Re: [EXTERNAL] Re: language identifiers for sign languages (incl. sgn) vs. attribute for indicating the representation of an individual language in "sign language modality"

Hi all,

I agree that the =91t=92 extension is not appropriate for the purpose we ar=
e trying to capture.

While I'm reluctant to propose an expansion to an already complex system, I=
'm wondering if there is openness to something like a "modality" tag that c=
ould be used to describe the expressed form of any language.

My thinking is that ISO 639-4 specifically refers to three different modali=
ties (written, spoken, signed). If we want to describe a language according=
 to its modality then why restrict that to just sign?

David.


________________________________________
From: Peter Constable <[email protected]>
Sent: Friday, November 22, 2019 6:23 PM
To: Doug Ewell; Christian Galinski; Fourney, David
Cc: ietf-languages; 'Sebastian Drude'; [email protected]
Subject: RE: [EXTERNAL] Re: [Ietf-languages] language identifiers for sign =
languages (incl. sgn) vs. attribute for indicating the representation of an=
 individual language in "sign language modality"

The scope of the =91t=92 extension is linguistic content that has undergone=
 some type of transform in its expression, and signed modality for a spoken=
 language could be considered a transform. But the =91t=92 extension as cur=
rently defined doesn=92t support this. What is supported is primarily deali=
ng with text transformations. Also, the way the =91t=92 extension works is =
that the additional information declares what content was transformed _from=
_, not what it is transformed _into_. For signed modality of spoken languag=
es, what=92s needed is a way to indicate signed modality as the final expre=
ssion, not the source.

So, I don=92t think the =91t=92 extension is appropriate.

I think a variant subtag =93signed=94 or =93signmod=94 would be better. The=
 main problem that would arise is that this is very generic (it could be us=
efully applied to any oral language), which there has been resistance to in=
 the past. A smaller issue is that, while variant tags for specific signed-=
modality variants could be registered, it might make sense to use a subtag =
sequence along the lines -signed-modvarnt, but it=92s currently not possibl=
e to specify a prefix as anything other than a valid language tag. (E.g., *=
-signed can=92t be a prefix specification.) That wouldn=92t be a problem as=
 long as the signed-modality variant is specific to a particular language, =
as would be the case for (e.g.) Signed Exact English.



Peter

From: Ietf-languages <[email protected]> On Behalf Of Doug Ew=
ell
Sent: Friday, November 22, 2019 1:05 PM
To: Christian Galinski <[email protected]>; 'Fourney, David' <da=
[email protected]>
Cc: ietf-languages <[email protected]>; 'Sebastian Drude' <Sebastian.=
[email protected]>; [email protected]
Subject: [EXTERNAL] Re: [Ietf-languages] language identifiers for sign lang=
uages (incl. sgn) vs. attribute for indicating the representation of an ind=
ividual language in "sign language modality"

Hi Christian,

> Many true sign languages (se definitions below), such as =93ase=94
> (American Sign Language [ASL], which /fictively/ might even have a
> Newfoundland and Labrador variety =96 to be coded ase-CA-NL in line
> with BCP47 rules) have already a language identifier.

This example is actually not valid BCP 47 syntax. The use of ISO 3166-1 cou=
ntry codes as region subtags doesn't extend to appending ISO 3166-2 subdivi=
sion codes directly. You would need to use "ase-u-sd-canl" or "ase-CA-u-sd-=
canl". See UTS #35, Section 3.6.5.

> The question to Doug is, how the BCP and Unicode rules deal with the
> above-mentioned difference between (true) =93individual sign languages=94
> and the =93signed language modality=94 (as a sort of =93transform=94 of a=
ny
> individual language)?

I don't believe there are or should be any "Unicode rules" (which I assume =
refers to CLDR and the 't' or 'u' extension) that deal with this.

One approach would be to request a variant subtag, such as 'signed', to rep=
resent the signed modality of a spoken language, such as (but not limited t=
o) Signing Exact English. See RFC 5646, Section 2.2.5 for details on varian=
t subtags and Section 3.6 for details on requesting a registration.

However, some may argue that modality is beyond the scope of BCP 47 variant=
s and would suggest a CLDR extension to deal with this within the 't' exten=
sion framework. In that case, your best bet would be to contact cldr-contac=
[email protected]<mailto:[email protected]> .

--
Doug Ewell | Thornton, CO, US | ewellic.org<https://nam06.safelinks.protect=
ion.outlook.com/?url=3Dhttp%3A%2F%2Fewellic.org&data=3D02%7C01%7Cpetercon%4=
0microsoft.com%7C19558599ac7d421157fc08d76f8fc274%7C72f988bf86f141af91ab2d7=
cd011db47%7C1%7C0%7C637100535555481188&sdata=3D5sWZ089qfuVRPWbmetKrFDHskz%2=
BETA2vY0ioACdSzos%3D&reserved=3D0>


-------- Original Message --------
Subject: language identifiers for sign languages (incl. sgn) vs.
attribute for indicating the representation of an individual language in
"sign language modality"
From: "Christian Galinski" <[email protected]<mailto:christian.g=
[email protected]>>
Date: Fri, November 22, 2019 11:48 am
To: "'Fourney, David'" <[email protected]<mailto:[email protected]=
a>>
Cc: <[email protected]<mailto:[email protected]>>, "'Sebastian Drud=
e'"
<[email protected]<mailto:[email protected]>>, <doug@ew=
ellic.org<mailto:[email protected]>>


Dear David,

First I have to apologize for my long silence =96 I was absorbed with work =
on several standards.

We are now at a crucial moment where things need to be clarified in ISO 639=
-4 =93language coding=94 (and ISO/TR 21636 =93Language varieties=94) =96 in=
cluding your issue of how to identify =93individual sign languages=94 (i.e.=
 true individual sign languages, which are not just a modality of spoken la=
nguage) and the =93signed language modality=94 which is a signed representa=
tion of a spoken language).


  1.  concerning the difference between =93individual sign languages=94 and=
 =93signed language modality=94, the use of the language identifier =93sgn=
=94 (in library use) is confined to an unidentifiable individual sign langu=
age =96 it is NOT referring to a =93signed language modality=94. According =
to the fundamental rules of language coding, we cannot change the scope of =
=93sgn=94, nor ignore the difference between sign language and the signed l=
anguage modality.
Therefore, for the sign language modality we need an =93attribute=94 to be =
added to the language identifier of an individual language, e.g. if the sig=
n language modality of the type of =93Signing Exact English=94 is used.
  2.  However, I could not find an identifier for signed language modality,=
 nor a mechanism for inserting an identifier for this in:
https://tools.ietf.org/html/bcp47<https://nam06.safelinks.protection.outloo=
k.com/?url=3Dhttps%3A%2F%2Ftools.ietf.org%2Fhtml%2Fbcp47&data=3D02%7C01%7Cp=
etercon%40microsoft.com%7C19558599ac7d421157fc08d76f8fc274%7C72f988bf86f141=
af91ab2d7cd011db47%7C1%7C0%7C637100535555486190&sdata=3DPWlDE0pdRCgLBG4wsnp=
rwit5%2B6EeB%2Fux%2FiApkJkmweg%3D&reserved=3D0>
https://tools.ietf.org/html/rfc6497#ref-UTS35<https://nam06.safelinks.prote=
ction.outlook.com/?url=3Dhttps%3A%2F%2Ftools.ietf.org%2Fhtml%2Frfc6497%23re=
f-UTS35&data=3D02%7C01%7Cpetercon%40microsoft.com%7C19558599ac7d421157fc08d=
76f8fc274%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C637100535555491183&s=
data=3DY3Zi1erRIWT%2F8K%2F5ZhtfSPCofmTkczyny89RagNWmhA%3D&reserved=3D0>
http://unicode.org/reports/tr35/<https://nam06.safelinks.protection.outlook=
..com/?url=3Dhttp%3A%2F%2Funicode.org%2Freports%2Ftr35%2F&data=3D02%7C01%7Cp=
etercon%40microsoft.com%7C19558599ac7d421157fc08d76f8fc274%7C72f988bf86f141=
af91ab2d7cd011db47%7C1%7C0%7C637100535555496180&sdata=3DfezBI46al7DmxtciBBI=
wI7Fj%2Fuuyor7d8uB7xdyvzM4%3D&reserved=3D0>
The regular order of attributes to a language tag (language identifier) is =
=93lang-geogr=94 (dialect), or =93lang-script=94 (language written in a cer=
tain script) or =93lang-script-geogr=94 (language in a script in a certain =
region).. In between, a =93t=94 (for =93transform=94 in the meaning of tran=
scription, transliteration, translation or other) may be inserted.

>From your experience/problems with video technology (and HTML), the questi=
ons to you would be:

  1.  Many true sign languages (se definitions below), such as =93ase=94 (A=
merican Sign Language [ASL], which /fictively/ might even have a Newfoundla=
nd and Labrador variety =96 to be coded ase-CA-NL in line with BCP47 rules)=
 have already a language identifier.
Does it need another attribute to further specify them as a sign language? =
In that case, an attribute must be found which is different from =93sgn=94.=
 How could it look like?
  2.  In the case of a signed language modality, such as =93Signing Exact E=
nglish=94 the core language identifier for English would be =93eng=94. It w=
ould need an attribute to identify it as the signed language modality (whic=
h could be followed by a country code, if there are =93dialects=94 of /fict=
ive/ eng-xxx-AUS meaning =93Signing Exact English as used in Australia=94. =
What could =93xxx=94 indicating =93signed language modality look like?
  3.  It probably would not help to use an attribute identifier =93Xxxx=94 =
in the slot of =93script code=94, as a signed language modality might sligh=
tly differ depending on the script used, even if it is the same spoken lang=
uage (represented in different scripts in different areas/communities).
  4.  Could the =93t=94 (transform) symbol be of help =96 as a given signed=
 language modality somehow is a =93transformation=94 of a spoken language?


  1.  The above questions (resp. the answer to them) could have an impact o=
n ISO 639 and ISO/TR 21636 insofar as we should not formulate provisions in=
 these documents which conflict with other standards. We should rather try =
to find generic solutions.

The question to Doug is, how the BCP and Unicode rules deal with the above-=
mentioned difference between (true) =93individual sign languages=94 and the=
 =93signed language modality=94 (as a sort of =93transform=94 of any indivi=
dual language)? see the respective terminology entries below

Best regards
Christian


p.s.
In the most recent revised version of ISO 639-4 we came up with the followi=
ng terminology entries:
individual sign language
NOT: signed language
individual language (3.1.3) having the visual-spatial language modality (3.=
5.1) as basic modality
Note 1 to entry: Usually =93sign language=94 appears as part of the name of=
 the respective individual language.
EXAMPLE: ASL (American Sign Language); )

signed language modality
NOT: sign language
visual-spatial language modality (3.5.1) that uses a combination of hand sh=
apes, palm orientation and movement of the hand, arm or body, and facial ex=
pression