For fuck's sake: Investigating the affective norms and auditory perception of British English taboo words

Swear words are paradoxically ubiquitous and taboo. Whilst there is evidence to suggest taboo stimuli are preferentially processed, most findings relate to visually presented stimuli, despite swear words being more frequently heard than read, and arguably more threatening when presented auditorily. Models of speech perception often include lexical influences on perception, but assume all words are equal (beyond frequency of use). However, more recent work suggests that emotional dimensions (Valence and Arousal) influence speech perception, perhaps as a way of alerting listeners to proximal threats or rewards. As swear words are generally rated on the extreme ends of these dimensions (i.e., low or high Valence, and high Arousal) and evoke physiological arousal, they might be preferentially processed in the auditory domain.

This thesis investigates the auditory perception of taboo words and explores why swears may be processed differently. Chapter 2 collates psycholinguistic norms for British English taboo and non-taboo words, and establishes the stability of these ratings in predominantly taboo, and predominantly non-taboo contexts. The results imply that participant ratings are not influenced by the quantity of taboo, or highly arousing words that participants are exposed to. From these norms, experimental stimuli were selected to investigate the intelligibility of taboo speech.

Chapter 3 examined whether vocoded taboo words presented in steadystate white noise, were more intelligible than non-taboo words. Results showed that listeners were significantly better at identifying taboo words, compared to non-taboo words. However, a follow up study using taboo and non-taboo stimuli with similar Valence and Arousal profiles, indicated that the improved intelligibility was likely driven by Valence and Arousal, rather than Tabooness.

Chapter 4 investigated whether listeners were biased towards hearing taboo words compared to phonemically similar non-taboo, and non-words. Results showed a bias towards hearing real words compared to non-words, replicating the Ganong (1980) effect. However, this effect was stronger in the context of nontaboo words, rather than taboo words. This suggests that taboo words are processed somewhat differently at a lexical (perceptual) level compared to nontaboo words, and non-words.

Chapter 5 investigated the influence of social context on the intelligibility of taboo words. Listeners were asked to identify words, ranging across the Tabooness scale, that were masked by multi-speaker babble (social masker), or speech modulated noise (SMN) (non-social masker). As the Tabooness of the words increased, so did the likelihood of correctly identifying the word, but unexpectedly this relationship was stronger in the context of SMN compared to multi-speaker babble. This suggests that Tabooness is a predictor of speech intelligibility, but the strength of this effect is influenced by the type of noise that is masking the target.

In conclusion, this thesis provides the first comprehensive affective norms dataset for British English taboo and non-taboo words, and used these norms to select stimuli in subsequent speech perception studies. These studies provide evidence that heard taboo words are processed differently to non-taboo words, though this appears to be driven by the emotion of the words, rather than their inappropriateness.