Knut Hofland(1); and Jarle Ebeling(2);
(1) P.O. Box 1003, Blindern, N-0315 OSLO
(2) Harald Hårfagres gt.
31, N-5007 Bergen, Norway
KEYWORD: alignment, browsing, parallel corpus
AFFILIATION: (1)The Norwegian Computing Centre for the Humanities,
(2)Department
of British and American Studies, University of Oslo.
E-MAIL: Knut.Hofland@hd.uib.no
Jarle.Ebeling@iba.uio.no
FAX NUMBER: +47 5558 9470(1)
+47 2285 6804(2)
PHONE NUMBER: +47 5558 9463(1)
+47 2285 6707(2)
Project Background
The English-Norwegian Parallel Corpus (ENPC)
project was started in the beginning of 1994 and it is expected to end in 1996.
The parallel corpus is meant to be a research tool for linguists and students
interested in contrastive linguistics. The corpus will contain English and
Norwegian originals and their translations (including both English-to-Norwegian
and Norwegian-to-English translations).
The corpus will only contain written material, but both fiction and non-fictional texts will be included. To make it possible to include as many different writers and translators as possible, text extracts of 10,000 - 15,000 words will be selected and not complete books. Each text extract will start at the beginning of the book, and if possible, end at a chapter boundary. The finished corpus will consist of 100 pairs of texts with a total of about 2.5 million words. The texts are marked up according to TEI P3. As regards the structure of the corpus and the projected uses of the material, see (Johansson and Hofland, 1994) and (Johansson, Ebeling, and Hofland 1996).
Alignment program
The alignment program has been written by Knut
Hofland. The program makes use of a simple bilingual lexicon (anchor words), but
in addition uses information like proper nouns, special characters and tags,
cognates and sentence length in characters. Statistics based on half the corpus
gives an error rate of approx 2 per cent. The program has also been used in
aligning texts from other language pairs like French-Norwegian,
English-French/German/Polish/Swedish/Finnish, Swedish-Estonian. The program
gives output in several formats, a TEI recommended format and one format also
suitable for use with ParaConc and WordCruncher for Windows.
Browsing tool
The browsing tool has been written by Jarle Ebeling.
The aligned and proofread texts are indexed, making it possible to search the
text database for words in one of the languages and retrieve the sentences
together with their translations in the other language. Words in the two
languages can be combined with the and or the not operator so
that only pairs of sentences with a specific word in the first language together
with (or not together with) another word in the second language are found. Words
can also be truncated and a distance (in number of words) between particular
words can be given.
References
Web page with more information and articles:
http://www.hd.uib.no/enpc.html
Johansson, S., and Hofland, K. (1994), `Towards an English-Norwegian parallel corpus', in U. Fries, G. Tottie, and P. Schneider (eds.), Creating and Using English language Corpora, (Amsterdam), 25-37.
Johansson, S., Ebeling, J., and Hofland, K. (1996) `Coding and Aligning the English-Norwegian Parallel Corpus', in K. Aijmer, B. Altenberg and M. Johansson (eds) Papers from Symposium on Text-based Cross-linguistic Studies, (Lund), 87-112.