Keywords accompanying abstracts are the metadata and topic representative features of research articles. They can enhance the retrieval, citation, and sorting of the studies in academia. Whereas keywords have been of interest to researchers in various fields of study, few, if any, studies have addressed them in a corpus-driven analysis. The present diachronic case study, then, aimed at analyzing 245 corpus-based studies to investigate their form, relevance, source, and frequency. Selected studies were published in International Journal of Corpus Linguistics from 1996 to 2016. Descriptive analysis of the data revealed that domain-specific keywords were the most frequent ones supporting the journal’s more specific taste. High percentage of keywords with one representation across the two decades alerted the chaos in selecting appropriate keywords. Lack of criteria in selecting the most relevant keywords and asymmetrical forms were among the salient results. Findings may be useful to the researchers, authors of research articles, as well as editors and publishers.
