Quechua Alphabets:

Which is Correct?


The View from Down Under


   ver esta página en castellano  


Article on the Quechua alphabets issue from the perspective of similar problems in devising spelling systems for Aboriginal languages in Australia, by the Australian linguist Gavan Breen

This article is written particularly in response to this article in Spanish by Daniel Cotari Gutiérrez, in which he criticises the official Bolivian & Peruvian spelling systems for Quechua, and advocates instead what he calls a ‘phonetic’ spelling system (with five vowel letters, for instance)

(While not in complete agreement with all the details of the arguments put forward here, the author of this Quechua website has offered to host this page for Gavan so that he can present his case, on the strength of his valuable  experience of precisely the same type of problem that presents itself in Quechua too.)


Back to Homepage



I am interested in a small way in the Quechua language (or languages), having travelled as a tourist in Bolivia and Peru in 2001. I observed there something of the variety of orthographies for the language; for example, I bought a small dictionary in which the name is written Kechwa; I saw another with the name Qheshua.

I read recently the articles on the orthographies in the web page http://members.tripod.com/~jlancey/, and I would like to make some observations especially on the article by Daniel Cotari Gutiérrez. I realise that it is several years since these articles were written, and that other linguists have replied, although perhaps not specifically. However, as an Australian linguist with much experience in the design and teaching of orthographies for Australian indigenous languages, I would like to offer a contribution from a perhaps different point of view.

It seemed to me, from my reading of Cotari’s article with very little other background knowledge of the topic, that both he and other protagonists of the five-vowel spelling system, and also the protagonists of the “normalised Quechua spelling system”, lacked a full understanding of the concept of the phoneme. This perception was based not on any understanding of Quechua, which I lack, but on general principles of linguistics. Having learnt more of the facts now (thanks to Dr Paul Heggarty) I have had to revise my perception of the protagonists of the normalised system, whom I now accept to be very competent in their linguistics.

Cotari, in his preamble, refers to “a system of spelling in which each sign (letter) expresses faithfully a real sound”. But this is impossible. Each time you pronounce a word, you pronounce it differently, minute though the differences may be. There is no such thing as a practical phonetic spelling system; a phonetic transcription is used for specific purposes by linguists who are studying a language whose phonemes they do not yet know or to describe the pronunciation of a language or justify their analysis of the phonology. In fact, a writing system must be a system in which each sign (a letter, or sometimes two, like sh or ng, or even three, rarely in some languages) expresses faithfully a class of sounds which monolingual native speakers of the particular language hear a single sound. What Cotari and others call phonetic spelling is by no means a phonetic spelling; it does not represent the pronunciation of words in all their fine detail, but groups sounds into classes according to close resemblances in order to reduce the number of sounds to a manageable figure. This is what phonemic orthographies do, but the so-called phonetic spelling of Cotari for Quechua does not make this grouping of sounds according to the structure of the Quechua language, but according to the structure of Spanish, with a few modifications.

We do not have here two types of transcription, one phonetic and one phonemic, but two types of transcription, one based on the structure of Spanish, the other on the structure of Quechua, not of one particular form, but taking into account the differences between the different forms and designed to be used easily by all.

I will deal with the issues, as presented by Cotari, in three sections:

1) The use of vowel letters which represent sounds which are not phonemic in Quechua;

2) Variants of consonants in initial and final position in syllables;

3) Spelling morphemes according to their pronunciation in a postulated proto-Quechua language.

I will occasionally employ the linguistic conventions of writing letters between <>, as for example <a>, phonemes between //, like /i/, and allophones between [], as [e]. In the last case I use a phonetic alphabet; where these letters appear in green, this is because I’ve had to use the symbols of the International Phonetic Alphabet, and for this I use the Arial Unicode MS font, which comes preinstalled on most modern computers.


1) The Vowels

Cotari says, for example: “It is written as i but it is pronounced as e.” This means: “It is written as <i>, but it is pronounced as /e/ in the hearing of a native speaker of Spanish.”  If Quechua has three vowel phonemes – as suggested very clearly by the examples given by Cotari and his declaration that [e] and [o] are conditioned by a neighbouring consonant, and as is also the opinion of many linguists – native speakers of Quechua who write their language with a five-vowel orthography cannot write this phoneme according to its pronunciation, because the pronunciation of <i> in [ima] and of <e> in [qena] are, for them, the same. On the contrary, such monolingual speakers have to pause a moment to think about what the neighbouring consonants are, because the pronunciation follows a rule that says:

“When this sound, /i/, is contiguous with one of the consonants q, qh, q’ (or in whatever other form these phonemes might be written, such as jj or even j) or separated from one of these consonants solely by one of the consonants r, ll, m, n, y, w (and others?), it is necessary to write e, otherwise write i.”

Alternatively, they must learn by heart all the words in which /i/ is written as <e>, and similarly those in which /u/ is written as <o>.

In fact, a sentence like, say, “It is written as i, and it is pronounced as i” uses i in two ways, as the expression suggests: first as the name of a letter of the alphabet, <i>, then as the representation of a sound, [i]. This same letter functions as the name of a phoneme, /i/. So it is possible to describe the situation of the sound [e] in Quechua when a three-vowel orthography is used: “It is written as <i>, but it is pronounced as [e]” and, much more important, “It is written as <i> and it is pronounced as /i/”. (Remember that /i/ does not represent a single sound but a class or group of sounds.) This is the situation for many words in Quechua in which the phoneme /i/ is pronounced as [e].

Let us make a comparison with English. In this language, the phoneme /n/ is pronounced [ŋ], which is written as <ng> in many languages, when it occurs before a velar consonant – /k/ (written as <k> or <c> or <q>) or /g/. (The situation is quite similar in Spanish, except that, unlike in English, there are no circumstances in which [n] can contrast with [ŋ], as it does in sinning compared to singing.) Using the logic of the supporters of the five-vowel orthographies, and using phonetic spelling, I would say that you must write <ng> in this situation; for example, bangk for bank, ingquiry for inquiry, angger for anger, tanggle for tangle. This would be patently ridiculous, but, adapting the expression of Cotari, in bank, for example:

“It is written as n, but it is pronounced as ng.”

It is ridiculous because ng [ŋ] is clearly an allophone of and not distinct from /n/, and, consequently, English speakers (or almost all English speakers) do not hear the difference between /n/ before k (etc.) and /n/ in other contexts, and would have to follow a rule like:

“When this sound, /n/, occurs before k, q, c (if a, o, u or a consonant letter other than y or h follows) or g (if a, o, u or a consonant letter other than y follows) it must be written as ng.”

This contrast between [n] and [ŋ] before /k/, which does not exist in English or Spanish, does in fact exist in many, perhaps all or nearly all, Australian languages; for example, wangka [waŋka] and wanka [wanka] are different words in some languages.

There are few examples of this type in Spanish, because it is closely related to the language for which our Latin alphabet was devised. There are many more examples in languages more distant from Latin, like the languages of the Andes or the languages of Australia. But even for French the system of five vowels is quite unsuitable; it has about fifteen vowel phonemes.

Many Australian languages have a system of three vowels, and I know no orthography nowadays which tries to write these languages using five vowel letters. I know one language in which missionaries formerly used such an orthography; it was easy to read, because the readers were reading religious songs and readings which they knew and needed an orthography just exact enough to remind them of the words. But to read unknown material or to write was very difficult. A native speaker wrote eight stories which were published in booklets; each had in its title a word which meant ‘story’, but in the eight books this word was written in seven different ways (and only one of these involved a typographical error).

I should add, however, that Spanish words which have not been adapted to the Quechua sound system in their pronunciation should be spelt in the Spanish way.


2) Consonants in Different Positions in Syllables

I am relying in parts of this section on some background information from Paul Heggarty, especially regarding [s] and [ʃ].

Clearly, for some consonants, it is a case of complementary distribution as for [n] and [ŋ], or for [d] and [ð], in Spanish. If, for example, [p] occurs only in the onset (beginning) of a syllable and the corresponding fricative [ɸ] occurs as the coda (ending) of a syllable, as it does in some Quechua dialects (which I will call, for simplicity, the southern dialects), the situation for this pair is similar to that in Spanish in which [d] occurs only in initial position in a word (or, more correctly, after a pause) or after [n] or [l], while [ð] occurs elsewhere. (In English, of course, [d] and [ð] belong to different phonemes; for example, despite the spelling, the only difference between day and they is that the former has [d] and the latter has [ð] ) Both [d] and [ð] belong to the phoneme /d/ in Spanish and are written with the letter <d>. In the same way, the phoneme /p/ in these Quechua dialects, sometimes pronounced [p], sometimes [ɸ], can always been written with the same letter, and <p> is suitable for this.

An Australian comparison is with the situation for the plosives in many indigenous languages; for example (with some simplification), the consonant written <p> (in many languages, <b> in others) is heard, by a native speaker of a European language such as English or Spanish, as a voiced plosive [b] in the combination [mb] but as closer to the voiceless consonant [p] elsewhere. Native speakers of these languages do not hear this difference (and in fact when speaking English often get them the wrong way round, calling a certain service station PB instead of BP, for example). The orthographies used respect this situation, and combine the voiced and voiceless sounds with a single symbol. The same applies in all these languages to the other three, four or five (according to language) plosives.

In the southern Quechua dialects the same situation as for the pair [p]-[ɸ] applies to the pair [q]-[χ], which is assigned to the phoneme /q/ and written with <q>. However, the situation is not exactly the same for some of the plosive/fricative pairs, because the fricative members of the pairs do occur in syllable-initial as well as syllable-final position and these sounds are clearly phonemes; thus one could write such a sound with the appropriate letter representing the fricative phoneme wherever it occurs. This applies to the sounds [s] and [x], which, when they occur initially in a syllable, are assigned to the phonemes /s/ and /h/ and written with <s> and either <h> or <j> (in different areas).

But [s] is the fricative corresponding to [t] and [x] is the fricative corresponding to [k], and words which have syllable final [s] in the southern dialects have [t] in that position in the northern dialects and words which have syllable final [x] in the southern dialects have [k] in that position in the northern dialects. Since the aim is to have a spelling system applicable to all dialects it is consistent with the approach adopted for [ɸ] and [χ] to group these syllable-final fricatives with the corresponding plosives and write them as <t> and <k>. These syllable-final letters are automatically pronounced as fricatives by speakers of southern dialects.

The situation for [ʃ] is slightly different again. [ʃ], and also in some cases [s], occurs in syllable-final position in words in which the other dialects have [] (ch) there. But, also, there is a single morpheme which in a few dialects can give rise to a contrast between [s] and [ʃ]. Without this it would be obvious that the situation of [ʃ] is the same as that of [ɸ] and [χ]. This very minimal distinction seems not to be a sufficient justification for introducing <sh> into the spelling system, with all the confusion that would cause in the other dialects.


3) Use of Proto-Quechua Forms

In this case, my initial response was to say that I agree with Cotari. The characteristics of a remote ancestral language are not relevant for the spelling of a modern language. It is necessary to write words according to their current form, and not according to their form in a past age. When there many allomorphs, as in the example of the ‘progressive’ given by Cotari, it seemed to me that it is reasonable to select one as ‘official’, but writers should be able to write them in their own dialect.

However, Paul Heggarty informs me that Cotari is not correct here, and that the spelling is not based on historical forms but on pronunciations still used in some areas. So it seems that Cotari is wrong again, for lack of knowledge if not lack of logic.




Back to Contents

Back to Homepage