Joseph Wright's English Dialect Dictionary (1898-1905) Computerised: architecture and retrieval routine

Comment on Chapter 0: By way of an introduction on this conference of relatively heterogeneous participants – philologists and computer scientists being the two main groups – I would like to emphasise the difficulty of finding a common language in the interdisciplinary dialogue. While English is, of course, the lingua franca amongst the participants, there is still a risk of nonand misunderstanding. As I see it, computer philologists tend to produce theoretical concepts of how problems can be solved with the help of software, but there is a limited competence in identifiying the problems. Philologists, on the other hand, with a good knowledge of the textual, historical and cultural embeddedness of their objects of research, often know the problems at issue, but may not be aware of the new computer-assisted tools and strategies waiting for them to solve their problems. Given this basic difference of competence, we should all try to „meet halfway“, i.e. to cooperate. The first step towards cooperation is to clarify one’s own position. I, therefore welcome the invitation of the organisers of the Dagstuhl conference to present a short selfintroduction. Most corpus-linguists either compile a corpus or use corpora compiled by others. My own academic past has caused me to do both, compiling and analysing. Details of my corpuscompiling activity can be seen from the survey given above. Wright’s English Dialect Dictionary is the last in a series of projects of corpus-compilation. As regards the application of corpora, it seems to me that both the historicity and complexity of corpus texts are often underestimated. The larger and the older the texts of a corpus are, the more are they likely to be based on variable and erstwhile norms which disallow computer accessibility of the texts in their original shape. Editorial practices (if editions are used for compilation) and the now unimaginable irregularities of spelling in at least medieval and Early Modern English texts are the two main arguments for the necessity of normalisation. If a word comes along in shapes that are incalculable at hoc, the results of query routines are bound to be misleading. For more detailed arguments along these lines see my eralier publications Markus 1997 and 2000.