ABSTRACT
One of the questions which any detailed investigation of antonymy would be remiss to ignore is this: exactly how widespread is antonymy in language? One would intuitively expect antonymy to be highly pervasive because ‘opposites’ are encountered so extensively in everyday life. However, an accurate assessment of the ubiquity of antonymy is very difficult to establish. First, there is the problem of defining antonymy: the broader the definition one uses, the more widespread the phenomenon will appear. Then there is the even greater problem of counting: in order to arrive at an estimate of the pervasiveness of antonymy within a corpus, one would need to identify every single antonymous pair in use, then retrieve every sentence which features both of those words,1 then manually edit all of these sentences (which would number over a million in my corpus) to eliminate those in which the word pair do not function antonymously. Only in this way could one arrive at an approximation of the proportion of corpus sentences which feature antonyms, and this approximation would still fail to account for inter-sentential antonymous usage.
