The integration of technology into language education presents a number of pedagogical challenges, particularly when it comes to determining how best to leverage digital tools to enhance teaching and learning outcomes. Educators must not only consider how technology can be incorporated effectively into instructional strategies, but also how they can guide students in selecting and using technological resources that contribute to their cognitive development and language acquisition.
In 2020, when pandemic forced educators to transition to remote teaching and platforms like Zoom, new challenges emerged—particularly regarding assessment integrity. With home-based exams, students were often confronted with the temptation to rely on unauthorized external resources. In language courses, some students resorted to tools like Google Translate (GT) as a shortcut for completing assignments, class activities, and assessments. In response, I developed an activity designed to help students recognize the limitations of Google Translate (see full document). The activity encouraged students to analyze GT translations and compare them with more contextually accurate alternatives.
Limitations of GoogleTranslate
English to French
Starting with a perfectly correct English sentence, compare the GoogleTranslate translation and a Human translation:
“In their house, everything comes in pairs. There’s his car and her car, his towels and her towels, and his library and hers.”
GoogleTranslate: Dans leur maison, tout vient en paires. Il y a sa voiture et sa voiture, ses serviettes et ses serviettes, sa bibliothèque et les siennes.
Human: Chez eux, ils ont tout en double. Il y a sa voiture à elle et sa voiture à lui, ses serviettes à elle et ses serviettes à lui, sa bibliothèque à elle et sa bibliothèque à lui.
Working with a partner, notice the issues with the GoogleTranslate translation. What is not accurately translated?
The activity also invited students to use WordReference to translate several sentences from English to French.
Practice using WordReference
Using Word Reference, translate into French the following sentences that use the words ‘own’ and/or ‘practice’:
My team has practice every day.
I want to have my own practice.
I do not own a car.
Do you have your own?
Did he own up to breaking the door?
Can you put that into practice?
Does your band have a practice this evening?
It is common practice.
She wants to work in the field of family practice.
In practice, it is much more complicated.
Do you know if he practices medicine or law?
However, I quickly realized that this activity was too ambitious for elementary-level learners. Since then, I have adapted my approach. For instance, during lessons focused on the French song "Les Questions," I demonstrate how GT inaccurately translates la petite souris as "little mouse" rather than the culturally and linguistically appropriate translation of "Tooth Fairy." I then guide students to use WordReference, where the correct translation is readily available. This method not only emphasizes the limitations of GT but also encourages students to engage with more reliable linguistic resources.
The release of ChatGPT in the fall of 2022 introduced a new challenge, as students quickly discovered its potential as a tool to assist with homework assignments. For years, at the elementary level, students were tasked with writing compositions at home. Initially, students were instructed to write drafts at home and use WordReference or Google for vocabulary support, with the expectation that they would flag any uncertain or machine-generated corrections. The goal was for students to engage in the process of revision and error correction, reflecting on and improving their language skills in collaboration with the instructor.
However, as the capabilities of AI tools like ChatGPT became more sophisticated, it became clear that the traditional approach of home-based composition assignments was no longer viable. To maintain academic integrity and ensure students' active participation in the writing process, we moved composition assignments to the classroom. Instructors would then provide targeted feedback on students' drafts, highlighting areas in need of revision. Students were subsequently tasked with making corrections at home and submitting a revised version via Canvas. This system encouraged independent revision, but the advent of ChatGPT posed a new challenge: students could easily use it to bypass the error correction process. To address this, we introduced a new component to the revision process starting in 2022. Students were required to write a self-reflection in which they identified and explained the errors they had corrected. This metacognitive task not only encouraged students to reflect on their learning process but also provided instructors with insights into students' understanding of their mistakes and the rationale behind their revisions.
In October 2023, we realized that ChatGPT itself could serve as a pedagogical tool to facilitate this process. Rather than being a shortcut, we experimented with having students use ChatGPT as a virtual tutor. Students were asked to input their first draft of a composition into ChatGPT, requesting the AI to correct errors and generate a list of explanations for the changes. The prompts used were:
- "Correct the following text in French. Take out the spelling and grammatical errors."
- "Correct the following letter and provide a list of the changes with an explanation."
After receiving the corrected text and the list of changes, students were instructed to review both the corrections and the explanations carefully. A week later, students wrote a second draft in class, which allowed instructors to assess how much of the corrected language students had internalized. To evaluate the effectiveness of this approach, I developed a brief survey asking students to reflect on their experiences and whether they felt the use of ChatGPT was helpful in improving their writing skills.
The results of the survey revealed that 58.8% of students found using ChatGPT more beneficial than generating the list of changes themselves. Less than 15% of students preferred generating the list on their own, while 26.5% expressed uncertainty about the usefulness of ChatGPT.
