A survey course that includes Till We Have Faces might enroll anywhere from sixty to three hundred students, and each of them owes the professor an essay on the same book. Grading that many papers on one text sounds repetitive, but it is actually an opportunity, since the shared subject makes it possible to build tight, reusable standards. The risk is drift, where papers read late in the pile are judged differently from those read early.

Start by writing down what a B paper looks like for this specific assignment, because the B is the grade that anchors everything else. For a prompt about Orual's narration, a B paper might make a clear claim, support it with three or four accurate quotations, and explain most of them. Describing that paper in concrete terms gives every grader the same picture of the middle of the class.

With the B defined, the A and C become easier to describe as departures from it. An A paper takes a more original or more rigorously supported position, and a C paper tends to summarize or to explain too few of its quotations. Teaching assistants who share this vocabulary can work independently and still produce grades that agree within a half letter.

Running a Calibration Session Before Grading Starts

Before anyone grades a full stack, have the whole team read and score the same five essays, then compare results. The disagreements are where the real learning happens, because they reveal which parts of the rubric each person interprets differently. Thirty minutes of discussion at this stage can prevent hours of regrading and several uncomfortable emails from students. It also gives newer teaching assistants a chance to ask questions in a low-pressure setting before real grades are on the line.

• Choose five sample essays that cover the full range of quality
• Have every grader score them independently before any discussion
• Compare scores row by row and talk through each disagreement
• Write down the rulings so they can be applied to later papers
• Repeat with two new essays halfway through grading to catch drift

Giving Useful Feedback at Scale

Students in large courses often receive little more than a letter grade and a few margin check marks, and they notice. A better approach is to write two or three targeted comments on each paper that address the biggest issue, one strength, and one concrete step for the next essay. For a Lewis paper, that might mean noting that the student has a strong thesis about Orual's self-deception but needs to explain the evidence from Part Two more fully.

Keeping a running list of comments you find yourself repeating saves enormous time. After the first twenty papers, most professors notice that the same six or seven issues account for the majority of their feedback. Turning those into a library of short, editable comments lets you give specific responses without retyping them each time. Sharing that library with your TAs spreads the same language across the whole course, so students get consistent comments no matter who grades them.

Where Software Helps and Where It Does Not

AI grading tools are well suited to the first pass in a large class. They can apply a rubric consistently across every paper, flag essays that lack a thesis or use no quotations, and draft criterion-level comments for the professor to edit. This does not replace the professor's judgment, but it handles the volume that makes consistency hard for a human reader.

Professors should still read a meaningful sample in full, including every paper that sits on a grade boundary. Those borderline cases are where a student's grade is most sensitive to small differences in judgment, and they are also where appeals tend to arise. Reading them yourself protects both the student and the integrity of the course. A good rule is to read any paper within a few points of a grade cutoff, plus any paper the first pass flagged as unusual.

Planning the Turnaround Time

Large courses rarely have the luxury of a leisurely grading schedule, since the next reading is already underway. Setting a published return date at the start of the assignment, and planning backward from it, keeps the work from piling up. A realistic target for a 150-essay stack with two TAs is about ten days, assuming calibration happens in the first two.

Students who get their essays back quickly can use the comments while the novel is still fresh and apply them to the next paper. Delayed feedback loses most of its value, because by the time it arrives the class has moved on to different texts and different questions. Fast, consistent grading turns an administrative burden into a genuine teaching tool.