This thesis examines the role played by language in the interpretation of emotional expression. Language labels may indirectly influence such tasks, organising and reactivating a useful repository of semantic knowledge (e.g. Barrett, Lindquist & Gendron, 2007). This proposal was explored using a series of semantic satiation experiments (Lindquist, Barrett, Bliss-Moreau & Russell, 2006). Participants repeated words 3 or 30 times before deciding whether two faces matched in emotional expression. Word type was manipulated across experiments (emotion labels, neutral labels and non-words); an indirect account would only predict reduced accuracy when participants experience semantic inaccessibility, achieved via massed repetition of an emotional label. However, reduced discrimination was found both after 30 (vs. 3) repetitions of any word, and two non-linguistic activities. Findings then suggest that the massed repetition decrement arises via a non-semantic mechanism, such as response uncertainty (e.g. Tian & Huber, 2010). However, an emotion-specific effect of language was also consistently observed. Participants showed facilitated performance after 3 and 30 repetitions of an emotion label, but only when it matched both expressions in the pair. This may suggest that language labels directly influence early emotion perception (Lupyan, 2007, 2012), or provide strategic support during paired discrimination (e.g. Roberson & Davidoff, 2000). A perceptual threshold procedure was used to test the direct assumption. Participants repeated an emotion or neutral label before deciding whether a briefly presented face did, or did not, display an emotional expression. In comparison to the neutral baseline, participants showed no facilitation in performance following exposure to emotion labels that were ‘weakly’ or ‘strongly’ congruent with the subsequently presented expression. Overall, findings inconsistently support the notion that language shapes the interpretation of emotional expression. This prompts discussion of how task demands may influence language-driven recruitment of conceptual knowledge, and the time-course across which these linked elements influence interpretation.
