Research progress on system-generated language: From the psycholinguistic perspective

Abstract

As the development of computer technology and Internet use, system utterance producing gains more and more attention. Research on utterance producing systems has been approached from two angles. In one research tradition, the analysis of corpus led to templates of system utterance generation. In another tradition, a natural language generation (NLG) system corresponding to human language production theories was founded. The NLG system was marked as flexible and trainable. In this paper, we introduce the progress on system-generated language from a psycholinguistic perspective and take an example to explain the function of human language production theories for the development of NLG systems. As more and more progress was made by psycholinguists on language production, there will be abundant room for NLG systems to improve in the future.