Linguistic enrichment and exploitation of the

1 Introduction The aim of the present paper is the investigation of the nature of translated text. This kind of analysis is based on the assumption that translations differ from their source language texts and from comparable texts in the target language, in the sense that they have specific properties which cannot be found in non-translated text. The present research focuses on the comparison of English translations from the register of narrative writing to English originals from the same register. Within this context, the analysis of the relevant register features shows whether or not the translations conform to the norms of the given register, i.e. whether or not normalization can be found in the translation corpus (see section 2). The corpus under investigation is composed of the fiction part of the Translational English Corpus and the fiction part of the British National Corpus (see section 3). This monolingually comparable corpus comprises about 10 million words in total. To cope with the this large amount of words included in the comparable corpus, computational methods which support corpus enrichment and exploitation are employed. Along these lines, the automatic annotation of the corpus and its representational format are presented (see section 4). Furthermore, the querying techniques as well as the significance tests are discussed (see section 5). Hereby, the comparability of the two sub-corpora (i.e. translated vs. original text) plays an important role for the methodology chosen to investigate the specific properties of translations. The analysis results in a profile of the nature of translated text, which clearly shows to what extent the translations differ from originally produced texts. Furthermore, possible explanations are discussed on the basis of examples taken from the translation corpus. The paper concludes with a summary and an outlook on related research perspectives (see section 6).