Template talk:pra-mah-decl-adj

@शब्दशोधक: When the hiatus-filler 𑀬 (ya) does not precede 𑀅 (a) or 𑀆 (ā), it should be removed. See prakrit.info/prakrit/grammar.html?r=phonology

This y-sound is properly found only between two a-vowels (i.e., a or ā).

For example, for masculine/neuter instrumental singular of 𑀓𑀺𑀡𑀺𑀬 (kiṇiya):

𑀓𑀺𑀡𑀺𑀬 (kiṇiya) + 𑀏𑀡 (eṇa) = 𑀓𑀺𑀡𑀺𑀏𑀡 (kiṇieṇa) instead of *𑀓𑀺𑀡𑀺𑀬𑁂𑀡 (*kiṇiyeṇa)

Since T:pmh-decl-adj just uses MOD:pra-decl/noun and this issue also affects nouns, it should be addressed at MOD:pra-decl/noun by someone who can code. So:

𑀚𑀡𑁆𑀡𑁄𑀯𑀯𑀻𑀬 (jaṇṇovavīya) + 𑀏𑀡 (eṇa) = 𑀚𑀡𑁆𑀡𑁄𑀯𑀯𑀻𑀏𑀡 (jaṇṇovavīeṇa) instead of *𑀚𑀡𑁆𑀡𑁄𑀯𑀯𑀻𑀬𑁂𑀡 (*jaṇṇovavīyeṇa)

This issue is one of the reasons why I made the non-𑀬 (ya) forms the main entries and made T:hiatus-filler form of redirects for 𑀬 (ya) forms. However, ideally both forms need declension. Kutchkutch (talk) 10:42, 4 March 2021 (UTC)Reply

@Kutchkutch: For now, I'll just remove the declension of 𑀓𑀺𑀡𑀺𑀬 (kiṇiya) since it's merely an alt form. 🔥शब्दशोधक🔥 14:19, 4 March 2021 (UTC)Reply

• RichardW57, RichardW57m Do you know how to fix this error?

  When the hiatus-filler 𑀬 (ya) does not precede 𑀅 (a) or 𑀆 (ā), it should be removed

  With this error, the declensions shown at 𑀡𑀺𑀓𑁆𑀓𑀺𑀬, 𑀩𑀁𑀤𑀻𑀓𑀬, 𑀆𑀡𑀁𑀤𑀺𑀬 would be considered incorrect (Svartava2). Many sources transliterate this hiatus-filler as ẏ, so should this be shown in the automatic transliteration to distinguish it from the etymological y (in Magadhi)? @Svartava2 Would य़ be the Devanagari representation of ẏ?Kutchkutch (talk) 03:21, 4 December 2021 (UTC)Reply

  Kutchkutch य़ seems very rare, even {{R:pra:Sheth}} just gives य. It's better and simpler to keep it as it is. —Svārtava 05:43, 4 December 2021 (UTC)Reply

  Yes. At the user level, I suggest adding an optional parameter |filler=. Giving a more general answer than I think is needed:

  1. Certain consonants are never hiatus consonants.
  2. Certain consonants are always hiatus consonants.
  3. Certain consonants are hiatus consonants unless the |filler= says it is not in this word.
  4. Certain consonants are only hiatus consonants if |filler= says it is in this word.

@Kutchkutch, Svartava2: I suggest the valid values be |filler=yes and |filler=no, य़ and its equivalents be treated as fillers by default, while I must take advice on the default assumption for य and its equivalents. I think by default य should NOT be a filter. --RichardW57m (talk) 11:16, 10 December 2021 (UTC)Reply

RichardW57, RichardW57m That seems fine. Hopefully, this solution would also work for terms without the hiatus filler such as 𑀕𑀅. Other than {ẏa / य़}, {va / व} is considered as secondary hiatus filler that is less productive than {ẏa / य़}. The secondary hiatus filler {va / व} is not indicated in transliteration (see 𑀣𑁄𑀅). As discussed at User_talk:Kutchkutch#ಬೋಲ್ಲಇ & Talk:𑀯𑀅𑀁, perhaps the Brahmi script does not have a way to distinguish {ẏa / य़} and {ya / य}, which may be an issue when converting between scripts. Kutchkutch (talk) 18:06, 13 December 2021 (UTC)Reply

@Kutchkutch, Svartava: Does <v> come and go in declension? <y> can make a surprise appearance. For example, in Ardhmagadhi, nāï 'kinsman' has a plural nāyao. It feels regular for Ardhamagadhi, but perhaps the ending is infrequent enough that its appearance alone can be treated as an irregularity. I've found Ardhamagadhi declension tables that completely ignore the -a(y)o type :-) but include the -avo type for masculine -u stems :-(. --RichardW57m (talk) 15:49, 30 May 2022 (UTC)Reply

@RichardW57, RichardW57m: It seems what you're asking is whether or not the <v> is always retained in declension when the penultimate letter of the stem is <v> (such as the <v> as 𑀣𑁄𑀯 (thova)) and the following vowel is not a or ā. {{R:inc:Woolner|135, 203}} mentions the nominative singular of Jain Maharastri 𑀣𑁂𑀅 (thea) as 𑀣𑁂𑀯𑁄 (thevo) = 𑀣𑁂𑀑 (theo), and the first half of line 596 in the Jain Maharashri text Maṇipati-Carita is

𑀊𑀲𑀸𑀲𑀸𑀑 𑀧𑀸𑀡𑀽 𑀢𑀑-𑀯𑀺 𑀣𑁄𑀯𑁄 𑀢𑀑-𑀯𑀺-𑀬 𑀇𑀦𑀼𑀳𑀼𑀢𑁆𑀢𑁄

ūsāsāo pāṇū tao-vi thovo tao-vi-ya inuhutto

so the <v> appears to be retained in the declension of Jain Maharastri. Kutchkutch (talk) 01:34, 31 May 2022 (UTC)Reply

@Kutchkutch: Actually Woolner says "theo = thevo", which should mean 'Prakrit form theo, 'Sanskrit form thevo ', but as he doesn't have a Sanskrit comparandum, he may mean 'Pali form thevo'. As he has it on p127 as part of theva-kaleṇa, also in JM, the 'v' isn't stable (which doesn't come as a shock), so we don't know whether its presence is affected by inflection. In general, the chances of 'v' dropping seem to depend on what vowels flank it, with 'a' encouraging retention. It looks as though there is a lot of work to do in this area. I'm inclined to be bold and assume that the ending does not significantly affect the retention of 'v'. You give me the distinct feeling that the answer is not being taught - possibly it isn't even known. --RichardW57m (talk) 17:16, 31 May 2022 (UTC)Reply

@RichardW57, RichardW57m: Yes, the order in Woolner is

𑀣𑁂𑀑 (theo) = 𑀣𑁂𑀯𑁄 (thevo), JM.

which Pischel has as

𑀣𑁂𑀯 (theva) AMg. JM. Pali, 130, 207

𑀣𑁂𑀅 (thea) M. S. Mg. 90, 230, 307

Kutchkutch (talk) 19:16, 31 May 2022 (UTC)Reply

@Kutchkutch: That last list of references appears to be for the form thoa (not indexed as such in the translation of Pischel). The alarming point is that Pischel quotes a JM stem theva, while Woolner exhibits a JM nominative singular theo. Pischel discusses rules for retention and loss at paragraphs 199 and 201. If we do have alternation (as opposed to a doublet as in the case thoa/thova), is the locative singular thee or theve? I much prefer what I did in Pali, which was to allow overrides (supplement or replacement) for specific forms in an otherwise regular paradigm. For Prakrit, there seems to be a belief that Indic script forms can be generated from a common Latin transliteration¹, so if you went that way (I don't see myself ploughing through Prakrits' indexes to collect the inflected forms²), you could allow overrides to be specified in the Latin script, so that cut and paste doesn't overly become cut, paste and transliterate. (Prakrit also exists in Grantha and Siddham, so be warned.) --RichardW57m (talk) 12:27, 1 June 2022 (UTC)Reply

¹ Pali has a lot of local subtleties that aren't captured by the standard transliteration - imagine if Devanagari ण were encoded differently for forms with and without a sloping bar below!.

² Recall Johnson's definition of a lexicographer as a 'harmless drudge'. --RichardW57m (talk) 12:27, 1 June 2022 (UTC)Reply

@Kutchkutch, Svartava, AryamanA: Are these supposed to be tolerated even though the Latin script is excluded from the language's script? (There's at least one useful source of quotes which gives the text only in the Roman script.) At the moment they're generating module errors, and I am current watching cat:E to convert the feminines to the same script as the masculine. --RichardW57 (talk) 13:56, 3 June 2022 (UTC)Reply

@RichardW57: Yes, they can be converted into the page-title's script. Earlier, though, when Latin script was entered in |f= it didn't trigger any module error. —Svārtava (t/u) • 14:00, 3 June 2022 (UTC)Reply

@Svartava: That's why I'm asking. I don't know whether to restore tolerance or merely make the error message more helpful. So far I've only found two cases, one of which I'm keeping for now for testing of the change. --RichardW57 (talk) 14:03, 3 June 2022 (UTC)Reply

@RichardW57: I think it's best to keep disallowing Romanisations in the parameters since now the declension is not only available in Brahmi but also other scripts. It's more consistent that way. —Svārtava (t/u) • 14:10, 3 June 2022 (UTC)Reply

@RichardW57, RichardW57m: Although it may be a bit more difficult to type and read the other scripts, I agree with I think it's best to keep disallowing Romanisations in the parameters ... It's more consistent that way.. By comparison, T:inc-extension does support MOD:typing-aids's system for Latin to Brahmi since the page title may not be in a script used for Prakrit. Kutchkutch (talk) 19:13, 3 June 2022 (UTC)Reply

Indeed, for there are languages like Romany and Pali. Non-Prakrit script feminines now provoke what I hope are helpful error messages. I only found two cases where a Roman script feminine had survived to the present. --RichardW57 (talk) 20:34, 3 June 2022 (UTC)Reply