Expressives are speaker-oriented, not-at-issue (NAI) content. How is the expressive dimension (Potts, 2007; McCready, 2010) of language processed and represented?  Multidimensional semantic accounts (Potts, 2007) remain influential, but several alternatives also exist, including semantic frameworks of continuations (Barker, Bernardi, & Shan, 2010), unidimensionally directly updating the Common Ground (Schlenker, 2010) and pragmatically treating expressives as independent speech acts (Frazier, Dillon, & Clifton, 2014/2017). The present study focusses on the most clearly expressive items, swear words, to offer new insight on this debate. The study of expressives is critical to a complete understanding of language, but remains a neglected area experimentally.

We present recent findings for swearing within a lexical decision task, influenced by an account which is rooted squarely in semantic-pragmatic theory. Stimuli included 30 swear words (e.g. shit, damn), 30 negatively valenced but non-swear words (kill, sick), 30 open class neutral words (e.g. wood, lend), 30 closed class neutral words, as swear words are thought to be a closed class as well (e.g. while, whom), and 120 pseudowords, for a total of 240 items. Behavioural results (N=34) show that swear words are more effortful for subjects than other words that are similar in their negative affect, meaning that there is more to the expressive dimension than merely a heightened emotional state. Regarding the EEG data (N=18), both negative words and swear words elicited larger late positivity effects (LPC) than neutral words, but scalp distributions suggest that the neural generators underlying these words are likely different (Figure 1). ERP data arguably provides support for a multidimensional account, as swear words appear to contribute a dimension of meaning that is separate from the descriptive content of classes like neutral words.

 

This claim is best evidenced by the fact that both the ERPs and scalp topographies are different for these two classes. The LPC indicates that the effects of swearing emerge relatively late, 550-750 ms after initial presentation of the word, which is also reflected in the behavioral data, with longer RTs to swear targets relative to negative and neutral words. After considering the alternative theories, our results are situated within the Potts framework, and ramifications for semantic theories of expressives will be discussed. This is the first study we know of exploring expressives neurophysiologically from a linguistic perspective.

​