What makes small-group instruction actually accelerate multilingual learners? This research-backed piece breaks down the science — from the Input–Output Loop to sentence-level instruction — and explains when small groups work, when they don't, and what to do differently.
“Small groups do not accelerate learning because they are small. They accelerate
learning because of what happens inside them.”
In Part I, we confronted an uncomfortable reality: small groups can either accelerate multilingual learners or quietly reduce their access to language and knowledge.
Part II moves us from structure to substance.
Because the real question is not whether we schedule small groups.
The real question is whether our small groups are engineered for acceleration.
In professional learning conversations, the debate often becomes polarized:
Both statements can be true.
The difference lies in design.
When small groups are layered on top of weak whole-class instruction, they become compensatory structures. When they are built from strong whole-class instruction, they become precision tools (Duke & Goldenberg, 2023).
Duke and Goldenberg (2023) argue that small-group instruction is most effective when it complements—not replaces—explicit, high-quality whole-class teaching. Likewise, Schmoker and Shanahan (2023) caution that poorly designed small groups can dilute instructional time and reduce access to rigorous content.
Small groups are expensive. Every minute a teacher works with one group is a minute other students are not receiving direct teacher-led instruction. For multilingual learners—who require both literacy and language development simultaneously—those minutes are not neutral.
The purpose of small groups must justify their cost.
Most educators agree with the phrase, “All teachers are teachers of language.” But agreement does not equal preparation. Small groups have a critical place here!
As Huynh and Skelton (2025) remind us, unless multilingual learners are explicitly taught to write and speak in academic registers, they default to social language in academic settings. This default is not a deficit; it is developmentally predictable.
The problem is not that students cannot think.
The problem is that they cannot yet express complex thinking in academic language.
Huynh and Skelton (2025) argue that academic growth requires intentional instruction in the sentence structures tied to disciplinary thinking. Without explicit attention to language form, multilingual learners remain dependent on surface-level scaffolds.
Small groups, when designed intentionally, create the protected space where that expressive gap can be closed.
By Grade 3, comprehension ranges in a single classroom can span nine grade levels; by Grades 4 and 5, that spread can exceed eleven (Firmender et al., 2013). This is not a classroom management issue. It is a structural instructional reality.
Whole-group instruction creates efficiency. It ensures every student receives exposure to grade-level vocabulary, discourse, and content knowledge.
Small groups create precision.
They allow teachers to narrow in on a single leverage point:
Acceleration occurs when precision increases learning density (Duke & Goldenberg, 2023).
Small groups should exist only for what cannot be fully addressed in whole-group instruction.
Small groups exist for three core reasons:
In early decoding, students benefit from rapid cycles of correct practice paired with immediate feedback. That density of interaction is difficult to achieve in whole-group settings.
Likewise, multilingual learners cannot refine articulation, syntax, or morphological awareness without someone responding to them in real time. Small groups work when they tighten the feedback loop (Huynh, 2025).
There is one question that prevents small groups from drifting into routine:
Who needs what next?
If the answer is “no one,” then whole-class instruction is more efficient.
If the answer is specific—“these students need support distinguishing /v/ and /f/,” or “this group needs explicit instruction in causal sentence structures”—then small groups are warranted.
When small groups become a daily habit rather than a strategic response, they drift from acceleration into scheduling tradition (Schmoker & Shanahan, 2023).
But identifying who needs support is only the beginning.
The more consequential question is how that support will be delivered once the group is formed.
Too often, small groups are justified by data but weakened by design. We identify the gap, assemble the students, and then revert to explanation. The result is a smaller version of whole-group instruction rather than a more precise one.
If small groups are to accelerate multilingual learners, they must do something qualitatively different from whole-class teaching. They must shorten the distance between instruction and evidence of learning. They must increase the frequency of student language production. They must make thinking visible.
In other words, once we know who needs what next, we must design the instructional rhythm that will move learning forward.
That rhythm is captured below:
Exposure alone does not produce learning.
Content must be delivered in manageable chunks.
Then students must process it.
Processing requires students to speak, write, label, draw, and explain. Without output, language does not develop. Without language production, comprehension remains fragile (Huynh, 2025).
Huynh (2025) describes this rhythm as the Input–Output Loop: instruction must move in tight cycles of input followed by immediate student output. This model emphasizes formative responsiveness—teachers gather evidence of understanding before moving on.
In high-functioning small groups, instruction follows a precise rhythm:
Teach → Student attempts → Immediate feedback → Retry → Clarify → Extend.
Acceleration lives in that tightening loop.
Many classrooms provide word banks and sentence frames. These supports feel helpful.
Yet Huynh and Skelton (2025) caution that traditional scaffolds—when overused—can create dependency rather than independence. Students may complete frames accurately without internalizing the underlying sentence structure.
Acceleration requires something deeper: internalized language structures.
Embedded scaffolds shift the focus from filling in blanks to producing academic syntax aligned to thinking (Huynh & Skelton, 2025).
For example:
Huynh and Skelton (2025) emphasize that the verb in the prompt determines the syntax required. Their analysis of thinking verbs across disciplines demonstrates how command terms correspond to linguistic structures.
Small groups are where teachers can explicitly teach that alignment between command term and sentence structure.
Small groups should not simply reteach content.
They should teach the language structure that allows students to express content understanding.
Acceleration is built from specificity.
Acceleration is not speed.
It is density of meaningful language practice (Duke & Goldenberg, 2023).
The traditional I Do → We Do → You Do sequence often assumes language competence that multilingual learners are still developing.
Huynh and Skelton (2025) argue that multilingual learners benefit from embedded modeling, structured language prompts, oral rehearsal before written production, and collaborative language practice prior to independence.
Element
Traditional Gradual Release
Gradual Release for Multilingual Learners
Assumption
Students already have the language needed.
Academic language must be explicitly taught.
Modeling (I Do)
Teacher models the skill.
Teacher models both the skill AND the language.
Guided Practice (We Do)
Students practice with support.
Students rehearse using structured sentence stems and talk.
Student Talk
Talk may occur but not always structured.
Structured speaking is required before independence.
Feedback
Feedback during or after practice.
Immediate correction on language, clarity, and structure.
Independence (You Do)
Students complete the task alone.
Students use academic language independently without frames.
Whole-group instruction introduces structures. Small groups ensure mastery.
Acceleration is measured not by how much support we provide—but by how effectively students internalize and generalize skills and strategies.
Even well-intentioned small groups can stall growth.
They undermine learning when:
The greatest risk is not small groups themselves.
It is small groups without cognitive demand.
Acceleration requires more teacher-led precision for the students who need it most—not less.
Summary: What Accelerates vs. Undermines Multilingual Learners in Small Groups
Small groups accelerate multilingual learners when they:
They undermine learning when they:
Acceleration is not a scheduling decision. It is a linguistic one.
Duke, N. K., & Goldenberg, C. (2023). Yes, small-group reading instruction works—but use it wisely. Education Week.
Huynh, T. (2025, June 11). The input–output loop. TanKHuynh.com. https://tankhuynh.com/164-the-input-output-loop/
Huynh, T., & Skelton, B. (2025). Sentence-level scaffolds that foster English learners’ independence and growth. Edutopia.
Larson, A. (2026). Small groups that work: Part II draft manuscript. Unpublished manuscript.
Schmoker, M., & Shanahan, T. (2023). Small-group reading instruction is not as effective as you think. Education Week.