Tucker, D. 1 , Tomaszewicz, B. 2, 3 & Wellwood, A. 1
1 Northwestern University
2 Universität zu Köln
3 Uniwersytet Wroclawski
What are the smallest units of compositional interpretation? Recent semantic analyses of comparatives have highlighted units below the word level: negative adjectives like 'shorter' contain a silent morpheme 'little' and thus are more complex than the positive 'taller' (Rullmann 1995, Heim 2006, Büring 2007). We find evidence for this decomposition in language understanding in English and Polish. In a visual verification task, images of two lines of different lengths were paired with statements containing 'taller' and 'shorter', and with equivalent mathematical statements, A > B, B < A. If 'negative' features are specifically linguistic (Just & Carpenter 1971, Clark & Chase 1972), then an asymmetry in processing might not be observed with math. Deschamps et al. (2015) find an effect of polarity only with quantifiers, not with their math counterparts. Our participants judged the truth of the statement with respect to the image displayed for 200ms (English experiment) and for 4s (Polish). Despite the different task demands, we find that both language and math statements lead to increases in processing load along with the number of negative symbols (RTs, accuracy). This finding is consistent with the decomposition hypothesis for comparatives, but not with the 'simple' hypothesis for math.