Using Semantics in Translation I. Morphologically rich languages pose difficulties for corpus-based methods of machine translation, because of the large number of word forms that are likely to occur rarely or be entirely absent from available corpora. Using semantic equivalents is one way of helping deal with that problem. II. Despite the current prevalence of corpus-based methods, the old interlingua paradigm retains its appeal. With the advent of semantic resources, such as WordNet, VerbNet and FrameNet, general-purpose semantic parsing becomes feasible. We report on one such attempt.