Few novels produce as much strong student writing as "Monster" by Walter Dean Myers, and few create as much grading pressure. A single set of character analysis essays about Steve Harmon can run to dozens of pages of dense argument about guilt, identity, and how other people see him. Teachers often read the first ten essays with real care and then start skimming, which means the students at the bottom of the stack get thinner feedback. AI essay grading can even that out when it is tied to the same rubric for every paper.

The first step is deciding what the essay is actually measuring. For a "Monster" unit, that might be a claim about whether Steve is guilty in a moral sense, supported by two or three pieces of textual evidence and explained in the student's own words. When those expectations are written down before grading begins, the tool has something concrete to score against instead of guessing at what a strong paper looks like.

Many teachers also worry that a machine will miss the emotional weight of the book. That concern is fair, and it is the reason the teacher should still set the criteria and review the results. The software handles the repetitive work of checking evidence, structure, and clarity on every paper, while the teacher spends saved time on the conversations that only a person can have about a difficult story.

Start With a Prompt That Produces Gradable Arguments

Vague prompts such as "discuss the themes in Monster" produce vague essays, and vague essays are the hardest to grade fairly. A better prompt asks students to take a position, for example whether the book wants readers to see Steve as innocent, guilty, or something more complicated, and to defend it with specific moments from the trial and his journal. That kind of prompt gives both the teacher and the grading tool a clear claim to evaluate.

• A defensible claim about Steve Harmon's guilt or responsibility
• At least two pieces of direct evidence from the text
• Explanation that connects each quote to the claim
• Awareness of the screenplay and journal format
• Clear organization from introduction to conclusion

Build the Rubric Around What Students Actually Write

Rubrics for a novel like "Monster" work best when each row describes observable behavior rather than a general feeling. Instead of a row labeled "analysis," try describing the difference between a student who retells the courtroom scenes and one who explains why the courtroom framing makes readers doubt what they think they know. Students read those descriptions and understand what to fix, and the grading tool can apply them without drifting.

It also helps to write one short anchor example for each performance level using a sample paragraph from a previous year. Teachers who do this tend to catch their own inconsistencies early, because the anchors force a decision about where a paper lands. The same anchors can be reused across sections, which makes shared grading with a colleague far less painful.

Keep the Teacher in Charge of the Final Grade

AI feedback should be treated as a strong first pass, not a verdict. A teacher who knows that a particular student wrote about Steve's fear in a way that echoes something from class discussion will notice things a rubric alone cannot capture. Reviewing the generated comments, adjusting a score when it does not fit, and adding a personal note takes far less time than writing everything from scratch.

Over a full unit, this workflow changes the rhythm of grading. Essays come back in a day or two instead of two weeks, which matters because feedback is only useful while students still remember what they were arguing. Faster turnaround also leaves room for a revision cycle, and revision is where most students improve their writing about literature.

Measure Whether the Workflow Is Working

After the first unit, compare a handful of AI-assisted grades against what you would have given by hand. If they cluster closely, the rubric is doing its job, and if they diverge, the gap usually points to a rubric row that is too vague. Adjusting that one row is often enough to bring the results back in line.

Pay attention to student reactions as well. When learners say the comments feel specific and fair, that is a good sign the process is respecting their work, and when they say the feedback feels generic, the prompt or rubric probably needs more detail. A few small adjustments each semester compound into a grading system that gets better every time "Monster" comes around on the syllabus.