Skip to main content

The best AI-proof homework tasks?

 Even when Google Translate was the main challenge, teachers have been concerned with designing homework tasks which make it harder for students to cheat. And I say ‘cheat’ unapologetically, since we know what we are talking about here: passing off work as your own when it isn’t. Because of Google, and now GenAI, teachers have resorted to setting vocab learning, or other tasks which don't require writing. You probably know what I think about vocab learning. If you don’t, look here. In a nutshell, not uselss but... ahem...sub-optimal.

Traditionally, many teachers liked to prioritise listening and speaking in class, reserving most writing for homework tasks. Is it time to reverse this? Maybe, to some extent. I still think that in some schools, where teachers are able to hold the line, it is possible to be pretty sure students have done their own work. They just need to know that cheating is unacceptable and will result in unpleasant consequences: a telling-off, detention or rewrite. This is down to individual teachers, departmental consistency and school culture. I understand why in many circmstances this may be a tough line to hold.

So, the basics: students have to process language they understand to build their knowledge of vocabulary and linguistic system. If AI does the job for them (cognitive offloading is the term used for this), their progress will be constrained.

Are there homework tasks we can set which are partially or wholly AI-proof? Don't bother asking AI, since its ideas (as Chat GPT acknowledged to me) usually allow students to write in the first language and get it translated. Even speaking tasks can fall victim to this - a student just has to get AI to produce the language, then read it aloud.

However...

Listening to audio or watching video, accompanied by comprehension exercises (true/false, questions, gap-fill, matching, etc) offers a useful solution. There are AI tools that can transcribe YouTube videos when they have captions, but this a fiddly affair for students, so they may not go to the trouble. If the audio or video is recorded by you or the department, AI can't do much with it. Note that I avoided including writing answers in the target language, since, once comprehension is sound, TL answers can be done by AI.

Audio and video will mainly suit intermediate and advanced learners, but you can find easier material in various languages, for example Audio Lingua. Your school’s learning platform may allow for course book audio files, or your own, to be shared for homework. This all assumes that every student has access to any equipment needed - a phone may be adequate. Listening work can be completed and uploaded to the school's platform for assessment. In due course, it may become commonplace for AI to do the assessment, though there are obvious caveats here. Students need to know you care about their work and are assessing at some of it yourself.

If vocab learning is so deeply ingrained in your language teacher psyche (!), then you might at least consider using sentence builders or parallel text knowledge organisers as your source material. This means that vocab is seen in contaxt and the chances of transferring chunked vocab knowledge to fluent use is greater. I would give a sentence builder and tell students to memorise as many sentences as possible from it, maybe adding their own if they are good enough. Back in class, they can then share their knowedge with a partner by playing "who can keep producing senetnces the longest", or writing down from memory as many sentences as they can. This may satisfy your "learn + test" itch!

To add, it may well be that interactive language learning tools may allow you to be certain students have worked intensively with language. These come at a cost, but many schools make much use of suitable sites (see below).

In sum, if I were working on a department on homework and the chalenges of AI, I would be suggesting:

1. Traditional written exercises and compositions, enforcing with sanctions where needed.

2. Greater use of video and audio listening, along the lines suggested above.

3. Chunked vocab and sentence memorisation using senetnce builders or even parallel text knowledge organisers or 'chat mats'.

4. Using interactive platforms to do process and manipulate language, where there is evidence of task completion. Tools such as Textivate, sentencebuilders.com, TeachVid, The Language Gym and Languagenut may fit the bill.

I would of course be getting students to think about how they learn a language, how processing language intensively is important and how passing off work as your own when it isn't is dishonest. I'd also make clear how important homework is, saying something along the lines "To make progress you need language you understand and a chance to use it. the more you do this, the more progress you make. Therefore, doing homework is a no-brainer." (Do not be lured by people who argue against homework full-stop, citing research - such research only holds true for primary learners.)


Comments

Popular posts from this blog

Zaz - Si jamais j'oublie

My wife and I often listen to Radio Paradise, a listener-supported, ad-free radio station from California. They've been playing this song by Zaz recently. I like it and maybe your students would too. I shouldn't really  reproduce the lyrics here for copyright reasons, but I am going to translate them (with the help of another video). You could copy and paste this translation and set it for classwork (not homework, I suggest, since students could just go and find the lyrics online). The song was released in 2015 and gotr to number 11 in the French charts - only number 11! Here we go: Remind me of the day and the year Remind me of the weather And if I've forgotten, you can shake me And if I want to take myself away Lock me up and throw away the key With pricks of memory Tell me what my name is If I ever forget the nights I spent, the guitars, the cries Remind me who I am, why I am alive If I ever forget, if I ever take to my heels If one day I run away Remind me who I am, wha...

Longman's Audio-Visual French

I'm sitting here with my copies of Cours Illustré de Français Book 1 and Longman's Audio-Visual French Stage A1 . I have previously mentioned the former, published in 1966, with its use of pictures to exemplify grammar and vocabulary. In his preface Mark Gilbert says: "The pictures are not... a mere decoration but provide further foundation for the language work at this early stage." He talks of "fluency" and "flexibility": "In oral work it is advisable to persist with the practice of a particular pattern until the pupils can use it fluently and flexibly. Flexibility means, for example, the ability to switch from one person of the verb to another..." Ah! Now, the Longman offering, written by S. Moore and A.L. Antrobus, published in 1973, just seven years later, has a great deal in common with Gilbert's course. We now have three colours (green, black and white) rather than mere black and white. The layout is arguably more attrac...

What is skill acquisition theory?

For this post, I am drawing on a section from the excellent book by Rod Ellis and Natsuko Shintani called Exploring Language Pedagogy through Second Language Acquisition Research (Routledge, 2014). Skill acquisition is one of several competing theories of how we learn new languages. It’s a theory based on the idea that skilled behaviour in any area can become routinised and even automatic under certain conditions through repeated pairing of stimuli and responses. When put like that, it looks a bit like the behaviourist view of stimulus-response learning which went out of fashion from the late 1950s. Skill acquisition draws on John Anderson’s ACT theory, which he called a cognitivist stimulus-response theory. ACT stands for Adaptive Control of Thought.  ACT theory distinguishes declarative knowledge (knowledge of facts and concepts, such as the fact that adjectives agree) from procedural knowledge (knowing how to do things in certain situations, such as understand and speak a langua...