Sharing Domesticity: What Can Online Rental Listing Texts Reveal about Homemaking Among Strangers?

Long-term home sharing now happens across longer periods, among wider demographics, in more and more cities and in a growing diversity of settings. Yet, we know relatively little about what kind of housing shared renting delivers. Sharing may happen by choice or by necessity but we know little about how tenants make home with strangers. This study leverages artificial intelligence and natural language processing tools to ask: What does the language of shared rental listing texts reveal about homemaking in shared rental housing? We draw on thousands of online listings for shared units in Los Angeles, California. We train a deep learning BERT model and analyze listing texts for verbal cues on the envisioned sharing relationship. For a subset of listings, we conduct qualitative thematic content coding for a deeper qualitative exploration. We find that language reliably distinguishes home sharing relationships focused on cost from those focused on building a congenial sharing environment. Texts signaling a transactional focus contain fewer personal details, use short sentences, and often signal a desire for minimal interaction. Those with a social orientation, on the other hand, tend to share more personal descriptors as well as subtle cues suggesting what types of applicants are welcome or unwelcome. Whereas transactionally-oriented listings talk about “tenants” rather than “roommates,” and describe “access” to different parts of the unit, socially-oriented listings call the unit a “home” and use evocative language to paint a mental image of it. Our results contribute to the nascent literature on the new sociogeographies of home. Understanding how shared unit tenants make home informs theorizing domesticity as it is evolving alongside changing housing strategies in traditional homeowner societies of the Global North.