In assessment contexts, the domain definition inference requires accurate documentation of linguistic demands in a specific target domain for precise measurement. The present study examines several aspects of repetitions in academic lecture settings to offer the domain definition inference for academic listening tests. To do this, I analyzed the Fluency Corpus of Academic English Lectures (FCAEL), a collection of 100 lecture videos from publicly available resources. I compared FCAEL to 54 academic lecture passages in Test of English as a Foreign Language (TOEFL) iBT to examine repetitions in terms of words, types of words (content words or function words), and the location (clause-initial or clause-within). I used descriptive statistics and visualizations to examine the two sources in terms of the proportion and frequency. The results showed that the two sources commonly had higher proportions of function words than content words and higher proportions of one-word repetitions than multi-word repetitions. However, some dissimilarities were also found. In FCAEL, "the" had the highest proportion, while "the" was ranked third in TOEFL iBT. Moreover, clause-within repetitions were more frequent than clause-initial repetitions in FCAEL, while the proportions were similar in TOEFL iBT. Overall, the results challenge the domain definition inference of academic lecture passages in TOEFL iBT. I propose thresholds of repetitions which would be useful for test development and revision.