A Corpus-Based Sentiment Analysis of University Students' Attitudes Toward Online Learning: Evidence from a Pakistani Public University

Abstract

This study explores the attitudes of the university students to online learning in a public university in Pakistan through a mixed method approach, both an explicit survey based quantitative approach and an implicit corpus-based sentiment analysis approach. With the rapid spread and adoption of digital learning platforms, the higher education system in Pakistan has experienced a major transformation, and it is imperative to understand the real cognitive and emotional reactions of students to the changes in this system. A specialized Learner Corpus was created out of 23 student essays (4396 word tokens and 999 word types), and survey data from 36 university students from different disciplines (English, Computer Science, Education and Business Administration). This study uses corpus analysis tools (AntConc v4.2.1) and a Corpus-Assisted Discourse Studies (CADS) framework to systematically find out what lies beneath the surface with regard to the semantic prosodies, in the form of absolute frequency lists, collocational networks, and concordance lines. The quantitative survey results reveal a high level of agreement with respect to the theoretical advantages of digital education with high agreement scores in the areas of spatial flexibility and convenience. But the power of the corpus analysis clearly outweighs such findings, with positive sentiments having strong linguistic associations with learner autonomy, and negative sentiments being strongly tied to infrastructural decay. The words that tend to group with internet (problems, issues) and with interaction (lack, poor), create a discourse in which students are passive victims of the power outages and the "digital divide. The study concludes that although the students of Pakistan are enjoying the time saving by using digital platforms, there are significant pedagogical barriers due to poor infrastructure and lack of face to face interaction. The results strongly recommend the use of a hybrid education model by institutions, and a strategic shift to using asynchonous digital interventions.

Keywords: Corpus-Based, Sentiment Analysis, University, Attitudes, Online Learning, Evidence, Public University.