ABSTRACT
This chapter uses the keywords approach pioneered by Scott to explore the data provided. In the keywords approach, two corpora are compared to reveal words which have unusually high or low frequencies and point to salient uses of language. A standard approach to comparing two corpora is to take the data one are interested in and then compare it to a larger general reference corpus. The keyword procedure uses a statistical test to determine whether a word is significantly more frequent or not. The chapter uses the log-likelihood procedure in WordSmith Tools to generate the wordlist. The procedure outlined earlier generated 83 keywords for the Indian English material, 48 keywords for Philippine English, 46 keywords for UK English, and 57 keywords for US English. Perhaps the most eye-catching of the fields that are unique to the UK data is the category of Politeness in which occurs the keyword sorry.
