Understanding ELL Stack Exchange: An Investigation into Discussion Topics and Answer Acceptance Criteria
Date
2024-05-30
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
ORCID
0009-0008-0481-0197
Type
Thesis
Degree Level
Masters
Abstract
Internet and technological advancements have revolutionized the way we learn and communicate. English language learners (Ells) are increasingly gravitating towards online platforms like Stack Exchange to seek help with their English-related queries. These virtual environments, rich with queries and discussions, present an untapped repository of insights. However, there is a lack of comprehensive studies on the questions Ells ask online and the criteria for answer acceptance within these communities. This thesis aims to delve into the content shared on ELL Stack Exchange, a question-answering online forum, designed for Ells to explore the type of questions posed by Ells, and factors contributing to answer acceptance.
Utilizing Latent Dirichlet Allocation (LDA) topic modeling, the study analyzed 27,847 questions from 2013, 2017, and 2021. The analysis identified seven key topics of discussions among which, Topic 6 (Grammar and Meaning) with 1,372 questions in 2013, 4,514 questions in 2017, and 4,232 questions in 2021 was the most common topic of discussion. In contrast, Topic 1 (Verb Usage) with about 1,000 questions over the whole dataset was the least discussed topic.
Further, by employing logistic regression analysis on a dataset of 29,700 responses, the study investigated the contribution of various factors to answer acceptance. The potential factors were the timeliness of answers (how quickly answers are provided after the question is posted), length of answers, readability score of answers, reputation score of answer providers, and the count of external links (URLs) within answers. The findings revealed that prompt, longer, and more comprehendible answers are more likely to be accepted. Users’ reputation score also plays a crucial role, highlighting the community’s trust in contributors with a proven track record of providing valuable content. The presence of URLs in answers does not significantly influence the acceptance rate.
By identifying prevalent topics of discussions and determinants of answer acceptance, this research contributes to the broader understanding of what English language learners find most challenging and how they prefer to get help. This information can be used to develop instructional materials based on learners’ needs and enhance online Q&A platforms for language learning based on learners’ preference of receiving answers.
Description
Keywords
English Language Learners (ELLs), Online Q&A Communities, Topic Modeling, Community Dynamics, Answer Acceptance Criteria
Citation
Degree
Master of Arts (M.A.)
Department
Linguistics
Program
Linguistics