Classrooms in the United States are more linguistically diverse than at any point in recent history. Approximately 10 percent of public school students are classified as English learners, and in many urban districts the figure is significantly higher. These students are not a monolithic group. They span a wide range of English proficiency levels, first-language literacy backgrounds, and academic preparation. What they share is the challenge of demonstrating their thinking in a language they are still developing. For writing teachers, this creates a grading dilemma: how do you assess argument quality, evidence use, and analytical thinking when the vehicle for that thinking, the English language, is still under construction?

The most common mistake is conflating language proficiency with thinking quality. A student who writes 'the author use emotional word to make the reader feel sympathy for the immigrant experience' has identified a rhetorical strategy, connected it to an effect on the reader, and named a specific thematic focus. The grammar is nonstandard. The analysis is solid. A rubric that penalizes surface-level language errors without distinguishing them from analytical errors will score this student lower than their thinking deserves. That is not just unfair. It is bad assessment practice, because the rubric is no longer measuring what it claims to measure.

The fix is rubric design. Effective rubrics for linguistically diverse classrooms separate what is being assessed (argument quality, evidence use, analytical depth) from how students express it (sentence structure, grammar, vocabulary). This does not mean ignoring language conventions. It means scoring them in their own dimension rather than letting them contaminate the score on dimensions that are supposed to measure thinking. A student can earn a high score on 'evidence and reasoning' while earning a lower score on 'language conventions,' and the resulting feedback is both more accurate and more actionable: the student knows their argument is strong and their grammar needs work, rather than receiving a single middling score that tells them nothing specific.

Differentiated instruction researchers emphasize that effective rubrics for multilingual learners maintain the same content expectations while adjusting the language-based expectations and scaffolding. The standard stays the same: all students must develop a defensible claim supported by evidence. The way students demonstrate that standard can look different depending on their language proficiency level. Sentence frames, graphic organizers, and first-language brainstorming are not accommodations that lower the bar. They are scaffolds that help students clear it.

Designing Rubrics That Separate Content From Language

The practical step is straightforward. If your rubric currently has a single dimension called 'writing quality' or 'expression and conventions' that blends analytical clarity with grammatical accuracy, split it into two separate dimensions. One evaluates the clarity and sophistication of the student's ideas and how well they organize their argument. The other evaluates language conventions: grammar, sentence variety, and academic vocabulary. Score them independently. This structural change benefits all students, not just multilingual learners, because it provides more specific diagnostic information. But it particularly benefits students whose thinking outpaces their English proficiency, which is most multilingual students in academic settings.

• Separate rubric dimensions for content/analysis and for language conventions so that surface-level errors do not suppress scores on analytical thinking.
• Write descriptors in clear, jargon-free language that multilingual students can understand; if a student cannot interpret the rubric, they cannot use it to improve.
• Provide sentence frames and graphic organizers as scaffolds for the writing process, not as simplifications of the assignment itself.
• Focus feedback on one or two language patterns rather than marking every error; a student who receives feedback on comma usage can address it, while a student whose paper is covered in red marks cannot prioritize.
• Use AI grading tools cautiously with multilingual student writing; verify that the tool does not disproportionately flag language-development features as errors that lower content scores.

AI Grading and Multilingual Students: A Caution

AI grading tools trained primarily on native-English student writing can struggle with the writing patterns common among multilingual students. Research has documented significant disparities in how AI detection tools score writing by non-native English speakers, and similar patterns can emerge in AI grading. A tool that penalizes unconventional syntax or limited vocabulary range will systematically under-score multilingual students on dimensions that are supposed to evaluate their thinking, not their English fluency. Before using AI grading with multilingual student populations, test the tool specifically on a sample of writing from your multilingual students and compare the AI scores against your own judgment.

When the tool is calibrated well and paired with a rubric that separates content from language, AI grading can actually benefit multilingual students. The consistency of AI scoring means that a multilingual student's essay is evaluated against the same criteria at the same level of attention as every other student's essay, without the implicit bias that can creep in when a human scorer is influenced by surface-level language issues. The key is ensuring that the rubric the AI applies is designed for equity, not just efficiency.

Building a Feedback Practice That Supports Language Growth

Feedback for multilingual writers should follow the same principle of focus that works for all students: prioritize one or two language patterns per assignment rather than marking everything. A student at an intermediate English proficiency level who receives twenty grammar corrections on a single essay cannot process all of them. But a student who receives feedback focused specifically on subject-verb agreement, with two or three examples from their own writing and a clear explanation of the pattern, can make meaningful progress on that specific skill. The next essay can target a different language pattern. Over time, the student builds proficiency one pattern at a time, which is how language development actually works.

The students in your classroom who are writing in their second or third language are doing something intellectually demanding that many monolingual adults have never attempted. They deserve assessment that recognizes the strength of their thinking while supporting the development of their English proficiency. That requires rubrics designed for equity, feedback focused on growth, and tools calibrated to avoid systematic bias. None of that is complicated. It just requires intentional design.