Wikipedia talk:WikiProject Ethiopia/Transliteration/Archive 1

From Wikipedia, the free encyclopedia

Archive This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.

Contents

Vowels

I've summarized what I think are the main issues in transliteration (what is called "naming conventions" in the other Wikipedia language communities with non-roman scripts) of Ge'ez/Ethiopic/fidel, some of the current systems, and a proposal for a standard on this page. I think deciding on the vowels and deciding on the consonants are pretty much independent issues. For the vowels, of course the main issue is how to map the seven Ge'ez vowels onto roman symbols (though there are a bunch of minor issues to be decided too). Here's a summary from my page of the existing systems I'm aware of, with the vowels in the traditional order and the l series in fidel at the top to illustrate. The symbols in parentheses in the Leslau row are what he uses in his Ge'ez dictionary. (Note that though Ethiopia has apparently adopted the UNEGN system for maps, it is the USBGN system that is most often used, according to the UN.) My proposal is to use the USBGN system, with the option of leaving off the macrons in the third and fifth forms when some ambiguity can be tolerated.

Ethiopian Semitic Vowel Transcription
ለ [ɐ,ǝ] ሉ [u] ሊ [i] ላ [a] ሌ [e] ል [ɨ] ሎ [o]
Wolf Leslau ä(a) u i a(ā) e ǝ o
Library of Congress
American Library Association		 a u i ā é e o
UN Group of Exports on Geographical Names e u i a e i o
US Board on Geographic Names e u ī a ē i o
F. Praetorius, Die amharische Sprache ă u i ā ē ĕ o
M. Cohen, Traité de langue amharique ɑ/ä/å u i a (ʸ)e ə/ŭ ʷo
Alone-Stokes a u ī ā ē i o
  • Franz Praetorius, Die amharische Sprache, Halle: Verlag der Buchhandlung des Waisenhauses 1879
  • Marcel Cohen, Traité de langue amharique, Paris 1936
  • Major J. P. H. M. Alone & D. E. Stokes, The Alone-Stokes Short Manual of the Amharic Language (with Vocabularies), 5th ed., Madras-Bombay-Calcutta-London: Macmillan and Co. Limited 1962 (1st ed.: 1909)

I wanted to add that, as someone who really believes in Wikipedia and its future, I think the decisions we make about transliteration of fidel could end up being universal. — MikeG (talk) 02:52, 5 June 2006 (UTC)

Just a quick pointer to those who might be confused (as I was for a second), but that "/" (slash) in front of series is actually the letter L italicized (i.e. the examples above are of the "L series") — ዮም (Yom) | contribs • Talk 03:14, 5 June 2006 (UTC)

1st and 6th vowels

Let me focus this a bit more and suggest we start with the two vowels that are the most common (at least in Amharic, Tigrinya, and Ge'ez), the first and sixth ones in the traditional order. However we choose to represent these vowels, I feel pretty strongly that because they are so common, the symbols should not have diacritics and should be part of the simplest fonts. This is the big problem I have with Leslau's system; he's got ä and ǝ for these vowels. There are other problems with these choices. First ä is not a standard phonetic symbol at all, and in alphabets where it's used (German, Hungarian, Finnish) it represents a completely different sound. Second the symbol ǝ conventionally represents a sound that's closer to the other (first) vowel. In fact some linguists use ǝ for this vowel rather than the sixth. If we choose to use simple unadorned roman characters for the first and sixth vowels, I think there are two possibilities, each with its own advantages: 1:e, 6:i (what I'm proposing) and 1:a, 6:e. Here they are in a table.

ES Vowels 1 and 6
ለ [ɐ,ǝ] ል [ɨ]
Leslau ä ǝ
USBGN e i
a e

— MikeG (talk) 13:54, 5 June 2006 (UTC)

I think the problem is that these two vowels in Amharic constitute two phonemes and not two phones. Each of these two has three allophones: one neutral (ɐ, ɨ), one in the proximity of a palatalized consonant (ɛ, ĭ), and one next to a labialized/rounded consonant (ɔ, ŭ). The question is whether the transliteration should show the phonemic, or the phonetic shape of these vowels.
Choosing to transliterate the phonemic shape means concealing the actual sounds from a reader who doesn't know the allophonic rules of Amharic. Think about transliterations such as: "wiha" for ውሃ, or "wend" for ወንድ. The letters "i" and "e" point to the fronted allophones, where the actual pronunciation is [wŭha] and [wɔnd].
On the other hand, choosing a more phonetic transliteration requires adopting more symbols, and thus, using diacritics. yhever 22:16, 6 June 2006 (UTC)
I don't think we should go for separate symbols for allophones. For one thing, that makes it harder for native speakers to transliterate their own languages. For another, there's little precedent for doing this in transliteration of other languages (though it does happen for Japanese consonants in one of the systems). The goal, I think, should be for people who know the languages (and who aren't linguists) to be able to transliterate them easily and for people who know nothing about the languages to have a rough idea of how the words sound. Yes, people may be led a bit off course with wiha or weha (depending on which symbol we choose), but not as much as they are with i in the standard (pinyin) transliteration of Mandarin Chinese, which differs much more than these two Amharic phonemes do. -- MikeG (talk) 01:36, 7 June 2006 (UTC)
To be honest, yhever, I didn't even realize that the sadis form of "w" in wiha was pronounced differently than in other texts until you pointed it out. While this level of detail is good for noting vowel variations and the like in linguistic articles, it's too specific for general articles. For one, as Mike pointed out, it makes correct transliteration too difficult even for native speakers, whereas systems based simply on the Ge'ez letters can be transliterated by anyone who knows the alphabet, whether or not they know the language (or they could just confer a table in an article to identify the form and transliterate).
Right now, I'm leaning towards USBGN for all articles, with concurrent use of IPA in linguistic articles for specificity and standardness. Right now I feel as if it's too few people contributing to create consensus, so I'll let more discussion occur before adding any proposal (i.e. to be voted on, not meaning it's already adopted) to the project page. Consonants shouldn't be too much of a problem, I don't think.
ዮም (Yom) | contribs • Talk 04:56, 7 June 2006 (UTC)
There hasn't been any discussion for a few days, so I'm just going to add my proposal now, though there need not be any voting yet. — ዮም (Yom) | contribs • Talk 22:55, 11 June 2006 (UTC)
I think my last attempt to explain my proposal was wrongly put. Let me try again...
The Ethiopian script distinguishes 7 vowels. I think it is agreed that as a minimum, a transliteration should show all 7. The problem is that these vowels are principally differentiated by quality, and not by quantity, so that 7 different symbols should be devised. Now it seems that the usage of diacritics in order to express these vowels is less preferable for readers who are non-linguists (e.g. 1:ɐ, 2:u, 3:i, 4:a, 5:e, 6:ɨ, 7:o), so a different system is needed.
In fact, these vowels are distinguished not only by quality, but also by length (though this distinction is secondary). This is somewhat like the difference in English between the vowels of bit and seen, or soon and put (cf. H. J. Polotsky, Notes on Gurage grammar, Jerusalem 1951, p. 16 [=Collected Papers by H. J. Polotsky, Jerusalem 1971, p. 531]). Anyhow, The distinction of length can be marked instead of that of quality, and thus we can have: 1:a, 2:ū, 3:ī, 4:ā, 5:ē, 6:i, 7:ō, which is my proposal.
My point is that if a macron is to be used anyway (as is the case with USBGN), then it should be used consistently, marking all long vowels. This suggestion is actually the mirror image of UNEGN that adds a diacritic mark to the short vowels rather than to the long ones.
Now, the two short vowels (the 1st and 6th "orders") are known to be much more influenced by the environment, having front and rounded allophones. My suggestion is to use a and i for their "neutral" allophones, but to allow the usage of "e" and "o" (without macrons) as the front and rounded allophones of "a", and also to allow "u" (again, without a macron) as the rounded allophone of "i". yhever 11:51, 12 June 2006 (UTC)
I'm aware of the allophone "u" (i.e. short u) for sadis (sixth) order "i," but could you give me an example of where ge'ez (first) order "a" (Leslau ä) has the allophone of "e" or "o"? — ዮም (Yom) | contribs • Talk 16:45, 12 June 2006 (UTC)
Sure. In fact this is sometimes even expressed in writing. Consider the pairs ወፍ ~ ዎፍ (both representing wof, or in IPA: [wɔf]) and: ሰምቸ ~ ሰምቼ (both representing samicche, or in IPA: [sɐmɨččɛ]). The vowels [ɔ] and [ɛ] (that Leslau transcribes as [å] and [ä]) are two allophones of the 1st order vowel.
Additional examples: ጨሰ ~ ጬሰ [č'ɛsɐ], መቸ ~ መቼ [mɐčɛ], በኋላ (sometimes pronounced: [bɔhala]), ነው [nɐw ~ nɔw], etc. yhever 18:51, 12 June 2006 (UTC)
Hmm. Well, wef is certainly pronounced wef (using USBGN) in my family, behwala the same, and new as new (CH'es is properly CH'is or CH'iis for smoke, unless you're talking about the Menz dialectal for Qes 'priest,' which is certainly regional), but meche is pronounced meché, and semichché as semichché. Perhaps it is a dialectal thing, in which case it wouldn't be proper to note the differences unless in a linguistic article noting the differences in dialects. — ዮም (Yom) | contribs • Talk 19:34, 12 June 2006 (UTC)
I was referring to the verb ጨሰ [č'ɛsɐ] (with its variant ጤሰ [t'esɐ]). I am well aware of the differences between dialects, and I would like to remind you that I have suggested e and o only as optional variants.
My main point was that long vowels should all have a macron and short vowels should be left without a macron, in order to preserve consistency.
By the way, the allophones e and o can also be demonstrated by the way the word ወንድ is transliterated in the Seleda online magazine: compare "wend lij" in this link and "lela wond alwodedhu" in this link (the same applies to the o in "alwodedhu"). Compare also weizero and woizero. yhever 21:10, 12 June 2006 (UTC)
I didn't mean to imply that you didn't know the differences between dialects (being a linguist I'm sure you do), it's just that my computer screen doesn't have good resolution and the Ge'ez font I'm using isn't good at differentiation between the ge'ez first and sadis sixth orders. Either way, I don't really have a problem with putting macrons on "o" and "e" (the latter already included), but why a macron on "ā" and first order as "a," given that the distinction is more of quality than of length, as well as the resultant inaccuracy in pronounciation by English speakers. We're making a transliteration system, but it has to take into consideration closeness to English transcription as well, as the transliteration will be used for some article titles. Your system with "e" and "a" for "a" and "ā," would be preferable to me. The latter shows better the connections to Proto-Semitic and other Semitic languages, but it doesn't reflect an accurate pronounciation for English speakers. — ዮም (Yom) | contribs • Talk 21:39, 12 June 2006 (UTC)
Ok. I am willing to compromise. There's nothing inherently wrong with using e for the 1st order vowel, as long as it is kept different from the 5th order vowel ē. I thought ā would be better than a, because it is more consistent with the other long vowels, but it isn't really that important.
ō (the 7th order vowel) and ū (the 2nd order vowel) should have their macrons so that they won't be confused with the rounded allophones of e and i.
So my suggestion is the following: 1:e (or o after w), 2:ū, 3:ī, 4:a (or ā), 5:ē, 6:i (or u after w), 7:ō. yhever 22:09, 12 June 2006 (UTC)
I don't actually care much between "a" and "ā," though I prefer the former out of simplicity, but is "o" necessary after "w"? I realize that it's often transcribed that way to better approximate the pronounciation, but it's not always the case, and may (?) in some cases depend on dialect (I'm certain of my family's pronounciation of ወፍ as Leslau "wäf," and asked them to say it to make sure). For example, take the verb to know, mawäq (Leslau; maweq USBGN, Ge'ez ማወቅ). I'm not sure if I've seen it transcribed/transliterated before, but the pronounciation is not mawoq. Furthermore, although wäld/weld (Leslau/USBGN), Ge'ez ወልድ is transcribed as wold (and may be pronounced that way by some?), I have difficulty imaginging someone saying wollädäch/wolledech (L/USBGN) instead of welledech. I don't have much problems with "u" for "i" after "w," but I don't think that "o" for "e" after "w" is universal, and should be restricted to linguistic articles and in dialectal context if appropriate. — ዮም (Yom) | contribs • Talk 22:28, 12 June 2006 (UTC)
I have already said that I suggest the usage of o for the 1st order vowel only as an optional variant. I have no problem with transcriptions such as welledech or wef. On the other hand, it should be kept in mind that native Amharic speakers transcribed ወይዘሮ, ወንድ and አልወደድሁ in the Seleda magazine as woizero, wond and alwodedhu, and they certainly did not do it in order to help me prove my point.
You are welcome to listen to Konjit Taye's news broadcast "Kezihem Keziam" from last Saturday. Around [00:54] you can hear her pronounce እንደ፡ወትሮው as [ɨndɐ-wɔtro(w)] (by the way, do you hear a final ው there?). At [01:12] she says [bɔhala] for በኋላ. These expressions may be transcribed as inde-wotrōw and bohālā -- but also as inde-wetrōw and behwālā. I have suggested allowing these variants, since some native speakers may find it more natural; others will stick to e.
And one more thing: the name of the singer አስቴር አወቀ is transcribed sometimes as "Aster Awoke" and sometimes as "Aster Aweke". You can check in google. yhever 00:49, 13 June 2006 (UTC)
That's very interesting. I didn't know that the use of [ɔ] for first form was so prevalent after "w." If it's simply an optional variant, then I don't have a problem. While some of the pronounciation is certainly actual conditioning of the vowel by its environment, some of the transliteration might simply be because it sounds intermediate. Maybe it's just me, but hearing those words with first order vowels after "w" as clear seventh order vowels sounds forced and awkward, but a vowel inbetween a full "o" and "e" (not sure of the IPA values) sounds perfectly normal. Either way, I support your proposal — ዮም (Yom) | contribs • Talk 01:02, 13 June 2006 (UTC)
Thank you. Let's wait and see what Mike Gasser has to say about it.
I would only like to add that o is not meant to denote the 7th order vowel (which is transcribed as ō) but rather that "intermediate" vowel that's conditioned by its environment in some dialects. yhever 01:05, 13 June 2006 (UTC)
I understand that, sorry. I just wasn't thinking about using diacritics. Is [ɔ] the IPA value for both the "intermediate" vowel and ō, or what would be the IPA values for these? Of course, we'll need Mike's opinion, but I'd also like to wait for at least Llywrch's opinion as a non-linguist & non-Ethiopian language speaker (the majority of our audience) before we make any decisions. Codex Sinaiticus's opinion would be helpful also, but I'm not sure if he'll be contributing to the discussion any time soon. — ዮም (Yom) | contribs • Talk 01:31, 13 June 2006 (UTC)
I used [ɔ] for the "intermediate vowel" (or more precisely: the rounded allophone of /ɐ/) and I think it is the correct IPA symbol, though Leslau uses å (Following his teacher Marcel Cohen). The 7th order vowel is [o] in IPA, but usually it take an on-glide and is pronounced [ʷo].
The rounded allophone of the 6th order vowel is transcribed by Marcel Cohen and Leslau as ŭ, and I'm not sure what's the correct IPA symbol (perhaps: [ʊ]).
The fronted (or palatalized, that is: near y or a palatalized consonant) allophone of /ɐ/ is [ɛ], though it is transcribed with ä by Leslau and M. Cohen, though unlike Leslau, M. Cohen used ɑ for the neutral allophone [ɐ]. Thus in Leslau's transcription the two allophones [ɐ] and [ɛ] are indistinguishable. yhever 02:38, 13 June 2006 (UTC)
(I actually have no idea how far to indent :-) For me a big issue is ease of use, especially by non-linguists, including native speakers. So I think yhever's examples of Ethiopians using both e and o for the /ɐ/ phoneme after /w/ are interesting. I hadn't thought of the possibility of allowing alternate spellings for the same word, but maybe that's a good compromise. The part that I worry about in yhever's proposal is that it adds diacritics (macrons) in two more places (ō, ū) to distinguish these from the allophones of /ɐ/. That would work against simplicity; it would be really easy for people to forget those (as well as extra work to type them). But one part of my original proposal was to allow in some contexts for the diacritics to be missing altogether (leading to ambiguity of course, but I think this could be tolerated, again in certain contexts, as it is, for example, for Japanese). In any case, I'd argue again the diacritic on ā. Without the diacritic, the symbol suggests something close to the Amharic (and Tigrinya) vowel, whether people are thinking of it as IPA or standard "European orthography" (Spanish, French, Italian, German, etc.).
By the way, what yhever is saying for Amharic applies to vowel harmony in Tigrinya too, where you can see words written in ways that reflect this sometimes. So the first vowel in ጽቡቕ 'good' is most often pronounced more like [u] (though I'm not sure about the length) to agree with the following [u].
There's a related issue that concerns both vowels and consonants. That's the alternation within fidel as well as in individual and regional accents (I think) between pairs like ጐ/ጎ and ጕ/ጉ. This holds for the velar consonants and /h/. One phonemic analysis (the right one, I think), for Tigrinya, Amharic and Chaha, in any case, treats the labialized variants of these consonants as separate phonemes, which would make the "right" transliteration gwe, gwi, etc. But maybe in these cases, we should allow both gwe and go and both gwi and gu. People would probably find it strange to write Mekwennin, Ankweber, etc. What do you guys think? — MikeG (talk) 15:11, 13 June 2006 (UTC)

(Let's restart the indenting for simplicity) Regarding the xw (x being a variable) consonants ('bastard letters'), I think the more common transliteration and fidel spelling be used with the other version listed as an alternate spelling and pronounciation. E.g. Gonder would be ጎንደር (gonder or gōnder), but would include the variant/alternate ጐንደር (gwender), which is important for the article, e.g. as the fact that it started with gwe is the reason it was chosen as a capital. ቈላ (qwella 'lowlands'), however, would have the ቆላ (qolla) spelling first, even though it's not very important to the article (listed as an alternate, again). ቋንቋ ('language' qwanqwa), however, would always be spelt that way, as I don't believe anyone pronounces it otherwise. — ዮም (Yom) | contribs • Talk 16:33, 13 June 2006 (UTC)

I'll respond later to the new points that have been raised by you, but just as an anecdote, ቋንቋ is actually spelt differently by Haddis Alemayyehu in his ፍቅር ፡ እስከ ፡ መቃብር (first page of the introduction, last line): as ቁዋንቁዋ, but this has to do with his suggested spelling reform that would abolish the "bastard" letters, and not with a different pronunciation. yhever 16:46, 13 June 2006 (UTC)

Scope

Before we discuss what system to use, I think it would be a good idea to decide what its usage would be. Would it be at all used in the naming of pages, transliterations for non-linguistic articles, transliterations for linguistic articles, etc.? I don't believe it should be used for naming articles unless the subject is obscure (which, admittedly, covers most of them) or has two different spellings with equal popularity, with the transliteration as a third alternative. With regards to transliterations for names, words, place names, and the like, I think the system ultimately adopted (assuming it's different from the Leslau one) would be fine. However, it'd actually be a bit weird for me to see the linguistic articles using this system instead of the common Leslau transliteration. Also, writing, Gi'iz instead of Ge'ez is just uncomfortable to me now because of the latters prevalence, but I would say this would fall into the first category (page naming), to be carried out throughout the page. I.e., if a page has a common spelling (like Haddis Alemayehu or Addis Ababa), then the common spelling should be used throughout the page, except for the transliteration in the opening sentence (or paragraph).

Another issue would be whether the transliteration system should be used in conjunction with the more popular Leslau one when transliterating the name at the beginning of an article (e.g. write [assuming USEGN] "Hadis Alemayehu" and "Haddis Alämayehu" [it doesn't do gemination either, not sure about the status of the y length, actually, but I think it's considered geminated]), or would we just use the adopted system?

ዮም (Yom) | contribs • Talk 03:14, 5 June 2006 (UTC)

Yes, I agree that lots of the decisions depend on scope. I think we could follow the other Wikipedia communities in how they've made these decisions. In particular I believe they've generally let common spellings override the standards they've adopted. This would apply to words like Ge'ez, Addis Ababa, Menelik. (What makes Ethiopian Semitic different from most of these other cases (Japanese, Chinese, Arabic) is that there already existed one or two standards that everybody already agreed on before the Wikipedians had to decide. So a lot of this is going to be harder with ES.)
We could have two separate systems, one for use within the linguistic articles, one for page naming and non-linguistic articles, though this would disagree with what's done with other non-roman scripts (at least Japanese, Chinese, and Arabic).
As for what's familiar within linguistics, I think the trend in recent years in the younger generation of Ethiopian Semiticists, many (most?) of them native speakers of the languages, is away from Leslau's system, though unfortunately there doesn't seem to be any agreement among these. I've seen at least three other combinations of symbols (besides Leslau's) to represent the vowels in linguistic articles on Amharic, Tigrinya, and the Gurage languages. I'm sure User:yhever will have thoughts on this too. — MikeG (talk) 12:59, 5 June 2006 (UTC)

Consonants

For consonants I can think of three issues:

  • Gemination
  • Pairs of symbols (or not) for some consonants
  • Symbols for /ḳ/, /ṣ/ (q/k', ts, ts', s')

For gemination, the main argument in favor is that consonant length is of course significant in the languages. The main argument against enforcing gemination is that it's not indicated in fidel or consistently in informal transliteration among native speakers. Note that there is a precedent for not marking contrastive supra-segmental features in transliterating languages where these features are not marked in the writing system, unless the topic is phonetics or phonology of course. This applies to Japanese pitch accent, to Russian stress, and to Chinese tone (though tone is indicated in linguistic articles on Chinese within Wikipedia). Maybe the right way is to allow both, as with vowel allophones? — MikeG (talk) 15:37, 13 June 2006 (UTC)

I do think gemination should be indicated (esp. wrt to Addis Abeba, e.g. – the spelling Adis Abeba would anger many as it's the Italian spelling), though it does pose a problem for some words (personally, it's a little hard to tell how long is considered geminated). My gut feeling is to allow either way for now, but if a certain gemination is known, then to replace the ungeminated version with the geminated one. That way, we can get mass transliteration done easily by non-speakers who can read fidel, but we can still have the most accurate transcription too, as other (Amharic/Tigrinya/whatever-speaking) users catch where there should be gemination and change it. Regarding /ḳ/ and /ṣ/, q and ts seem to be the easiest way to transcribe them, but both betray their true sounds to a degree. Q invokes a more pharangealized k given that the languages we're transcribing are going to be mainly Semitic, while ts (maybe tz like in Hebrew?) could be mistaken for the two individual consonants in word-medial positions (take the verb to write in Amharic metsaf, e.g.), and when the two occur together naturally (nothing comes to mind right now), the transcription would be confused. k' and ts' might solve that problem, but the ' could easily be mistaken for a glottal stop. /ḳ/ and /ṣ/ would be the most accurate in some sense (though they're both a bit ambiguous), but I fear that the difficulty of finding such letters makes its use limited to linguistic articles involving Proto-semitic. There's also the option of ts to show it's one phenome, though that looks a bit weird. Perhaps the best options would be k' and s', both relatively widely used, and with ' being distinguishable from a glottal stop through its italicization. Also, would we use ' or ʾ (the latter being more often used for glottal stops, though)? — ዮም (Yom) | contribs • Talk 16:49, 13 June 2006 (UTC)
I also think that gemination should be indicated if it known. As for q, I think it is much preferrable to use a special latin letter for an Ethiopic letter, when it is possible, rather than using an underdot or an apostrophe.
ts is fine for non-strict transliteration. In fact, I am considering the option of not indicating emphaticness in non-strict transliteration but using a diacritical dot in the strict transl., similarly to what has been decided in the Arabic Manual of Style. Thus we would have non-strict: p, t, ts, ch, q, x (or: kh?, qh?) for ጵ, ጥ, ጽ, ጭ, ቅ, ቕ; but strict: ṗ, ṭ, ṣ, ċ, q (or: ḳ?), ẋ. What do you think about it? yhever 22:40, 13 June 2006 (UTC)
The problem with following the Arabic MOS wrt to emphatic consonants is that emphaticness in Arabic is completely different from Ethiosemitic emphaticness. While saad and sin are relatively close in sound (the pharyngealization is slight when present in my experience), as are Ta' and ta' and daad and dal, s' (Ṣädäy) vs. s (Sawt/Shin) is completely different in pronounciation, and the pronounciation differences b/w (simplified transcription out of laziness) ch and CH, t and T, and k and K are evident to non-speakers. If the transcription to be used is to be simplified and used throughout an article (which I wasn't thinking about earlier, I was more considering its use for transliteration in the opening sentence), then a simpler version using q and ts would preferable, as well as ch and sh for č and š (though ñ can probably be kept, though maybe dropped for the rest of the article gn as usually used for modern names?). Expressing emphatic ch and t without extra characters, however, will be a difficult task. Most Ethiopians either ignore it (e.g. Petros - emphatic "p" and "t") or use capitalized versions (e.g. CH for emphatic ch and T for emphatic t), which looks unprofessional. Perhaps the best way to resolve this issue is similar to the Arabic MOS. We might have to have a transcription system for use throughout articles, and a more complex one for the beginning of the article. In which case, the vowels above would be the more complex one (since writing ā, ē, ī, ō, or ū throughout an article would be asking too much). — ዮም (Yom) | contribs • Talk 23:53, 13 June 2006 (UTC)
I'm not sure I understood your point. In fact, I think emphaticness is as important to Arabic as it is to Amharic (or any other Ethiopian language), but a simplified transcription may be needed for non-specialists, and in order to avoid using diacritics throughout the article.
My suggestion (which I myself don't find entirely satisfactory) was to use diacritics (by which I mean both macrons and "emphatic dots") only in the more complex transcription in the beginning of the article. Now, ts and q can be used also in the simpler transcription since they don't require diacritics -- but the other emphatic consonant would not be distinguished in the simpler transcription under the suggested method.
You wrote "perhaps the best way..." -- Do you think it is the best way, or do you think a different method should be used? yhever 01:00, 14 June 2006 (UTC)


I think this is the best way. I hadn't been thinking of what would be used for the article name and throughout the article (which obviously should avoid diacritics except when necessary). Consonants should still be geminated, IMHO, except in the case of "sh," "ch" and other letters that require two letters (also gn/ny, though not ñ). Wrt to simplified vowels, I do believe we should adopt the same system, with the change that ā, ō, and ū be simplified to a, o, and u. There is a question of gn vs. ny vs. ñ for the simplified version, as you wouldn't see the name 'Tegegne' written as 'Tegenye' or 'Tegeñe'. — ዮም (Yom) | contribs • Talk 01:40, 14 June 2006 (UTC)

General comments

I'll admit that I don't know much about the problems of transliterating from Amharic/Tigrinyan into Latin characters, so I'm willing to consent for the most part on whichever system you guys agree to. However, I do have some specific concerns about whichever system is adopted that I'd like to mention.

  • Respect established spellings. There a few words & proper names from these languages that have acquired an accepted spelling -- which may be incorrect. Most significantly these include "Addis Ababa", "Haile Selassie", & "Ge'ez". I've seen the city spelled both "Axum" or "Aksum", so as long as either is acceptible there shouldn't be a problem with them.
  • Simplicity. Let's try to avoid unusual characters, like the schwa or diacritical marks. The reason for this is that although this might bring additional accuracy to the transliterations, it confuses the reader (most of whom will not have prior lingusitic training or familiarity with the languages) & breaks in some browsers. (For example, the verison of Opera that is installed on my Nokia 770 displays a surprising share of the unusual letters with a broken box.)
  • Consistency. This might be the most controversial point, because it dictates that if we decide to accept "Haile Selassie", then we should use "Haile" whenever that set of sounds appear. The advantage of consistency is that it encourages the reader to notice the reuse of these forms, & help make the proper names more familiar.

Of course if these principles are followed, either in whole or part, we will have a transliteration system unlike any other & which will likely baffle the experts. For this reason, if I convince you to use a simplified method for representing vowels, then in the opening paragraph we will need to add the technical transliteration of the word -- hopefully with an abbreviation to show which transliteration system is being followed.

Are these conditions something you experts can work under? -- llywrch 23:17, 13 June 2006 (UTC)

I fully agree to your first two points. (1) I agree that established spellings should be maintained. (2) I have already suggested using two kinds of transcription: a simpler one for the name of the article, and to use throughout the article (without diacritics or special symbols); and a strict one which would appear once in the first paragraph which would be more accurate and reversible (or almost reversible).
Your last point, which you have admitted yourself to be the most controversial, I believe is in contradiction with the first point. If "Haile Sellasie" is to be maintained, what should be done with "Hayq"? In fact the first vowel in "Sellasie" is identical to the vowel in "birr", but both are established spellings. The problem is that "established spellings" are inconsistent.
Anyhow, thanks for your comments. yhever 01:18, 14 June 2006 (UTC)
One point, somewhat related to this discussion, that occured to me as I was looking at Battle of Adowa: when writing the plural form of a noun borrowed from Amharic or Tigrinyan, do we use the standard English form (i.e., -s or -es), or also use the proper plural form from the original language? Right now, the practice on WP is to use kililoch as the plural of kilil (because a user can find it stated in Regions of Ethiopia), but for the plural of Ras I used "Rasses" for lack of a better alternative. (And I'll admit that I'm unsure whether the "s" in Ras should be pronounced like a soft "s" or like a "z"; if the latter, then the proper English plural would be more accurately written "Rases" with one "s".) -- llywrch 15:59, 14 June 2006 (UTC)
I won't address the question yet, but the s in Ras is pronounced like a soft s. No consonants in Ethiopian Semitic languages technically have allophones, as they are written as they are pronounced (though in reality, there are tons of allophones, like k/x and q/x' in Tigrinya, and s/sh, d/dj, z/zh, t/ch, T/CH, l/y, and n/ñ, all of which are conditioned by their environments for most cases, but appear sometimes not as allophones and are written with a letter that's a variant of the original sound, usually a line over it) — ዮም (Yom) | contribs • Talk 16:35, 14 June 2006 (UTC)

Simpler version vs. Strict transliteration

It seems that ዮም and llywrch agree that we should use two systems of transliteration, a simpler one for article names and as the common transliteration throughout the article, and a strict/complex one that would be given once at the first appearance of the transliterated term that would be more accurate and hopefully reversible. I believe this solves also Mike Gasser's uneasiness with too complex transliteration that won't be understood by non-specialists. If this can be agreed upon, then the proposals on Wikipedia:WikiProject Ethiopia/Transliteration should be divided into two -- one for each system of transliteration. So, what do you think? yhever 02:33, 14 June 2006 (UTC)

Sure, but I don't think that we need two pages. Two subsets would be fine. I will add a proposal for both relatively soon. — ዮም (Yom) | contribs • Talk 02:39, 14 June 2006 (UTC)


Yom's proposals

Here are my proposals for both consonants and vowels for both the complex and simple transliteration systems. Discussion can start either under each header or in a separate header, it doesn't matter, just follow whatever precedent the first commenter sets. — ዮም (Yom) | contribs • Talk 03:13, 14 June 2006 (UTC)

Complex

Consonants

(box stolen from MG)

Ethiopian Semitic Consonant Transcription
IPA p' t' k' ʧ' s' ʧ ʤ ʃ ʒ ɲ ʔ x,χ x',χ' ħ ʕ
Proposal ṗ ṭ ḳ č̣ ṣ č ǧ š ž ñ ʾ x or ḫ q/q̄/x' ??? ḥ ʽ

ዮም (Yom) | contribs • Talk 03:13, 14 June 2006 (UTC)

For ኸ and ቐ I would propose using x and ẋ. I also agree that ḳ is better than q for the strict transl. For ኘ I think ñ is better than ň. yhever 11:45, 14 June 2006 (UTC)
I have no problems with x and ẋ. ň was just a typo for ñ because of the bad resolution on my screen. I've changed it now. — ዮም (Yom) | contribs • Talk 03:13, 14 June 2006 (UTC)
The trouble with those dots is that they're not found in many fonts, and in fact the one under č doesn't look at all right in the font I have (the dot is in the wrong place). Why not apostrophe, which is a pretty standard way of representing ejectives across languages (much more so than the dots in fact)? Also, do we really have to go with ǧ? This is not part of IPA or any language's orthography that I know of (except Esperanto!), and j would agree with the simple notation. Shouldn't we be trying to make the two transcription styles resemble each other as far as possible? For the same reason, why not q in place of ḳ? OK, I understand the reasons; it represents a different consonant in IPA and Arabic transliteration, but it is pretty standard in linguistic transcriptions of Ethio-Semitic. — MikeG (talk) 03:49, 15 June 2006 (UTC)
Hmm, I hadn't thought of that problem (it shows up fine for me). The "č dot" problem is from something else, though. All the other characters are one letter, but the "č dot" is two separate characters (I copied it from your transcription, and in the editing box, the dot is clearly separate from the č, over to the right some, but when parsed it comes out a little off to the left, but clearly under the č). The first letter is č, while the second is a dot, but different from a period, and it is supposed to be parsed as going under the preceding character. If there's a single character for č dot, however, I don't think this will be a problem. Do you have a problem with the other ones? I'm not certain, but I do believe that all of the others should be completely supported by most browsers (ignoring relatively old ones). Someone who's knowledgable in these things should confirm or deny this, though (is there some sort of place where we can ask or look it up in Wikipedia?). regarding ǧ, I picked it because it's one character, unlike dj, which can often be confusing (I know many an American who is baffled regarding the pronounciation of Djibouti) and therefore no better than ǧ (you're objecting because of its uncertain pronounciation for non-specialists, right?). It's also used in Arabic transliteration, I believe, and it's the transliteration used for the Encyclopaedia Aethiopica. I have no objection to "j" for simplified, but it does represent a sound more like "y" for IPA and, being the complex transliteration, could easily be confused for that sound. We could use the IPA ʤ, but that seems to confound your concerns even further (that people wouldn't know how it's pronounced). Regarding q, the letter used to represent is cognate with Latin Q, but, as you said, it represents a diffrent sound than what we're transcribing. As q and k aren't pronounced differently, using a diacritic is actually better in showing the distinctness of the sound. I'm flexible with regards to ḳ vs. q, but I'm worried about what we would use for Semitic comparison articles (e.g. List of Proto-Semitic stems and Proto-Semitic language), where q is used for a pharyngealized k, and an Ethiosemitic "q" would fail to show the distinction in sound. Perhaps we could make "q" standard for most linguistic articles but not for ones comparing pronounciation with other Semitic languages? But that would be confusing, as only a few consonants would change and readers would be confused. To be honest, I like the look of q better as it's more often used, but there are problems for the two articles I mentioned above. — ዮም (Yom) | contribs • Talk 04:50, 15 June 2006 (UTC)
On second thought, this might not be a problem, however, as there's already a general consensus for transliteration for Semitic languages in general (which may or may not agree with the individual language transliterations; I'm not sure). — ዮም (Yom) | contribs • Talk 04:52, 15 June 2006 (UTC)
I used two characters for the č with dot because there's no single Unicode character for the combined symbol. I used the "combining dot below" character (hex #0323) to make it. I think it would be preferrable to have a single Unicode character for this phoneme, but there doesn't even seem to be a c with dot below (though there is one with dot above: #010B). As far as j is concerned, it does represent a different sound in IPA (we're using y for that), but it's pretty standard for ʤ in the American variant of IPA with the caron added: ǰ (Unicode #01F0, see the "Americanist" characters on the IPA page, which include the other characters with carons that we're talking about using). — MikeG (talk) 15:26, 15 June 2006 (UTC)
One possibility would be to use c for ʧ and ċ (#010B) for ʧ'. c does represent a somewhat different sound in IPA (palatal stop), but most people probably would not know this (it's a pretty rare phoneme). A compromise would be to stick with č for ʧ and use the single character ċ for ʧ'. — MikeG (talk) 15:46, 15 June 2006 (UTC)
I guess we use a superscript w for the labialized consonants (kʷ, etc.)? And for the palatalized velars of Western Gurage a superscript y (kʸ, gʸ, ḳʸ, xʸ)? For Ge'ez ሠ and ፀ, I suggest we use the "standard" ś and ṣ´ (except that, as you can see, I couldn't figure out how to add both the dot and the acute accent to the s for ፀ). Also Z(w)ay has an implosive d ([ɗ], apparently borrowed from Cushitic, the consonant that's written dh in Oromo orthography). I propose we use ḍ for this since it's glottalized like the other consonants we're using dots for. — MikeG (talk) 02:41, 17 June 2006 (UTC)
Regarding the superscripts, no disagreement here. Regarding Sawt, however, I'm not sure if the standard ś is such a good idea, as it represents a proto-semitic phenome continued by śat, which could have indeed been pronounced [ś] (later s, though I'm not sure how early this is). ḍ might also be a problem, since it is a variant transliteration of Ge'ez Tsappa/ḍappa, which is no different from modern Tsaday (Tsade (letter)). Is there anything wrong with using the IPA ɗ in this case? I would rather prefer the IPA transcription. — ዮም (Yom) | contribs • Talk 03:25, 17 June 2006 (UTC)
No, ɗ would be fine (and for this very limited case there are no competing traditions to worry about). — MikeG (talk) 21:33, 17 June 2006 (UTC)

Outstanding issues (I think)

  • Ejectives generally: dot or apostrophe?
  • ጨ: č+dot, ċ, č', something else?
  • ቀ: ḳ, q?

— MikeG (talk) 13:18, 25 June 2006 (UTC)

I think it should be like that: ቀ should be q in the simple transcription (since it is a simple letter) but ḳ in the complex transcription (to make it more coherent with the other emphatic consonants). Thus, similarly, ጨ should be č̣ in the complex transcription. As for its simple transcription, it depends on the decision about emphatic consonants in general (beside q and ts) in the simple transcription. I have not made up my mind yet. yhever 14:33, 25 June 2006 (UTC)
I agree with what yhever stated above wrt the complex transliteration. The simple transliteration is still up for grabs. The real outstanding issue is, as yhever said, the simple transliteration (i.e. do we distinguish between ተ and ጠ, ቸ and ጨ, and ፐ and ጰ or no?). I don't think that the č + dot transliteration will be that big of a problem. Remember, it's the complex one, so there's always the possibility that people won't be able to see the symbols (we could have a superscript question makr next to it or the Ge'ez script, like they have done for Japanese, to link people directly to various fonts they can get for Ge'ez as well as to see all of the transliteration symbols) — ዮም (Yom) | contribs • Talk • E 18:07, 25 June 2006 (UTC)
OK. I'll go along with you guys on the dots for the ejectives, in the interest of making the relationship to emphatics in other Semitic languages clear. But, as yom suggests, I think we need to make it easy for people to read these articles, which should not be just for linguists and Semiticists. If I'm not mistaken, we've now agreed on everything (both consonants and vowels) for the complex transcription. — MikeG (talk) 13:01, 26 June 2006 (UTC)

Vowels

This is Yhever's proposal with the more specific vowel lengths included. I don't believe anyone has a problem with this for the specific transliteration.

Ethiopian Semitic Vowel Transcription
IPA ለ [ɐ,ǝ] ሉ [u] ሊ [i] ላ [a] ሌ [e] ል [ɨ] ሎ [o]
Proposal e (or o) ū ī ā ē i (or u) ō

ዮም (Yom) | contribs • Talk 03:13, 14 June 2006 (UTC)

My original proposal included allowing o and u optionally as the rounded allophones of e and i, respectively. yhever 11:46, 14 June 2006 (UTC)
Sorry, I forgot to add that; I added it now. — ዮም (Yom) | contribs • Talk 15:07, 14 June 2006 (UTC)
I assume it will be OK to modify this for other ES languages. For example, in Tigre the 1st and 4th vowels differ mainly in length, and Raz uses a for the 1st. Leslau does the same for Ge'ez in his concise dictionary (1: a, 4: ā). And Silt'e has the 5 short and 5 long vowels borrowed from Cushitic, with short a filling the 1st order position. — MikeG (talk) 03:45, 16 June 2006 (UTC)
Yes, for tigre, anything but a and ā for 1st and 4th, respectively, would be less than desirable. The same goes for Ge'ez. Also, Shin/sin (Ge'ez Sawt) will have to be distinguished from Sat. The former represents both š and s (with the shift being š->s; cp. USBGN negash and nigus); this is a consonantal issue, but we obviously can't apply these vowels to all Ethiosemitic languages (Gurage languages all have 10 vowels, don't they?). The Gurage languages should be relatively simple using the letters aeiou for short (those are their vowels, right?), and with lines over them (āēīōū) for long). What about Harari/Haderé? Did it adopt/keep long [a:] from Arabic too (well, we're already using ā for all, but I mean is first order short a?)? I don't think this'll be too big of a problem, though. — ዮም (Yom) | contribs • Talk 05:00, 16 June 2006 (UTC)
Northern Gurage (Soddo) and some Western Gurage (at least Chaha) have the familiar seven vowels of Amharic (more or less). Eastern (Silt'e and I think Zway, though I'm not sure) have the Oromo vowels. yhever should know about the other Gurage languages. I'll check on Harari. I know there are some long vowels. yhever probably knows. In any case, I doubt there will be a problem. — MikeG (talk) 13:28, 16 June 2006 (UTC)
It turns out that Z(w)ay has the same 5 short and 5 long vowels that Silt'e has (I found a great new book on Zay). In addition to 1st order, Harari has long and short [a], so we could use e, a, and ā for these. Inor (Ennemor) will be the hard one since (according to Berhanu and Hetzron's book), it has 8 short and 8 long vowels. They argue the the long vowels are really double vowels, so we could represent them by doubling the short vowel symbol. The short vowels include roughly the Amharic set plus /æ/, so I guess we could just use the symbol æ for this 8th vowel. More complicated in Inor is the fact that vowels and some consonants (other than consonants that are already nasal) can be nasalized. I guess we could use the standard tilde for this, but that might not work together with the macron. Maybe we should worry about Inor when get to that article. — MikeG (talk) 02:25, 17 June 2006 (UTC)

Simple

Consonants

Ethiopian Semitic Consonant Transcription
IPA p' t' k' ʧ' s' ʧ ʤ ʃ ʒ ɲ ʔ x,χ x',χ' ħ ʕ
Proposal p t q ch ts ch j sh zh gn ʾ k or x q or x'??? h ʽ

I chose gn vs. ñ as in Italian & French based on how most Ethiopians trancribe it. K/X in Tigrinya will probably depend on the circumstance. A Tigrinya-speaker with the name Tekle (w/o an established spelling) should stay Tekle despite the pronounciation Texle, but there may be some names that don't usually use k for the x phenome, though I don't know them. The same goes for q vs. x' (e.g. Amh. Meqellē, even though in many Tigrinya dialects it's Mex'elle, with a short last e, and some dialects [southern?] have Meqelle). The ayin and alif marks could be compressed into one ' for simplicity instead of the spiritus asper and the other mark for alif (upside down/mirrored spiritus asper).

ዮም (Yom) | contribs • Talk 03:13, 14 June 2006 (UTC)

I think ny is better than gn since "gn" is ambiguous in English (compare "sign" and "signature"). ny has the disadvantage of confusing ኘ with ንየ, but I think there are very few minimal pairs, so that this isn't really an issue.
As for ኸ and ቐ, what do you think about using kh and qh? You have suggested yourself using k and q (and to neutralize the difference between a fricative and a plosive); A non-specialist seeing kh and qh may either ignore the h (and pronounce k and q, or k for both), or understand that these digraphs denote fricative consonants. Anyhow, no harm done. yhever 11:53, 14 June 2006 (UTC)
I agree with yhever on ny vs. gn. Since this is the English Wikipedia, we shouldn't be using conventions confined to French and Italian orthography (though I agree that this is something a lot of Ethiopians are used to). gn can be very confusing for English readers and has even led to the spread of a bunch of languages names, repeatedly copied on different websites from some initial offender (including that of the National Geographic Society) with -nga in place of -gna (Tigringa, Orominga, etc.). For ኸ and ቐ, I think we should go with something that's close to k and q (or whatever is used for ቀ), so yhever's proposal for using h to show the fricatives makes sense. But since is the simplified transciption, why distinguish these at all? Though the precise rules differ with dialect, they're perfectly predictable allophones of ከ and ቀ that just happen to be distinguished in fidel (according to everything I've ever read or heard). I agree with everything else you propose, Yom. (I'll comment on vowels tomorrow; you guys are way ahead of me!) — MikeG (talk) 03:40, 15 June 2006 (UTC)
I'll concede the gn vs. ny as I don't think this will come up much (not many names have the phenome and those that do have established spellings, like Tegegne). For the Tigrinya allophones of k and q, I wasn't actually sure what to use, so my proposal for those was up in the air. Actually, seeing as this is the simple transliteration, I'm not sure if any distinction should be made at all. No Tigrinya speaker would write their name as Tekhle or the capital of Tigray as Meqhelle. Though the kh and qh is a good transcription method in general, it's better as a dirty way of representing the sounds when more specific single characters aren't available. Since this is the simple transliteration, I propose we simply use k and q, since the allophones will be identified in the complex transliteration (this was your proposal, too, right?). — ዮም (Yom) | contribs • Talk 04:58, 15 June 2006 (UTC)
Right. As you say, nobody seems to distinguish the allophones in informal transliteration. (Another problem with representing the allophones is that there might be confusion across the dialects that have different rules for where the fricatives are used.) — MikeG (talk) 14:53, 15 June 2006 (UTC)
Agreed, dialectal variation is another problem (e.g., don't Southern Tigrinya speakers say Meqelle, while others say Meqhelle?). If yhever agrees with this, then this section, too, is basically decided (unless Llywrch has any objections, which I doubt, or unless Codex/Feqade decides to join the discussion and disagrees). — ዮም (Yom) | contribs • Talk 21:34, 15 June 2006 (UTC)
Also, what do you guys think about the possibility of using p', t', and ch' for simple transliterations? The thought came to mind because of its already prevalent use (e.g. P'ent'ay, and Silt'e - the latter not yet made, but already spelled that way). It would be a bit of a burdon, though. — ዮም (Yom) | contribs • Talk 00:53, 16 June 2006 (UTC)
(Actually there is a Silt'e language article.) The apostrophe could be an option, but I guess we don't want too many options. I haven't seen it in the transliteration styles that Ethiopians use for their own names. But it is weird that we're distinguishing /k/ from /ḳ/ and /s/ from /ṣ/ but not the other pairs (especially one that's as important as /t/ vs. /ṭ/). I could go either way. — MikeG (talk) 03:39, 16 June 2006 (UTC)
As for the allophones ኸ and ቐ in Tigrinya, I am not competent to say anything, so I'll leave that to you. But ኸ also occurs in Chaha where it is not, strictly speaking, an allophone (e.g. ኸረ(ም) 'be(come)' vs. ከረ 'day'; or: ከና(ም) 'ascend, go east' vs. ኸና(ም) 'forbid, refuse'). I suppose there won't be many (or any) articles with names in Chaha, but this should be kept in mind. In linguistic articles, a more complex transcription will anyhow be needed.
As for using an apostrophe to mark emphatic consonant (other than "ts" and "q"), I am not sure. This opens a new problem: what should be done when we have a digraph such as "ts" (ጽ) and the sequence "t-s" (ትስ) or "sh" (ሽ) and "s-h" (ስህ); in the Arabic Manual of Style they used an apostrophe to split such digraphs (e.g. "dh" = ذ but "d'h" = ده). In "mets'haf" the apostrophe may serve both functions, but what about the name ይስሐቅ, should it be transliterated as "yis'haq"? If so, then the apostrophe is used for two different functions, and may be confusing.
I have not made up my mind, but these problems should be taken into consideration. yhever 09:52, 18 June 2006 (UTC)
Since we're already allowing some ambiguity in the simple transliteration, I think we could just go with Yishaq for ይስሐቅ. This kind of situation probably wouldn't come up very often. I agree, though, that we should use the apostrophe for only one purpose, either ejectives that don't already have their own symbol or separation of symbols that look like digraphs. — MikeG (talk) 14:18, 18 June 2006 (UTC)
Yishaq is fine, but mets'haf wouldn't be a problem since we would be using the complex transliteration. Plus, we could always insert an i for the super-short sort schwa. In simple transliteration, I don't think t followed by s happens often. — ዮም (Yom) | contribs • Talk 18:52, 18 June 2006 (UTC)
But don't you think that "Yishaq" may be misunderstood as ይሻቅ? yhever 00:16, 19 June 2006 (UTC)

(restarting indentation) Yes, but I don't know what else we could do. The pronounciation will be made obvious in the first sentence, and Yis'haq would make the s look like a variant spelling of ts. The s-h pair doesn't happen often, and we can note that an alternate transliteration (but not pronounciation) is Yis'haq, or some other method that shows that the two are not part of a single digraph. — ዮም (Yom) | contribs • Talk 01:08, 19 June 2006 (UTC)

To summarize, these are the places where there would be ambiguity in the simple transliteration, given what (I think) we've agreed on so far:

  • p: /p/, /p'/
  • t: /t/, /t'/
  • ch: /ʧ/, /ʧ'/
  • ts: /s'/, /ts/
  • sh: /ʃ/, /sh/
  • zh: /ʒ/, /zh/
  • ny: /ɲ/, /ny/

Using apostrophe for ejective p, t, and ch would solve the first three. Using apostrophe to separate ambiguous pairs of characters would solve the last four. — MikeG (talk) 17:47, 19 June 2006 (UTC)

I don't think ch will ever be a problem since c doesn't represent any phenome by itself in our transliteration. ʧ for simple transliteration is more specific than necessary. I also think ʃ, ʒ, and ɲ are more specific than necessary for simple transliteration (except perhaps for ancient names in which only the consonants are known). The alternatives for ts (s' or ts') could be adopted, though I think ts or ts' are best. I can't see us adopting anything other than sh, zh, and ny for the simple transliterations for cases where there is no h sound being represented. Sh is certainly read as [ʃ] inherently, so only when it is not read thus should we disambiguate it. Zh is a little ambiguous as well, but I can't think of a better digraph or single letter to represent the sound for simple. ny we've already discussed (and is rare aside from ñ cp. metoñña and metonnya - both acceptable for 'one hundredth' or 'cent'), so I really think the question is mainly whether or not we should distinguish between p and p', t and t', and ch and ch'. I have no problem with including the apostrophes, but it seems to me as if the spellings would just look weird. I guess this would only apply to new articles without established spellings (e.g. P'ét'ros would still be spelled Petros), so I'm flexible with regards to this. With regards to situations like Yeshaq, I think the best thing to do would be to do something like this: Bahr negus Yeshaq (or Yes'haq Ge'ez ....) was....
I think yhever is most adamant on this, so what do you think, yhever? — ዮም (Yom) | contribs • Talk 19:11, 19 June 2006 (UTC)
For perspective, I searched through the 68 page CSA pdf on all Ethiopian towms and woredas, and in every instance of "sh," the sound being denoted was [ʃ] (and not s followed by h). — ዮም (Yom) | contribs • Talk 23:09, 19 June 2006 (UTC)
Oops. Ignore most of my first response. I admit I was on a computer that saw all the IPA symbols you wrote as boxes and assumed they stood for symbols like č and ñ, rather than the actual IPA symbols. I also missed that you were representing what the consensus seems to be, rather than proposing alternatives to use (which were boxes for me, so I couldn't see that it was just IPA). I agree with the above, then. Let's see what yhever thinks. — ዮም (Yom) | contribs • Talk 23:52, 19 June 2006 (UTC)

Vowels

In order, i.e., by Leslau: ä | u | i | a | e | ə | o .

(again box from MG)

Ethiopian Semitic Vowel Transcription
IPA ለ [ɐ,ǝ] ሉ [u] ሊ [i] ላ [a] ሌ [e] ል [ɨ] ሎ [o]
Proposal e u i a ē i o
More distinctive alternate e u ī a ē i o
More simplified e u i a e i o

Remember that for article names without preferred spellings, that this will be used throughout the article. Personally, the distinction between ī and i isn't as important as between e and ē. The more simplified one is for if we decide to do away with diacritics for all vowels, in which case confusing e and ē seems like the best choice as the diacritic is already not often marked (e.g. Meqelle, libbe, Harerge, etc.).

ዮም (Yom) | contribs • Talk 03:13, 14 June 2006 (UTC)

I agree to the "more simplified" proposal, except for one detail. I think it is important enough to distinguish between the 1st and the 5th order vowels to allow for a slight complication.
I am not sure what would be the best option. One way is to use é for the 5th order vowel; é occurs in the names of e.g. French articles, and is also known in English in the loanword(s) "fiancé(e)", and thus it won't seem completely strange to someone who can read English.
One other option, which is applicable only for Amharic, is to take advantage of the fact that in some Amharic dialects the 5th order vowel is pronounced with an on-glide as ye, so that something like ie can be used (e.g. "biet", "Blattiengieta"). This option has been used in the common spelling "Haile Sellasie".
yhever 12:09, 14 June 2006 (UTC)
Oops... I wasn't paying attention... Your two other proposals can do just as well. yhever 12:12, 14 June 2006 (UTC)
My proposal is the one that distinguishes between the first and fifth order vowels (e and ē/é), but not the third and sixth (i and ī). The other two are alternatives in case someone wants a transcription system that distinguishes every vowel, or one without any diacritics at all. French é can work as well (in fact, it was an alternative in my original proposition), though I do like the look of ē better for some reason. I'm flexible, though. The "ie" is a possibility, but the "ie" pronounciation isn't that common. It's more of a peasant dialect and makes someone sound uneducated, IMHO. — ዮም (Yom) | contribs • Talk 15:02, 14 June 2006 (UTC)
If we're going to distinguish the 1st and 5th vowels, I think é is a better option than ē for 5 since it's a part of every conceivable roman font whereas ē is not. — MikeG (talk) 15:52, 15 June 2006 (UTC)
Okay, I'll concede that. This section (Simple vowels) seems pretty much decided, then. — ዮም (Yom) | contribs • Talk 21:33, 15 June 2006 (UTC)