I'd like to propose that those of us who edit pages that have to do with Ethiopia and Eritrea agree on an official policy to transliterate the languages that are written in the Ethiopic script, as has been done for some other languages, including Japanese, Korean, and the Indic languages and is currently under discussion for Arabic and Hebrew.
There are two places for transliteration in an encyclopedia: for articles on the languages themselves and for writing words that come originally from the languages, especially the names of people and places and the titles of works, in non-linguistic articles. We wouldn't have to necessarily agree on the same conventions for both purposes, though it would obviously be simpler for users if they were at least similar. The discussion here is meant to deal with words originating in these languages that appear in non-linguistic articles. There is currently informal agreement on using a variant of the WL system described below for linguistic articles (see Amharic language, Tigrinya language, Soddo language). A separate issue is when to use Oromo orthography for names that also have Amharic spellings. (Another possibility is to regularly use both.)
There are at least three well-accepted sets of conventions for romanizing Amharic, and in some cases other Ethiopian Semitic languages, and a number of variations on these. These include
The vowels present an obvious problem because the seven of them need to be distributed among the five roman vowel letters. Here is how the WL, LOC/ALA, UNGEGN/EMA, and BGN/PCGN system represent the vowels (in their traditional order).
ɐ,ǝ | u | i | a | e | ɨ | o | |
---|---|---|---|---|---|---|---|
WL | ä | u | i | a | e | ǝ | o |
LOC/ALA | a | u | i | ā | é | e | o |
UNGEGN/EMA | e | u | i | a | e | i | o |
BGN/PCGN | e | u | ī | a | ē | i | o |
There is another minor difference: for some reason, the BGN/PCGN system uses ā to represent the vowel for the 1st order characters which have the /a/ vowel: አ ሐ ሀ ኀ ዐ: Ādis Ābeba.
The consonants that differ in the systems are the following. (The last four columns are not relevant for Amharic but are for some other Ethiopian Semitic languages.)
p' | t' | k' | ʧ' | s' | ʧ | ʤ | ʃ | ʒ | ɲ | ʔ | x,χ | x',χ' | ħ | ʕ | |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
WL | ṗ | ṭ | q | č̣ | ṣ | č | ǧ | š | ž | ň | ’ | k | q | ḥ | ‘ |
LOC/ALA | p̣ | ṭ | q | ċ | ṣ | c | ǧ | š | ž | ñ | ’ | x | q̄ | ḥ | ‘ |
UNGEGN/EMA BGN/PCGN |
p' | t' | k' | ch' | ts' | ch | j | sh | zh | ny | (’) | h(h) | k'(k') | ḥ(h) | ‘ |
The goals of LOC/ALA are to accurately reproduce what appears orthographically in a title or author name, so they do not indicate gemination (because it's not indicated in the orthography) but do distinguish the consonant letters with the same pronunciation (for example, ሀ and ኀ). UNGEGN/EMA and BGN/PCGN have optional ways of distinguishing these letters and, like LOC/ALA, do not indicate gemination.
Here are some desirable properties for a transliteration scheme for non-linguistic articles, in no particular order.
The BGN/PCGN system has several advantages.
BGN/PCGN does not, however, handle the non-Amharic consonants found in other Ethiopian Semitic languages. For these, we could use x for the ኸ series, ‘ for the ዐ series, x' for the ቐ series, and perhaps ḥ for the ሐ series.
Note that BGN/PCGN does deviate considerably from the WL system that people have informally agreed to use for linguistic articles.
So here is the proposal. There are two levels of transliteration, one more precise, with diacritics, and one less precise, with no diacritics. Without diacritics some of the distinctions are lost, but this is common in other transliteration schemes, for example, what is being proposed for Arabic. What I give here is only the more precise scheme.
IPA(WL) | Ethiopic | Transliteration | Comments |
---|---|---|---|
ɐ,ǝ(ä) | ለ መ ... | e | |
u | ሉ ሙ ... | u | |
i | ሊ ሚ ... | ī | |
a | ላ ማ ... | a | Also for ሀ ሐ ኀ አ ዐ, a minor simplification of the BGN/PCGN convention, which would have ā for those. |
e | ሌ ሜ ... | ē | |
ɨ(ǝ) | ል ም ... | [i] | |
o | ሎ ሞ ... | o | |
h | ሀ | h | |
l | ለ | l | |
ħ(ḥ) | ሐ | ḥ | Except for Tigrinya, Tigre, Harari, and Ge'ez, this could be h. |
m | መ | m | |
s | ሠ ሰ | s | ሰ and ሠ need to be distinguished for Ge'ez. |
r | ረ | r | |
ʃ(š) | ሸ | sh | |
k'(q) | ቀ | k' | |
x'(q) | ቐ | x' | Only needed for Tigrinya. Could also be k'. |
b | በ | b | |
v | ቨ | v | |
t | ተ | t | |
ʧ(č) | ቸ | ch | |
h | ኀ | h | Needs to be distinguished from other hs for Ge'ez. |
n | ነ | n | |
ɲ(ň) | ኘ | ny | The usual gn came from the Italian (and French) spelling convention, but I believe this is confusing for most English readers. |
ʔ(’) | አ | ’ | Could be omitted in initial position. |
k | ከ | k | |
x(k) | ኸ | x | Could be h except for Tigrinya, Tigre, and West Gurage. Could be k for Tigrinya. |
w | ወ | w | |
ʕ(‘) | ዐ | ‘ | Except for Tigrinya, Tigre, Harari, and Ge'ez, can be ’ or omitted when initial. |
z | ዘ | z | |
ʒ(ž) | ዠ | zh | |
j(y) | የ | y | |
d | ደ | d | |
ʤ(ǧ) | ጀ | j | |
g | ገ | g | |
t'(ṭ) | ጠ | t' | |
ʧ'(č̣) | ጨ | ch' | |
p'(ṗ) | ጰ | p' | |
s'(ṣ) | ጸ ፀ | ts' | Could also be s'. ጸ and ፀ need to be distinguished for Ge'ez. |
f | ፈ | f | |
p | ፐ | p |
Further:
Here are names of some familiar places and people as they would appear in the less precise version of the proposed transliteration (without any diacritics) and as they appear in the modified WL system ("WL*") that we are using for articles on the languages.
Ethiopic | Proposed Transliteration | WL* |
---|
As you can see, one drawback of the proposal is that it leaves us with two quite different ways of transliterating. I would argue against adopting WL for non-linguistic articles because it uses unusual characters (ǝ, ä) for very common sounds and because it deviates a lot from what people (other than linguists) are used to.