Issues in Generating Turkish from Interlingua

fhakkaniturkogcsbilkentedutr fterukoehngcscmuedu This report presents the generation of Turkish text from the interlingua representations used in the KANT knowledgebased machine translation system and some issues that we have encountered during the development of this system The target language Turkish can be considered as a subjectobjectverb SOV language in which constituents can change order rather freely Morphologically Turkish is an agglutinative language with very productive inectional and derivational suxation processes In this system the interlingua representations are mapped to Turkish fstructures which are the inputs of the Turkish sentence generation system Then the surface form of the sentences are realized by the morphological generation system This study is the rst application of knowledgebased machine translation from English to Turkish and the results are very promising For an example set of sentences the system was able to translate o