EVALUATION OF GRAMMARLY IN DETECTING GRAMMATICAL ERRORS: THE CASE OF ENGLISH ESSAYS FROM THE BRITISH NATIONAL CORPUS (BNCWEB CQP – EDITION)

Abstract

The study evaluated Grammarly in detecting grammatical errors in English Essays from the British
National Corpus (BNCweb CQP – Edition). The Corder (1973) framework of Error Analysis
which consists of four sub-categories of errors which are omission, addition, selection, and
misordering was used to analyze the errors that were detected by Grammarly. The study used a
quantitative method approach. With systematic random sampling, a sample of nine English Essays
from British National Corpus (BNCweb CQP – Edition) was extracted. Corrective Feedback (CF)
and Automated Written Corrective Feedback (AWCF) as concepts of language learning provided
a context for the study. The results showed that selection errors were the most dominant 39%
followed by addition errors with 32%, and the least being omission errors with 29%. Spelling
errors were at 60.5%, followed by article errors with 15.8%, punctuation errors with 7.9%, subject
– verb agreement with 5.3%, while preposition errors, word form errors, pronoun errors, and
number errors were each at 2.6%. There are implications that Grammarly is effective in detecting
errors which could be classified into omission, addition, selection, and misordering, and
grammatical categories, such as spelling, article, preposition, punctuation, word form, subject –
verb agreement, pronoun, and number. As shown by the findings of the study, with the use of
Grammarly, writers would produce error free written texts. Therefore, the study recommends
Grammarly as a tool that should be used to ensure error free writing in different phases of life
where writing texts is involved, such as academia. Even so, further research needs to be carried
out on the effectiveness of Grammarly in detecting grammatical errors in the best interest of
encouraging the improvement of its features.