Comparison_on_Applications_and-RRRJ-3_2_4
Comparison_on_Applications_and-RRRJ-3_2_4
https://irojournals.com/rrrj
Abstract
English grammar plays an important role in the learning of English language skills.
However, it has always been challenging for students to understand grammar right from school.
Complicated rules make the learning process even harder. Teachers have tried to come up with
many ways to make the learning process interesting and simple yet there exists a gap. Artificial
Intelligence bridges the gap and achieves the goal of attaining grammar learning in interesting
ways. The integration of artificial intelligence (AI) in grammar learning applications represents
a transformative approach to language education. Many applications focus on teaching
grammar by gamifying the learning process. AI-powered applications provide personalized
learning experiences, immediate corrective feedback, and adaptive difficulty adjustments,
leading to improved grammatical competence and sustained user engagement. This research
attempts to make a comparative study of English grammar learning platforms and applications
by analyzing their impact on the autonomous learning of students. Applying AI has been
considered learner-centric as it stimulates learning and fascinates the learners with expected
learning outcomes.
1. Introduction
The use of Artificial intelligence has revolutionized the education sector. According to
the report of the Congressional Research Service, “AI can generally be thought of as
computerized systems that work and react in ways commonly thought to require intelligence,
such as the ability to learn, solve problems, and achieve goals under uncertain and varying
Recent Research Reviews Journal, December 2024, Volume 3, Issue 2, Pages 357-369
DOI: https://doi.org/10.36548/rrrj.2024.2.004 357
© 2024 Inventive Research Organization. This is an open access article under the Creative Commons Attribution-Non-commercial International (CC BY-NC 4.0) License
V. Thamil Selvi, Vaishnavi. B
conditions, with varying levels of autonomy.” [1] In recent times many educationalists have
started to acknowledge the importance of integrating AI in classrooms to enhance the learning
of students. In English language learning one of the topics in which students still struggle is
learning basic grammar concepts and integrating them in communication effectively. This is
because of the complexities of grammar rules and lack of practice. But now with the help of
AI, a teacher can have as many teaching assistants as possible in various forms of applications
and software. However, deciding on the right tool for the students is a challenge. For example,
even though gamification is an effective tool in grammar learning, it is not apt for every
situation and all students. For an advanced learner, gamification is not necessary. This gives
rise to a need for further comparative study in the field of Grammar learning through AI as
studies so far have studied applications in isolation.
Learning grammar is an essential part of language learning. The role of AI is to make
learning easy, effective and interactive. Many free tools are available online which help
learners in improving their grammar. Prominent among them are Grammarly and QuillBot.
Grammarly and QuillBot are grammar correction tools which help students correct their basic
grammar and paraphrase the sentence. Speak and Improve and Deep English are speech-based
AI chatbots that help improve learners' grammar through interactive conversation. This study
attempts to make a comparative study of Grammarly, QuillBot, Speak and Improve, and Deep
English to analyse the role of AI in grammar learning. It also attempts to study the advantages
of simultaneously using grammar correction applications and conversational chatbots to
improve the English language of college students.
2. Problem Statement
Learning grammar can be difficult for various reasons. Many students learn grammar
mechanically in the form of rules rather than understanding the concept. In addition to this,
students have very little conversational practice. Because of this, they find it difficult to use
grammar elements in both written and oral communication. This study is an attempt to analyze
the role of AI in enhancing grammar competency among college students by using a
collaboration of grammar correction applications like Grammarly and QuillBot and
Conversational Chatbots like Speak and Improve and Deep English.
3. Literature Review
In current times, much research is conducted in the field of AI and Language learning.
In the book “Intelligent English Grammar: AI Strategies to Master the Rules”, Infante Vera
Recent Research Reviews Journal, December 2024, Volume 3, Issue 2 358
Comparison on Applications and Impact of AI in English Grammar Learning
argues that, “Artificial Intelligence (AI) has the potential to radically transform the way
grammar is taught and learned in English. AI technologies, such as Natural Language
Processing (NLP), Machine Learning, and Intelligent Tutoring Systems, offer new and
powerful tools to address common challenges in grammar learning and significantly improve
the student experience.”[2] Hea-Suk Kim’s Findings align with these findings as she concludes
in her article that, “while grammar proficiency improved after using AIs grammar correction
during the English composition process, no discernible differences in grammar improvement
were observed across the different AI used.” [3] In research conducted by Nawsha Ghaleb
Shareef, “EFL students perceive AI-powered platforms as beneficial tools for improving their
English grammar proficiency as they can serve as valuable tools for practising grammar,
providing personalised feedback, and adapting to individual learning needs, pace, and styles.”
[4]
In research study entitled “The Role of Grammarly in Assessing English as a Foreign
Language (EFL) Writing”, the use of grammar software was evaluated. The findings of the
study confirm that when students' work is assessed by an AI grammar checker, there is a
substantial decrease in their mistakes. (indirect corrective feedback). The application is more
effective in eliminating mistakes in terms of language use (grammar), and writing mechanics
(spelling and punctuation). [5] Umu Fadhilah also concludes that “Grammarly application can
provide comprehensive feedback to students about grammatical errors, thus saving students
and lecturers time in improving writing, while also training students with the ability to learn
independently.[6]
According to W. Alharbi, AI-powered synchronous text editors are more recent than
asynchronous ones. Examples of those include Grammarly, ProWritingAid, and Writing
Mentor applications which have been gaining popularity in educational, professional, and
personal settings. These tools intelligently provide users with automated written corrective
feedback (AWCF)” [7] Furthermore, in another study conducted by Tran Tin Nghi, Tran Huu
Phuc, and Nguyen Tat Thang, it was concluded that “Chatbots have proved themselves as a
handy tool to help students engage in the lessons, especially in practice activities with mobile-
based application through social networks. Students are eager to share their understandings and
their performance with their peers.[8]
In a similar study, Burr Settles analyzed the impact of AI in Duolingo and concluded that
“The strong relationship between scores based on ML/NLP estimates of item difficulty and the
IRT estimates from operational data provides evidence that our approach—using items’
359
V. Thamil Selvi, Vaishnavi. B
AI plays a vital role in improving language by providing platforms for learning grammar,
the base of a language. Grammarly AI and Quillbot AI use Natural Language Processing (NLP)
to understand and interpret human language, enabling tools to detect errors, suggest
corrections, and even offer style improvements. They employ machine learning algorithms to
analyze vast amounts of text data, learning from millions of writing samples to refine their
error detection and suggestion capabilities. This allows for more nuanced and accurate
grammar and style suggestions. AI tools often incorporate feedback from users, learning from
corrections and suggestions made by human editors to improve their algorithms further.
Grammarly’s AI models use two techniques to analyse a sentence, Syntactic parsing and
Tokenization. Grammarly’s AI breaks down sentences into their grammatical components (like
nouns, verbs, adjectives, etc.). This process, known as syntactic parsing, helps the tool
understand the structure of the sentence, making it possible to detect grammatical errors such
as misplaced modifiers, subject-verb disagreement, and incorrect punctuation. The process of
breaking down a sentence into smaller units called tokens (words, punctuation marks, etc.) is
called Tokenization. Grammarly uses tokenization to analyze each part of a sentence
individually while understanding its role in the larger context.
Grammarly employs supervised Machine learning where the AI is trained on vast
amounts of labelled data. This means the system has seen thousands (if not millions) of
examples of correct and incorrect grammar, allowing it to learn patterns and make predictions
about new, unseen text. For more accurate grammar checking, Grammarly uses feature
engineering to create features (or attributes) from the text. These features might include parts
of speech, word patterns, and sentence structure, which the machine learning models use to
detect errors and suggest corrections. The models used by Grammarly are trained on a variety
of data sources, including books, websites, and user-generated content, to understand different
writing styles and contexts. This broad training helps the tool make accurate suggestions across
different types of content, from formal essays to casual emails.
Grammarly leverages deep learning, particularly neural networks, to improve its
language understanding. Neural networks can model complex relationships between words and
their meanings, allowing Grammarly to make more sophisticated corrections, such as
rephrasing awkward sentences or suggesting style improvements. Grammarly doesn’t just
point out mistakes; it also generates suggestions to improve the text. Using NLG, Grammarly’s
AI can create alternative phrases or sentences that are clearer or more concise, helping users
express their ideas more effectively. Grammarly’s AI models continually improve by learning
from user interactions. When users accept or reject suggestions, this feedback is used to refine
the models, making them smarter over time. Grammarly uses reinforcement learning to adapt
to individual users' writing styles. Over time, it learns from the corrections and suggestions that
users accept or reject, tailoring its advice to better fit their preferences and writing habits.
QuillBot’s core feature is its ability to paraphrase sentences. This involves understanding
the input text and generating a new version that conveys the same meaning using different
words and structures. QuillBot also uses NLP techniques like sentence tokenization and
syntactic parsing to understand the grammar and context of the original sentence before
attempting to rephrase it. QuillBot uses transformer-based models, similar to those in OpenAI’s
GPT series or Google’s BERT. Transformers excel at understanding the context within a piece
of text, which is crucial for generating coherent and contextually relevant paraphrases. The
self-attention mechanism in transformers allows the model to focus on different parts of a
sentence when generating a paraphrase. The models used by QuillBot are fine-tuned on specific
paraphrasing tasks. This involves adjusting the pre-trained models using a smaller, more
focused dataset, which helps the AI generate better paraphrases that are in line with the user’s
expectations. QuillBot often generates paraphrases by replacing words with synonyms. NLG
techniques help ensure that these replacements are contextually appropriate and that the
paraphrased sentence maintains the original meaning.
Grammarly and QuillBot, while serving slightly different purposes, share several
similarities in their AI technologies, particularly in their use of NLP, machine learning, deep
learning, and natural language generation. Both tools are designed to understand context, offer
real-time suggestions, and adapt to user preferences, making them powerful writing assistants.
361
V. Thamil Selvi, Vaishnavi. B
While their specific applications may differ—Grammarly focuses more on grammar and style,
and QuillBot on paraphrasing and summarization—the underlying AI techniques they use are
remarkably similar.
363
V. Thamil Selvi, Vaishnavi. B
engineering students were involved in the study from Tamil Nadu, 10 boys, and 10 girls, all
aged between 18 and 19 years, with similar academic backgrounds. These students were
divided into groups of 5 boys and 5 girls each. To track the level of improvement a pre-test
was given, where the students were asked to write an essay on a selected topic. The proficiency
of the students was assessed and they were categorised as beginner, elementary, intermediate
and upper-intermediate. The Criteria for evaluating the data set is given in the Table 1 below:
Table 1. Evaluation Criteria
Beginner 0-39%
Elementary 40-59%
Intermediate 60-79%
Upper-Intermediate 80-100%
One group of students practised English grammar using Grammarly and Speak &
Improve application and another group was asked to practice using QuillBot and Deep English
application. After 15 days the students were again asked to write an essay on a similar topic.
The details of the pre-test and post-test results for both groups are given in the Tables 2 and 3
and the results are depicted graphically in Figure 1 and 2.
Table 2. Test Results of Students using Grammarly and Speak & Improve
Figure 1. Test Results of Students using Grammarly and Speak & Improve
Table 3. Test Results of Students using QuillBot and Deep English
A single-factor ANOVA test was conducted to determine the similarity between both
groups. ANOVA (Analysis of Variance) is a statistical method used to compare the means of
two or more groups to determine if there are statistically significant differences between them.
When applied to determine students' performance, ANOVA helps in understanding whether
different factors such as teaching methods, study time, or class sections lead to different levels
of student performance. The ANOVA test results for the pre-test scores of Group 1 and Group
2 are as follows:
• F-value: 0.096
• P-value: 0.760
8.1 Interpretation
• The F-value is very low, indicating that the variance between the group means is
small relative to the variance within the groups.
• The P-value is 0.760, which is much higher than a typical significance level of 0.05.
This means that there is no statistically significant difference between the pre-test
scores of the two groups.
Proficiency
Group Test Mean SD T
Growth
Pre-test 38.5 15.31
G1 (N= 10) 30.1 -19.66
Post-test 68.6 12.2
Pre-test 40.6 15.01
G2(N=10) 29.6 -18.62
Post-test 70.2 10.48
Both groups show a significant improvement in the proficiency of the students as shown
in Table 4 . The t-statistics indicate a statistically significant difference between the Pre-Test
and Day 15 scores, suggesting that the use of AI had a significant positive effect on students'
performance. The extremely low p-value confirms that the observed improvement in scores is
highly unlikely to be due to chance. There is a noticeable improvement in the average scores
from the Pre-Test to Day 15, indicating that the student's proficiency increased over time. The
Recent Research Reviews Journal, December 2024, Volume 3, Issue 2 366
Comparison on Applications and Impact of AI in English Grammar Learning
decrease in standard deviation suggests that students' scores became more consistent by Day
15.
8.3 Comparison of the Proficiency Growth of Group 1 and Group 2
The first dataset exhibits slightly higher proficiency growth (30.1 points) compared to
the second dataset (29.6 points). This indicates that the students in the first dataset showed a
marginally greater improvement in proficiency from the pre-test to Day 15. Compared to the
combined use of QuillBot and Deep English, the students using Grammarly and Speak &
Improve show a higher percentage of improvement with a difference of 2.46%. Students in
Group 1 demonstrated an average percentage improvement of approximately 36.08% Whereas
students in Group 2 exhibited an average percentage improvement of about 33.62% This is
because Grammarly offers more accurate grammar corrections and suggestions compared to
QuillBot and the user interface is simple and effective. All 20 students demonstrated
improvement in their grammar skills after using AI for learning. This shows that AI has a
positive impact on the field of autonomous learning of students.
9. Conclusion
Grammar is an essential part of language development as it is the backbone that structures
and refines communication. Studying grammar alone as rules and using traditional learning
methods makes the process tedious. But when integrated into daily communication and
simplified practice, it becomes easy for the learners as the rules get subconsciously embedded
in their minds and become part of their communication skills. However, many students who
graduate from government and government-aided schools learn English at a later stage of their
lives, mostly after entering college. At this stage of their life, they hesitate to seek help as they
are afraid of humiliation and hence their confidence level drops. AI-powered tools play a
crucial role in enhancing grammar learning for various proficiency levels. Even though
grammar learning is very efficient with the help of tools like Grammarly and QuillBot, when
it is paired with conversational chatbots like Deep English and Speak & Improve, it increases
the efficiency as chatbots focus on overall communication development through practice.
Since software like Grammarly, QuillBot, and Speak & Improve are free and user-friendly,
students belonging to any strata can utilise it. Despite recent criticisms of the internet by
Generation Z, it remains a valuable tool for their learning and development
367
V. Thamil Selvi, Vaishnavi. B
References
[11] Younis, Hussain A., Nur Intan Raihana Ruhaiyem, Wad Ghaban, Nadhmi A. Gazem,
and Maged Nasser. "A systematic literature review on the applications of robots and
natural language processing in education." Electronics 12, no. 13 (2023): 2864.
[12] Fuchs, Kevin. "Exploring the opportunities and challenges of NLP models in higher
education: is Chat GPT a blessing or a curse?." In Frontiers in Education, vol. 8, p.
1166682. Frontiers Media SA, 2023.
[13] Fitria, Tira Nur. "Grammarly as AI-powered English writing assistant: Students’
alternative for writing English." Metathesis: Journal of English Language, Literature,
and Teaching 5, no. 1 (2021): 65-78.
[14] Nicholls, Diane, Kate Knill, Mark JF Gales, Anton Ragni, and Paul Ricketts. "Speak &
improve: L2 English speaking practice tool." In Proceedings of Interspeech 2023, pp.
3669-3670. Sheffield, 2023.
[15] Mohammad, Taj, Ali AF Alzubi, Mohd Nazim, and Soada I. Khan. "Paraphrasing
Prowess: Unveiling the Insights of EFL Students and Teachers on QuillBot Mastery."
International Journal of Information and Education Technology 14, no. 5 (2024).
Author's biography
369