Most English departments already own a literary analysis rubric, and most of them fit Till We Have Faces poorly. Standard criteria such as "thesis," "evidence," and "organization" are fine in the abstract, but they say nothing about the unreliable narration, the two-part structure, or the religious ambiguity that define this book. A rubric tailored to the novel tells students what a strong paper looks like before they begin writing.

Start by deciding how many criteria you can realistically assess. Four to six is usually the limit before a rubric turns into a checklist that nobody reads, and each criterion should describe something a student can visibly improve. For this novel, that often means interpretation of narration, use of textual evidence, treatment of the two-part structure, engagement with the myth source, and clarity of writing.

The performance levels matter as much as the criteria. Vague labels like "excellent" and "good" force the grader to guess, while descriptive levels let students see exactly what separates a proficient paper from an advanced one. Writing each level as an observable behavior, such as "explains how Orual's word choices reveal what she refuses to admit," keeps scoring consistent across graders.

Criteria That Fit This Novel Specifically

Interpretation of narration deserves its own criterion because it is the skill the book tests most directly. A top-level paper recognizes that Orual is writing an accusation against the gods and reads her account with suspicion, while a lower-level paper repeats her claims as if Lewis endorsed them. This single distinction predicts overall quality better than almost any other feature of a student's essay.

• Interpretation of Orual as a narrator with motives, not a neutral witness
• Use of quoted evidence from both Part One and Part Two
• Explanation of how the second part reframes the first
• Connection to the Cupid and Psyche myth without relying on plot recap
• Control of claims, paragraph structure, and sentence-level clarity

Writing Performance Levels That Students Can Act On

Descriptors should name what is present at each level rather than what is missing. Instead of writing that a lower-level paper "lacks depth," describe it as one that summarizes events from the novel and mentions themes without tying them to specific passages. Students respond better to descriptions of what they did than to judgments about what they failed to do.

It also helps to include a short example sentence for each level. A thesis at the developing level might say that Orual is jealous of Psyche, while one at the advanced level might argue that Orual's jealousy is disguised as concern, and that her narration works to hide this from herself. Seeing the two side by side teaches more than a paragraph of abstract description.

Weighting and Scoring Without Overcomplicating It

Not every criterion should count equally. Interpretation and evidence usually deserve more weight than surface polish, because a student who misreads the novel cannot be rescued by clean grammar. Many teachers weight the interpretive criteria at roughly sixty percent and divide the remainder among structure, myth connection, and mechanics. A weighting that mirrors this logic also tells students where to spend their revision time, which tends to improve the quality of second drafts.

Resist the temptation to add extra categories for every skill you care about. A rubric with ten rows invites inconsistent scoring and overwhelms students, who end up focusing on whichever row has the lowest score. A tighter rubric gets cleaner data and produces feedback that students can reasonably absorb in a single sitting. If a department truly needs to track more skills, it is better to run a second rubric on a different assignment than to cram everything into this one.

Using the Rubric With AI Grading and Feedback Tools

A well-written rubric is also what makes AI-assisted grading reliable. When each criterion has observable level descriptions, a tool can compare a student's paragraph to those descriptions and draft comments tied to the right row. Vague rubrics produce vague machine feedback, so the time spent sharpening the language pays off twice. Teachers who write descriptors with a tool in mind often find that human graders and students benefit from the extra clarity too.

Teachers should still review the output, especially on criteria that involve judgment about interpretation. A quick scan of a handful of papers at the start of a batch shows whether the rubric language is being applied the way you intended. If the tool is misreading a criterion, the fix is usually a clearer descriptor, which improves the rubric for human graders as well.