Sign languages (was: Re: additions to ISO 639 and the
IANAlanguage subtag registry)
petercon at microsoft.com
Tue Feb 21 22:38:12 CET 2006
> From: ietf-languages-bounces at alvestrand.no [mailto:ietf-languages-
> bounces at alvestrand.no] On Behalf Of John Cowan
> > "sgn-CR" is perfectly 3066-valid for "sign languages as used in
> > Rica," but if anyone has used it to mean specifically "Costa Rican
> > on the basis of Michael's page, they have no assurance that it will
> > interpreted as such by 3066-conformant processors.
> I think that's splitting hairs. sgn-US unquestionably means American
> Sign Language, by RFC 3066 registration; it would be entirely proper
> for people to use sgn-CR for Costa Rican SL.
Picking this arbitrary point to respond and give my opinion.
First, I agree with others that signed expression of spoken languages is
simply another modality, like writing, and a variant subtag -signed is
the appropriate solution; e.g., en-signed for Signed English.
As for tags for signed languages (not signed expression of spoken
languages), I think the use of region IDs in the template "sgn-XX" to
identify *languages* is a bad idea; (note: with "sgn-XX" tags, the "sgn"
is largely redundant, and a large part of the semantic distinction is in
the country subtag):
- As Mark Davis pointed out, many matching implementations will treat
"sgn-XX" and "sgn-YY" as though there was some significant common
relationship even if in fact the two are entirely unrelated.
- The "sgn-XX" template cannot accommodate cases in which multiple
signed languages are spoken in a given country. This is a real scenario
with multiple instances, and where it occurs you necessarily end up with
- Because a country ID is required to indicate the *language* identity,
a region subtag is no longer available to support region-based
sub-language distinctions (e.g. ASL as used in Canada rather than the
US). I don't know of specific scenarios requiring this, though it's
I don't particularly see what benefit is gained by having a common
subtag prefix for all signed languages. But I'm willing to consider
arguments that there is some significant benefit. For sake of
discussion, let's assume that there is: then I strongly prefer that
3066ter would handle "sgn" like a macrolanguage ID and use ISO 639
alpha-3 extlang subtags; thus, "sgn-csd" for Chiang Mai Sign Language
and "sgn-tsq" for Thai Sign Language rather than using "sgn-TH" for
More information about the Ietf-languages