Writing the Situated Self: Intersectional and Geographic Patterning in College Admissions Essays

Written language is often considered a reflection of the author's internal thoughts, states, and characteristics. This is why writing is used in evaluation to make inferences about authors, but also why social scientists use text as a lens into psychological states. We present a complementary perspective: that writing systematically reflects the social environments of authors. We explore these dynamics in a large corpus of college admissions essays ( n = 232,838 essays submitted by 76,510 self-identified Hispanic applicants from 2015 to 2017) and metadata about author sociodemographics, including the ZIP code of their high school, gender identity, and household income. We find that intersectional demographic markers are highly salient. We further demonstrate that ZIP code-level socioeconomic measures are strongly correlated with the writing style and content of local applicants. Finally, we show that misclassification rates vary geographically and are modestly associated with local demographic composition and social capital, indicating that writing is influenced both socially and geographically. By documenting strong alignment between text and authors that is observable through machine reading, our findings clarify how author demographics and context structure what is written and, stylistically, how. Framing text as an external, social process rather than purely internal creates opportunities to bridge psychological and sociological data, methods, and theories to better understand the connections between language and society, especially in moments of evaluation. Because these texts were produced before the popularization of large language models (LLMs), our results serve as useful benchmarks for archival work and also generate new questions as LLMs mediate written self-presentation.