Indeterminacy of Translation

The indeterminacy of translation is the thesis, introduced by W.V.O. Quine in Word and Object (1960), that there is no fact of the matter about which of multiple mutually incompatible translation manuals is correct. When a linguist confronts a radically foreign language — one with no historical or cultural connection to known languages — multiple translation schemes can be constructed, all compatible with the totality of behavioral evidence, yet attributing different meanings to the same utterances.

Quine's argument turns on the thought experiment of radical translation: a linguist observing a native speaker who utters 'gavagai' in the presence of a rabbit. The linguist might translate this as 'rabbit,' but alternative translations — 'undetached rabbit part,' 'rabbit stage,' 'instantiation of rabbithood' — are equally consistent with all observable stimulus-response pairs. The linguist can choose a manual, constrained by pragmatic criteria of simplicity and coherence. But the constraint is not ontological. There is no deeper fact that makes one manual correct.

This is stronger than the underdetermination of theory by evidence. Underdetermination says multiple theories fit the data; indeterminacy says there is no fact of the matter at all. The consequence for philosophy of language is that meaning is not a natural kind — not a feature of the world independent of our descriptive practices. It is a posit: a useful fiction for organizing verbal behavior.

The thesis connects to meaning holism (the meaning of a term is its place in the entire network) and to Quine's skepticism about the analytic-synthetic distinction. If there are no analytic truths fixed by meaning alone, and if meaning itself is indeterminate, then the traditional framework of semantics collapses and must be rebuilt on behavioral foundations. The indeterminacy is not merely a problem for translation between radically foreign languages. Quine extended the insight in Ontological Relativity (1969) to the claim that reference itself — not just translation — is inscrutable without a background theory. When two speakers 'agree' on a claim, they are agreeing within a shared interpretation scheme, not accessing identical propositions. The scheme is never neutral. What looks like cross-theory agreement may be agreement within a theory about what both parties are saying — a loop that never makes contact with a theory-independent world.