Carter's collection often appears in introductory literature courses, women's studies seminars, and surveys of twentieth-century fiction, where enrollments can reach two hundred students or more. A professor with a single teaching assistant may face a stack of several hundred essays at the same moment, each one asking for thoughtful engagement with dense, allusive prose. Without a deliberate system, the quality of feedback will decline sharply as the stack shrinks.

The first step is to narrow the assignment so that essays are comparable. A prompt that lets students write about any story in any way produces hundreds of unrelated arguments that are slow to evaluate. A prompt that asks everyone to analyze how one story, such as The Werewolf or The Snow Child, revises its source tale makes it far easier to recognize strong and weak work quickly.

Teaching assistants need calibration, and the time to do it is easy to underestimate. Before grading begins, the professor and the TAs should score a shared set of five or six sample essays independently, then discuss where scores diverge and why. Those conversations typically reveal that two graders are interpreting a rubric row in different ways, and fixing that early prevents hundreds of inconsistent scores later.

Structuring the rubric for scale

A rubric used by multiple graders must be more explicit than one used by a single instructor. Each performance level should be defined with concrete language and, ideally, an anchor example drawn from a real student essay. This reduces the interpretive latitude that leads to different scores for similar work, and it makes appeals easier to resolve because the standard is documented.

• Write descriptors that name observable features of the essay
• Include a short anchor excerpt for each score level in the thesis and analysis rows
• Run a calibration round with every grader before live scoring begins
• Check a random sample of each TA's scores against your own midway through
• Keep a log of rulings on edge cases so decisions stay consistent

Where AI-assisted grading helps most

In a large course, AI feedback is most useful as a first-pass layer that applies the rubric to every essay the same way. It can note whether each paper contains an arguable thesis, how much of the evidence comes from the assigned story, and whether the writing has major structural problems. Those observations let a TA move through the stack faster and focus attention on the essays that need judgment.

Professors should still decide how much of the final score depends on human review. Many programs treat AI output as advisory, with the TA confirming or adjusting each score and adding at least one personal comment. This arrangement preserves accountability, which matters when students dispute grades or when the department is audited.

Protecting the quality of feedback

Large courses risk turning feedback into a numerical exercise, and students notice quickly when the comments are generic. One remedy is to require every graded essay to carry a specific next step tied to the rubric, such as "develop your claim about the Marquis's collection of objects by explaining what it reveals about his view of women." A concrete instruction like that costs seconds to write and can transform how a student revises.

Finally, build in a short revision opportunity if the schedule allows. Students who apply feedback to a second draft learn far more than those who simply receive a grade, and the revised essays are often easier to grade because the major problems have already been addressed. The combination of consistent scoring and real revision is what makes large-class literature teaching sustainable.