Change of word types to word tokens ratio in the course of translation (based on Russian translations of K. Vonnegut novels)

The article provides lexical statistical analysis of K. Vonnegut's two novels and their Russian translations. It is found out that there happen some changes between the speed of word types and word tokens ratio change in the source and target texts. The author hypothesizes that these changes are typical for EnglishRussian translations, and moreover, they represent an example of Baker's translation feature of levelling out. Translation study has always experienced a kind of difficulty with defining its subject. The issue of methods is even more critical for this branch of linguistics. How can one estimate equivalence or adequacy of translation in an impartial and dependable way? Are there any laws of translation process which influence its result? Linguists and translatologists have been studying these problems for the last fifty years. In the nineties the field saw the first attempts to combine translation studies and corpus linguistics with its strict methods of analysing large text collections. Quantitative method allow to describe translation result in a detailed and reliable way and to come to some conclusions about the process itself. Review of research within this topic is presented, for example, in [Olohan, 2004]. It should not come as a surprise, that this inevitably lead translatologists to a disputable question: are there any objective statistical dependencies between source text and target text? This question is closely related to Baker's hypothesis about objective differences between any translated and non-translated texts [Baker, 1993]. If we solve this issue, we would be able to compare translations in an objective way; moreover, it would give additional clues to understand how human brain functions when we translate. Can we say that some statistical invariants of the text remain after translation or it is a kind of random choice? This is the topic of the present article. Parallel corpus for preliminary research of this problem is: Kurt Vonnegut's novels «Cat's Cradle» (1963) and «Slaughterhouse-Five Or The Children's Crusade» (1969) and their Russian translations by Rita Rait-Kovaleva «Колыбель для кошки» and «Бойня номер пять или Крестовый поход детей» respectively. We took the texts from Moshkov e-library (http://www.lib.ru) and from the site «Vonnegut Books in English» (http://bg-studio.newmail.ru). The corpus was analysed with AntConc software (http://www.antlab.sci.waseda.ac.jp/software.html). Symbol case was not taken into account and we did no lemmatization. In the source text of «Slaughter-House Five» there are 6247 word types and 50848 word tokens. Thus, the ratio is 0.12. Types used only once (hapax legomena) constitute 50% of word types. In the source text of «Cat's Cradle» there are 6372 word types and 54353 word tokens. Thus, the ratio is also 0.12. 52% of word types are hapax legomena. If we look at the frequency dictionary of both novels, we will see how the frequency is decreasing when the rank of a word type increases. This is the consequence of Zipf Law, which states that frequency of word types in any natural text is inversely proportional to their rank. It means that frequent word types are few and rare word types are many [Zipf, 1932]. The novels were split into equal fragments in order to find out whether type/token ratio coincides in different pieces of the texts. «Slaughter-House Five» was split into 10 fragments by chapters, and «Cat's Cradle» into 11 fragments 300 lines each. Tables 1 and 2 show type/token ratios (TTR) in all the fragments: Table 1: Lexical statistical features of source text of «Slaughter-House Five» Fragment Number of word types Number of word tokens Types to tokens ratio (TTR) * Types to tokens ratio 1 1411 5494 0,26 2 1805 7328 0,25 3 1376 4663 0,3 4 1099 3536 0,31 5 2389 11217 0,21 6 1207 4357 0,28 7 62