Argues that words with different meanings normally appear in discriminably different contexts, and that the cue for learning to associate direct antonyms is not their substitutability but rather their frequent co-occurrence in the same sentence. (Author/VWL)