The Fashion Paratext Dataset focuses on the collection and analysis of fashion captions within contemporary fashion media. Although often overseen and rendered subordinate, these paratextual snippets of text – such as titles, introductions and captions – work as strategic value producers within fashion media and shape our fashion narratives and vocabulary. Accordingly, these textual elements play an important role in our relation to fashion and clothes as readers, wearers, makers and consumers. Through its dominance and vast networks, industrial market-based fashion language is becoming our fashion mother tongue. And as professor of applied linguistics Robert B. Kaplan writes, our first language, or mother tongue has a powerful influence on the way we shape our thoughts and organize our ideas.

Using data science methods, this project collaboratively produces a large scale dataset of fashion captions that can be mapped and analyzed using Natural Language Processing. The dataset takes the captions out of the saturated pages of fashion media (both print and online), offering the opportunity to read the attentively, isolated from their original contexts. As such, it can open up space to think about an alternative vocabulary.

Project leads: Laura Gardner and Femke de Vries.
Website and consultation: Rowan McNaught.
Research with: Charlotte Plumb, Charlotte Verdegaal, Julia Berg, Leanne Choi, Liance van der Merwe, Lindy Boerman, and Yashna Seethiah.