category: Differentiation
---
Most teachers know the ELL trap: you're trying to support a student's English while teaching them actual content, and every simplification feels like it's dumbing down the idea. AI can help with this paradox. It can generate language scaffolds—sentence frames, vocabulary support, syntax models—without watering down the cognitive demand.
The problem with "simplified" versions
When you simplify a science lesson for ELLs, you often strip out the precise language AND the rigor. The student learns a diluted version of the concept, and then when they encounter the full text or assessment, it's a shock. The vocabulary changes. The grammatical complexity jumps. They have to learn the concept again.
What you actually need is the same rigorous content with language supports built in. That's what AI can scaffold for you.
The sentence-frame move
One teacher I know works with high-school physics. She has newcomers in class who have strong math skills but limited English. Here's what she does:
1. She takes the physics concept and the text she'll teach from
2. She asks an AI: "Create 5-7 sentence frames that help a newcomer express understanding of [concept]. Use academic language, but give them structure. Include one frame for explaining, one for comparing, one for disagreeing."
3. She gets back frames like:
- "The force of __ causes the object to _. This is similar to _ because __."
- "I disagree with __ because the evidence shows __."
4. She projects these during class. The ELL student uses them to build full sentences, but the thinking is grade-level.
Two layers of scaffolding
The work needs two scaffolds running at the same time:
- Language scaffold (the frame, the vocabulary list, the syntax model)
- Conceptual scaffold (breaking the idea into sequence, giving examples, connecting to prior knowledge)
AI is phenomenal at the language layer. You still do the conceptual scaffolding—you know your students and what they already understand. But when you ask the AI: "Here's what we're learning. Give me the English structures a newcomer would need to express this," you get real support.
What NOT to do
- Don't ask an AI to simplify the text. You'll get baby versions.
- Don't expect it to know your students' L1 (first language). If you have Spanish speakers, you CAN ask it to generate Spanish sentence frames too, but that's explicit.
- Don't skip the reading/checking. Generated frames sometimes miss important nuance.
The follow-up
After your ELL student uses these frames to participate, your job is to gradually fade them. By mid-unit, they should be moving toward full sentences. By end of unit, just the vocabulary list. By next unit, independent.
The frame doesn't cage them—it launches them.
One more move: translation as a learning tool
This is different from giving them translated worksheets. Instead:
- They learn in English (with frames)
- You give them 2-3 key sentences from the lesson in both English and their L1
- They find the differences in how the languages handle the same idea
This actually deepens their English learning AND their conceptual understanding. They see that "photosynthesis" is a specific term in English, but their L1 might describe the process more literally.
Ask an AI: "Here are the key sentences from today's lesson [paste]. Translate into Spanish and align them line-by-line with the English. Highlight words that don't have direct equivalents."
You get something that teaches language AND meaning.
Real talk
This doesn't work for every ELL in the same way. A newcomer with no literacy in their home language needs different support than a student who reads well in Spanish but is developing English. The frames help; your knowledge of what each kid needs does the heavy lifting.
But with this move, you're not choosing between rigor and support. You're layering them.
