Research on the Computational Effects of Natural Language Processing Software Coh-Metrix

• Published in: 2024 International Conference on Data Science and Network Security (ICDSNS) pp. 1 - 6
• Main Author: Liu, Liping
• Format: Conference Proceeding
• Language: English
• Published: IEEE 26.07.2024
• Subjects: Coh-Metrix; Complexity theory; Data science; Natural language processing; natural language processing software; Network security; Software; syntactic complexity; Syntactics; Systematics; Writing
• DOI: 10.1109/ICDSNS62112.2024.10690999

This paper takes iWrite automated essay scoring as an example to investigate the extent to which Coh-Metrix can reveal syntactic complexity of writing by intermediate English learners. This study uses the two-sentence complexity analyzer to analyze the overall complexity, dependent complexity and phrase complexity[1]. The results show Coh-Metrix can reveal the improvement the overall complexity, he complexity of coordinate phrases and the complexity of complex noun phrases, but it reveals no significant effect on the dependency complexity. This paper suggests that this may be related to the content of feedback provided by the system. In addition, Coh-Metrix analysis also shows that automatic essay feedback can reduce the complexity of phrasal verbs in second language writing, which may be related to the avoidance strategy adopted by the subjects.