Antonym sequence in written discourse: a corpus-based study

• Members of an antonymous pair show asymmetrical distribution in actual language use.
• The concept of markedness is the most important for antonym sequence in text.
• Ordering principles of antonym use in text reflect general cognitive principles.

This article investigates which principles govern the order of antonyms in language use and why antonyms tend to be used in a particular sequence in the sentence. The traditional notion about the paradigmatic relations is that the words in a paradigmatic relation must be replaceable, one for the other, in all contexts. Although antonymy is a paradigmatic relation and antonyms are similar in many respects, members of an antonymous pair may not show symmetrical distribution in actual language use. This study is motivated by an observation, based on corpus data, that certain antonymous pairs tend to be used in a preferred sequence in the sentence. Of the fifty-six English antonymous pairs investigated by Jones (2002), only fourteen did not show any marked preference towards either of the two possible orderings in the sentence. This research was conducted in the electronic corpus of contemporary Serbian language and differs from the study of antonym sequence in English done by Jones with respect to certain aspects of methodology, size of the database and the treatment of the concept of markedness. The results show that three-quarters of the fifty-seven antonymous pairs pre-selected for this study are used in a preferred sequence in text and that there is a strong correlation between antonym sequence in language use and the concept of markedness. It is argued that antonym sequence in written discourse is influenced by the following principles: word frequency, neutrality, temporal and visual-spatial ordering and gender. The conclusion is that the distributional patterns of antonym sequence reflect general cognitive principles and conceptual representations.