After Linguistics - based MT

The 80s can be characterized as the era of Linguistics-based MT (LBMT) and of its failure in the history of MT, in which (computational) linguists have initiated the first serious attempt at constructing scientific or computational theories of MT. Partly because of a large discrepancy between scientific interests and engineering practices, this work has little influence on the performance of commercial MT systems in the market. Because of this failure of commercial exploitation of LBMT, the pendulum has swung suddenly to the other extreme of the spectrum and researchers are now interested in developing, without any theory, practical systems which can readily be used in actual translation environments. Some have even abandoned the idea of machine translation altogether and switched to development of tools for translators, including those of retrieving translation examples from data bases.