Do two negatives make a positive? Language and logic in language processing

This study focuses on a factor known to increase sentence processing complexity–negation. We sought to distill out of negation a logical property–Inference Reversal–to see whether it, and not an actual negation word, determines this complexity. First, we tested a negation-less pair of polar operators (at most, at least) in Hebrew. We found that processing time for sentences containing the Inference Reversing at most lagged behind those with at least. Second, we compared the processing of sentences containing two Inference Reversing operators (not less) to sentences with zero (ø, more) and one (not more, less). Since two Inference Reversingoperators annul Inference Reversal (“two negatives make a positive”), we asked whether their processing cost is annulled, or rather cumulative. Surprisingly, RT _{not less} was shorter than RT _{not more}. These findings lead to the conclusion that even when covert, Inference Reversal is an important determinant of processing complexity.