Language tags that denote writing systems

Jon Hanna jon@spin.ie
Mon, 9 Dec 2002 16:32:40 -0000


> "Can I have a sound file please? zh would be acceptable." is
> worse nonsense; it's unambigious what yi-Latn would sound like,
> but what zh would sound like is very ambigious.

Exactly. Jumping from "yi-Latn" to "yi" isn't reliable without a lot of
information about each code (actually that information would be very useful,
an RDF service with subDialectOf predicates would be cool, but that's
another matter).
The solution to this ambiguity is to not lump orthographical distinctions in
with the language code.

The language tags are already geared towards
> writing in many cases; give tags as needed to differentiate sound
> content, and accept that the tags won't always clearly divide
> both perfectly.

I don't think so. The current use, and hence the current registrations
(aside from the sgn-* registrations) may, but I think 3066 was very
successful in arriving at a reasonable way to encode languages.