About the Translation Project

This project is not presented as a replacement for the work of Tibetan scholars, experienced translators, or teachers of the tradition. It is an experimental computational research project investigating how complex Tibetan texts can be processed, segmented, analyzed, interpreted, translated, and reviewed.

Translation is one of the results of the project, but it is not its only object. The purpose is to study and develop a process in which sources, linguistic decisions, alternative readings, uncertainty, and the evidence used can be preserved and examined.

The work combines deterministic processing, structured research data, AI-assisted analysis, lexical and traditional resources, published translations, and human review. Existing translations are used as important comparative resources, not as templates.

An important part of the Tibetan-to-English translation work uses MITRA, a language model developed specifically for Buddhist Tibetan translation by Kurt Keutzer and his collaborators. MITRA provides an initial computational translation that is then reviewed, revised, compared with other resources, and preserved together with the source material used in the project.

When a reading remains uncertain, we attempt to preserve that uncertainty and the evidence surrounding it rather than silently resolving it.

The objective, therefore, is not simply to produce another translation, but to construct a reviewable record of how the translation was reached.