Wiktionary talk:About Ancient Greek/Archive 5

Dialectal forms in inflection tables

I'm starting to think that maybe we should do this differently.

We obviously want to show κούροισῐ(ν) and not **κούροις in the inflection table for κούρη. This is non-negotiable in my eyes. But the tables we have are, well, first of all, they're huge. Anywhere from half to two-thirds of the cells have footnotes. There are upwards of a dozen footnotes per adjective, or as many as forty per verb, and keep in mind this is only a summary—I'm generalizing the dialects into Attic, Epic, "Aeolic", and ""Doric"" and simplifying many of their features, while leaving out completely dialects like Boeotian or Arcadian or Cretan. The space for dialectal forms is nearly as tall as (or even taller than) the table itself, even when split into two columns. Second of all, it's an overload of information, information which is probably irrelevant to most viewers, who either (a) don't know Ancient Greek and don't care about dialectology, or (b) are classicists and also don't care about dialectology. Third of all, the program currently puts every marker in by default, which means that even hapax legomena (and, we had a discussion once about this, and I support putting whole tables for them) have full information on what the word would like in the dialects, which is simply not information that we can confidently provide. Plus it means we have to do things like explicitly set dial=~dor/aio for any word that has a non-original eta in it. Every single derivative of δῆμος should not need to be qualified that way.

Therefore, I propose that we get rid of the "dialectal forms" box. We should probably replace it with a note like "For inflection of dialectal forms, see Ancient Greek dialectal inflection." which isn't great wording but you get the idea. —ObsequiousNewt (εἴρηκα|πεποίηκα) 05:00, 16 August 2016 (UTC)

Pings only work when followed by a signature in the same paragraph. Chuck Entz (talk) 02:26, 17 August 2016 (UTC)

Seems like they also have to be added in the same edit as the signature. — Eru·tuon 04:16, 18 August 2016 (UTC)

Let's try this again: User:I'm so meta even this acronym, User:JohnC5, User:Angr, User:Erutuon — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 04:30, 18 August 2016 (UTC)

I don't like the ugly mess of dialectal forms at the bottom of Attic–Koine inflection tables, but it's really cool that Wiktionary actually displays the forms. I don't know of any other website where you can see them. Perseus can recognize dialectal forms, but it doesn't display paradigms (or maybe I'm just not technologically adept to figure out how to do it).

I would rather see the dialectal forms split into separate tables, at least for the dialects that are actually attested in literature. Then you can view coherent paradigms, rather than a jumble of forms.

That would make Wiktionary more useful for someone who's reading Homer or some other work not in Attic or Koine, and wants to figure out what case and number (or person, number, voice, etc.) a form is, or wants to get a feeling for what the paradigm looks like in a particular dialect. If you're reading Homer, you could click on the Epic table and view the inflection.

Maybe that's a tiny fraction of Wiktionary users, though. Hard to know. If so, perhaps the work involved in making sure δῆμος (dêmos) doesn't get Aeolic and Doric endings stuck onto it isn't worth it. I like to be able to see dialectal forms (actually, pretty much just Epic; I don't know Aeolic and Doric), but I'm probably atypical. — Eru·tuon 06:01, 18 August 2016 (UTC)

There's always the option of nested collapsible boxes, so that those who don't want to see the extra content don't have to. — This unsigned comment was added by Chuck Entz (talk • contribs). 13:19, 18 August 2016 (UTC)

I think showing separate tables for different dialects is a good idea. Certain forms always correlate to certain other forms. For example, an Ionic writer will always use forms with ē and never those with ā, so it's not really possible to find one case form with ē next to another case form with ā in the same Ionic text. —CodeCat 14:32, 18 August 2016 (UTC)

@ObsequiousNewt, Chuck Entz, Erutuon, CodeCat: I would also be sad to see all the dialectal variants go, but agree that the current presentation is rather cluttered; Chuck's idea of presenting them in nested collapsible tables is the best solution to this that I've heard so far. — I.S.M.E.T.A. 17:35, 18 August 2016 (UTC)

@I'm so meta even this acronym, Chuck Entz, Erutuon, CodeCat: I like this solution better than what we have now, but I'm not sure I like it enough to be satisfied with it. Pursuant to my third reason, I would prefer to limit it to those dialects that are actually attested, rather than showing everything by default. For example, μοῖρα would have collapsible tables for Epic and Ionic (but not Aeolic?) but τετράμοιρος, being a hapax limited to Attic, wouldn't have any. The problem with this is that figuring out every dialect that something exists in can be difficult (I had to search through a dozen words until I found an example that I could easily check was Attic only) and only changes the form of my third problem with the current situation. There's also the problem that, as I stated in my first reason, there are a lot of dialects I'm leaving out, some of which vary by just one form, so it's wise to keep in mind that there are more than five dialects that we're dealing with... I think Buck has something like two dozen, plus groups ("Doric", "Aeolic", Northwest, Ionic, East, West, etc.) which should ideally be all considered, and this not only increases the amount of space that will be necessary, but also the amount of work that will have to be done when looking for attested dialects. — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 20:21, 18 August 2016 (UTC)

Well, only some of the dialects have literature written in them. Other dialects, I assume, are only found in inscriptions: Arcadocypriot, Cretan, Elean, Pamphylian, Boeotian, Euboean, Northwest Greek, etc. At the minimum, the tables should cover literary dialects. I don't know where one would find information on epigraphical dialects anyway, and what the purpose of covering them would be. — Eru·tuon 20:37, 18 August 2016 (UTC)

Some people study epigraphy. Myself, for example. Here is a database of what is probably most inscriptions, and here is a database of what is probably most papyri. "Literary dialects" are important, to be sure—it's where the Attic/Epic/Ionic/Doric/Aeolic distinction comes from—but I see no reason to say they are more important than non-literary dialects. — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 00:26, 19 August 2016 (UTC)

Okay, so there would be a group of people interested in inscriptions and their inflected forms. I would still say literary dialects are more important in the sense that there are likely to be more people reading Ancient Greek literary works than inscriptions. I wouldn't object to including non-literary dialects. It would be interesting. — Eru·tuon 01:29, 19 August 2016 (UTC)

I think we should have a distinction between dialects that are displayed by default and dialects that are not displayed by default. The former would be automatically populated unless suppressed for that table, while the later could either be only manually populated, or automatically populated only when enabled for that table. Chuck Entz (talk) 01:33, 19 August 2016 (UTC)

@ObsequiousNewt, Erutuon, Chuck Entz: The declension tables have (or, at least, used to have) a footnote saying "Not all forms, especially dialectal forms, are necessarily attested. Use with caution." Does that not mitigate this concern? — I.S.M.E.T.A. 16:50, 19 August 2016 (UTC)

I'm sure there are dialects with small corpora where many words simply aren't attested in any form. I'm not familiar with the workings of the templates, so I could be mistaken, but- in such cases, aren't we showing inflections for words that may have never existed at all in those dialects? After all, it's one thing to apply attested inflectional morphemes to attested roots to produce unattested forms, but applying them to unattested roots risks possibly masking regional distribution patterns. Chuck Entz (talk) 18:04, 19 August 2016 (UTC)

@Chuck Entz: Even for widely attested dialects, it's generally impossible to search the entire corpus to find if a word is attested, which is why I don't think any dialect should be shown by default. — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 02:53, 22 August 2016 (UTC)

Okay, so I feel right now like the best solution is to get rid of showing any dialect by default, but have—honestly, I would prefer just having separate tables rather than trying to do something with Javascript, as it's easiest—for any dialect explicitly mentioned by LSJ. So, to pick a word at random, οὐρανός would have tables for Attic (unmarked), Severe Doric, Bœotian, and Æolic (which I'm inclined to call Lesbian per Buck) and rely on an Appendix:Ancient Greek dialectal declension for any other forms, because even though they exist (e.g. Epidauran), searching for all possible forms is difficult to impossible—I'd have to check at least eleven forms in Packhum, which wouldn't even be exhaustive because of line breaks, *and* there are plenty of texts (e.g. most papyri) that aren't digitized or public or searchable. Is this solution acceptable to everyone? Pinging again, since it's been a month: @I'm so meta even this acronym, JohnC5, Angr, Erutuon, CodeCat, Chuck Entz — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 18:02, 16 September 2016 (UTC) — IFYPFY. — I.S.M.E.T.A. 23:10, 12 October 2016 (UTC)

That is acceptable to me. —JohnC5 18:08, 16 September 2016 (UTC)

Me too. And don't worry too much about obscure dialectal forms. If something attested in a papyrus that isn't digitized or public or searchable, then it isn't "likely that someone would run across it and want to know what it means". —Aɴɢʀ (talk) 18:35, 16 September 2016 (UTC)

I was going to suggest that editors explicitly select which dialects to display, and then the module would generate separate tables for each. The current behavior of selecting which dialects not to display and sticking all but the Attic in a bunch of footnotes is a bit odd. Regarding οὐρανός (ouranós), it should also have Ionic and Epic. — Eru·tuon 19:24, 16 September 2016 (UTC)

@ObsequiousNewt: This is not my favorite solution, but I am content with it, especially because of the excellent and detailed survey of dialectal variation afforded by Appendix:Ancient Greek dialectal declension. One thing, though: how do I get multiple dialects to appear? I was thinking of Δῑδώ (Dīdṓ), for which I need the table to display both the Attic and the Ionic forms; I tried adding a |dial=att/ion parameter, but that didn't work. What's the secret? — I.S.M.E.T.A. 23:30, 12 October 2016 (UTC)

@I'm so meta even this acronym: There currently isn't one, but I think it should be... at least moderately easy to implement. I'll get onto it sometime in the near future, I have a few things on my plate—I'm currently working on rewriting grc-translit to make it more editor-friendly and also probably more accurate, and I might prop up a survey regarding it as soon as I have finished reading through Gignac's grammar of Koinê (and whatever else I find to cross-reference it; if you know any good resources I'd be grateful.) — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 23:46, 12 October 2016 (UTC)

@ObsequiousNewt: Cool. In the meantime, I'll just add a second table, redundant as it might seem. Just holler at me when they can be combined; no rush, of course. I wish I could help re Koinê, but you are definitely the superior of the two of us when it comes to Greeking, so I couldn't even recommend anything. Sorry. — I.S.M.E.T.A. 00:02, 13 October 2016 (UTC)

Most common words

Wiktionary doesn't have a frequency list for Ancient Greek. Is there any place we can get some frequency lists, say for different types of literature: Attic prose, tragedy, Epic poetry, or the Septuagint and New Testament?

Perseus has a Vocabulary Tool, but it doesn't let you select all the AG works on the list. (I tried doing a list from Plato's works, and the output is messy, with many repetitions and some doubtful results. Not sure how to clean it up.) I do have an Attic Greek vocabulary book from the University of St Andrews that I could type up a list from, but that's a lot of work. I generally use the Dickinson Greek Core Vocabulary when trying to come up with example words, or to get some inspiration on which entries to work on, but the site doesn't explain how that list was compiled.

Some lists would be helpful, because they would help me determine which entries most need work, with the goal of helping readers of AG works. (Other somewhat different goals would be creating entries for words that are used as roots for English words or taxonomic names.) — Eru·tuon 06:19, 8 October 2016 (UTC)

Frequency is hard, because inflected forms. Perseus' vocabulary tool does not work like that; it just estimates based on I don't know what. It is 100% unreliable. In terms of what is useful, I'm more inclined to fix existing entries, and add important entries which are missing, than to try to compile frequency-lists. That's the kind of thing I'd want a TLG subscription for anyway, although Perseus does have a fairly good corpus. — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 07:03, 8 October 2016 (UTC)

Ancient Greek has big holes in its attestation, so frequency is a poor indicator of importance. Besides, there are Ancient Greek writings that codify core values of Western Civilization, and others that are corpus filler. A hapax legomenon in the former is a big deal, but almost everything in the latter isn't. Chuck Entz (talk) 15:54, 8 October 2016 (UTC)

Perhaps I'll type up the St Andrews lists then. They probably have some sensible way of selecting the most useful Attic vocabulary. — Eru·tuon 03:41, 25 October 2016 (UTC)

Perseus under PhiloLogic has word frequency tools, but I haven't figured out how to get them for more than one document yet. But you could certainly grab them for the Iliad and Odyssey, and then that's already two huge corpora. In fact, here is a page of Homer. — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 16:29, 25 October 2016 (UTC)

Hi @Erutuon, I just saw this discussion. I have made these types of lists in the past. It's possible to somewhat mitigate the problem with the Perseus lists using a Bayesian filter. The other issue with Perseus is the slowness and brittleness of the interface, especially when using multiple works. I will post something here soon, but it will be preliminary. I was going to do this eventually for the purpose of guiding some LSJ imports I have been contemplating. Isomorphyc (talk) 12:31, 26 October 2016 (UTC)

Here is a preliminary list: User:OrphicBot/Sandbox/Naive Perseus Concordance. It stands to be improved, but it is probably pretty close to what one would get from selecting everything in Perseus. Isomorphyc (talk) 18:20, 26 October 2016 (UTC)

A few more:

* User:OrphicBot/Sandbox/Naive Perseus Concordance - Homer

* User:OrphicBot/Sandbox/Naive Perseus Concordance - Plato

* User:OrphicBot/Sandbox/Naive Perseus Concordance - Attic Drama

* User:OrphicBot/Sandbox/Naive Perseus Concordance - Koine

Isomorphyc (talk) 18:57, 26 October 2016 (UTC)

Concordance:New Testament Greek includes frequencies. —Aɴɢʀ (talk) 14:53, 27 October 2016 (UTC)

@Isomorphyc: Thanks! There are clearly errors in some of the entries in the lists, but it helps to be able to see at a glance which terms have articles and which don't. I've already created a few entries for some of them. — Eru·tuon 19:46, 29 October 2016 (UTC)

Symbol to mark apocope

What symbol do we want to use to mark apocope? I just created δ' and τ' using ' (U+0027 APOSTROPHE), with hard redirects from forms using ’ (U+2019 RIGHT SINGLE QUOTATION MARK) and ᾿ (U+1FBF GREEK PSILI). Then I found ἀλλ᾽ using ᾽ (U+1FBD GREEK KORONIS). So what should we use as the primary form? Whichever we pick, I do think there should be hard redirects from the others. —Aɴɢʀ (talk) 23:30, 25 October 2016 (UTC)

I like the look of the coronis better, but the practice you described elsewhere relating to French, of using the plain apostrophe in entry names and a nicer-looking character in headword templates, would make sense here too. — Eru·tuon 23:34, 25 October 2016 (UTC)

I prefer the coronis. The argument that is usually employed in favour of ' vs. ’ is that the former is a lot easier to type on QWERTY keyboards than the latter; that doesn't really apply to polytonic Greek. I'm happy with redirects from the ' form, of course. Are the psile redirects necessary? Do spellings with it ever occur? — I.S.M.E.T.A. 23:38, 25 October 2016 (UTC)

Well, they occur at User:ObsequiousNewt/freq-hom and at τε#Derived terms, so certainly some people use psili in digital renditions of Ancient Greek. —Aɴɢʀ (talk) 23:49, 25 October 2016 (UTC)

Perhaps we should ask the question of which is most commonly used. I searched my text files of the whole Iliad and Odyssey (since that's the only giant work that I have a text file of), and they have far more ugly regular apostrophes than pretty right single quotation marks (i.e., about 8–10,000 versus fewer than 100). The spacing smooth breathing and spacing coronis did not occur at all. If there are redirects, it doesn't really matter which one we use, but hey, the ugly apostrophe seems to be more common. — Eru·tuon 03:49, 26 October 2016 (UTC)

Pinging ObsequiousNewt for his opinion since he hasn't stated one yet. —Aɴɢʀ (talk) 15:29, 26 October 2016 (UTC)

The character used by Perseus is U+1FBD, the coronis, which oddly doesn't normalize to any other character but does normalize to space + psili (U+0020 U+0313). However, the coronis is properly the symbol used to mark crasis, not elision. The symbol used to mark elision is rather the "apostrophe", which does seem to imply we should be using U+0027. I put coronides on my list because I did not know this. — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 19:44, 26 October 2016 (UTC)

Still, if you look at a modern printed text (e.g. OCT or Loeb), the apostrophe looks identical to the smooth breathing and the coronis, and true crasis always has a nonspacing coronis over a vowel (doesn't it?), not a spacing coronis next to a letter. Since Unicode provides us with a spacing coronis, why not make use of it? Also, I notice Perseus uses U+1FBD GREEK KORONIS as its apostrophe. —Aɴɢʀ (talk) 20:43, 26 October 2016 (UTC)

@Angr, Erutuon, ObsequiousNewt: The derived term «οἷός τ᾿ εἰμί» was added to τε by Erutuon in this revision (in case that matters at all). ObsequiousNewt stated that he “put coronides on (mostly) everything > 100” in User:ObsequiousNewt/freq-hom, although the character that actually occurs there is ᾿ (U+1FBF GREEK PSILI); I think this is because ObsequiousNewt's coronides underwent NFKC normalisation (᾽ [U+1FBD GREEK KORONIS] and ᾿ [U+1FBF GREEK PSILI] both undergo compatibility decomposition to ␠ [U+0020 SPACE] +  ̓ [U+0313 COMBINING COMMA ABOVE]; ␠ +  ̓ then undergoes canonical recomposition to ᾿ [U+1FBF GREEK PSILI]; see w:Unicode equivalence). As Erutuon said “if there are redirects, it doesn't really matter which one we use”, but using a properly-displaying character in the PAGENAME at least means that we don't need to specify that character using |head= parameters. ObsequiousNewt correctly states that the coronis properly marks crasis, not elision; however, crasis is just a kind of contraction, which leads me to regard the analogy of English (and other languages), wherein the apostrophe indicates both omission and contraction, as encouraging the use of the coronis for marking elision as well as crasis. The psile is unsuitable for marking elision because it indicates smooth breathing, whereas some cases of elision involve rough breathing (e.g. ἐπί → ἐφ᾽ and κατά → καθ᾽); the dasia is even less suitable because it indicates rough breathing (whereas most cases of elision involve smooth breathing) and because it isn't even the right shape. To my mind, our choice is between ᾽ (U+1FBD GREEK KORONIS) and ʼ (U+02BC MODIFIER LETTER APOSTROPHE), with the latter arguably representing the better semantic choice (though cf. my argument from the analogy of other languages, above); to my eyes, the former looks right, whilst the latter is the wrong shape. — I.S.M.E.T.A. 22:07, 26 October 2016 (UTC)

Could Module:headword or Module:headword/templates be made to automatically replace a plain apostrophe with coronis? Then we wouldn't have to specify the character in the |head= parameter. — Eru·tuon 22:14, 26 October 2016 (UTC)

Why do we need to use U+02BC and not just U+0027? In fact, I'm pretty sure that U+02BC is meant for glottalization, ejective, etc., it being a modifier letter. — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 23:38, 26 October 2016 (UTC)

I agree that U+02BC is wrong, but U+2019 is an option. I'm beginning more and more to come around to the idea of using U+1FBD GREEK KORONIS, though, because really crasis and apocope are very similar processes. They're both methods of avoiding vowel hiatus across a word boundary, but in one the two vowels merge and in the other the first vowel is deleted. Are there any objections to settling on U+1FBD GREEK KORONIS as the Ancient Greek apostrophe? —Aɴɢʀ (talk) 13:08, 27 October 2016 (UTC)

@Erutuon: Could that functionality also be extended to {{citation}}, {{alter}}, {{quote}}, {{Q}}, {{m}}, {{l}}, {{bor}}, {{inh}}, {{der}}, and {{cog}}? If so, then that could be quite acceptable (although a lot of hassle, IMO).
@ObsequiousNewt: I stand corrected re ʼ (U+02BC MODIFIER LETTER APOSTROPHE).
@Aɴɢʀ: Given ObsequiousNewt’s correction, to my mind our only choice now is to use ᾽ (U+1FBD GREEK KORONIS).
 — I.S.M.E.T.A. 16:55, 27 October 2016 (UTC)

@I'm so meta even this acronym: You should be able to do that just by adding it to the entry_name field in Module:languages/data3/g. DTLHS (talk) 16:58, 27 October 2016 (UTC)

@DTLHS: Doesn't all that just change link text? It needs to change the display text, too. — I.S.M.E.T.A. 17:01, 27 October 2016 (UTC)

@I'm so meta even this acronym: Yeah, ideally the same replacements should happen in all of those templates as in {{head}}. {{m}}, {{l}}, {{alter}}, {{der}}, {{bor}}, {{inh}}, {{cog}} depend on Module:links, so if a function for automatic replacements were built into the module (or one of the modules it uses), those would be taken care of. Not sure about {{Q}}, {{quote}}, {{citation}}. — Eru·tuon 18:08, 27 October 2016 (UTC)

@Erutuon: I would say that {{alter}}, {{Q}}, {{quote}}, and {{citation}} are the ones that most need that functionality since, AFAIK, there is no way of piping which form they display. — I.S.M.E.T.A. 18:13, 27 October 2016 (UTC)

@I'm so meta even this acronym: Hmm, I am confused what we are talking about with regard to {{Q}}, {{quote}}, and {{citation}}, since they are not linking templates. With the linking templates, I mean that Module:links could automatically replace plain apostrophes (') with a better-looking character in the displayed link text, while converting a spacing smooth breathing or spacing coronis with a plain apostrophe when generating the URL. Are you proposing that {{Q}} also automatically display plain apostrophe as a nicer-looking character in Ancient Greek quotations? — Eru·tuon 18:59, 27 October 2016 (UTC)

@Erutuon: Yes, that is what I am proposing. — I.S.M.E.T.A. 21:37, 27 January 2017 (UTC)

I object to using the coronis, because it's simply not correct. It doesn't matter how similar they are; they're not the same process. I don't really think it's that important which apostrophe we really use, I mean, they look the same in most books I've seen, but saying we should use the coronis because they're "similar processes" does not seem like a good reason. — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 17:44, 27 October 2016 (UTC)

@ObsequiousNewt: Is it a reason against using the coronis? I'm confused because of your “I don't really think it's that important which apostrophe we really use” statement. — I.S.M.E.T.A. 18:15, 27 October 2016 (UTC)

My point is, there are more important things at stake than which of several nearly-identical Unicode characters we should be using for the apostrophe, but if there is a most correct character to be found it is not the coronis. — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 19:12, 27 October 2016 (UTC)

Actually, it seems they are the same character: according to w:Smooth breathing#Coronis, the crasis-marking coronis over a letter started out as an apostrophe after the letter. So the nonspacing U+0343 COMBINING GREEK KORONIS is just a development of the apostrophe, suggesting that the spacing U+1FBD GREEK KORONIS is really intended for no other purpose than to be the Ancient Greek apostrophe. —Aɴɢʀ (talk) 07:54, 28 October 2016 (UTC)

That sentence doesn't have a citation, and I'm not sure that "apostrophe" is being used there in the truest sense of the word. Unicode was not always designed with perfect knowledge of Greek in mind, either, so we cannot assume that U+1FBD was intended to be an apostrophe—or, if it was, that such intent was correct. — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 17:03, 28 October 2016 (UTC)

@ObsequiousNewt: When you wrote "I've put coronides on (mostly) everything > 100" at the top of User:ObsequiousNewt/freq-hom, weren't you referring to the symbols at the end of apocopated words, which Unicode converted to psilis? Why do you so strenuously object to using coronides as apostrophes now when you apparently did it (or attempted to do it) yourself there? —Aɴɢʀ (talk) 17:28, 28 October 2016 (UTC)

@Angr: Yes, I put coronides there, because Perseus does. That was before someone raised the question of what symbol should be used, and I looked it up in Smyth, who states that the coronis is used for crasis, but the apostrophe for elision. So that could be considered a mistake on my part. — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 17:32, 28 October 2016 (UTC)

@ObsequiousNewt: So shall we keep using the typewriter apostrophe in page names and the curly apostrophe in the headword line? Since Perseus uses the spacing coronis as an apostrophe, I feel that we should at the very least have a hard redirect from that form, in case people copy and paste over from Perseus (or any other website that follows that convention). —Aɴɢʀ (talk) 18:09, 28 October 2016 (UTC)

Sure. Frankly, if most electronic websites use the coronis, then I'm in favor of using the coronis as well. Honestly, since the apostrophe is often rendered as a vertical line and not a comma—whereas most elision marks I've seen in print resemble a (9-shaped) comma—it might be better to use the coronis anyway. I guess I just wanted to point out that an argument from similarity of meaning, as opposed to form, seems faulty. — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 19:37, 28 October 2016 (UTC)

Well, let's see: Perseus uses the coronis; TLG and el-wikisource use the curly apostrophe. All the online editions of the Greek New Testament I looked at use either the curly apostrophe or the typewriter apostrophe. What other sites are there? —Aɴɢʀ (talk) 20:06, 28 October 2016 (UTC)

One obvious question that no one has thought to ask: if we put one on the head of a pin, will there still be room for angels to dance on it? ... Chuck Entz (talk) 20:39, 28 October 2016 (UTC)

@ObsequiousNewt (and @Isomorphyc, and any other Lua-experienced people), do you have any opinion on the idea of making {{head}} and link templates automatically replace plain apostrophes with a nicer-looking character, while automatically replacing non-plain-apostrophes with plain apostrophe in URLs? This would be useful not only in AG entries, but also French, as I recall there was a conversation in the Beer parlour in which @Angr said that the current practice was for French entry names to use the plain apostrophe, and for French headwords to display the right single quotation mark (curly apostrophe).

I think it is ridiculous to have to manually type the correct character in the |head= parameter of {{headword}}, and the |alt= parameter of link templates; this should be done automatically. But I vaguely realize (though I am not quite able to understand all the Lua in Module:headword and Module:links) that nothing like this yet exists, and I don't know if anyone is able or willing to figure out how to do it and add the appropriate functions. — Eru·tuon 20:00, 29 October 2016 (UTC)

I don't know about the first part, but the second part is easy. within a given language, it's as easy to replace ’ or ᾽ with ' inside linking templates as it is to replace ᾰ and ᾱ with α, which we already do. And even without that, the redirects mean that alt= should never be necessary. —Aɴɢʀ (talk) 20:21, 29 October 2016 (UTC)

Both can be done, and probably the place in Module:headword to make this change is the part of the preprocess() function in which the pagename is taken as a default headword if none is given. We would need to test for Ancient Greek as the language and swap out one punctuation mark for the other. Given the module does not make provision for this kind of prestidigitation, we ought to make sure the regular maintainers of that module do not mind our barnacles here. If every language did this, there would be either a general system or a mess. Isomorphyc (talk) 03:25, 30 October 2016 (UTC)

There is some resistance to this idea. I proposed that something like this be done for French and Ancient Greek at Module:links and @Wikitiki89 objects to it. This is probably the first time it has ever been proposed that a module modify the displayed text that an editor has entered; as of yet, the only replacements relate to entry names. (I could be wrong about this.) It's different, since it would restrict editors from displaying particular characters. But I see no reason why it should not be done, if there is indeed a consensus that Ancient Greek entry names always use the plain apostrophe, but that the single right quotation mark (or another usually comma-like character) always be displayed. If such consensus exists, it's easier to enforce it with modules than to find all AG text with plain apostrophe and do manual replacements, even if it was done by a bot. Anyway, it should certainly not be done for all languages. Editors for a particular language might prefer the plain apostrophe, or curly apostrophe might mess up language-specific uses of the apostrophe.

@CodeCat mentioned that some entry names might use plain apostrophe as a quotation mark. In that case, a replacement would incorrectly replace both the opening and the closing plain apostrophe with a right quotation mark. But she did not give an example and I am not sure if that ever happens with English entries; I know it doesn't happen with Ancient Greek entries.

I have halfheartedly wondered if perhaps it would be useful to have language-specific display-related replacements in the data submodules of Module:languages, distinct from the replacements used to generate entry names. The existing entry name replacements can automatically change curly apostrophes (or coronides or spacing smooth breathings) to plain apostrophes to enforce the rule that entry names should use plain apostrophe, but there would have to be a replacement applying to displayed text to enforce the rule that a curly apostrophe be used in displayed text. Then this could be used in Module:headword and Module:links to generate displayed text, and each language's display could easily be changed if consensus changes. That would be much better than creating separate but basically identical replacement rules in each module, which would have to be changed individually. — Eru·tuon 04:02, 30 October 2016 (UTC)

It will be much easier to have a robot iterate over the 57 Greek entries containing apostrophes, and add head= arguments, than to add and maintain a new feature in the display machinery. However, I think the apostrophe forms should hard redirect to the coronis forms, if editors wants the coronis to be the Greek apostrophe. For even if the head= argument is changed surreptitiously, the large pagename heading will still have an apostrophe in it. Isomorphyc (talk) 05:29, 30 October 2016 (UTC)

The reason I started this thread is that I want to make more entries for apocopated forms, but I feel like we haven't reached any agreement as to what form they should have. Right now, there are 6 entries in Category:Ancient Greek apocopic forms, and they all use the typewriter apostrophe in the page name and the spacing coronis in the headword line. They also all have hard redirects from the curly apostrophe, the spacing coronis, and the spacing psili. Does anyone object to this state of affairs? If not, I'll continue making entries that way. —Aɴɢʀ (talk) 13:42, 30 October 2016 (UTC)

Sorry this became so protracted. From what I understand of the past discussion either settling on apostrophe or coronis for the main form is fine, and hard links as you have done them are good. I don't recommend editing the core modules for this. What you are doing is a little bit arduous. I made a list of `contractions' in Perseus here: User:OrphicBot/Sandbox/Greek_Contractions. It is obviously problematic, but if you wanted to cull it for the forms you wanted, I could make the stubs and hard links if you would like. The forms you have created are so formulaic that, excluding the stems, I don't think you would have to do much except create missing stem entries. If this summary is about right, we only have to decide on whether the main forms should be with the coronis or apostrophe. I think either is good, and unfortunately this conversation was so confusing I am not sure at all what the show of hands looks like. Isomorphyc (talk) 16:45, 30 October 2016 (UTC)

Edit : as should be obvious, the contracted forms are contracted forms of any oblique form of the stem. I do have the intermediate, uncontracted oblique forms too, but I have not added them to the table yet. I realise this makes it a little bit hard to read until I do this. Isomorphyc (talk) 16:51, 30 October 2016 (UTC)

I have emailed the Perseus webmaster to ask why that website uses ᾽ (U+1FBD GREEK KORONIS) to mark apocope in Ancient Greek text and I have posted a question to the Unicode public general mail list asking what the intended use(s) of ᾽ (U+1FBD GREEK KORONIS) is/are. I shall post any information I receive hereto if and when I get a response. If anyone would like a copy of either or both of those messages, please ask, and I shall post them hereto. — I.S.M.E.T.A. 23:23, 30 October 2016 (UTC)

I got a prompt and full reply from Perseus; the answer surprised me:

A technical question concerning the character used to mark apocope in Ancient Greek text

Thanks for writing to Perseus.

I take it you mean elision? Perseus 4 texts are stored in beta code, not Unicode. The source texts therefore use a simple single quotation mark ("straight quote") that has been transformed via browser cookies. Perseus 4 text rendering predates much of the recent Unicode development and is not the best guideline for any current font rendering.
In the process of transforming this beta code into Unicode we have had discussions on areas where the current transformations are inadequate representations.
As noted here, 02BC is the desired apostrophe substitution:
https://github.com/PerseusDL/tei-conversion-tools/wiki/Greek-Betacode-to-Unicode-Transformations
I believe this is based on TLG recommendations
https://www.tlg.uci.edu/encoding/quickbeta.pdf which may shed further light on your questions.

Best,
Lisa

---

Dear Perseus Webmaster,
I am an administator of the English Wiktionary (the lexicographical counterpart to the English Wikipedia): <https://en.wiktionary.org/wiki/Wiktionary:Main_Page>. I and other editors of the project's Ancient Greek entries are currently discussing which Unicode character to use to mark apocope in Ancient Greek text. We have noticed that your website, <http://www.perseus.tufts.edu/hopper/>, uses ⟨᾽⟩ (U+1FBD GREEK KORONIS). Could you explain why this character was chosen for this purpose, as opposed to whatever alternatives that were considered, please? Thank you for your time.

Yours faithfully,
I.S.M.E.T.A.
<https://en.wiktionary.org/wiki/User:I%27m_so_meta_even_this_acronym>

Pinging @Angr, Erutuon, ObsequiousNewt, DTLHS, Chuck Entz, Isomorphyc. — I.S.M.E.T.A. 19:37, 31 October 2016 (UTC)

I'm surprised too, because of all the possible characters to use as an apostrophe, U+02BC is the worst choice. It isn't intended as punctuation at all, but as a letter that happens to look like an apostrophe. Apparently the Perseus people saw that as an advantage, but I don't. Also, in the discussion linked to, they complain that apostrophes have been conflated with the smooth breathing, but what's output at Perseus is not the smooth breathing U+1FBF but the coronis U+1FBD. —Aɴɢʀ (talk) 20:12, 31 October 2016 (UTC)

Thanks for emailing the Perseus people and asking them about this. I agree that the modifier letter is a strange choice. I'm convinced now that the most appropriate character is the right single quotation mark, ’, ever since I read on Wikipedia that it is the preferred character to use as an apostrophe, despite its official name. — Eru·tuon 20:30, 31 October 2016 (UTC)

I'm sorry, guys, but I'm really busy with classes right now, and I'm going to have to withdraw from commenting on this issue. I'll try to stick around to fix the most important Lua bugs but things are going to be going a bit slow for a while. — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 05:21, 1 November 2016 (UTC)

I got responses from two members of the Unicode public general mail list; the most relevant one referred me to the Unicode Standard (v. 9.0.0), chapter 7, § 7.2, subsection “Greek Extended: U+1F00–U+1FFF”, page 310, whose pertinent part reads:

• Spacing Diacritics. Sixteen additional spacing diacritical marks are provided in this character block for use in the representation of polytonic Greek texts. Each has an alternative representation for use with systems that support nonspacing marks. The nonspacing alternatives appear in Table 7-3. The spacing forms are meant for keyboards and pedagogical use and are not to be used in the representation of titlecase words. The compatibility decompositions of these spacing forms consist of the sequence U+0020 space followed by the nonspacing form equivalents shown in Table 7-3.

That seems to prohibit the use of ᾽ (U+1FBD GREEK KORONIS) for the purpose of marking elision. Pinging @Angr, Erutuon, ObsequiousNewt, DTLHS, Chuck Entz, Isomorphyc. — I.S.M.E.T.A. 20:32, 1 November 2016 (UTC)

OK. I'm working on replacing the coronis with the curly apostrophe in heardword lines, quotations, etc. —Aɴɢʀ (talk) 20:37, 1 November 2016 (UTC)

Thank you; this is very useful information. I will correct any errors I have made. Isomorphyc (talk) 22:30, 1 November 2016 (UTC)

@Wikitiki89 noticed my replacing plain apostrophes with curly ones and said that the data modules should be updated to automatically convert curly apostrophes to plain ones when generating entry names for links. I asked him to add the replacement "["..u(0x2019)..u(0x1FBD)..u(0x1FBF).."]" → "'" to Module:languages/data3/g (the three codepoints are for right single quotation mark, spacing smooth breathing, and spacing coronis), since I do not have template editor privileges. I should perhaps have asked here first. I think this replacement wouldn't cause any problems (there are no correct uses of spacing smooth breathing and spacing coronis in Greek words); if someone wants to link to the entries on these characters, they would have to use the mul language code. — Eru·tuon 00:51, 2 November 2016 (UTC)

I was given template editor privileges, so I've added the replacement to the module. Links should bypass the redirects now. — Eru·tuon 22:31, 13 January 2017 (UTC)

Entries for grave-accent forms?

(1) At the moment, we have AFAIK no entries for oxytones where the acute accent has been replaced with a grave accent; rather, the grave-accent version is a hard redirect to the acute-accent version. But I've noticed that German Wiktionary does have separate entries for the grave-accent versions. In other words, while en:δὲ is a hard redirect to en:δέ, de:δὲ is a separate article from de:δέ. Is everyone in agreement that we want to keep doing it this way here at en-wikt, or does anyone think we should have separate articles for words with grave accents?

(2) Whatever we decide to do about grave accents, can we agree that we should do the same for double-accented words that may appear before enclitics? In other words, if we decide to stick with hard redirects for things like δὲ → δέ, can we also have hard redirects for things like ἄνθρωπός (as found in a phrase like ἄνθρωπός ἐστι (“it's a human being”)) → ἄνθρωπος? —Aɴɢʀ (talk) 13:34, 30 October 2016 (UTC)

@Angr: I support hard redirects to oxytones from both barytones and diploxytones. — I.S.M.E.T.A. 18:22, 30 October 2016 (UTC)

@Angr: In favour of hard redirects as well (in both cases). --Fsojic (talk) 09:20, 31 October 2016 (UTC)

Ancient Greek given names

I'm having a hard time figuring out what qualifies as a "given name" in Ancient Greek; but I have the feeling that Category:Ancient Greek given names is widely incomplete, and that many items from Category:Ancient Greek proper nouns should appear there as well. An example: Ἀνδροκλῆς (Androklês) (or even Δημοσθένης (Dēmosthénēs), Ἀριστοτέλης (Aristotélēs), for that matter). --Fsojic (talk) 14:35, 1 November 2016 (UTC)

I think the category is populated by using {{given name}}. In what cases are you unsure as to what qualifies as a given name? —Aɴɢʀ (talk) 17:39, 2 November 2016 (UTC)

@Angr: the ones I mentioned above, for example. --Fsojic (talk) 17:50, 17 November 2016 (UTC)

@Fsojic: I can't think of any reason why those wouldn't be considered given names. —Aɴɢʀ (talk) 21:06, 17 November 2016 (UTC)

@Angr: I guess I'm confused by the fact that (some of) these names almost always refer to a specific, well-known individual. I feel like I have to check that other people than the philosopher are called "Ἀριστοτέλης" to make sure that "Ἀριστοτέλης" is indeed a given name, and not some kind of... well, "proper noun"; I mean, given our current definition of proper noun ("A noun denoting a particular person [...]"), using the header "proper noun" for "given names" is a bit contradictory, don't you think? --Fsojic (talk) 21:26, 17 November 2016 (UTC)

If Aristotle's parents named him Aristotle (rather than it being a family name handed down automatically), then it's a given name, right? As for being a proper noun, I've always been skeptical of the supposed difference between proper nouns and common nouns. I think the distinction is artificial and has no linguistic merit. But as long as we're stuck with the distinction, names are proper nouns, because they're most commonly used to refer to specific individuals. I must know at least 2 dozen guys named David, but I still say "I saw David at the store yesterday", not "I saw a David at the store yesterday". —Aɴɢʀ (talk) 21:33, 17 November 2016 (UTC)

"If Aristotle's parents named him Aristotle (rather than it being a family name handed down automatically), then it's a given name, right?" Exactly; but to assert with confidence that it is indeed the case that his parents named him "Aristotle" (and that it isn't a "nickname"), I think it's a somewhat good clue that there are other ancient "Aristotle". Actually, I didn't pick the best examples: "Πλάτων" seems more problematic to me. --Fsojic (talk) 21:54, 17 November 2016 (UTC)

A nickname fulfills the same function as a given name and can be considered a subset (imo). There are actually multiple Platos as well (check the disambiguation page on wiki), it's a given name. — Kleio (t · c) 22:34, 17 November 2016 (UTC)