Transliteration of Ancient Greek into English

[Anaxandridas ho Skandiates 155]

As pointed out in the other thread, transliterations for ancient Greek units displayed are inconsistent or wrong in many places, or represent late medieval/late byzantine pronounciation.

I will just start with one civ and move on day by day as I work myself through it. Seleucids part one:

Thureoporos Polites = Thyreophoros Polites
Pezakontistes Aravikos = Arabikos
Neos = Naos
Bibliothikon = Bibliotheke (or if you insist, bibliothēkē - but use of ē and ō for Η and Ω should be consistent and uniform, so team please decide and I can help you fix this!

Antiokhos = Antiochos? I can live with "Kh" for "χ" instead of my suggestion "Ch" for "χ".
But let us settle on a variant and standardize it all.

 

I recommend plain e and o instead of ē and ō, and a permanent "Ch" for "χ". But I am perfectly fine with "Kh" and ē and ō.

ē and ō have an educational and accuracy advantage! but then I must add it in the many places where it is missing. As you like.

(also, the 'successor' Greeks should rather have a 'Basileion' building instead of an 'Agora' - where do I suggest adjustments of actual names? help?)

Edited January 3, 2019 by Anaxandridas ho Skandiates

[wowgetoffyourcellphone 6.234]

I would say post a list of units/buildings, their current names, and then your corrections. 

[Itms 2.464]

A reference of the Greek transliteration conventions we use would be this commit from me: https://trac.wildfiregames.com/changeset/16666

Since then, the Seleucids were completed, so there could be a new commit in that fashion. But I still haven't gotten around writing a comprehensive document for transliteration for all the ancient languages in the game I had started it then, but sadly didn't go further than this commit.

On 1/3/2019 at 10:14 PM, Anaxandridas ho Skandiates said:

also, the 'successor' Greeks should rather have a 'Basileion' building instead of an 'Agora' - where do I suggest adjustments of actual names? help?

Are you comfortable enough with software to propose patches?

[Anaxandridas ho Skandiates 155]

On 1/11/2019 at 10:24 PM, Itms said:

A reference of the Greek transliteration conventions we use would be this commit from me: https://trac.wildfiregames.com/changeset/16666

Since then, the Seleucids were completed, so there could be a new commit in that fashion. But I still haven't gotten around writing a comprehensive document for transliteration for all the ancient languages in the game I had started it then, but sadly didn't go further than this commit.

Are you comfortable enough with software to propose patches?

My dear friend, I think that transliteration is not any worse or better than what I proposed. (although the ancients of course used no accents, and I recommend just ê and ô for heta and omega. It would be great to write Greek and Latin words transliterated with Latin capitals at any event.)

Both are fine with me - but the question is, how do we standardize the WHOLE GAME, every Greek word?

The most important thing is, whichever system we use, it is CONSISTENTLY APPLIED. Now different civs have same words spelled differently or have just spelling mistakes. Just little things here and there, as could be expected with such a big enterprise.

 

I cannot at all find my way through your references though. I can only code HTML.
What I can do, is meticulously go through every structure and every unit in all Greek civs, and post the corrections needed here.
But one transliteration-standard must be decided, then I can do the entire lookup work, and maybe you can add to the game?

We need some leading figure to decide on a transliteration-standard, now, so we can do this.

[Stan` 6.885]

If you can edit HTML you can edit XML. Same thing

[Nescio 1.482]

The purpose of transliteration is to be precise.

• Transliteration has to be a bijection:
  • if Ζεύς is transliterated Zeús, then υ has to be an u everywhere else as well, and not an y
  • if e and o represent ε and ο, then η and ω shouldn't be written as e and o (ē/ê/ę and ō/ô/ǫ are fine)
• Accents matter: ὧς and ὡς (Hom ὥς) are two different words
• Be consistent: φ, χ, θ are the aspirated versions of π, κ, τ, so if κ is transliterated c, then χ ought to be ch, and if κ is transcribed k, then χ ought to be kh

[Itms 2.464]

I agree with you Nescio, except on the υ bullet point: ευ is not ε+υ, it is a single grapheme. So deciding to transliterate this grapheme into eu while transliterating the grapheme υ into y would be possible. Same thing for αυ, ου, ει, etc.

[Nescio 1.482]

3 hours ago, Itms said:

I agree with you Nescio, except on the υ bullet point: ευ is not ε+υ, it is a single grapheme. So deciding to transliterate this grapheme into eu while transliterating the grapheme υ into y would be possible. Same thing for αυ, ου, ει, etc.

Not exactly true. A grapheme is the smallest possible unit in a writing system, e.g. a word in a logographic system, a syllable in a syllabary, and a letter in an alphabet (e.g. Greek).

αυ /au̯/ and ευ /eu̯/ were diphthongs (“double sound”), a single syllable consisting of two vowels pronounced together (cf. “cowboy” /ˈkaʊˌbɔɪ/); in contrast, ει /eː/ and ου /oː/ had become monophthongs (“single sound”) in Classical Greek (cf. “voodoo” /ˈvuːduː/). However, none of these has a single character, all are written by two separate letters, thus two graphemes. In contrast, ᾳ, ῃ, and ῳ (initially diphthongs, later monophthongs) are single graphemes (as are e.g. å, æ=ä, ø=ö, and ü).

Although pronunciation matters for one type of transcription (how to convert sounds into signs), it is unimportant for the other type (how to convert one writing system into another).

Edited January 15, 2019 by Nescio

[Nescio 1.482]

Anyway, for the transliteration of Greek, we should discuss and decide upon the following points:

• Do we want to distinguish between ᾰ, ῐ, ῠ and ᾱ, ῑ, ῡ? (Dictionaries do, text editions don't.)
• What to do with the iota subscriptum (ὑπογεγραμμένη): ᾳ, ῃ, ῳ? E.g. Θρᾴκη (Thrace)
• As for the transliteration of η, ω:
  • classicists prefer not to transliterate Greek, because they assume everyone can read it fluently
  • linguists prefer ę, ǫ
  • historians prefer ê, ô
  • Wikipedia prefers ē, ō
• As for the transliteration of υ and ευ, either u and eu, or y and ey
  • in German and Latin, y represents a sound similar to Greek υ, but u is quite different
  • in English and French, u represents a sound similar to Greek υ, but y is quite different
• ρ: r or rh?
• ξ → x, ψ → ps?
• β, π, φ → b, p, ph; γ, κ, χ →g, k, kh; δ, τ, θ → d, t, th; no objections?

Personally I'd recommend ę/ǫ for η/ω, because ê/ô causes problems with accents, and ē/ō suggests η/ω were simply longer versions of ε/ο, which is not true for Classical Greek, while ę/ǫ makes it clear they're different sounds. Also, I think transcriibing ρ as simply r is fine.

[Anaxandridas ho Skandiates 155]

On 1/15/2019 at 5:10 PM, Nescio said:

The purpose of transliteration is to be precise.

• Transliteration has to be a bijection:
  • if Ζεύς is transliterated Zeús, then υ has to be an u everywhere else as well, and not an y
  • if e and o represent ε and ο, then η and ω shouldn't be written as e and o (ē/ê/ę and ō/ô/ǫ are fine)
• Accents matter: ὧς and ὡς (Hom ὥς) are two different words
• Be consistent: φ, χ, θ are the aspirated versions of π, κ, τ, so if κ is transliterated c, then χ ought to be ch, and if κ is transcribed k, then χ ought to be kh

Anyway, for the transliteration of Greek, we should discuss and decide upon the following points:

• Do we want to distinguish between ᾰ, ῐ, ῠ and ᾱ, ῑ, ῡ? (Dictionaries do, text editions don't.)
• What to do with the iota subscriptum (ὑπογεγραμμένη): ᾳ, ῃ, ῳ? E.g. Θρᾴκη (Thrace)
• As for the transliteration of η, ω:
  • classicists prefer not to transliterate Greek, because they assume everyone can read it fluently
  • linguists prefer ę, ǫ
  • historians prefer ê, ô
  • Wikipedia prefers ē, ō
• As for the transliteration of υ and ευ, either u and eu, or y and ey
  • in German and Latin, y represents a sound similar to Greek υ, but u is quite different
  • in English and French, u represents a sound similar to Greek υ, but y is quite different
• ρ: r or rh?
• ξ → x, ψ → ps?
• β, π, φ → b, p, ph; γ, κ, χ →g, k, kh; δ, τ, θ → d, t, th; no objections?

Personally I'd recommend ę/ǫ for η/ω, because ê/ô causes problems with accents, and ē/ō suggests η/ω were simply longer versions of ε/ο, which is not true for Classical Greek, while ę/ǫ makes it clear they're different sounds. Also, I think transcriibing ρ as simply r is fine.

Gentlemen, I propose:

e and o represent ε and ο
ē and ō represent η and ω (I also like ę/ǫ better but TBH the lines are easier to see, let's just continue using them)

b-p-ph represent β-π-φ
g-k-kh represent γ-κ-χ
d-t-th represent δ τ θ

rh represents ρ
x represents ξ
ps represents ψ

y represents υ - but the dipthongs are distinguished by consistently using u :

au represents αυ
eu represents ευ
ou represents ου
ai represents ai
ei represents ει
oi represents οι

the only accents are ´ and ` and ^ (acute, grave and circumflex) The diacritics will then look something like this: ō̂ ē̂ ḗ ṓ ò í 
ng represents γγ ΦΑΛΑΓΓΕΣ = Phálanges (and so forth: γξ = nx, γχ = nkh, γκ = nk)

iota subscriptum always written out: ai=ᾳ ēi=ῃ, ōi=ῳ

[Anaxandridas ho Skandiates 155]

ει and ου did eventually become monophthongs just as @Nescio wrote, the transition phase however stretched over a very long time. For indeed already as archaic diphthongs, spoken fast enough they sound nearly like a monophthong (try it out for yourselves).

[GunChleoc 464]

I don't speak Greek, but visually, ę would look better than ē because ḗ  is hard to read.

Just my 2 cents from the sidelines

[wowgetoffyourcellphone 6.234]

I'm no linguist, but for readability I think it would be best to leave out all of the accents and stuff. So, instead of Hoplítēs Athēnaîos (or whatever it's supposed to be), we have Hoplites Athenaios (Athenian Hoplite). 

[nani 520]

2 hours ago, wowgetoffyourcellphone said:

I'm no linguist, but for readability I think it would be best to leave out all of the accents and stuff. So, instead of Hoplítēs Athēnaîos (or whatever it's supposed to be), we have Hoplites Athenaios (Athenian Hoplite). 

Unlike in English accents can have a very important role in other languages so I would rather remove the least possible of the transliteration.

Edited February 15, 2019 by nani

[wowgetoffyourcellphone 6.234]

32 minutes ago, nani said:

 Unlike in English accents can have a very important role in other languages so I would rather remove the least possible of the transliteration.

Okay then.  Can we get an option to just turn off the specific name? Been asking for this for ages (I think it clutters the UI). 

[Nescio 1.482]

7 hours ago, nani said:

Unlike in English accents can have a very important role in other languages so I would rather remove the least possible of the transliteration.

Exactly!

What is currently needed is someone (@Itms?) to make some actual decisions; see earlier posts.

[Nescio 1.482]

@Itms, the problem is also present in the design document, e.g. https://code.wildfiregames.com/source/design/browse/master/book/gameplay/civs/seleucids.md

• line 13: "Dory."
• line 15: "thureos"

[Nescio 1.482]

Put a proposal at https://trac.wildfiregames.com/wiki/SpecificNames

[Anaxandridas ho Skandiates 155]

I disagree in the strongest terms with Y → u , it is a bad decision, it must be Y → y

Otherwise it looks fine, but @wowgetoffyourcellphone RIGHTLY said that to most people, all the accents are just confusing. I suggested a simplification, you have ignored it and persist in using all the accents, offering no simplification whatsoever.

Are we really sure that we do not just want to stick to three?
´ and ` and ^ (acute, grave and circumflex)

Leaving them out completely as suggested by @wowgetoffyourcellphone is also an option at this point, I just want stalemate of months to END with some kind of decision. But to make Y into u will misrepresent the main pronounciation in the whole period of 0 A.D.

The minuscule Greek "u" for ypsilon belongs to a later age, and the SOUND of u for ypsilon securely  does not belong in the era of the game but is earlier - our Spartans, Athenians, Macedonians, Ptolemies, Seleucids, said 'y' not 'u'.

Edited June 4, 2019 by Anaxandridas ho Skandiates

[Stan` 6.885]

@Nescio what's your take on this ? I can't commit nor accept https://code.wildfiregames.com/D1935 until this is settled.

[Nescio 1.482]

First of all, it's a draft proposal, to have something to work with, not a final decision. It includes diacritics per https://code.wildfiregames.com/D1935#80248

1 hour ago, Stan` said:

@Nescio what's your take on this ? I can't commit nor accept https://code.wildfiregames.com/D1935 until this is settled.

That doesn't involve Greek, only Old Persian.

22 hours ago, Anaxandridas ho Skandiates said:

Are we really sure that we do not just want to stick to three?
´ and ` and ^ (acute, grave and circumflex)

Actually I'm not quite sure what you mean here; Ancient Greek has three accents (rising tone, falling tone, rising-then-falling tone), Modern Greek just one (stress).

22 hours ago, Anaxandridas ho Skandiates said:

The minuscule Greek "u" for ypsilon belongs to a later age

Yes, I'm aware that minuscules are an early Medieval invention (same for Latin); nor did Greek have word-separators yet. However, 0 A.D. is a 21st C game, so we use spaces, punctuation, upper case and lower case distinction, diacritics, modern fonts, etc.

22 hours ago, Anaxandridas ho Skandiates said:

and the SOUND of u for ypsilon securely  does not belong in the era of the game but is earlier - our Spartans, Athenians, Macedonians, Ptolemies, Seleucids, said 'y' not 'u'.

Again, there is a fundamental difference between orthographic transliteration (converting one writing system into another) and phonetic transcription (converting sounds into written signs). How something is pronounced is of fundamental importance for the latter (audio voices in 0 A.D.), but quite irrelevant for the former (written specific names in 0 A.D.).

22 hours ago, Anaxandridas ho Skandiates said:

I disagree in the strongest terms with Y → u , it is a bad decision, it must be Y → y

It doesn't really matter whether we transliterate Υ/υ→Y/y or Υ/υ→U/u, as long as it's done consistently, i.e. map every upsilon to one and the same letter (e.g. ευ→ef, ου→ou, υ→y is understandable but ugly).

[Anaxandridas ho Skandiates 155]

@Stan` @Nescio

Let us take the American Library Association and Library of Congress Ancient and Medieval romanization scheme  of 2010 and be done with this.

https://www.loc.gov/catdir/cpso/romanization/greek.pdf

Exactly as it is in that document (only excepted that we do not need Byzantine ΜΠ and ΝΤ to signify Beta and Delta) :

Flat line over ō and ē as suggested by Nescio previously to denote Ω and Η.
No diacritics.
Y is y, unless with the dipthong vowels where it is au, ou, eu etc.
Gamma as in the guideline PDF:
Γ G
γ g
γγ ng
Γκ Gk
γκ gk (initially and finally)
nk (medially)
γξ nx
γχ nch

Text will then look like in the guideline PDF:

Lētous kai Dios huios

Hēsiodou tou Askraiou Erga kai hēmerai

[Anaxandridas ho Skandiates 155]

I really to not like "huiothesia" for υιοθεσία, or "Harpuies" for Άρπυιες -

But we must compromise, and this is neither my standard nor Nescio's nor anyone else's.

Let us settle on the 2010 guidelines and finally get to work.

[Anaxandridas ho Skandiates 155]

@Itms @Stan` @wowgetoffyourcellphone @Nescio
Can we reach an agreement to standardize, according to the Ancient and Medieval romanization scheme of 2010 used by the American Library Association and Library of Congress?

I am no American and do not agree with all in the standard, but it is bearable for me from a scholarly viewpoint and provides high enough quality that I think we can be well served in it. This has dragged on for a very long time now.

[Stan` 6.885]

I realize I failed to answer here, so I will do it now.

On 6/5/2019 at 1:03 PM, Nescio said:

That doesn't involve Greek, only Old Persian.

I like consistency, so we should not use diacritics nor different letter for Greek and not for old Persian for consistency's sake.

6 hours ago, Anaxandridas ho Skandiates said:

Can we reach an agreement to standardize, according to the Ancient and Medieval romanization scheme of 2010 used by the American Library Association and Library of Congress?

As we already use american English for the game it might make sense to follow american standards.

So here is my take on this:

• I like diacritics and flat lines and whatnot
• I also like that the game displays text in Greek 

However:

• I can totally understand how specific names can be confusing (eg. Kushites)
• The player should not be confused by the interface.

 

So I vote for the American Library's convention.

---

As @wowgetoffyourcellphone said, specific names could be an option. Maybe @Freagarach could submit a patch for that if she wants to. There are not many occurrences of the word so it should be easy to make it togglable 

• https://github.com/0ad/0ad/search?l=JavaScript&q="specific+name"&type=Code
•  https://github.com/0ad/0ad/search?l=JavaScript&q="specificname"&type=Code