Module talk:el-translit

At video, in the second trans-table, it seems to be transliterating as vínteo even though it should be vídeo. —Μετάknowledge^{discuss/deeds} 00:24, 17 November 2013 (UTC)

Transliteration of γκ at the begining of a word should be g, not nk (except if an ν exists before). --Xoristzatziki (talk) 06:12, 28 November 2014 (UTC)

Hopefully not too boldly, I've made a diaeresis be added to the transliteration of α, ε, ο, ω followed by η, because those are pronounced as diphthongs in Modern Greek, but in the transliteration they can't otherwise be distinguished from αι, ει, οι, which are monophthongs. For example, with this change βοήθεια is transliterated as voḯtheia, rather than voítheia, as voítheia could equally represent *βοίθεια (*voítheia), which is pronounced as *βίθια (*víthia). — Eru·tuon 21:16, 21 January 2018 (UTC)

β, γ, δ vs. μπ, γκ, ντ

Some thoughts related to recent changes in the testcases. It is ambiguous to transliterate γκ, ντ as g, d as suggested by this edit by sarri.greek, while also transliterating γ, δ on their own as g, d. Then g or d in the transliteration represents either a stop (/ɡ d/) or a fricative (/ɣ ð/), and there is no way to know which, without looking at the original Greek. The fricative should be represented with a different symbol from the stop. In my cute little script, I use ġ, ḋ for the stops (or nasal–stop sequences), but that was a random decision and probably no one else uses that. There is an additional option: γκ, ντ also represent /ŋɡ nd/ in some words, sometimes reduced to /ɡ d/ by younger speakers, sarri.greek told me.

The easiest and cleanest solution would be to render μπ, γκ, ντ as mp, nk (or gk), nt. Then there's no confusion between γκ, ντ and γ, δ, and no need to distinguish between the two different pronunciations of μπ, γκ, ντ, /mb ŋɡ nd/ and /b ɡ d/. That would make the transliteration less phonemic.

Or editors could manually enter transliterations in order to distinguish between /mb ŋɡ nd/ and /b ɡ d/. For instance, ντ could be transliterated as nd by default, and editors could manually enter a transliteration with d when that is necessary. (Or the other way around if the d pronunciation is more common.) That's assuming that there is no rule that the module can use to determine which way μπ, γκ, ντ are pronounced. (Linking templates only show automatic transliterations for Greek, even if you supply a manual transliteration. This is controlled by the override_translit field in Greek's data table; see Module:languages/data2.)

I noticed this ambiguity a while ago, but didn't post because I don't know what the best solution is. Just rendering the Greek letters directly would be easy to implement, but I do like a more phonemic transliteration system because it's easier to read correctly. But phonemic transliteration probably requires manual input, which is a hassle. — Eru·tuon 07:39, 3 December 2018 (UTC)

Thank you so much @Erutuon for your detailed analysis. I was NOT thinking of phonetics for transliteration, but of the way transliterations are done in passports. Before 1996 δ was dh, etc. But it was dropped. That override_translit sounds great. How is it done? It would be very nice for μπαμπάς (bampás /⁠babás⁠/) with |ts= (the ts= I cannot use in headword lines). And the wrong bampás is always visible and will not go away. sarri.greek (talk) 08:04, 3 December 2018 (UTC)

@sarri.greek: When override_translit is set to true (for Greek, in Module:languages/data2), the automatically generated transliteration is always shown, so if you do {{l|grc|μπαμπάς|tr=this is a test}}, the transliteration generated by Module:el-translit will be displayed after the link, instead of "this is a test". Removing override_translit allows manual transliteration to be shown. — Eru·tuon 06:27, 5 December 2018 (UTC)

Diaeresis for translit of ωι, ωί

The module currently adds a diaeresis for combinations with an eta like αη and οη, which become aï and oï to indicate the correct disyllabic pronunciation. For consistency, the same should be done for omega, specifically ωι, ωί (as seen in πρωί (proḯ) and its derivatives) should be transliterated oï, oḯ to avoid them being read as οι, οί > oi, oí. 86.145.59.95 10:24, 12 January 2021 (UTC)

Manual tr

Sometimes, a manual tr= is desired as in the Etymology of lemma ψεύτρα (pséftra), where a parenthesis mark is not expected in the transliteration of
ψεύτης (pséftis) as {{l|el|ψεύ(της)}}. (2022.05.05). Discussed at Wiktionary:Grease_pit/2022/May#Module:el-translit
|tr=- works, ψεύ(της gives psévtis (expected:pséftis) A manual correction as tr=pséf(tis) is not allowed. Or, a dashes or parentesis mark could be added as 'ignored' in the Module:el-translit at section
text = gsub(text, "([αεηΑΕΗ])([υύ])(.?)",
Thank you, ‑‑Sarri.greek ^♫ I 16:12, 5 May 2022 (UTC)

If we allow manual transliteration, then someone will have to monitor Category:Terms with manual transliterations different from the automated ones/el and remove incorrect or inconsistent manual transliterations. I don't like allowing people to err. Vahag (talk) 16:42, 5 May 2022 (UTC)

Thank you @Erutuon for your improvements! ‑‑Sarri.greek ^♫ I 23:30, 6 May 2022 (UTC)

Unjustified revert

For observers, past discussions are Wiktionary:Beer parlour/2023/September § About the transliteration of Greek initial ντ- and User talk:Catonif § changes to Module:el-translit.

@Benwing2, I am terribly sorry to revert you once again but I do not believe you had any right to do that, note the situation:

1. You claim this change never had consensus. Note the discussion that took place ended with a 6–3 majority for an ISO transLITERAtion or more generally orthography based transliteration, yet note, as I explained more thoroughly on my talk page, that was not even what I did! I did not in any way alter the nature of the transliteration, I merely limited myself to fixing the small yet pressing issue of initial /d/ and /ð/ transliterating the same, and the result of my edit happened to match exactly the ELOT tranSCRIPTion. Anyone has their preferences on orthography vs. phonetic translitera/scriptions, but I do not believe anyone, not even you, would prefer the version of the transliteration scheme before my edit to the one after my edit. It is the same exact scheme, except one has a flaw and the other one doesn't.
2. You claim this change is causing issues as highlighted by Sarri.greek, and this is simply not true. You are assumingly referring to diff. The problem which Sarri faced has nothing to do with my change, actually, it has nothing to do with this module entirely! The transliteration cannot be overridden because of the override_translit = true over at Module:languages/data/2#L-527 and I am fairly sure you know that very well. Moreover, the Leibniz issue is about a medial mp, while my edit only concerned initial nt's. The transliteration of Lampniz would have been the same before my edit, as demonstrated by the fact that the "issue" is still there, you did not fix anything. Whether the transliteration in this case should be overridden at all or not is a separate matter (personally, I do not believe so). This can mean you either did not look at my edit or you did not look at Sarri's, or if you did, it could mean you seized an opportunity to revert the edit and make it look reasoned to by-standers which do not know the situation.

I would rather keep discussing about the issue at hand rather than digressing on the motives that made you revert me, which should normally be irrelevant, but the question naturally arises: why did you revert this? (1) Did you not understand I did not actually change the nature of the translit scheme but merely amended an objective flaw it had? I assume not, since I've already told you in the last talk. (2) Did you not understand that Sarri actually wanted the override_translit to be set to false? Or (3) did you think reverting this change would make the scheme go closer to your ideal phonetic-based transcription? It doesn't, it simply makes it objectively worse. Or finally, (4) do you hold a grudge against me? If so, please don't, let's be friends.

Anyways, I am sorry for being lengthy. I do not like reverting people, and I am ready to change my mind about anything if good reasons to do so are put forward. I wish good both for the project's outcome and for its community, and I am sad to see things like these happen, which are detrimental for both. Happy Women's Day. Catonif (talk) 14:38, 8 March 2024 (UTC)

My mistake, then, M @Catonif -thanks for the 'Happy Women's Day-, and I apologise, M @Benwing2 for my 'ντ as initial' remark (diff). It is true that ΝΤ ντ = Nt nt according to the official ISO843@iso.org.TypeA (unique.coversion=transliteration for machines, customs, etc). I read at the official Greek translation (pdf@sete of ELOT, 2001) that TypeB (the more phonemic one) is allowed to be altered according to the needs of a text, and has footnote about (ντ), plus a special note about foreign names. Unfortunately, all wiktionaries are obliged to provide the TypeA, but can also provide a TypeB more phonemic too. I see that the issue has been debated for a long time. Thank you, and sorry! ‑‑Sarri.greek ^♫ I 16:33, 8 March 2024 (UTC)

I am very sorry Ben I apparently did not read Sarri's edit summary entirely, I now notice it does mentions initial nt... and I even accused you of not examining things properly... yikes, embarassing. 😬 Anyways, your actions make much more sense now and I do not feel threatened by some sort of mighty anger anymore. Again, sorry for my oversight and for the accuses trying to explain your motives. (Point 1 above still stands however, so a revert would still unjustified.) Catonif (talk) 18:02, 8 March 2024 (UTC)

@Catonif Given that you counted Sarri among the "6 in favor" and she appears to now be opposed, I don't see how 6-3 in favor makes sense any more (in any case this is barely 2/3 majority with significant opposition, not something I'd count as consensus). As for whether I'd prefer the previous scheme, yes I would, with override_translit turned off so it's possible to fix the cases where the translit generates wrongly. Benwing2 (talk) 20:39, 8 March 2024 (UTC)

Translit el needs update

Notifying MM @Benwing2, Catonif, Erutuon, Saltmarsh, about discussion on transliterations going on at Wiktionary:Beer parlour/2024/March#Two transliterations.
According to ISO843 TypeA = ELOT743 TypeA (pdf.Table1), our translit at Module:el-translit needs update with i macron for eta (η), o macron for omega (ω). Thank you. ‑‑Sarri.greek ^♫ I 23:32, 10 March 2024 (UTC)