Rediscovering the French Dictée

When I was in ninth grade, we had a large world language (then “foreign” language) department at my high school. My homeroom classmate had an older teacher for French I and I had her daughter for my teacher. For some reason, I recall a conversation about a dictation exercise, the dictée, that the older teacher (who was from France) regularly did. My teacher didn’t do dictées. My homeroom classmate was not too keen on them. That was the first time I heard of the dictée.

I couldn’t tell you why I remember that little homeroom conversation. But I never did have a dictée in any French class right up through my BA degree in French Language and Literature. The dictée had become old fashioned. Its emphasis on correct grammar and spelling were shoved aside as too rigid in the “notional-functional” approach that was growing in the early ’80s and which came to replace grammar-translation.

The dictée returned to my attention last fall when I read an article in FranceInfo about a dictée contest. In fact, such “concours” are pretty common now and the dictée, that classic French pedagogical tool, has enjoyed a resurgence of popularity.

Scroll ahead some months and I find myself teaching a student in French V for an online school. The student is brilliant and already fluent in French, so I am challenged to devise lessons for him when my usual stock of lessons fall within his mastery or when the grammatical studies in the syllabus provided by the school are things he easily tests out of. I asked him what he thought he needed to work on instead, and he said spelling. The dictée sprung to mind as a spelling activity.

Concurrently, I am teaching an AP French course through another company. It’s a small class of eight very bright students who are a pleasure to teach. Unfortunately, they have gotten into the habit of relying heavily on AI to generate their work. Even in improvised, spontaneous chat assignments, some of them repeatedly leave the page (the Innovation synchronous chat app tracks this) to no doubt consult the AI as to what to say. The work many of them submit is polished and perfect beyond their years. Language learning involves detecting common errors and refining the language. But if the students never reveal their errors or lack of knowledge, I cannot easily correct. It feels like an arms race to continually devise activities that are resistant to AI assistance. The dictée sprung to mind as a virtual classroom task that would be difficult to get AI help on.

Since I began teaching French (1990!), I have used composition assignments to look for errors to work on with students. These are assignments modeled on the New York State proficiency test and Regents exams of the era. They measure each clause by comprehensibility, appropriateness, and form. Students have a free error allowance (one for French III and up). These assignments were done in class under supervision with no references (although for classes needing extra support I could allow a certain number of questions). When I began teaching online, I wanted to use this assignment as I had. I coded a World Language App here at Innovation that provides a digital space for students with a proctor and an easy scoring page for teachers. A large percentage of my students use AI extensively to generate these even though I ask them not to. They present polished work that I know they could not have written themselves. One student I had actually created a few errors on purpose to cover the AI consultation. They were random and not the kind of natural “errors” that naturally occur. I don’t make a big deal. I rarely even let on that I know. Some students use AI less as the course goes on when they learn to feel comfortable. I contented myself with turning the perfect compositions into an exercise: I could ask students about what they wrote, the tense they chose, or just offer grammatical descriptions of the work. It was interesting watching them explain the presence of complex structures that they had not been taught yet! 🙂

The Dictée gives me important information about student proficiency and direction for lessons

I tried out a dictée with three students this week, French II, III, and V. I used the Innovation short answer digital learning space in two ways: I did one lesson “live” during our session and one was a recording I made for homework. I was pleased with the results. I learned a lot more about my students’ language abilities than from the weekly composition assignment.

My purpose is to discover student errors so we can correct them and polish them. The errors fall into two categories, lexical (spelling) and grammatical (conjugation and agreement, etc.). But I also learned that the dictée can reveal something about students’ vocabulary knowledge too. Words they do not know they are likely to skip or render phonetically. My guess is that native speakers would do a better job rendering unknown words spelled correctly because they are more familiar with the writing-phonology system of French.

I was happy enough with the results to modify the composition app to allow attaching an audio clip and a model answer. I worked with ChatGTP myself a little to code a function that would quickly assess the students’ spelling. This app is easy to use during synchronous sessions: I merely generate a link from the course playlist, the student saves their dictation, and then on submission I can display on a shared screen to debrief. I can also assign these for homework.

The student’s digital space for submitting dictée.
The scoring page where teachers can highlight errors with different colors for lexical errors and grammatical errors.

Valuable Information about Student Skill Levels

Since it is difficult and unlikely that students will have the time and opportunity to check this with an AI, I get valuable and authentic data about student skill levels. Namely, their lexical spelling, grammatical knowledge, and a good picture of their vocabulary. Research has shown that dictation can indirectly reflect a student’s vocabulary knowledge, since a richer vocabulary base enables more accurate transcription of spoken language. Currently, in remote teaching contexts it is difficult to get this information. Students at all levels are becoming adept at AI queries. They polish and submit work that defeats the purpose of assigning it!

The debriefing on the task is as important as getting the information about student skill levels. Reviewing the corrections and creating custom exercises to train students out of the errors or teach them the grammar structures and vocabulary they need: these are necessary to fully profit by the task.

The senior teacher in my high school, all those years ago, who maintained the tradition of the dictée would smile now, I suspect, to find that some of us are returning to that ancient practice. Keep an eye on the Innovation app that assesses dictés! I plan to refine it as I use the activity more and more.