Tuesday, December 23, 2008

Trope-ing the light fantastic (universal translators, part II)

About a week ago, I dismissed the know-it-all, doesn't-need-training, universal translator as a trope. I left for another day -- which, it turns out, is today -- the matter of universal translators (UTs) able to learn all other languages. Call these Type II UTs.

At a minimum, the Type II UT is more credible than the know-it-all variety. Type II presumes that not all translations are self-evident. The Type II has to learn something. Maybe that's vocabulary. Maybe it struggles a bit with idiom. Still, Type II assumes that all languages ultimately fit within its array of lexical, syntactic, and semantic models. It still presumes some significant intersection of world views between us (as prospective builders of the UT) and the hive-mind slime molds of Rigel III.

But perhaps there is a loophole. Stories generally gloss over how the UT works. What if the UT is telepathic and thus gains insight into the utterly alien world view of the other mind? Another of my trope pieces concluded that technology-enabled telepathy might be possible.

Reading the hive mind of the slime molds might facilitate learning to translate -- but how would one read a mind whose thought processes, vocabulary, and symbolism are unknown? Wouldn't you need a Type I, doesn't-need-training UT to read minds on behalf of the Type II?

In hindsight, my "maybe technological telepathy can happen" post applies only to species with whom we already communicate. The aggregated output of a Martian's billions of neurons (or whatever he/she/leur/it uses) would surely be only so much noise to a telepath -- mechanical or otherwise -- never before exposed to it.

Can technology learn some languages? Most likely. Will it learn some faster than others? Sure, that makes sense -- the more the new language resembles a language the translator already knows, the better. But a universal translator? Trope city.

No comments: