An inter-semiotic approach to translation : Leonard Cohen in AfriKaans

A song constitutes a piece of music and lyrics designed for a singing performance (Franzon 2008:376). Popular songs usually have a regular structure consisting of verses and a chorus or refrain (Chan 2009:109). The chorus often contrasts with the verses in terms of the rhythm and the melody, as well as the lyrics. It is also quite common for the verses to be translated and the chorus to be kept in the source language as the chorus is the song’s most important feature (Gritsenko & Aleshinskaya 2016:166). The main concern in song translation is to create a singable target text to ‘permit the actual performance in the target language of foreign songs, with their pre-existing music’ (Low 2008:2). Song translation, therefore, constitutes a ‘linguistical and cultural confrontation between two linguistic codes and between the expressive singability within poetry and music’ (Gorlée 2002:166). The song translator has to pay attention to the rhythm of the melody, the text in the form of a lyric (often also with a rhyme scheme), hidden and private meaning, the intended message and the medium of presentation. This is the process of relating two different semiotic systems, each having its own requirements that constrain and enable the functioning of the specific system. This article specifically focuses on the translation of Leonard Cohen’s songs into Afrikaans by a South African gospel singer and preacher, Koos van der Merwe. The data were collected from an original Leonard Cohen CD and the translated versions thereof from the Van der Merwe CD (Leonard Cohen in Afri-Kaans1).


Introduction
A song constitutes a piece of music and lyrics designed for a singing performance (Franzon 2008:376).Popular songs usually have a regular structure consisting of verses and a chorus or refrain (Chan 2009:109).The chorus often contrasts with the verses in terms of the rhythm and the melody, as well as the lyrics.It is also quite common for the verses to be translated and the chorus to be kept in the source language as the chorus is the song's most important feature (Gritsenko & Aleshinskaya 2016:166).The main concern in song translation is to create a singable target text to 'permit the actual performance in the target language of foreign songs, with their pre-existing music' (Low 2008:2).Song translation, therefore, constitutes a 'linguistical and cultural confrontation between two linguistic codes and between the expressive singability within poetry and music' (Gorlée 2002:166).The song translator has to pay attention to the rhythm of the melody, the text in the form of a lyric (often also with a rhyme scheme), hidden and private meaning, the intended message and the medium of presentation.This is the process of relating two different semiotic systems, each having its own requirements that constrain and enable the functioning of the specific system.This article specifically focuses on the translation of Leonard Cohen's songs into Afrikaans by a South African gospel singer and preacher, Koos van der Merwe.The data were collected from an original Leonard Cohen CD and the translated versions thereof from the Van der Merwe CD (Leonard Cohen in Afri-Kaans 1 ).This article is thus not so much concerned with translation as a strategy to bridge the language barrier, but rather with the nature of the constraints of multi-semiotic phenomena.A number of ways in which Van der Merwe negotiated the complex multi-semiotic interface between text and music will be investigated through a discussion of inter-semiotic translation in a song selected from the Van der Merwe CD, namely 'Halleluja'.

Multimedial translation
Language, according to Buhler's (1990:35) organon model, has three communicative functions: representation, expression and appeal.Drawing on this model, three basic communicative text types can be derived: informative, expressive and operative texts (Reiss & Vermeer 2013:182).Informative texts refer to texts having the intention of passing on news and information, expressive texts 'convey artistically organised content' or act as works of art (Reiss & Vermeer 2013:182), while operative texts convey information organised in a way that the receiver may be convinced to act in accordance with the sender's intentions (Reiss & Vermeer 2013:182).In addition, Reiss 1.The spelling 'Afri-Kaans' is as found in the title of the Koos van der Merwe CD.
Whether or not song lyrics should be translated has been debated by researchers, translators, artists and audiences.Some are of the opinion that songs should not be translated as singing in translation produces a weak version of the source text, while others argue that a song in the language of the audience fosters better understanding.The translation of song lyrics goes beyond linguistic aspects and includes musicological aspects such as the melody, rhythm and mode of presentation.Because of the interaction between the music (the melody) and the lyrics, the music in some cases obscures the lyrics and in other cases prolongs the lyrics.Therefore, the song translator faces a constant negotiation of inter-semiotic elements with regard to, among others, functionality and singability.This study provides an overview of the musicological aspects of song translation, with reference to Low's pentathlon and Franzon's layers of singability.As an illustration, this article provides a discussion of the translation of a Leonard Cohen song into Afrikaans by a South African gospel singer and preacher, Koos van der Merwe.The data have been collected from an original Leonard Cohen CD and the translated versions thereof from the Van der Merwe CD (Leonard Cohen in Afri-Kaans).
adds another function to these three written text types, namely, the multimedial, which is of specific interest to this study (Reiss & Vermeer 2013:137).The multimedial text type is more than just the mere fourth category of text types; it is a hypertext including songs, stage plays, film scripts, opera libretti, comics and advertising material (Surugiu 2010:118).In contrast to novels, short stories and poems, the verbal elements in a multimedial text are part of a larger body or parts that form an artistic whole.
The semiotic material in multimedial texts can include aspects such as pictures, music and facial expressions (Surugiu 2010:118).All of these aspects play an important role in contributing to meaning, which confirms that meaning in multimedial texts, especially where music is involved, cannot be determined by a textual analysis only.
To consider the meaning in multimedial musical texts, a textual analysis must be combined with an analysis of other musical aspects, such as the melody, tempo, pitch, duration, loudness, timbre, rhythm, harmony, pause, stress and articulation.It is thus clear that one needs an inter-semiotic approach to study song translation.In language, there are a number of ways in which form expresses meaning, as languages use different forms which have secondary and figurative meanings, adding even more complication (Shamami 2012:247).Furthermore, the translator might be required to use a different form to convey the same meaning in another language.If one should translate the form of the source language according to the corresponding form in the target language, it is quite possible that the meaning would be changed or result in an unnatural form in the target language (Shamami 2012:247).

The inter-semiotic translation of a multimedial text
The translation of song lyrics differs vastly from interlingual translation.The term 'inter-lingual translation' is taken from Jakobson's (1959Jakobson's ( /2012:233) :233) distinction between three types of translation, namely intra-lingual, interlingual and intersemiotic translation.The first two types can be studied within one semiotic mode, namely, language.Inter-lingual translation is 'translation proper' between different languages, while intra-lingual translation refers to translation within the same language.Translation proper, specifically, is generally associated as a faithful, word-forword translation of the source text, despite the possibility that the result may be seen as inappropriate with regard to the intended purpose (Nord 1997:4).Skopos theory argues that the intended purpose of the translation determines the translation process, in other words, the methods and strategies used to produce the target text (Nord 1997:27).Skopos is the Greek word for purpose (Nord 1997:27).Reiss and Vermeer (2013:89) state that: '… the skopos of an action takes precedence over the mode of action, that is, the purpose determines whether, how and what is done'.Low (2008:1) agrees that the translator's choice of translation strategy is determined by the skopos or the end-purpose of the song translation.Metin Tekin and Isisag (2017:135) argue that the skopos theory is the most suitable theory to be applied in song translation as the aims and needs of the target culture are more important than the source text.
Aligned with the skopos of the translation, Jakobson's third type of translation, inter-semiotic translation, is the most appropriate form to adhere to the purposes of song translation for a specific audience.Jakobson's inter-semiotic translation requires a move away from spoken language and involves a different complex system.Ramos (1980:376) explains that inter-semiotic translation can be between language and music, or language and paintings or even without any participating verbal sign system, such as the musical versions of many ballets.In vocal music, both verbal and musical systems are represented, making it an example of Jakobson's inter-semiotic translation because of the similarity in the interaction between words and lyrics and the interaction between two different languages.
According to Gorlée there is also a creative side to intersemiotic translation or recoding that depends on 'improvised desire and free will (on the part of the receptor) translations', taking the meaning of the written signs in a verbal language and transforming them into a: '… language in a mixed, metaphorical manner of speaking ' (2008:346).Inter-semiotic translation is a triadic (or more complex) activity as it involves the 'union of inter-medial translations into an embedded one' (Gorlée 2005:38).Inter-semiotic translation is also the: '… decentering of verbal language to transpose it into nonverbal languages' (Gorlée 1997:240-244).However, it can also happen the other way round, that is, non-verbal meaning can be translated into verbal meaning.As Low (2008:2) puts it 'translation is a complex activity, and the devising of singable texts is more difficult than most translating tasks'.
A song consists of various elements each of which has an important role in rendering the song a complete product.These elements include linguistic aspects, such as the song lyrics or text, and musicological aspects, such as the melody and rhythm.To produce a functional, singable inter-semiotic translation, the translator has to pay attention to all the elements present in the phenomenon to be translated.The translator, therefore, has to keep in mind the complication of a multiplicity of semiotic elements that interplay with the words.A word that would fit perfectly in a poem would not necessarily fit well in a lyric (Gorlée 1997:236) and thus a substitute has to be found.

Song translation
The song-lyric translator is challenged by a poetical text that has to be rendered in the target language and culture and, at the same time, is limited by the various constraints presented by the music to which the lyric is set (Surugiu 2010:120).The musicological semiotic system thus constrains the choices available to the translator.It is generally assumed that the melody is fixed and the translator is not at liberty to change the music to fit the words.Thus, the song-lyric translator must choose the target lyrics carefully and adjust where necessary to fit the rhythm of the music.However, there are cases where slight melodic or rhythmic changes are inevitable for various reasons.Taking all of this into consideration, it becomes clear how complex song translation is, as the translator has to translate according to the limitations set by the music.
Low's pentathlon Low (2005:191) has a practical approach to 'naturalness, rhythm, rhyme and fidelity' and developed the pentathlon approach involving the following five elements: singability, sense, naturalness, rhythm and rhyme.
In terms of the skopos of any song translation, singability is probably the most important aspect of the pentathlon.Singability implies that the translated song must effectively be performed so as to satisfy the audience.Low (2008:13) suggests that translators should pay special attention to vowels.It is not necessary that each vowel in the target text would match those of the source text, but rather fulfil the needs of the melody.Therefore, song translation must produce singable lyrics.
The second element in Low's pentathlon is sense, referring to keeping as much as possible of the original meaning of the source text in the target text.Low states that in normal translation (inter-and intra-lingual translation), semantic accuracy is most important, but because of the constraints of vocal-music translation, 'stretching or manipulation' (flexibility) of sense may become necessary (Low 2005:194).The meaning remains important, but flexibility also plays a role in the sense that syllable counts may be altered or that a near-synonym may replace an exact word.Low (2008:13) argues that the meanings of the words are not equally important in all songs.In some instances, song translators need to make semantic compromises -something that would be unacceptable in, for example, a scientific translation (Low 2008:12).The translator, therefore, has to look at the song as a whole and aim to reproduce the spirit and mood of the original song (Drinker 1950:235).
Low's third element is naturalness, which is usually associated with the translator's responsibility towards the audience (Low 2005:195).One aspect that receives attention with regard to the degree of naturalness evident in the target text is whether the target text should reveal the fact that it is a translation.The audience has to receive the message on their first 'encounter' with the song and, therefore, naturalness plays an important role as the audience must understand the song while it is being performed (Low 2005:196).Unnaturalness would demand extra processing effort from the audience (Smola 2011:109).
The fourth element in Low's pentathlon is rhythm, notated in the music score, and is applicable to both the source and target texts.The rhythm influences the song translator's choice of words, as he needs to respect the rhythm that pre-exists the target text.Song translators also pay attention to syllable count, considering the source text and the target text to have the exact same syllable count (Low 2005:196).This, of course, is desirable but not always possible.The stresses of the text and the length of the notes must be taken into consideration as well.The melody, therefore, influences the length of the vowels, as well as the consonants available for use by the translator (Low 2005:198).However, a rest in music (a moment of silence in the melody) should not be ignored either and the translator needs to be careful where he or she places the gap.Although the pentathlon principle also allows for flexibility with regard to syllable count, it should be seen as a last resort, and preferably in a recitative phrase rather than in a lyrical phrase (Low 2005:197).He also argues that it is better to add a syllable on a melisma and to remove a syllable on a repeated note as these methods alter the rhythm without influencing the melody too much.It is also possible that insignificant changes may be made to a melody in order to have a successful target text.According to Low, the translator sometimes has to choose between adding or repeating a new word or phrase, or removing a note or two from the melody (Low 2005:197).However, if the translator decides to add words, it must sound as if the words have been in the source text, adding to the naturalness of the target text.
The final aspect of the pentathlon principle is rhyme, an element that Low believes has rendered many target texts useless, mainly because the translators of these texts placed too high a premium on rhyme, losing sight of more important elements and in the end overshadowing the melody (Low 2005:198).
Franzon's layers of singability Franzon (2008:373), assuming that a song has three properties (music, lyrics and performance) and music has three components (melody, harmony and musical sense), suggests that the song translator has five options in the translation of a song: (1) not translating the song, (2) translating the lyrics but without considering the music, (3) providing new lyrics, (4) adapting the music to the translation and (5) adapting the translation to the music.Singability (as also argued by Low in his pentathlon) refers to the phonetic suitability of the translated lyrics and how easy the words are to be sung to specific note values.In addition, it refers to the assessment of original and translated lyrics (Franzon 2008:374).Singability may also be described in the same way as the skopos theory describes a good translation -'suitable in every relevant way for the particular purpose' (Franzon 2008:375).Thus, singability occurs when the music and the lyrics have reached a 'musico-verbal unity' (Franzon 2008:375), enabling the lyrics, together with the music, to deliver the intended message.Franzon (2008:389) provides three layers of singability that should also be included in the target lyric: obtaining a prosodic, poetic and semantic-reflexive match with the music.For the lyricist, music consists of a melody, a harmonic structure and a specific meaning that has to be conveyed (Franzon 2008:390).The function of the lyrics will be influenced, if not determined, by the way the music functions simultaneously.
To reach a prosodic match to the melody, a singable lyric uses prosodic elements such as rhythm, stress and intonation.Gorlée (2002:168) states that the prosodic patterns are fixed by the linguistic structures in collaboration with the musical rhythms.Franzon refers to this as: '… universal speech phenomena that appear in singing in a stylized and controlled form ' (2008:390).Different musical genres have different demands with regard to phonetic suitability.In opera, for example, both vowels and consonants must be 'easy to vocalize' (Franzon 2008:390).
The poetic match, or the partnership that exists between text and music, is closely integrated with the harmonic structure of a piece of music, in other words, when the lyrics of a piece of music verbally 'mirrors' musical elements such as stylistic figures, climax, contrast and rhyme (Franzon 2008:390;Surugiu 2010:120).In vocal music, certain words fall on specific notes.Meaning is thus shaped through stress and intonation, as well as tone and modulation of the voice.Meaning is also conveyed through melodic phrasing such as keywords, climaxes, pauses and rhyme (Gorlée 2002:168).However, these devices occur in language as well and are not in themselves musical.It is important that these elements, in intra-lingual translation, are recognised and reproduced with new linguistic material relating to the music and producing the required function.
A semantic-reflexive match is concerned with expressions, such as mood, description, metaphor and story (Stopar 2016:141).There are also universally accepted structures, such as happiness depicted by joyful music, or moments when a specific musical movement are depicted through words (Franzon 2008:391).Gorlée (2002:168) adds that rhyme plays an important role in poetry and song translation and often poses many difficulties for the translator.However, the translator is urged to rather use imperfect rhyme than using unusual word choices or inverted word order.Franzon (2008:391) presents these three elements of singability, prosodic, poetic and semantic-reflexive match, as layers, building upon each other to present a complete work.In song translation, the music (melody) is the most important aspect as it determines the prosody, as well as the stylistic choices and, at the same time, adds semantic value to the lyric content.The most basic requirement is prosody because without it, singability might be completely lost, or it might technically be impossible to sing.The character of the song determines the need for a poetic or semantic-reflecxive match.The translator may also decide to play around with individual characteristics such as creating perfect rhythm, reaching different levels of success.The translator can decide which layer to pursue, and in most cases, the first two layers will be chosen because the presence of the third layer, the semantic-reflexive match, seems to indicate that the other two are present as well.
In the following section, we present a case study, discussed and analysed in terms of the complex characteristics and challenges of song translation.

Case study: 'Halleluja'
For the purposes of this discussion, the focus falls on the translation, into Afrikaans, of one of Leonard Cohen's songs, 'Hallelujah'.Cohen originally wrote 80 verses for 'Hallelujah', but the final version only contains 15 verses.It happens very often that various versions of a particular song are recorded, and for various purposes.This particular song has been used in television shows such as The West Wing, ER, Scrubs and Holby City.Many artists have recorded this song and up to December 2008, about 1000 versions of the song had been recorded (Cheal 2016).The translation into Afrikaans is the work of a South African gospel singer and preacher, Koos van der Merwe.The source text was taken from an original Leonard Cohen CD and the translated version thereof from the Van der Merwe CD (Leonard Cohen in Afri-Kaans).
The version used for discussion in this article contains seven verses.Firstly, we discuss the musical aspects of the song and then present a discussion of the lyrics (the chorus and each verse).

Discussion: Musical aspects
Leonard Cohen has written the melody in the key signature of D major, but the melody moves through different keys during the course of the song.The introduction moves between D major and A major with the lyrics beginning in D, in bar 1 (Figure 1).In bar 2 it changes to B minor, enhancing the feeling of secrecy to fit the meaning of [a secret chord].
The key signature returns to D major in bar 3, with the lyrics in this bar beginning with the word [David].The following bar is in B minor again, but this time the minor key enhances a reserved feeling of inner peace and pleasure, as indicated by the lyrics of this bar.It is as if the king is composing a hallelujah but he is not quite sure whether it is a holy or a broken hallelujah.The rest of line 3 (bars 5-7) stays in major keys and moves from G-A-D major.It is also worth mentioning that Cohen specifically chose B minor as it is (musically) related to D major.In music notation moving from D major to B minor is moving backwards (falling), while moving from D major to A major is rising.The lyrics of line 5 of the source text are imitated in the melody.The melody moves from B minor ('the minor falls') to G major ('the major lifts') in the following two bars, corresponding with the lyrics.Van der Merwe stays true to Cohen's imitation between lyrics and melody with the target text 'die vierde eers, die vyfde volg' to imitate the movement in the melody.Traditionally, a minor key indicates that emotions are falling (sadness) and a major key indicates joy, and, therefore, uplifts the emotion of the listener.In this case, the target text follows the source text pattern between keys and words, with the translated words appearing in the same bars as the source text.
In bar 14, the key signature (Figure 2, bar 14: F# major) beautifully expresses and strengthens the feeling that the king is baffled.Up to now the key signatures (D, G, A, b) were all musically related to one another.However, the use of F# major creates a feeling of complexity in the listener as it appears unexpectedly and is unrelated to D major.Bar 15 once again moves to B minor, combining the feeling of uncertainty (baffled) and pleasure (hallelujah) as indicated in the lyrics (lines 6 and 7).
Later in the song, Cohen often refers to 'a cold and a broken hallelujah', as well as 'a perfect hallelujah'.These images are strengthened in the melody by the alternating use of minor and major keys, with the minor indicating coldness and brokenness, and the major indicating holiness and perfectness.

Chorus
Cohen used only the word 'Hallelujah' for the chorus.It appears 5 times after every verse, except the final verse, where it appears 17 times.'Hallelujah' is a Hebrew word, which means 'Glory to the Lord'.Van der Merwe stayed true to the source text with regard to the melody, rhythm and the lyrics, as it is the chorus that attracts the listener's attention and, as indicated earlier, the chorus is one of the most important characteristics of a song (Gritsenko & Aleshinskaya 2016:166).The melody and rhythm of the chorus are repeated unchanged during the song, creating a feeling of stability, peace and serenity.

Verse 1 (lines 1-6)
In line 1 (Table 1), the source text, 'Now I've heard there was' is translated with 'Dit was glo' [it was believed] in the target text.In Afrikaans 'glo' [believe] means the same as 'I've heard' in the source text, indicating something like a 'gerug' [a rumour].By doing so, Van der Merwe succeeds in keeping the same number of syllables between the source text and target text.Keeping the number of syllables unchanged also adds to the rhyme scheme of the source text, with no loss of meaning in the target text.At the same time, the translation of inter-semiotic aspects enhances singability as the words fit the rhythm of the melody.
In line 1 (cf.Table 1 and Figure 1), Van der Merwe changes the melody and rhythm to fit the target lyrics to emphasise the word 'was'.This adds to the naturalness of the target text (cf.Low's pentathlon) as the Afrikaans natural accent has to fall on the first beat of the bar.Van der Merwe then uses a shorter note for 'glo', placing emphasis on the word 'was' to enhance the meaning thereof and correspond with the importance of the source text.
In line 3, Van der Merwe uses ellipsis by omitting 'do you' from his translation of the source text.The words he chooses for the Afrikaans translation have the same number of syllables as the source text and this allows for the melody and lyric to fit and enhance singability (cf.Franzon's layers of singability).Although words have been omitted in the target text, there is no loss of meaning (cf.Low's pentathlon element of sense).
Nearly the same happens in line 4 (cf.Figures 1 and 2, bars 8 and 9), where Van der Merwe omits 'It goes like this' from the target text without losing any meaning.In this instance, he adds the words 'eers ' [first] and 'op' [up] to the Afrikaans translation with the effect of naturalness, an aspect of Low's pentathlon.
An inter-semiotic element is evident in lines 4 and 5, namely, the movement of the melody.In line 4 the source text lyrics ('the fourth, the fifth') are mirrored in the melody with the keys moving from G to A in bar 10 (Figure 2).The lyrics, however, would still make sense to the target listener without any musical training as fifth always follows fourth.
In music notation G is the fourth note from D and A the fifth, and the melody moves D-G-A major, corresponding to the source text lyrics.However, the target text lyrics do not follow the same pattern as the source text because the important words 'vierde eers die vyfde volg' [fourth first the fifth follows] in the lyrics are spread over two bars (Figure 2, bars 9-10) and not in the same bar as is the case in the source text.Van der Merwe changed the rhythm of bar 9 of the target text melody to the same rhythm as bar 10 (the same rhythm in both melodies).As the target text lyrics are spread over two bars, the rhythms of the two bars are the same to imitate the rhythm of the lyrics in line 4 of the source text.
In bars 12 and 13 (Figure 2), Van der Merwe changes the rhythm of the target text melody to enhance the feeling of 'the baffled king'.He alters the rhythm for the three syllables of 'oor-bluf-te ' [baffled] to three short notes.The first note is the last beat of bar 12, with the other two short notes the first two beats of bar 13.The source text (Figure 2, bar 13) uses a short note followed by a long note, which fits the two-syllable word 'baffled'.This corresponds with Low's flexibility syllable count, which forms part of the pentathlon.
Another inter-semiotic element is evident when Van der Merwe changes the rhythm of the original melody by introducing a moment of silence on the next beat after he sings 'oor-bluf-te' [baffled] (Figure 2, bars 13 and 14).The short notes, immediately followed by a short silence, momentarily create a feeling of uneasiness in the listener, strengthening the corresponding lyrics.
Verse 2 (lines 9-14) In line 12 (Table 2), the source text has 'a kitchen chair', but Van der Merwe replaces the indefinite article with a possessive pronoun by using 'haar kombuisstoel' [her kitchen chair].The possessive pronoun has a more personal effect on the target text.
Van der Merwe omits 'and she' (line 13) from the target text, as well as 'she drew' (line 14) without losing any of the intended messages.By doing this, Van der Merwe is able to fit the translation to the music, improving singability.This is a good example of how he negotiates the inter-semiotic nature of the source text.In line 13, the number of syllables is the same in both the source text and the target text and improves singability.In line 14, however, he replaces the omitted words with 'laat' [let].
In terms of melody, Van der Merwe changes the rhythm in bar 35 (Figure 3), as he uses more words in the target text than the source text and needs to find a way to make them fit.He arranges it in such a manner that both the source text 'you' and the target text 'jy' is sung on long notes, although these are not the same in length.Van der Merwe also changes the rhythm of the melody to fit the syllables of 'kombuisstoel' [kitchen chair].It would sound very unnatural if he kept the rhythm the same as in the source melody, as the long note would then fall on 'stoel' [chair].He reverses

Die vierde eers, die vyfde volg
The fourth first, the fifth follows 5 The minor fall, the major lift

Die mineur af, die majeur op
The minor down, the major up 6 The baffled king composing Hallelujah

Oorblufte koning speel sy Halleluja
Baffled king plays his Hallelujah the rhythm of the source text with the result that 'kit-chen' and 'kom-buis' both sing the first syllable on a short note and the second syllable on a long note (bars 35 and 36).By spreading 'kom-buis' [kitchen] over two bars he also ensures that the stronger syllable falls on the stronger beat, adding to the naturalness of the translation.The rhythm of bar 37 is adjusted to fit the target text words 'sny jou hare af' [cut your hair off].'Sny jou' [cut your] are sung on two shorter notes (the shortest notes in that specific bar), imitating the movements of the scissors when hair is being cut.In the source melody (Figure 3, bar 38), Cohen also uses a short note for the word 'cut', although not as short as Van der Merwe's.By doing this, Van der Merwe negotiates the inter-semiotic nature of the text to increase singability.

Verse 3 (lines 17-22)
In line 21 (Table 3), Van der Merwe adds 'Die' [the] at the beginning of the sentence for the sake of rhythm, accent and singability.If he did not add a word, the target text would sound awkward and not fit the melody and rhythm.This is another example of negotiating the inter-semiotic nature of song translation -deciding whether to change the rhythm and the melody, or the lyrics, or maybe both in some cases.
In line 22, however, Van der Merwe omits the second 'it's a' from the target text to keep the same number of syllables as the source text, still bringing the message across while fitting the rhythm and melody (cf.Franzon's layers of singability).
Verse 4 (lines 25-30) In line 27 (Table 4), Van der Merwe omits 'do you' as he did in line 2 (Table 1).In line 3 (Table 1), Van der Merwe uses ellipsis by omitting 'do you' from his translation of the source text.The target text still has the same number of syllables as the source text, allowing a perfect fit between the melody and the lyric and enhancing singability (cf. Franzon's layers of singability).Although words have been omitted in the translation between the two texts, no loss of meaning is evident (cf.Low's pentathlon element of sense).
In line 30, Van der Merwe omits 'we drew' and translates the single syllable word 'breath' with the three-syllable word 'asemteug'.By doing so, he keeps the total number of syllables equal in this line (30) in both the source text and target text, and stays true to the meaning of the source text.This is a good example of the importance of negotiation between inter-semiotic elements with regard to functionality and producing a singable product.Verse 5 (lines 33-38) In line 37 (Table 5), Van der Merwe adds words to the target text to keep the same number of syllables as the source text and enhance singability.The words 'it's not somebody' is translated as 'dis nie een', with a loss of two syllables.By adding 'wat sê' Van der Merwe produces a singable target text without changing the meaning of the source text.
The same technique used in line 22 (Table 3) occurs in line 38 (Table 5).Van der Merwe omits the second 'it's a' from the target text to keep the same number of syllables as the source text, still bringing the message across, while fitting the rhythm and melody (cf.Franzon's layers of singability).

Verse 6 (lines 41-46)
In line 46 (Table 6), Van der Merwe omits the second 'die' (the) from his Afrikaans translation.The reason for the omission relates to a negotiation to increase singability by choosing either a word-for-word translation that would have increased the number of syllables and forced the translator to change the rhythm and melody to fit the lyrics or by keeping the source text syllable count and changing the lyrics to fit the original rhythm and melody.Once again, a careful negotiation between inter-semiotic aspects of song translation takes place to increase functionality, corresponding with Franzon's layers of singability that aims at a functional translation.
Verse 7 (lines 49-57) Van der Merwe changes the word order of the target text in line 53 (Table 7) for a number of reasons.He does not only follow the rhyme of the source text (lines 52 and 53) but also creates a translated text in Afrikaans that is natural and does not lose any meaning (cf.Low's pentathlon).
In line 54 Van der Merwe uses 'hierdie tong' [this tongue], rather than 'my tongue' as Cohen did.The reason for the choice is to fit the rhythm of the melody by keeping the same number of syllables as the corresponding line in the source text.In so doing, Van der Merwe adds to the singability and eventually the functionality of song translation by negotiating the inter-semiotic nature of the text.

Conclusion
As evident in the discussion of the translated version of Cohen's 'Hallelujah', song translation aims to integrate the verbal content of a source text (the song) into a different language, incorporating elements of rhythm and melody in the translated version of the song.The song translator faces different challenges than those faced in inter-and intra-lingual translation, as he has to produce a singable text.Low's pentathlon and Franzon's layers of singability address aspects that the song translator has to take into consideration, and suggest ways to overcome these challenges.Various aspects with regard to song translation have been mentioned and a few discussed in lesser or more detail.Van der Merwe used quite a number of translation strategies, depending on the various challenges faced in the translation of the song lyrics.Often these techniques were applied because of syllabic prominence that must be adhered to in order to fit the rhythm, in which case the rhythm was left unchanged in most of the cases, most of the time.In other songs, however, the rhythm was changed to place the syllables in the correct position so as to enhance pronunciation and increase the singability and naturalness of the target text.
There were instances when rhyme determined the meaning, and at some other instances rhythm determined the meaning.

Die heilig of gebreekte Halleluja
The holy or broken Hallelujah In some cases, Van der Merwe changed the rhythm of the source melody to fit the target text and in other cases he changed the target text to fit the rhythm of the source melody.However, the pre-existing melody placed constraints on the translator.
Van der Merwe also made use of specific instruments to enhance the message of the source text, integrating all aspects of the target text to enhance functionality and creating inter-semiotic unity.The key signatures chosen for the melody often also played an important role in creating emotion and feeling that strengthens the message of the source text.Van der Merwe succeeds in creating poetic, prosodic and semantic-reflexive matches in his translations.
The various techniques that he applies also allow him to create a target text that makes sense, sounds natural in the target language, enhances singability and adheres to the skopos.This research study adds to the literature on song translation by showing a number of challenges that a song translator faces and how these can be addressed.
s a cold and it's a broken HallelujahDis 'n koue en gebreekte HallelujaIt's a cold and broken Hallelujah