|
Abstract : |
For more than forty years machine translation has been an active area of research and development. Several variants of MT systems have been proposed e.g., interactive systems, batch-mode systems requiring a large amount of human post-editing and systems that generate text in several languages from a common structured set of data rather than a source text. Although there are some successful systems and several products on the market, it is still the case that most translators and translation companies prefer not to use MT-systems. In fact, it has been estimated ?that the current share of anything that could be called MT, pure or mixed, is well below 1 % of the total translation market ? (Isabelle et al, 1993). This situation is likely to change, however. The disappointment over the performance of automatic systems has caused a change of direction in system development, putting the focus on systems that will support the translators rather than replace them. The ever increasing processing capacity of PCs and workstations will also enable translators and companies to purchase powerful tools as add-ons to their ordinary word processing programs. Furthermore, as we show below, already very simple tools can speed up the, |
