My Experience with AI-powered Digital Writing Assistants: An Illustrated Autoethnographic Study

Here is a first draft of my Research Proposal. I still need to develop the Literature Review. I look forward to your feedback. Thank you!

Introduction: Although online grammar checkers have gained popularity in recent years, there has been limited research conducted on the impact of AI-powered assisted digital writers on students’ revising and editing skills (Cavaleri, 2000). Therefore, my interests lie in the intersection of writing, specifically grammar, and technology; the focus of my research is Grammarly, which is considered one of the most accurate grammar checkers and the most user-friendly. My study examines the algorithm of Grammarly, evaluates other AI-powered assisted digital writers on the market and offers my perspective as a high school English teacher who offers students access to Grammarly Premium and as a graduate student who uses Grammarly Premium for my writing assignments. Overall, my study will provide best practices and recommendations to enhance the overall learning and teaching experience for students and teachers who are interested in purchasing a paid personal or institutional subscription to Grammarly Premium.

From a High School English Teacher’s Perspective:

Some teachers are reluctant to embrace online grammar checkers because they feel that the machines will replace them. My belief is that AI-assisted digital writing assistants will help teachers rather than hurt them. The reality is that there is overcrowding in public schools, and unlike private schools, public school teachers do not have the luxury of having a class size of 10-15 students. In 2020, public school teachers face a class size of 20 – 30 students per class, and with the increase of students in English classes, there has been an impact on the quality of writing instruction. As a teacher in a large public school in New Jersey, I have had 28 students in my AP Language and Composition class. (There are teachers across the country with class sizes up to 30 students.) In my high school, we have 40-minute class periods. During writing workshops, it is mathematically impossible to revise and edit every students’ essays in a timely manner. Students would sign up for a writing conference with me and would want me to revise and edit their entire essay. It takes me at least 10 minutes to look over a draft and multiply that by 28, which is 280 minutes or over 4 hours per one assignment. However, some of the writing conferences would take longer than ten minutes. I tried limiting the writing conferences. I tried teaching the students to revise and edit their essays. I tried scheduling writing conferences during the students’ study hall, and after school, but it became overwhelming and time-consuming. So, this year, in 2019, I took a leap of faith and purchased Grammarly Premium ($139.95/year). I allowed my 10th-grade students in my Level I (college prep) and Level II (regular track) to use my personal Grammarly Premium account to revise and edit their synthesis essays. Overall, my students found Grammarly Premium more helpful than the free, basic version. I would have a student on my desktop computer while conferencing with another student writer. Once the students finish going over their essays, I would ask:

“What is a common error that you noticed?”

“Did you notice a pattern of errors such as subject-verb agreement?”

The most common alert is that of passive voice. From there, I have a conversation with a student regarding the difference between active and passive voice. I follow up with a mini-lesson on the difference between active and passive voice and provide a grammar handout on this grammatical rule for additional practice. I find this backward design in teaching grammar in context more meaningful to the student than starting off with a random worksheet on passive and active voice. Therefore, AI-powered digital writers do not replace the teacher; instead, it allows teachers to help two writers simultaneously and to have more in-depth conversations about grammar. However, there are times where I would observe some students mindlessly clicking the suggestions without reading them or without rereading their essays. Sometimes, Grammarly would make suggestions that are out-of-context, and the writer may need to reject those suggestions. I would have these students read their essays out loud to me.

Literature Review: A majority of the studies on the impact of online grammar checkers involve non-native English speakers in other countries. Two Australian researchers Michelle Cavaleri and Saib Dianati contend that online grammar checkers such as Grammarly promote “self-directed learning and student self-efficacy” (2016). Furthermore, research from Potter and Fuller found that the use of online English grammar checkers increased students’ motivation, engagement, and confidence in grammar rules and English language proficiency” (2008).

METHODOLOGY

I plan to use autoethnography, which is a [qualitative] research method that “seeks to describe and systematically analyze (graphy) personal experience (auto) in order to understand cultural experiences (ethno)” (Ellis, 2011). Specifically, I plan to use a “layered account which focuses on the authors’ experience alongside data, abstract analysis, and relevant literature” (Ellis, 2011). In terms of data, I plan to collect the first drafts of synthesis essays from my current 10th students during the 2019 -2020.

The data that will be collected: 1.) essays without any type of revisions; 2.) essays with just peer conferencing; and 3.) essays with Grammarly Premium. The essays are from students with varying writing abilities, ranging from recently-exited ESL to more advanced students. I plan to notify my supervisor of my research project. In terms of data analysis, I offer “vignettes of my experience with various online grammar checkers, multiple voices, and introspection” (Ellis, 2011). Based on my research, I intend to make recommendations to enhance the AI-powered digital writing assistant experience such as Paperrate, Grammark, After the Deadline and Language Tool.

I plan to collect essays for graduate classes and for blogs that I have written with and without the use of Grammarly Premium. I want to examine the reduction of the errors in my students’ essays and my essays. I want to assess the types of errors that online grammar checkers successfully were able to discern. Cavaleri and Dianati make an important distinction that “grammar checkers do not claim to teach grammar; they are tools to bring potential problems to the writer’s attention” (2016). Grammarly cannot replace a human editor.

Briefly on Grammarly for those who are unfamiliar with it. It was founded in 2009 by Max Lytvyn and Alex Shevchenko. To use Grammarly, users copy and paste a text into the input box, or upload a document. “Grammarly’s free version provides grammar, punctuation, spelling, sentence structure and style support” (see Figure 1). “The premium subscription, which costs $139.95 a year, checks an additional 150 grammar points and provides plagiarism detection, vocabulary enhancement suggestions and a contextual spelling feature and gives users a score out of 100” (Grammarly, 2015) (see Figure 2). It also provides short explanations of each grammar issue it addresses and provides corresponding feedback, which often includes examples of both correct and incorrect usages in green and red respectively (see Figure 3). Users can click the suggested correction to apply it to the text, or click “ignore‟ to move on. Users can also simply read through the feedback without needing to accept or ignore each comment. Before reviewing the text, the premium version also asks users to select audience, formality, domain (academic, business, general, technical, casual, and creative), tone, and intent” (Grammarly, 2020) (see Figure 4.) (Moré, 2006). 

Originally, Lytvyn and Shevchenko developed Grammarly for the college market, thereby offering longer explanations of grammatical rules. As an educator, I appreciated this feature of Grammarly. However, this feature was removed once Grammarly was introduced to the mass market. Recently, Grammarly started hiring human editors to check writing for an additional fee. There were style and syntactical errors that Grammarly was unable to detect, thus requiring human expertise. In terms of the “score” that Grammarly provides it does not necessarily correlate for the grade on the assignment. Teachers take into consideration the context of the essay along with the quality of ideas, the development and elaboration of the ideas, and the insights of the writer — all of which a machine, no matter how good the algorithm, cannot assess. 

2 thoughts on “My Experience with AI-powered Digital Writing Assistants: An Illustrated Autoethnographic Study

  1. Pingback: My Experience with AI-powered Digital Writing Assistants: An Illustrated Autoethnographic Study — Peace & Quiet | OT's with Apps & Technology

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s