Scalar inferences and cognitive load


  • Bob van Tiel
  • Paul Marty
  • Elizabeth Pankratz
  • Chao Sun



A number of studies have found that participants are less likely to interpret the scalar words ‘some’ and ‘or’ with an upper bound when their cognitive resources are burdened, thus suggesting that the computation of scalar inferences is cognitively effortful. We conducted two sentence-picture verification tasks to determine whether this finding generalises to other scalar words. In Exp. 1, we manipulated cognitive load by asking participants to memorise simple or complex grid patterns during the experiment (cf. De Neys and Schaeken, 2007). In Exp. 2, we manipulated cognitive load by varying the time participants could take to process the sentences and pictures (cf. Chevallier et al., 2008). In this way, we tested seven scalar words: ‘some’, ‘or’, ‘low’, ‘scarce’, ‘might’, ‘most’, and ‘try’. We expected to find lower rates of scalar inferences when participants experienced greater cognitive load, i.e., when they had to memorise complex grid patterns in Exp. 1, and when they had less processing time available in Exp. 2. We find significant effects of memory load in the expected direction, but only for positively scalar words, i.e., for scalar words that denote a lower bound. We fail to find any significant effects of processing time. We explain these findings by arguing that the scalar inferences of positively scalar words introduce negative information into the meaning of the sentence, and that the processing of such negative information is cognitively demanding.




How to Cite

van Tiel, B., Marty, P., Pankratz, E., & Sun, C. (2019). Scalar inferences and cognitive load. Proceedings of Sinn Und Bedeutung, 23(2), 427–442.