Module talk:grc-decl/Archive 1

It seems based on this page (which looks fantastic, by the way) that words with parentheses and Epic datives like πολῐήτε͜ω are causing linking problems. Thought I'd mention. —JohnC5 06:23, 27 April 2015 (UTC)Reply

Yeah, I know, it was late and I was saving my work so I didn't lose it. I'll fix it soon. ObsequiousNewt (εἴρηκα|πεποίηκα) 15:06, 27 April 2015 (UTC)Reply

Cool beans. I am always here to complain about things you haven't implemented yet. ;D —JohnC5 21:45, 27 April 2015 (UTC)Reply

An odd problem at Κῦρος (Kûros): the inflection template deletes the initial Κ. I don't think I did anything wrong.... — Eru·tuon 08:50, 23 February 2016 (UTC)Reply

@Erutuon: Fixed. It was a bug in Module:grc-accent. —ObsequiousNewt (εἴρηκα|πεποίηκα) 15:35, 24 February 2016 (UTC)Reply

In many cases I wish I could turn off the dialects note ("Not all forms ... are attested"). It should be displayed in most cases, but in, for instance, the Attic table in ἡμεῖς (hēmeîs), it's obviously inaccurate. Probably the note gets more likely to be accurate the rarer the word is, and the more dialectal forms the table includes. So it should be a toggle that can be used with discretion. — Eru·tuon 09:08, 2 March 2016 (UTC)Reply

The operative word is 'necessarily', I guess, but I can add an option to remove it. —ObsequiousNewt (εἴρηκα|πεποίηκα) 16:01, 2 March 2016 (UTC)Reply

@Erutuon: I set it to remove the notice when the declension type is irregular. I'm a little hesitant to do it otherwise, but if you think it's best I will. —ObsequiousNewt (εἴρηκα|πεποίηκα) 18:57, 7 March 2016 (UTC)Reply

Sounds good. I'll consider if there are other cases where the note isn't needed. By the way, you might want to look at the tables in ἡμεῖς (hēmeîs). Looks like the "dialectal forms" header is still there and the syntax for table row isn't working. — Eru·tuon 19:07, 7 March 2016 (UTC)Reply

Fixed. Actually an old bug that would, if fixed, have obviated the need for this proposal (basically, if there were no dialectal forms listed, the row should have been omitted completely.) —ObsequiousNewt (εἴρηκα|πεποίηκα) 22:42, 7 March 2016 (UTC)Reply

Wondering if the suffixes -φι, -θι could be included in the table. They only occur for some nouns. -θι clearly has a locative meaning, while -φι is a crazy mix of instrumental, locative, ablative, genitive, dative. Not sure about their PIE etyma, but they might be originally from case endings.

For instance, ἠώς (ēṓs) has the locative ἠῶθι (ēôthi, “at dawn”), βία (bía) has the "instrumental" βίηφι (bíēphi, “by might”), ὄρος (óros) a locative ὄρεσφι (óresphi, “in the mountains”), ναῦς (naûs) an ablative ἀπὸ ναῦφιν (apò naûphin, “from the ships”).

These would have to be turned on for individual nouns, since they aren't exactly fully productive, and only for the Epic dialect. (Or maybe for others, but I haven't read Doric or Aeolic.) They don't have distinct singular and plural forms, it seems. Not sure what to call φι(ν) (phi(n)), since it would certainly need a label. — Eru·tuon 20:15, 9 March 2016 (UTC)Reply

I'm hesitant to put them in the tables proper, largely because I don't know how they should be displayed. It may be easier just to treat them as adverbs in most cases. —ObsequiousNewt (εἴρηκα|πεποίηκα) 14:50, 10 March 2016 (UTC)Reply

I concur with Newt on this one. —JohnC5 19:55, 10 March 2016 (UTC)Reply

Well, I don't think they're exactly adverbs. They're sometimes used with prepositions (or postpositions): ἠῶθι πρό (ēôthi pró, “before dawn”), ἀπὸ νευρῆφιν (apò neurêphin, “from the bowstring”).

I think you're right they shouldn't go in the table proper. They don't have distinct singular and plural forms, so they wouldn't fit in the table like the regular cases. They would make the most sense as notes below the table. Either they have to be put in the table or in Usage notes or Derived forms, and it's neater to have them in the table.

Also, we should include forms in -θεν (-then) (ablative) and -δε (-de) (maybe could be called allative). Those are very similar: I think -θεν (-then) is used with ἀπό (apó) sometimes.

Even if these are adverbs, adverbs from adjectives are displayed in tables, so it's odd to use that as justification for excluding them. We just have to decide what to call these forms, and add a parameter that lets us choose which of them to display in a given entry and probably parameters to specify the forms when they are irregular. Perhaps -θι locative, -φι oblique, -θεν ablative, -δε allative. — Eru·tuon 21:03, 10 March 2016 (UTC)Reply

They can be put in the table or directly below it (still in the Inflection header.) Although, frankly, since these are all Homeric forms (right?) we can easily put them in the dialectal forms section. —ObsequiousNewt (εἴρηκα|πεποίηκα) 14:03, 11 March 2016 (UTC)Reply

No, not all of them are only Homeric or Epic. -φι is only Homeric, and probably combining them with prepositions is, and certainly the other forms are more common in Homer, but several forms with -δε and θεν are used in later Greek. It's a confusing situation where endings are becoming less productive, but some forms with the endings are still used. Not sure what you mean by dialectal forms section; Alternative forms? — Eru·tuon 18:27, 11 March 2016 (UTC)Reply

Is there any record of a non-Epic dialect using cases in -θε(ν) or -θι in a non-adverbial sense? Buck doesn't mention it, but Buck doesn't mention any exotic case forms. (Not sure why. Perhaps they were only literary?)

Homer has Ἰλιόθι πρό, ἠῶθι πρό; ἐξ οὐρανόθεν, ἀπὸ Τροίηθεν; ἐκ ποντόφιν, ἀπὸ ναῦφιν. LSJ also lists σὺν ὄχεσφι in Ibycus (a Lyric poet). -σε is only used in six substantives at a maximum, and I can find no attestation with a preposition or postposition. -θι is attested only with postpositions (and at that only πρό), except possibly in A.R.2.355 with διὰ νειόθι.

-δε is most certainly not a case form, however; rather, it is an enclitic suffix. This is shown by ἀγορήνδε (not **ἀγορῆνδε), Αἴγυπτόνδε (not **Αἰγύπτονδε or **Αἰγυπτόνδε). (There are three forms οἴκαδε, ὀρθιάδε, and φύγαδε, which may have simply been extended from consonant stems. ὀρθιάδε is only attested once.) —ObsequiousNewt (εἴρηκα|πεποίηκα) 20:38, 16 March 2016 (UTC)Reply

Attic doesn't use either one with prepositions. Not sure about Doric or Aeolic.

Yes, accents for words with -δε are assigned according to the rules for enclitics. However, there is at least one example in which an adjective takes the exotic suffix to agree with an exotic-suffixed noun it modifies: ὅνδε δόμονδε (hónde dómonde, “to his house”). That seems very case-like. The only example I know; not sure how commonly adjectives modify nouns with these exotic suffixes, or if the adjectives ever take a different case-marking from the noun. I'll keep an eye out for it. — Eru·tuon 05:55, 18 March 2016 (UTC)Reply

δόμονδε is consistent with the rules of enclitics; paroxytones don't change their accent. —ObsequiousNewt (εἴρηκα|πεποίηκα) 13:12, 18 March 2016 (UTC)Reply

I know. I wasn't commenting on the accent, but on the fact that ὅν (hón, “his”) took the suffix -δε to agree with δόμονδε. That's un-enclitic-like behavior, since enclitics typically attach to a whole phrase rather than each constituent of the phrase. — Eru·tuon 18:54, 18 March 2016 (UTC)Reply

It's an anomaly, although it certainly deserves pointing out on the -δε page. I'm inclined however to believe that the -δε there is possibly just the pronominal -δε, which serves rather to emphasize a pronoun (ὅδε 'this very', etc.) —ObsequiousNewt (εἴρηκα|πεποίηκα) 16:01, 22 March 2016 (UTC)Reply

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘ Well, if you're suggesting ὅν is a demonstrative here, that's certainly not the case. The reading "to this very house" doesn't make sense in context. For instance, in one example of the phrase (Odyssey 1.83) Athena is talking to Zeus about Odysseus returning to his own house. If she meant "this very", she'd (I assume) be talking about Odysseus returning to Olympus. And the demonstrative would usually have the form τόν (though the relative can sometimes be used as a demonstrative). So it only makes sense for the word to be a possessive adjective, and possessives don't usually take the deictic suffix -δε (-de). — Eru·tuon 19:51, 22 March 2016 (UTC)Reply

ὅνδε is not a demonstrative, but it is a pronoun, and it may have accordingly taken emphatic -δε. However, in any case, it is an anomaly. There is no other instance of -δε acting specifically like a case, and all instances have it acting like an enclitic. —ObsequiousNewt (εἴρηκα|πεποίηκα) 18:28, 27 March 2016 (UTC)Reply

Okay, I misunderstood what you were saying. Look, both things are anomalies: an enclitic acting as a case suffix and a possessive adjective taking a deictic suffix. I am more inclined to go with the first anomaly, though, because nowhere else that I know of does a possessive adjective take the deictic suffix (no *ἐμόνδε, *σόνδε), and the introduction to my edition of books 1-12 of the Odyssey by Stanford analyzes ὅνδε as having the "allative" suffix. I'm also not sure what motivation there is for adding the demonstrative suffix to an possessive adjective, and what meaning would be intended: "for Odysseus to return to * home"? It's a little weird. Reflexives don't need extra deixis. They are already unambiguously referring to something. But I don't know if this is cross-linguistically plausible or not.

In any case, I agree -δε (-de) is an enclitic generally; ὅνδε δόμονδε is, if anything, an exception. So, do you have an opinion either way on adding these forms to tables? Are you bringing up the point about enclisis because you would argue forms with added enclitics should not be displayed in inflection tables? — Eru·tuon 20:10, 27 March 2016 (UTC)Reply

I can support adding -φι(ν), -θι, and -θε(ν) to inflection tables—although -φι(ν) does not have a consistent identity, -θι is rare, and all are apparently Epic only. -σε does not show any instance of clearly case-like behaviour, and so is best simply treated as an adverbial suffix. -δε shows one probable instance of agreement, and one possible instance of prepositional usage, but also acts syntactically like an enclitic used with the accusative rather than a proper case, and so I am hesitant to include it in inflection tables. —ObsequiousNewt (εἴρηκα|πεποίηκα) 20:41, 27 March 2016 (UTC)Reply

Looks like nouns in -τύς (-tús) currently aren't covered. — Eru·tuon 00:37, 15 March 2016 (UTC)Reply

Don't the oblique cases have a short vowel? —ObsequiousNewt (εἴρηκα|πεποίηκα) 15:23, 15 March 2016 (UTC)Reply

Hm, yes, it appears so. So that was the problem. — Eru·tuon 19:57, 15 March 2016 (UTC)Reply

I just spent a couple days rewriting the module to make it cleaner (and in the process cut 20 kilobytes.) I believe I've tested everything for regression/bugs, but I'd just like to ask that everyone be wary and watch for any bugs or inaccuracies. Thanks. —ObsequiousNewt (εἴρηκα|πεποίηκα) 20:01, 17 April 2016 (UTC)Reply

Cool! Well, one problem so far: βασιλεύς (basileús) somehow has ᾰσ added to the stem, though ἱερεύς (hiereús) does not, perhaps because of the ᾰσ within the stem. — Eru·tuon 20:31, 17 April 2016 (UTC)Reply

There was an error mentioned here that I think does not pertain to these changes but still persists. —JohnC5 20:50, 17 April 2016 (UTC)Reply

Then there's the striking coincidence that 106 of the 113 entries in CAT:E are in Greek script... Chuck Entz (talk) 21:43, 17 April 2016 (UTC)Reply

@Chuck Entz: Well, considering how many entries use the declension module, it's surprising there aren't more. And quite a few errors relate to -ος (-os) neuter nouns and -ής (-ḗs) adjectives, so probably at least half can be wiped out at once. — Eru·tuon 00:42, 18 April 2016 (UTC)Reply

Oh dear, I did not anticipate this many problems. Well, let's see what I can fix... —ObsequiousNewt (εἴρηκα|πεποίηκα) 01:08, 18 April 2016 (UTC)Reply

Looks like -ος neuters are fine now, but -ής (-ḗs) adjectives show incorrect comparative and superlative forms (αλήθτερος, αλήθτᾰτος; αληθώτερος, αληθώτᾰτος), and Περικλῆς (Periklês) seems to be displaying adjective forms rather than the kles forms given in the module. And Σωκράτης is having problems. — Eru·tuon 03:12, 18 April 2016 (UTC)Reply

Update: 98 module errors down, 8 to go. Thanks! Chuck Entz (talk) 14:02, 18 April 2016 (UTC)Reply

The display where there are multiple forms, such as in the dialectal form of κεχηνώς (kekhēnṓs), needs spaces for legibility. —JohnC5 14:38, 18 April 2016 (UTC)Reply

Through our combined efforts, we seem to have cut it down to one page, δῖος. I'm a little hesitant to fix this error just yet because I want to ask first—is there any other first-and-second declension adjective that has a short feminine? —ObsequiousNewt (εἴρηκα|πεποίηκα) 21:42, 18 April 2016 (UTC)Reply

Not that I know of. So, maybe best to just give the irregular forms using |FNS= and |FAS=. Actually, I tried to do that, but not in the correct way. I'll fix it. — Eru·tuon 23:24, 18 April 2016 (UTC)Reply

Okay, ἁπλόος and the legibility problem are both fixed. ἐΰς is broken, though, and there may still be others. —ObsequiousNewt (εἴρηκα|πεποίηκα) 13:42, 19 April 2016 (UTC)Reply

After your fix, there are 19 Ancient Greek entries at CAT:E. Chuck Entz (talk) 02:02, 21 April 2016 (UTC)Reply

Seems to be a problem affecting irregular paradigms. — Eru·tuon 04:11, 21 April 2016 (UTC)Reply

Fixed, hopefully for good. Now we need to find the more insidious bugs... —ObsequiousNewt (εἴρηκα|πεποίηκα) 23:18, 21 April 2016 (UTC)Reply

Hey, I wonder if the module could be modified so that a table can be tagged with multiple dialect names. That would be appropriate in ἠώς (ēṓs), where the forms are the same in both Ionic and Epic. (Then again, maybe I'm wrong about that.) I tried |dial=epi/ion and |dial=epi, ion and neither worked. — Eru·tuon 21:40, 23 September 2016 (UTC)Reply

@Erutuon: In general, they're not. They're similar, and Epic is largely derived from Ionic, but Epic borrows features from several dialects, as well as preserving things (e.g. -οιο) not preserved by any dialect. — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 01:19, 24 September 2016 (UTC)Reply

@ObsequiousNewt: I'm aware that Epic borrows from several dialects, but that does not mean that forms are different for all words. I think ἠώς (ēṓs) may have identical, or mostly identical, forms in Epic and Ionic. — Eru·tuon 01:46, 24 September 2016 (UTC)Reply

@Erutuon: Assuming that is correct, it is easy to specify a custom header, as I have now done for ἡώς. — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 01:50, 24 September 2016 (UTC)Reply

For the first-person plural pronoun tables at ἡμεῖς (hēmeîs), the vocative, which automatically displays because vocative plurals of nouns and adjectives are the same as the nominative, should be turned off. Vocative is equivalent to second person, so it is mutually exclusive with first person. Could an option be added for that? (The vocative cell should probably just have a dash, like it does at ἐγώ.) — Eru·tuon 03:05, 9 October 2016 (UTC)Reply

You have to set it explicitly—the 'irregular' declension is set to take in order all of the (potentially) unique parameters, and vocative plural is normally never unique. — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 15:22, 9 October 2016 (UTC)Reply

@ObsequiousNewt: the module is getting the accentuation of feminine plural adjectives wrong. For example, at ἄξιος it's giving nom/voc.fem.pl ἀξίαι and gen.fem.pl ἀξιῶν. Those would be right if this were a noun, but in an adjective it should be ἄξιαι and ἀξίων (see Smyth §287a). —Aɴɢʀ (talk) 18:44, 29 October 2016 (UTC)Reply

@Angr Fixed. Thanks for pointing that out, that was quite an error. — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 01:36, 30 October 2016 (UTC)Reply

@ObsequiousNewt: The nominative neuter plural of the article in the Doric declension needs to be τᾰ̀, like the accusative plural, not τοὶ (toì), which is masculine. See the Doric inflection of μνημεῖον, for example. —Aɴɢʀ (talk) 22:10, 31 October 2016 (UTC)Reply

@Angr Fixed. — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 19:38, 5 November 2016 (UTC)Reply

I think I must have broken something while adding error messages, because now a fair number of Greek nominals appear in Cat:E. ObsequiousNewt informed me on Talk:τετράπους that irregular adjective declensions require numbered parameters (even if they are empty), so I added code that checks for undefined parameters and outputs a more informative error message than "gsub received a bad argument #1". But apparently some tables didn't have error messages before, because they got around the requirement of numbered parameters somehow, perhaps by using named parameters.

Anyway, I think it would be much more convenient if the module accepted templates in which not all numbered parameters are entered. But I haven't figured out how the module works enough to do that. @ObsequiousNewt, JohnC5, Angr, I'm so meta even this acronym, what do you all think? — Eru·tuon 04:58, 3 February 2017 (UTC)Reply

Sorry, this is all over my head. —Aɴɢʀ (talk) 07:20, 3 February 2017 (UTC)Reply

I should apologize—what you wanted was actually being prevented by a bug, which I think is fixed now. — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 00:08, 6 February 2017 (UTC)Reply

@ObsequiousNewt: Thank you. I just said you didn't listen to my suggestions on the other talk page, but then I see you did, so I'm sorry too. — Eru·tuon 00:14, 6 February 2017 (UTC)Reply

@ObsequiousNewt, Erutuon Hey, on pages like Θέτις (Thétis) and others, the |dial= parameter is not working correctly anymore in terms of changing the forms or displaying the dialect name. What's going on here? —JohnC5 18:33, 29 March 2017 (UTC)Reply

@JohnC5: I've pinged @ObsequiousNewt about this on some of the pages. I think the forms are fine; it's the title text that's wrong. — Eru·tuon 20:48, 29 March 2017 (UTC)Reply

Oops! It was because of a change I made in Module:grc:Dialects. The correct dialect labels should now be showing up again. — Eru·tuon 20:51, 29 March 2017 (UTC)Reply

@Erutuon: Thanks! —JohnC5 21:45, 29 March 2017 (UTC)Reply

Anyone have any idea why there's a module error at Μενέξενоς? I don't see anything wrong. —Aɴɢʀ (talk) 17:02, 15 April 2017 (UTC)Reply

@Angr: There is a Cyrillic character in the page title. The omicron is not an omicron. --Barytonesis (talk) 17:06, 15 April 2017 (UTC)Reply

@Barytonesis:: OK, thanks for fixing it. It occurred to me that the omicron might have been a Latin o; I didn't think of Cyrillic. —Aɴɢʀ (talk) 17:16, 15 April 2017 (UTC)Reply

That was tricky. I copy-pasted the character in the search box, that's how I found out; otherwise I don't think I would have thought about Cyrillic either. --Barytonesis (talk) 17:40, 15 April 2017 (UTC)Reply

I can't get names in -κλῆς (Special:WhatLinksHere/Template:grc-decl-3rd-εης-con) to work under {{grc-decl}}. Any ideas? —Aɴɢʀ (talk) 18:42, 15 April 2017 (UTC)Reply

@Erutuon, and idea on this? I also need help adding the alternative accusative singular -ιν (-in), -υν (-un) as seen in {{grc-decl-3rd-dn-prx}} or on ἁψίς (hapsís). —JohnC5 04:46, 17 April 2017 (UTC)Reply

@JohnC5, Angr: I'm not sure how to fix this. There is supposed to be a kles declension in the module, but for some reason it is not being used when I try to create a table for Περικλῆς (Periklês). It's using the es adjectival declension instead. I wish, again, that the module were easier to understand. — Eru·tuon 18:06, 17 April 2017 (UTC)Reply

I could use a good refactoring, but I'm not sure I'll be up to the challenge any time soon. I wish NEwt would come back. —JohnC5 18:10, 17 April 2017 (UTC)Reply

Yeah, him and Atelaes. —Aɴɢʀ (talk) 07:37, 18 April 2017 (UTC)Reply

@Erutuon, JohnC5, ObsequiousNewt (on the off chance that he comes back soon): There's an error in u-stem adjectives like βραχύς (brakhús): the genitive singular masculine and neuter should end in -εος, not -εως (the way nouns do); see Smyth § 296–97. —Aɴɢʀ (talk) 08:01, 29 April 2017 (UTC)Reply

There's a second error: the dative singular of the neuter is given as βρᾰχεῖ / βρᾰ́χέῐ̈; the first one is all right, but the second one has two accent marks. It should be βρᾰχέῐ̈, just like the masculine. —Aɴɢʀ (talk) 09:05, 2 May 2017 (UTC)Reply

Just spotted this too:
3rd decl. adj. GEN sg in -ύς -εῖα -ύ (like πλατύς) and -υς -υς (like δίπηχυς) are -έος and -εος respectively, with an omicron, not an omega.
Alsο I don't presently recall ever seeing uncontracted NOM pl in these groups (in general or in particular).
I was about to change the above (basically the former) but then I bumped onto the dialforms_weak_u(_prx) functions whose function I don't understand and since I have never edited a module I decided not to edit but rather to comment here.
P.S. Also on e.g. βαθύς, LSJ has an uncontracted DAT sg but my faithful grammar book from highschool has contracted ones in both particular and general, so I guess having them both...Thanatos|talk|contributions 01:36, 20 May 2017 (UTC)Reply

Also, there is no equivalent to {{grc-decl-3rd-ευς-con}} in this module. We're gonna have to fix this at some point. —JohnC5 15:16, 3 June 2017 (UTC)Reply

When I use |form=sing-plur to specify that a form is not attested in the dual (e.g. Koine forms), the title line of the table shows the plural instead of the singular. See for example ἁμαρτία#Inflection, where the title line says "First declension of αἱ ἁμαρτῖαι; τῶν ἁμαρτιῶν (Attic)" instead of "First declension of ἡ ἁμαρτίᾱ; τῆς ἁμαρτίᾱς (Attic)" (as it would if I didn't suppress the dual). —Aɴɢʀ (talk) 12:50, 21 June 2017 (UTC)Reply

@Angr: Fixed. — Eru·tuon 17:05, 21 June 2017 (UTC)Reply

Both {{grc-decl}} and {{grc-adecl}} are accentuating nominative/accusative plural feminine forms incorrectly when under recessive accent. At κύριος (kúrios) and κυρία (kuría), it's generating κυρίαι instead of κύριαι, and at κεφάλαιος (kephálaios) it's generating κεφαλαῖαι instead of κεφάλαιαι. —Aɴɢʀ (talk) 09:34, 21 July 2017 (UTC)Reply

@Angr: For the noun κῡρῐ́ᾱ (kūríā) at least, the right form (in Attic) is κυρίαι, not **κύριαι. Not because -ai is long, but because other cases retain the place of the accent in the nominative singular. Please see what I wrote here. Now, for the adjectives, I'm wondering: is it the place of accent in the nominative masculine singular that matters for all three genders, or are the feminine forms based on the nominative feminine singular? --Barytonesis (talk) 12:58, 3 August 2017 (UTC)Reply

@Barytonesis: Smyth §287a: "The accent in the feminine nominative and genitive plural follows that of the masculine: ἄξιαι (áxiai), ἀξίων (axíōn), not ἀξίαι (axíai), ἀξιῶν (axiôn), as would be expected according to the rule for substantives (205), e.g. as in αἰτίᾱ (aitíā, “cause”), αἰτίαι (aitíai), αἰτιῶν (aitiôn)." I don't whether substantivized adjectives like κῡρίᾱ (kūríā) behave like nouns or like the adjectives they're derived from, though. —Aɴɢʀ (talk) 13:15, 3 August 2017 (UTC)Reply

@Angr: Well, I think the second part of what you wrote answers that question. αἴτιαι (adjective) but αἰτίαι (noun); ἄξιαι (adjective) but ἀξίαι (noun); κύριαι (adjective) but κυρίαι (noun), φίλιαι (adjective) but φιλίαι (noun), etc. (without taking other dialects into account) --Barytonesis (talk) 13:26, 3 August 2017 (UTC)Reply

In that case, Module:grc-decl/sandbox needs to be corrected (@JohnC5, Erutuon) because at the moment it's generating φίλιαι for the noun. —Aɴɢʀ (talk) 15:03, 3 August 2017 (UTC)Reply

@Barytonesis, Angr, JohnC5: Okay, that issue is fixed. The modules can handle the difference in accent between adjectival ἄξιαι (áxiai), substantival ἀξίαι (axíai), and adjectival ὀλίγαι (olígai). — Eru·tuon 18:50, 4 August 2017 (UTC)Reply

(It seems a little weird to discuss the module remodeling at Template talk:grc-adecl.) @JohnC5: The "contracted, uncontracted" tables of es-stems bother me. It leaves it up to the reader to guess which dialects might use these forms. It would be better if we could connect the contracted or uncontracted forms to specific dialects, and probably omit the uncontracted forms by default, because we default to displaying the Attic–Koine forms in other declensions, and we choose to place the main entry under the Attic or Koine form. For instance, I would rather we just display the table containing the genitive singular -ους (-ous) by default, and only add -εος (-eos) if someone puts in |dial=ion or |dial=epi. — Eru·tuon 05:38, 24 July 2017 (UTC)Reply

Another thing, which might be more controversial: I wonder if it would be good to automatically add a dialect label to all declension tables, if there are significant dialectal differences in that particular declension type. (This in combination with adding term-labels to all entries.) So, if you don't add a dialect, the table will automatically say "(Attic)", because Attic forms display by default. Without that, it's misleading: if the word with that particular nominative singular or masculine nominative singular form is used in other dialects than Attic, the forms are likely to be wrong.

I would also argue for always adding term-labels and tables for every dialect in which a word is used. Even when a word is only used in Attic or Koine, it is good to be explicit about that. — Eru·tuon 05:44, 24 July 2017 (UTC)Reply

@Angr, Barytonesis, what do y'all think of this? I can implement these changes, but I need some agreement. I personally prefer having uncontracted and contracted, but that's because I like seeing the morphological etymology—not for any good reason. —JohnC5 05:49, 24 July 2017 (UTC)Reply

Hmm, then we should mention both the dialect and the phonological description, linking to the Appendix: (Attic, contracted). It is useful information.

To add to my point, it is rather odd to display the uncontracted table for words like ἐργώδης (ergṓdēs), which judging by the author names in the LSJ entry, isn't used in Ionic or Epic. — Eru·tuon 06:12, 24 July 2017 (UTC)Reply

I'm open to whatever—just a statement of preference. —JohnC5 06:18, 24 July 2017 (UTC)Reply

Are there any differences between Attic and Koine declension? If so, I'd rather have it show Koine when the dialect is unspecified, since we usually use Koine forms as our main entries (we list γλῶττα (glôtta) as a variant form of γλῶσσα (glôssa), not vice versa). If not, I'd rather have it automatically display "(Attic/Koine)" when no dialect is specified. Otherwise I think it's a great idea. —Aɴɢʀ (talk) 09:34, 24 July 2017 (UTC)Reply

@Angr: At the moment, the module does not contain any mention of the code koi (viz. It treats Attic and Koine indentically). —JohnC5 18:16, 24 July 2017 (UTC)Reply

This page gives a quick overview of the differences between Attic and Koine. It doesn't look like there's anything worth mentioning in the declension of nouns and adjectives, except the lack of a dual in Koine. The verb module, however, may want to add -αν and -οσαν as third-person plural aorist 2 endings in light of Koine forms like ἤλθοσαν (ḗlthosan) and ἦλθαν (êlthan) for classical Attic ἦλθον (êlthon). —Aɴɢʀ (talk) 19:20, 24 July 2017 (UTC)Reply

So, what's the plan for the moment? —JohnC5 01:57, 25 July 2017 (UTC)Reply

I'm in favor of marking as Attic everything that isn't explicitly marked as something else. And I'm in favor of removing dual forms from any table marked as Koine. —Aɴɢʀ (talk) 08:08, 25 July 2017 (UTC)Reply

I would prefer having to mention the dialect as well. And I agree with Angr about duals. --Barytonesis (talk) 17:14, 25 July 2017 (UTC)Reply

Is contracted ἱππεύς, ἱππῶς (hippeús, hippôs) (currently output by {{grc-decl/sandbox}} for Attic) actually a thing? I hadn't heard of that. I thought the Attic was uncontracted ἱππεύς, ἱππέως (hippeús, hippéōs). — Eru·tuon 16:38, 25 July 2017 (UTC)Reply

It should be ἱππέως (hippéōs), yes. --Barytonesis (talk) 16:43, 25 July 2017 (UTC)Reply

Huh, I wonder if the contracted forms exist at all or if they're just a mistake... — Eru·tuon 17:57, 25 July 2017 (UTC)Reply

@Barytonesis, Erutuon: They exist but only when εύς is preceded by a vowel, and even then they aren't common; see Smyth §276. —Aɴɢʀ (talk) 18:33, 25 July 2017 (UTC)Reply

I fixed the current problem. I don't know, however, whether the contracted form exists. —JohnC5 18:01, 25 July 2017 (UTC)Reply

Question: should we make providing the whole genitive form (not just the ending) compulsory in all cases? I'm having a hard time remembering in which cases the ending is enough, and in which cases it's not. At least I think we should make it clear somewhere, if it isn't already. --Barytonesis (talk) 18:16, 25 July 2017 (UTC)Reply

Another thing I'd love to have is categories for the different variants of each declension (first, second, third), for accent types under Ancient Greek nouns by inflection type, or for dialectal considerations (nouns with long alpha in Attic but not Ionic). The categories would probably intersect: while I like to divide feminine first-declension nouns by short alpha vs. eta or long alpha, others use different divisions. I'm also not sure how to name some of the declension variants for the third declension, particularly those for nouns with stems ending in historical s, i/y, and u/w: Σωκρᾰ́της (Sōkrátēs), πόλῐς (pólis), βᾰσῐλεύς (basileús).

But anyway, at least some of these categories could best be added by Module:grc-decl. It would also be nice to mention something about declension subtypes in the header for the table. — Eru·tuon 18:20, 25 July 2017 (UTC)Reply

@Erutuon: I agree. I love categories. If we come up with a full list of categories and naming conventions, I'll gladly add them. Particularly, if we have declension, declension modifier, accentation, and dialect names, we can assemble whatever categories we want (e.g. "Ancient Greek third declension s-stem nouns with properispomenon"). —JohnC5 18:34, 25 July 2017 (UTC)Reply

@JohnC5: I'm doing brainstorming at User:Erutuon/Ancient Greek categories. You and other Ancient Greek editors are welcome to add your ideas there or modify things. — Eru·tuon 19:19, 25 July 2017 (UTC)Reply

I like that idea. --Barytonesis (talk) 21:35, 25 July 2017 (UTC)Reply

Anyone have any idea how to deal with νόος (nóos), νόος (nóos)? Normally, when the nominative and genitive are the same, we say indeclinable. Should there be a declension override or should it try to find a declension instead, and then have an indeclinable override (i.e. should it assume indeclinable until told otherwise or vice-versa)? —JohnC5 03:48, 26 July 2017 (UTC)Reply

Displaying a declension table for indeclinable nouns seems pointless. It wouldn't contain any real information. So maybe the module should just find a declension or return an error. Are there other declined nouns with nominative and genitive identical? — Eru·tuon 04:22, 26 July 2017 (UTC)Reply

At present, I am only familiar with this one. Also, I think I've placed all the accent-declension combinations I can find in Appendix:Ancient_Greek_declension_tables. Did I miss any major ones? Can some one start hte adjectives? —JohnC5 06:03, 26 July 2017 (UTC)Reply

Unfortunately Ima have to slow down for a few weeks, so I prefer not to get down to it now. But this is quite addictive so we'll see. --Barytonesis (talk) 02:00, 27 July 2017 (UTC)Reply

@JohnC5: I've given in and started it anyway: Appendix:Ancient Greek adjective declension tables. --Barytonesis (talk) 15:17, 30 July 2017 (UTC)Reply

@Angr, Barytonesis, Erutuon: I've added Attic as the default. If we finish the category list, I can add them. Is there anything else to be done? —JohnC5 03:51, 27 July 2017 (UTC)Reply

I'm working on the category list. A little unsure how to handle the third declension. We could easily mix a bunch of different category structures, though.

One thing I would request is that you add the "contracted" blurb linking to Appendix:Ancient Greek contraction for those declensions that involve contraction. That would include the sigmatic nouns and the rare nouns in -εύς (-eús) with some contracted forms, which Barytonesis mentioned above. Maybe it should link to the Attic section of the appendix, when the dialect of the table is Attic.

Or how about adding more sound changes? I wonder if we could also have a note regarding compensatory lengthening (probably below the table) for nouns with a stem ending in a short vowel and ντ or ν but a nominative singular with a long vowel and sigma: ὀδούς (odoús), ἐλέφᾱς (eléphās), μείς (meís). Or in plosive-stems a note about final stops being neutralized to ks, ps, s in the nominative singular and dative plural. Or a note for nouns that have different vowel-grades of their stem. — Eru·tuon 18:13, 27 July 2017 (UTC)Reply

Regarding -εύς (-eús) nouns: I think ἁλιεύς (halieús) in Attic has both uncontracted and contracted genitive and accusative singular and plural forms, while ἱππεύς (hippeús) does not. Can the module distinguish between those by whether the -εύς (-eús) is preceded by a vowel? Smyth mentions other examples: Εὐβοεύς (Euboeús), Παιραιεύς (Pairaieús), Πλαταιεύς (Plataieús), maybe Εὐβοιεύς (Euboieús). He mentions something about contracted forms being less common after about 350 BC, but isn't very clear on just how common they were in each time period.

Also, the ending έῐ̈ / εῖ (éï / eî) should probably just be εῖ (eî) for all Attic nouns. Pronouncing that ending with two short syllables is a Homeric feature, I think. — Eru·tuon 18:32, 27 July 2017 (UTC)Reply

@Erutuon: There's a lot of stuff for which you just asked. Could you make a list of features to add for my reference (to which others may add as necessary), and also provide examples of these notes you'd likes to see? As for the -εύς (-eús) forms after vowels, I could implement that, but I wonder whether we are getting too specific on that point. There's already a note mentioning that not all forms may be attested, so if a few extra unlikely contracted forms slip in, I'm not super worried about it. —JohnC5 04:39, 28 July 2017 (UTC)Reply

@JohnC5: Okay, I'll make a list and flesh out my idea of notes. As to contraction in -εύς (-eús) nouns, given that Smyth says it only happens in a few nouns of that category, I'd rather remove the contraction altogether than indicate that it can happen with all -εύς (-eús) nouns. — Eru·tuon 04:56, 28 July 2017 (UTC)Reply

@Erutuon: Please note the list I've started. :) —JohnC5 05:18, 28 July 2017 (UTC)Reply

@JohnC5: Right, I saw that right before I saved my comment (edit conflict). — Eru·tuon 05:39, 28 July 2017 (UTC)Reply

Nice! You anticipated something I was going to say about required parameters. Very pleased with the new error messages too. — Eru·tuon 21:51, 29 July 2017 (UTC)Reply

@Barytonesis, concerning ὗς (hûs), this will have to be hard-coded. Because of the complexity to join a rough breathing mark with the desinence, it's just too much effort. I plan to make a set of irregular nouns that are hard-coded, but that will be a bit of time from now. —JohnC5 19:23, 30 July 2017 (UTC)Reply

@JohnC5: For ὗς (hûs), no worries. But what do you mean by "make a set of irregular nouns that are hard-coded"? --Barytonesis (talk) 23:02, 31 July 2017 (UTC)Reply

• Content
  • Better handling of title blurbs so that "(Attic, contracted)" and the like would be possible.
  • Show declension categories with links to appendices in title of tables: Declension of ᾰ̓γᾰθός; ᾰ̓γᾰθή; ᾰ̓γᾰθόν (first and second, Attic).
  • Add footnotes to forms exhibiting particular sound changes that explain the irregularities of a particular declension. (Please provide the information you would like to see)
    • for 3rd-decl. with nom. sg. in -ς with a long vowel, and stem ending in a sequence of the corresponding short vowel followed by ν, ντ:
      • The long vowel x in the nominative singular arose by compensatory lengthening of the short vowel y in the stem z after the loss of *n in the cluster *ns, *nts (pick the one that applies).
    • for 3rd-decl. with nom. sg. in -ς, -ξ, -ψ, dative plural in -σῐ, -ξῐ, -ψῐ and stem in τ, δ, θ; κ, γ, χ; π, β, φ, with no vowel change:
      • The Proto-Hellenic cluster *ts, *ds, *tʰs (pick the one that applies) simplified to -ς, -σ- (-s, -s-) in Ancient Greek.
      • Forms containing -ξ, -ψ (-x, -ps) arise from a respelling (in the case of κ, π or χ, φ) or neutralization of κ, γ, χ· π, β, φ (k, g, kh; p, b, ph) (pick the one that applies) when followed by a -ς, -σ- (-s, -s-).
    • 3rd-decl. with vowel preceding -ευς (-eus):
      • The first form is uncontracted, the second contracted (regarding the gen. and acc. sg. and pl).
  • Transliteration (compare {{ar-decl-noun}}, with transliteration in gray below each form)
  • Addition of categories (see User:Erutuon/Ancient Greek categories)
• Editor interface
  • Shorter or simpler |form= parameter: MS or ms rather than M-sing; MP or mp rather than M-plur (For the sake of backwards compatibility, I have merely aliased “S” to “sing”, “D” to “dual”, “P” to “plur”, and “F” to “full”. I'm not gonna do the lowercase aliases because they could cause a lot of trouble.)
  • A way to disable the dual (or is there already one that isn't documented?): perhaps |form=sp (“SP” should work now. “sing-plur” worked in the past and will continue).
  • Module errors when first or second parameter is missing, ideally explaining what each parameter is (at least for nouns, where second parameter is genitive; for adjectives, second parameter is either feminine or neuter)
  • Hidden category for second-declension nouns without gender specified: in the great majority of cases it's masculine, but some could be feminine.
• Module cleanup
  • An entrance point for unit testing
  • A simpler table markup creation system (Not necessary at the moment)
  • Cleaned up entrance points with less declension-specific logic hard coded
  • Add classes and move CSS to a stylesheet (disadvantage: won't display for mobile users if it's in MediaWiki:Common.css) (I don't know how to do this, —John)

• Addition of categories (see User:Erutuon/Ancient Greek categories)
  • @Erutuon: For the accent placement one, do we want the functional accent placement, or the placement on the nominative? —JohnC5 03:39, 30 July 2017 (UTC)Reply
  • Actually, could you write a function in mod:grc-accent that detects the placement of the accent? —JohnC5 03:54, 30 July 2017 (UTC)Reply
    • That's a good idea. I'll give it a try. — Eru·tuon 05:27, 30 July 2017 (UTC)Reply
      • @Erutuon: Thanks! I'm looking to for it to either return a number encoding the accent or (better) the words “oxytone”, “paroxytone”, and so on. Please have it return a nil instead of an error. Also, I'm guessing that we want to naïvely assume that an acute on the penult is a paroxytone, even if the word is only two syllables and could have a proparoxytone? —JohnC5 18:00, 30 July 2017 (UTC)Reply
        • @JohnC5: Done! (See here for examples.) I'm surprised how quick that was. Well, the function currently uses that surface assumption. It doesn't figure out where the accent could be if the word were longer. — Eru·tuon 18:25, 30 July 2017 (UTC)Reply
          • @Erutuon: Thank you so much. I think the surface assumption will be best. I'm gonna set up a unit testing framework in a bit (pace @Benwing2, I have a different system in mind). I would appreciate your and @Barytonesis's help filling it with the contents of these appendixes once it is done. —JohnC5 19:18, 30 July 2017 (UTC)Reply
            • @JohnC5 Interested to see what you have in mind. I admit my system is not a proper unit test but rather a practical way of checking for regressions. It has worked well for me but YMMV. Benwing2 (talk) 19:22, 30 July 2017 (UTC)Reply
              • @Benwing2: I was thinking something like this. @Barytonesis, Erutuon, does this setup make sense? —JohnC5 21:08, 30 July 2017 (UTC)Reply
                • @JohnC5 Of course; that's a proper unit test. In Module:ru-pron we actually have both sorts of tests: Module:ru-pron/testcases and test_new_ru_pron_module in the module code. Both of them are useful, for testing different things. Benwing2 (talk) 21:11, 30 July 2017 (UTC)Reply
                • (edit conflict) Yep, that's pretty much the setup I was thinking of, except that I wanted to eliminate syncretic forms in the same number (for instance, not display the vocative singular of δῐ́κη, or display three forms each in the singular and plural of neuter second-declension nouns). — Eru·tuon 21:15, 30 July 2017 (UTC)Reply
                  • @Erutuon: I only kept the VS in because it can vary in some declensions. —JohnC5 21:27, 30 July 2017 (UTC)Reply
                    • Well, I guess it's a different way of conveying the same information. — Eru·tuon 21:30, 30 July 2017 (UTC)Reply
                      • @Erutuon, Barytonesis, Angr: Well, I've added everything to the test cases except the 3rd declension vowel stems. Could I ask y'all to look it over and determine if they are all correct? Thanks for all y'all's help. —JohnC5 07:23, 31 July 2017 (UTC)Reply
                        • @JohnC5: I don't see anything obviously wrong, though subtler errors might have gone unnoticed, among so many forms. Also, I've thought of a few other cases which I've added to the appendices. --Barytonesis (talk) 13:18, 31 July 2017 (UTC)Reply
                        • It's not really important, but should we leave οἶνοψ (oînops), which is an adjective, in that list? --Barytonesis (talk) 10:52, 31 July 2017 (UTC)Reply

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘ οἶνοψ (oînops) was using {{grc-decl}} instead of {{grc-adecl}}. I've fixed it now. —Aɴɢʀ (talk) 11:54, 31 July 2017 (UTC)Reply

Just a comment ... I have implemented many of these features already in the Russian module Module:ru-noun. Perhaps you can get some ideas there about how to implement these features, esp. testing. The way I test changes involves the setting test_new_ru_noun_module in the module; see the code that implements this (starting at line 1908). The idea is that if you turn this on, it compares the declension of each noun to the corresponding declension generated by a user-space version of the module, and adds each noun that differs to a tracking category. I can use this to see the effect of a change before committing it. To see exactly what differs, I uncomment line 1938 and then use the "Preview page with this template" feature to view a page that shows up in the tracking category. This throws an error where the contents of the error indicate exactly which parts of the declension differ. Benwing2 (talk) 04:54, 30 July 2017 (UTC)Reply

@JohnC5: I can try refining the notes on sound or spelling changes, as I don't mind mucking around in the complicated logic. — Eru·tuon 20:20, 7 August 2017 (UTC)Reply

@Erutuon: Please do. Tell me when it's time for adding categories. —JohnC5 22:11, 7 August 2017 (UTC)Reply

@Erutuon: Sorry I've taken a hiatus on this. Do you want to finish up the notes and categories? Then, once we figure out how to fix the error in Appendix:Ancient_Greek_declension_tables/third/vowel, we should be good to go. —JohnC5 23:23, 23 August 2017 (UTC)Reply

@JohnC5: I was planning to do the notes at least, but I didn't because I couldn't figure out a good way to word them. I hope to get back to it sometime. — Eru·tuon 21:00, 26 August 2017 (UTC)Reply

@Erutuon: Any idea on the error at Appendix:Ancient_Greek_declension_tables/third/vowel? —JohnC5 23:53, 26 August 2017 (UTC)Reply

@JohnC5: No idea from just looking at it, but I can do some testing and see if I can find the problem. — Eru·tuon 22:17, 28 August 2017 (UTC)Reply

Oh, it's probably somehow because the length of the iota isn't marked. That's annoying. — Eru·tuon 23:07, 28 August 2017 (UTC)Reply

We need a function that adds macrons and breves based on the accent: in this case, changing ἴχνος to ῐ̓́χνος. It's quite bizarre having Module:grc-accent assume that an unmarked ambiguous vowel is long when the accent makes it clear that it can't be. The function can add macrons or breves when the two first arguments are processed. Then there will be no obvious errors from Module:grc-accent's flawed way of determining vowel length. — Eru·tuon 05:01, 29 August 2017 (UTC)Reply

Short vowels are so much more common than long vowels that the module should assume unmarked ambiguous vowels are short rather than assuming they're long. —Aɴɢʀ (talk) 07:05, 29 August 2017 (UTC)Reply

That makes sense, but I have to think about how to incorporate that into the logic of the modules. For instance, to make it clear that it is merely a fallback and that the length still needs to be marked explicitly. — Eru·tuon 20:16, 29 August 2017 (UTC)Reply

See Module:grc-accent/sandbox for a function that marks implied length. — Eru·tuon 20:16, 29 August 2017 (UTC)Reply

Now to fix cases like Αἰθίοψ, Αἰθίοπος that result in Αἰθῐ́οψ, Αἰθίοπος, Αἰθίοπῐ, .... — Eru·tuon 23:10, 29 August 2017 (UTC)Reply

Will Module:grc-accent accept ε̄ and ο̄? The "spurious diphthongs" ει and ου are sometimes spelled ε and ο in early inscriptions, so we'd probably want to be able to mark them long in entries for the inscriptional spellings without triggering a module error. —Aɴɢʀ (talk) 06:39, 30 August 2017 (UTC)Reply

The only way for this module to handle contracted o-stems like Πειρίθους (Peiríthous) is if we pretend it's Πειρίθος and then manually override the nominative, accusative, and vocative singular (i.e. {{grc-decl|Πειρίθος|ου|AS=Πειρίθουν|NS=Πειρίθους|VS=Πειρίθου|form=sing}}). For ὀστοῦν (ostoûn) we have to pretend it's ὀστόν and then manually override the nominative singular and nominative plural: {{grc-decl|ὀστόν|ου|NP=ὀστᾶ|NS=ὀστοῦν}}. Not very efficient. —Aɴɢʀ (talk) 12:14, 27 July 2017 (UTC)Reply

@Angr: I already fixed this here. —JohnC5 15:57, 27 July 2017 (UTC)Reply

@JohnC5: It seems clunky how the various modules have to search the args.form parameter to determine the gender and number. It might be better to generate a table that can be referred to. I'm working on that idea. — Eru·tuon 20:25, 2 August 2017 (UTC)Reply

@Erutuon: Be careful. |form= doesn't only contain MFN, sing, dual, plur, full, but also X, con, and open. It's not so easy. —JohnC5 21:19, 2 August 2017 (UTC)Reply

@JohnC5: Right. I've avoided modifying it when extracting gender and number from it. — Eru·tuon 21:34, 2 August 2017 (UTC)Reply

I've integrated args.g . It seems to work; I checked that the masculine–feminine terms still showed both genders. — Eru·tuon 23:12, 2 August 2017 (UTC)Reply

I've modified the adjective declension-matching function and its data so that it returns an error giving the possible feminine endings. So {{grc-adecl/sandbox|στᾱ́ς|στᾶσα}} yields the error message Declensions were found that matched the ending of the masculine “στᾱ́ς”, but the corresponding feminine endings were “-αινᾰ”, “-ᾱσᾰ”, none of which matched the ending of the feminine “στᾶσα”. This should be more helpful, because editors can then see that they need to add a breve. — Eru·tuon 19:02, 3 August 2017 (UTC)Reply

@Erutuon, thanks for all of your recent changes. Please tell me when you are done with this round so I can start adding more features (and maybe refactoring) without creating merge conflicts. Cheers! —JohnC5 14:39, 4 August 2017 (UTC)Reply

@JohnC5, I think I'm done for now. I don't have any immediately implementable big ideas for improvements. I wish the accent were handled differently, but I have no clearly formulated idea of how to do that. (Wow, I'm amazed how much faster the new noun-declension identification system is in Module:grc-decl/sandbox/testcases and the appendices.) — Eru·tuon 20:21, 4 August 2017 (UTC)Reply

I've made the adjective declension detection indifferent to length, when there are no conflicts. Conflicts are detected automatically, so new adjective declension types can be added with no extra worry. If there is a conflict, and length of the ending absolutely has to be specified, the error message will help the editor resolve the conflict.

I'm sort of puzzled, though, because {{grc-adecl/sandbox|ποιήσας|ασα}} isn't working. Maybe it's a hiccup in Lua's handling of tables, or something else I've overlooked. — Eru·tuon 20:22, 6 August 2017 (UTC)Reply

@Angr, Barytonesis, Erutuon, sorry to bother, but could someone work out the distribution of adjectives in -εις (-eis) ~ -εσσᾰ (-essa) vs. those in -εις (-eis) ~ -εισᾰ (-eisa)? The two section in Smyth are §299 (and §299D) and §307. Is the different just in the tone placement (which mod:grc-decl assumes) or is it participles vs. adjectives? —JohnC5 15:19, 9 August 2017 (UTC)Reply

Yes, χᾰρῐ́εις (kharíeis) and its sisters are adjectives not connected to a verb. — Eru·tuon 17:47, 9 August 2017 (UTC)Reply

I'm working on making the module output the correct tables for, for example, ἀδεής (adeḗs). For Ionic, I assume it should be uncontracted in some cases where Attic is contracted.

But there are a number of forms with three consecutive vowels, such as nominative, accusative, and vocative dual *ᾰ̓δεέε (*adeée). I suspect at least that one is impossible, but it is likely unattested. The module thinks the adverb is *ᾰ̓δεέως (*adeéōs), but it should be ᾰ̓δεῶς (adeôs) (like the Attic form), as can be seen in Herodotus, Histories 3.65.

I'm guessing that's because of the vowel in the root, which would result in three consecutive vowels if two of them did not contract. Compare a root ending in a consonant: the Ionic adverb from ᾰ̓ληθής (alēthḗs) is ᾰ̓ληθέως (alēthéōs), contrasting with the Attic ᾰ̓ληθῶς (alēthôs). So perhaps the genitive plural of ἀδεής (adeḗs) would be the contracted ᾰ̓δεῶν (adeôn) in Ionic as well as Attic. If so, the module could contract some or all of the forms if ε (e), or any vowel, is the last letter of the root. But I don't have enough specificity on this yet. — Eru·tuon 22:20, 10 September 2017 (UTC)Reply

@JohnC5 called my attention to the fact that Ῥοδῶπις (Rhodôpis) is being accented Ῥόδωπῐς (Rhódōpis). During the declension detection process, Module:grc-decl/decl places the word in the accent category pax. Then it determines the stem forms Ῥοδώπ, Ῥόδωπ, Ῥοδωπ, and uses the second one for the nominative singular. But that yields the wrong form because there's a vowel in the ending -ῐς (-is), though it's right for some other third-declension nouns without a vowel in the ending.

In most cases, accent, in a manner of speaking, doesn't move from its position in the nominative singular unless forced forwards. The exceptions that I can think of are adjectives like ᾰ̓ξῐ́ᾱ (axíā), ᾰ̓́ξῐαι (áxiai), where accent moves backwards, and third-declension monosyllabic nouns like πούς (poús), ποδός (podós), where it changes position based on the case.

The four categories recognized by the module are antepenult (prx, or proparoxytone), penult (pax, standing for paroxytone, though it also includes properispomenon), acute on ultima (unmarked; oxytone), circumflex on ultima (con; perispomenon). But it might be easier to separate between ultima accent and non-ultima accent, and then, for non-ultima, measure accent placement from the beginning of the word rather than the end. Then, for instance, the module would make ᾰ̓́νθρωπος (ánthrōpos) have accent on the first syllable unless the rules of accent prohibit it, Ῥοδῶπῐς (Rhodôpis) on the second syllable, and so on. I'm not totally sure how this would work, but I'll look into it sooner or later. — Eru·tuon 06:27, 23 September 2017 (UTC)Reply

Update: I've made functions for detecting and adding accent in Module:grc-accent. The accent-adding function is safe: for instance, if you try to add accent to the first syllable of ᾰ̓νθρωπου (→ *ᾰ̓́νθρωπου (*ánthrōpou), after ᾰ̓́νθρωπος), it will actually be added to the second syllable (ᾰ̓νθρώπου (anthrṓpou)); and an accent can't be added to a word that already has one. — Eru·tuon 00:10, 14 October 2017 (UTC)Reply

Okay, Module:grc-decl/sandbox and its submodules now use the new accent accent functions, and Ῥόδωπῐς (Rhódōpis) is accented correctly. I discovered some errors in the testcases while refactoring the modules. The testcases for the sandbox are about twice as slow as the testcases for the main module, perhaps because of the time required to add accents to each form. But the important thing is that the accentuation system is more theoretically correct, and should be easier to modify when there are errors.

There are some remaining issues. εω is now treated as a diphthong when adding accents to nouns, so that πόλεως isn't accented as *πολέως, but there may be cases where this results in undesirable behavior. — Eru·tuon 01:11, 20 October 2017 (UTC)Reply

@Erutuon: So to clarify, εω only patterns like -ηο- (its origin) under all of the following limited circumstances combined:

• in the final syllable
• in Attic
• as a result of -ηος

So these words are gonna cause problems. It should not be a rule that -εω- pattern as a diphthong, but only when it is the result of -ηο- in the final syllable in Attic. —JohnC5

I'm not sure how often it'll cause problems, but I probably should figure out some kind of switch just in case, and to be theoretically correct. At the moment the short αι, οι and synaeresis switches are combined, which is messy. — Eru·tuon 08:20, 21 October 2017 (UTC)Reply

Done: args.synaeresis. — Eru·tuon 00:44, 27 October 2017 (UTC)Reply

@Erutuon, I'm not sure I know what do about this one. —JohnC5 02:49, 24 September 2017 (UTC)Reply

@JohnC5: If it's one-of-a-kind, as seems to be the case, it should probably just be entered as irregular. If another noun like it turns up, we could program the pattern into the module. — Eru·tuon 01:59, 25 September 2017 (UTC)Reply

@Erutuon: Here's this and this, including φωΐς, δαΐς, ξοΐς, Τρωΐς, and νῆϊς as examples. I also wonder what ever declensions are broken by diaereses. Is it possible to remove the diaereses somehow, process the forms, and then add the diaereses back? —JohnC5 02:17, 25 September 2017 (UTC)Reply

@JohnC5: Certainly (regarding the diaeresis). I can create a function for adding a diaeresis in Module:grc-accent.

Hm, those other nouns are different. They have a genitive in -ῐδος (-idos) and hence are pretty regular, while Σάϊς (Sáïs) has a genitive in -εως (-eōs) (along with its other odd forms). Oh, duh. Σάϊς is like πόλις (pólis). Brain fail. So it could be supported if the module is made to see the pattern behind the diaeresis. — Eru·tuon 03:49, 26 September 2017 (UTC)Reply

At Βορέας an anon just pointed out that even though the declension table is marked with |dial=dor, the note at the bottom still says, "This table gives Attic inflectional endings." Can that be fixed? —Aɴɢʀ (talk) 21:19, 16 October 2017 (UTC)Reply

@Angr: Yep, I had accidentally reversed the notes. Fixed. — Eru·tuon 21:23, 16 October 2017 (UTC)Reply

At ὄνειδος I notice that only the contracted forms are being shown, even though the table does not say |form=con. —Aɴɢʀ (talk) 11:03, 20 October 2017 (UTC)Reply

Right, that's a result of having the dialect default to Attic, which has contracted forms (see above). To include non-contracted forms, one has to add a table for the dialect that uses them. — Eru·tuon 20:42, 20 October 2017 (UTC)Reply

Another goal I have is to fix the parameters for irregular paradigms. At the moment, for a plural paradigm like ἡμεῖς (hēmeîs), you have to enter a lot of empty numbered parameters for the singular and dual ({{grc-decl|irreg||||||||ἡμεῖς|ἡμῶν|ἡμῖν|ἡμᾶς|VP=-|dial=att|form=X-plur}}). The numbered parameters should start with the nominative plural instead when the paradigm is plural-only. — Eru·tuon 00:50, 27 October 2017 (UTC)Reply

Done. Now the declension template for ἡμεῖς (hēmeîs) will look like {{grc-decl|irreg|ἡμεῖς|ἡμῶν|ἡμῖν|ἡμᾶς|VP=-|dial=att|form=X-plur}}. And the module checks for any numbered parameters that don't have a corresponding form. — Eru·tuon 21:38, 31 October 2017 (UTC)Reply

@Erutuon, Barytonesis, Angr, for the question about εὔελπῐς (eúelpis) and its neuter ending, are there any dental adjectives that don't have neuters like εὔελπῐ (eúelpi)? The deletion of a word-final stop is regular in AG, so I'm curious whether there are other outcomes. —JohnC5 02:20, 2 November 2017 (UTC)Reply

@JohnC5: I'm having trouble finding any other dental adjectives at all... But yes, I don't see what other outcome there could be. --Barytonesis (talk) 08:37, 2 November 2017 (UTC)Reply

I found a reference to stems in dentals by looking up εὔελπις (eúelpis) in the Greek index of Smyth: § 291a. It mentions εὔχαρις (eúkharis). I guess they're rare, and there probably isn't another pattern to the neuter nominative singular. — Eru·tuon 19:49, 2 November 2017 (UTC)Reply

So, we should be good to implement this? —JohnC5 21:08, 2 November 2017 (UTC)Reply

@JohnC5: I've implemented it in Module:grc-decl/sandbox/decl/data. It's harder to figure out how to do it in Module:grc-decl/decl/data. — Eru·tuon 21:24, 2 November 2017 (UTC)Reply

@Erutuon: I think this is fine to leave until our next "production release". Thanks for all your recent work, by the way. —JohnC5 03:50, 3 November 2017 (UTC)Reply

@Barytonesis notes on the talk pages of βελτίων (beltíōn) and χείρων (kheírōn) that comparatives in -ων (-ōn), have Attic masculine and feminine accusative singular and nominative and accusative plural forms showing contraction, from a stem in pre-Greek -os- (Smyth § 293).

To add these alternative forms, the crucial question is how to distinguish them from adjectives like εὐδαίμων (eudaímōn) (and nouns like δαίμων (daímōn)), which do not have these forms, but are handled by the same functions in Module:grc-decl/decl/data as the comparatives (namely "3rd-cons" and "3rd-N-cons"). Maybe the only solution is a boolean parameter. Both the comparatives and the non-comparatives are input using the same parameters otherwise: bare stem βελτῑ́ον (beltī́on) for the adjective, and nominative and genitive δαίμων, -ονος (daímōn, -onos) for the noun. (Part of speech being "adjective" should not be relied on; there could be an idiomatic substantivization of a comparative in -ων (-ōn).) — Eru·tuon 22:56, 7 November 2017 (UTC)Reply

I arrived at a solution. Now (once the sandbox code is moved to the main module) the code comp in the |form= parameter will add the Attic contracted forms: {{grc-adecl|βελτῑ́ον|form=comp}}. — Eru·tuon 21:06, 26 November 2017 (UTC)Reply

Eru @Erutuon: gave me permission to write a few observations-feedback from an average user. I don't wish to spoil your discussion page which is reserved for serious subjects, so i posted them at my page; may be of interest for you to know how laymen see your ancient greek lemmata. A fan of your (Eru, John and all AnGree team) work, sarri.greek (talk) 19:33, 29 November 2017 (UTC)Reply

Thank you for your attention to previous post. In Cat-άρχης, isn't sing.voc -άρχα instead of -άρχη? (as in -άρχης, -μέτρης, -πώλης,.. 2nd component from verb). All words-in-list=hellenistic (checked @perseus, mediaevalKriaras@pyli, DSMG@pyli) Lots of infl.types are missing from perseus, prob because hell not anc. I will add |period=koi1}} at IPA. But what do I add at Etymology:? I'll also check gre.mod.-άρχης to make sure of correpsondance. Thanks sarri.greek (talk) 22:38, 29 November 2017 (UTC)Reply

I couldn't remember the rules, but Smyth § 226 confirms this: masculines in -ᾱς have a vocative singular in -ᾱ, those in -ης have -η except for nouns in -της (-tēs), demonyms, and compounds, which have -ᾰ. So, words in -άρχης (-árkhēs), as compounds, have a vocative singular in -ᾰ. There is actually a parameter to change the vocative ending to -ᾰ: |voc=α. (Seems not the most convenient for me, as I don't have a Greek keyboard and have to use {{chars}}.) — Eru·tuon 23:04, 29 November 2017 (UTC)Reply

O dear. @Erutuon: Me neither (a polytonic I mean). I saw your nice CE coming through. Don't bother with me about trivials. But infl is important. And here is the rule from my old Tzartzanos gammar:

nouns FistDecl masc -ης, form voc.sing -α βραχύ are:

1) ethnics (ώ Πέρσα, ω Σκύθα -you got this ok I think-.

2) the -της (ω πολίτα, ω δικαστά, BUT ω δέσποτα -you got them OK-)

3) the -άρχης, -μέτρης, -πώλης, -τρίβης, -ώνης etc ήτοι compounds with 2nd synthetikcon a verb (I think there are the ones you need.

I was wondering if στοιχειολάτρης falls into this category. IF ETYMO is from verb λατρεύω, then it does. My instinct goes with ω ...λάτρα, but I am NOT sure). sarri.greek (talk) 02:28, 30 November 2017 (UTC)Reply

I fixed the template so that you can do, for example, {{grc-decl|ἐθνᾰ́ρχης|ου|voc=α}} to indicate that vocative singular ends in short alpha.

Most likely στοιχειολάτρης (stoikheiolátrēs) falls in this category; it is a compound and its meaning does suggest it is derived from λατρεύω (latreúō). However, Smyth does not say that the second component has to be from a verb. It does seem that most compounds in this declension class do have a second component from a verb. I did find an exception in Category:Ancient Greek masculine nouns in the first declension – μῑσογῠ́νης (mīsogúnēs) – but the vocative is probably unattested. — Eru·tuon 03:29, 30 November 2017 (UTC)Reply

yes, you are right @Erutuon: I would never say μισαγύνα. P.S. Whenever you have spare time (which I guess you do not) check ἐθνάρχης I did infl.types and the gre.mod versions. When you say it is ok, I can do more. sarri.greek (talk) 03:43, 30 November 2017 (UTC)Reply

@Sarri.greek: I actually edited ἐθνάρχης (ethnárkhēs) after you did, and everything you did was correct. — Eru·tuon 06:29, 30 November 2017 (UTC)Reply

Thanks @Erutuon: and bravo. I copied. sarri.greek (talk) 06:44, 30 November 2017 (UTC)Reply

I am going to update Module:grc-decl and its submodules now with the new version from Module:grc-decl/sandbox and its submodules.

Irregular paradigms in {{grc-decl}} or {{grc-adecl}} will probably display incorrectly or have module errors until I correct the parameters.

There will probably be other bugs or errors as well. I will deal with module errors as they come up. If you notice inaccuracies, please post them here. — Eru·tuon 02:24, 2 December 2017 (UTC)Reply

Bugs or missing features:

• ἠώς (ēṓs): title of table
• σύμπαν (súmpan): no dative plural generated for substantivized adjective
• τράχηλος (trákhēlos): heteroclitic (masculine singular, neuter plural); can't put both in one table
• Homeric first-declension genitive plural in -ῶν (-ôn) mainly after a vowel; does Ionic use -ῶν (-ôn) after a vowel (or elsewhere), or only -έων (-éōn)?
• Allow πῑ́ων, πῑ́ειρᾰ, πῖον (pī́ōn, pī́eira, pîon) to be shown in table. Add parameters for feminine nominative and genitive singular (or just nominative), to be declined like a noun and added to the feminine part of table? Or, since πῑ́ων is used for both masculine and feminine, allow single-gender adjective table?
• Would be nice to be able to automatically decline ὗς (hûs). Can't currently because stem is the rough breathing!
• εἷς (heîs), οὐδείς (oudeís): declension could be automatically generated if masculine–neuter and feminine were supplied separately.
• ἀαγής (aagḗs): Doric has the same forms in the supposedly "contracted" and "uncontracted" tables.
• ὄν (ón): dative plural is missing a smooth breathing.
• Allow removing case-rows or gender columns from table: for instance, nominative and vocative for ἑαυτοῦ (heautoû), vocative for ἐκεῖνος (ekeînos), neuter for ἐμαυτοῦ (emautoû), σεαυτοῦ (seautoû) and their dialectal forms.
• Need some way to handle θήλεα (thḗlea), feminine of θῆλυς (thêlus).
• Add "fun" Epic dative plurals in -εσσῐ, either always or whenever some algorithm predicts that they make sense metrically in hexameter.

O @Erutuon: you are working so hard checking all templates. Was your question: Γανυμήδης, ...How many are there...? rhetoric? I hope it is not ῠ῞βρῐς but i've made a list. (I also made up some fonts). I could post the list somewhere if you need it, any time. sarri.greek (talk) 10:42, 10 December 2017 (UTC)Reply

@Sarri.greek: It wasn't rhetorical. But I just now realized that we already support this declension pattern in nouns like Σωκράτης (Sōkrátēs). The module just had to recognize the genitive -εος (-eos). — Eru·tuon 02:42, 11 December 2017 (UTC)Reply

Yes @Erutuon: I was wondering... what is the Grammar-book one uses as standard source in wiktionary? I have never seen a grammar in english. I dont' want to repeat things that are common knowledge for philologists.

• list of names with 2nd component verb μέδω (μεδ- or μηδ-) MASC -ης: Ἀρχιμήδης Διομήδης Ἀγαμήδης Γανυμήδης Ἐπιμήδης Εὐμήδης Εὐρυμήδης Ἐχεμήδης/Ἐχεμέδης Θρασυμήδης Κλεομήδης Κλυτομήδης Λυκομήδης Μεγαμήδης Μεγαμηδείδης Παλαμήδης Περιμήδης Πολυμήδης, Φρασιμήδης. FEM -η: Ἀγαμήδη Ἀλιμήδη Ἐκαμήδη Μεγαμήδη Περιμήδη Πολυμήδη Πυκιμήδη MASC -μέδων: Μέδων, Ἀλκιμέδων Ἀμφιμέδων Αὐτομέδων Εὐρυμέδων Λαοδέμων Ναυσιμέδων FEM: Ἰφιμέδεια Μήδεια Πρωτομέδεια. Also: Μέδη Ἀλκιμέδη Ἰφιμέδη ...
• from my schoolbooks, just a ref: @Tzartzanos: Thrid Decl. simgmolecta -σ. ... ... Compound Propria sigmolecta with second component Noun neuter sigmolecton (but only in stem -εσ form). Διογένης<Διογένεσ (γένος). Περικλῆς<Περικλεεσ- (κλέος). Δημοσθένης (σθένος). Πολυνείκης (νεῖκος). Θηραμένης (μένος). Διομήδης (μῆδος). Nominativus of such neuter compounds appear in the long stem and τhe other cases the full stem -es. Voc is the very stem, with stress ascending. @Stamatakos: The -ης Propria, because of their remarkable resemblance to masc.1stDecl. -ης ] form besides the normal Acc. -η one more Acc by its side -ην (Διομήδη & Διομήδην, Δημοσθένη & Δημοσθένην)... Homer and Ionic have uncontracted types θεοειδέα (acc.sing), ὁμηγερέες (Nom.pl) Διομήδεος, Διομήδεϊ.

as for the irregulars: --heterogenus (their plural): 1) with only one pl: ὁ σῖτος - τα σῖτα. ὁ λύχνος - τα λύχνα. Ι think your τράχηλος is here. 2) double: ὁ δεσμός - οἱ δεσμοί/τα δεσμά. το στάδιον, οἱ στάδιοι/τα στάδια. --diplogenus (at sing): ὁ ζυγός/το ζυγόν, one pl: τα ζυγά. ὁ νῶτος/το νῶτον, τα νῶτα. --heteroclitic for number or certain cases=crazy! Each, a template of itsself. e.g. η γυνή, το πῦρ τα πυρά.... got 3 pages of these. sarri.greek (talk) 01:11, 12 December 2017 (UTC)Reply

I typically use Smyth, that is, Herbert Weir Smyth's Greek Grammar, because that's the only grammar I have. But @ObsequiousNewt created the module, and he may have used other references; certainly he did when adding more obscure dialectal forms.

Forms with -μήδης (-mḗdēs) and similar components was not quite what I meant, but it is interesting that there are so many. They could perhaps be listed at μέδω (médō) under the related or derived terms header.

I guess I will consider adding support for heteroclitic nouns. I don't know how complex it would be, but it would be nice to be able to automatically generate the forms and display them in a single table.

I wonder if the first-declension accusative singular for (Attic and Ionic) third-declension nouns in -ης are common enough that we should automatically list it. (Δημοσθένην alongside of Δημοσθένη in the table for Δημοσθένης.) It would make the header "Third declension of ..." technically inaccurate, but it would be more convenient than manually adding an alternative accusative singular using the parameter |AS=. — Eru·tuon 02:07, 12 December 2017 (UTC)Reply

My sources for the module were Smyth's grammar and Carl Darling Buck's Introduction to the study of the Greek dialects. There's a link to the latter on Appendix:Ancient_Greek_dialectal_conjugation. Of course, now that the module doesn't display dialectal forms (as it shouldn't), the second is only relevant to the pages that that feature was replaced by. Re: the -ην accusative, I would tentatively vote for not putting it in the table, or at least marking it as nonstandard (but really, I don't think nonstandard forms should be in the table. We learned that lesson ages ago.) — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 04:31, 12 December 2017 (UTC)Reply

Thank you for the Smyth Gramm, @Erutuon: (found this pdf too) & thanks @ObsequiousNewt:, I have downloaded Buck, for my archive, it is of course too diff for my level. You experts, me layman... I don't know how you do all these magic templates. My dream when i was student, was to have decl. table for each word of the comments under the few paradigmata of my schoolbook. And now, you are making dream come true (alas, 40 years later!) Thanks sarri.greek (talk) 09:38, 12 December 2017 (UTC)Reply