASA-TLX questionnaire; Hart & Staveland, 1988, doi:10.1016/S0166-4115(08)62386-9).
Methods: Native and nonnative English listeners performed a primary word recognition task with sentences produced by one native and one nonnative accented English talker presented in quiet or mixed with speech-shaped noise. They also performed a secondary visual response task during which they pressed keyboard keys to indicate the orientation of an arrow on the screen. The two tasks were first performed separately and then simultaneously. Listening effort was assessed objectively as an increase in response times (RTs) on the visual task when it was performed in combination with the word recognition task. Listeners also provided ratings of perceived effort for a subjective assessment of listening effort.
Results: RTs were longer in nonnative-accented and in noise-masked speech conditions, compared to native-accented speech and speech in quiet, for both native and nonnative listeners suggesting increased listening effort. The longest RTs were found in the combined noise-masked and nonnative-accented speech condition indicating that listening effort was increased most when the two acoustic challenges were combined. While nonnative listeners self-rated the listening task as more effortful than native listeners, all listeners perceived nonnative-accented speech as more effortful and they gave up on the task most frequently when the nonnative-accented speech was mixed with noise.
Conclusions: Even though native listeners had better word recognition accuracy than nonnative listeners overall, the effect of nonnative-accented and noise-masked speech on listening effort, as indicated by RTs, was similar for both groups. All listeners recruited additional cognitive resources when processing signal-intrinsic and signal-extrinsic sources of degradation, and the effect on listening effort was cumulative.