Corpus-based approaches and discourse analysis in relation to reduplication and repetition

Reduplication is important in language studies. Its linguistic form at the lexical level has long been explored in terms of various formalist theories. However, the linguistic function at other levels such as the discourse layer tends to be ignored. A reduplication corpus (ongoing compilation; 1687 items in total thus far) has been constructed as the baseline for an integrated approach to the interplay of various kinds of repetition in the use of language. The frequency of each token was calculated based on its occurrence in the British National Corpus (BNC). Then a wordlist with the top 102 items was proposed for related research topics such as frequency, percentage coverage, concordance, and collocation in terms of McCarthy's framework (1990 and later) using MonoConc Pro, WordSmith 4.0 and the SARA 3.2 software. The probability of collocation was calculated in terms of mutual information (MI). The higher the MI score, the more genuine the association between two items (Church and Hanks, 1990). A powerful search engine, Google, was further employed to locate relevant texts on websites for the analysis of reduplication from lexical to discourse levels. Both reduplication and repetition do play a significant role and exhibit extensively a certain language musicality in our everyday life.