Era code for the Indian National calendar

[sffc]

Currently it is "saka", but @robertbastian suggested in unicode-org/cldr#4519 (comment) that maybe it should be "shaka".

Which is the more correct transliteration to ASCII?

@Manishearth

[Manishearth]

Shaka seems acceptable, and in-use.

General rule of thumb: if there is a common accentless transliteration, we should use it, otherwise we should strip accents.

I certainly pronounce it Shaka. Technically, Shak because of schwa deletion in modern languages.

[justingrant]

If we do change the name here, then I'd suggest that the translation strings in CLDR should also be changed, because it'd be weird for the name to be "Saka" in CLDR but shaka in ICU4X. FWIW, we used "saka" as the era ID in the Temporal polyfill because we were copying what CLDR did.

new Intl.DateTimeFormat("en", {calendar: 'indian', era: 'long'}).format(new Date())
// => '1/12/1947 Saka'

[ptomato]

Temporal meeting 2025-04-03: We support whatever is the appropriate transliteration and agree with Justin's comment that the output strings should be changed to agree with whatever that is.

[sffc]

TG2 discussion: https://github.com/tc39/ecma402/blob/main/meetings/notes-2025-04-03.md#era-code-for-the-indian-national-calendar-31

We would like to see additional evidence to break the tie between "saka" and "shaka".

[Manishearth]

Wikipedia calls it both (with an accent on Śaka): https://en.wikipedia.org/wiki/Indian_national_calendar, https://en.wikipedia.org/wiki/Shaka_era, with a preference to Shaka.

Indic Wikipedias use some equivalent of शक or शाक (usually, शक संवत, संवत means era)

https://hi.wikipedia.org/wiki/%E0%A4%B6%E0%A4%BE%E0%A4%B2%E0%A4%BF%E0%A4%B5%E0%A4%BE%E0%A4%B9%E0%A4%A8_%E0%A4%B6%E0%A4%95

Googling "Shaka" and "Saka" gives me a fair number of similar results, with Google being smart enough to show me both. A lot of the pages for "Saka era" spell it "Śaka", but not all of them.

What's happening is that standard transliteration (IAST) for Indic languages uses diacritics; with unaspirated consonants all represented with a single letter. So श is represented as ś, and ष is represented as ṣ, despite both representing "sh" sounds.

In practice, people do not transliterate this way. IAST requires a lot of diacritics; most people transliterate more informally, using "h" digraphs like <sh> and <ch>, omitting retroflexing distinctions (IAST t vs ṭ), omitting vowel length distinctions (a vs ā), ultimately relying on peoples' vocabulary to disambiguate. Even amongst people who use diacritics, diacritics are mostly used for vowel length distinctions, and some people use capital/lowercase consonants to mark retroflexing. Some South Indian orthographies use the presence/lack of an h to mark retroflexing, as well. It's kind of all over the place, though श almost always gets represented with <sh>.

All of this is to say: IAST is used in an academic context, but is not very similar to the actual schemes people use. While for codes I'd typically be fine with using something academic, this is further complicated with the fact that we can't actually use IAST here, we can only use ASCII, and IAST is very much not ASCII friendly. The diacritic matters here.

An interesting example I found was this site, which spells it "Saka", but also spells the first month as "Chaitra" (which is not IAST, IAST would be Caitra), presumably because a diacritic got lost along the way. It's not uncommon for transliteration schemes to get mixed when people are used to spelling a word a certain way, but spell new words a different way.

Overall I think "shaka" is a spelling using common-but-nonstandardized transcriptions that are ASCII friendly, "śaka" is a spelling using an uncommon-but-standardized-and-academic transcription system, and "saka" is a version of that made to be ASCII-friendly, but in doing so loses information.

A wrinkle is that in most modern languages it's actually pronounced "shak", and if you said the word to people they woudl likely transcribe it "shak", but it's pretty common for people to consider certain words as "sanskrit-ish" when transcribing and transcribe with the deleted schwa included.

Is that a useful amount of evidence? I don't think you'll get a clear answer on what is used in the wild here beyond that, Indic transcription is a mess.

From looking at this I lean towards "shaka".

[sffc]

CLDR Issue: https://unicode-org.atlassian.net/browse/CLDR-18487

Kirill voiced support for Shaka.

I've updated unicode-org/cldr#4519

[ptomato]

Closed in #39

[justingrant]

If we've settled on shaka then:

• Should there be an alias of saka given that the experts on this thread had trouble coming to agreement, so end user developers might have the same trouble? Related: should śaka also be accepted as an alias, even though it's not ASCII?
• Should CLDR's localized output be changed from the current "Saka"? If yes, then should it change to "Shaka" or "Śaka"? And who's working with CLDR to make this change?

date = new Date('2000-01-01');
formatter = new Intl.DateTimeFormat('en-u-ca-indian', { era: 'long' });
formatter.formatToParts(date).find(part => part.type === 'era').value;
// => 'Saka'

[ryzokuken]

https://github.com/unicode-org/cldr/pull/4519/files#diff-5ea2936f709d9dc83508c70c9a589a5bdaadf9392a0ca96d9b56507621f4090bR4868

@justingrant thanks for raising the point about the alias, I'll make a tiny PR to add saka as an alias.