Emotional language appears to support the inference process in a hierarchical nature (Shears, et al., 2011). However, Nasrallah, Carmel and Lavie (2009) suggest that the negative valence should be primary in supporting inferences because it is survival based. Further, Gygax, Garnham and Oakhill (2004) claim the importance of context is critical when readers are processing emotional language. Here, we extend previous findings using two sentence pairs, by examining longer, more natural story contexts. Similarly, we hypothesized that if emotional language supports the formation of causal inferences, then positive stories should cause more false alarms to inference-related target words than negative stories. Participants made key press responses to words either in the story (control) or words related to the inferred information (experiment). Both accuracy and reaction time data were used to measure the formation of inferences across valences. Results suggest readers formed inferences equally from positive and neutral stories, but did not form inferences from negative emotional stories. These findings imply a unique quality of negative emotional language that resists typical comprehension processes of knowledge-based inferences.
Ariza, Adriana; Shears, Connie; Lam, Maisy; Cohen, Amy; Bond, Melissa; Smith, Mackenzie; Sam, Erika; and Kim, Jay, "Why Doesn't Negative Behave? Inferences from Emotional Language" (2015). Student Research Day Abstracts and Posters. 114.