Annotated data of English antonyms

The data set consists of a good 500 randomly selected occurrences of each of the above adjectives in their contexts in the BNC (British National Corpus) (some 21,000 occurrences in total). The UNIX command grep was used to retrieve the sentences containing the target words tagged as adjectives in the BNC, and the nominal heads of the adjectives were then identified using a head finder script. The sentence in the written part of the corpus data and the corresponding chunk for the spoken occurrences for each of the adjectives were imported into FileMaker Pro and the adjectives were then manually coded. The methodological procedure used in the analysis of the data proceeds from the lexical items in each case to their actual discursive interpretations in context, i.e., from lexical items to their contextual readings. For instance, if the actual reading of say short report refers to the paper copy, it was analyzed as a concrete object since its basic domain of instantiation is space/concrete object, and if it refers to the content it was coded in its domain of instantiation which is neither space nor time, but abstract/mental space. Crucially, this method then also involves a close analysis of the combining nominals and the meanings they express in each instance. The method of identifying discursive meanings of the antonymic word pairs in their contexts serves to make it possible to make generalizations across the interpretations of the lexical items rather than focusing on the lexical items as such without taking their meanings into account.