Natural Language Generation in Student Writing: Language Features, Strategies and Links to Successful Writing

Abstract

AI Natural Language Generation (NLG) is a process where computer systems generate human-comprehensible language texts from information. This study explores students’ writing quality with NLG tools that students use to generate content for their writing. We analyzed the language features of four short stories composed by students with NLG tools in a Hong Kong mainstream secondary school. In the stories, we found different patterns in basic organization and structure and syntactic complexity for words from NLG tools. Human raters scored the stories and we found more sophisticated strategies for student interaction with NLG tools appear more effective for human ratings. The findings provide pedagogical strategies and technical specifications to use NLG tools for writing in classroom contexts.