Wiktionary talk:About Ancient Greek/Archive 3

Template:grc-pron

This template is still under development, but I think it's good enough to be used in entries, and so I thought I'd let folks know about it. The idea is to improve upon Gilgamesh's already excellent work with {{grc-ipa-rows}} using the newfound powers of a proper server-side scripting language. The first problem with ipa-rows is that it's tedious to enter in all the information. {{grc-pron}} auto-detects the entry title and does all the work for you. The second problem is that I, for one, can't remember all the stuff; it is possible that I'm just remarkably stupid and the only one with this difficulty. I remember that αι and οι are short at the end of a word, and that the token replaces the i with a j. I remember that a gamma before itself or another dorsal is replaced with a nasal, and the token has it as an n. Past that, there's all sorts of stuff that I always forget to encode that is meaningful, like certain combos of nasals and plosives. grc-pron is designed to remember all of those rules for you. The other beauty is that grc-pron is more flexible. If there is a rule which is not already encoded into tokens in ipa-rows, then all the entries would have to be edited. grc-pron has no such limitations. Anyway, please keep in mind that the code's ink is still drying, so any use should undergo careful scrutiny. Please let me know if you see any mistakes or have suggestions. Thanks all. -Atelaes λάλει ἐμοί 04:48, 24 March 2014 (UTC)

At Βορέας it complains of "ambiguous vowel α at 5" and at ἰτέα of "ambiguous vowel ἰ at 1, ambiguous vowel α at 4", but the template has no documentation telling us how to disambiguate those vowels. —Aɴɢʀ (talk) 07:59, 13 April 2014 (UTC)

Huh. I could have sworn I had written documentation for it. Up now. -Atelaes λάλει ἐμοί 08:15, 13 April 2014 (UTC)

Thanks. It's pretty dense prose, though; could you provide some examples? Should I write {{grc-pron|w=Βορέᾱς}} and {{grc-pron|w=ῐτέᾱ}} or what? Also, maybe it should be moved to {{grc-pronunciation}} or {{grc-pronun}} since {{grc-pron}} looks like a headword-line template for Ancient Greek pronouns (cf. {{en-pron}}, {{el-pron}}, {{de-pron}}). —Aɴɢʀ (talk) 08:25, 13 April 2014 (UTC)

It doesn't mark vowel length for 1st-century BC Egyptian Koine, while {{grc-ipa-rows}} does. Is that intentional? —Aɴɢʀ (talk) 08:30, 13 April 2014 (UTC)

Yeah, I can see how the prose would be hard to take in at a glance. I'll try and scour up some good examples. But I think you do have the right idea on how to use it. {{grc-pron|w=ῐτέᾱ}} looks wrong to my eyes, but since a smooth breathing mark is identical to an absent one, I have to admit that it's the most sensible way to format it, even if I would use {{grc-pron|w=ἰ˘τέᾱ}}.....for no good reason other than my own personal obsessive-compulsive nature. As for the name, I quite agree; I've never been happy with it either. I was kind of leaning towards {{grc-P}}; as a commonly used template (or at least one which will hopefully be commonly used), I like the idea of something short and sweet. {{grc-pronunciation}} is rather long, and {{grc-pronun}} rings too much like pronoun in my mind, even if the absence of the second o should make it clear on closer examination. As for the missing vowel length in Egyptian, yes, that was deliberate. At least according to the 'pedia that distinction no longer existed. Mind you, I'm no expert. I was taking most of my cues from Gilgamesh's work and what the 'pedia had to say. -Atelaes λάλει ἐμοί 20:15, 13 April 2014 (UTC)

While we're discussing the various nuances of the template, I should probably mention that there's some javascript which goes along with it, which isn't being forced upon the general public yet. If you want to try it out, it resides at User:Atelaes/viewSwitching.js. Basically, what it does is to make the default view a single line, showing the Classical, 4th cent, and 15th cent pronunciations, separated by arrows, with a button to expand it to the standard five lines. The idea is that the current implementation is great for detail, terrible for taking up a lot of space before what most people are probably interested in, the definitions. I've proposed inserting the script into our common JS pool in the GP, and I want to give it a week to stew, just in case anyone realizes that it's going to cause the entire project to permanently crash. -Atelaes λάλει ἐμοί 21:10, 13 April 2014 (UTC)

Dialectical variation in conjugated forms

I've been contributing a bit more to Wiktionary lately, especially in the area of Ancient Greek— not that I know anything about it professionally, but I've been working off LSJ on Perseus. So I just looked at the list of requested terms and found "αἴθυια", which, according to LSJ, has an Epic/Ionic declension in αἰθυίης, αἰθυίῃ, &c. but an Attic/Doric/Aeolic declension in αἰθυίας, αἴθυιαι, &c. This is probably a stupid question, but how should I indicate the declension(s) on the page? ObsequiousNewt (talk) 22:32, 25 March 2014 (UTC)

The answer necessarily owns up to some of the failings in our inflection templates which were written long ago (back when documentation pages didn't exist ;-)), and have gotten various refreshes over time. The two templates are {{grc-decl-1st-ala-prx}} and {{grc-decl-1st-ets-prx}}. ala takes a parameter "titleapp", which is short for title append, while ets does not, taking only title, which overwrites the default title. So, if I were to do it I would use the following code:

{{grc-decl-1st-ala-prx|titleapp=({{grc-att}}, {{grc-dor}}, {{grc-aio}})}}

{{|grc-decl-1st-ets-prx|title=[[Appendix:Ancient Greek first declension|First]] declension of {{l|grc|αἴθυθα}}, {{l|grc|αἰθυίης}} ({{grc-ion}}, {{grc-epi}})}}

Does that makes sense? -Atelaes λάλει ἐμοί 02:22, 26 March 2014 (UTC)

Yeah, I see what you mean. Is this a consistent pattern in (proper) nouns? Should I extend the titleapp parameter to the other templates? And I'd be glad to contribute to the documentation of existing templates; I've noticed that some of them are lacking. ObsequiousNewt (talk) 02:55, 26 March 2014 (UTC)

I suspect it is. Ionic often ends on eta when other dialects end on long alpha, but it does depend on what kind of long alpha it is. I think that titleapp is a useful enough feature that it's worth putting on templates which lack it. Also, if you're interested, one template task that I've been meaning to do, but haven't gotten around to yet, is to update the title bar of all the verb templates with the indicative 1st singular forms. -Atelaes λάλει ἐμοί 03:26, 26 March 2014 (UTC)

Okay, I'll start working on the templates. Thanks for your help! ObsequiousNewt (talk) 03:34, 26 March 2014 (UTC)

I've done a bit of research, and now I understand how this works—the ā changes to ē in Epic, Ionic, and conditionally Attic—so perhaps the best thing to do would be to have two inflection tables for most 1st-declension nouns, in a format much like you just described? ObsequiousNewt (talk) 04:24, 26 March 2014 (UTC)

Well.....often enough the long alpha/eta exists in the nominative, which makes for separate entries, which negates the need for separate entries. However, we always get the short alpha in proparoxytonal words, so we'll likely see this issue come up often, though not always, as some words are simply not attested in Ionic. So, I think it would be a good idea to create such a dual template, perhaps at {{grc-decl-1st-prx}}, but keep the existing two. -Atelaes λάλει ἐμοί 07:08, 26 March 2014 (UTC)

New template is live at λύρα—how is it? ObsequiousNewt (talk) 02:26, 27 March 2014 (UTC)

Very nicely crafted. I like it. -Atelaes λάλει ἐμοί 05:30, 28 March 2014 (UTC)

Actually, on a similar topic: it seems to be current practice to put alternate imperfect, aorist, future, etc. forms on the lemma page while alternate presents are conjugated on their own page. Would it not make more sense to conjugate alternate presents on the lemma page? Or is this already the practice? ObsequiousNewt (ἔβαζα|ἐτλέλεσα) 18:22, 27 April 2014 (UTC)

Luacizing the conjugation tables

Why? Because {{grc-conj-aorist-1}} takes eleven parameters and in fact should take twelve (the first aorist active infinitive should have properisopomenon but currently has proparoxytone.) Writing functions to modify the root to give it the correct accent is simple, but the problem lies in how to structure the system (and keep backwards compatibility until we've changed all the conjugation tables.) I can come up with a couple options, neither of which look very appealing:

• Option 1: create new templates, perhaps grc-conj-pres (imp, fut, aor, perf, plup, futp) which call the corresponding existing templates with parameters gained from an existing module; it would look something like
  

{{grc-conj-perf-labial|λελοιφ|λελοι}}

on the page, and

{{grc-conj-perfect-labial|{{#invoke:grc-accent|penult|{{{1}}}}}|{{#invoke:grc-accent|ult|{{{1}}}}}|{{{1}}}|{{#invoke:grc-accent|penult|{{{2}}}}}|{{#invoke:grc-accent|ult|{{{2}}}}}|{{{2}}}|{{#invoke:grc-accent|circ|{{{2}}}}}}}

on Template:grc-conj-perf-labial.

• Option 2: rewrite {{grc-conj-perfect-labial}} etc. in Lua (well, to invoke a Lua template anyway) where said template would provide both accent (probably #include from a common Module:grc-accent) and the entire template text. To provide backwards compatibility, we'd check if the template was called with more than two parameters (or four for aorist, or one for present and imperfect) and, if so, use the original behavior. But again, module calls are more expensive than template calls, IIRC.

Thoughts? ObsequiousNewt (ἔβαζα|ἐτλέλεσα) 16:27, 17 May 2014 (UTC)

First, yes, they absolutely do need to be luacized. Inasmuch as I'm quite proud of the templates as they stand, they are nightmarish to use; most people simply don't feel comfortable using them at all, and they're tedious and time-consuming for those few of us who do. I would lean towards option 1, as I suspect that option 2 would be hamstrung with legacy issues. Following option 1 would still allow us to follow option 2 once everything is up and running, if we feel it's feasible ({{grc-conj-perfect-labial}} could certainly invoke Module:grc-conj-perf). -Atelaes λάλει ἐμοί 04:16, 18 May 2014 (UTC)

This is what I came up with before I gave up on Lua in disgust. My god that is the worst language I have *ever* used. ObsequiousNewt (ἔβαζα|ἐτλέλεσα) 19:22, 18 May 2014 (UTC)

I think you may just need more time to get used to it. Most problems that people have when learning new programming languages is that they think too much in terms of their "old" language, and try to make the new one fit into that. It may help to see Lua on its own terms. In any case, you should never use :sub in Wiktionary, always use mw.ustring.sub. The former will break up UTF-8 characters into bytes, which messes things up quite badly.

In any case, I may be able to help, but I don't know anything about Ancient Greek so I'm not sure what the module is meant to achieve. Could you give a concise step-by-step explanation? —CodeCat 19:46, 18 May 2014 (UTC)

Hmmm.....for some reason I figured you were made of tougher stuff. I got penult working on non-diphthongs, which you can see at User:Atelaes/Sandbox. -Atelaes λάλει ἐμοί 22:28, 18 May 2014 (UTC)

Is that a challenge, sir? But in all seriousness, Lua defies all the conventions (array indices start at 1‽) and lacks basic functions present in every other language I've worked with (list contains element) and yes, I was just in the process of changing things to mw.ustring.sub... basically I'm really wondering what the advantage of Lua really is supposed to be. Anyway, I'll take another stab at it later if I have time. ObsequiousNewt (ἔβαζα|ἐτλέλεσα) 01:03, 19 May 2014 (UTC)

I've always assumed that WM's top two priorities when they picked a server-side scripting language were: 1. light-weight, light-weight as hell, and 2. safe enough (or modifiable to safe enough) that you might consider letting a bunch of random strangers write code for your site to run without any oversight whatsoever. Ease of use is nice, but WM's wikis have an over-abundance of really smart people who like learning and figuring things out. -Atelaes λάλει ἐμοί 02:54, 19 May 2014 (UTC)

ArielGlenn (talk • contribs) would probably know, if they were still around. -Atelaes λάλει ἐμοί 03:02, 19 May 2014 (UTC)

Also, before this gets too far, I think that {{grc-conj-perf|λέλοιφα}} should be sufficient. Phi is a labial, which the template should know gets combined with the various endings. This won't work all the time, of course, but a well scripted module could handle it most of the time. It will, of course, have to be flexible enough to accept correction when its assumptions turn out to be incorrect. Also, is that a real perfect? -Atelaes λάλει ἐμοί 07:20, 19 May 2014 (UTC)

That would seem to imply option 2, though. (And yes, that should be λέλοιπα.) ObsequiousNewt (ἔβαζα|ἐτλέλεσα) 22:02, 19 May 2014 (UTC)

Why would that imply 2? -Atelaes λάλει ἐμοί 06:27, 20 May 2014 (UTC)

It is accomplished. ObsequiousNewt (ἔβαζα|ἐτλέλεσα) 02:47, 12 August 2014 (UTC)

History of palatalization in Greek

I asked this question at User:Atelaes's talk page, but he doesn't seem to be around much anymore, so I'll ask again here. {{grc-IPA|w=κι}}, {{grc-IPA|w=κει}}, {{grc-IPA|w=κη}}, and {{grc-IPA|w=κοι}} all show palatalization of /k/ to /c/ in the 10th and 15th century lines, but {{grc-IPA|w=κυ}} doesn't show palatalization. Is that right? (The same applies to the equivalents with γ.) —Aɴɢʀ (talk) 06:26, 8 October 2014 (UTC)

A quick look at w:Medieval Greek tells me that palatalization should be triggered by everything but α,αυ,ο,ου,ω. A quick test reveals that this is the case with everything but αι and υ. I trust Atelaes' knowledge, but I agree that he should look at this. ObsequiousNewt (ἔβαζα|ἐτλέλεσα) 02:14, 9 October 2014 (UTC)

The problem is he's currently inactive. I would go ahead and edit Module:grc-pronunciation myself, except I can't figure out which part of it says what vowels to palatalize before. —Aɴɢʀ (talk) 05:47, 9 October 2014 (UTC)