TRANSLATING CHILDREN’S LITERATURE: SOME INSIGHTS FROM CORPUS STYLISTICS

In this article I explore the potential of a corpus stylistic approach to the study of literary translation. he study focuses on translation of children’s literature with its speciic constrains, and illustrates with two corpus linguistic techniques: keyword and cluster analysis—speciic cases of repetition. So in a broader sense the article discusses the phenomenon of repetition in diferent literary (stylistic) traditions. hese are illustrated by examples from two children’s classics aimed at two diferent age groups: the Harry Potter and the Winnie the Pooh books—and their translations into Czech. Various shits in translation, especially in the translation of children’s literature, are oten explained by the operation of so-called “translation universals”. hough “repetition” as such does not belong to the commonly discussed set of translation universals, the stylistic norms opposing repetition seem to be a strong explanation for the translation shits identiied.


Introduction
"Classic" children's literature is oten exceptional in its intertextuality, since it may transcend national literary traditions in ways adult literature more rarely does.In this sense, translation, as a form of adaptation and intertextuality, plays an extremely important role.he importance of translation, however, varies in diferent national literary traditions.For small languages, such as Czech, translations constitute a substantial part of the canon of children's literature.Arguably, this canon is to a certain degree international and to some extent common to many literary traditions (e.g.the Swedish Pippi Longstocking, the French he Little Prince, and the English Alice in Wonderland or the Harry Potter series have been appropriated by children all over the world)."Smaller" languages in general seem to be more accepting of foreign books, and the proportion of children's classics that actually are translations, albeit entirely domesticated, may be fairly high if not in the majority1.he situation for English is entirely diferent; as Lathey (2011: 203) says, it is estimated that children's book production in the UK yearly involves only about 2% of translations.
Translating for children involves a number of speciics; especially developmental stages need to be taken into account -"[t]ranslators should be aware of the stylistic features and modes of address appropriate for diferent age groups" (Lathey 2011: 199).And as Lathey further says: "Fortunately, the speciic demands of translating for children are now the subject of discussion and debate in both professional and academic circles" (ibid).he following work is situated in the ield of corpus stylistics and aims to show how the corpus linguistics methods used in the analysis of literary texts can be also beneicial in the analysis of literary translation and even support the translation process itself.A translator performing a stylistic analysis of the text to be translated should not only look at "how the text means what it does" but ideally also proceed to a stage "which seeks to explain, not only how the text means what it does, but also why a writer may have chosen to shape the text in a particular way" (Malmkjaer 2003: 38).
I will illustrate this with two case studies of two frequently used methods in corpus linguistics, keyword and cluster analysis, and these will be examined as speciic cases of repetition.In the irst study I revisit my earlier work on keywords in Harry Potter; the second study of clusters will be illustrated by examples from Milne's Pooh books. he texts were chosen on the basis of their target audience: while with Harry Potter older children are the assumed readership, the Pooh books are traditionally aimed at smaller children and oten read aloud to them by their parents.I will examine and discuss repetition in these texts and their translations in relation to difering stylistic traditions and translation universals.

Translating Children's Literature
he label "Children's Literature" may seem self-evident but in fact it is not straightforward to deine. he position of children's literature within the ield of literary criticism is not entirely clear and it is far too oten regarded as somewhat inferior to "general" literature.As Nikolajeva (2005: xiii)

says (my emphasis):
Some common prejudices about children's literature, maintained not only by its adversaries but even by its most ardent supporters, include the claims that children's literature is simply action oriented rather than character oriented, optimistic and with happy endings, didactic, and repetitive.
It is especially its didactic nature that is one of its distinguishing features.And as Nikolajeva (1996: 3) further says: "he very emergence of children's literature on a large scale is due to the fact that sometime in the seventeenth century society became conscious of childhood as a special period of life" and children's literature has become to be regarded as an important educational resource.However, the notion of childhood has been substantially changing with time and place (Hunt 2005: 3;Grenby & Immel 2009: xiii) and no doubt that the role of literature in children's lives keeps changing as well.
he "simplicity" of children's literature is perceived both at the narrative and discoursal levels (Nikolajeva 2005).Characters are expected to be uncomplicated, the storyline straightforward, preferably linear, the setting familiar-the same goes for the language: the language of children's literature is assumed to be "a 'scaleddown' version of 'language in general' , simpliied to be made accessible to these young readers" (hompson & Sealey 2007: 2).In fact, the language of children's literature has been researched only marginally with a few notable (book length) exceptions (Stephens 1992;Knowles & Malmkjaer 1996;Sunderland 2011)2 and recent corpus based studies by hompson and Sealey (2007) and Wild, Kilgarif and Tugwell (2013)3.his perception of "simplicity" and the didactic character of children's literature arises from the asymmetrical nature of the communication between the reader and the writer: he speciic feature of communication between a children's writer and a young reader is that it is asymmetrical.A young reader's cognitive capacity, life experience, and linguistic skills are normally diferent from those of an adult writer….Some scholars claim therefore that children's literature is always adapted to the needs of its audience… involving subject matter as well as form.(Nikolajeva 2005: xv) Translation is also a form of an adaptation-through time and space-and it is therefore not surprising that another speciic area of research in children's literature is its translation (e.g.Puurtinen 1995Puurtinen , 1998;;Oittinen 2000;Tabbert 2002;Lathey 2006Lathey , 2011)).In fact, children's literature has become "a playield" for some translation scholars aiming at identifying some of the translation universals as it has been assumed that these may manifest themselves in the translation of these texts more clearly than in the translations of adult literature.It is especially the demand for accessibility (i.e.adaptation to the needs of the target reader) that are assumed to cause the emergence of translation universals, namely-simpliication, explicitation, and normalization (for discussion see e.g.Laviosa 1998, Malmkjaer 2011, Lapshinova-Koltunski 2015, and Ippolito 2013).Baker (1993) deines "simpliication" as "the tendency to simplify the language used in translation" (Baker 1996: 181-182), "explicitation" as the tendency to "spell things out rather than leave them implicit" (Baker 1996: 180) and "normalization" (also called "standardization", "conservatism", "conventionalization"), as "the tendency to conform to patterns and practices which are typical of the target language, even to the point of exaggerating them" (Baker 1996: 176-7).
Although "repetition" as such is not mentioned among the "universals", scholars such as Toury (1977Toury ( , 1995) ) and Ben-Ari (1998) see the common translation strategy of avoiding repetition as a "universal" feature of translated language.While the terms simpliication, explicitation, and normalization are not generally understood as evaluative, repetition, especially in translation, is judged undesirable.Avoiding repetition is a common translation strategy operating across many languages.Ben-Ari (1998: 1) comments that in the case of repetition "considerations of adequacy come second to considerations of acceptability in the target language", repetition in the translated text is to be avoided even if the source text uses repetition.He then showcases various translators' techniques of avoiding repetition.Even though avoiding repetition in translation seems to be a nearly universal translation strategy, there are still considerable diferences in tolerating repetition among the various rhetorical and stylistic traditions (Abdulla 2001) and translators' dealing with repetition is thus a part of the "normalization" process.
A speciic case of repetition, which illustrates the diiculty, concerns reporting verbs. he reporting verb said is the most frequently used reporting verb in English; however, it is perceived as stereotypical by the translation scholar Levý (2011: 113): Most professional translators are aware these days that the stereotypical repetition of said in English introducing direct speech quite simply belongs to a diferent literary convention, and as a rule they vary the way they represent this reporting verb in translation.Corness (2009) investigates translation of said into Czech in a parallel corpus of 22 English source novels translated into Czech, i.e. nearly 10,000 occurrences of reporting said and inds that it was translated into Czech by no less than 1,323 diferent translation equivalents (and only 136 occurred 10 times or more).Farová (2016) carries out a similar study in a parallel corpus of English iction texts and their translations into Czech and Finnish.While in Finnish the translators keep the nearest equivalent (sanoi) in the majority of cases, there is a clear tendency in Czech for variation (and avoiding repetition, cf. Bečka 1992: 46).Nádvorníková (2017) examines translations of reporting verbs between Czech, French and English using parallel translation corpora and again conirms on larger data the tendency of Czech translators to avoid repetition and especially the repetition of the reporting verb said.However, reporting verbs may, as Mahlberg and Čermáková (forthcoming) show in their analysis of Alice in Wonderland and its Czech translation, substantially contribute to the characterization process and variation of the translation equivalents just for the sake of stylistic variation, i.e. avoiding repetition, may not always be the most suitable translation strategy.
Some researchers argue that readability and naturalness are key expectations in the translations of children's literature, which may provide translators more space for creativity (Oittinen 2000).At the same time, the translations are expected to be ideologically/educationally in line with the target literary setting and other speciic features are also relevant; e.g.Alvstad (2010) discusses ive issues that need to be considered when translating children's literature.hese are: 1. cultural context adaptation, which means modiication of the original aimed at adjusting a text for the intended reader's frame of reference (Klingberg 1986), which may include adjusting historical and cultural backgrounds such as measurements or names; 2. ideological manipulation, which Klingberg (1986) calls "puriication" and which is basically a form of censorship that may include stylistic changes or even more drastic content adjustments, 3. dual readership (children and adults), 4. features of orality, concerns texts that are intended to be read aloud, which may force the translator to chose between the content and sound, and 5. relationship between text and image as the "coexistence of a verbal and visual code is common in children's literature" (Alvstad 2010:24).Illustrations should support the text's content and in cases of content shits in the course of the translation process the original illustration may no longer be adequate for the translation.
To this list of issues I would add another one and that is sensitive adaptation of the translation to the stylistic norms of the target language.While it is clear that children's literature, including the translated part, as one of the earliest literacy resources, should more or less conform to the rhetoric tradition with which the children are gradually becoming familiar, this also poses problems if the two rhetoric traditions, i.e. the child's mother tongue and the source text language, are divergent.I will illustrate this (Section 4) with speciic cases of repetition and their translation into Czech, as repetition is arguably one of the features of children's literature (see Nikolajeva's quote above).

Corpus Stylistics in Translation
he term "corpus stylistics" is most oten used as a cover term for the application of corpus linguistics methods to the analysis of literary texts (Mahlberg 2015)4.Corpus stylistic studies may look at a work of a particular author (e.g.Mahlberg 2013) or one particular text (e.g.Mahlberg & McIntyre 2011).Essentially, corpus stylistics, being derived from corpus linguistics, relies on comparisons-the text under study is being compared to another text and oten to a large collection of texts; these are called reference corpora, e.g. one text by a particular author may be compared to his/her other texts, or to a corpus of texts of the same period etc.One of the strengths of corpus linguistics is identiication of repeated textual patterns and other repetitive textual features.
Repetition-repeated words-are oten seen as "the most uncreative of strategies" (Toolan 2012, p. 17).However, as himself Toolan (2012) argues, "repetition is central to literariness and literary creativity" but, clearly "not just any repetition will do: some kinds of repetition work brilliantly, others we generally agree are disastrous".And how to distinguish them?here are various kinds of repetition and repetitions fulil various functions in the text; they work diferently in poetry (e.g.Toolan 2012) and in iction texts.A commonly used method both in corpus linguistics and corpus stylistics is a keyword comparison.Keywords are words that occur statistically signiicantly more oten; i.e. they are repeated more oten than expected in the text under study in comparison with our reference data.Keywords oten show the "aboutness" of the text (Scott 2010) and applied to the study of iction they may help to identify themes and narrative progression (Toolan 2009), or they may help diferentiate among various characters, as in Culpeper's (2009) study of characters' speech in Shakespeare's Romeo and Juliet.In the translation, by identifying the keywords, the translator may use them to support the translation process itself (Čermáková 2015).Keywords as a speciic form of repetition may create lexical networks that contribute to textual lexical cohesion, and in order to render this kind of cohesion in the translation appropriately, these networks need to be identiied (Mastropierro & Mahlberg 2017).
Another speciic form of repetition are clusters, i.e. sequences of repeated words.5 Mahlberg (2013) in her cluster analysis in the work of Dickens identiies several types of clusters and their functions.One type of cluster she calls "labels": these contain a name or are a part of an expression that is used in a way similar to a name (Mahlberg 2013: 152) and "[i]n addition, Labels relate to the more striking repeated phrases associated with characters that are typically discussed in literary criticism" (Mahlberg 2013: 152).Labels have highlighting and contextualising functions and in Dickens they are oten associated with body language, e.g.Monsieur Rigaud in Little Dorrit who is brought to readers' memory with the help of a cluster and his nose came down.In this sense, the verbatim repetition of ive or even more words has an important role and as such we would expect this nature of repetition to be rendered in translation appropriately.

Repetition in Translation: Harry Potter and Pooh
I will now present for illustration two very brief case studies of repetition and its treatment in translation6.In Section 4.1 I revisit an earlier study of keywords in Harry Potter and the Philosopher's Stone (Čermáková & Fárová 2010), and in Section 4.2 I briely look at clusters in Milne's Pooh books.

Identifying Repetition: Keywords in Harry Potter
An earlier study (Čermáková & Fárová 2010) examined the Czech and Finnish translation equivalents of some of the English keywords in Harry Potter and the Philosopher's stone (1997) in the translations by Pavel Medek and Jaana Kapari, respectively.he hypothesis of this research was that keywords are suiciently important words in the original that their translation should show particular consistency.he list of the most frequent keywords itself did not contain any big surprises: it contained mostly proper names and words relating to the novel's theme, e.g.wand, cloak, owl, wizard, broomstick, troll.And as expected, the translator treated most of them with consistent translation equivalents.Some, however, seemed to be problematic, e.g.troll.Troll is not a part of the "domesticated" (i.e.Czech) ictional and fairy-tale world, and the translator was clearly struggling for an equivalent, which ended up in a somewhat distorted cohesion of the story of trolls invading Hogwarts.So, in this case, the inconsistency of the translation equivalents was less caused by the repetition in the source text, and more by the translator's strategy for adapting the ictional world at least partly to the target audience.I will now examine another keyword-wizard-in more detail.
he word wizard appears in the novel 65 times, 14 times as an adjective or in possessive use (e.g.wizard coins, bank, gold, sport, chess), 51 times as a noun.When we examine the concordance lines and sort them to the let, we can notice a number of positively evaluating adjectives (best, great, greatest, famous) in contrast to a single negatively evaluating one, dark: If we further examine the list of collocates (span -5/5), we get words like duel, witches, witch, defeat, greatest, great, dark, wand, bad.his already gives us some idea about the semantic prosody7 of the word in this text: wizards may represent the "good" or "evil" side and there is lots of competition going on between these two sides.his is important for the translation as there is no straightforward translation equivalent in Czech.
Mad?" said Percy airily."He's a genius!Best wizard in the world!But he is a bit mad, yes.My dad says it must've been a powerful Dark wizard to get round Gringotts, but they don't think Harry looked at the Famous Wizard card."Dumbledore again, " he said.He wasn't in Great Wizards of the Twentieth Century, or Notable Magical widely believed to be the work of dark wizards or witches unknown.Gringotts' goblins today particularly famous for his defeat of the dark wizard Grindelwald in 1945, for the discovery of the my master' s instructions -he is a great wizard and I am weak -" " You mean he was there in "Hermione!""Harry -you' re a great wizard, you know." "I'm not as good as you, "Professor Dumbledore is a very great wizard, Potter, he has many demands on his time -" Considered by many the greatest wizard of modern times, Professor Dumbledore is it' s logic -a puzzle.A lot of the greatest wizards haven't got an ounce of logic, they' d be stuck When we examine the translation solutions in Czech, we ind three major equivalents, the most frequent kouzelník (29 times -44.6%), the next čaroděj (9 times -13.8%) and černokněžník (5 times -7.7%).hese equivalents have diferent semantic prosodies in Czech, which I examined in a large corpus of children's literature (12.6 mill.words).8 he most frequent in the latter was čaroděj with a frequency of 54.42 ipm (= instances per million), and occurring in both positive and negative contexts.Kouzelník was less frequent (35.46 ipm), oten associated with magicians and illusionists as performers. he least frequent, and clearly negative, was černokněžník (2.7 ipm).
What is interesting in the Czech translation of Harry Potter and the philosopher's stone is the development of translations of the word wizard throughout the translation process.In the irst ive occurrences of wizard the translator opts for čaroděj.With the sixth occurrence the equivalent changes for čarodějník further modiied by zlej ('bad'), which is not present in the source text, see example 1: (1) See, there was this wizard who went ... bad.As bad as you could go.Worse.Worse than worse.His name was ... " Hagrid gulped, but no words came out.
"Could you write it down?"Harry suggested.
"Nah -can't spell it.All right -Voldemort." Hagrid shuddered."Don' make me say it again.Anyway, this -this wizard, about twenty years ago now, started lookin' fer followers.
Here the translator clearly feels that he needs to make a distinction between the good and bad wizards; so he chooses a diferent translation equivalent (though having the same word stem čaro-), and by modifying it with the adjective makes the evaluation explicit (this equivalent is used in the whole book series just once). he next three occurrences of wizard are translated by a newly introduced equivalent kouzelník; then we have four more equivalents čaroděj.Ater that, the translator adopts a systematic policy of using kouzelník for the good wizards and another newly introduced equivalent černokněžník for the bad wizards, see example 2.
(2) Considered by many the greatest wizard of modern times, Professor Dumbledore is particularly famous for his defeat of the dark wizard Grindelwald in 1945, for the discovery of the twelve uses of dragon's blood and his work on alchemy with his partner, Nicolas Flamel.
For the adjectival/possessive use of wizard, the translator consistently uses the adjective kouzelnický (adjective derived from kouzelník, which suggests this is the unmarked term, the "norm") (ex.3): (3) Nah, irst stop fer us is Gringotts.Wizards' bank.
here is one more interesting observation concerning the translation equivalents of wizard.he words oten occur with their female counterparts witch(es).here are two main translation equivalents in Czech, both derived from the same stem (čaro-) as the translation equivalent čaroděj (which-as noted above-has been dropped by the translator ater the irst 13 occurrences of wizard in the text): čarodějka (usually a "good" witch) and čarodějnice (usually a "bad" witch).In these cases the translator opts for the pair kouzelník and čarodějka for the good ones and the pair černokněžník and čarodějnice for the bad ones.In both cases he clearly prefers to have a pair of words not based on the same root, though alliteration (still weakly present in the second pair) might have been an interesting stylistic efect (sounding perhaps too repetitive?) and chooses the consistency he adopted earlier, see examples 4 and 5: (4) Each house has its own noble history and each has produced outstanding witches and wizards.
(5) Investigations continue into the break-in at Gringotts on 31 July, widely believed to be the work of dark wizards or witches unknown.
Even though a word like wizard in a book like Harry Potter and the philosopher's stone is a predictable keyword, the development of the translation equivalents above clearly shows that the translator aims at a consistent approach, which is adopted ater the irst 13 occurrences of the source word in the text.In this case the consistency amounts to explicitation, as the translator decides to draw a clear line between the good and bad wizards.Keeping the irst intuitive equivalent čaroděj, would have allowed him to maintain the ambivalent meaning (in terms of the semantic prosodies) of the English wizard.However, when we look at the equivalents of wizard in the remaining books in the Harry Potter series, it seems that the frequent repetition of this word is uncomfortable for the translator, and he drops the translation strategy adopted in the irst book: all three equivalents are usedkouzelník most frequently (605 times), then čaroděj (234 times) and černokněžník only rarely (45 times).his suggests the translator aimed to avoid repetition and "normalization" taking over.Considering the intended audience of these books are older children, this is in line with the Czech stylistic tradition, which prefers variation.However, I would argue, if the intended audience were smaller children, consistency (one way or other) would have been a better strategy.

Identifying Repetition: Clusters in Winnie the Pooh and he
House at Pooh Corner he following probe into cluster translations loosely follows up on my analysis in Čermáková (2015), in which I examined the translation of John Irving's Widow for One Year.Irving is an author who heavily relies on repetition (oten of long word sequences-eight or even more words) and, as the analysis has shown, this is stylistically unacceptable (at least from the translators' point of view) both in Finnish and Czech.I aimed to show that performing a basic corpus stylistic analysis may be a useful and powerful procedure for the translators before they embark on the translation to make their translation decisions more objectively informed (by quantitative analysis).he individual translation solutions and decisions are then obviously a complex procedure.Here we will look at two classic stories for younger children, by A. A. Milne: Winnie the Pooh and he House at Pooh Corne.Both have been translated into Czech by Hana Skoumalová.
I have generated four-word clusters from these two texts and picked two for further brief analysis: A Very Small Animal relating to Piglet (occurring 8 times) and the oten quoted Bear of Very Little Brain (occurring 7 times).hey operating here, consciously or subconsciously on the part of the translator, with some negative consequences.

Conclusion
In this article I have aimed to show how corpus stylistics can be used in the analysis of literary translation and further how it can support a translator's stylistic analysis by enlarging the toolkit with corpus linguistics techniques.By identifying keywords and clusters as starting points, we are able to uncover lexical networks that contribute to the construction of literary meaning and cohesion in the text that is of speciic importance for the young reader.We have seen that repetition seems to be a source of discomfort for many translators.Lexical repetition is most oten compensated for by the use of synonymy (in its broadest sense).he resulting semantic networks convey a similar meaning but the original lexical networks are altered or even lost.Coming back to the quote from Malmkjaer (2003) in the introduction, I would argue that in some cases, due to persistent stylistic norms, the translators remain at stage one -"how the text means what it does" and do not proceed to the next stage "which seeks to explain … also why a writer may have chosen to shape the text in a particular way" (Malmkjaer 2003, my emphasis).In the case of children's literature this may be mainly due to constraints imposed by the acceptability of the inal text. he various shits in translations I demonstrated here (and that have been demonstrated elsewhere) may be explained as a natural part of the translation process in which translation universals operate.But, in fact, what many of these shits seem to show is the translators' strategy of simply avoiding repetition and conforming to the target stylistic norms.
As is the case in all translation, "[t]ranslators are the artisans of compromise" (Lefevere 1992: 6); they are in a position where they have to interpret and mediate between two texts (Lefevere 1992: 10-11), and in translating children's literature they need to take into account their immature reader for whom higher levels of explicitation, normalization and simpliication may seem perfectly suitable translation strategy.10However, with stylistically carefully crated source texts, the "why" is as important as the "how" and "what".
because "stylistics (of the corpus and non-corpus varieties) is not solely concerned with the analysis of literature" and proposes the following deinition of corpus stylistics: "the application of theories, models and frameworks from stylistics in corpus analysis".5. he term 'cluster' is oten interchangeably used with the more general term 'n-gram' , where the n speciies the number of consecutive words that are repeated.Another widely used term in corpus linguistics for repeated sequences of words is 'lexical bundle' (Biber et al. 1999).
6. he parallel corpus InterCorp available at www.korpus,cz was used throughout the analysis.
8. Texts with the label JUN in SYN_v4 corpus, available at www.korpus.cz9. here is a diference between přihlouplý and hloupoučký.hey both mean "a bit silly", but while the irst adjective achieves a toning down with the preix při-but remains still fairly negative, the latter case is a diminutive form, in which the negative evaluation is somewhat weakened.
10.In addition, we should remind ourselves that the inal translation product, the printed book we ind on the bookshelves, has undergone also an editing process in the publishing house, where editing may take place without consultation of the original text.