Second Language Acquisition from Aligned Corpora

The paper describes a system for automatic aligning and searching for translation equivalents in large bilingual corpora. This implementation was developed to facilitate our tasks in GLOSSER #343 Coper-nicus'94 Joint Research Project, where Linguistic Modeling Laboratory was charged especially with preparation of bilingual material. The Gale-Church algorithm is chosen as aligning procedure for parallel texts. The main functional characteristics of our system MARK ALISTeR (MARKing , ALIgning and Searching TRanslation equivalents) are described. The program package can be used under MS Windows as an autonomous procedure of second language acquisition and CALL instrument. Evaluation of the results and the alignment errors of the algorithm and the tool is presented for diierent types of texts.