4.2.1 Rationale of Corpus
Toury emphasizes the significance of empirical research by suggesting that:
the introduction of empirical methods proved to be a true landmark in the evolution of the discipline...and the greatest contribution of experimentation lies precisely in its potential for shedding new light on the interdependencies of all factors which may act as constraints on translation and on the effects of these interdependencies on the process,its products,and the functions which any of them may serve in the recipient culture,and in increasing their predicative capacity.(1995: 221-222)
Ma and Miao explain that an empirical model of translation studies should be improved through the observation,description and analysis of phenomena in translation,which not only leads to the empirical accumulation of knowledge,but also makes translation studies descriptive and multidisciplinary (2009: 336).Within current translation studies,an increasing number of scholars have shown interest in carrying out research into such issues as “translation universals” using corpus translation.From 1993 onwards,corpus linguistics has developed rapidly thanks to scholars such as Baker,Laviosa,Shlesinger and Zanettin et al.all of whom began to use corpora as a tool to investigate translation at both a descriptive and practical level (Baker 1993,Laviosa 1998,Shlesinger 1998,Zanettin et al.2007).
In current manifestations of translation research,there are four types of electronic corpora that are specifically useful: namely,parallel corpora,comparable corpora,monolingual corpora,and multilingual corpora.
According to Baker,corpora applied in translation studies refer to “collections of texts held in machine-readable form and capable of being analysed automatically or semi-automatically in a variety of ways” (1995: 225).
Corpus-based translation studies is a newly developed field based on the theories of corpus linguistics,linguistics and translation studies,and mainly focuses on investigating authentic bilingual language data by using computer applications to analyse and calculate probability and statistics.According to Wang,“the development of corpusbased study and bilingual corpora in particular,may help Descriptive Translation Studies develop in a deeper way” (2006: 6).
Baker is considered to have been the first scholar to introduce the corpus as a research method in translation studies.She applied the research method of corpus linguistics study to the collection of a great deal of authentic translation materials,and in turn she described the nature of translation and even established many regulations and norms of translation.Baker pointed out that,with the development of corpus-based translation studies,scholars may use the comparison of the original language text and the translated language text to explain the communication role of the translated text in the nature of translating (1993: 243).Baker’s idea led scholars to certify,modify and clarify the nature of translation in order to better describe the nature of translation.(https://www.daowen.com)
Baker divided the corpora of translation studies into three main types: parallel corpora,multilingual corpora and comparable corpora according to different study purposes (1995: 230).Parallel corpora highlight the differences between the original text and the translated text at the level of lexis,sentences or paragraphs.The second type of corpora is multilingual corpus.It is normally a collection of two or more versions of text; they are all original texts as opposed to translated ones.This type of data collection can be regarded as a kind of pure “parallel corpora” that offers information to scholars and enables them to tell the difference in writing styles between each text,and helps them to set up criteria to evaluate the quality of translation.Last but not least,comparable corpora refer to two independent collections of texts in the same language; for example,texts originally produced in a given language (English),and texts translated into the same language (English).Scholars can test their theoretical hypotheses through vast computer-stored collections of comparable texts,which can easily reveal the distinctive linguistic features of translated texts.They can also offer learners a better comprehension,not only of target but also of source texts,allowing them to compare language and cultures in terminology,phraseology and textual conventions.Comparable corpora may help scholars to indicate the universality of translation.It is clear that the application of corpora not only supplies translation scholars with empirical research tools to investigate the nature of translation and describe the features of translated texts,but it also provides practical translators with an invaluable tool in their practical work.
According to Ma and Miao,
parallel corpora consist of original texts and their translated texts.They can be used to provide information on specific language-pair translational behaviour,to positcertain equivalence relationship between lexical items or syntactical structures in the two languages concerned,or to explore the phenomenon of translationese.They can also offer translation learners the possibility to observe what strategies translators appear to privilege,and to learn how to overcome specific translation problems based on their observation.(2009: 356)
The acknowledged advantage of corpus is its “data-driven feature [that] helps researchers conduct their investigation from the bottom up,using concrete facts to make generalizations about languages”,which can be tested repeatedly in order to avoid subjectivity and randomness in the process of study (Baker,1996: 186).In particular,the use of corpus in translation studies,such as in relation to film subtitling,can enable researchers to find some peculiar semantic features,differences in the style of the texts or language customs,differences in the length between the original versions and the translated manifestations,co-occurrence,coherence,and even the distinctive features of using punctuation within the translation.However,the disadvantages are also obvious.On the one hand,the data collection process is labour and time intensive.Any researcher copiously needs to transfer the format of the selected texts in order to meet the requirements for collecting data in the corpus.On the other hand,most scholars tend to proceed with their research from the top down in order to find evidence in support of their abstract hypotheses (ibid: 186).Therefore,the use of corpus requires the researchers’ patience and intelligence.I think with full preparation works,these difficulties are resolvable.
The development of parallel corpus studies in European countries is at a mature stage.Meanwhile,in China,the study of corpus linguistics and corpus translation has developed very rapidly in the last 10 years.The most common types of corpus undergoing development are general corpus,specialized corpus,synchronic corpus,diachronic corpus,and learner corpus (Liang et al.2013: 4-5).Although some of these corpora can be found online (e.g.,http://mtgroup1ict1ac1cn/corpus),these corpora rarely meet the needs of language scholars because of the limited data that is available to the public.Moreover,instances of E-C parallel corpora are even rarer,particularly in relation to the subtitling of animation films.
In this research,I have opted to set up a self-built corpus in order to compare the original subtitles and the translated subtitles in my selection of 15 feature length animation films.The purpose behind this discussion of suitable translation strategies is to take a scientific approach with a view to identifying “subtitling universals”,particularly in relation to E-C subtitling.
In the introduction chapter of this book,I explained my research purpose and hypotheses.I plan to test and evaluate all of these hypotheses using the chosen methodologies.The self-built corpus will help to evaluate the effectiveness of the first hypothesis (about the length difference between the translated Chinese subtitles and their corresponding English original texts) and analyse the proportions of the use of traditional expressions and popular expressions in the data.