Machine learning techniques for decoding GP sentences: effects of processing breakdown

Abstract

This paper discusses the decoding of garden path sentences by machine learning technologies. The parsing of garden path sentence comprises the pseudo- and genuine-analysis. The pseudo-decoding leads to processing breakdown and backtracking. The genuine-decoding is the result of integration of machine learning knowledge and linguistic cognition of human beings. Context free grammar, recursive transition network and CYK algorithm can be used to decode garden path sentences effectively and efficiently. Machine learning development needs the involvement of linguistic and computational information.