Common words, uncommon uses: The most “discoursally significant” language edits in a corpus of Chinese medical manuscripts

• Two medical research manuscript corpora (edited and unedited) were compared.
• A word list of differences between corpora was created using a log likelihood test.
• Most these words were simple with discourse functions like register and stance.
• Implications were offered for teaching and materials development.

With developing countries publishing more research and with China poised to become the world leader in scholarly publications by 2020 (The Royal Society, 2011), there is a widespread pedagogical need for language learning and language editing services, especially for Chinese first-language (L1) researchers in the field of English for Research Publication (ERP). The purpose of this data-driven study was to compare an edited and unedited corpus of 52 medical research manuscripts prepared by Chinese-speaking doctors for journal submission, in order to produce a frequency wordlist for each corpus. A log likelihood goodness-of-fit statistic was first used to compare the two wordlists to compile a list of words representing the most significant differences in frequencies, followed by a qualitative analysis to determine the functions of these words. Apart from the most frequently underused words being function words like the and of, the majority of significant edits were made involving common words that could be categorized into three main discourse-level functions: register (that), stance (can, has/have been), and discourse organizers (although, as, this). It is thus important for materials and course developers as well as non-English speaking researchers to pay special attention to these simple words laden with discourse and rhetorical functions.