Translation, Authority, and the Risk of Meaning: From Mahāvyutpatti to the AI Era
The formation of the Mahāvyutpatti was not merely a lexical project but an institutional response to a perceived risk: that the Dharma, once transferred across languages, could be altered not by loss but by misinterpretation. Indian paṇḍitas and Tibetan translators alike understood that translation was not a neutral act. It was a site at which meaning could either be preserved or quietly transformed.
A foundational principle preserved in Tibetan scholastic culture states:
ཚིག་མ་འཁྲུལ་ན་དོན་མི་འཁྲུལ།
དོན་མ་འཁྲུལ་ན་ལྟ་བ་མི་འཁྲུལ།¹
If the words are not mistaken, the meaning is not mistaken;
if the meaning is not mistaken, the view is not mistaken.
This formulation reflects a concern inherited from Indian epistemological traditions: that linguistic precision is not merely technical, but doctrinal. A shift at the level of terminology propagates upward into philosophical error. Translation, therefore, required control—philological, institutional, and interpretive.
For this reason, early translation practice was structured around asymmetry. The paṇḍita functioned as the authority on meaning; the lotsāwa as the mediator of language. Tibetan reflections on this structure preserve a clear warning:
པཎྜི་ཏ་ལ་མ་བསྟེན་ན་ལོ་ཙཱ་བ་མཁས་པ་མ་ཡིན།²
Without relying on the paṇḍita, the translator is not learned.
The implication is precise: translation cannot be autonomous. It must remain anchored in sources of authority beyond the translator’s individual judgment.
Later Tibetan scholasticism—particularly in figures such as Sakya Paṇḍita—tightens this position by identifying the specific conditions under which translation fails. The first is the reduction of translation to lexical substitution:
ཚིག་ཙམ་འཛིན་པས་དོན་མི་རྟོགས།³
Grasping mere words, one does not understand meaning.
The second is the abandonment of the two core authorities of interpretation—scripture (lung) and reasoning (rigs pa):
ལུང་དང་རིགས་པ་མ་བསྟེན་པས་དོན་ལ་འཁྲུལ་པ་འབྱུང་།⁴
Without relying on scripture and reasoning, error arises in meaning.
And finally, the most direct formulation of the risk:
དོན་མ་རྟོགས་པར་བསྒྱུར་ན་བསྟན་པ་འགྲོ།⁵
If one translates without understanding, the teaching declines.
These statements together outline a complete theory of translational failure. Error does not arise merely from ignorance of vocabulary, but from a breakdown in the relationship between three authorities: textual transmission (lung), rational analysis (rigs pa), and lineage-based instruction (man ngag). In modern terms, we might add a fourth: institutional authority, which governs consistency, publication, and dissemination.
The early translation enterprise thus operated under a regime of controlled meaning. The Mahāvyutpatti itself can be read as an attempt to stabilize equivalence across languages so that interpretation would not proliferate uncontrollably. It is, in effect, a technological solution to a philosophical problem: how to prevent semantic drift.
A traditional prophetic formulation—often attributed to Padmasambhava—marks the moment when this control begins to loosen:
བྱ་ལྕགས་འཕུར་དང་རྟ་འཁོར་འགྲོ་བའི་དུས།
བོད་ཀྱི་ཆོས་ནི་མར་འགྲོ།⁶
When iron birds fly and wheeled horses move,
the Dharma of Tibet will go outward.
This outward movement—historically realized in the transmission of Buddhist texts into European languages—recreates the original problem under new conditions. The translator is no longer embedded within a tightly controlled system of paṇḍita oversight and shared doctrinal training. Instead, translation occurs across institutional, cultural, and philosophical gaps.
In this context, the earlier warnings acquire renewed force. Without shared terminology, fixed lexicons, or unified interpretive authorities, translation becomes increasingly dependent on individual judgment. The risk identified by the Indian paṇḍitas re-emerges in a new form: not the disappearance of the Dharma, but its gradual transformation through interpretation.
The contemporary use of AI in translation intensifies this condition. Where earlier translators risked misunderstanding due to limited knowledge, AI systems risk producing fluent renderings without understanding at all. The distinction identified by Tibetan scholastics—between grasping words and understanding meaning—becomes critical:
Grasping words is now trivial.
Understanding meaning remains difficult.
The professional translator, therefore, stands within a field defined not by certainty but by constraint. The task is not to eliminate error entirely, but to manage it responsibly—by remaining accountable to the four voices that govern translation:
• the text (lung)
• reason (rigs pa)
• tradition (man ngag)
• institutional context
To these, a fifth must now be added:
• the limits of what can be responsibly known and said.
⸻
Footnotes (Apple Notes / Pages ready)
¹ Traditional Tibetan scholastic formulation reflecting the word–meaning–view relationship; consistent with Indo-Tibetan epistemological principles.
² Tibetan translation-era principle emphasizing dependence on Indian paṇḍitas; reflects hierarchical translation model.
³ Attributed doctrinally to the scholastic position of Sakya Paṇḍita; captures critique of lexicalism.
⁴ Standard scholastic formulation of reliance on lung (scripture) and rigs pa (reasoning).
⁵ Widely attested scholastic warning: translation without understanding leads to decline of the teaching (bstan pa).
⁶ Traditional prophetic verse attributed to Padmasambhava regarding the global spread of the Dharma.
Comments
Post a Comment