Designing and compiling parallel aligned corpora: Pitfalls and (some) solutions on the example of a corpus of translated musem texts (English-Spanish)

Abstract

Text corpora are tools having both a long tradition in research and a variety of applications. Of all existing types, this paper focuses specifically on parallel, aligned corpora. By taking one of this corpora as a starting point-a parallel, aligned corpus from museum texts originally written in English and subsequently translated into Spanish-, the aim of this article is to propose a methodology that consists of four basic stages. By the revision of previous literature on the topic, and by using multiple software programs-proprietary and free, specifically created for corpus compilation and created for other purposes-, it is concluded that, although the compilation of corpora such as the one that was intended is a feasible task, the procedure is full of obstacles. Some obstacles were overcome, while some were not; that is the case, for example, of the repetitions on the aligned corpus, which are not present in the corpus.