We have tried in this chapter to give a brief overview of some of the
issues and techniques which are being actively researched today in MT.
Of course, there is not enough room in one chapter to do justice to
the field, and we have of necessity omitted much work that is of
interest. In particular, we have restricted our discussion to MT
itself and have said nothing at all about recent work in the
development of translators aid s, multilingual
authoring packages and
terminological systems of various sorts. Nonetheless we have
identified three important trends in current research in MT. The
first is the exploitation of current techniques from computational
linguistics to permit a multidimensional view of the translational
relation between two texts. The second in the increasing orientation
of the research community towards the use of existing resources of
various sorts, either to extract useful information or directly as
components in systems. The third, related, trend is towards
statistical or empirical models of translation. Though we have dwelt
in some detail in this short survey on `pure' statistical and simple
pattern matching methods, in fact much recent work advocates a mixture
of techniques, for example with statistical methods supplementing
rule-based methods in various ways.