Date of Degree

2-2020

Document Type

Thesis

Degree Name

M.A.

Program

Linguistics

Advisor

Martin Chodorow

Subject Categories

Bilingual, Multilingual, and Multicultural Education | Computational Engineering | Education | Educational Technology

Keywords

Computational Linguistics, Second Language Acquisition, NLP, English as a New Language

Abstract

Although educators and linguists have described and studied L2 learner errors, often from the perspective of positive and negative transfer (Krashen, 1981), there is relatively little empirical data on the frequencies of L2 error types. Learner error rankings do exist (Donahue, 2001) (Cambridge learner corpus; (Nicholls, 1999)), but these rankings are often too general and may overlook specifics when it comes to particular language learner groups. The goal of our work is to develop a tool that can be used to explore error patterns, address educator needs, and help answer research questions in L2 language learning. The tool we have developed combines an online corpus of corrected learner sentences (Lang-8; Mizumoto, T. et. al. 2013), an Error Annotation Toolkit (ERRANT; (Bryant, C. et. al. 2017)), and a dependency parser (SpaCy; (Honnibal, M. et. al. 2015)) to provide a more informative view of learner errors.

To test the usefulness of the tool, we asked educators of Spanish-speaking English learners from an international high school to provide a list of error types that they commonly encounter in the classroom. One of the most frequently mentioned error types was Pro-drop of the 3rd person neuter singular (It), which is the focus of this thesis. Using our tool, we investigated Pro-drop instances within learner sentences relative to the syntactic role of the pronoun and its corresponding correction. Results showed a significantly higher error rate of Pro-drop with the 3rd person neuter singular particularly in nominative position in relation to other pronoun types. Investigating errors, such as Pro-drop produced by specific L1 groups, has implications for discourses formed around such topics and may prove to be beneficial to educators, researchers, and developers of educational technology.

Share

COinS