The aim of this study is to describe the ancillary antonymy in Turkish, which is observed in the discursive realization dimension of antonymy and is assumed to be widespread in all languages, and aims to form a methodological tool in determining the non-canonical relations. Ancillary antonyms (B-pairs), depending on the semantic power of canonical antonms (A-pairs), occurs in the unit of sentence or discourse in which A-pair is involved. in order to observe the examples of this formation in Turkish, the use of canonical pairs determined in accordance with the coexistence hypothesis are examined in Turkish Natioanal Corpus (TNC) and a ancillary antonymy database was created. the organisations in which the pairs A and B ara located are considered as cognitive contstructions and these constructions are classified. in the study is also determined the semantic productivity of ancillary pairs and typology of relations between the concepts forming these pairs. Within this framework, It seems that ancillary pairs are formed by nouns and verbs, although the canonical antonyms are typically made up of adjectives. At the same time, it has been shown that the examples of ancillary antonyms established with concepts belonging to different domains are based on both intellectual creativity and primitive oppositions.