Identification of noun-noun compounds in the context of speech-to-speech translation

Abstract

Translation of noun compounds has been a challenging task in machine translation for many years. Noun compounds are used very productively and constitute a large amount of the text vocabulary. We present a work on translation of noun compounds from English to German, focusing on identifying the nominal structures in English which represent compounds. This work is part of developing a speech-to-speech translation system, and the input to the machine translation component comes from automatic speech recognition. This brings in a further difficulty to the translation of noun compounds because the input might introduce recognition errors and it does not have typical features of a written text such as punctuation and capitalization. We present a method of noun compound identification based on syntactic analysis and lexical information and discuss the challenges that we encountered during the compound identification process.