Automated Writing Support Using Deep Linguistic Parsers (original) (raw)

NTUCLE: Developing a Corpus of Learner English to Provide Writing Support for Engineering Students

2017

This paper describes the creation of a new annotated learner corpus. The aim is to use this corpus to develop an automated system for corrective feedback on students’ writing. With this system, students will be able to receive timely feedback on language errors before they submit their assignments for grading. A corpus of assignments submitted by first year engineering students was compiled, and a new error tag set for the NTU Corpus of Learner English (NTUCLE) was developed based on that of the NUS Corpus of Learner English (NUCLE), as well as marking rubrics used at NTU. After a description of the corpus, error tag set and annotation process, the paper presents the results of the annotation exercise as well as follow up actions. The final error tag set, which is significantly larger than that for the NUCLE error categories, is then presented before a brief conclusion summarising our experience and future plans.

Essay Scoring with Grammatical Error Detection

English proficiency is an important skill in the world today. For non-native En-glish speakers, an automated system capable of reviewing practice essays would assist English learners without requiring involvement of native essay graders. This project explored the use of an LSTM-RNN machine model for identifying existence of grammatical errors in essays written by non-native learners of English. The relatively high accuracy obtained suggests that using LSTM-RNN models is a promising approach and recommends future opportunities for refinements.

Grammar Correction AI Tools for English Language Teachers in Higher Education

IJIRT, 2024

The advent of AI-enhanced grammar correction tools marks a significant evolution in language education, shifting from traditional rulebased systems to sophisticated applications that leverage artificial intelligence (AI) and natural language processing (NLP). These modern tools provide real-time feedback, personalized instruction, and context-aware suggestions, catering to individual learner needs. By utilizing advanced algorithms and deep learning techniques, these tools excel in detecting complex grammar patterns and subtle errors that conventional checkers often overlook. Key features include adaptive learning capabilities, which track user progress and deliver tailored recommendations, thereby fostering learner autonomy and confidence. Recent studies indicate that the integration of AIdriven grammar correction tools significantly improves writing fluency and reduces error frequency among language learners. Furthermore, these tools enable educators to enhance traditional teaching methods by providing scalable support that addresses diverse learning styles within the classroom. As technology continues to advance, AI-enhanced grammar correction tools promise to further transform language education by making it more accessible and effective.

Semi-automated typical error annotation for learner English essays: integrating frameworks

Proceedings of the 4th workshop on NLP for Computer Assisted Language Learning at NODALIDA 2015. NEALT Proceedings Series114: 35–41. 35 26 / Linköping Electronic Conference Proceedings 114:35-41, 2015

This paper proposes integration of three open source utilities: brat web annotation tool, Freeling suite of linguistic analyzers and Aspell spellchecker. We demonstrate how their combination can be used to pre-annotate texts in a learner corpus of English essays with potential errors and ease human annotators’ work. Spellchecker alerts and morphological analyzer tagging probabilities are used to detect students’ possible errors of most typical sorts. F-measure for the developed pre-annotation framework with regard to human annotation is 0.57, which already makes the system a substantial help to human annotators, but at the same time leaves room for further improvement.

Computer-Assisted Writing Revision: Development of a Grammar Checker

A computerized grammar checker was developed to assist teachers of English as a Second Language in editing student compositions. The first stage of development consisted of an error analysis of 175 writing samples collected from students. The 1,659 errors found were classified into 14 main types and 93 subtypes. This analysis served as the basis for constructing a taxonomy of mistakes and ranking the categories according to frequency of occurrence and comprehensibility. The grammar checker was then designed with a small electronic dictionary containing, 1,402 word stems and necessary features, and a suffix processor to accommodate morphosyntactic variants of each word stem. An augmented transition network parser equipped with phrase structure rules and error patterns was then constructed. In addition, a set of disambiguating rules for multiple word categories was designed to eliminate unlikely categories, increasing the parser's efficiency. The current implementation detects seven types of errors and provides corresponding feedback messages. Future research will focus on detecting more kinds of mistakes with greater precision and on providing app:opriate editing strategies. (Author/MSE)

ERRANT: Assessing and Improving Grammatical Error Type Classification

2020

Grammatical Error Correction (GEC) is the task of correcting different types of errors in written texts. To manage this task, large amounts of annotated data that contain erroneous sentences are required. This data, however, is usually annotated according to each annotator's standards, making it difficult to manage multiple sets of data at the same time. The recently introduced Error Annotation Toolkit (ERRANT) tackled this problem by presenting a way to automatically annotate data that contain grammatical errors, while also providing a standardisation for annotation. ERRANT extracts the errors and classifies them into error types, in the form of an edit that can be used in the creation of GEC systems, as well as for grammatical error analysis. However, we observe that certain errors are falsely or ambiguously classified. This could obstruct any qualitative or quantitative grammatical error type analysis, as the results would be inaccurate. In this work, we use a sample of the F...

Corpus-based Error Detector for Computer Scientists

2018

This study describes the design and development of a corpus-based error detector for short research articles produced by computer science majors. This genre-specific error detector provides automated pedagogic feedback on surface-level errors using rule-based pattern matching. In the corpus phase, a learner corpus of all theses (n = 629) submitted for three academic years was compiled. A heldout corpus of 50 theses was created for evaluation purposes. The remaining theses were added to the working corpus. Errors in the working corpus were identified manually and automatically. The first 50 theses were annotated using the UAM Corpus Tool. Errors were classified into one of five categories (i.e. accuracy, brevity, clarity, objectivity and formality). By the fiftieth thesis, saturation had been reached, that is the number of new errors discovered had dropped considerably. Annotated errors were extracted into an error bank (xml file). Each error was assigned values for severity, detecta...

A corpus-based lexical and grammatical error identification: L2 learners academic writing

2021

Writing in English has never been an easy task to many second language (L2) learners. Many of them perform poorly in their English academic writing where numerous lexical and grammatical errors are found in their report. Therefore, this thesis investigates the difficulties faced by UTHM learners involved in academic writing by identifying and analyzing errors made by them with the application of error analysis procedures. This research attempts to find out the types and patterns of errors in which it focuses on the frequency of the lexical and grammatical errors of the L2 learners in their writing. Errors were investigated and identified based on students’ 36 progress and final reports which were assembled from first year engineering students; named as the Learner Corpus Universiti Tun Hussein Onn Malaysia(LCUTHM). The LCUTHM was analyzed by means of linguistics Natural Language Processing tools (NLP) such as CLAWS 5 tag set, Markin Version 4 and categorized by MonoConc Pro II in th...

Application for Grammar Checking and Correction

2020

1,2,3Student, Dept. of Computer Engineering, Vidyalankar Institute of Technology, Mumbai, India 4Professor, Dept. of Computer Engineering, Vidyalankar Institute of Technology, Mumbai, India ---------------------------------------------------------------------***---------------------------------------------------------------------Abstract This paper identifies and examines the key principles underlying building a state-of-the-art grammatical error correction system. Techniques that are used include rule-based, syntax-based, statistical-based, classification and neural networks. This paper presents previous works of Grammatical Error Correction or Detection systems, challenges related to these systems and at last suggests future directions. We also present a possible scheme for the classification of grammar errors. Among the most observations, we found that efficient and robust grammar checking tools are scarce for real-time applications. Natural Language consists of the many sentence...

The QuillBot Grammar Checker: Friend or Foe of ESL Student Writers

Creative Practices in Language Learning and Teaching, 2022

The use of automated written corrective feedback applications in the writing classroom has been widely researched. However, as the QuillBot grammar checker was newly launched, no studies had been conducted on its accuracy and usability among English as a Second Language (ESL) student writers. To address this gap, a small-scale sentence-and paragraph-level test was carried out to determine whether it would be a useful tool for ESL students to obtain feedback in realtime instead of waiting for teacher corrective feedback. The new, freely available grammar checker was tested along with the free version of two other popular grammar checking tools, Grammarly and Ginger. Actual writing samples from second-year diploma students were used. The results indicate that the QuillBot grammar checker outperformed the other two software. However, this tool is not perfect and users will still have to manually check for undetected errors and decide if suggestions should be accepted or ignored. Therefore, it is recommended that QuillBot should be used to supplement and not replace teacher feedback. With students using the QuillBot grammar checker in their writing tasks, teachers do not only save time on checking language errors but also have more time to provide feedback regarding global writing concerns, namely, content and organization.