The first thing you notice about San Cristóbal de las Casas is that there are other languages spoken in the markets besides Spanish. Speaking in clipped, almost melodic syllables, women wearing dark wool skirts have a weighty cadence that is older than both Mexico and the surrounding colonial structures.
Tsotsil is frequently that language. And until recently, practically no modern technology could comprehend a single word of it.
| Project Profile | Details |
|---|---|
| Initiative Name | Bilingual Tsotsil–Spanish AI Literacy Platform |
| Lead Institution | School of Humanities and Education, Tec de Monterrey |
| Project Lead | Alejandro Martín del Campo |
| Co-Researcher | Gabriela Salas (previously worked on Spanish–Nahuatl translation accepted by Google Translate) |
| Language Family | Mayan (agglutinative structure) |
| Estimated Tsotsil Speakers | Around 550,000 people, mostly in the Chiapas highlands |
| Region | Highlands of Chiapas, Mexico |
| Government Partner | Chiapas Ministry of Education, under the “Chiapas Puede” social program |
| Technology Base | Four neural-network models for natural language processing, designed to run offline |
| Capabilities | Speech-to-text, bidirectional translation, text-to-speech with native accent |
| Linguistic Corpus So Far | ~17,000 phrases and 20,000 words collected; goal of 21,000 validated phrases |
| Project Completion | Roughly 40%; pilot testing expected within 18 months |
| Related Context | Chiapas has Mexico’s highest illiteracy rate, at 11.5% among people 15 and older |
Slowly but surely, that is beginning to change in a way that feels both modest and subtly radical. For the past two years, a group of researchers from Tec de Monterrey has been collaborating with the Chiapas Ministry of Education to develop an artificial intelligence platform that can read, speak, and translate Tsotsil, a Mayan language spoken by about 550,000 people. Speaking with those involved gives me the impression that they don’t want to oversell it. They have witnessed what transpires when outsiders show up with software that promises salvation.
The platform is bilingual in Spanish and Tsotsil, and its offline functionality is the most notable design choice. No fiber optic cables winding through the cloud forest, no signal towers. Just preloaded computers in community centers and classrooms, the same kind that are already sitting in rural schools, only partially utilized. The project’s leader, Alejandro María del Campo, describes it more as a literacy companion than a translation tool, meeting people where they are rather than where Silicon Valley thinks they should be.
Tsotsil is more difficult than it appears. According to Gabriela Salas, who previously assisted in training an AI model for Nahuatl that Google Translate eventually adopted, Tsotsil defies Spanish logic in ways that cause discomfort for engineers. It has agglutinative properties. Words have multiple meanings. Because even a slight grammatical error in training data can lead to something culturally incorrect, the team must first have native speakers validate every sentence they feed the model.
So far, about 20,000 words and 17,000 phrases have been gathered through interviews with teachers, weavers, elders, and anybody else who is willing to sit and talk. At least 21,000 validated phrases are the goal. It’s laborious work, and the deadline may slip—these things nearly always do. However, the corpus is expanding weekly in highland living rooms and community centers.
Interestingly, this isn’t the only endeavor. Andrés ta Chikinib, a young teacher in Zinacantán, has been feeding Tzotzil grammar into ChatGPT for months—almost as a side project, almost as a dare. He made the chatbot his student because he lacked textbooks. It’s difficult to ignore how frequently these tales start in the same manner: the child grows up attempting to regain the language after their parents stopped speaking it due to discrimination.
All of this raises a deeper question. Although the National Institute of Indigenous Languages discovered that almost half of all variants are still in danger of going extinct, Mexico passed a law in 2003 designating Indigenous languages as official. Laws are silent. Individuals do. And more and more, machines may as well.
It’s still unclear if an offline AI operating on Chiapas classroom computers can change that course. However, there is something genuinely strange about this project as it develops—technology designed to endure rather than grow.
London Bilingualism's content on health, medicine, and weight loss is solely meant for general educational and informational purposes. This website does not offer any diagnosis, treatment recommendations, or medical advice.
We consistently compile and disseminate the most recent information, findings, and advancements from the medical, health, and weight loss sectors. When content contains opinions, commentary, or viewpoints from professionals, industry leaders, or other people, it is published exactly as it is and reflects those people's opinions rather than London Bilingualism's editorial stance.
We strongly advise all readers to consult a qualified medical professional before acting on any medical, health, dietary, or pharmaceutical information found on this website. Since every person's health situation is different, only a qualified healthcare provider who is familiar with your medical history can offer you advice that is suitable for you.
In a similar vein, any legal, regulatory, or compliance-related information found on this platform is provided solely for informational purposes and should not be used without first obtaining independent legal counsel from a licensed attorney.
You understand and agree that London Bilingualism, its editors, contributors, and affiliated parties are not responsible for any decisions made using the information on this website.
