<abstract xmlns="http://www.w3.org/1999/xhtml">

<p>The traditional English listening test system does not focus on analyzing students’ strengths and weaknesses, and designing a test bank with targeted exercises will be beneficial for students to grasp their actual situation. In this paper, after combining the twin network structure with the pre-trained language model, the Fusion-LM language matching model is constructed, which is utilized to calculate the matching similarity of students’ language features in the test. The adaptive grouping module and automatic scoring module are designed with constraints in mind and the adaptive test question bank for college English listening comprehension is established together. The validity test of the test question bank revealed that the test results of each module of the system meet the requirements, and the errors generated by automatic scoring are basically less than 2 points. The results of the teaching control experiment on whether to apply the test bank show that there is a significant difference between the post-test scores of the experimental and control classes of college English listening utterance comprehension (P=0.025&lt;0.05) and the listening and pronunciation recognition abilities of the experimental class students have also improved significantly after the teaching. This paper lays a necessary foundation for the effective development of university English listening teaching activities and provides an effective method for improving students’ listening comprehension abilities.</p>
</abstract>

The traditional English listening test system does not focus on analyzing students’ strengths and weaknesses, and designing a test bank with targeted exercises will be beneficial for students to grasp their actual situation. In this paper, after combining the twin network structure with the pre-trained language model, the Fusion-LM language matching model is constructed, which is utilized to calculate the matching similarity of students’ language features in the test. The adaptive grouping module and automatic scoring module are designed with constraints in mind and the adaptive test question bank for college English listening comprehension is established together. The validity test of the test question bank revealed that the test results of each module of the system meet the requirements, and the errors generated by automatic scoring are basically less than 2 points. The results of the teaching control experiment on whether to apply the test bank show that there is a significant difference between the post-test scores of the experimental and control classes of college English listening utterance comprehension (P=0.025<0.05) and the listening and pronunciation recognition abilities of the experimental class students have also improved significantly after the teaching. This paper lays a necessary foundation for the effective development of university English listening teaching activities and provides an effective method for improving students’ listening comprehension abilities.

Construction and Application of Adaptive Test Bases for College English Listening Comprehension Based on Natural Language Processing

Hope College Southwest Jiaotong University

Applied Mathematics and Nonlinear Sciences

This work is licensed under the Creative Commons Attribution 4.0 International License.

The traditional English listening test system does not focus on analyzing students’ strengths and weaknesses, and designing a test bank with targeted exercises...

Construction and Application of Adaptive Test Bases for College English Listening Comprehension Based on Natural Language Processing

Published Online: Jul 05, 2024

Page range: -

Received: Feb 22, 2024

Accepted: May 21, 2024

DOI: https://doi.org/10.2478/amns-2024-1697

Keywords
Fusion-LM model, Twin network structure, Matching similarity, Constraints, Adaptive question bank

© 2024 Jing Fang et al., published by Sciendo

This work is licensed under the Creative Commons Attribution 4.0 International License.

Construction and Application of Adaptive Test Bases for College English Listening Comprehension Based on Natural Language Processing

Published Online: Jul 05, 2024

Page range: -

Received: Feb 22, 2024

Accepted: May 21, 2024

DOI: https://doi.org/10.2478/amns-2024-1697

KeywordsFusion-LM model, Twin network structure, Matching similarity, Constraints, Adaptive question bank

© 2024 Jing Fang et al., published by Sciendo

This work is licensed under the Creative Commons Attribution 4.0 International License.

Keywords
Fusion-LM model, Twin network structure, Matching similarity, Constraints, Adaptive question bank