Although there are millions of Nahuatl speakers, the language is highly threatened. The dominant language of Coatepec de los Costales, a small village in Guerrero, Mexico, was historically Nahuatl, a Uto-Aztecan language, referred to by some as “Mexicano” (Messing, 2009). In the last 50 years, there has been a pronounced shift from Mexicano to Spanish in the village, and fewer than 10% of the residents currently speak Mexicano. Without intervention, the language will be lost in the village. The ultimate cause of language shift is a disconnect in transferring the Indigenous language from the older to the younger generations. In Coatepec, older Nahuatl speakers are not teaching their children the language. This recurring theme appears in case studies of language shift around the world. Using a conceptual framework that combines (1) a critical sociocultural approach to language policy; (2) Spolsky’s (2004) definition of language policy as language practices, ideologies or beliefs, and management; (3) the ethnography of language policy, and (3) Indigenous knowledges, I collected and analyzed data from a six-month ethnographic study of language loss and reclamation in Coatepec. Specifically, I looked closely at the mechanisms by which language ideologies, management, and practices were enacted among members of different generations, using a combination of observation, archival analysis, and in-depth ethnographic interviews. Seidman’s (2013) three-part interview sequence, which includes a focused life history, details of experience, and reflections on meaning, provided the framework for the interviews. What are the language ideologies and practices within and across generations in this setting? What language management strategies – tacit and official – do community members of different generations employ? This in-depth examination of language ideologies, practices, and management strategies is designed to illuminate not only how and why language shift is occurring, but the possibilities for reversing language shift as well.