Romance languages are well known for their use of expletive negation (henceforth, EN), i.e., the occurrence of a negator in the complement clause of certain verbs, adpositions or adverbs that is “illogically” not part of the meaning of the sentence. This study explores the hypothesis that such “illogism” that recurs across languages must be due to universal properties of the message to be encoded and the language production system. Jin & Koenig (2019) proposed a language production model to account for the striking similarity of EN-triggers between two unrelated languages (French and Mandarin). Their model makes several predictions which our paper tests: (i) languages like English where EN is purported not to occur should in fact include the same range of EN-triggers; (ii) English speakers can understand a negator within the scope of an EN-trigger expletively; (iii) the likelihood a speaker of English will understand a negator expletively is correlated with how frequently she has encountered an expletive interpretation of negators for that particular trigger. To test the first prediction, we conducted a corpus study of unrehearsed English speech on Google. To test the second prediction, we conducted a semantic Stroop-like comprehension experiment where participants’ semantic judgements (both logical accuracy and response time) was dependent on whether a negator was interpreted logically or expletively. Overall, this paper suggests that EN is by no means specific to Romance languages and that expletive uses of negators occur in the same contexts in both production and comprehension in languages where EN is not conventionalized to the same degree it is in Romance. Overall, our results support the claim that “illogical” properties of natural languages that recur across languages of the world reflect universal properties of the language production system.