[Journal of the Simplified Spelling Society, J19, 1995/2, p30-36.]
[Note: this page is set to display International Phonetic Alphabet characters. If your browser over-rides this setting, by default or by choice, the IPA characters may not show.]

The Ethiopic Writing System: a Profile

Thomas Bloor.

See Note [1]

Thomas Bloor is a lecturer in Applied Linguistics in the Language Studies Unit of Aston University. He has taught and researched general and applied linguistics and English in Britain and overseas for many years, including twelve years in Africa, eight of these in Ethiopia. He is editor of CLIE Working Papers and co-author with Meriel Bloor of The Functional Analysis of English Edward Arnold (1995).

1. Origins and current use.

The Ethiopic [2] writing system has its origins in the same ancestral writing systems as those of European alphabets, namely the Semitic scripts that proliferated in the Middle East more than three thousand years ago (Coulmas 1989). Little is known about the precise timing and location of the emergence of the earliest Semitic phonetic writing system though speculations abound.

All that seems reasonably certain is that a consonantal script developed among Semitic people on the Eastern shore of the Mediterranean some time between 1800-1300 BC. (Gaur 1987: 88).

Fig. 1 (adapted from Coulmas).

A family tree model of the writing systems (eg, Coulmas: 142; see Fig.1) shows two main branches descending from Proto-West Semitic: North Semitic and South Semitic. Among the descendants in the North Semitic branch are Hebrew, Arabic and Greek (and hence Roman and Cyrillic). Note that we are speaking here of writing systems and not languages. The South Semitic side is usually held to have produced Ethiopic via the Sabean system, which is speculatively dated as emerging in the 11th and 10th centuries BC, but there are dissenting voices. Bernal (1990), rejecting the family tree model, dates the origins of Ethiopic script earlier, relating it to Thamudic, an older script (1990: 64).

The Ethiopic system is used on a large scale in the representation of three Semitic languages, all confined to Ethiopia and Eritrea (the latter being formerly part of Ethiopia but now an independent state). These three languages are Giiz (also spelt Gi'iz, Geez or Ge'ez), Amharic and Tigrinya.

Giiz is a dead language, ie, not the mother tongue of any living person, but it still has a very significant role in the culture of highland Ethiopia as the traditional language of literature and religion. One of Ethiopia's many distinctive features is that it is the only country in Africa whose dominant religion is a non-European form of Christianity dating back to the 4th Century AD. This is embodied in the Ethiopian Orthodox (or Coptic) Christian Church.

The role of Giiz in Ethiopia resembles that of Latin in Europe in pre-modern and even recent times. The liturgy and religious texts, such as books of the Bible, translations of Arabic Christian texts from Egypt and many original Ethiopian writings are in Giiz, including a large body of literature such as the qine /k'ɨne/, poems of remarkable subtlety compiled in the 'wax and gold' mode, the essence of which is a play on words in accordance with strict conventions (Levine 1965). The compilation of qine is a living tradition both within the church (in Giiz) and outside it (formerly in Giiz but nowadays in Amharic and Tigrinya). The wax and gold technique seems to date back only to the 14th century AD, some hundreds of years after Giiz ceased to be a living language in the technical sense; and indeed the heyday of Giiz literature is held by some to be the 14th to 16th centuries. Bender, Head and Cowley (1976: 99) point out in a footnote that "several important Judaeo-Christian documents (such as the full text of the Book of Enoch) were preserved only in a Geez (sic) translation". This classical language is still a source for the coining of Ethiopian literary and technical terms, much as Latin has been in Europe, and, like Latin, it is sometimes perceived as a criterion by which to judge aspects of modern languages.

Giiz inscriptions in the Ethiopic script can be traced back to the 4th century AD when Giiz was the language of the empire of Aksum, a flourishing Semitic civilization based in what is now northern Ethiopia, but with wide military and trading contacts in the neighbouring territories and far beyond. Tablets from that period relating to the Emperor Ezana, the first known Ethiopian convert to Christianity, feature inscriptions in Giiz, Sabean and Greek. Sabean was a language of what is now Yemen, across the Red Sea from Aksum and the putative homeland of the ancestors of the Semitic highland Ethiopians. Ethiopian monarchs down to and including the last one, Haile Selassie (1891-1975), claimed descent from Solomon and the Queen of Sheba (Saba), and are often referred to as the Solomonic line. (Readers will recall that a visit by the Queen of Sheba to Solomon's court is recorded in the Old Testament (Kings I.10; Chronicles II.9). Ethiopian traditional paintings go one better and show them - discreetly - in bed together.)

Amharic and Tigrinya are probably as closely related as, say, Spanish and Portuguese. It is not clear whether they are directly descended from Giiz, as Romance languages are from Latin, or whether all three are the progeny of an earlier language. What is fairly certain is that the Ethiopic writing system was passed on from Giiz to Amharic and Tigrinya. Tigrinya is the majority language of Eritrea and of Tigray, a province of Ethiopia. Amharic has been the language of the politically dominant ethnic group in Ethiopia for many hundreds of years, and, with the exception of one Tigrinya speaker in the nineteenth century, it has been the language of the emperor, the /nɨgusə nəgəst/, literally, 'king of kings' as the Giiz title puts it. It has also been the official language of the state, the day-to-day language of the Church (outside the liturgy, gospels, etc.), the language of primary education as well as a widespread lingua franca (there are others) and the mother tongue of over fifteen million people. [3] Ethiopic script is still the normal medium for newspapers, magazines, novels, poetry, primary school texts, official and legal documents and other printed matter as well as for private correspondence.

There are many other languages in Ethiopia, but most have no established written form. Those that have been written have normally used the Ethiopic script. Representatives of other language groups are currently debating whether to use the Ethiopic or the Roman system. The trend seems to be towards the latter, with possible major long-term effects on Ethiopian culture, but the outcome is still in doubt at the time of writing.

2. The system.

There is some dispute as to whether or not this writing system is a syllabary. In theory, an alphabet has individual symbols (letters) representing phonemes with individual vowel and consonant symbols; a consonantal system represents only consonants, leaving the reader to guess the vowels; a syllabary has individual signs for syllables (consonant + vowel combinations); in practice, the systems are often mixed. Ethiopic essentially uses one character per syllable, exploiting a repertoire of 275 symbols. It is often referred to as a syllabary (eg, Bender 1968, Bender, Head and Cowley 1976), but Sampson (1985) explicitly rules it out of the syllabary category. His argument is that what he considers to be a true syllabary, such as the Mycenean Greek system known as Linear B, has unrelated symbols for phonologically similar syllables; for example, the sign for the syllable /da/ has nothing in common with the sign for the syllable /do/ or the syllable /ka/. The Ethiopic system, on the other hand, can be analysed as thirty-three basic consonant forms with relatively systematic variations to indicate vowels and/or labialization. Thus it could be argued that Ethiopic to some extent resembles other Semitic scripts such as the Arabic and Hebrew consonantal systems, which basically indicate consonants but, for teaching purposes, etc, have developed optional diacritics to signify vowels. However, there is the significant difference that in Ethiopic the variations are obligatory and an integral feature of the symbol, more akin in this respect to some Indian writing systems (Gaur 1987: 100).

Unlike Arabic and Hebrew, Ethiopic is written from left to right. This orientation may have developed via an earlier boustrophedon style in the ancestral form, that is, 'ploughing back and forth' or left-to-right and right-to-left on alternate lines. The Ethiopic system makes no distinction between upper and lower case letters and has no conventional cursive form, though, of course, rapid handwriting can result in an ad hoc cursiveness and often a lack of clear distinctions. Unlike Hebrew and Arabic, there are no systematic variations in the form of the symbol according to its position in the word.

Word boundaries were originally unmarked and later were indicated by two vertically placed dots like a colon, ...:..., though, with foreign influence, letter spaces are now often used instead of the traditional symbol. A sentence boundary is indicated by four dots ...::, and, less frequently, what would in Roman script be a comma by the symbol ...+. The old form of the question mark, three vertically placed dots, has been largely superseded by the question mark ?. Quotes are usually in the French style <<...>> and parentheses and exclamation marks are as in the Roman system: ( ... ), !. The last three are modern innovations.

To simplify somewhat, the system works on the basis of thirty-three base symbols with seven 'orders', representing seven vowels, for each base symbol. (Out of context, the first order is the normal reference term for each set of items, its 'name', as it were.) To consider some examples, the seven orders of the syllables identified with the consonants /b/, /l/, /k/ and /z/ are as in Fig. 2:

Fig. 2.

Key: Note: correspondences are very approximate.


It is easy to see that there is a consistent pattern in the shapes given in the chart where:

• the 1st order is the base symbol with no explicit vowel indicator (though a vowel is pronounced); C(onsonant) + /ə/;
• the 2nd order is the base symbol modified by a projection half-way down the right leg; C+/u/;
• the 3rd order has a projection at the base of the right leg; C+/i/;
• the 4th order has a short left leg; C+/a/;
• the 5th order has a loop on the right leg; C+/e/;

The 6th and 7th orders are less systematic, but some regularity can be discerned:

• the 7th order corresponding to /bo/, /ko/ and /zo/ all feature a short right leg;
• the 6th order corresponding to /bɨ/, /lɨ/, /kɨ/ and /zɨ/ all have something happening on the left though it is something different in each case.

This gives a rough idea how the system works. There is considerable regularity of letter shapes, but some orders are more regular than others. The shapes are most consistent in the 5th order, slightly less in the 3rd, slightly less again in the 2nd, still less in the 4th and even less in the 7th. The 6th order is least consistent, with the greatest number of patterns, so that the form is largely unpredictable although in the entire system some clusters of similar 6th order patterns can be found.

Fig. 3

Unfortunately, the system is not quite as regular as the examples in Figure 2 suggest. Thus the set of syllables with /g/ starts off regularly enough but is unpredictable in the 4th, 6th and 7th orders. The /r/ set is even more irregular and /w/ is especially unpredictable with regard to the 2nd and 6th orders. (Fig. 3.)

The system has other kinds of regularity, however. For example, one form of /s/ (as in English see), written has a corresponding form for /ʃ/ (as in English she), phonetically its palatal equivalent (Fig. 4). Once the learner knows the forms of one of these, the forms of the other are entirely predictable.

Fig.4

Moreover, the same phonetic relation of palatalization works equally well for the pairs /z/: /ʒ/; /t/:/tʃ/; /d/: /dʒ/; /t'/: /tʃ'/(Fig.5). The last pair are ejectives:, where the speaker pronounces a 'glottal stop simultaneously or almost simultaneously with another sound.' (IPA 1949: 17); /t'/ is phonemically distinct from /t/; /tʃ'/ from /tʃ/, etc, in Amharic and many other Ethiopian languages.

These parallel forms indicate that somewhere in the history of the development of the system there was considerable phonetic expertise involved, either conscious or intuitive. Naturally, most learners of this system know nothing of these formal phonetic categories, but the correspondences are to some extent identifiable by intuition.

Fig.5

Of the 33 base forms, two represent vowels in isolation and the rest consonants and the semivowels /w/ (as in English well) and /j/ (as in yes) (henceforth classed as consonants). In addition to the (33 x 7 =) 231 major symbols, there are 44 additional variants for labialized consonants (plus vowel), that is, syllables involving consonants with lip-rounding, for example, /k_wa/.

3. Redundancy and ambiguity.

In Giiz times, the system added two extra symbols (ie, 2 sets of 7) to cope with two new sounds (/p/ and /p'/), originally required for use in ecclesiastical Greek and Latin borrowings and names (eg, Paulos, Paul); these now show up in loanwords like /polis/ (police) and, of course, in the name of the country: /itjop'ja/. Scholars who believe that Ethiopic is derived from Sabean claim that when Giiz adopted the Sabean system, it dropped a number of unwanted symbols. This did not happen when Amharic and Tigrinya took up the system from Giiz. They kept all the symbols (later adding 8 more for consonants not used in Giiz, and even later adding a modified form of the /b/ symbol to represent the foreign phoneme /v/). The result is that there is a considerable amount of systemic redundancy, particularly in the case of Amharic, which lacks several consonant sounds found in the phonology of Giiz. Thus, 4 distinct sets of 7 can represent the sound /h/ + vowel: 2 sets represent /s/: and 2 /s'/ (ejective) The 44 labialized consonant symbols (/k_wa/, etc.) are also arguably redundant, wholly or partially, and there is considerable variation in the spelling of many words that may involve them.

In addition to the irregularity in its forms, the 6th order presents another problem. This is the systemic ambiguity of its representation: it can stand for either a consonant in isolation or a consonant with the short vowel /ɨ/. For example, can represent either the vowelless consonant /b/ at the start of the name Bloor, or /bɨ/ (similar to /bι/ in English bit /bιt/). In reading, this is a problem largely for foreign users of the system since there is rarely, if ever, any ambiguity about which word is intended. It could indeed be regarded as an advantageous economy in the writing system since in pronunciation C+/ɨ/ and C+zero (ie, vowelless consonant) alternate fairly freely without affecting meaning; the /ɨ/ tends to creep in between two consonants to maintain the Consonant-Vowel pattern with lack of consonant clusters that prevails throughout Ethiopian languages. (This is known to phoneticians as an epenthetic vowel.) This ambiguity may occasionally present spelling problems, though; Bender, Head and Cowley (1976) cite the example of (dove, pigeon), which is variously pronounced /rɨgɨb/ and /ɨrgɨb/ but normally spelt as above, ie, with an initial /r/ equivalent: /rɨgɨb/ rather than as ie, with an initial vowel symbol: /ɨrgɨb/, which is equally plausible.

Usually, a word-initial vowel is represented by one of the two basic sets reserved for consonantless vowels (as in the words for Addis or Ethiopia). These occurred in Sabean (and probably old Giiz) as consonant symbols, but as the consonants in question were not required for Amharic, the two sets became used for vowels alone. One set would have been enough, of course, and even inside the sets there is redundancy since, with consonantless vowels and /h/, the 1st order vowel is pronounced identically to the 4th order: /a/.

Fig. 6

Thus the name of the capital, Addis Ababa (/addis abəba/, literally 'New Flower') is written as Characters for Addis Ababa using the 1st order form for /a/. (Note: the second syllable in Ababa is not stressed). The two consonantless vowel sets are as in Fig.6:

Transcriptions of foreign names might present difficulties for reading aloud: for example, Bloor (in my pronunciation /bluə/ - or /bluəɹ/ before a vowel) is not a problem for Amharic phonology unlike, say, Smith, which has a non-Amharic word-initial consonant cluster /sm/ and a non-Amharic consonant /θ/, or Upward (with its non-Amharic initial vowel). Bloor could be transcribed as but this in principle could be read as /bluwr, bɨluwr, bɨluwɨr/, etc. However, this is hardly more variable than the pronunciations the name receives from English speakers, and, even without undue modesty, cannot be conceived of as an issue for Amharic spelling.

Slightly more significantly, loanwords involving vowelless consonants produce some variations in spelling and in pronunciation: the English word carbon (RP /kɒbən/) is normally transliterated as (Bender, Head and Cowley, 1976: 127) but could be read as /karbon, karɨbon, ɨkarɨbon/, etc. In fact, a more phonetic transliteration would be since the 1st order vowel is closer to the English neutral vowel sound in the second syllable than is the 7th order /o/ vowel, but Amharic transliterations tend to follow English spelling rather than English pronunciation, with the result that when spoken they are sometimes unrecognizable as words of English origin. Once again, though, this is not likely to be a problem for Ethiopians.

The greatest problem for the reader (especially the foreign reader) is perhaps the failure of the system to indicate gemination (the 'doubling' or 'lengthening' of consonants in pronunciation). In Amharic, gemination is phonemically meaningful, as it is in many other languages such as Italian (eg, Italian mola 'grindstone' versus molla 'spring'), but not in English. This phenomenon does occur in English, but, as it is not within a single morpheme, it is not classed as gemination. We find it in the difference between the /l/ sounds in (1) cool leaves and (2) cool eaves. There are numerous cases of minimal pairs of words in Amharic where the only difference in pronunciation is the presence and absence of gemination, eg, /alə/ ('said') and /allə/ ('is present'), but this difference is not reflected in the orthography since both are written Amharic speakers, of course, perceive that there is a difference, but, unless they are phoneticians, they do not perceive the feature as a doubling or lengthening of the consonant so much as a matter of slight stress variation, which is in fact sometimes a corollary of gemination.

4. Transliteration into Roman.

Most basic consonants transliterate without problems into the Roman alphabet. No distinction is usually made between ejective and non-ejective pairs, though sometimes a systematic distinction is made using K and Q to distinguish between and /kə: k'ə/ Most of the other consonants are fairly straightforward.

The English spelling of the first word in Addis Ababa accidentally represents the gemination feature better than the Amharic spelling, which does not distinguish between /d/ and /dd/. But the fact that double letters do not normally indicate gemination in English orthography means that this transliteration has no significance for English readers, though it might well have some for Italians. However, in language teaching books for foreigners, dictionaries, scholarly papers and so on, the double letter convention can be - and sometimes is - easily implemented with a clear explanation. Unfortunately, the frequent wrong placement of stress and lengthening of the vowel in the second syllable of Ababa by English and other foreign speakers could not easily be contra-indicated by alternative transliterations without the use of stress-diacritics.

The question of transliteration of vowels is perhaps the most difficult. There are more than seven vowel sounds in the phonology of Amharic, just as there are more than five in English; Armbruster's (1908) classic study of Amharic lists eighteen. Even so, the Amharic vowel system is fairly well represented by the seven orders in Amharic but is difficult to represent with the resources of the Roman alphabet. For example, there are not two, but three, distinct vowel sounds in the Amharic pronunciation of the name Addis Ababa: /i/, /a/ and /ə/. Of the vowels represented by A in the Roman rendering, all except the middle one in Ababa are the 4th order /a/ (close to a Northern English /a/ in bad), but the middle vowel in Ababa is closer to the 'neutral' schwa /ə/ than to an /a/. (In fact, in some Ethiopians' casual pronunciation the middle vowel is almost lost and the B sounds become a fricative /ß/, between /b/ and /v/ as in the Spanish haber, but lengthened or repeated.) When the stress is roughly right, the English pronunciation of the name accidentally gets a reasonable approximation to the Amharic 1st order vowel because unstressed vowels are usually neutralized in English, but this also means that the final vowel gets neutralized too, which is not the case in the Amharic pronunciation.

Sometimes this 1st order vowel, which I have very loosely transcribed as /ə/, is transliterated by an E: Abeba; but this tends to produce the English reading /ə'beibə/ with stress on the second syllable (also found in Italian and French renderings). The city which used to be written in Roman script as Harar now tends to be transliterated as Harer to avoid the common mispronunciation /hə'ɹɒɹ/ (US) or /hə'ɹɒ/ (RP), with heavy stress on the second syllable, but it remains open to misinterpretation. ( is 1st order /rə/ and is 6th order /rɨ/ or, as in this case, /r/ without a vowel.) This discussion is not intended to suggest that English should try to emulate the Amharic pronunciation exactly; after all, we do not pronounce Paris in the modern French manner. It is simply meant to illustrate the problems of transliteration with a familiar example.

The 3rd order vowel /i/ (as in English see) is fairly straightforwardly transliterated by I, as in Addis, but then the 6th order vowel /ɨ/ in /dʒɨmma/ (the name of a city) also has I in the transliteration Jimma. In other instances, this 6th order vowel may be transliterated as E; for example, in the common spelling of the language name Giiz /gɨʔɨz/ as Geez (/ʔ/ indicates a glottal stop). Mockler, a historian, uses the spelling enjerra for the noun /ɨnjəra/ (the staple bread of Ethiopia), ie, he uses E for 1st and 6th order vowels; but he is cavalier about transliteration on the grounds that: "Amharic names won't go into English exactly; their alphabet has too many consonants and vowels" (Mockler, 1986: xiv). Frydenlund and Svensen in their Amharic primer (1967) and Levine also use E for the 6th order, if more methodically.

In the usual Roman alphabet version of the name of the last Ethiopian emperor, Haile Selassie the first E represents the 1st order vowel (approximately /ə/); the second E represents the 6th order vowel (/ɨ/); and, oddly, the digraph IE represents the final vowel, which is the 5th order vowel /e/ (somewhere between the E in English bed. and the É in French blé). So it seems that here the strongest candidate for transliteration as E is the one that does not get a straight E; perhaps this is to avoid its interpretation by English (or French) readers as a silent E: /səlas/. (In fact, the usual French rendering is Selassié, which has the last vowel about right but includes a redundant I, leading to another mispronunciation.)

A source of great confusion for English speakers reading about Ethiopian affairs is the pair of terms Tigray (a province of Ethiopia) and Tigre (a province in Eritrea). [4] The confusion is worse confounded by the frequent transliteration of both as Tigre, whereas, besides being pronounced differently by Ethiopians, the Ethiopic written form clearly distinguishes them: /tɨgray/and /tɨgre/ hence the most recent transliteration given above.

One way of indicating these distinctions is to use diacritics. Thus Ä may be used to represent the 1st order vowel: hence Abäba; but Levine uses Ä for the 4th order vowel and A for 1st order. The 5th order vowel /e/ is usually represented by E but is sometimes represented as É: hence, Tigré. Frydenlund and Svensen use É in this way, and, as already stated, reserve unaccented E for the 6th order. The tilde is sometimes used, as in Spanish, to indicate the palatalized /n/: hence Tigriña instead of Tigrinya. (Possibly under Italian and/or French influence, the digraph GN is commonly used: Tigrigna.) Dawkins (1960) eccentrically but effectively places the numerals 1 to 7 as diacritics over the letters E, U, I, A, E, I, O for the seven vowels.

Alternatively, phonetic symbols can be exploited. However, even when this is done, there is enormous variation. Obolensky et al. (1964) use /ə/ for the 1st order vowel (as I have done), but Bender, Bowen, Cooper and Ferguson (1976) use /ɛ/. For the 6th order vowel, Bender et al. use /ɨ/ and Obolensky et al. /ι/, but /ə/ is used by some writers.

5. Spelling reform?

It has to be said that Amharic orthography using the Ethiopic syllabary (or 'consonantal system' as Sampson would have it) seems much more regularly phonetic than English orthography (or at least, pace Chomsky and Halle 1968 or Stubbs 1986, the regularities are more transparent). However, it is also true that the task of learning over 270 symbols is daunting, and, even if we allow for the systematic regularities in the vowel indicators, palatalization markers, and so on, the memory load is quite considerable. (I have omitted discussion of a number of confusing exceptions.)

The practical case for reform is fairly obvious and there are simple improvements which leap out at the impartial observer. The reduction of the four /h/ sets to one would immediately reduce the overall number of symbols by twenty-one (since there are seven versions of each consonant). Loss of the duplicates for /s/, /s'/ and /a/ would free the system of a further twenty-one.

With more difficulty, the seven vowel-indicator shapes could be virtually standardized for all sets. Gemination could be indicated by doubling consonants, or perhaps by a diacritic, since two consonants together would invite an intrusive /ɨ/ in the reading. Granted, as stated above, this last is a much greater problem for foreigners than for the Amharic speaker, and orthographies do not usually develop for the convenience of foreigners. Even so, as long as Amharic retains its special national status, the problems of second language learners need to be considered since the majority of Ethiopians are not mother-tongue speakers of Amharic.

The forty-four symbols indicating labialized consonant+vowel syllables might, as sometimes happens already, be replaced by the use of the symbol for /w/ after the relevant consonant or an added vowel /o/. Some of these symbols are already becoming obsolete without any deliberate action being taken. On the other hand, they are unambiguous and economical in terms of the number of symbols required to write a given word; for example, in the traditional way, /k'_wank'_wa/ ('language') is written but with the suggested change would be More drastically, the system could be converted to a simple alphabet with one symbol per phoneme instead of one per syllable; thus we could have the 1st order symbols representing vowelless consonants and use the seven forms of as the vowels, giving a total of twenty-seven letters. This would make the system vastly simpler, but, among other disadvantages, it would result in some very long written words since Amharic words have on average more syllables than English. How serious this disadvantage would be is a moot point since, for example, the current system involves two strokes on the typewriter for most symbols; one for the base and one for the vowel indicator. It just takes up less space on the page. In any case, as reformers have found elsewhere, such considerations are fairly insubstantial when set against the weight of centuries of tradition.

Such suggestions have been considered over the years. Cowley (1967:1) refers to an Amharic document of 1940 (Ethiopian Calendar, ie, 1947 Gregorian Calendar) where the reform of the alphabet "received a thorough airing". Cowley's scholarly article concerns "logical standardizations, not plans for radical reform" (ibid.). Conceding briefly that the simplest solution for the multiple forms would be to use the same form in all instances, so reducing the total number of symbols in the system, he goes on to argue, on the assumption that all the symbols will be retained, for using Giiz spelling for obvious Giiz words wherever possible as the basis for selection among the four /h/ symbols, the two /s/, the two /s'/ and the two consonantless symbols. On other issues he argues variously on etymological, grammatical and phonological grounds, and seems to opt for clarity rather than economy. The article indicates that alternative spellings occur for a considerable number of individual words in written Amharic.

Under the pre-1991 military government, the Ethiopian Academy worked on possible reforms but did not implement them, and since 1991 the new government has sponsored investigation of the issue. Presumably, these efforts have come to nothing for the usual reasons: among others, natural human conservatism and emotional attachment to traditional modes (not a trivial matter); the fear of losing access to the body of literature, which is here further intensified because of the intimate connection between writing and religion; and the immediate practical difficulties and expense of making the changes. Even more significant perhaps is the relative lack of urgency of the problem since from 1971 to 1991 - and before, though to a much smaller extent - Ethiopia was in an almost constant state of war, civil strife and famine, and since then has been undergoing major reorganization - conditions which may push even spelling reform down the agenda. It is the post-revolutionary period that is often a time for innovative language planning (Yule 1994) and Ethiopia is no exception to this trend. Thus, reform at some time in the near future remains a strong possibility.

References.

Notes.