Sign languages (was: Re: additions to ISO 639 and the IANA language subtag registry)

Peter Constable petercon at
Tue Feb 21 22:54:14 CET 2006

> From: ietf-languages-bounces at [mailto:ietf-languages-
> bounces at] On Behalf Of Michael Everson

> Signed English in the US and Ireland is different
> from signed English in the UK. 

Just like Pinyin and Wade-Giles romanizations of Chinese are different.
It's still Chinese in both cases, so the tag should begin with "zh"; the
fact that it is Latin transliteration is the next most significant
attribute, and the choice of Romanization is a further refinement of
that. Similarly, the language is English whether signed using the
gestural "notation" conventions of US or of Ireland; the next most
significant attribute is the signed modality, and the choice of signing
conventions is a further refinement of that.

> >  > If there is a future revision to RFC 3066bis to incorporate the
> >>  forthcoming ISO 639-3, which is based on Ethnologue codes but
> >>  should replace "Ethnologue" in our discussions, then I strongly
> >>  including the ISO 639-3 code elements for sign languages as
> >>  language subtags at that time.
> I think this is a mistake. I think Ethnologue's
> taxonomy is immature with regard to these
> languages.

Taxonomy? What exactly does taxonomy have to do with this? If we weren't
sure what exactly the genetic relationships were between (say) English,
French, Basque and Spanish, that wouldn't make the tags "en", "fr", "eu"
and "es" any less useful.

> >I, on the other hand, support keeping the 'sgn' subtag with the
> >country codes, for reasons of backward compatibility.  The difference
> >between (true) signed and spoken languages is huge, much larger than
> >barriers between different signed languages (it is not unknown for
> >with radically different languages to learn to communicate in one
> >In addition, backward compatibility is very important.
> Country designations (or some other designation)
> are necessary in some instances.

The key here is "or some other designation". In no case is a country
subtag *necessary* to provide a distinction between language identities.

> >     Adamorobe SL  [sgn-ads] (Ghana)
> I think using the 2-letter code for Ghana plus a
> modifier might be better here. That is what we
> used for Martha's Vineyard SL, isn't it?

Blech! So you want to have "sgn-GH" for 'Ghanian SL' and something like
"sgn-GH-admrb" for 'Adoromobe SL', even though the two are probably
quite distinct, if even related at all. That will work beautifully
(NOT!) with left-prefix matching algorithms.

> In general I favour whaat John has suggested. I
> would LIKE to work with Phil Blair and his
> contacts at Gallaudet on this. Can we take time
> to do that? It will take time.

Frankly, it seems to me that a need to take some time working out a
scheme arises only when we've got a hair-brained scheme that tries to
use country subtags in some cases and (necessarily) not in others. If we
just use treat signed languages like any other normal language using
tags like "ads", or if we agree on a consistent template using alpha-3
extlang subtags e.g. "sgn-ads", then there's nothing to be worked out as
far as the tagging scheme is concerned; the only open issue is
identifying what are all the distinct signed languages out there.

Peter Constable

More information about the Ietf-languages mailing list