ESL/EFL support — common Nepali-learner mistakes

The patterns are not random

Linguistics has a name for what happens when you learn a second language on top of a first one — L1 transfer. The first language quietly tells you what is “natural” in the second, and where the two languages disagree, you make a predictable mistake. For Nepali learners of English, the disagreements are well-mapped:

• Articles. Nepali has no a / an / the. The student either drops articles entirely (“I went to market”) or sprinkles them randomly (“The Kathmandu is the capital”). The fix is targeted practice on the function of articles, not a generic grammar lecture.
• Prepositions. Nepali postpositions do not map one-to-one onto English prepositions. “I am studying in class nine” vs “at class nine”; “He is at home” vs “in home”; “I will reach office by five” vs “in five.” These have to be learned phrase by phrase. There is no underlying rule that resolves them.
• Present perfect vs simple past. This is the famous one. “I have gone yesterday” is wrong in English because yesterday fixes the action in completed past time, which forbids present perfect. In Nepali, the equivalent forms overlap — गएको छु and गएँ are not as sharply divided as have gone and went are. Students need to feel where the boundary actually lies.
• Question word order. In Nepali, intonation can turn a statement into a question without inverting subject and verb. “You are coming?” sounds fine to a Nepali ear. English requires “Are you coming?” — auxiliary inversion is the rule, not the option.
• Plural marking. Childrens, sheeps, informations, advices, furnitures. English irregular plurals and uncountable nouns do not match Nepali pluralisation patterns, and the natural extension of the regular -s rule produces these.
• Subject-verb agreement. “My friends is coming.” Nepali verb agreement is shaped by honorific level rather than number-of-subject in the same way English uses it. The result is third-person-singular -s errors that persist long into intermediate English.
• Tag questions. “You are going home, isn’t it?” The Nepali equivalent — होइन? or है? — is one form that works for everything. English tag questions agree with the verb in the main clause, which is fiddly and gets borrowed wrong.
• Phrasal verbs. “Open the light. Close the AC. Discuss about the problem. Comprise of three parts.” Phrasal verbs are non-transparent and have to be memorised; the patterns the student transfers from Nepali are rarely the English ones.

One pattern at a time

The most common mistake in teaching against these errors is to do all of them at once. The grammar lesson covers articles, prepositions, tenses, and agreement in a single forty-minute period, the students take notes, nothing transfers to the writing in their notebooks the next week. Generic grammar instruction does not move the needle because attention is spread too thin to change any actual habit.

The pattern that works is one mistake, one period, twenty examples, all from contexts the student recognises. You pick the present-perfect-with-yesterday error. You generate, with AI, twenty practice items where the student has to choose between I went and I have gone with adverbs that distinguish them. You drill those twenty items, mark them together, and the next time a student writes “I have gone yesterday” in a free-write, you point and say “yesterday — which tense?” — and they fix it themselves. This is what changing a habit looks like. Twenty examples on the one pattern, several times across a term, with the teacher pointing every time the pattern reappears in the wild.

The AI prompt: “Generate twenty short practice sentences for Class 9 Nepali learners of English on the difference between simple past (I went) and present perfect (I have gone). Each item should have a blank to fill in and should include a time adverb — yesterday, last week, since 2020, ever, never, just — that determines which tense is correct. Use contexts a Nepali student would recognise — school, family, festivals, cricket, daily life. After the items, give the answer key with a one-line explanation of which adverb forced which tense.” You get a printable worksheet in under two minutes.

The minimal-pair drill

The single highest-leverage format for pattern-specific practice is the minimal pair — two near-identical sentences, one correct and one wrong, and the student spots the difference. Minimal pairs force the learner’s attention onto exactly the contrast that matters and nothing else.

The AI prompt: “Give me fifteen minimal pairs for Nepali Class 8 students learning English articles. Each pair has one correct sentence and one with an article error — missing, extra, or wrong (a vs the). Both sentences should be otherwise identical. After the fifteen pairs, give the answer key with a one-line reason for each.” Examples:

• “I am going to school.” / “I am going to the school.” (zero article for institution as institution; the article would mean a specific physical building)
• “She bought a sari.” / “She bought the sari.” (first mention takes a; the would require previous mention)
• “My father is a teacher.” / “My father is teacher.” (occupations take a/an in English)

Twenty minutes on fifteen minimal pairs does more for article use than two weeks of explanation. The learner is not memorising a rule; they are training a pattern-detector. After enough exposure, articles start to feel right or wrong rather than requiring conscious decision, which is the only stable end state for this kind of error.

Mother-tongue-aware corrections

When a student makes one of these patterned mistakes, the most useful correction is not “this is wrong” — it is “this is wrong, and here is why your Nepali ear told you it was right.” The student then understands the mistake as a difference between two languages they both know something about, not as a mysterious error they cannot predict.

AI is good at producing these explanations in Nepali. Prompt: “A student wrote ‘I have gone to market yesterday.’ Explain in plain Nepali why this is wrong in English. Reference how the same idea is expressed in Nepali (गएको छु vs गएँ), and why English does not allow present perfect with yesterday. Keep it to four sentences, in language a Class 9 student understands.” The output is a short, targeted Nepali explanation the student can read in thirty seconds. The next time they write I have gone yesterday they catch it themselves because they remember the explanation, not just the correction.

This is one of the cleanest demonstrations of why AI is genuinely useful for Nepali ESL teaching specifically. A monolingual English explanation of the same error rarely transfers. The bilingual explanation — naming both the English rule and the Nepali habit that led to the mistake — sticks. Producing such explanations for forty students with no AI is impractical. Producing them with AI takes seconds per student.

Why specific-pattern practice beats generic drilling

A useful way to think about all of this. The student’s English ability is the sum of many small, separately-trained habits. Each Nepali-learner pattern is a habit that needs to be retrained. The retraining requires time on the specific habit — the time the student’s brain spends actually noticing the pattern, choosing correctly, getting feedback. Generic grammar lessons spend forty minutes on twelve different patterns and give each pattern a few seconds of attention. Targeted lessons spend forty minutes on one pattern with twenty examples, and give that pattern forty minutes of attention. The math is not subtle.

The teachers who get the best results from AI for ESL are the ones who keep this in front of them. Pick one pattern. Generate the drill. Run the drill. Track which students still make the mistake. Drill again next week. Move to the next pattern only when the first one has substantially settled. Boring, methodical, and exactly the kind of work that AI has just made cheap enough to actually do.

Across a year, eight to twelve patterns can each be drilled four or five times, with practice spaced out to take advantage of how memory consolidates. That is roughly fifty focused drills per year. The student’s English does not improve through inspiration; it improves through fifty small fixes to fifty real habits, each one targeted at a real Nepali-to-English transfer error their language background made predictable. The materials for those fifty drills, in 2026, take less than a school week to generate. The teaching is still you.

Check your understanding

Quick check

—

True or false — the recurring English mistakes Nepali learners make (article confusion, present-perfect-with-yesterday, isn't it tag questions, childrens-style plurals) are largely random individual errors with no underlying pattern AI can usefully teach against.

1. ATrue — every student is different, and grammar mistakes do not follow predictable patterns; AI can only help with vocabulary.
2. BFalse — these mistakes are well-documented L1-transfer patterns rooted in specific differences between Nepali and English (no articles in Nepali, postpositions that do not map to English prepositions, overlapping past tenses, intonation-based questions, different pluralisation), and they are exactly the kind of focused, contextualised drilling AI is unusually good at producing.
3. CTrue — they are individual mistakes that only a one-on-one tutor can address.
4. DFalse — but they are too complex for AI to handle and require Cambridge-published textbooks instead.

Check answer→

What comes next

You have now seen, across the chapter, how AI changes language work in the Nepali classroom — curriculum translation, scaffolding for non-English-medium students, targeted ESL pattern practice. The work the teacher does most often, though, may not be in the classroom at all — it is the parent meeting, the SMS to the parent who did not show up, the admin form that has to go to the municipality. The next chapter, Parent communication and admin, is about how AI helps with everything that happens around the lesson.