Language Subtag Registration Form: variant "signed"

[Peter Constable]

> From: ietf-languages-bounces at alvestrand.no [mailto:ietf-languages-
> bounces at alvestrand.no] On Behalf Of Michael Everson


> >I have no particular reason to rush signed languages in IETF
> >language tags. I very much wish you hadn't rushed a bunch of tags
> >through five years ago.
> 
> Five years ago I was responding to a user request for a number of tags.

For how many of the 22 registered tags can you point to real uses? I tried some searches on MSN and Google and (apart from discussions of language tags) found only the SignBank site, which I'm guessing was the source of your user request. 

The SignBank site is hardly a stellar example of use of registered language tags: 

- One of the 22 registered tags, sgn-ZA, they don't use at all.
- They've been using 21 tags that weren't registered.

Of the un-registered tags, there are twelve they can claim are sanctioned (quite unfortunately, IMO) by 3066bis:

sgn-AU: 17 entries
sgn-CZ: 533 entries
sgn-ET: 87 entries
sgn-FI: 32 entries
sgn-MT: 0 entries
sgn-MY*: 3 entries
sgn-NG: 0 entries
sgn-NZ: 37 entries
sgn-PH: 693 entries
sgn-PL: 1644 entries
sgn-SI: 14 entries
sgn-TW: 57 entries

(*I'll note that I questioned the merits of this scheme precisely in relation to signed languages of Malaysia because there are at least three distinct signed languages spoken in Malayasia and "sgn-MY..." leads to matching problems.)

They've used six that follow the examples of certain registered tags but that *cannot* be registered under 3066bis:
sgn-CA-fr "Quebec SL": 5 entries
sgn-CH-fr "Swiss-French SL": 622 entries
sgn-CH-it "Swiss-Italian SL": 5 entries
sgn-ES-ct "Catalonian SL": 11 entries
sgn-GB-ie "Northern Ireland SL": 54
sgn-US-lit "ASL Literature": 22

They've also used three that actually contradict ISO 3166-1:
sgn-AE "Arabic Nations" (AE = United Arabic Emirates) -- 4 entries
sgn-EO "Signuno" (EO is unassigned) -- 2808 entries
sgn-WO "International Signs" (WO "must not be used in ISO 3166-1"): 42 entries

We need not have rushed back then since it appears they were going to do their own thing whether we registered tags or not.

John Cowan has insisted that we can't deprecate the existing tags, but we might keep in mind that tags have been deprecated before, and for these cases we have one known user. And it's not like they got a million records: for a handful of tags they have thousands of records in their lexical DB:

sgn-CH-de: 4765
sgn-US:    4515
sgn-DE:    3061
sgn-BE-nl: 2235
sgn-BE-fr: 1381


For most of the registered tags, the numbers are rather smaller:
sgn-NI: 894
sgn-FR: 840
sgn-NO: 379
sgn-GB: 183
sgn-JP: 172
sgn-MX: 172
sgn-BR: 171
sgn-GR:  24
sgn-CO:  21
sgn-ES:  21
sgn-IE:  17
sgn-IT:   9
sgn-NL:   9
sgn-SE:   3
sgn-DK:   0
sgn-PT:   0

And as mentioned above, they don't appear to use sgn-ZA at all.




> I think that Sign Languages work differently and interact differently
> with spoken languages than spoken languages interact with one
> another. I think they ought to have their own namespace (e.g., sgn-),
> not just be lumped in with 6000 other spoken languages.

We're not discussing records that document how languages work; we're talking about metadata elements that merely *identify* languages, and that get used in processes for matching content to requests based on language identity.


> I think the
> Ethnologue taxonomy doesn't handle "signed spoken languages" well,
> and I think that all of those three-letter codes should be pulloed
> from 639-3 until we can discuss and clarify the matter with Gallaudet
> etc. They can always been added to 639-3 in due course. But I think
> the whole area needs another serious look. (And I'm willing to put
> the RFC 3066 tags on the table if the 639-3 tags are.)

Please clarify that last statement.



Peter Constable