Template talk:grc-adecl

Doesn't seem to work with adjectives such as ὠκύς (ōkús), ὠκεῖα, ὠκύ yet:

— Eru·tuon 22:23, 1 March 2016 (UTC)Reply

It needs a length mark on the alpha. I could try and program it to assume one, though. —ObsequiousNewt (εἴρηκα|πεποίηκα) 15:58, 2 March 2016 (UTC)Reply

Oops. That is what the documentation says. It would be nice if the template could take -υς, -εια with no length marks, and there aren't adjectives in -υς, -είᾱ, so it wouldn't cause problems. But not necessary if it's difficult. — Eru·tuon 19:38, 2 March 2016 (UTC)Reply

I have removed the strict requirement for length marks on adjectives. Exercise caution, though; I'm not sure that removing the mark will work in all cases. —ObsequiousNewt (εἴρηκα|πεποίηκα) 15:49, 4 March 2016 (UTC)Reply

Sweet! I'll look out for any problems. — Eru·tuon 21:16, 4 March 2016 (UTC)Reply

The template gives ἡλίκος (hēlíkos) recessive accent, yielding *ἥλικος; ἡλίκη; *ἥλικον (*hḗlikos? hēlíkē? *hḗlikon) in place of ἡλίκος; ἡλίκη; ἡλίκον (hēlíkos? hēlíkē? hēlíkon). Probably would happen with ὀλίγος (olígos) too, though the inflection template hasn't been updated yet. Please fix! — Eru·tuon 03:28, 30 March 2016 (UTC)Reply

It needed length marks. —ObsequiousNewt (εἴρηκα|πεποίηκα) 15:31, 30 March 2016 (UTC)Reply

Looking to make a table of crasis for ἐμός (emós) (see here), but not sure if {{grc-adecl}} allows it... — Eru·tuon 07:25, 14 April 2016 (UTC)Reply

Looking through the declension of ἁπλοῦς, I noticed that the neuter forms in the plural (ἁπλᾶ, ἁπλοῦς) do not coincide. Is this correct? My (paper) grammar says, as would be expected, that nom. and acc. for the neuter forms should be identical (both ἁπλᾶ < ἁπλόα)...

NvdS 86.88.25.125 09:36, 16 April 2016 (UTC)Reply

Fixed, thanks. —ObsequiousNewt (εἴρηκα|πεποίηκα) 15:11, 19 April 2016 (UTC)Reply

@ObsequiousNewt, the declension for πεντεκαιδεκᾰπλᾰσῐ́ων (pentekaidekaplasíōn) should work but doesn't. —JohnC5 01:35, 27 April 2016 (UTC)Reply

@JohnC5 Consonant stems take as a single argument the bare stem of the word (πεντεκαιδεκᾰπλᾰ́σῐον, εὔχαριτ, etc.). This is necessary—while it would be possible to program the module to recognize -ων, -ον as third declension endings, the stem would ambiguously be either -ν- or -ντ-. —ObsequiousNewt (εἴρηκα|πεποίηκα) 01:44, 27 April 2016 (UTC)Reply

@ObsequiousNewt: Makes sense. Thanks! :) —JohnC5 01:49, 27 April 2016 (UTC)Reply

@ObsequiousNewt: Can we turn off the comparative? —JohnC5 02:01, 27 April 2016 (UTC)Reply

@JohnC5: If you want, but colourless green ideas sleep furiously. —ObsequiousNewt (εἴρηκα|πεποίηκα) 02:07, 27 April 2016 (UTC)Reply

@ObsequiousNewt: Thanks, Chomsky. :P I meant is there a way to remove the comparative and superlative forms from an inflection table of an adjective (perhaps, to take an example at random, the incomparable adjective πεντεκαιδεκᾰπλᾰσῐ́ων (pentekaidekaplasíōn))? —JohnC5 02:11, 27 April 2016 (UTC)Reply

If you set comp=|super=}} it should work. —ObsequiousNewt (εἴρηκα|πεποίηκα) 02:16, 27 April 2016 (UTC)Reply

So that's how to do the declension of μείζων (meízōn). Yay! (I should've asked.) — Eru·tuon 05:38, 27 April 2016 (UTC)Reply

@ObsequiousNewt: it's having trouble with the dative plural, though. For the two adjectives mentioned above, it's giving πεντεκαιδεκᾰπλᾰσῐ́οσῐ (pentekaidekaplasíosi) and μείζοσῐ (meízosi) instead of -ουσι. And for aorist participles like ἰδών (idṓn) the only way I can get it to work is by (1) giving the nonexistent form "ἰδούς" as parameter 1 and (2) explicitly providing the correct MNS as well as the NDP (which would otherwise appear as "ἰδονῐ"). —Aɴɢʀ (talk) 17:33, 2 November 2016 (UTC)Reply

@Angr: It's supposed to be -οσῐ. -νσ- only lengthens the preceding vowel if it comes from *ντ-σ (see Smyth 293, and Perseus bears him out.) I believe I have fixed the aorist participles.

@ObsequiousNewt: Wow, you're right. I've never seen a dative plural in -οσι before. As for the aorist participles, now it's giving the masculine nominative singular of ἰδών as "ἰδόν", the masculine dative plural as "ἰοῦσῐ" (lacking the δ, how did that happen???), and the neuter dative plural is still "ἰδονῐ". —Aɴɢʀ (talk) 20:10, 2 November 2016 (UTC)Reply

@Angr: Should be all fixed, at least for now. — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 20:22, 2 November 2016 (UTC)Reply

At σεσημειωμένος, the template is giving the nom./voc. fem. pl. forms antepenultimate stress instead of penultimate stress. —Aɴɢʀ (talk) 23:30, 23 November 2016 (UTC)Reply

User:Angr: Fixed. — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 20:44, 29 November 2016 (UTC)Reply

@ObsequiousNewt: Is there any way to get this template to work at πᾶς? I can't figure out a way. —Aɴɢʀ (talk) 22:00, 25 January 2017 (UTC)Reply

@Angr: I think I've got it. Had to make a few changes to the module. — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 19:07, 26 January 2017 (UTC)Reply

For adjectives like ἀληθής (alēthḗs), the table generates the masc/fem acc.pl form ἀληθέες in the uncontracted table and ἀληθεῖς in the contracted table, i.e. the same as the nom/voc pl. But according to Smyth §292 D, although the acc.pl is the same as the nom/voc.pl in the contracted forms, in the uncontracted forms (which appear only in Homer and Herodotus anyway) the masc/fem acc.pl is ἀληθέας. This needs to be corrected in the module. —Aɴɢʀ (talk) 12:45, 15 July 2017 (UTC)Reply

@Angr: I am not sure why "uncontracted" forms are listed at all. If they don't occur in a dialect, and presumably then are just hypothetical reconstructions of the Attic forms before contraction, they shouldn't be given, or should be marked with asterisks. If they do occur in a dialect, as these ones do, they should be marked with that dialect and maybe not with "uncontracted". Unfortunately, I don't know how to modify the module to do this. — Eru·tuon 16:44, 15 July 2017 (UTC)Reply

I've gone and changed the form you mention, though it's a makeshift solution... ideally there should be no "uncontracted" and "contracted" tables and you would just get the Attic table by default, and the Ionic with |dial=ion. — Eru·tuon 16:53, 15 July 2017 (UTC)Reply

Does someone know why the neuter column is left empty in ἑτερώνῠμος (heterṓnumos) and other proparoxytone two-endings adjectives? --Barytonesis (talk) 02:27, 22 July 2017 (UTC)Reply

@Erutuon --Barytonesis (talk) 02:30, 22 July 2017 (UTC)Reply

Nevermind, I see JohnC5 has been doing some work on the template, it probably comes from there. I will continue replacing instances of {{grc-adecl-1&2-eta-prx}}, it will be a minor inconvenience. --Barytonesis (talk) 02:35, 22 July 2017 (UTC)Reply

Fixed, sorry about that. —JohnC5 03:58, 22 July 2017 (UTC)Reply

Doesn't work with γήινος (gḗinos) (last instance of {{grc-adecl-1&2-eta-prx}}). --Barytonesis (talk) 03:19, 22 July 2017 (UTC)Reply

One last thing: for three-endings adjectives, neuter forms have been replaced by duplicated feminine forms. --Barytonesis (talk) 03:45, 22 July 2017 (UTC)Reply

@Angr, Barytonesis, Erutuon: As far as I can tell, the errors we've been facing seem to be stemming from the penult function in Module:grc-accent. It is supposed to find penultimate stress, or if nothing is found, call the circ function to assign an ultimate circumflex or acute. This is important because the penult function is called in the makeconj function of Module:grc-decl on the root in order to make an antepenultimately stressed word. For some reason, though, the penult function, which should be returning the stem γήιν- (gḗin-), is returning γηῖν- (gēîn-). Somehow it seems to be failing to find the second-to-last stem vowel. I'm not sure I can tell why at the moment, however. —JohnC5 04:26, 22 July 2017 (UTC)Reply

Ok, in the case of γήινος (gḗinos), the ήι (ḗi) was being tokenized together, so you have to use γήϊνος (gḗïnos) (which is the DGE's spelling).

Also, on further examination, I noticed the a line in makeconj, which is now commented out in Mod:grc-decl/sandbox: if not mw.ustring.match(ct, 'prx') then stem = m_accent.circ(root) end. The module works by taking the stem and assigning as ultimate stress (1), penultimate stress (2), and no stress (3) (again this a being described for the stem). The line of code that I commented out made it so that, if the declension pattern was not proparyoxytonic, the stem stress would be ultimate acute (1), ultimate circumflex (2), and none. I removed the line and got this correct result:

I don't know why this line of code was there, but we need a way to unit test this code. Before I continue messing around, we should get a list of all the different types of declensions we expect and put them in one place so that I can ensure that nothing is broken. @Angr, Barytonesis, Erutuon, could I entreat you to assemble a list of examples of all the declensions and tone patterns? This module needs to be massively refactored for it to be maintainable, but I'm scared to fool with anything without being able to see all the effects. —JohnC5 05:05, 22 July 2017 (UTC)Reply

I could start on that, but where should the list be: Appendix:Ancient Greek declension tables? — Eru·tuon 05:10, 22 July 2017 (UTC)Reply

@Erutuon: Thanks. Appendix:Ancient Greek declension tables could work, but honest, since we're going to fiddling with the sandbox module for a while, I was hoping you could whip up one of your fancy unit test modules to test some or all the expected forms. One of the main issues however is the fact that this code uses a ridiculous number of global variables, making it very hard to debug or extract data from it. My first task is going to try to localize all the variables. —JohnC5 05:18, 22 July 2017 (UTC)Reply

@JohnC5: Hmm, if we can put all the unique forms in a list like this, then the UnitTests module can do its work:
δῐ́κη δῐ́κης δῐ́κῃ δῐ́κην
δῐ́κᾱ δῐ́καιν
δῐ́και δῐκῶν δῐ́καις δῐ́κᾱς.
There would be a different layout for the list depending on the declension: for instance, neuter second-declension nouns would have fewer forms. Adjectives would have many more, though. Is there somewhere in the module where the forms are all found in a table, from where they can be extracted?

I'm likewise frustrated by the global variables. I wanted to clean them up, but never did it because it was so complicated. — Eru·tuon 05:40, 22 July 2017 (UTC)Reply

I've put it in an IDE and am slowly trying to disentangle the code. I'm also gonna make subpages eventually. —JohnC5 05:59, 22 July 2017 (UTC)Reply

@Erutuon: I've refactored the so that the only global variables are those listed at the top. Everything else is being passed around via the args parameter. This isn't the most elegant solution, but it's better than before. Please feel free to move things around/move them to separate pages/write unit tests/etc. I hope I haven't introduced too many bugs. :P —JohnC5 08:40, 22 July 2017 (UTC)Reply

N. B.: we already have Category:Ancient Greek declension appendices, but these are old, and potentially incomplete (I see Appendix:Ancient Greek third declension is missing the declension pattern of λεχώ (lekhṓ), for example). --Barytonesis (talk) 12:43, 22 July 2017 (UTC)Reply

There's still a problem with γήινος (gḗinos). Since the use of the trema ¨ prevents us from using the breve ({{subst:chars|grc|i^¨}} returns ῐ¨, but maybe there is no breve + trema combining character in Unicode?), the iota is treated as long; so feminine nominative plural is given as **γηῗναι instead of the correct γηΐναι.

Now I also notice that {{subst:chars|grc|i¨/}} returns ι΅ ({{subst:chars|grc|i/¨}} > ί¨) instead of ΐ. --Barytonesis (talk) 13:02, 22 July 2017 (UTC)Reply

@Barytonesis: To get the diaeresis, you need to use + rather than the spacing diaeresis ¨: {{subst:chars|m|grc|i/+}} → ΐ (ḯ). — Eru·tuon 17:12, 22 July 2017 (UTC)Reply

User:Erutuon/sandbox1 Appendix:Ancient Greek declension tables is where it's at. --Barytonesis (talk) 23:33, 22 July 2017 (UTC)Reply

@Angr, Barytonesis, Erutuon: I think that I've fixed most (all?) of the currently known problems with the module with the sandbox. You can test it out below. Please tell me if I've missed something. Erutuon, I'm working on making an entrance point for unit testing. —JohnC5 01:20, 23 July 2017 (UTC)Reply

Also, κεχηνώδης (kekhēnṓdēs) should have a properispomenon for the genitive, right? —JohnC5 01:48, 23 July 2017 (UTC)Reply

Which genitive? singular κεχηνώδους (kekhēnṓdous), dual κεχηνώδοιν (kekhēnṓdoin) (not **κεχηνῶδοιν, since -οι- is long here) and plural κεχηνωδῶν (kekhēnōdôn) are the right forms, if I remember well. --Barytonesis (talk) 02:03, 23 July 2017 (UTC)Reply

@Barytonesis: Sorry, I meant the neuter nominative singular. —JohnC5 02:06, 23 July 2017 (UTC)Reply

Ah, then yes; that's what I mentioned here, actually. But this only applies to adjectives suffixed in -ώδης (-ṓdēs) (ἐργῶδες (ergôdes)) and -ήρης (-ḗrēs) (ποδῆρες (podêres)), apparently. It's recessive in other cases: συνήθης (sunḗthēs) > σύνηθες (súnēthes). --Barytonesis (talk) 02:14, 23 July 2017 (UTC)Reply

Wait, the genitive plural might be κεχηνώδων instead of **κεχηνωδῶν (but I feel like this question has been resolved before, so take this with a grain of salt. @Angr?). --Barytonesis (talk) 02:23, 23 July 2017 (UTC)Reply

@Barytonesis: I think I've fixed them. Anything else? —JohnC5 02:53, 23 July 2017 (UTC)Reply

Can't think of anything right now. Thanks a bunch! --Barytonesis (talk) 03:11, 24 July 2017 (UTC)Reply

I copied the list of declensions from my sandbox page to Appendix:Ancient Greek declension tables, and switched to {{grc-decl/sandbox}}. — Eru·tuon 06:27, 23 July 2017 (UTC)Reply

Could someone research the rules about 3rd declension dental stems and the alternation of the accusative singular between -δᾰ and -ν (for example κλείς (kleís)). Anyone have any advice? —JohnC5 08:01, 23 July 2017 (UTC)Reply

Mmh... I don't have any, but I'll mention ἔρις (éris) (ἔριν~ἔριδα), ὄρνις (órnis) (ὄρνιν~ὄρνιθα) and κόρυς (kórus) (κόρυν~κόρυθα) which display a similar alternation. --Barytonesis (talk) 03:11, 24 July 2017 (UTC)Reply

@Barytonesis, FYI, we're gonna need to change {{grc-decl/sandbox}} and {{grc-adecl/sandbox}} back when we actually move the sandbox code into production. Also, please continue to add words as you find them to the appendix for both nouns and adjectives. Thanks for your help. —JohnC5 03:33, 24 July 2017 (UTC)Reply

Now I regret we have orphaned and deleted the old templates so fast; keeping them a bit longer would have made it much easier to complete the page with all the imaginable paradigms. Oh well. --Barytonesis (talk) 16:33, 25 July 2017 (UTC)Reply

I would have recommended simply deprecating them so that the page histories remain legible... — Eru·tuon 16:35, 25 July 2017 (UTC)Reply

@Angr, Erutuon, JohnC5 I've started gathering the adjectives: Appendix:Ancient Greek adjective declension tables. --Barytonesis (talk) 15:19, 30 July 2017 (UTC)Reply

• βαρύς (barús) is not handled properly.

@Angr, Erutuon, JohnC5: Should we categorise participles by voice and tense, or is it useless? --Barytonesis (talk) 12:04, 6 August 2017 (UTC)Reply

@Barytonesis: Not sure whether it's worth doing, but there are several places in the entry where it could be done: the headword, the definitions, or the declension table. It would be easiest in the definitions, assuming all participles use {{inflection of}}, because the voice and tense information is already there in the template. — Eru·tuon 17:13, 6 August 2017 (UTC)Reply

Yeah, the inflection table is not the right place to categorize it. —JohnC5 18:52, 6 August 2017 (UTC)Reply

While |comp=- and |super=- successfully suppress the comparative and superlative, the template won't allow us to specify irregular comparatives and superlatives. For example, at πολύς (polús), the template syntax says |comp=πλείων, πλέων|super=πλεῖστος, but the template is still generating the nonexistent "πολλότερος" and "πολλότᾰτος". —Mahāgaja · talk 23:07, 8 February 2019 (UTC)Reply

@Mahagaja: Fixed, I think! — Eru·tuon 23:57, 8 February 2019 (UTC)Reply

At ὅς (hós), the module seems to be getting the neuter dative plural from the neuter genitive plural, but it shouldn't. It should get it from the masculine dative plural. —Mahāgaja · talk 22:58, 12 February 2019 (UTC)Reply

@Mahagaja: Okay, fixed that, and filled in the unintentionally blank cells and made sure that the vocative is blank. — Eru·tuon 01:07, 13 February 2019 (UTC)Reply