“Playing” with a subcorpus extracted from a larger Q&A web corpus. I wanted to do some KW analysis, but the potential reference corpora at my disposal were too different to allow for any meaningful comparison - they foregrounded differences in mode/genre. Then it came to me that the best corpus to compare it with may be a second subcorpus from the very same dataset. Meaningful differences can indeed stand out only against a background of sameness. #corpuslinguistics@linguistics
Add comment