Rolf A. Stachowitz beyond the Feasibility Study: Syntax and Semantics

During the second part of the last decade, considerable advances had taken place in hardware and software development and-most important-in linguistic theory, through the incorporation of a semantic component into the theory of grammar. In view of these developments the Linguistics Research Center submitted a proposal to Rome Air Development Center with the purpose of ascertaining in some depth the then current opinions of the linguistic and computational linguistic community on the feasibility of fully automatic high quality translation. In particular, we were interested in obtaining the views of Professor Bar-Hillel. Since the funds available for this study were limited , a large number of scholars from this country and abroad, whose opinion would surely have contributed to the effort, could not be invited ; we considered mainly those linguists who, due to their theoretical background, regarded mechanical translation, as, at least theoretically possible, i.e., proponents of the universal base hypothesis. The report on this study has been published. One of the conclusions is the statement that in spite of the progress that had been made in linguistic analysis, linguistic research had dealt primarily with syntactic analysis of individual sentences and less with semantic problems and discourse analysis. As a result, " current linguistic theory is inadequate for machine translation ". The recommendations indicate that for improved machine translation, research in the areas of descriptive linguistics, theoretical linguistics, and comparative linguistics among others was necessary and should be supported. Moreover, research in discourse analysis and production of coherent discourse should be encouraged and various grammatical models different from transforma-tional grammar should be investigated. There was general consensus among the various participants about these points. The main differences pertained to the extent to which pragmatic information, that is, knowledge of the world, had to be