Machine Translation

Mark Davis ☕ mark at
Fri Sep 11 19:30:26 CEST 2009

Yes, and as technical terms used in the industry, *dictionary lookup* is
very strongly distinguished from *machine translation*. Machine translation
take whole texts, has to handle grammar, and produces whole replacement
texts. Dictionary lookup does none of those things.

In a certain sense, I can see what Michel is saying; that the variants
indicate a result, and not a methodology used to produce that result. So
either sr-mactrans or sr-m-footrans would in practice just mean "badly
written Serbian", nothing more.

I agree with Peter that unless we have some concrete use cases -- and for
that matter, someone who would be actively using a mechanism -- the whole
discussion is pointless. I don't forsee Google, for example, distinguishing
MT text via a language tag either with a variant or with an extension.


On Fri, Sep 11, 2009 at 10:10, Peter Constable <petercon at>wrote:

> From: ietf-languages-bounces at [mailto:
> ietf-languages-bounces at] On Behalf Of Michael Everson
> > Rendering of one language into another is translation, whether it is a
> word or phrase or sentence.
> I think there are practical distinctions such as Martin is suggesting that
> are made in some contexts. For instance, in preparing linguistic data in a
> form descriptive linguists often refer to as interlinear text,
> morpheme-by-morpheme lookups are often called the "gloss", and distinguished
> from the "free translation".
> Peter
> _______________________________________________
> Ietf-languages mailing list
> Ietf-languages at
-------------- next part --------------
An HTML attachment was scrubbed...

More information about the Ietf-languages mailing list