Survey courses in world literature often place Lorca's Collected Poems alongside works from several other traditions, which means professors have only a week or two to cover him. Enrollment may reach well over a hundred students, and teaching assistants vary in experience. Assigning a poetry writing task in this setting is pedagogically valuable but logistically demanding. A clear structure for the assignment and the grading keeps the course manageable without reducing it to multiple choice.

The first design decision is length. In a large class, a focused response of 600 to 800 words on a single poem or a pair of poems is far easier to grade fairly than a long research essay. Shorter assignments also let students write more of them, which builds skill faster than a single major paper. Specific prompts that name the poem and the skill, such as analyzing how imagery builds mood, reduce the variation graders must handle.

Second, think about how the course staff will apply the rubric. A rubric that works for the professor may be applied inconsistently by graduate teaching assistants unless it includes clear descriptors and sample responses. Providing annotated examples at each performance level reduces disagreement and speeds up training. A short calibration meeting before grading begins pays for itself many times over.

Building a Rubric for Scale

A rubric for a large course should be compact, perhaps four criteria with three or four levels each, and written in plain language. Each descriptor should describe observable features of the writing rather than impressions. For example, instead of saying the analysis is insightful, say that the analysis explains how a specific image affects the poem's tone. Observable descriptors let different graders reach similar scores independently.

• Thesis: makes a specific, arguable claim about the poem.
• Evidence: uses short quotations that are directly relevant.
• Analysis: explains the effect of language, image, or form.
• Writing: organizes ideas clearly with minimal distracting errors.
• Context: uses background information only where it supports the reading.

Calibrating Graders

Calibration begins with a norming session in which everyone grades the same three or four papers independently and then compares results. Disagreements are not problems but opportunities, because they reveal ambiguities in the rubric. Adjust the descriptors until graders converge, and record the decisions in a short guide. Revisit calibration midway through the grading period to prevent drift.

Spot checks are another useful safeguard. The professor can review a random sample of graded papers from each assistant, looking for scores that appear unusually high or low. This is not about distrust but about ensuring that students in different sections are treated equally. Sharing the findings with the team improves both fairness and morale.

Where Technology Fits

Large classes are where AI-assisted tools can make the biggest difference. A platform like GraideMind can generate rubric-aligned first-pass feedback on every submission, flagging common issues and offering suggested comments that graders can accept, edit, or reject. This reduces the time spent on repetitive notes and lets staff focus on interpretation and individual cases. Consistent first-pass feedback also supports uniformity across sections.

Institutions should decide in advance how such tools are used and communicated. Many universities require disclosure to students and some have policies about grading automation. A brief statement in the syllabus describing that staff review all scores and that software supports feedback preparation helps maintain trust and meet institutional expectations.

Keeping Feedback Meaningful at Scale

Even in large classes, students benefit from feedback that identifies one or two priorities for improvement. A short summary comment at the end of each paper naming the most important issue and one concrete strategy is more useful than scattered marginal notes. Encourage staff to quote a phrase from the student's own writing in that summary, which signals that the paper was actually read.

Finally, use the grading period to collect information about the class as a whole. Note the most common errors and share them in a lecture or discussion section. Addressing patterns publicly reduces the need to repeat comments hundreds of times and helps students see that their challenges are shared. Effective large-class writing instruction depends on turning grading data into teaching.