Six example sentences appear on a worksheet, followed by "What do you notice?" The strongest learner spots a pattern, two classmates copy the answer, and three others invent rules that fit only the first line. Discovery has become a guessing contest. For an English teacher working with a small A2 class, examples can lead to durable grammatical knowledge when the comparison is guided, the evidence contains a meaningful contrast, and the final rule receives an answer check.
Discovery needs an object
Guided induction starts with language data and moves learners toward a generalisation. The teacher may ask learners to sort examples, compare two meanings, mark a changing form, or predict the next item. That differs from leaving a class alone with a page of sentences. Guidance tells learners what relation to inspect while preserving the intellectual work of forming the rule. Consider four lines:
- Mei has worked here for six months.
- Mei worked here for six months in 2024.
- I have known Leo since March.
- I knew Leo at university.
"What do you notice?" permits dozens of answers. Better prompts narrow the search: Which actions connect to now? Which sentence gives a finished time? Underline the time phrase, then match each sentence to a timeline. The learner still constructs a generalisation, though the task points at tense, time reference, and current relevance.
The contrast must carry meaning.
Before class, the teacher can test the design by asking a colleague to solve the examples without the intended rule in view. If the colleague reaches three equally plausible generalisations, the data need a sharper contrast. If the answer appears from typography alone, perhaps because every target form is bold and every distractor is grey, learners are matching visual signals instead of relating form to meaning. A useful set makes the relevant distinction recoverable from the sentences, while one carefully chosen counterexample prevents a shortcut from working. This ten-minute rehearsal often improves the task more than adding another page of practice.
Evidence for guided induction
Carrie Haight, Carol Herron, and Steven Cole studied 47 second-semester college learners of French. Each student learned eight grammatical structures, four through deductive teaching and four through guided induction. Immediate quizzes favoured guided induction, and the delayed results showed a strong trend in the same direction. The 2007 Foreign Language Annals study used a within-subjects design, which helped compare treatments inside one cohort.
Its boundary matters. These were beginning college learners of French, the teachers covered eight selected structures, and the outcomes came from the study's own tests. The result supports guided induction under those conditions. It does not establish that learners should discover every English rule or that a teacher explanation damages learning. A second experiment points the other way. Rosemary Erlam compared deductive and inductive instruction for French direct-object pronouns with 69 secondary learners. The deductive group received rules and metalinguistic information. The inductive treatment focused on form through language-use tasks without explicit grammar explanation. Deductive instruction produced an advantage on the study's comprehension and production measures. Erlam also warned that results depend strongly on what a test measures. The 2003 Modern Language Journal article gives teachers a reason to resist a method slogan.
These findings can coexist. The Haight treatment was guided: teacher and learners built the pattern through directed comparison. Erlam's inductive group received no explicit rule. The target structures, ages, materials, and tests also differed. "Examples first" names an order. It says little about the quality of the questions between the first example and the final rule.
A classroom comparison can preserve the useful parts of both approaches.
Give learners four minutes to test a guided rule, collect two competing versions, and then supply the standard description. The explanation arrives after learners have a concrete problem for it to solve, and nobody leaves with an unchecked private theory.
Build a useful contrast
Start with a decision the grammar changes. For articles, compare "I spoke to a manager" with "I spoke to the manager you emailed." For conditionals, ask whether the speaker treats a deadline as open, unlikely, or finished. For the present perfect, connect the form to a time window that includes now. Learners need a reason to inspect form beyond satisfying the teacher. I prefer six sharply chosen examples to 18 sentences that repeat the same surface pattern. Include a near neighbour that learners should reject. If every present-perfect sentence contains "since," a learner may infer that since mechanically causes the tense. Add "I lived there from 2018 to 2021" and ask why a completed interval changes the choice.
Prompts should move from observation to explanation:
- Circle the form that changes.
- Match each sentence to one of two timelines.
- State a rule that covers all six examples.
- Test the rule against a seventh sentence.
Four stages are enough. More prompts may turn discovery into a disguised lecture written as questions. Fewer may reward learners who already know the rule while leaving everyone else's misconception hidden.
Check the rule before practice
Ask pairs to write their rule in 17 words or fewer. Put two competing versions on screen without student names. The class tests both against a fresh example and identifies the version that predicts the choice. Only then does the teacher give a concise conventional rule and label, correcting any missing condition. This answer check protects accuracy. It also produces useful assessment evidence. A teacher can see whether the problem lies in form, time meaning, or an overgeneralised cue. That information should shape the next task: more interpretation when learners confuse the meanings, a short form drill when endings are unstable, or freer production when both are ready.
This sequence connects with grammar practice that begins with interpretation. Learners first decide what a sentence means, then connect that meaning to form.
Production comes after the contrast is visible.
Direct explanation remains sensible when time is tight, the form has a clean and memorable rule, or learners need a reference before an exam, whereas guided discovery earns its minutes when comparing examples reveals a choice that a one-line rule tends to flatten. A mixed approach is often the honest outcome of the evidence.
For Monday's lesson, replace one "What do you notice?" box with two timelines, six contrasting examples, and a seventh sentence designed to break the first rule learners are likely to propose.