Oriya and Odia

[Doug Ewell]

This is a discussion about subtags for the Odia (formerly "Oriya")
language, and specifically about whether we should add a single,
secondary Description field beyond what appears in ISO 639-3.

In 2011 the Indian constitution was amended to change the spelling of
the language name "Oriya" to "Odia," and the spelling of the state name
"Orissa" to "Odisha." In brief, the new spellings were considered better
representations of actual pronunciation and less evocative of the
British colonial era in India. These changes are less drastic than some
other name changes that have occurred in India, such as the renaming of
Madras to Chennai in 1996.

The changes were apparently met with a great deal of celebration and
displays of pride in Odisha. However, both old and new names have
continued in use, especially outside India. Many software localization
efforts, including CLDR, have switched to the new name; for example,
Windows 10 has "Odia" language settings instead of the previous "Oriya."

In the ISO 639-3 data files dated 2016-01-15, reflecting the 2015 change
series, one of the changes was an name update to code element [ory]:

2015-05-05:
	Reference name: Oriya (individual language)

2016-01-15:
	Reference name: Odia
	Additional name: Oriya (individual language)

The "(individual language)" part exists because the original ISO 639-3
code element for Odia, [ori] was converted to a macrolanguage in 2012,
encompassing both Odia proper [ory] and Sambalpuri [spv], which is
claimed to have "75-76% of lexical similarity" with Odia. The
macrolanguage code [ori], which is implemented in the Language Subtag
Registry using its 2-letter, ISO 639-1–based form ('or'), was not
renamed and still has only one name:

	Reference name: Oriya (macrolanguage)

There is a third language, Adivasi Oriya [ort], which was also not
renamed. Ethnologue says that Adivasi Oriya has only "38%–42%" lexical
similarity "with standard Odia." It is not encompassed by the [ori]
macrolanguage and is not discussed further here.

It seems inconsistent (to me) that the name associated with the
individual language [ory] was changed but the name for the macrolanguage
[ori], which probably sees much wider use, was not.

Section 3.1.5 of RFC 5646 says that for subtags in the Registry derived
from core standards, the first Description field corresponds to the name
used in the core standard, but additional Description fields can be
added (at the Reviewer's discretion). So for 'ory' we need to apply the
2016-01-15 changes above, and for 'or' we need to keep the existing
Description field. However, we can — and I argue that we should —
add a second Description field:

	Type: language
	Subtag: or
	Description: Oriya (macrolanguage)
	Description: Odia (macrolanguage)
	Added: 2005-10-16
	Suppress-Script: Orya
	Scope: macrolanguage

Then, for consistency, we can apply a modification of the ISO name to
the first Description field for 'ory', without (IMHO) violating the
intent of 3.1.5:

	Type: language
	Subtag: ory
	Description: Odia (individual language)
	Description: Oriya (individual language)
	Added: 2012-08-12
	Macrolanguage: or

The names are out of order with respect to each other ("Oriya" first for
'or', "Odia" first for 'ory') because of the requirements of 3.1.5, but
the bottom line is that the Registry would show both names for both
subtags. This would allow implementers of BCP 47 to use either name
without deviating from the Registry unnecessarily, and should satisfy
those who prefer either the "old-school" name or the "new-school" name.

The alternative would be to adhere strictly to 639-3 names as shown
above, without adding anything new.

I will post records and registration forms for whichever option is
decided upon, as soon as the Reviewer declares consensus one way or the
other. The remaining 639-3–based changes are now in their two-week
review period, and will not necessarily have to wait for the Odia issue
to be resolved.

--
Doug Ewell | http://ewellic.org | Thornton, CO 🇺🇸