📄 The impact of flipped classroom model on teaching English grammar at 10th grade of high school

Summary

This Iranian study asks whether the flipped classroom can help with a chronic problem in Iranian school English: very little weekly class time, which is often spent on teacher-led grammar explanation. In a flipped design, the explanation is recorded as video and watched at home, so class time goes to practice, group work and higher-order tasks. In autumn 2016 (Aban–Azar 1395), one experienced teacher taught the same grammar unit from the 10th-grade English textbook to two comparable classes of girls in Tehran's District 4. One class was taught conventionally and the other was flipped. Evidence came from three sources: a researcher-made multiple-choice test analysed in SPSS, a validated learner questionnaire, and a semi-structured interview with the teacher. Test scores did not differ significantly between the groups. Students and teacher nonetheless reported higher satisfaction and motivation, better use of time, and less repetitive drilling. The author concludes that flipping is a feasible way to recover class time, provided principals and parents are briefed and supportive.

Key points

• The design was quasi-experimental: one teacher, two same-level 10th-grade girls' classes in Tehran, one identical grammar unit, and conventional versus flipped delivery. The footnotes point to will vs. be going to as the grammar topic.
• Data were triangulated at three levels: (a) a multiple-choice achievement test, (b) a learner feedback questionnaire, and (c) the teacher's expert view gathered through a semi-structured interview.
• The SPSS analysis found no significant test-score difference between the experimental and control groups.
• The qualitative levels showed gains in satisfaction, motivation, time management and avoidance of tedious repetition.
• Context and prior work:
  • English has had one two-hour weekly session since the 6-3-3 reform.
  • The Prospect textbook series began in 2012 (1391).
  • Flipped teaching traces back to Bergmann & Sams (Colorado, 2006) and to Mazur's peer instruction.
  • Flumerfelt & Green (2013) report a failure rate falling from 52% to 19% in a low-achieving school.

Interesting ideas

• I1. The study measures one intervention through three independent lenses (an objective test, learner self-report and an expert interview), so a null score result still shows where any effect actually lies.
• I2. The paper reports a null result on the main metric alongside process benefits rather than hiding it, which keeps claims honest.
• I3. Holding the teacher, the material and the level constant, and varying only the delivery mode, isolates the effect of the one design choice.
• I4. The appendix test uses grammatical distractors that differ minimally from the correct form, such as "does he starts" or "didn't played", to probe specific grammar knowledge.
• I5. Flipping moves passive content intake out of scarce time so that limited interactive time is spent on harder, higher-order work.

For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

• I1 (learn): Evaluate a parser or prag change with three separate signals: benchmark score, witness-parser agreement shifts, and an LLM-teacher qualitative review. This separates "no accuracy gain" from "a cheaper or cleaner pipeline".
• I2 (learn): Log non-significant LAS or answer-accuracy changes explicitly next to cost and repair-rule metrics. Pair each with a negative control, so that "no harm, saves effort" is a documented outcome.
• I3 (learn): Run ablations where only one stage, such as the perceptron parser versus gparse, changes while the corpus and frames remain fixed. This attributes downstream answer differences cleanly.
• I4 (tokens): Such minimal-pair multiple-choice items make a cheap diagnostic set for the en tagger and morphology. They test whether agreement and tense features are assigned correctly rather than guessed.
• I5 (answer): School-course readers could precompute static knowledge (word senses, verb classes) offline. Query-time effort then goes to reasoning over the world of the text for the actual question.