Few narrators in literature test a student's analytical skill like the Underground Man, whose contradictions are the entire point rather than a flaw to explain away. When grading essays on unreliable narration, instructors frequently encounter two extremes: students who take every claim at face value, and students who dismiss the narrator entirely as a liar with nothing meaningful to say. Neither extreme reflects the sophistication Dostoevsky built into the character. A grading approach that names this middle ground explicitly helps students understand what a genuinely strong essay looks like.

Strong essays on unreliability usually identify specific moments where the narrator contradicts himself, such as his shifting claims about whether he wanted Liza to visit him. Weaker essays gesture vaguely at "the narrator is unreliable" without locating that unreliability in the text. When grading, look for whether the student explains the function of a contradiction rather than just flagging it as an inconsistency. A student who notices the narrator's self-aware admission that he is lying, and explains why that admission itself is suspect, deserves significantly more credit than one who simply lists examples.

It also helps to distinguish between essays that treat unreliability as a psychological symptom and essays that treat it as a philosophical strategy. Some students argue the narrator lies because he is mentally unwell or socially damaged, which is a legitimate reading if supported with textual evidence. Others argue the unreliability is Dostoevsky's method for critiquing rationalist philosophy, since a narrator who cannot even trust his own motives undercuts any claim that reason alone can govern human behavior. Both readings can earn full marks if the student commits to one and defends it thoroughly.

A Grading Framework for Unreliability Essays

A practical framework asks three questions of every essay: does the student identify specific instances of unreliability, does the student explain a plausible reason for that unreliability, and does the student connect it to a larger claim about the novel. Essays that answer all three with textual support belong in the top band regardless of which interpretation they choose. Essays that only answer the first question, simply cataloguing contradictions, belong in the middle band because they show comprehension without full argument. This framework keeps scoring consistent even when student interpretations diverge widely.

• Look for specific quoted contradictions, not general claims about the narrator
• Reward essays that explain the purpose of the unreliability, not just its presence
• Distinguish psychological readings from philosophical readings, and accept either if well supported
• Flag essays that treat the narrator as simply insane, since this often avoids real analysis
• Check that the conclusion ties unreliability back to the essay's central argument

Feedback That Moves Students Forward

When a student's essay stops at cataloguing contradictions, the most useful feedback asks a pointed follow-up question rather than simply marking the essay as underdeveloped. A comment like "you note three contradictions here, but what do they suggest about why the narrator is telling this story" gives the student a concrete next step. Vague comments such as "go deeper" rarely help a student understand what deeper analysis actually requires. Specific, text-anchored feedback is especially important on a novel this psychologically dense, where students can easily feel lost about what "deeper" even means.

It also helps to model the move from observation to argument directly in the margin, showing students how a single contradiction can support a thesis. For example, noting that the narrator's claim to be "a sick man" followed immediately by claims of superior intelligence sets up a pattern worth tracking across the whole text. Pointing this out once, clearly, often does more for a student's next essay than a full page of general comments. Feedback on this text works best when it teaches a transferable reading skill, not just a fix for this one paper.

Scaling This Approach Across a Class

Grading unreliable narrator essays well takes time, since each paper requires checking specific textual claims rather than skimming for keywords. For classes with heavy writing loads, AI-assisted feedback tools that are trained to flag the difference between "observation" and "argument" can draft a first pass of comments tied to this exact distinction. That draft still needs an instructor's eye to confirm the interpretation is textually sound, since a tool cannot always judge whether a reading of Dostoevsky is defensible. Used this way, the technology handles repetitive pattern-spotting while the instructor keeps final judgment on interpretation.

Over a semester, tracking which contradictions students tend to notice and which they miss can also inform how you teach the novel the next time around. If most students catch the Liza contradiction but miss the narrator's shifting claims about the Crystal Palace, that gap is worth addressing directly in class discussion before the next essay is due. Grading, in this sense, becomes diagnostic as well as evaluative. A consistent framework applied across many essays reveals patterns that a single paper never could.