Speech is remarkably multidimensional. Not only do listeners attend to the sounds speakers make, but also to characteristics such as visible lip and mouth movements that speakers make, as well as the context in which a conversation occurs. Recent research has examined how listeners’ use of visual information (i.e. lip and mouth movements) is affected by the shielding of respiratory masks worn during the COVID-19 pandemic. However, relatively little research has examined how other types of masks, such as those worn for religious or cultural practices, impact listeners’ speech understanding. In addition to auditory or visual disruptions caused by masks, social associations the listener makes about the speaker based on the type of mask they wear could influence their perception of the speaker’s intended message. We hypothesized that if a mask is associated with an out-group or has an otherwise negative association with the listener, it is possible that listeners may commit errors consistent with those negative associations.
In the current work we tested the above hypothesis within the domain of Western attitudes towards Muslim speakers, in particular those that wear Niqab, a head covering worn by some Muslim women that occludes all parts of the face and head except for the eyes and nose. We held that ambiguous utterances made by speakers in Niqab would be more likely to interpreted as negative than utterances made by control speakers wearing masks used as personal protective equipment (PPE). Participants were presented audiovisual stimuli of a masked speaker in either a KF-94 PPE mask or Niqab. Prior to each stimulus participants read a word (e.g., ATOMIC) that had a semantic association with the target word (e.g., BOMB) or a phonological neighbor of that word that served as a lure (e.g., MOM). Targets and lures were manipulated to have opposing positive or negative emotional valences (e.g., MOM/BOMB) according to normative data from the English Lexicon Project. This led to four cue-stimulus pairings: positive target (e.g., DAD-MOM), positive lure (e.g., DAD-BOMB), negative target (e.g., ATOMIC-BOMB), and negative lure (e.g., ATOMIC-MOM). Targets and lures were masked at a moderate signal-to-noise ratio of -5 dB. Listener performance was measured by the number of correctly-identified words. An outgroup-exposure questionnaire was presented to the participants immediately following experimental trials, asking the participants to rate their level exposure to diverse religious and cultural identities, conservatively religious women and outwardly-presenting Muslims. Results will be discussed in terms of number of correctly identified words based on the visual presentation of the speaker (Niqab vs KN-94), the semantic congruency of the two words, the valence of the cue-words correctly identified, and the level of exposure to Muslim women as reported through the questionnaire.