Sanna Maarit Ruokolainen Semantic song translation strategies in the Finnish subtitled and dubbed versions of Disney s Frozen

Transcription

1 UNIVERSITY OF EASTERN FINLAND PHILOSOPHICAL FACULTY SCHOOL OF HUMANITIES English Language and Translation Sanna Maarit Ruokolainen Semantic song translation strategies in the Finnish subtitled and dubbed versions of Disney s Frozen MA Thesis March 2018

2 ITÄ-SUOMEN YLIOPISTO UNIVERSITY OF EASTERN FINLAND Tiedekunta Faculty Philosophical faculty Osasto School Department of Humanities Tekijät Author Sanna Maarit Ruokolainen Työn nimi Title Semantic Song Translation Strategies in the Finnish Dubbed and Subtitled Versions of Disney s Frozen Pääaine Main subject Työn laji Level Päivämäärä Date Sivumäärä Number of pages English Language and Translation Pro gradu -tutkielma x Sivuainetutkielma Kandidaatin tutkielma Aineopintojen tutkielma March 28, pages + Appendix Tiivistelmä Abstract This research looks into semantic translation strategies used in the dubbed and subtitled songs of an animated musical film. The aim of this study is to compare how the two translated versions differ from each other in terms of content and what kind of semantic translation strategies has been used in them. As songs in a musical have a narrative function, a further aim of this study is to examine if the used translation strategies affect the conveying of the plot of the movie. The research material consists of six songs and their Finnish dubbed and subtitled versions in the movie Frozen. The research method relies on Ritva Leppihalme s (2007) and Andrew Chesterman s (2016) classifications of translation strategies. The analysis consists of two phases: first the lines of the song lyrics will be categorized as literal translations, substituted lines and lines which contain semantic changes after which the lines containing semantic changes will be analyzed further and the semantic changes categorized following Chesterman s semantic translation strategies. The semantic translation strategies which are in focus in this study are paraphrase, synonymy/antonymy, hyponymy, abstraction change and trope change. According to the results, 59.5% of the lines in the dubbed version contain semantic changes whereas in the subtitled version, semantic changes occur in 44% of the lines. The most significant difference on line level appears in the number of literal translations and substitutions as in the dubbed version, 19.5% of the lines are translated literally and 21% are substituted whereas in the subtitled version, the share of literal translations is 53% and only 3% of the lines are substituted. The most commonly used semantic changes were synonymy/antonymy and paraphrase. In the dubbed version, synonymy/antonymy occurs 36 times and paraphrase 25 times. In the subtitled version, the numbers were 24 and 12. The next most common semantic change was trope change with 18 occurrences in the dubbed version and 15 in the subtitled version. Hyponymy occurred 13 times in the dubbed version and five times in the subtitled version. Abstraction change occurred only five times in the dubbed version and once in the subtitled version. The differences in the numbers of literally translated and substituted lines implicate that the subtitled version may convey the plot better than the dubbed version which has to take singability into account. However, it seems that semantic translation strategies may support the conveying of plot rather than disturb it. Thus, to better estimate the conveying of the plot, a further study could be conducted with focus on the lines categorized as substitutions in this research. Nevertheless, the results implicate that paraphrase and synonymy/antonymy may be useful strategies in song translation in order to convey the message of the song. Avainsanat Keywords translation strategy, song translation, audiovisual translation

3 ITÄ-SUOMEN YLIOPISTO UNIVERSITY OF EASTERN FINLAND Tiedekunta Faculty Osasto School Filosofinen tiedekunta Humanistinen osasto Tekijät Author Sanna Maarit Ruokolainen Työn nimi Title Semantic Song Translation Strategies in the Finnish Dubbed and Subtitled Versions of Disney s Frozen Pääaine Main subject Työn laji Level Päivämäärä Sivumäärä Number of pages Date Pro gradu -tutkielma x Englannin kieli ja kääntäminen Sivuainetutkielma liite 6 sivua Kandidaatin tutkielma Aineopintojen tutkielma Tiivistelmä Abstract Tämän pro gradu -tutkielman aiheena on tarkastella ja vertailla semanttisia käännösstrategioita animaatioelokuvan laulujen dubatuissa ja tekstitetyissä versioissa. Tavoitteena on tutkia, kuinka eri versiot eroavat toisistaan sisällöllisesti, ja millaisia semanttisia käännösstrategioita niissä on käytetty. Koska tutkimusmateriaalina on animaatiomusikaali, jonka lauluilla on narratiivinen funktio, tutkielman tavoitteena on myös tarkastella, vaikuttavatko käytetyt käännösstrategiat juonen kulun välittymiseen. Tutkimusmateriaali koostuu elokuvan Frozen kuudesta laulusta ja niiden suomenkielisistä dubatuista ja tekstitetyistä versioista. Tutkielman menetelmä nojaa Ritva Leppihalmeen (2007) ja Andrew Chestermanin (2016) käännösstrategialuokitteluihin. Analyysi on kaksivaiheinen: ensin laulujen säkeet jaetaan Leppihalmeen strategioita mukaillen suorasti käännettyihin säkeisiin, korvattuihin säkeisiin ja semanttisia muutoksia sisältäviin säkeisiin. Tämän jälkeen tarkastellaan lähemmin niitä säkeitä, jotka sisältävät semanttisia muutoksia ja luokitellaan ne Chestermanin semanttisten käännösstrategioiden mukaan. Semanttiset käännösstrategiat, jotka ovat tämän tutkielman keskiössä, ovat parafraasi, synonymia/antonymia, hyponymia, abstraktisuuden muutos ja kielikuvan muutos. Tulokset osoittavat, että dubatussa versiossa 59,5 % säkeistä sisältää semanttisia muutoksia, kun taas tekstitetyssä versiossa vastaava luku on 44 %. Suurempi ero säetasolla on suorien käännöksien ja korvattujen säkeiden jakautumisessa, sillä dubatussa versiossa suorasti käännettyjä säkeitä on 19,5 % ja korvattuja säkeitä 21 % kun taas tekstitetyssä versiossa suorasti käännettyjen säkeiden osuus on 53 % ja korvattujen säkeiden vain 3 %. Semanttisista muutoksista yleisimmät olivat synonymia/antonymia ja parafraasi. Dubatussa versiossa synonymiaa/antonymiaa esiintyi 36 tapauksessa ja parafraasia oli käytetty 25 kertaa. Tekstitetyssä versiossa vastaavat luvut olivat 24 ja 12. Seuraavaksi yleisin semanttinen muutos oli kielikuvan muutos, joka esiintyi dubatussa versiossa 18 ja tekstitetyssä versiossa 15 kertaa. Hyponymiaa esiintyi dubatussa versiossa 13 ja tekstitetyssä versiossa 5 kertaa. Abstraktisuuden muutos esiintyi vain 5 kertaa dubatussa ja kerran tekstitetyssä versiossa. Erot suorasti käännettyjen ja korvattujen säkeiden määrässä voivat tarkoittaa, että tekstitetyt versiot lauluista välittävät juonen kulun paremmin kuin dubattu versio, jossa pitää priorisoida laulujen laulettavuus. Toisaalta vaikuttaa siltä, että semanttiset käännösstrategiat pikemminkin tukevat juonen välittymistä kuin häiritsevät sitä, joten erot laulukäännösten välillä eivät anna suoraviivaisia tuloksia juonen kulun välittymisestä. Jotta juonen kulun välittymistä voisi arvioida tarkemmin, voisi tutkimukseen ottaa mukaan säkeet, jotka tässä tutkimuksessa luokiteltiin korvatuiksi, sillä niissä merkitys on kauimpana alkuperäisestä versiosta. Joka tapauksessa, tulosten perusteella synonymia/antonymia ja parafraasi voivat olla hyödyllisiä käännösstrategioita laulujen kääntämisessä. Avainsanat Keywords käännösstrategiat, laulujen kääntäminen, av-kääntäminen

4 Contents 1. Introduction Theoretical Background Translation Strategies General Viewpoints to Song Translation Musical Song Translation Audiovisual Translation: Dubbing and Subtitling Translating Songs for Subtitles and Dubbing Translating for Children Material Frozen Songs in Frozen Method Analysis Frozen Heart Do You Want to Build a Snowman? For the First Time in Forever Love is an Open Door Let it Go For the First Time in Forever (Reprise) Discussion Conclusion References Appendix... 83

5 1. Introduction As Finland is a predominantly subtitling country, research in audiovisual translation has for long focused on subtitling and not so often taken dubbing into account. Still, the majority of young children s television shows and movies are dubbed in Finland, as well as in other subtitling countries. In recent years, translating for children has been noted as an important field of study (Lathey 2010) and in its wake, the dubbing of TV shows and movies for children has also been studied to some extent. Similarly, song translation research in Finland has only recently achieved attention as a field of research. Therefore, song translation for children in dubbing and subtitling is an almost non-researched topic of research in Finland. This MA thesis intends to fill this research gap in translation studies. This MA thesis is a case study on song translation strategies in the Finnish subtitled and dubbed versions of Disney s animated motion picture Frozen. In this study, song translation and its strategies will be studied from the perspectives of both dubbing and subtitling. The aim of this study is to compare the subtitled and dubbed versions of song translations in Frozen, focusing on the semantic translation strategies used in them. As the two translated versions feature different song lyrics in the movie Frozen, it pays to take a closer look into the differences between them. Song translation for different purposes includes various considerations, for example the dubbed versions of songs are meant to be sung and therefore their primary function is to work as singable songs. By contrast, the subtitled versions are expected to present an illusion of a song but need not work as a singable one. However, since Frozen is a musical, its songs have an important function in conveying the story of the movie; in other words, the song lyrics have a narrative function. Consequently, both the dubbed and subtitled versions are expected to convey this narrative function and therefore, a further aim of this study is to examine how the story of the movie is delivered in the two mentioned forms of audiovisual translation. The analysis will consist of two phases: first the lyrics will be examined on line level and secondly, a more thorough word-level analysis will take place. The analysis will be done by comparing the semantic contents of the song lyrics. The method of the analysis will rely on Ritva Leppihalme s (2007) and Andrew Chesterman s (2016) categorizations of translation strategies with focus on Chesterman s semantic translation strategies. Chesterman (2016: 90) classifies the semantic strategies into 10 1

6 different subcategories. Chesterman s model will be introduced in section 2.1 and how his model will be adapted into this study will be discussed in section 4. Although the analysis is conducted as a comparative text analysis between the two translated versions focusing on the semantic content of the lyrics, other important aspects of song translation and audiovisual translation will be taken into account. For example, among the most important points to consider in song translation are rhythm, rhyme and singability (Low 2003, 2013; Franzon 2005, 2008) and it is likely that the lyrics of the dubbed version of Frozen have been influenced by these requirements. In addition, both dubbing and subtitling have several restrictions which are caused by the spatial and temporal constraints of audiovisual translation and the restrictions have presumably affected the translation choices in Frozen. Since the dubbed and subtitled versions of the movie Frozen are made for different purposes e.g. dubbing is meant to be sung, subtitles not it is not relevant to compare how the translations vary in rhythm or singability. Nevertheless, these restrictions and requirements of both song translation and audiovisual translation will be acknowledged as influential factors although they will not be a crucial part of the actual analysis. In addition, this study will not attempt to evaluate whether one of the translations delivers the plot conveying function better. Rather, this MA thesis will study the differences in an exploratory manner and attempt to find out which kind of semantic changes have been made to reach a functioning translation in the demanding context of song translation in audiovisual form. The overall focus of the present study will be on comparing the semantic translation strategies for song translation in different types of audiovisual translation. My research questions for this study are: How are the song translations different in the subtitled and dubbed versions? More specifically: How does the semantic content of the lyrics vary between the translations and compared to the original English lyrics? What kind of semantic translation strategies have been used in the two translations? Which of the translations includes more semantic changes and are certain kind of semantic translation strategies prevalent in one or the other translated versions? Finally: Can any conclusions about restricting factors or the conveying of plot be made on the grounds of the semantic changes observed? 2

7 This topic is relevant to study for numerous reasons. In addition to the research gap mentioned earlier, translating for children requires attention. In Gillian Lathey s (2010: 205) words, Translating for children is a delicate art and many researchers agree that translating for children is a responsible task because the quality of the translations may have an impact on children s developing language skills. This responsibility is even more clearly present in dubbing for children because children tend to watch their favorite programs and movies multiple times over (Tiihonen 2007). In addition to the responsibility dictated by the age of target audiences, subtitle quality is of great importance regardless of the target group. This is because a considerable part of text read daily in Finland consists of subtitles (Salmi & Vihonen 2007). Additionally, as subtitles can be used as a language learning tool (Caimi 2013), their quality should be taken into careful consideration. Song translation in subtitles has been studied in a few Finnish studies. For example, Leena Hyttinen s (2007) MA thesis, Oikopolkuja toimiviin laulukäännöksiin ruututeksteissä, discusses the functionality of song translation in the subtitles of a television show Short Cuts. Similarly, Laura Siitonen s (2014) MA thesis on subtitling songs for the Eurovision song contest looks into the translation of songs in subtitles. However, research on song translation in dubbing is almost non-existent. The only such research I am aware of is Jenni Koski s (2003) MA thesis on song translation in Disney s motion picture Tarzan. Koski s study is a reception study of how children of a certain age understand the song lyrics in the dubbing and thus its topic is quite closely related to my study. However, this study attempts to analyze the differences in song translation strategies between two modes of audiovisual translation and this, to my knowledge, has not been studied before. Therefore, it is a relevant topic to study and it can offer clarifying information about how different song translations can be depending on their function. It will also be interesting to see if the results of this study can be used to describe the semantic translation strategies available to song translators. This research paper will proceed as follows: translations strategies, song translation, audiovisual translation as well as translating for child audiences will be introduced in Section 2, and the material, Frozen and its six songs to be analyzed, in Section 3. Section 4 will discuss the methodology of this study as well as the adaptation of Chesterman s translation strategy model whereas Section 5 will discuss the results of the analysis. Finally, Section 6 will conclude this study as well as discuss the validity of the results elicited. 3

8 2. Theoretical Background The topic of this study includes characteristics from many subcategories of translation. This section will introduce these different theoretical aspects relevant to the present study. Translation strategies will be discussed first, in section 2.1. Song translation in general will be discussed in section 2.2, musical song translation in and song translation in audiovisual translation in , whereas section 2.3 will introduce audiovisual translation in general. Finally, section 2.4. will introduce translating for children and its effects on the focus of this study Translation Strategies Translation strategies refers to a translator s solutions to specific translation problems (Leppihalme 2007). A translation strategy can also be seen as a process of solving such a problem (Chesterman 2016). Other terms sometimes used which carry the same meaning include, for example, translation rule, translation solution or translation method. Ritva Leppihalme (2007: 365) points out that the term strategy emphasizes the translator s active role in the translation process and it has become a more established term in the field. Although the term strategy suggests that the solution for a translation problem is chosen carefully, translation strategies can be either conscious or unconscious choices. However, to delimit unconscious strategies from conscious ones is, as Andrew Chesterman puts it, a tricky business (Chesterman 2016: 88) as conscious choices can become automatized, and thus unconscious, when repeated numerous times and similarly, unconscious choices can become consciously planned. In addition, how conscious translation choices are, is arguably dependent on the translator and his or her experience (Chesterman 2016: 89). Thus Chesterman (2016: 88) refers to translation strategies being potentially conscious. Research related to exploring translators problem-solving processes is a wide field in itself and therefore, the level of consciousness in translation strategies will not be further debated in this study. Chesterman (2016: 86) states that a translation strategy can be observed, for example, as a problem-solving process, or a goal-oriented activity. In his view, translation strategies are a translator s way of conforming to norms in translation studies. In this context, conforming to 4

9 norms does not necessarily pertain achieving equivalence. Rather, translation strategies are tools which help translators produce a best version in the current translation situation. Chesterman (2016: 87) also sees that strategies are a type of linguistic behavior and therefore always result in some kind of textual manipulation. Translation strategies can be divided into global and local strategies (Leppihalme 2007). These terms were first used in the 1980s by Candace Séguinot (1989) and over the years they have become established in the field of translation studies. A global strategy is a general choice how to translate a whole text or factors that are present in the whole text, for example a dialect choice or how to translate cultural items in a text (Chesterman 2016). A global strategy can entail, for example, a choice between domesticating and foreignizing a text. A local strategy, however, deals with finding a solution to a specific translation problem within a text (Leppihalme 2007). Local strategies can cover translating certain structures or, on a lexical level, for example culture-bound items. As Leppihalme (2007: 366) points out, local strategies are often but not always dependent of the global strategy chosen. For example, if the global strategy is to modernize the language of a historical novel in its retranslation, the local strategy for archaic words in the novel is to find modern equivalents for them. This study deals only with local strategies and thus they will be further discussed in the following paragraphs. Several different classifications for local strategies have been made in the course of time. There are classifications for translation strategies which can be applied to all kinds of translation and those which concern specific sub-fields of translation. Among the most widely known classifications are those of Vinay and Darbelnet, (1958 as quoted in Chesterman 2016: 90) which lists borrowing, calque, literal translation, transposition, modulation, equivalence and adaptation as strategies, and Eugene Nida s (1964 as quoted in Chesterman 2016: 90) which consists of four categories: change of order, omission, change of structure and addition. On the basis of previous researcher s works, Leppihalme (2007) sums up the general local strategies into four categories: preservation, change, addition and omission (Leppihalme 2007: 368). The local strategies which Leppihalme (ibid.) mentions, will be used as a basis for the analysis in this study and thus they will be further discussed in the next paragraph. In addition to general classifications of translation strategies, classifications have been compiled 5

10 and drawn up for more specific purposes in individual subcategories of translation, for example, audiovisual translation and song translation to name but a few. As local strategies are often solutions to very specific and individual translation problems, Leppihalme (2007: 368) defines them on word-level, but mentions that they can also be used with different parts of texts varying in length (translation units). The first strategy, preservation, entails that a word is transferred in the translation as such or with minor changes. The strategy of change is defined rather loosely in Leppihalme s text since she describes it as simply changing the text from one language to another. However, later on Leppihalme quotes Chesterman (2016) of the possible ways to change a text and presents synonymy and hyponymy as examples of local changes. The strategy of omission means leaving out a word or phrase in a source text in such a way that the target text does not contain an equivalent for it. Conversely, addition means that a word or phrase that does not exist in the source text is added in the translation. According to Leppihalme (ibid.), this can be done for example to compensate some omitted words. For example, if wordplay has been omitted earlier on in the text, some wordplay can be added somewhere else to maintain a playful tone in the target text (Leppihalme 2007: 368). Chesterman (2016: 86) points out that strategies are always a kind of textual manipulation, and therefore he concisely suggests that on the most general level, there is only one strategy to change something. He points out that despite its vagueness, change is actually a good term to use, because the source text is always changed in some ways when it is translated. However, he specifies that change does not refer to all kinds of changes, but: The changes in focus here are those that involve a choice between possibilities. This means we are not concerned with changes that are obligatory for obvious grammatical reasons, like adding a definite article in English when translating from a language which does not use one. (Chesterman 2016: 90) Chesterman s own classification for translation strategies, which Leppihalme (2007) refers to, has been developed on the basis of Vinay and Darbelnet s (1958) and Nida s (1964) as well as a few other classifications. He has attempted to draw up a heuristic classification which could be useful for translators in practice. Chesterman s classification is comprised of three primary categories: Mainly syntactic/grammatical, mainly semantic and mainly pragmatic (2016: 90). 6

11 He has named the groups thus to acknowledge that they often overlap and that they can be broken up to subcategories in several different ways. In total, Chesterman s classification comprises of 30 different strategies. I will not introduce every one of them in detail as some of them are beyond the scope of this study. However, all the primary strategies and their subcategories can be seen in Table 1 below: Table 1. Chesterman's classification of translation strategies Syntactic strategies Semantic strategies Pragmatic strategies G1: Literal translation S1: Synonymy Pr1: Cultural filtering G2: Loan, calque S2: Antonymy Pr2: Explicitness change G3: Transposition S3: Hyponymy Pr3: Information change G4: Unit shift S4: Converses Pr4: Interpersonal change G5: Phrase structure change S5: Abstraction change Pr5: Illocutionary change G6: Clause structure change S6: Distribution change Pr6: Coherence change G7: Sentence structure S7: Emphasis change Pr7: Partial translation change G8: Cohesion change S8: Paraphrase Pr8: Visibility change G9: Level shift S9: Trope change Pr9: Transediting G10: Scheme change S10: Other semantic changes Pr10: Other pragmatic changes According to Chesterman (2016: 91), syntactic strategies primarily manipulate form whereas semantic strategies manipulate nuances of meaning (lexical semantics as well as clause meaning). However, pragmatic strategies involve a selection of information according to the translator s knowledge of the intended readership. Pragmatic strategies are larger changes and they include also syntactic and semantic changes. If syntactic strategies manipulate form and semantic strategies meaning, pragmatic strategies manipulate the text s message itself. Chesterman (2016: 104) states that pragmatic strategies are often a result of a global decision concerning how to translate the entire text. As this thesis studies semantic translation strategies, only those will be discussed further here. 7

12 The first strategy, synonymy, pertains that instead of the most commonly used equivalent for a word, a synonym or near-synonym is chosen. According to Chesterman (2016: 99), the strategy of synonymy is used, for example, to avoid repetition. An example of synonymy is using issue instead of magazine. Antonymy means that an antonym with negation is used instead of a synonym. An example of antonym is using not inclusive instead of exclusive. The strategy of hyponymy can entail three different subclasses which Chesterman classifies as follows: a) ST superordinate -> TT hyponym, in which a source text word has been substituted with a hyponym in the translation. This type can also be called specification (Chesterman ibid.). b) ST hyponym -> TT superordinate, in which a source text word has been substituted with its superordinate in the translation. This type can also be called generalization. c) ST hyponym X -> TT hyponym Y, in which the words in both source text and target text are hyponyms of the same superordinate. The next category, converses, express the same situation from different viewpoints, for example buy vs. sell (Chesterman, 2016: 100). Abstraction change changes the level of abstractness or concreteness, whereas distribution change expresses the same message with fewer or more items. The strategy of emphasis change adds, reduces or alters emphasis. For example, a common way to add emphasis in English is to use the adverb highly as a premodifier to adjectives. Emphasis can also be altered so that a different thematic element is emphasized in the translation. The next category, paraphrase, results in a translation described as loose or undertranslated. In Chesterman s model, paraphrase favors the pragmatic sense or meaning of a larger unit, for example a clause, and to render this meaning, semantic components are not as important. In other words, paraphrase renders the same meaning by completely different choices on a lexical level. Chesterman states that paraphrase is often used in the translation of idiomatic expressions (2016: 101) when there is no corresponding expression in the target language. Similarly, the strategy of trope change is used in translating figurative speech, for example metaphors. Chesterman has divided it into four subclasses which are presented below: 1. ST Trope X -> TT Trope X: a figurative element, for example a metaphor, is retained. It can be retained in three different ways: the ST trope can be transferred in the TT as the same term 8

13 in the level of lexical semantics, as a trope of same type, but not semantically identical, or thirdly, as of the same type, but not at all related semantically. 2. ST Trope X -> TT Trope Y: the general feature of figurativeness is retained, but its realization is different both semantically and pragmatically. 3. ST Trope X -> TT Trope Ø: a figurative element is dropped altogether. 4. ST Trope O -> TT Trope X: a figurative element is added. In addition to these categories, the last category of other semantic changes includes, for example, change of physical sense of deictic direction. Because Chesterman places much value on defining changes, most of the categories involve a change of some kind and therefore his classification is quite different from Leppihalme s (2007) one. However, some parallelism occurs between Leppihalme s preservation and Chesterman s Literal translation (G1) although Chesterman places it in the category of syntactic strategies and defines it as literal in form, not in meaning (Chesterman 2016: 91). Chesterman (2016: 92) does not define if literal translation entails semantic equivalence, but mentions in an example that slight differences in meaning can occur in literal translations. Similarly, Chesterman includes addition and omission in his strategy of Information change (Pr3) although in his view (2016: 107), information is added or omitted only if the translator deems it relevant or irrelevant to the readers of the target text. Leppihalme s four strategies and Chesterman s classification will be used as the basis of the analysis in this study. However, not all categories will be used in this study, and some of the categories will be defined somewhat differently for the purposes of this study. Section 4 will discuss how these two classifications will be adapted to this study in detail. This section has discussed translation strategies and introduced Chesterman s (2016) classification for translation strategies which will be used as the basis of analysis in this study. The next sections will introduce audiovisual translation, song translation and translating for children which all pose their own restrictions and challenges as well affect the scope of the study. 9

14 2.2. General Viewpoints to Song Translation Song translation comprises several different forms which are the result of a multitude of music genres, the instances where the songs are performed and the purpose and functions of the songs, song lyrics and their translations. As Marta Mateo (2012: 115) notes, song translation comprises several different forms, ranging from opera translations, popular song translations or adaptations to subtitling or dubbing songs. Peter Low (2013: 72) points out that often when song translation is discussed it is regarded as one entity and it is easily forgotten that song translations have various functions. For example, some song translations are meant to be sung and some to be presented in text form. The translation strategy used in song translation is arguably very different depending on the purpose of the translation, for example between surtitling opera and translating a popular song. Research in song translation has been scarce until recent years (Mateo 2012; Franzon: 2008) but lately it has been studied in different instances. The term song translation is not the only established term in the field as others such as music-linked translation or translation of musical texts are also used (Mateo 2012: 115). Due to the nature and limitations of song translation, researchers have argued whether to call it translation at all or to name it adaptation or rewriting (Franzon 2008; Mateo 2012). As Mateo (2012) points out, song translation requires a multidisciplinary approach demanding not only expertise in translation but also knowledge about music and lyrics and this may be another reason why its research has been relatively scarce. Other reasons contributing to the small number of research can be that the amount of song translations done is much smaller than other types of translations and as Johan Franzon (2008: 374) notes, songs are more often translated by other professionals singers, songwriters or playwrights rather than professional translators. When translating a popular song, the purpose is first and foremost to produce a singable translation. A singable translation is also the desired outcome of a musical song translation but it has several other functions as well. The term singability has been defined in different ways by different researchers. For example, Low (2013) defines singability as phonetic suitability of the translated lyrics. However, according to Low (2013: 73), the definition of singability depends on whether the target user of the translation is the singer or the audience. 10

15 Thus, Low s definition of singability as phonetic suitability could be seen as important from the singer s point of view. On the other hand, Franzon (2008) proposes that singability should be understood in a broader sense: according to him, a singable translation is suitable in every relevant way for the particular purpose as a good translation is described to be according to the skopos theory (Franzon 2008: 375). On the basis of both this broader viewpoint and Low s interpretation, Franzon (ibid.) defines singability as the attainment of musico-verbal unity between the text and the composition. This consists of several elements of music and text which will be introduced in more detail later in this section. To evaluate whether a song translation is singable, Low (2003: 92) introduces his Pentathlon principle to song translation. In this approach, the criteria which he deems as most important are singability, sense, naturalness, rhythm and rhyme. According to Low, sense refers to semantic accuracy whereas naturalness entails e.g. word order and register. Rhythm and rhyme, in turn, refer to the poetic nature of lyrics. However, Low does not offer a detailed explanation what his interpretation of singability includes. In accordance with his definition of singability as phonetic suitability, he mentions the importance of phonetic features, e.g. favoring easy consonant clusters, but claims that it is up to the singers to judge what a singable text is like. However, it seems to me that using the same term singability in different levels, i.e. that a singable translation consists of five different criteria, one of them also being singability, can create misunderstandings. Thus, I agree with Savonen (2008: 239) who points out that Low s criteria are in some ways overlapping. Savonen (ibid.) comments that the five criteria are intertwined, as he points out that rhythm and rhyme, for example, can bear meaning or facilitate the transfer of meaning (sense) which in turn can affect the singability of a song. Although Low (2003: 101) emphasizes that no criterion should be considered as the most important at the expense of other features, he mentions that sense or semantic accuracy is the one of the elements which can be stretched or compromised most often for the sake of a singable translation. Tiina-Elina Mannila (2005) agrees with Low s definition of singability as phonetic suitability and adds that vowels, diphthongs and consonants all together with the pitch of notes have an effect on singability. Mannila (2005: 194) points out that the pitch of a note has an impact on how phonemes are produced and for example, some syllables are easier to produce on a 11

16 higher pitch than others. Mannila (ibid.) argues that singability can also be defined as presentability, which in addition to the phonetic features also entails that the meaning of the song can be easily interpreted. This supports Savonen s (2008) view that rhythm and rhyme can facilitate the transfer of meaning. In other words, different textual features together with the music support each other and work together to carry out the function of the song. To explain his definition of singability as musico-verbal unity, Franzon (2008: 390) proposes an approach which he calls the three layers of singability. These layers consist of three criteria, or matches of song lyrics that are: 1. prosodic match, 2. poetic match and 3. semanticreflexive match. The prosodic match includes that the lyrics sound natural and are understandable when sung, which on a textual level can be observed via the syllable count, rhythm or intonation as well as sounds that are easily vocalized. The notion of vocalization is in accordance with Low s phonetic suitability. The poetic match entails that the lyrics grab the audience s attention with a poetic effect. According to Franzon (ibid.), a poetic effect can be achieved with the use of rhyming, segmentation of phrases and locating key words carefully in the lyrics in a way that they work together with the music. Lastly, the semantic-reflexive match means that the lyrics are perceived as meaningful and that meaning is transferred to the audience. This third match can be evaluated, for example, on the terms of how the story of the song is told and its characters portrayed and more specifically in the use of description and for example, metaphors. Although Franzon (2008) regards all of the layers or matches as important, he points out that the first one, prosodic match, is the most important to render in the translation because it enables the lyrics to be sung aloud technically. The need for the other two layers should be decided according to the nature of the song. It needs to be noted, however, that this approach is dedicated to songs which need to prioritize singability and Franzon himself has other function-based approaches for songs that have other important functions, for example songs in a musical. These musical song translation strategies will be introduced in section According to Savonen (2008), Low is one of the greatest contributors to research in song translation and arguably his Pentathlon principle has been used in many song translation analyses. However, it needs to be noted that Low himself has mostly studied popular song 12

17 translation and therefore it may not be the best method in analyzing musical song translations, where a visual communication channel is added to that of the music and lyrics. Low s emphasis on popular song translation can be seen in his statement how a singable translation is never ideal for other purposes (2013: 73). In his opinion the constraints of producing a singable translation distort the text and the result is always an adaptation of the original lyrics. However, when translating a musical song, its singability and its narrative function need to be considered as equally important. In a musical or a movie where songs are used, they have the function of conveying the plot and story with their lyrics and therefore the translations should have a similar function. In my opinion, a song translation in a musical must also fulfil the function of conveying the story and cannot be an adaptation but have semantic equivalence at least to some extent. Thus, Low s approach will not be used as a method in this study although his notions of song translation will be taken into account in the analysis. Since musical song translation is a special case of song translation, it will be discussed in more detail in the next sub-section. Not all song translations are meant to be sung. These can include, for example, opera surtitling or subtitling songs in a movie or a television show. For example, in opera surtitling, the song translation is needed for the audience to be able to follow the story of the opera which is often performed in a language that the audience does not understand (the most common languages in opera being Italian, German, French, English, Russian and Czech [Mateo 2012: 116]). Subtitling songs for cinema or TV can be said to have a similar function: to convey the meaning of the lyrics. However, song subtitling can have differing guidelines and restrictions due to the constraints of audiovisual translation. Thus, song translation in audiovisual instances will be discussed in more detail in section 3.2.1, after audiovisual translation in general has been introduced Musical Song Translation The term musical song translation most often refers to song translation for staged musicals in theater productions. Sometimes it is also more explicitly called musical theater translation. In the context of this study, musical song translation is used for both theater and movie musicals for the sake of clarity and because the most important theories of musical theater translation 13

18 also apply to musical songs in movies with some minor differences. Both theater musicals and movie musicals have a visual communication channel which means that the same meaning transferred through songs is also presented visually by the actors and the surroundings. Songs in both types of musicals have narrative functions and they can also support characterization. In the context of this study, musical songs refer to songs that are sung by the actors (or characters) as songs that are used as background music are beyond the scope of this study. In addition, in this study, function of songs means first and foremost the song lyrics function. Songs can have additional functions, such as creating a certain ambience or mood, which are created by musical elements, such as tone or what instruments are used. However, since this study is concerned with song lyrics and the music-related aspects are beyond the scope of this study, song function is used as the primary term. Franzon (2005, 2009) has analyzed and compared the Swedish, Norwegian and Danish translations of the stage musical My Fair Lady extensively. On the basis of his studies, Franzon proposes a functional approach to musical song translation, in which the strategy for musical song translation depends on the overall functions of the songs but it also distinguishes functional units in the song lyrics which will be introduced later on in this section. The functions of songs are connected to their content and Franzon (2005: 272) categorizes songs into song modes which depend on which level or function is most important. The three levels he names are narrative, staging and presentational. The narrative and staging levels are further divided into two modes respectively. Thus, according to Franzon, musical songs sung by the actors can be divided into the following five modes: 1. Narrative: Soliloquy 2. Narrative: Unison singing 3. Staging: Addressed monologue 4. Staging: Addressed dialogue 5. Presentational: Song as singing The narrative level consists of the transfer of the plot but it also conveys the characters feelings or motives. Soliloquy is a term used in drama, defined in Merriam-Webster Online Dictionary as a dramatic monologue that represents a series of unspoken reflections or a long, usually serious speech that a character in a play makes to an audience and that reveals the character's thoughts. In the context of song translation, soliloquy is therefore a song 14

19 which is sung by one character and which reveals this character s thoughts and feelings. Unison singing, according to Franzon (2008: 272), usually includes shared convictions presented as some kind of collective monologue. In the staging level, the interaction between actors [ ] shares the social, spatiotemporal context of the spoken dialogue (Franzon 2008: 273). The staging level can be either monologue or dialogue but in contrast to the narrative level, the song is addressed to other characters or the audience. In the four previous modes, the characters are not aware that they are singing but the presentational level entails songs as singing. It can, for example, appear as one character presenting a song to another. Songs can belong to more than one level, for example, a presentational song can also be an addressed monologue and often the staged monologues or dialogues have also a narrative function. Therefore, Franzon (ibid.) suggests that songs can be categorized according to their primary function. Franzon s (2008: 263) functional approach also entails that semantic meaning, or fidelity, as he prefers to call it, must be present to some extent. However, in his view, the similarity between the source and target texts works better on a contextual-functional level rather than a textual-semantic level which has been the traditional approach in translation studies. What Franzon (2008: 267) means by this is that, similarly to the skopos theory, the translator should recognize what the source text wants to say i.e. what is the function of the text and how the function is delivered on a contextual level and render the same meaning into the target text without having to repeat every linguistic or structural feature of the source text. As Franzon puts it: Rather than direct translation, equivalent to the quoting of verbal content, the song translator s task would be to perform a creative transposition, approximating the source text as much as possible or as little as necessary, and perhaps negotiating between the formats determined by the functional units. (Franzon 2005: 275) A functional unit, which Franzon refers to, is a term coined from another term, translation unit which is an old and established concept in translation studies. For example, Leppihalme (2007: 368, see p. 9) defines a translation unit as a part of text to be translated (and possibly requiring a translation strategy). The term functional translation unit has been used, for example, by Christiane Nord (1997). In Nord s view, texts have several functions which are conveyed with markers: textual, structural, syntactic or lexical ones. Nord (1997: 45) states that one function can be marked with several different markers and all markers that convey a particular function 15

20 thus form a functional unit. According to Franzon (2005), functional units that may be found in musical song texts include deictic reference, plot information, cultural reference and musical and intermedial coherence (Franzon 2005: 263). On a textual, or contextual, level they can be marked with pronouns, verbal modes, utterances implying emotion or attitude, deictic or spatiotemporal references [e.g. demonstrative pronouns] [ ] and implied staged activity, including gestures or verbal behavior (Franzon 2005: 274). In his analysis of translations of My Fair Lady, Franzon noted that although the three translations were all different from the original text in their own ways, they had similar strategies to prioritize intratextual coherence. This had been done with interpreting and rearranging the functional units, as Franzon summarizes: A link to a functional context is created by using other words than those of the closest possible semantic approximation of the source text (Franzon 2005: 276). Interestingly, although Low does not underline functionality or use the same terminology in his Pentathlon principle, his guideline for transferring meaning without clinging to the exact semantic equivalents is actually quite similar to Franzon s functional approach: Thus a precise lexical equivalent may be replaced by a nearsynonym, a narrow term by a superordinate term, a particular metaphor by another one which functions in a similar way in the context. (Low 2003: 94). Both Low s guidelines and Franzon s functional approach have similarities with Chesterman s (2016: 92) literal translation strategy as well as his semantic translation strategies which were discussed in section 2.1. Chesterman (2016) mentions that slight semantic changes can occur even in literal translations and his categories of synonymy, hyponymy and trope change are similar to Low s guidelines above. Franzon s (2009) study of the Scandinavian (Swedish, Danish, Norwegian) translations of My Fair Lady consists of fidelity towards the songs music, their theatrical function and their contents. Although Franzon refers repeatedly to the functional units, he does not use them as a method for evaluating the content of the song translations. Rather, he (2009: 185) mentions text approximation as a tool for evaluating the fidelity towards content. He describes text approximation as a combination of three textual strategies: metaphrase, paraphrase and addition which he bases on an age-old model of John Dryden (1680) who described 16

21 metaphrase, paraphrase and imitation as the three main strategies for translators. In Franzon s interpretation of the strategies, metaphrase means a direct, word-for-word translation and semantic equivalence in its strictest form. Paraphrase, on the other hand, means that the translator has strived for retaining the meaning of the original text but with words other than direct equivalents. Franzon (2009: 191) states that there are virtually countless ways to paraphrase words, but gives a few examples. For example, paraphrase can be carried out with changes in style (eat nibble) or specificity (chocolate confection). A paraphrased word can also be a part of a larger whole (face chin) or it can entail changes in word class or verbal mode. (Examples from Franzon 2009: 191, my translations). These changes also have connections to Chesterman s (2016) categories of semantic changes discussed in section 2.1. Finally, addition means that the translator has used words which do not appear at all in the source text. Franzon (2009: ) divided the lyrics into the smallest units possible, which means that the categories he compared in the translations included, for example, subject, object, predicative expressions, lexical phrases and modal verb phrases. Franzon admits that the way of comparing single words and their equivalents is timeconsuming (Franzon 2009: 192) but it is an easier method quantitatively than to look at, for example, lines at a time. Franzon s (2009: 198) results indicate that there were significant differences in which strategy was the most frequently used between different kinds of songs and the three languages. The mean percentual value (of all the languages and songs) showed that metaphrase was the most often used strategy and was used in 48% of the cases. Paraphrase was used in 19 % and addition in 33 % of the cases. It needs to be noted, however, that all the three Scandinavian languages are cognate languages of English and the large number of metaphrased words may be attributed to this fact. If this analysis were to be used with different languages, it could reveal significantly different distribution of the strategies Audiovisual Translation: Dubbing and Subtitling Audiovisual translation refers to a type of translation which deals with multimodal texts, i.e. texts that are linked to sound and picture. It is used, for example, in the translation of movies and television programs, in opera and theater as well as the localization of games. Audiovisual translation (from now on, AVT) has many different types, including subtitling, surtitling, dubbing, voice-over, audiodescription and subtitling for the deaf and hard-of-hearing 17

22 (Chaume 2012b: 1), of which the two most common are dubbing and subtitling. In Frederic Chaume s (2012a: 1) words, dubbing consists of replacing the original track of a film s [ ] source language dialogues with another track on which translated dialogues have been recorded in the target language. Subtitling, on the other hand, presents a written translation of the dialogue in the bottom of the screen without replacing the original dialogue (Diaz Cintas & Remael 2007). Usually one or other of these types is the preferred type of AVT and used more widely in a country although sometimes there is no clear division between them. However, in most countries young children s TV shows, films and cartoons are dubbed regardless of the most common type of AVT in the country. Subtitling and dubbing pose several challenges to translators due to their multimodal nature. The dialogue to be translated is always linked to the picture presented on screen and it needs to be taken into account. Other challenging factors in AVT are the spatial and temporal restrictions. The restrictions of both dubbing and subtitling will be discussed further in this section. Dubbing (also known as revoicing) replaces the source language dialogue with a target language one. Dubbing usually has separate voice actors for each role as opposed to voiceover, where one actor can be used for all the roles (Chaume 2012: 3). At its best, dubbing should create an illusion that the movie or TV program was originally done in the target language (Heikkinen 2007: 237). To achieve this illusion, many linguistic factors as well as synchronization must be taken into account. Chaume (2012) defines realistic and plausible dubbing as something which does not distract viewers from the storyline of the program or movie in question (Chaume 2012: 83). Chaume (2012) lists acceptable lip synchrony, credible and realistic dialogue, coherence between images and words, loyalty of the translation, clear sound quality and acting as the most important factors in producing a good quality dubbing. However, there are many more factors that require consideration. For example, synchrony is not limited to lip-sync but consists of different forms such as kinesic synchrony the synchrony between the translation and the actors body movements and isochrony the synchrony between the duration of the original and translated utterances (Chaume 2012: 69). Linguistic factors in producing a dubbing that sounds realistic include the use of register and dialects, colloquial vocabulary and grammatical and syntactical inconsistencies since all of these features are used by people in daily spoken discourse. Tatu Tiihonen (2007: 175) points 18

23 out that lines should sound natural and be easily pronounceable. In practice, this can be achieved by avoiding difficult phonemes or clause structures, as this is what people do in their normal spoken communication as well (Tiihonen 2007: 179). Chaume (2012) describes the linguistic factors needed for good quality dubbing in more detail, as he quotes the guidelines made for dubbing for Spanish television. To achieve credible dialogue, the guidelines suggest, for example avoiding passive voice and encourage using short and unfinished sentences, topicalization, ellipsis, onomatopoeia, puns and informal register. According to Chaume (2012: 85), most guidelines and publications consider credible language and dialogue more important in dubbing than the various forms of synchrony. In even more detail, credible dialogue consists of elements on different language levels, such as prosody, morphology, syntax and lexis. Credibility on these levels can be achieved with, for example, considering intonation and articulation as well as favoring grammatical inconsistencies, short and simple syntactic structures and simple and colloquial vocabulary. However, Chaume (2012: 91) points out that universal guidelines for the language of dubbing cannot be made, as the guidelines should be drawn up bearing in mind the features and norms of the target language. Realistic language is not only important to create an illusion of speech but it is also related to characterization. As Tiihonen (2007) points out, dubbed lines should not only be synchronized but the language should fit the character speaking it. Different characters speech can be distinguished by dialects, accents, vocabulary as well as level of formality to name but a few examples. In subtitling, spatial and temporal restrictions are among the most significant constraints. The number of characters and words that can be fitted on screen is limited and so is the time reserved for the audience to read them (Diaz Cintas & Remael 2007; Vertanen 2007). The maximum number of characters per line (generally approximately 37 characters) varies somewhat between countries and whether the translation has been done for TV, cinema or DVD publication. The character number for cinema subtitles is usually the largest since the size of the screen allows more text. However, the temporal restrictions are the same in cinema translation and thus the differences in character numbers cannot be substantial. In Finland, the conventions vary somewhat between YLE, the national broadcasting company, and commercial subtitling companies, such as Pre-Text and SDI Media (Lång 2013). For example, 19

24 in the television channels of YLE, the maximum number of characters per line has traditionally been 33 (Vertanen 2007) but the number has risen by a few characters over the years. The temporal restrictions can vary according to the films s or program s target group and content. According to Juha Lång (2013: 58), subtitles in children s TV programs and programs which include difficult terminology should stay on the screen for a longer time than average. Although time restrictions can vary according to different factors there is a general six-secondrule in the subtitling industry. This rule entails that a full two-liner consisting of 35 characters on each line can be read in six seconds (Diaz Cintas and Remael 2007). Thus, if the rule is correct, average reading speed of subtitles would be approximately 12 characters per second. In addition to the restrictions of time and space available, subtitles must also present a semantically adequate rendering of what is being said (Diaz Cintas and Remael 2007). To be able to produce a semantically accurate text with a restricted number of characters, translators must condense what is being said. As Diaz Cintas (2010) points out, reduction is the most frequently used strategy in subtitling. Esko Vertanen (2007) also observes that reduction is an integral part of subtitling and sometimes extensive amounts of spoken text must be cut out to be able to fit the essential information in the subtitles (Vertanen 2007: 150). Vertanen points out that the information presented in the subtitles is incomplete as such it is only understandable when presented together with the sound and image. What information is left out depends on many factors but as a general rule, Vertanen (2007) mentions that those parts of speech which are not essential in understanding the plot, can be left out. More specifically, they can include (in Finnish) introductory clauses (e.g. I think that ), names, and to some extent, exact attributes of time and place names. Moreover, anything that the audience is already assumed to know can be left out, for example, things that have already been mentioned or can be seen on screen. Vertanen points out that reduction can sometimes be taken too far. According to Vertanen, it should be remembered that subtitles are always a representation of someone s speech (Vertanen 2007: 153). Therefore, subtitles should not be stripped too bare and careful consideration should be given to word choices to create an illusion of speech. It is also important from characterization s point of view. Other important guidelines for subtitling include that a line should form a meaningful whole. This entails that line breaks are done in a way that does not disturb reading. For example, compound words should never be separated or clauses cut arbitrarily. In Finnish, odd line 20

25 breaks can almost always be avoided by changing the word order (Vertanen 2007) which is very flexible in Finnish. Because subtitles leave the original dialogue present, they have a significant effect in foreign language learning. They can also be used as a language learning tool in education. As Annamaria Caimi (2013) points out, studies show that subtitles can affect language learning positively. It is also suggested that people in countries which favor subtitling learn foreign languages with less effort than people in dubbing countries (Caimi 2013: 168). Subtitles used as a learning tool can be either interlingual (subtitles in the viewer s native language) or intralingual (subtitles in the same language as the soundtrack) (Caimi 2013: 169). In addition to their use as learning material, subtitles are read daily in countries where it is the primary type of AVT. As Vertanen (2007: 147) points out, in Finland (as probably in other subtitling countries as well) subtitles constitute a considerable part of all texts that people read. In Finland, this has been studied by for example Leena Salmi and Inkaliisa Vihonen (2007), who conducted a survey of Finnish adults reading habits as a diary study. The research material consisted of 54 diaries where the participants made notes of all text they read during a day. Although the amount and type of text read differed between the participants, the overall results support the view that subtitles are an important part of text read daily. According to the results, approximately 40% of all read texts were translations, and up to 45% of the translations were subtitles. Therefore, as Vertanen (2007: 147) and Salmi and Vihonen (2007: 6) state, the quality of the language in subtitles must be taken into careful consideration as they can have a large impact on the audience s reading skills. The quality of subtitles is also important from another viewpoint. Tiina Tuominen s (2012) dissertation looked into the audience reception of subtitles. The results of Tuominen s (2012) study indicate that subtitles have an important role as a part of the viewing experience, a conscious or unconscious one. Their fluency and invisibility seem to be most important to viewers, and the absence of translation errors is less important since errors are rarely noticed (Tuominen 2012: 280). One of the most distinct differences between subtitling and dubbing is that the original dialogue is present in the subtitles whereas in dubbing, it is replaced with the target language 21

26 one. Because the original dialogue and the subtitles coexist side by side, they can be compared by those who understand the source language and thus the subtitles are more vulnerable and available for scrutiny than the dubbing (Diaz Cintas 2010). This may mean that there is more pressure to stay loyal to the content in the subtitles than in dubbing. Another significant difference between subtitling and dubbing is that whereas subtitling can usually be done independently, dubbing is always team work (Tiihonen 2007: 181). Tiihonen points out that a translator can only go so far himself/herself with the dubbing and the lines will almost always change in the recording process. The recording situation is the first time when the lines are spoken out loud by the voice actors and they may sound different than what the translator has intended. In addition, if there is a separate director in the dubbing process (in Finland, the translator is often also the director), he or she can have opinions or views on how the lines should be composed. Furthermore, the voice actors themselves may have some suggestions about the lines. Because subtitling can be done independently by the translator, it is clearly the cheaper of the two alternatives. However, the difference in costs is greater in countries with fewer inhabitants. When the number of viewers is larger the costs of AVT (per viewer) are significantly lower and thus dubbing is more profitable (Vertanen 2007: 150). Very often countries with low population count favor subtitling and those with high population tend to lean towards dubbing. This section has discussed AVT and its restrictions on a general level. The following section will focus on how song lyrics are rendered in these two forms of AVT and what kind of challenges the temporal and spatial restrictions of AVT pose for song translation Translating Songs for Subtitles and Dubbing How songs are translated or whether they are translated at all in subtitles or dubbing depends on their function in the movie or TV program. The question whether songs are translated at all is relevant when the songs are for example, played as background music. Generally, in subtitles, if a song bears any relevance to the plot of the program or movie, it is translated. 22

27 In dubbing, as Chaume points out, song translation is not always as straightforward. According to Chaume (2012: 104), in some dubbing countries songs tend not to get translated at all while in some other countries they are subtitled even though the rest of the movie is dubbed. Chaume states that the most important factor in deciding whether to translate a song and this includes both subtitling and dubbing is the function the song has in the film or program: When the lyrics refer to the plot, the song should be translated to give the target audience the same access to the meaning of the lyrics as the original audience has. (Chaume 2012: 104) Chaume states that the preferred mode of AVT to use in the translation of songs is subtitling, with the exception of children s cartoons. He adds that songs in cartoons require an adaptation of the lyrics that matches the music s rhythm in four different poetic rhythms, which are the rhythm of quantity on number of syllables, rhythm of intensity or accentual distribution, rhythm of tone and the rhythm of timbre and rhyme (Chaume 2012: 103). As Diaz Cintas and Remael (2007: 209) point out, if the movie or program is a musical, the choice between translating and not translating is easier, as songs in a musical are almost always translated. If the musical is dubbed, the translated songs are meant to be sung and therefore the choice of the strategy to be used is quite straightforward; it is necessary to create a singable translation which follows the content of the original song and delivers the same function as the original lyrics. Thus, the guidelines for musical song translation apply also to dubbing songs as long as the visual restrictions and different forms of synchrony are taken into account. However, when subtitling a musical, there are different options. Interestingly, several researchers (Chaume 2004; Ivarsson & Carroll 1998) seem to think that other features of song translation, for example rhythm, are as important to present in the subtitles as the content of the song. Ivarsson and Carroll even claim that a song translation in subtitles needs to be singable which is why the rhythm of the lyrics must be rendered in the subtitles. However, they also provide detailed reasons why rhyming is as not important in the subtitles which is presented in the citation below: It should be possible to sing a translation of a song to the music, and the rhythm is therefore more important than the rhymes. The audience can read only one subtitle at a time, and intricate rhyming schemes will often be missed since the rhymes will normally only be noticed if they occur in both lines of the same subtitle or if they are repeated at least three times (Ivarsson & Carroll 1998: 121) 23

28 As stated in the quotation above, Ivarsson and Carroll argue that rhymes are not important to preserve in the subtitles since they are easy to miss because of the limited time the subtitles stay on screen. By contrast to the previous researcher s viewpoints, Chaume (2004) does not mention singability but he considers rhyming and rhythm equally important with content. It also needs to be noted that different companies which provide AV translations can have their own guidelines for translating songs. Hyttinen (2007) conducted her MA thesis on song translation strategies in the subtitles of the TV show Short Cuts and she also discussed subtitling norms in Finland. She contacted several AV translation companies (Broadcast Text, SDI Media, and Pre-Text) and two TV channels (Yle, MTV3) about their guidelines for translating songs. Although the subtitle providers had different opinions about translating songs, the majority preferred that the translations should follow the content of the original lyrics whereas rhyming and rhythm were a good addition, but not necessary (Hyttinen 2007: 29 32). The only other common guideline was that the lyrics of songs should be written in italics whenever possible. According to the results of Hyttinen s analysis, the guidelines of translation companies did not seem to have a significant effect on how the songs were translated (2007: 80) in practice, and the genre of the song seemed to affect the translation strategy more than the guidelines. According to Hyttinen s results, the number of syllables had been kept close to the syllable count of the original lyrics in all of the songs but rhyming, for example, was only used in two of the songs which happened to be children s songs (2007: 81). In her MA thesis, Laura Siitonen (2014) analyzed subtitles of nine songs presented in the Eurovision Song Contest with the help of Low s Pentathlon principle. She studied how four of Low s features, sense, naturalness, rhythm and rhyme, were delivered in the subtitles. Instead of singability, she looked into the restrictions and conventions of subtitling. Siitonen found out that sense was the feature that was most often retained in song translations (2014: 102). Somewhat similarly to Hyttinen s results, Siitonen concluded that rhythm and rhyme were sometimes delivered in the translations but most often not consistently which could indicate that the restrictions for subtitling have affected the translation in many ways (Siitonen 2014: ). In addition, she noted that the restrictions and conventions were followed very closely unlike rhythm, which was the criterion which was most often left out of the subtitles (Siitonen 2014: 104). 24

29 Despite the different guidelines of AVT companies and recommendations of researchers in translation studies, there seems to be no general consensus or conventions about the strategy of subtitling songs. In contrast, the choice of translation strategy to be used seems to depend largely on the context and function of the song as well as the restrictions of AVT. The fact that no strict or consistent guidelines exist may create some leeway for the translator who can polish up the translation to follow the rhyme scheme or the rhythm of the original lyrics whenever allowed by the space and time constraints. However, given the fast-paced work environments of AV translators, it is likely that the translator will focus on what he or she regards as most essential which arguably is the content of the lyrics. Thus, the other features, e.g. rhyme and rhythm, may not gain as much attention Translating for Children Translating for children needs to be addressed since the main target audience of the research material are young children. Translating for children requires special attention from the translator. Children s language skills are still developing and this makes the translator s role a responsible one. Some researchers, for example Tiina Puurtinen (1995), suggest that because of this, translating for children may be more difficult than translating for adults. The responsibility of translating for children is even clearer in dubbing because, as mentioned before, children like to watch their favorite movies and TV shows again and again. It should also be noted as Riitta Jääskeläinen (2007) and Tiihonen (2007) both point out that the language of TV shows and movies can have affect children s language and children can also practice their skills with the help of them. However, some important elements that are present in dubbing for adults do not need similar attention when dubbing for children. For example, lip synchrony is not as important for two reasons. Firstly, a significant portion of the programs and movies which require dubbing are animations and although the technology for animating characters and their mouth movements has developed rapidly, they are still not as clearly visible as with live people. Secondly, as Heidi Heikkinen (2007: 237) points out, children, especially young children, are not as capable in lip-reading as adults are. Similarly, when dubbing for adults, the viewers are aware that the language of the translation is not the 25

30 original and this requires the translator to create an illusion of a text that is originally made in the target language. However, children are not necessarily aware of the language change. Although some constraints of dubbing can be given minor focus in translating for children, some constraints require more attention than in dubbing for adults. Especially when translating songs of a musical movie for children. Since the songs in a musical take the plot forward it is important that children understand the lyrics because following the movie could become more challenging if the lyrics are not understood. The conveying of the story requires using language that is understandable for children. Translating songs for a child audience has not, in my knowledge, been studied in detail but some factors regarding understandability in translating for children can be applied to it. Riitta Oittinen (1997: 103) has studied translating for children extensively and she gives a few examples of textual characteristics which affect a text s understandability for children. These are for example, familiarity of words and simple sentence structures. The notion of sentence structure can be of more significance in Finnish because it is an inflected language and thus, word order in Finnish is relatively free and allows different kinds of sentence structures that are grammatically correct and bear the same meaning. This can offer some leeway for song translators. However, some sentence structures can make the utterance sound too poetic or even comical because their use is not commonplace. Therefore, playing with word order may not be a good strategy in translating songs for children because the lyrics may become harder to understand for them. Another factor that needs to be taken into account when talking about translating for children is the dual readership or dual audience in children s literature and it can also be applied to movies and TV shows targeted at children. As Cecilia Alvstad (2012: 24) notes, children are not the only people reading children s literature or watching children s movies. In addition to children, the books are read by publishers, parents, teachers or librarians, and most often it is the adults who choose what kind of stories are available for children. Dual readership or audience can show in different ways. According to Alvstad (ibid.), some children s books can contain irony, satire or jokes which are clearly targeted at adults reading aloud to children. Sometimes the dual audience is not clearly visible, but it can be shown in the values or ideology of a text or story. Alvstad (ibid) points out that the adult-child dual audience is a 26

31 factor exclusive to children s literature. She also notes that dual audience can be very difficult to render in translations. When all the challenges of translating for children are combined with the restrictions of AVT and song translation respectively, it can be said that translating songs for dubbed children s movies poses notable challenges for translators. This section has presented the theoretical background for this study. The following sections will present the source material for this study and the method which combines the different features of theory presented in this section. 27

32 3. Material The material for this study consists of six songs in the animated musical Frozen. Frozen is a product of Walt Disney Company which has a long tradition of animated musicals, dating back to its first animated feature film Snow White which was published in 1937 (Cooke 2008). Walt Disney himself regarded music as an essential part of his work, stating that Action controlled by a musical pattern has great charm in the realm of unreality (Barrier 1999, as quoted in Cooke 2008). As his legacy, the Disney musicals have placed great value to the quality of the music in the movies and the films have also been nominated for the Academy Award for Best Original Song multiple times (Cooke 2008: 293). This section will further introduce the film Frozen and its songs Frozen The material chosen for the present study includes six songs and their translations in the subtitles and dubbing from the children s movie Frozen. Frozen is a Disney animation published in Disney s motion pictures are very popular worldwide but Frozen reached overwhelming popularity all over the world. Although the movie did not receive unanimous critical acclaim, it was the highest-grossing film of 2013 and at the moment of writing, number 9 in the list of all-time highest grossing motion pictures worldwide (Box Office Mojo 2016) and it has been dubbed into 41 languages. When a children s movie reaches this kind of immense popularity, it can be said that the quality of the translations requires even more attention. In Finland, the film was published as Frozen huurteinen seikkailu [lit. frosty adventure )] and both dialogue and song lyrics were translated by Marko Hartama. The plot of the movie revolves around sisters Elsa and Anna who are princesses in the fictional kingdom of Arendelle. The older sister, Elsa, can control ice with her magical powers which are kept secret from everyone, including her sister Anna. When Elsa comes of age and is crowned the queen of Arendelle, her powers are revealed in a dispute with her sister and she flees to the mountains after accidentally freezing the entire kingdom. Anna sets out to seek her sister to reconcile with her and to return summer to Arendelle. Disney states that the plot is loosely based on Hans Christian Andersen s fairy tale The Snow Queen. Frozen is a musical 28

33 and includes nine songs in total. This means that songs constitute a significant portion of the overall length of the movie and have a function of delivering parts of the story with their lyrics. Frozen is rated PG (Parental Guidance Suggested) in the United States and the age limit of the film in Finland is seven. However, it is likely that children younger than this also view the film. Because of the young age of the target audience, the language used in the dubbing needs careful attention. The age of the movie s and the dubbing s primary target group can thus be estimated to be somewhere between 4 and 12. However, to estimate the target group of the subtitles is not as straightforward. Jääskeläinen (2007) points out that generally, in Finland, programs targeted at children older than 10 are usually subtitled because they are assumed able to read subtitles. However, it does not mean that children older than 10 do not watch the dubbed version. In addition, Disney s movies are also popular among adults but it depends on the viewer if he or she prefers the original or dubbed version. This varied target group is also related to the dual audience which was mentioned in section 2.4. As children s movies and books tend to have two intended target groups, adults and children, this can show to some extent in the original dialogue as well as in the dubbing and subtitles Songs in Frozen There are nine songs in Frozen and all of them have been composed by Kristen Anderson- Lopez and Robert Lopez. The song Let it Go won the Academy Award for the Best Original Song in The source material for this study comprises of six songs of Frozen. The choice of songs was done on the basis of their functions. The delimitation was necessary because including all the songs would have made the source material too large for the purposes of this study. However, some of the songs include dialogue between the lyrics and these parts of dialogue are beyond the scope of the present study. As discussed in section 2.2.1, musical songs can have various functions and this also applies to the songs in Frozen. It seems that every song has a narrative, plot-conveying function, at least to some extent since some of the songs may have other functions as well. Many of the songs are also important in characterization. Furthermore, some songs are seemingly there only for the fun factor, to entertain, but when taken a closer look, they also bear other significant functions. For example, the songs that seem most important in terms of the plot are Frozen 29

34 Heart, Do You Want to Build a Snowman?, Love is an Open Door, Let it Go, For the First Time in Forever as well as its reprise and to some extent, the last song of the movie, Fixer-Upper. In addition, the songs that are important in portraying the characters which sing them include at least For the First Time in Forever and Love is an Open Door which introduce the character of Anna more closely, and Let it Go which is crucial in depicting the character of Elsa. As for the songs Reindeer(s) are Better than People and In Summer are clearly more humorous in nature than the others and entertainment is probably more important in them than conveying plot. Therefore, the songs Reindeer(s) are Better than People and In Summer were left out of the analysis in this study. Similarly, the last song Fixer-Upper was left out because although it has a plot-conveying function, it is not as important as in the other songs that were chosen. In section 2.2.1, Franzon s (2005) five modes of musical songs were introduced. The modes are: 1. Narrative: soliloquy 2. Narrative: unison 3. Staging: Addressed monologue 4. Staging: Addressed dialogue and 5. Presentational: Songs as Singing. The songs were categorized in terms of how many singers they have (soliloquy vs. unison, dialogue vs. monologue), whether the songs are addressed to someone or appear as background narrative separate from the dialogue (eg. monologue vs. soliloquy), and whether the songs appear as speech or singing in the movie. The table below presents the songs of Frozen classified according to Franzon s song modes: Table 2. Categorization of the songs in Frozen according to Franzon s (2005) five modes of musical songs. Narrative soliloquy Let it Go Narrative unison Frozen Heart, For the First Time in Forever Addressed monologue Do You Want to Build a Snowman?, In Summer, Reindeer(s) are Better than People Addressed dialogue Love is an Open Door, For the First Time in Forever (Reprise), Fixer-Upper Song as singing Reindeer(s) are Better than People, In Summer, Fixer-Upper 30

35 As can be seen in Table 2, Let it Go is the only narrative soliloquy as it is sung by Elsa and it depicts her thoughts and plans. Moreover, it is not addressed to any other character in the movie. Frozen Heart and For the First Time in Forever are both categorized as narrative unison, because they both have more than one singer and they are not addressed to any other character. Although For the First Time in Forever does not precisely sit into Franzon s (2008: 272) definition of unison singing as containing shared convictions (Elsa and Anna present their own varying viewpoints in the song), it cannot be categorized as any other mode either. The songs falling into the categories of addressed monologue or dialogue are quite straightforward because they are clearly addressed to other characters and sung by one or more characters. In addition to their places in other categories, the songs Reindeer(s) are Better than People, In Summer and Fixer-Upper also appear as singing in the movie. It is interesting to note that these songs which appear as singing, have least value in conveying the plot and are the most humorous ones. Every one of the songs certainly present their own challenges for a translator. For example, five of the songs have more than one singer and they feature different lines sung simultaneously and the lyrics are also quite fast-paced. This is especially challenging to render in the subtitles. The songs do not include many culture-bound items if any since the story is set in a fairy tale world with little cultural items or phenomena relevant to the plot. However, some wordplay is present in the lyrics, especially in the songs In Summer and Fixer-Upper. Fixer-Upper especially contains a lot of wordplay because the whole song is built around the idiomatic meanings of the verb to fix. This line, for example, can be tricky to translate: The way to fix up this fixer-upper is to fix him up with you. Another example of wordplay is the line Winter s a good time to stay in and cuddle/but put me in summer and I ll be a from the song In Summer. The song is sung by a live snowman, Olaf, who wonders how his life will change when summer comes. The line stops mid-way and the viewer can deduce the missing word with the help of rhyming and the visual channel when a small puddle is presented on screen. The two pictures below illustrate how the wordplay is constructed. In these kind of situations, the relationship between the lyrics and what is seen on screen is crucially important to notice when translating. 31

36 Picture 1. Disney 2013 Picture 2. Disney, 2013 The relatively small number of wordplay, idioms and other such features can be due to the fact that the primary target group are young children. However, some colloquial language does occur in the lyrics. Although there are not any particularly challenging lexical features in the songs, excluding the wordplay, all of the restrictions of AVT and song translation contribute to the fact that translating this movie s songs has not been an easy task. 32