An Attempt to Create a Machine-readable and Aligned Corpus Essay

Submitted By newyearbb00
Words: 4203
Pages: 17


Josef Schmied*
Translation and Cognitive Structures1
This project is based on a corpus of English and German source and target texts, ranging from contemporary literature to scientific textbooks. We try to create a machine-readable and aligned corpus which will allow us to discover and categorize translation equivalents for a number of linguistic items, such as prepositions, subordination, deictic elements, metaphors or culture-specific structures. On this basis we look for regularities in the configuration of factors that influence equivalent choices for each of the phenomena in question. Apart from theoretical insights into contrastive language structures as well as cognitive aspects of the translation process, the purpose of the project is to discover and categorize prototye and non-prototype equivalents in two closely related languages. Research results could, for instance, be applied to bilingual lexicography or other language learning and translation aids.

1. A new approach to contrastive and cognitive issues: an example German and English are genetically and typologically closely related languages. Thus language learners and bilingual language users (e.g.

The project group consists of a team at Chemnitz (led by Josef Schmied) and a team at Cambridge (led by Kirsten Malmkjær) and it works parallel to other translation workgroups (cf. Johansson/Hofland 1994). The basic research design was agreed on in
February/March 1993, when I worked as an honorary research fellow at the Research
Centre for English and Applied Linguistics in Cambridge. I have to thank all supporting agencies that have enabled me and my colleagues to pursue the project so far (the
German Research Association/DFG, the British Council and the Technical University of Chemnitz-Zwickau), all the colleagues who have assisted and encouraged us (Gillian
Brown, Cambridge; Michael Claridge, Chemnitz; Elke Esders, Brussels and Paul Procter, Cambridge) and all the publishing houses and agencies that have contributed texts
(sometimes even in machine-readable form).
Although the basic research framework is the same for both groups, the Chemnitz team is specialized in corpus collection, whereas the Cambridge team adds their translation experience. Analyses will be carried out parallel (in each case source- and targetlanguage specialists have to work together), but each team will be responsible for separate sections of the comparison.

* Josef Schmied
Englische Sprachwissenschaft
TU Cemnitz-Zwickau
Postfach 964
09009 Chemnitz /D)
Hermes, Journal of Linguistics no. 13 – 1994

170 translators) can in many cases start from the assumption that structures are parallel. Thus it appears fairly safe to expect a German in as an equivalent for an English in. This crude juxtaposition however does not consider external and internal boundaries of the underlying cognitive2 principles, i.e. the basic status of in in the respective prepositional system and the polysemous subcategories (including the respective quantitative weight). Therefore, most dictionaries would also give at least into as an English equivalent for directional German in (the choice is normally easy for the dictionary user as the form is usually disambiguated by case morphology in German; local in entails dative, directional in accusative inflection). However, this only applies for prepositional English in, adverbial in has various other equivalents in German
(herin, hinein, herein, ein, etc.). Thus it would be interesting to analyse all cases where German in is not rendered by English in on the most general level of analysis. As we have seen, this example is subcategory- and word-class sensitive, thus we can establish a lower, more detailed level of analysis where in, even local in, is not rendered by in in
English, but by at for instance - and exclude adverbs from the analysis.
For other prepositions this juxtaposition of structures can be easier