ABSTRACT
Existing classes of antonymy provide a useful mechanism for making logical distinctions between different word pairs, but they are not sufficient to describe the often complex ways in which antonyms operate in text. For this reason, a complementary, data-based system of classification has been developed. However, an empirical system is not inherently better than a logical system; corpora simply provide evidence upon which fresh observations may rest. If these observations are to be accurate, it is essential that the corpus is interrogated in a methodologically sound fashion. Antonymy is no more problematic to examine than other linguistic phenomena, but a number of methodological issues arise when corpus data is used. This chapter will explore these issues, justify the approaches taken, and report on each stage of the classification process.
