RSS feed

Singing in Unknown Languages: a small exercise in applied translation theory

Brian Mossop, York University School of Translation, Toronto


When choirs sing in languages unknown to most of their members, they are faced with the questions: what do these words mean, and how do I pronounce them? Translation theory can help provide practical phonetic and semantic aids to choir members. Catford’s notions of phonological translation and transliteration are extended to solve the phonetic problem. The semantic problem is solved by writing multiple translations into the singers’ scores.


Choral music, phonetics, spelling, transliteration, meaning levels, applied theory.

1. Phonetics and semantics in choral music

In the English-speaking world, choral singing is an extremely popular pastime and, interestingly, amateur choirs often sing works in languages which most members of the choir do not know. The question thus arises: What role does translation play and what role could it play as a choir rehearses and performs works in unfamiliar languages?

This is a question which, as far as I know, has never been addressed in the literature on music and translation. Most of that literature is concerned with translations which will themselves be sung—not with aids to singing in the original language of a work. The remainder of the literature is about opera surtitles, which are aids for the audience, not the singers.

To begin, a few words about musical communication. Speech communicates by relating sound to conventional linguistic meanings (semantics), but music communicates directly through sound. This is obvious with instrumental music, but choral music is no different. For the professional musicians who conduct amateur choirs, the human vocal apparatus is simply another instrument that produces sound. For them, the phonetic aspect of linguistic signs tends to be more important than the semantic aspect. To put it another way, musical meaning in choral music is conveyed more through phonetics than through semantics.

It is important to bear in mind that composers of choral music are setting words to music; that is, the words are given, and the composer then creates the music, taking the phonetics of the words into account1 as one aspect of the resulting overall sound. Consequently, failure to pronounce words correctly may result in a departure from the composer’s musical intent.

In addition, conductors need to focus on phonetics because of the difficulty of learning to produce the exact sounds required to give the right musical effect. Even when singing in English, choir members tend to use pronunciations from their normal speech, which are often not appropriate. Thus a conductor may focus on the quality of a vowel (“give me a pure Italian e,” rather than the vowel of English great with its diphthongal quality) or the treatment of a consonant (“spit out that final ‘t’ in caught” to convey anger, or “leave out the ‘r’” in the most common Canadian/American pronunciation of words like far, because it is ugly). Furthermore, if enunciation is not much more careful than in casual conversation, the audience may not be able to identify a word, and its semantic content will then be lost.

The relative importance of the semantic aspect of the choral text varies with genre. In the English-speaking world, semantics is quite important in folk songs (whether sung in English or some other language), in popular music and in Broadway musicals. With operas, when sung in the original language, many audience members are primarily interested in the emotions conveyed by the singers’ voices, though the use of surtitles beginning in the 1980s does indicate a demand on the part of some opera-goers to know the detailed semantics of the singer's German, French, Italian or Russian words while they are being sung, rather than just having a printed plot summary. With musical settings of the Latin mass, when the choir sings agnus dei qui tollis peccata mundi, it is doubtful that many members of the audience (or indeed many members of the choir) know—or even care—that this means ‘lamb of god who bears the sins of the world’. They came to the performance to hear an interpretation of Mozart’s or Bach’s music, that is, the sounds produced by the combined instruments and human voices.

How then is semantics typically treated during rehearsals of works to be performed in a language which most choir members do not know? Based on my experience of nearly 25 years in six different amateur choirs2, what happens is that either the conductor or a choir member who does know the language stands at the front and goes through the song, pronouncing a phrase at a time, which is then imitated by the choir. Typically this happens just once, at a single rehearsal early in the preparation of the work. Occasionally a quick sight translation is given, though one doubts that many people remember the translations of individual phrases since there is no time to write them down. In later rehearsals, the conductor or a knowledgeable choir member may correct the choir’s pronunciation, though this tends to be haphazard. Meanwhile the semantics are typically never discussed again, or discussed only at whole-text level (“keep in mind that this song is about war and death”)3.

The scores distributed to choir members may contain translations. Sometimes the translations appear on a separate page, either alone or side-by-side with the source text. In other cases, the translations appear under the musical notation (on a separate line below the source text, or sometimes as a replacement for the source text). However these translations are designed to be sung; they are not usually much use as a semantic aid by those singing in the source language. That is because the translator had to choose words with the right number of syllables to fit the notes, and the right stress pattern to match the rhythms of the music. As a result, the translations are often extremely free, but even when they are closer, they typically do not follow the precise word order of the source, because of grammatical differences between the languages. The word which appears under a given source-language word will only be a translation of that particular word by accident. A singer who is rehearsing to sing in the source language will thus typically not find out, by consulting the translation, the semantics of the word currently being sung.

This state of affairs would seem to create a problem if we compare the situation to one in which the choir is communicating to an audience in a language that is known to all its members. For then, each member knows the semantics of each word as it is being sung, and presumably this helps the choir to communicate with the audience.

Look at Figure 1, which shows a page from the vocal score of Rachmaninoff’s 1902 composition Vyesna, commonly known in English as the Spring Cantata (Rachmaninoff n.d.: 6). The words for the four choral voices to sing appear under the musical notation, first in untransliterated Russian, and then below that in German translation. The translation is very free: whereas the Russian source text means ‘a green noise is coming, is buzzing,’ the German means ‘spring is coming, spring is drawing near.’ If a German-speaking choir were singing this in Russian, the singers would be quite mistaken if they thought that shoom — the Russian word at the beginning of the second last line of music — means the same thing as naht (‘is drawing near’), this being the German word that is found under shoom; the Russian word in fact means ‘noise.’

Mossop 1

Figure 1. Passage from Rachmaninoff’s Spring Cantata, with words in original Russian and German translation. © Copyright 1903 by Hawkes & Son (London) Ltd. Reproduced by permission of Boosey & Hawkes Publishers Ltd.

2. Translation theory applied — Semantics

The subtitle of this article is ‘a small exercise in applied translation theory.’ It is ‘applied’ in the sense that I will not be reporting observations of singers at rehearsals or of audience members at concerts, and I will not be setting out the results of surveys of choir members, conductors or concert-goers. Rather I will be describing how, in one particular instance, I was able to apply translation theory to the two practical problems confronting singers: How do we pronounce these words? And what do they mean?

Translation theory has a great many things to say about the semantic aspect of language. One of these is that meaningful units exist at several levels: words, phrases, sentences, and larger units each have ‘meaning’ in some sense. When someone sings in their own language, they grasp meaning at each of these levels. Ideally, then, a semantic aid for choir members singing in a language they do not know will convey meaning at more than one level.

2.1 Word-level meaning

If singers are to communicate successfully to the audience, it is best that they know the meaning of the particular word they are currently singing. For this purpose, a word-for-word gloss is needed. Glosses are metalinguistic rather than metatextual in nature; that is, they are explanations, in the target language, of the ‘dictionary meaning’ of the source-language lexical items. Glossing in this sense cannot of course be completely context-free, since the dictionary will very often give more than one sense for a word, and a suitable selection must be made. Still, it is best to choose the most common equivalent for the sense that is relevant. Gloss French pauvre, in the sense ‘lacking money,’ as poor, not impoverished, penniless or broke.

Look at Figure 2, which shows what the score seen in Figure 1 looked like after I whited out the Russian and German texts and wrote a variety of wordings on the page4. Under the musical notation can be seen a phonetic representation to guide the choir’s pronunciation. The nature of this representation, which is not a traditional transliteration of Cyrillic, will be discussed later in the article.

Mossop 2

Figure 2. Passage from Rachmaninoff’s Spring Cantata, showing phonetic representation (top left), word glosses (middle left), sentence-level meaning (bottom left) and contextual meaning (right). © Copyright 1903 by Hawkes & Son (London) Ltd. Reproduced with changes by permission of Boosey & Hawkes Publishers Ltd.

At the top left of the page, the phonetic representations are repeated, with an English gloss under each word. The glosses could also have been placed in a second line under the musical notation, but I wanted to avoid cluttering the score and distracting attention from the phonetics. As a result, though, the glosses will mainly be useful to a choir member who is rehearsing at home, since it would be hard to look up to the top of the page while actually singing during a rehearsal or performance.

2.2 Sentence-level meaning

In the glosses at the top of the page, I added some syntactic assistance in brackets. For example, Russian has no definite or indefinite article (the grammatical category of definiteness is conveyed through word order), so I added the indefinite article to help the user: ‘(it’s) coming, buzzing, (a) green noise.’ Under the glosses, I rewrote the sentence using English word order: ‘A green noise is coming, is buzzing.’ The need for rewriting may not be apparent in this particular case, but consider a later gloss: ‘as in milk bathed stand orchards cherry quietly (they) make noise.’ The structure of this sentence is not immediately clear; a reformulation is needed: ‘cherry orchards stand as if bathed in milk; quietly they make noise.’

2.3 In-context meaning

The sentence-level meaning ‘A green noise is coming, is buzzing’ is rather obscure: how can a noise buzz, and what is a green noise? The Russian text is the first line of an 1862 poem by the chief figure in late 19th century Russian poetry, Nikolai Nekrassoff. The poet added a footnote mentioning that zellyonee shoom (green noise) was the peasants’ way of describing the awakening of nature in spring, which is the subject of the opening portion of the poem.

To convey in-context meaning, at the top right of Figure 2, in brackets, I attempted a more poetic translation. For the verb goodyot, I avoided buzz since it wrongly suggests insects in summer, and instead chose hum—a more spring-like sound. For the noun shoom, I rejected suggestions in bilingual dictionaries such as roar (of battle) (too loud), rustle (of leaves) (which suggests autumn) and even murmur (of the forest), because I needed a word that could be the subject of the verb hum. The Russian word shoom also means ‘stir’ in the figurative sense of someone ‘creating a stir’ (i.e. being talked about), so I decided to use stir non-figuratively to suggest the movements associated with spring awakening. This yielded the sequence stirring…is humming, which makes more sense than noise…is buzzing: nature is stirring, which is creating a humming sound. I also changed the gloss green to verdant because this word suggests greenery growing rather than simply the state of being green (the Merriam-Webster online dictionary defines verdant as “green with growing plants”). Of course, it is unlikely that an uneducated peasant, such as Nekrassoff had in mind in his footnote, would speak of a verdant stirring, but I decided to sacrifice this aspect of the poet’s intent.

I now had a total of three semantic representations of the sentence, along with the phonetic representation (to be discussed in the next section):

Phonetics:                 ee-dyot          goo-dyot     zell-yon-ee    shoom
Word meaning:           (it’s)  coming, (it’s) buzzing, (a) green      noise
Sentence meaning:     a green noise is coming, is buzzing
In-context meaning:    a verdant stirring is drawing near, is humming

Normally, translators must choose among various strategies on a literal-to-free scale, but in this case there was no need to make a choice. I was able to give choir members the benefit of three different ways of conveying meaning. The gloss tells them the dictionary meaning of each word they are singing; the sentence representation conveys the meaning in familiar English word order; the in-context meaning gives (I hoped) a bit of the poetry.

3. Translation theory implied — Phonetics
3.1 The phrasebook system

Now let’s turn to the pronunciation problem. Here, a minor aspect of translation theory is indeed applicable but it is not really needed since the solution is fairly obvious, namely, the same solution that is used in many foreign language-learning textbooks and in phrasebooks for tourists. Still, even if theory is not needed to solve the pronunciation problem, the system used in textbooks and phrasebooks is based on an implied theory, which is worth spelling out. But first, an example of what I shall call the phrasebook approach:

Vous deviez être très belle quand vous étiez jeune.
You must have been very beautiful when you were young.
Voo DUH-vyay et-ruh TRAY bell con voo zay-tyay juhn (Tomb 1989: 23).

The author of Wicked French for the Traveler here draws on the English speaker’s knowledge of how English words (like tray)5 are pronounced and also on the speaker’s knowledge of common pronunciations of letter sequences (like voo).

This approach is not perfect. For example, the French ‘r’ sound is quite different from the ‘r’ of English tray. Similarly, the last sound of French belle is not like the ‘l’ of English bell but like the ‘l’ of English leak. Unfortunately, in English, this latter l-sound never occurs at the ends of words, so there is no way to represent it using the phrasebook system.

Another example: the first sound of French jeune sounds like the ‘s’ of measure, not the ‘j’ of joint. Since the former sound practically never occurs at the beginnings of words in English, there is again no way to represent it. The author of a phrase book can include a guide to pronunciation (“‘j’ at the beginning of a word is pronounced like the ‘s’ of measure”), but that requires users to learn pronunciation rules. One wonders how many tourists do in fact bother to read such pronunciation guides, commit them to memory and practice the rules before leaving on their travels.

Even if a user were to read and memorise the pronunciation guide, there remains the problem of sounds which do not exist in the user’s own language. Take the word jeune again. Its vowel is represented by the sequence uh in the third line of our sample sentence. This is the same representation used earlier in the sentence for the first syllable of deviez. However the actual French sounds involved are not the same: the first syllable of deviez has the mid central vowel that is represented in the International Phonetic Alphabet as [ə], the same vowel which occurs in English the. The word jeune, however, has the mid front rounded vowel [œ], which does not occur at all in English. Still, the two sounds are somewhat similar, and if singers pronounce them the same way, that will not really matter. French speakers in the audience might notice the incorrect pronunciation of jeune if a single person were speaking, but not if eighty voices are singing, with instrumental accompaniment.

The letter combination uh exemplifies a further problem with the phrasebook system: it is not always obvious how to pronounce a given group of letters. The sequence uh is not common in English orthography. It does occur in the written representation of a well-known speech particle and in the unstressed version of the second person pronoun (“Uh, what d’yuh mean?”), but the user of the tourist guide may not think of this when confronted with duh-vyay. There is no obvious way to represent this sound, even though it is very common in English. It occurs in the, but French deviez cannot be represented as de-vyay since the ‘e’ will probably be given some other sounding, perhaps that of the first ‘e’ of deviate. With this particular word, one possibility might be dove-yay, as long as readers take dove to be a type of bird, not the past tense of the verb dive, which has the wrong sound.

Now, a choir that is going to sing regularly in a particular language might want to spend time on phonetics exercises to learn sounds that do not occur in English as well as difficult consonant combinations (e.g. the Russian word vzglyat). In addition, all of the problems of representation cited above can be solved if a choir learns the International Phonetic Alphabet, which has a unique symbol for every sound found in human languages. A few choirs have done this, but it involves a very steep learning curve and a very considerable amount of time, which many amateur singers will not be willing to invest. Choirs that do not want to learn the International Phonetic Alphabet or do phonetics exercises will have to be content with an approximation to the right sound.

For many choirs, the goal will be to avoid the really gross errors that arise when choir members try to work directly from the orthography of the foreign language. The Romanian word spelled române (which means ‘Romanian’) is pronounced something like row-mi-nay: despite the spelling, the vowel in the second syllable sounds somewhat like the ‘i’ of English it, as I discovered when my choir sang some Romanian carols at a Christmas concert. And of course there are so-called silent letters: if even a single voice in the choir pronounces the ‘s’ of French très in the word sequence très belle, the audience will hear it. If the performance is being given in a multilingual city like Toronto, there will almost certainly be audience members who know the language of any given piece, and they may react negatively (“why couldn’t they learn my language properly?”). This problem, of eliminating really gross errors such as pronouncing silent letters, can certainly be avoided with the tourist phrasebook system; indeed, that is the system’s single greatest merit.

Curiously, the phrasebook approach is only rarely used in the published scores of works of choral music, even though the publishers must realise that sales will be to choirs singing in a language which many members do not know (the Rachmaninoff score discussed earlier was originally published in Moscow in 1903 but the version my choir used was obtained from a publisher in California). It is as if English-speaking choirs were just assumed to know how to pronounce French, Latin, German, Italian and so on. Or perhaps it is assumed that a single demonstration of the correct pronunciation by a knowledgeable choir member will suffice — a truly laughable assumption. A great many people do not pick up on the sounds of other languages easily; even after months of rehearsing, one can always hear a few voices making gross pronunciation errors.

The only exception to the lack of phonetic assistance in published scores is with Russian. Here one often finds transliterations, but these are usually based on a 1-1 system, whereby each Cyrillic letter is replaced by a given Latin letter or digraph (ч becomes ch). This approach, as it happens, can be unhelpful because Russian orthography, while not nearly as bad as English or French in terms of correspondence between spelling and sounding, is not quite as good as Spanish in this regard. Thus the second word of музыка Чайковского (‘music of-Tchaikovsky’) will typically be transliterated chajkovskogo (whereas if we follow the tourist phrasebook system, we will get something like tchy-COUGH-skuh-vuh): the letter transliterated as ‘g’ is actually pronounced ‘v’; the letter transliterated as ‘v’ is actually pronounced ‘f’; and the vowels of the last two syllables, being unstressed, are reduced from ‘o’ to ‘uh’.

3.2 Literary uses of the phrasebook system

This defect of traditional 1-1 transliteration systems brings me to the implied theory behind the superior system found in tourist phrasebooks. What do writings in the field of Translation Studies have to say about phonetics? Aside from incidental discussions in connection with film dubbing (lip synchronisation), the issue does of course arise in writings on poetry translation, with regard to rhyme, alliteration and meter. Such writings might prove useful when preparing translations which are themselves to be sung, but it is hard to see any relevance to the problem of helping choirs sing in languages they do not know.

More relevant are references in Translation Studies to literary translations where the translators have actually employed the phrasebook system—though that term is not used, and the translator’s purpose is of course not that of providing a pronunciation guide to the source text. For example, one sometimes comes across references to Luis van Rooten’s version of the Mother Goose Rhymes, translated at the phonetic level into French. If you know the pronunciation rules for French orthography and read the following aloud:

Un petit d’un petit
S’étonne aux Halles
Un petit d’un petit
Ah! Degrés te fallent (van Rooten 1967: 1).

it sounds very much like the first lines of Humpty Dumpty spoken with a French accent:

Humpty Dumpty
Sat on a wall;
Humpty Dumpty
Had a great fall.

Van Rooten’s French version is remarkable in that it consists entirely of real French words inserted into more-or-less grammatical syntactic structures—unlike typical phrasebook examples, which contain only a few actual words of the target language. A further interesting aspect of the translation is that van Rooten presents his book not as a translation but as a long lost original French manuscript entitled Mots D’Heures: Gousses, Rames (which if spoken aloud sounds like Mother Goose Rhymes). Each page is heavily annotated with English footnotes on the ‘French’ text. The first line, un petit d’un petit, the words of which mean ‘a little-one from a little-one’ is annotated: “the inevitable result of a child marriage.” The phrasebook system is thus used to create humour.

Another, somewhat different use of the phrasebook system can be seen in the translation of the Ancient Roman poet Gaius Valerius Catullus by Louis and Celia Zukofsky (Zukofsky 1969) and in David Melnick’s translation of Book One of Homer’s Iliad (Melnick 1983). Unlike van Rooten, the Zukofskys and Melnick select a few English words that are related either to the meaning of the particular source passage being translated or to a theme of the source text6. Melnick’s word choices are of the latter type: they reflect the homoerotic theme in Homer’s poem, as seen notably in the relationship between Achilles and Patroklos. The poem begins with an invocation of the muse: ‘Tell us, goddess, about the anger of Achilles, son of Peleus, the dreadful anger that brought countless ills to the Achaeans’ (Homer’s word for Greeks). The first two words in Ancient Greek can be rendered with the phrasebook system as men-een ah-ay-deh, meaning ‘anger’ and ‘tell’ (or ‘sing’ as the second word is commonly translated, since the Homeric bards are thought to have played a string instrument while reciting their poems). Melnick’s translation begins:

Men in Aïda they appeal, eh? A day, O Achilles
Allow men in, emery Achaians. All gay ethic, eh?

The wording of the translation is fairly funny in itself, and in 1983 (just a dozen years after the advent of the gay liberation movement in the United States) there would have been the added amusement of seeing this founding document of Western literature turned into a gay text. For those few who could follow along in Ancient Greek, further humour would arise from seeing how close Melnick comes to the way it is thought the source text was pronounced.

3.3 Catford’s transliteration system extended

Now, interesting as cases like the Homer and Mother Goose translations are, we are still left with the question: What about theory? Do theoretical writings about translation have anything to say about phonetics and spelling? To my knowledge, the only attempt to integrate phonetic considerations into a larger theoretical framework was Catford’s, almost half a century ago now. His A Linguistic Theory of Translation has chapters entitled “Phonological Translation,” “Graphological Translation” and “Transliteration,” but they are usually dismissed as oddities if mentioned at all. There is in fact nothing odd about them at all. Like anyone trained in linguistics, Catford attends equally to both sides of the linguistic sign—the semantic and the phonetic.

What is interesting about Catford’s treatment is that it fits into a systematic discussion. He begins his book by setting out a general linguistic theory, in the form of Michael Halliday’s ‘scale and category grammar’ of the early 1960s (1965: 3-4). Drawing on this approach, Catford then makes a distinction between total translation and restricted translation.
Total translation is “replacement of SL grammar and lexis by equivalent TL grammar and lexis with consequential replacement of SL phonology/ graphology by (non-equivalent) TL phonology/graphology” (1965: 22). That is, in normal translation, semantic but not phonetic/graphic equivalence is sought. Restricted translation is defined by Catford as “replacement of SL textual material by equivalent TL textual material, at only one level” (1965: 22). Thus the chapter on phonological translation begins by saying that “phonological translation is restricted translation in which the SL phonology of a text is replaced by equivalent TL phonology. The grammar and lexis of the SL text remain unchanged” (1965: 56). The significance of this somewhat obscure statement will soon become clear.

The chapter of most relevance in Catford’s book is the one on transliteration. Traditionally, transliteration is thought of as involving two languages with different scripts, such as French and Russian. Catford’s examples are indeed of this sort, but as we shall see, his description of transliteration can be applied to any pair of languages, whether or not they use different scripts. Catford describes transliteration not as a direct relationship between the units of two writing systems but rather as a three-step process: “SL graphological units are replaced by SL phonological units; these SL phonological units are translated into equivalent TL phonological units; finally the TL phonological units are placed by TL graphological units” (1965: 24).

Consider a Russian word written in the Cyrillic alphabet: ЗЕЛЕНЫХ. In the first step, we replace writing with speech, by speaking the word aloud, the result being representable in the International Phonetic Alphabet as [zjɛljɔnɨx]). (The raised j represents palatalisation of the preceding consonant, which to an English ear sounds like the ‘y’ of yak between the consonant and the following vowel.) In the second step, the spoken form [zjɛljɔnɨx] is phonologically translated into equivalent English sounds, this being representable in the International Phonetic Alphabet as [zɛljɔnɪk]. Since the two final Russian sounds [ɨ] and [x] do not exist in English, the closest (‘equivalent’) English sounds have been used. In addition, English words never begin with the sequence [zj], so [z] alone is used as the nearest equivalent.

In the third and final step, the English phonological translation is represented using the Latin alphabet. How this is done is the key question for present purposes. In traditional transliteration, the Latin letters are selected from some conventional list of 1-1 correspondences between Cyrillic and Latin letters. This might yield something like ZELENYKH (or ZELYONYH, depending on the transliteration system selected — there are several). Here the Latin digraph KH represents Cyrillic X, while Y represents the Cyrillic letter Ы, which is the vowel of the last syllable.

Transliterations like ZELENYKH are found in Russian books that have been purchased by libraries in the English-speaking world. They are needed for cataloguing purposes, but there is no need for anyone to speak the words aloud. Indeed, it will not be clear to non-Russian speakers how to pronounce such representations. How is the Y to be spoken? Or the KH?

The 1-1 rule, when followed strictly, places a severe limitation on the phonetic value of transliterations. Thus the fact that the two instances of Cyrillic E in ЗЕЛЕНЫХ have very different pronunciations is not represented at all in ZELENYKH. Now, some Cyrillic-Latin transliteration systems do not follow the 1-1 rule strictly; for example, our composer’s name is sometimes transliterated Rachmaninov rather than Rachmaninoff. The former is the result of the 1-1 rule; the latter represents the actual pronunciation of the final letter, by making the reader think of the English word off. The composer’s name ends with the Cyrillic letter B; when this letter appears at the end of a word, it is always pronounced ‘f’ rather than ‘v’ (its pronunciation before a vowel). However, the 1-1 principle dictates that the letter ‘f’ cannot be used to transliterate B, because it has been set aside to represent the Cyrillic letter ф, which is always pronounced ‘f’. The transliteration Rachmaninoff solves the problem by disregarding the 1-1 rules, but it leaves another problem unresolved: how is one to pronounce the ch? Unless one has already heard the composer’s name spoken aloud, one might be tempted to pronounce Rach- to rhyme with English catch, whereas in fact it sounds more like rack.

Clearly a more general departure from the 1-1 rule is called for if the words in a singer’s score are to be helpful. The problem for choir members when they are practicing at home, or reading the words at rehearsal or during performance, is that they have to engage in a fourth step, which Catford does not consider. Whereas the transliterator had to follow the three-step sequence SL writing — SL sound — TL sound — TL writing, the choir now has to speak the TL written representation aloud while singing. They have to get from TL writing back to TL sound, which is being used as an ‘equivalent’ of SL sound. And they have to do this without knowing the International Phonetic Alphabet and without a speaker of the source language whispering the pronunciation in their ear.

To help with this task, we can simply set aside completely the 1-1 convention of traditional transliteration and instead draw on the pronunciation conventions of English orthography for Catford’s third step — the move from TL sound to TL writing. Using this approach — the phrasebook system — choir members will be able to move, quickly and without assistance, from the written representation to a reasonable facsimile (i.e. no gross errors) of the source-language pronunciation. Our sample Russian word, traditionally transliterated as ZELENYKH, might be represented in the phrasebook approach by ZELL-YON-NICK. This representation is visually divided into syllables, each of which is to be sung while producing the musical pitch (or pitch sequence) written above it in the musical notation. As much as possible, actual English words (YON and NICK) or parts of words (-ELL) have been chosen to ensure the desired pronunciation. Which particular English words and part-words are chosen will depend on the dialect of English spoken by most choir members. For example, the French word botte (‘boot’) may or may not be phonetically representable as BOUGHT — a word which is pronounced with very different vowel sounds in different parts of the English-speaking world.

4. Conclusion

I have spent much more space discussing the phonetic problem facing choristers (how do I pronounce these words?) than the semantic problem (what do the words mean?). That is partly because — as indicated at the outset — phonetics is more important than semantics in conveying musical meaning. It is also because the solution to the semantic problem was fairly straightforward, given the familiarity within translation theory of the idea that meaning is expressed at different size levels (word-phrase-text). As for the solution to the phonetic problem, the phrasebook system provided a fairly obvious practical solution, but its theoretical underpinnings needed to be spelt out in detail. The solution is an extension of Catford’s ideas that ‘equivalence’ can be phonetic as well as semantic, and that transliteration is a three-step process. Once it is seen that Catford’s ideas can be applied to any pair of languages regardless of the script they use, we can then understand, in a theoretical way, how it is that choirs can avoid gross error by using the spelling-sound correspondences of their own language in order to sing in a language they do not know.

  • Catford, John Cunnison (1965). A Linguistic Theory of Translation. London: Oxford University Press.
  • Catullus, Gaius Valerius (2005). The Poems of Catullus (tr. Peter Green).Berkeley: University of California Press.
  • Melnick, David (1983). Men in Aida. Tuumba 47. Berkeley, California: Tuumba Press. (consulted 1 June 1 2012)
  • Rachmaninoff, Sergei Vasilievich (n.d. [1903]). Der Frühling, Op. 20. A Cantata for Baritone Solo, Chorus and Orchestra. With Russian and German Text. Choral Score. A Kalmus Classic Edition. Van Nuys, California: Alfred Music Publishing.
  • Tomb, Howard (1989). Wicked French for the Traveler. New York: Workman Publishing.
  • van Rooten, Luis (1967). Mots D'Heures: Gousses, Rames.New York: Grossman.
  • Zukofsky, Celia and Louis (1969). Catullus. London: Cape Goliard Press.

Mossop 3Brian Mossop has been a staff translator for Canada’s federal government for the past 39 years, and he has been a part-time instructor at the York University School of Translation in Toronto for the past 32 years. He holds a bachelor’s degree in French and Russian, and a master’s degree in linguistics. He also sings with the Toronto Choral Society. Email:


Note 1:
Or more precisely, taking into account the composer’s notion of how the words are pronounced: the composer may have had an imperfect knowledge of the language, or the pronunciation of a word during the composer’s lifetime may have differed from the modern pronunciation. In the case of works in Latin, standard pronunciations have always varied considerably from country to country. The choral director will often have to decide which of various pronunciations of a word will be used (to take an English example: will the first syllable of neither be pronounced NIGH or KNEE?).
Return to this point in the text

Note 2:
At one time or another, these choirs have sung works in Cree, Chinese, French, German, Italian, Japanese, Polish, Romanian, Russian, Spanish, Swedish and Zulu.
Return to this point in the text

Note 3:
In fairness, it should be understood that amateur choirs typically rehearse for only two hours a week, and most of this time needs to be devoted to other aspects of music: the often difficult pitch sequences and rhythms, variations in loudness and so on. These days, choir members can hear pronunciation at home using Google Translate, which includes an icon that can be clicked to hear the foreign expression once it is typed into the source-language box. This may be helpful with sounds that also occur in the singer’s own language, but with other sounds, merely hearing them is not enough. Singers need to know where to position their tongue and lips. Even watching someone pronounce a word containing the high front rounded vowel of French lune or German müde is of little value; watching can help with lip position, which is visible, but not with tongue position, which is not. Most people, including every choir director I have encountered, are completely unaware of the articulatory mechanics of producing linguistic sounds.
Return to this point in the text

Note 4:
In March 2010, a choir of which I am a member sang a concert entirely in Russian, and I served as one of two phonetics coaches. For this purpose I prepared the modified version of the vocal score seen in Figure 2 (I have slightly changed the wordings for the purposes of this article).
Return to this point in the text

Note 5:
The capitalisation of DUH and TRAY wrongly suggests that these syllables are stressed. With music, stressed syllables do not need to be capitalised to help with pronunciation because the music itself most often creates the right stress automatically.
Return to this point in the text

Note 6:
Unfortunately, no example of the Zukofskys’ homophonic translation can be given because their son Paul, the copyright holder, is in the habit of threatening legal action, even for short quotations which one might think constitute fair use.
Return to this point in the text