What College Was Actually Selling

Translation, Weighed. From early words we live in to the language institutions reward.

Source: DALL·E (AI-generated image)

Spanish and English have been part of my life ever since I can remember. My parents were from Mexico, but I grew up in Northern California. When I was in ninth grade, my mom decided she wanted me to be fully literate in Spanish, and so I spent a year with my aunt and her family in Saltillo, Mexico. The dollars that my mom had made stretch to send me to parochial school in Oakland, California, stretched even further in Saltillo. There I was placed alongside expat children of the local GM plant managers and the local elite. It was full immersion in Spanish for one year. When I got back to high school in the U.S., there were so many words that I knew naturally—words like penultimate or remunerative. I sailed through high school. In my second year of college, I went to São Paulo for two years, fully immersed in Portuguese. This jiggled my Spanish and my English, stretching sounds and combinations in new directions. Words traveled across all three languages, and many of those words helped me when I got to graduate school, where I got a Ph.D. in cognitive science and psychology.

The language of science, even of law and of medicine, had always been second nature, and it made my studies easier, especially in STEM fields, which are so Latin-heavy. If it mattered for me, did it matter for others? The data bore it out. Steele and colleagues' evaluation of Portland's dual-language immersion lottery found that everyone benefited, with a larger benefit for those in Spanish programs.

College is the first place many people get fully immersed in Norman English—the formal, Latin-derived register that runs through institutions, science, and law. This is the part of the conversation about higher education that almost nobody has, and I think it is the part that matters most.

Those who arrive at college from educated, wealthier families have generally been exposed to Norman English their entire lives. If they grew up in a household with science, science terms are almost entirely Norman. Myocardial. Phylogeny. Spectroscopy. If they grew up in a house with lawyers, legal terms are Norman too. Habeas corpus. Tort. Promissory.

First-gen and underprivileged students often arrive at college without that immersion. My hypothesis is that this is one of the mechanisms by which elite universities actually work for the students who need them most. It is not the seminars or the small class sizes or the libraries, though those help. It is the four years of being soaked in Norman, in community with people who already speak it.

But here is the harder claim. Norman English functions like a second language for native English speakers. There are two pathways to fluency: early exposure or rare adult talent. Most people get neither. By the time first-gen students arrive at college at 18, their Norman system is relatively underdeveloped. Many catch up. But many do not, and the ones who do not are not failing.

English is two languages masquerading as one. There is a Saxon underbelly that forms the household, almost all of it derived from Old English. Norman in English is learned later, mostly through school and books rather than through the body. Tranquil means very soft and easy on a calm day. Calm you can feel in your body. Tranquil you cannot. Tranquil is a word learned from a vocabulary list.

The seam is the Latinate layer of English, absorbed after 1066, when William the Conqueror put a French-speaking ruling class atop the Anglo-Saxon population.

Jay Caspian Kang has a piece in The New Yorker this week wondering whether AI will make college obsolete. He is not negative on college, just unsure whether his daughter will need it. He cites Gallup: In 2013, 74 percent of young adults said college was very important. By 2025, 35 percent. He worries only elite colleges will survive—places where the privileged buy a credential of declining value.

But reading his essay, I kept asking: Aren't we putting the cart before the horse? Kang sees the stratification coming. He does not see what college has been doing all along. Elite colleges work for first-gen students partly because they provide sustained Norman immersion that wealthier students absorbed long before college. That is the actual machinery.

AI does not threaten that machinery by producing better Norman text. Work I did with Ferenc Bunta, Juliana Ronderos, My Nguyen, and colleagues, published in 2021, showed that Germanic words form the basis of language two to three years earlier than their Latinate equivalents. The Saxon substrate is what LLMs train on. The model has only the statistical version. It produces the surface without ever undergoing Normanization.

Cognition Essential Reads

AI and the Problem of Polished Prose

How to Throw the Right Kind of Party

AI can generate Norman prose. It cannot generate Norman immersion.

This might be one reason that first-gen students who graduate from elite universities thrive. They have been immersed long enough to communicate in the language of the elite, to move in elite spaces in increasingly natural ways, to at least fake elitism even when they do not feel it.

Which is why the policy answer is not about saving college. It is about reducing the seam itself. We should be teaching Norman earlier, in embodied form, to those who do not get it at home.

When I first talked about the seam at a bilingualism conference, a 0-3 policy researcher told me that some preschools in the Washington, D.C., area were teaching preschoolers Latin. Latin is great, but nobody speaks it. Perhaps we should focus on a form of Latin-derived language that might be functional in the United States. That language is Spanish.

One way to save college: dual immersion in preschool with Spanish as the live language, formalized through high school. By high school, the Norman would already be there. Not as an alien register encountered for the first time at 18, but as a second language acquired the way I acquired it—early, oral, written, lived. Through the only window that reliably opens for everyone: childhood.

Early Norman is what saves college from becoming a playground that only the Norman-immersed elite can play in.