Helping non-English-medium students with English material

Pre-teach the vocabulary, every chapter

The single highest-leverage move you can make for a non-English-medium student facing English material is to front-load the vocabulary. Before they open the chapter, they already know the ten or fifteen words that will otherwise stop them on every page. This is not a new idea — ESL pedagogy has known it for decades — but in a classroom of forty students with no AI, producing the vocabulary sheet every week is the kind of work that quietly stops happening.

The AI workflow is short. Paste the English chapter into the model and ask: “Extract the ten to fifteen most important content words from this chapter that a Class 8 Nepali-speaking student would not know. For each, give: the English word, the Nepali meaning using the NEB textbook term where one exists, a simple definition in Nepali, and one example sentence in simple English drawn from a context the student would recognise — school, home, or a Nepali festival. Format as a table.”

The output is a one-page vocabulary sheet you can photocopy and distribute the day before the chapter starts. The student arrives at the English text with the key words already in their head. The text stops being a wall and becomes a sequence of mostly-known words with occasional unknowns the student can guess from context. The cognitive load drops by half. The student starts to feel that English is something they can read, not something that is happening to them.

Bilingual reading guides — Nepali summary per paragraph

Once vocabulary is pre-taught, the next scaffold is a bilingual reading guide. Take the English chapter and produce a one-page companion document with the same paragraph structure — but each paragraph gets a one-line Nepali summary alongside. The student reads the English paragraph, gets stuck or unsure, glances at the Nepali line to confirm the gist, returns to the English with comprehension intact, and keeps going.

The prompt: “For this English chapter, produce a reading guide. Number each paragraph 1, 2, 3 and so on. For each paragraph, write one Nepali sentence that captures the main point a student needs to take away. Keep the Nepali sentence short — fifteen words or fewer. Do not translate the paragraph; summarise the point.” The output is intentionally not a translation. Translation would replace the English text in the student’s attention. Summary is a check the student can glance at without ever leaving the English page.

This is the workflow that converts a chapter from “I cannot read this” to “I can read this, slowly.” Over a term, the student builds reading stamina in English. Over a year, you start removing the Nepali column for the second half of chapters, then the third quarter, then only for the hardest paragraphs. The scaffold shrinks as the student grows. That shrinking is the whole point.

Explain-it-back — comprehension in the strong language

The third scaffold is a routine, not a document. After a student reads a section in English, you ask them to explain it back to you in Nepali. Not translate it — explain it. Tell me what this is about. Why does it matter. What is the example. The student is allowed to use any language they like. You are checking comprehension, not English production.

This routine is good pedagogy on its own. AI makes it scalable to a class of forty by generating the prompts. “For the photosynthesis chapter we just read, give me eight ‘explain it back’ questions in Nepali that I can ask students one at a time. The questions should check whether the student understood the concept, not whether they remember the English vocabulary. Vary the difficulty — start easy, get harder.” You print the questions, fold the paper into eight strips, and the next class period is a structured Nepali-language comprehension check on English-language reading. Forty students, eight rounds, every student speaks at least once.

The reason this routine matters more than it sounds is that it separates two skills that English-medium schooling otherwise conflates: understanding the science, and producing English sentences about the science. A weak English speaker who has fully understood photosynthesis deserves credit for that understanding. The routine lets you see it. Without the routine, the student looks like they have not learned the science when really they have only not learned the English.

The honest limit — scaffolds do not build English

It is tempting, especially in a Nepal-context conversation where AI suddenly makes Nepali-English bilingual materials almost free, to oversell what these scaffolds do. They do not build English proficiency. They let a student with limited English access English-language content. Those are different things, and conflating them produces a Class 10 student who has technically passed every exam by relying on the Nepali column and cannot read a paragraph of English without it.

Building English is a separate workstream. It needs hours of input — graded readers, listening practice, conversation, writing with feedback. AI helps there too — that is what the ESL section, two sections from now, is about — but it is its own work. The scaffolds in this section let the student keep up with science, math, and social studies while the English work happens. They do not substitute for the English work.

Be honest with the students about this distinction. “This sheet is here to help you read the chapter. It is not how you will learn English. We will work on English itself in the other class periods, in your reader, in conversation. This sheet is so the rest of school does not stop while your English grows.” The student understands the division of labour. The parents do too. Everyone is more honest about what is happening.

The over-reliance trap, and how to step out of it

The single failure mode of this whole approach is scaffolds that never come off. The vocabulary sheet that started life as a leg-up becomes a permanent prosthetic. The Nepali summary column that helped the student through a hard chapter becomes the only thing the student reads — the English column is on the page, but their eyes go straight to the Nepali. After a year, the student has built no English reading stamina at all. They have just built familiarity with your Nepali summaries.

The escape from this trap is your judgement, and it has to be deliberate. Plan, at the start of the year, how the scaffolds will shrink. “For chapters one through four, full vocabulary sheet and full bilingual guide. For chapters five through eight, vocabulary sheet but reading guide only for the hardest two paragraphs. For chapters nine through twelve, vocabulary sheet only — the reading guide goes away. For the SEE prep chapters, no scaffolds.” You do not have to follow the plan exactly — some students need more, some less — but having a plan changes the default. The default with no plan is that everything stays forever.

A useful indicator: when a student stops glancing at the Nepali column to check, and instead just reads it instead of the English, the scaffold has gone too far. Pull it. Have a one-minute conversation. “You have been reading the Nepali side. From next week, the sheet only has the English. The vocabulary list stays. You can do this.” They almost always can.

This whole section is, in the end, a single instruction repeated three ways: the scaffolds are excellent, and they are temporary. The teacher who builds great scaffolds and never removes them has done less good than the teacher who builds adequate scaffolds and removes them on schedule. AI made the building cheap. The removing is still your call, and it is the harder part of the job.

Check your understanding

Quick check

—

Two teachers use AI to support Class 8 community-school students with the English Science textbook. Teacher A produces vocabulary sheets, bilingual reading guides, and explain-it-back questions for every chapter all year, every chapter, in the same depth. Teacher B starts with full scaffolds for the first chapters, then deliberately reduces the Nepali column for later chapters, removes the reading guide entirely by mid-year, and keeps only the vocabulary list for the final term. Which workflow more closely matches the approach this section recommends, and why?

1. ATeacher A — more support is always better; the scaffolds make every chapter accessible all year.
2. BTeacher B — the scaffolds are designed to be temporary and to shrink as the student’s English grows, so deliberately removing them on a planned schedule is the whole point; full-strength scaffolds all year build comfort with the scaffold rather than reading stamina in English.
3. CBoth are equally good — the scaffolds are the support, and removing them is just optional.
4. DNeither — the right approach is to give students only the English text with no scaffolds at all.

Check answer→

What comes next

The scaffolding workflow lets a Nepali-comfortable student access English content while their English grows. The English itself grows through separate work, and one of the most efficient parts of that work is targeting the specific mistakes Nepali learners make in English — the article confusion, the preposition misuse, the present-perfect tangles that are not random at all but deeply patterned. The next section is about using AI to teach against those patterns directly.