Session 4: Machine Translation

About 2~ years ago, ARPA initiated a new program in Machine Translation (MT). Three projects were funded: CANI)II)E, built by IBM in New York; I,INGSTA'I~, built by Dragon Systems in Boston; and PAN(;I,()SS, a col laborat ion of the Computing Research Laboratory at New Mexico State University, the Center for Machine Translation at Carnegie Mellon University, and the Information Sciences Institute of the University of Southern California. All but one of the papers in this section is directly related to these systems. In one way or another, each paper addresses one of the following two major dimensions of variation: basic ap. proach (i.e., operation and data collection by statistical vs. symbolic means) and depth of analysis (i.e., direct replacement, transfer, or interlingual). This overview first explains these terms and then describes the import of the papers.