Template talk:mr-noun-ī-f

@Aryamanarora Hi! Marathi has a character called 'eyelash ra'. Unicode has chosen to represent this with ऱ [1].

'Eyelash ra' appears in the declined forms of words with र as the final consonant of the word's stem such as नोकरी (nokrī). In these declined forms र would be replaced ऱ in a declension template. Since a template cannot automatically detect the final character of the word stem, a declension template would require some way to manually tell it that the final character of the word stem is र. The best solution I could come up with was to add the parameter '|r=1' when the declension template is used. So with this solution नोकरी (nokrī) would have {{mr-noun-i-f|नोकर|nokr|r=1}} in its declension section, and the template {{mr-noun-i-f}} (and all applicable declension templates) would have '{{#if:{{{r|}}}|़|}}' when a declined form requires it. Do you know if there's better way to tell the template that the final character of the word stem is र? Kutchkutch (talk) 22:48, 30 October 2017 (UTC)

@Kutchkutch: I think this is a sign that we should start using modules for declension templates. AFAICT, Nepali's conjugation template is entirely Luacized and so {{ne-conj}} is really lightweight, similarly MOD:sa-decl automatically handles the quirks of Sanskrit declension that are really difficult to implement in wikicode, such as the Vedic accent and the various poetic and archaic variants. I don't know much Lua, but I could try to make a MOD:mr-decl that could automatically detect the stem, with some help from our resident Lua-experts ^_^. —Aryaman ^{(मुझसे बात करो)} 01:34, 31 October 2017 (UTC)

Also, the templates seem to be missing some parameters? At नोकरी (nokrī) I still see triple braces for direct plural. —Aryaman ^{(मुझसे बात करो)} 01:35, 31 October 2017 (UTC)

@Aryamanarora: Thanks for the information about modules! There are perhaps more situations like the one I tried to explain that a module could be better suited for (such as stem-final स → श in some declined forms as in आरसा (ārsā)). I wish there was a way to learn Scribunto/Lua fast so that I could at least make a few edits Module:mr-translit on my own and try to make a Module:mr-IPA just like Sagir recently made Module:as-IPA for Assamese. There appear to be a few guides on MediaWiki and WikiBooks.

It's interesting that you mentioned Nepali because according to the Unicode link I provided above, the 'eyelash ra' is also present in Nepali. Wyang appears to have done all the module work for Nepali. I have discovered Module:ne-utilities, but I didn't know there was a module Sanskrit. Even though the lemma form of Marathi verbs has -णे following the verb stem, it's rarely pronounced that way in actual speech except in formal contexts, and it's actually pronounced closer to /ɳə/ and this pronounciation is orthographically written as -णं. Nepali words such as चिनी (cinī) has ', pronounced चिनि (cini)' so perhaps such a system could be useful for Marathi verbs and neuter words with word-final ए such as सोने (sone).

Thanks for your willingness to try to make a MOD:mr-decl! Perhaps that would reduce the number of templates needed for each paradigm if it knows the last two characters a word. If I ever get to verb inflection templates, there could be quirks there too that I haven't found about yet.

Re: नोकरी (nokrī): I made some significant implementation changes at T:mr-decl-noun and I haven't updated all the templates using it yet to adjust for the changes made there. One of the changes was renaming all the parameters. At first there were only a few parameters so I gave them names such as 'OS', but as the number of parameters increased I named them 'PA', 'PS', etc. After a while the meaning of each parameter became less clear especially with the default numbered parameters such as '1', '2', '3', etc. So to make the meaning of each parameter clearer, I renamed them with more descriptive names. The correct direct plural of नोकरी (nokrī) is नोकऱ्या with the 'eyelash ra' (Some fonts may not support the correct appearance of ऱ्).

I've tried to limit the number pages the declension is deployed on until all the declension paradigms have a template and until many of the issues can be addressed so that they don't show incorrect declined forms. If T:mr-decl-noun had been deployed on more pages before the changes, the changes would have caused even more pages to have undefined parameters.

I was thinking of making T:mr-noun-c-f, but I found out that feminine nouns with a word-final consonant are declined in two ways with one being more common than the other. So perhaps this means there should be a T:mr-noun-c-f-exception for the less common way with around ~300 nouns (according to one source) being declined that way.

Do you think it's better to make a thousand basic entries first and then worry about declension/inflection at a slower pace? Or is it better to address this first before there's hundreds of entries that will need to have inflection/declension templates deployed on manually? Kutchkutch (talk) 02:08, 31 October 2017 (UTC)

@Kutchkutch: Ideally, a module would work without having to be told the stem (unless it's a case where the declension is irregular). We could always ask someone with a bot to add the templates later. —Aryaman ^{(मुझसे बात करो)} 13:31, 31 October 2017 (UTC)