When I first started thinking about doing Honours in corpus linguistics, I knew I wanted to look broadly at sex, sexuality and/or gender. When I approached Monika Bednarek to be my supervisor I didn’t have any experience with corpus linguistics, but it came to be the perfect method for my project. My data were the advice columns from the magazine Dolly (known as ‘Dolly Doctor’), which I collected into a corpus of about 90,000 words. I also wanted to see how the advice had changed over time, so half my data was from the 1990s (1994 and 1995) and half was from the 2010s (2014 and 2015), which I compared for their differences and similarities.

I first looked at the data in terms of differences, performing a keyword analysis with the 1990s subcorpus as the study corpus and the 2010s subcorpus as the reference corpus (and then vice versa). The value of using corpus linguistics was immediately apparent: the keywords in the 1990s suggested a preoccupation with sexual health (e.g. clinic, STDs [sexually transmitted diseases], pregnant, pill, condom), while keywords in the 2010s suggested a preoccupation with mental health and emotions (e.g. anxiety, down, angry, depression, upset). I used the keywords as a starting point for more detailed manual discourse analysis, and you can read up on my findings in an article recently published in Discourse & Society.

In terms of the similarities between the subcorpora, keyword analysis showed that normal was a key evaluative adjective in both of the decades (when compared to a reference corpus of general Australian English). Concordance analysis helped me see how this was used in the data; for example it is frequently amplified as completely normal or very normal. A particularly interesting finding here was how the word normal was distributed in the corpus, appearing in almost three times as many answers as questions. I used this finding as a starting point to then look at texts where normal didn’t appear in the question, but was instead introduced in the answer. This revealed interesting patterns of evaluation within the question-and-answer pair, including how the answer shifted from evaluations like I’m too scared to use a tampon to The fears you have are quite normal.

Corpus linguistic analysis was a great tool for getting into the data and pinpointing things which were worth looking at in more detail. Above all, it was fantastic data to work with, and I still secretly enjoy the reactions I get at conferences when I present texts about tampons or masturbation…

You can read about this research in more detail in my thesis: https://ses.library.usyd.edu.au/handle/2123/18592