How to use the Web to make better translations
Mirosława Podhajecka
University of Opole
It has been observed that students perceive the translation process as an activity in which bilingual dictionaries play a key role. No wonder the final product results in translationese, i.e. vocabulary typical for translated, not original texts (Gellerstam 1986). The problem becomes conspicuously clear in the case of specialized areas of English (business, medical, technical, etc.), with their focus on factual information. As philology students don’t usually have the expertise in the given field, turning to authentic English texts is the only safe way to make good translations.
Traditional printed materials are often unavailable, but the Web offers us easy access to virtually unlimited electronic resources. Though the structure of the Web is chaotic and uncontrollable, the collection of texts can nonetheless be used successfully as a language corpus (Kilgarriff and Grefenstette 2003). Some of the prerequisites for the computer-assisted translation process are: computer literacy, access to the Internet and a few search engines (concordancers). Below are some of the tips to avoid pitfalls on the Web:
Authenticity Web sites are often made in different language versions, therefore the information that has just been retrieved may not be native English. Search engines that allow for restrictions to a particular country (e.g. Google) will be helpful.
Varietal differences There is a range of varieties of English, and subtle differences are not traced easily. Intuition plus good reference books should guide us through a jungle of possible options so that the translation will be consistent in its lexical content.
Techniques It is worth comparing traditional search engines and the Web concordancers (e.g. WebCorp). In order to minimize the number of incidental errors, at least two different techniques should be used. Simple queries (by general keywords, e.g. forestry, to get background information), and advanced queries (by strings of words, e.g. sylvicultural purposes / forestation purposes) will give us diverse insights into the factual language of specialized texts.
Assessing results High frequencies are usually, but not always, indicators of authentic use. The more concordancing lines we get, the higher the probability that the phrase has wide currency in English, but for reliable results we need to take into account only one occurrence per site.
It has been found that although the methodology of computer-assisted translation is time-consuming, it is also stimulating. The students’ feedback has been very positive, which proves that, regardless of its limitations, the method has a great potential. Indeed, not only does it help to make better-quality translations, but it also enhances the students’ autonomy in the learning process.
References:
Gellerstam, M. (1986) "Translationese in Swedish Novels Translated from English". In: Wollin, Lars and Hans Lindquist, Translation Studies in Scandinavia, CWK Gleerup, Lund, pp. 88-95.
Grefenstette, G. and J. Nioche (2000). Estimation of English and non-English Language Use on the WWW. http://arxiv.org/ftp/cs/papers/0006/0006032.pdf
Kilgarriff, A. and G. Grefenstette (2003). Introduction to the Special Issue of Web as Corpus. http://www.mitpress.mit.edu/journals/pdf/coli_29_3_333_0.pdf
Volk, M. (2002). Using the Web as Corpus for Linguistic Research. In: Renate Pajusalu and Tiit Hennoste (ed.): Tähendusepüüdja. Catcher of the Meaning. Festschrift for Professor Haldour Õim. University of Tartu, Estonia: Publications of the Department of General Linguistics 3.