Teaching tips for better responses, or
Sometimes silence is a really good answer, said no language teacher ever.
How do you ensure that your students are giving their best responses?
I’ve recently added a really effective strategy to enhance students’ answers to my toolkit. I came across this approach in the context of machine learning and prompt engineering, yet I find it extremely valuable for my students. I thought it would be worthwhile to share it on the blog [#bloggingisnotdead…well, not yet], along with other strategies to facilitate better student responses.
I believe that in the era of AI, where a ‘wave of speech’ can be generated within 30 seconds, the ability to construct meaningful and in-depth responses is a skill that will be increasingly in demand.
The approach goes by the name of the Cognitive Verifier Pattern. What it does is guide the large language model to make a series of follow-up questions for the user to answer before combining those responses to give an answer to the main question. The reason behind this approach is simple: sometimes, when people are unfamiliar with a topic or unsure of how to phrase a question, they might ask questions that are too broad to get a precise answer. By generating these additional questions, the language model helps ensure that the final answer is thorough and accurate. It also encourages users to think critically and can lead to fresh insights and better responses that they might not have considered initially.
I had the same line of thinking – when a learner is given a set of subquestions on a vague or broad topic, they are more likely to provide a meaningful in-depth response. I also thought that this was where AI could come in handy to assist learners in brainstorming these subquestions. Remarkably, after a few rounds of questioning with the help of AI, learners learners start firing off their own follow-up questions way quicker.
To test this approach, I’ve developed Q-Smart, a customized app designed to help learners generate subquestions. You can try the demo version on My English Domain, or use ChatGPT or Bing directly. In the app, I’ve streamlined the initial prompt pattern: ‘When I ask you a question, generate three additional questions to aid in providing a more comprehensive and nuanced response to the initial question.’ We’re currently piloting a series of Employability Skills e-modules and have implemented this approach to assist students in handling challenging job interview questions, like ‘Where do you see yourself in 10 years?’ and ‘Why should we hire you?’—all without model answers but utilizing subquestions. The students’ initial responses were remarkably impressive. It was probably the first time I’d heard such comprehensive answers from students without any additional preparation.
Students can also choose one of the subquestions and use Q-Smart to generate more nuanced questions for a deeper exploration.
If you’re making questions for your students, just mention who they are—like pre-service students or teenagers—in the first prompt. It makes the questions more on point for them.
I also tried the approach for Cristina Cabal’s challenge – The Great Silence Breaker Silly Challenge: Can my Students Talk about a Simple Object for 2 Minutes?. In this challenge, students are asked to speak non-stop for about two minutes on a simple object, such as a spoon or a cup of coffee.
‘Talk about a spoon. You have two minutes. Go.’
As readers of this blog know, I’m a big fan of spoons, so two minutes would not be enough to tell everything I’d like to about spoons. Nonetheless, it’s quite a challenge.
Now, let’s take the same challenge but use Q-Smart: types, designs, materials, use cases, symbolic and ritualistic roles, your first spoon, idioms with a spoon, and whatnot. There is a whole saga behind!
Combined with other tools and strategies aimed at helping learners practise giving better answers (and use wait time in the classroom more effectively), this approach may assist learners in delivering more substantial and meaningful responses.
Ask better questions, get better answers—it’s that simple!
