Eliciting Techniques That Make Students Do the Work
Eliciting is the technique of drawing language out of students rather than feeding it to them, and it's the single highest-leverage skill for online tutors because it inverts the tutor-talk-time problem at the root. If you're explaining vocabulary, defining grammar rules, or filling silences with model sentences, you're doing the cognitive work the student should be doing. The working rule is simple: ask before you tell, wait three seconds longer than feels comfortable, and treat the student's first attempt as data rather than a failure.
Why does my student rely on me to fill silences?
Students rely on the tutor to fill silences because every prior lesson has taught them that the tutor will. This is the trained-helplessness pattern of marketplace tutoring — drop-in lessons with high pressure to look "engaging," which translates into tutors talking 70–80% of the time and students getting comfortable in a passive role. The student isn't lazy; they've learned the rules of the game.
The classroom-discourse research is unambiguous on the cost. Tutor-talk-time (TTT) is inversely correlated with student speaking proficiency gains — the more the tutor talks, the less the student produces, and language is a production-driven skill (Swain, 1985, on the output hypothesis; Ellis, 2008, on the role of pushed output). The marketplace incentive structure pushes tutors toward over-talking because high TTT looks like "value delivered" to a student rating an isolated lesson; it just doesn't produce learning gains over a 50-lesson arc.
The remedy isn't to talk less — it's to ask better questions and wait through the discomfort. The screen-mediated awkwardness of silence is real, but the silence is where the cognitive work happens. A student who's never given three seconds of think-time never develops the production muscle. A student who's given three seconds, then three more, then prompted gently, develops a productive vocabulary at twice the rate.
How do I check understanding without asking "do you understand?"
"Do you understand?" is the worst question in language teaching because the answer is always yes. Students nod to be polite, to avoid embarrassment, to move the lesson forward — and you've learned nothing about whether the input landed. Concept-check questions (CCQs) replace the yes-no with questions that can only be answered correctly if the concept was understood.
The CCQ pattern: identify the meaning components of the target language, then ask questions that test each component independently. For the structure "I used to live in Paris":
- "Do I live in Paris now?" (Tests the past-not-present component.)
- "Did I live in Paris one time only, or for a long period?" (Tests the habitual/repeated past component.)
- "Is this about now or about the past?" (Tests tense.)
Three short questions, twenty seconds of student work, total clarity on whether the structure was understood. Compare with "Do you understand used to?" — which produces a head-nod and zero information.
The same principle scales to vocabulary. For the word furious:
- "Is furious a strong feeling or a weak one?" (Intensity.)
- "Is furious a positive feeling or a negative one?" (Valence.)
- "What might make someone furious?" (Production check.)
CCQs work because they convert receptive recognition into production. A student who can answer them has the word; a student who can't, doesn't — and now you know which side they're on.
What are five eliciting prompts I can use today?
Eliciting prompts are the verbal moves that shift cognitive work from tutor to student. Below are the five highest-leverage prompts working tutors use in online lessons — they cost nothing to deploy and transform tutor-talk-time within a few sessions.
| Prompt | When to use | What it elicits |
|---|---|---|
| "Try that again — how would you say it?" | After an error you'd normally correct | Self-correction; metalinguistic awareness |
| "What word might you use here?" (with a pause for a visual or context) | Vocabulary teaching | L1 transfer attempt + receptive vocabulary check |
| "Tell me more about that." | Anytime a student's answer is one sentence | Extended output; spontaneous structure |
| "What's the opposite of...?" / "What's another way of saying...?" | Vocabulary recycling | Lexical chains; antonyms; paraphrase |
| "What do you think — is this true or false?" | Reading or listening tasks | Comprehension + opinion + structure |
The first prompt — "Try that again — how would you say it?" — is the single highest-impact change a tutor can make. It substitutes for almost all real-time corrections. Lyster and Ranta's correction-uptake research (1997) showed that prompts (the "elicit a correction" type) had higher uptake than recasts in classroom interaction; the effect transfers to online tutoring where the friction of stopping the conversation is even higher.
The second prompt requires a visual or a context — show a picture, type a sentence with a gap, point at something on the screen. Eliciting vocabulary from a definition you give is harder than eliciting it from a context the student can interpret. The visual or context does the heavy lifting.
How do I elicit grammar without lecturing?
Grammar elicitation works by presenting examples and asking the student to extract the rule, rather than presenting the rule and asking the student to apply it. This is the discovery approach to grammar teaching, and it's slower per concept but produces longer-lasting retention because the student did the analytical work.
The pattern is straightforward:
- Show 3–5 examples of the target structure in context. For the second conditional: "If I won the lottery, I'd buy a house." "If I had more time, I'd learn Spanish." "If she lived closer, we'd see her more often."
- Ask diagnostic questions about the examples. "Are these about now, the past, or the future?" "Are they real situations or imaginary?" "What tense is in the if-clause?"
- Let the student state the rule, with prompts as needed. "So when do we use this structure?"
- Confirm or refine. "Yes — that's exactly right. We use the second conditional for imaginary situations in the present or future."
The student arrives at the rule themselves. The tutor's verbal contribution is questions, not exposition. The total tutor-talk-time for a 5-minute grammar focus drops from 80% to maybe 30%, and the student leaves the lesson having actively constructed the grammar point rather than passively received it. The contrast with the marketplace "I'll explain the second conditional now" model is night and day in retention terms — see how to teach grammar online without lecturing for the longer treatment.
How long should I wait after asking a question?
The minimum useful wait time is three seconds; the optimum is closer to five. The classroom interaction research (Rowe, 1986, on "wait time"; replicated repeatedly since) shows that teachers who extend wait time from the typical 1–1.5 seconds to 3+ seconds see student-response length triple, response accuracy increase, and student-initiated questions multiply. The effect is large, well-established, and almost completely unused by online tutors who feel screen pressure to fill silence.
Three practical tactics for extending wait time:
- Count silently to four. The discomfort peaks around three seconds and then dissipates. Counting gives you a physical mechanism to override the urge to interrupt.
- Use micro-encouragement. "Take your time." A nod. A brief "mm." These signal that the silence is OK and you're not waiting impatiently — they extend wait time without filling it.
- Build wait-time into your lesson plan. If you've designed an activity that needs 2 minutes of student preparation, give the full 2 minutes. Don't compress it because the screen feels quiet. Use the time to read your own notes, type a question into chat, drink water.
The hardest case is the student who asks for help after one second of silence. The polite response is to acknowledge the request and extend the wait: "Take a moment — what's your first instinct?" Most students will produce a usable answer in the next 5 seconds. The few who genuinely can't will signal it more clearly, and that's your cue to scaffold.
How do I elicit from a low-level student without exposing what they don't know?
Low-level elicitation requires scaffolded prompts — questions that progressively narrow the range of acceptable answers until the student can produce something. A B1 student asked an open question ("Tell me about your weekend") freezes; the same student asked a closed-then-open sequence ("Did you work on Saturday? OK — what did you do instead?") produces 3–4 sentences. The prompts are doing the scaffolding work the student can't yet do internally.
The progressive scaffolding pattern:
- Closed question to anchor. "Did you cook dinner last night?" Yes/no answer is achievable.
- Wh-question on the same topic. "What did you cook?" Single word or short phrase.
- Extended prompt. "How did you make it — what did you do first?" Multiple sentences, sequencing language.
- Spontaneous follow-up. "What's your favourite thing to cook?" Open-ended, building on the established topic.
Each prompt builds on the previous answer; the student isn't being asked to generate language from nothing, they're being asked to extend language they've already produced. This is the elicitation equivalent of i+1 input (Krashen, 1985): one step beyond the student's current production, scaffolded by what they just said.
For students who freeze even on the closed question, drop further: "Did you cook? Or did you eat out?" Forced choice between two given options. This is the lowest scaffold; if the student can't pick between two, they probably need vocabulary support, not eliciting. Speaking confidence builds from a low scaffold floor, not from being asked to perform above their current level.
How do I track tutor-talk-time across lessons?
Tutor-talk-time is impossible to fix if you don't measure it. The crude method is to record a 5-minute segment of a lesson and time how much of it is your voice versus the student's; the better method is to use a structured lesson recording with timestamps you can review. Either way, the first measurement is usually a shock.
The benchmarks worth aiming for:
| Lesson stage | Target tutor-talk-time |
|---|---|
| Lead-in / topic introduction | 40–50% |
| Presentation (new language) | 30–40% (eliciting-based) |
| Controlled practice | 20–30% |
| Freer practice / production | 10–20% |
Across a 60-minute lesson, the weighted average should land around 25–30%. Anything above 40% is a signal that you're over-talking; anything above 50% is a signal that you're lecturing.
Tuton's in-lesson classroom includes optional lesson recording (with student consent) that you can review afterwards to gauge talk-time without needing a separate tool — useful for the first 5–10 lessons with a new student when you're calibrating to their level. The recordings also let you spot eliciting moments you missed in real time and plan to deploy them next lesson. Tracking student progress across lessons includes tracking your own teaching patterns, not just the student's output.
How does Tuton help with eliciting and lesson structure?
Tuton's lesson library includes elicitation-first lesson templates with embedded CCQs and discovery-grammar sequences — pulling one in saves you the prep work of designing scaffolded prompts from scratch. The AI lesson-planning assistant can generate CCQs for any target structure you give it, at the CEFR level you specify, and surface them in the lesson plan automatically. Combined with the in-lesson classroom's recording feature, you get the input (better-designed elicitation prompts) and the feedback loop (post-lesson talk-time review). See all features or pricing.
Frequently asked questions
What's the difference between eliciting and prompting?
Eliciting draws language out of the student that they already possess receptively (the word is in their head, you're surfacing it); prompting nudges the student to produce language they're capable of but didn't initially attempt. In practice the two overlap heavily — a good eliciting question often functions as a prompt — and both contrast with "telling," which is the opposite of either.
How many CCQs should I ask per new language item?
Two to four — enough to check the key meaning components without turning the lesson into an interrogation. The minimum is one for each distinguishing feature of the target language; the maximum is the point where the student starts feeling tested rather than guided. If a student is answering all your CCQs correctly, you can move on; if they're answering them all wrong, you need to re-present the language, not ask more CCQs.
How do I elicit from a student who's a fast speaker — they fill the silence themselves?
Lean into it. Fast speakers are usually generating more language than slow speakers; your job is to refine and extend rather than create silence. Use "tell me more about that" prompts, focus on collocation and register upgrades, and save the silence-extension techniques for the slower, more reflective students.
Can I elicit grammar from absolute beginners (A1)?
Partially. At A1, the student lacks the metalinguistic vocabulary to discuss grammar, so eliciting the rule itself doesn't work — but eliciting the pattern through examples does. Show three positive sentences with the present continuous, then ask the student to produce a fourth following the pattern. The grammar is internalised inductively, even if it can't be articulated.
What if I elicit and the student gives the wrong answer?
Treat the wrong answer as useful data. "Interesting — let me show you" is a non-judgmental way to pivot. Then either present the correct answer and move on, or re-scaffold with a smaller prompt to see if the student can self-correct. The wrong answer tells you exactly where the gap is; you've learned something the "do you understand?" yes-nod would have hidden.
How do I avoid eliciting feeling like an interrogation?
Mix your prompts. Pure elicitation across a whole lesson feels like an exam. The trick is to deploy elicitation at the high-leverage moments — meaning checks, error self-correction, vocabulary recycling — and to talk normally (including telling and explaining) in the connecting moments. Eliciting is a tool, not a posture.