Text Corpus as an Abstract Data Structure

Linguistic corpora have become one of the most important sources of evidence for empirical scientists in a number of disciplines connected with the study of language, language behaviour and text. Consequently, the methods of compiling and analysing text corpora play a central role for data acquisition, observation, and the testing of hypotheses, as well as for inductive, heuristic attempts such as data or text mining. In linguistics, corpus linguistics has become established as an academic branch and many researchers consider themselves, in the first line, as corpus linguists. The number of publications in this field is increasing continuously, their topics varying from purely notational questions about the use of mark-up languages over corpus classifications, directives for the compilation of corpora, and individual empirical studies using corpora, to methodological and mathematical text books on quantitative text analysis. However, there is one very practical, technical aspect which has not yet been approached until know (to my best knowledge) although is bears far-reaching consequences for the work with corpora. The following considerations, which deal with this aspect, are based on theoretical work and practical experience in the field of software engineering in computational, quantitative, and corpus linguistics.