Wiktionary talk:Greek transliteration


 * Moved from Wiktionary talk:About Greek/Transliteration-new— Saltmarsh απάντηση 14:30, 1 October 2009 (UTC)
 * And inserted initial talk from Appendix talk:Greek transliteration/Old &mdash; Saltmarshαπάντηση 05:04, 31 August 2012 (UTC)

It is suggested that any talk about the transliteration/romanisation of Modern Greek should be carried out here. The initial ground rule being that we should try to emulate the tables shown at About_Greek/Transliteration, but this ground rule can also be discussed!
 * &mdash;Saltmarsh 14:43, 3 August 2007 (UTC)

Distinguishing between ο and ω
Some editors distinguish between ε and η by transliterating the latter with a macron, as “ē”. Should we similarly distinguish between ο and ω by writing the latter with a macron, i.e. as “ō”? Rod (A. Smith) 19:31, 30 September 2007 (UTC)


 * Yes, the less ambiguity, the better. † Raifʻhār Doremítzwr 02:28, 1 October 2007 (UTC)


 * Although I can remember seeing it done (I forget where) none of the 'Authorities' quoted on this project page transliterate to a 'ē', 'ō', etc. When I put the table together it seemed sensible to use and cite other authorities. Does this need deeper study? — Saltmarsh απάντηση 14:32, 27 August 2009 (UTC)

Stress marks
Should Greek vowels with stress marks (i.e. ά, έ, ή, ί, ό, ύ, and ώ) be transliterated into Latin (roman) script vowels with the corresponding diacritics (i.e. á, é, ḗ, í, ó, ý/ú, and ó/ṓ)? Rod (A. Smith) 19:32, 30 September 2007 (UTC)


 * As hereinbefore: yes, the less ambiguity, the better. † Raifʻhār Doremítzwr 02:28, 1 October 2007 (UTC)


 * Although I can remember seeing it done (I forget where) none of the 'Authorities' quoted on this project page transliterate to a 'ē', 'ō', etc. When I put the table together it seemed sensible to use and cite other authorities. Does this need deeper study? — Saltmarsh απάντηση 14:32, 27 August 2009 (UTC)


 * (I now recall) The table does not mention stress marks - their transfer into the Roman script was assumed. — Saltmarsh απάντηση 05:54, 3 September 2009 (UTC)

Move to Appendix:Greek transliteration
This article is referenced, and the wiktionary system appears to be closely based on an established system, adding only clarifications rather than modifying it.

Is it safe to move this to Appendix:Greek transliteration in category:Transliteration appendices? —Michael Z. 20:31, 12 April 2008 (UTC)

Distinguishing between η/ι and ο/ω
I think that I have taken care of any other points raised on the old discussion page Wiktionary talk:About Greek/Transliteration. The question was should ISO843 and ELOT743 be ignored and an alternative used:

Is there a consensus? — Saltmarsh απάντηση 06:09, 3 September 2009 (UTC)


 * Who is going to add romanizations in the future, bots or users? If the latter, I'd vote for simplicity (i,o). I've already found it difficult to type accented characters, especially when I add "assisted" translations and haven't the help of edittools. On the other hand, a bot could easily (?) create transliterations and save us all that trouble. In that case, ō could be used. But still, I'd think of ē as "unnatural". Maps and road signs always have i. --flyax 07:38, 3 September 2009 (UTC)


 * I haven't asked User:Conrad.Irwin this question; but my preference is that (in future) all transliterations should use the automatic Transliterator, which would remove the problem of later changes to the mapping of characters between alphabets. Bots are outside my knowledge - I don't know what complications they would be capable of dealing with. Some transliteration of existing Greek entries is idiosyncratic so automation would help! I am talking to User:Rodasmith and User:Doremítzwr this. — Saltmarsh απάντηση 14:01, 4 September 2009 (UTC)


 * Wow. The transliterator seems like a great tool.  Thanks for putting it together.  My preference would be for ō and ē, but if that preference somehow impedes adoption of the transliterator, I'd be happy to set that preference aside.  —Rod (A. Smith) 15:55, 4 September 2009 (UTC)


 * It will be a 5 minute job (I think) to make any changes. (I am off line soon - til 22 Sep) — Saltmarsh απάντηση 09:16, 5 September 2009 (UTC)


 * I suppose that we will not be able to prohibit humans from entering transliterations in the future (nor should we try to). While it would be a matter of a few seconds for a bot to change them, I'd still rather see us move towards something simpler to enter than more complicated (with fewer diacritics rather than more). And my preference is also for relying on the standard (which should produce results more familiar to people who work with the language) as opposed to ignoring it.  What did you have in mind with the proposal of the alterative vowel forms given above? -- ArielGlenn 14:02, 5 September 2009 (UTC)


 * My primary concern is that this transliteration scheme be as bijective as possible. As for the specific alternatives proposed above, I’d prefer <ē> for <η> and <ō> for <ω>, chiefly to keep the scheme more in line with the one for Ancient Greek. IMO, the opposition to diacritics on the grounds that it would make it more difficult for users to enter transcriptions is a red herring, given that the transliterator allows the process to be automated and that a user must already enter diacritics, namely acute accents, diæreses, and combinations thereof (e.g., <ḯ> and <ÿ́>), even if macra are eschewed. However we represent those η–ι and ω–ο distinctions, the importance is that they be made; I’d more than likely be fine with whichever scheme we ultimately adopt, provided it preserve bijection.  †  ﴾(u):Raifʻhār (t):Doremítzwr﴿ 02:19, 9 September 2009 (UTC)


 * i. And what do you suggest that we do with μπ? What about ευ=εφ=ef and ευ=εβ=ev? No transliteration system can be fully reversible and I don't see why it should be. It's not a new kind of alphabet for the Greek language, it's just for help. Imagine that you're in Greece and looking for the road to Κορώνη. You'll never get there if you expect to see Korōnē anywhere :P. Road signs have Koroni.
 * ii. Please, see my list of contributions. You'll discover that most of my assisted translations have no diacritics. Thank you. --flyax 06:03, 9 September 2009 (UTC)


 * (In re this: see .) I’m not insisting that it be “fully reversible”, only as bijective as practically possible; do not let the perfect be the enemy of the good. If undefined: for is problematic, let’s use undefined: instead. If, in the assisted addition of translations’ transcriptions, you have no access to the edit tools, then that is a problem that needs fixing; I remember seeing a box of edit tools beneath the left-hand–side search box — I’m sure something similar could be done for assisted translations. Of supplementary concern to the bijection issue is whether any of those homographic transliterations obscure any phonemic distinctions; do they?  †  ﴾(u):Raifʻhār (t):Doremítzwr﴿ 13:38, 9 September 2009 (UTC)


 * red herring: A clue that is misleading or that has been falsified, intended to divert attention.
 * So, do you think that my intention was to divert attention, consequently do you assume that I didn't argue in good faith? --flyax 15:05, 9 September 2009 (UTC)


 * Say wha’?! Please, chill; I meant no such thing. I gave you the link because I interpreted your comment as you not knowing what the term meant. Perhaps I used it incorrectly; if so, I apologise for the misunderstanding.  †  ﴾(u):Raifʻhār (t):Doremítzwr﴿ 21:49, 9 September 2009 (UTC)


 * (After edit conflict, *grumble*) Why is using a variant of the system with partial bijectivity preferable to using a variant of the system which is in widespread use and hence familiar to people with some exposure to romanized terms in the language? As to diacritics being hard for the user to enter, I was saying above that I would prefer to move to a system with fewer such markings rather than more if we are going to make changes at this point. Not only is this approach in much more widespread use, but as things currently stand, I currently am forced to use a script at the command line to generate transcriptions, and I don't like to have to do so.  -- ArielGlenn 15:29, 9 September 2009 (UTC)


 * Phonetic representation and familiarity are, of course, also relevant considerations. As I said, if it is suspected that transcribing <η> as <ē> would cause confusion because it’s usually transcribed &lt;i>, let’s use the far more similar-to-the-familiar <ī> instead. As for the rest: “I currently am forced to use a script at the command line to generate transcriptions” — please explain, since I don’t understand what you mean.  †  ﴾(u):Raifʻhār (t):Doremítzwr﴿ 21:49, 9 September 2009 (UTC)


 * I mean that, while I much prefer typing plain text in the same way that I am typing this message, in order to enter translations I can't do that. I have a script I run at the command line that will produce a transliteration given the Greek word, and I have been using that since the adoption of the system.  I could hunt and peck to find the characters among those in the editttools, but that is very tedious.  Anyone who adds more than a few such entries is going to want to be able to type them using one of their standard keyboard layouts... at least, I certainly do. -- ArielGlenn 14:45, 10 September 2009 (UTC)


 * I can understand that, but doesn’t the automated transliterator obviate the need for such manual input?  †  ﴾(u):Raifʻhār (t):Doremítzwr﴿ 14:57, 10 September 2009 (UTC)


 * That is where this discussion started :-) I don't know how it's meant to interact (or not) with the user.  But I do think it's unwise to rule out manual editing of information. -- ArielGlenn 14:57, 12 September 2009 (UTC)


 * Pulling it together (I hope these summaries are fair)
 * 1 (flyax) - Who does it ?
 * 2 (flyax) - I'd vote for simplicity (i,o) - but if auto translit. then ō although I think of ē as "unnatural". Maps and road signs always have i
 * 3 (RodASmith) - My preference would be for ō and ē
 * 4 (ArielGlenn) - I'd still rather see us move towards something simpler to enter than more complicated (with fewer diacritics rather than more)
 * 5 (Raifʻhār (t):Doremítzwr) - I’d prefer <ē> and <ō>
 * 6 (ArielGlenn) - reply ... ... ...
 * 7 (Raifʻhār (t):Doremítzwr) - compromise <ī> and <ō>
 * All current templates will need modifying to automatically translate the relevant word, but each should have the option of allowing users to enter manual transliterations - but I believe that these should be strongly discouraged, we should aim for consistancy. Would everyone be more or less happy with the suggested compromise of <ī> for <η> and <ō> for <ω>. Once the Transliterator is installed we will at least have an easy way of making change of mind "global" with minimum effort. — Saltmarsh απάντηση 17:31, 22 September 2009 (UTC)


 * Support the compromise of <ī> for <η> and <ō> for <ω>. —Rod (A. Smith) 19:14, 22 September 2009 (UTC)


 * Support per Rod A. Smith.  †  ﴾(u):Raifʻhār (t):Doremítzwr﴿ 19:45, 22 September 2009 (UTC)

ευφ, ευβ
What about Εύβοια and ευφυΐα; Are they transliterated as Evoia or Evvoia, efyia or effyia? --flyax 07:26, 9 September 2009 (UTC)
 * They transliterate as follows Εύβοια > Évvoia and ευφυΐα > effyḯa two tourist maps to hand have (for Εύβοια) Evia and Evvoia, Collins complete English &amp; Philips world atlas (1972) have Évvoia "a transliteration", Webster's new world college &amp; Britannica atlas (1950) have Evvoia — Saltmarsh απάντηση 16:36, 22 September 2009 (UTC)

ISO843/ELOT743
Greetings,

Am I correct in my undestanding that ISO/ELOT is used as a base for the transliteration rules to be used? It looks to me that when you are referring to ISO/ELOT, you are NOT using the values from the transliteration (type 1) tables but from the transcription (type 2) tables. If that is intentional please ignore me, however if what you wanted in the first place was the transliteration as defined by ISO 843 then a couple of assumptions I read here are wrong:
 * ISO already specifies ī,ō to be the equivalent of η,ω
 * ISO 843 transliteration (and not transcription) is FULLY reversible. ANY Greek text when transliterated as defined in ISO 843, can be transliterated back to Greek without any loss of orthography, without human intervention. I have heard of at least one implementation able to do just that: grconv by Diomidis D. Spinellis

ISO/ELOT doesn't specify when to use transliteration or transcription, however the goals of the two methods are very clear: Type 2 (transcription) is supposed to be phonologically accurate, and nothing else. Type 1 is designed to preserve the etymology of the word as much as possible, being usable by non-Greek speakers, using the least conversion rules possible, and at the same time any Greek speaker should be able to reproduce the original orthography without effort.

IMHO you share a much greater overlap of goals with the type 1 conversion method, rather than type 2, which is why it looks so strange to me that you are using type 2 as a reference. I suspect this is because you are using the PDF from Eesti Keele Instituut as a reference rather than the original ELOT spec. Unfortunately this pdf has a few errors on Type 1, and it doesn't really explain what each type really is designed for.

About Évvoia: geographical places, and peoples names are usually transcribed phonetically using type 2. But I don't think this should affect your decisions on how you transliterate actual words, because you don't want to convert those phonetically, but keeping orthography and etymology is more important. Ευφυΐα should be just eufyḯa and not effyḯa because the latter is not a transliteration but a phonetical transcription.

I hope any of this doesn't seem too intrusive or anything. It wasn't clear to me if this information was available to you when making your decisions, but in any case I hope my feedback can be useful to you, even if just a little.

Cheers