Re: BPMLOD and string metadata

Am Do., 2. Feb. 2023 um 11:37 Uhr schrieb r12a <ishida@w3.org>:

> hello Christian,
>
> BCP47 does in fact represent all the ISO 632-3 languages.
>

You're right, ISO 639-3 had been added. What led to my confusion is the
following paragraph in the spec:

"When a language has no ISO 639-1 two-character code and the ISO 639-2/T
(Terminology) code and the ISO 639-2/B (Bibliographic) code for that
language differ, only the Terminology code is defined in the IANA registry."

I think that should be updated to reflect where ISO 639-3 stands in the
ranking. It is based on ISO 639-2/T, but has a considerable number of
internal redefinitions. For example, I have the strong impression that ISO
632-2 frs ("East Frisian") referred to a dialect of the Frisian language
(the Saterland dialect) whereas ISO 632-3 frs ("East Frisian" refers to a
dialect of Low German spoken in Eastern Frisia). I'm thinking this because
ISO 632 subclassifies Frisian into its member branches, but doesn't do that
for Low German ("nds"). ISO 639-3 subclassifies Low German ("nds", "wep",
"frs", "pdt", numerous codes for Low German in the Netherlands, and, along
the way, redefines the "nds" language tag to everything not assigned to a
particular dialect), and this actually seems to be triggered by the
existence of a separate language code for East Frisian (this is hinted at
by https://iso639-3.sil.org/code/frs *officially* pointing to
https://en.wikipedia.org/wiki/East_Frisian_Low_Saxon).

Anyway, this example neatly illustrates that leveraging ISO language codes
from different institutions and highly underspecified documentation (esp.
for ISO 639-2 and ISO 639-1, which only give the name) ends up in something
rather messy. In consequence, some people produce invalid BCP47-alike tags
by just using ISO639-3 tags or URIs  all the time. This was done in the
GOLD ontology, for example (seems to be offline, but cf. the second entry
under https://lov.linkeddata.es/dataset/lov/terms?q=prefix*).

Best,
Christian