Adapted from a 2020 graduate Reading Methods essay for Sharon Marcus. The original combines close reading, quantitative literary history, statistical NLP, film, gender and sexuality studies, and methodological criticism.
Literary criticism often treats interpretation as almost synonymous with reading. But scholars do more than interpret. We describe, evaluate, explain, and interpret—and those operations make different kinds of claims.
Separating them can make an argument more precise. It can also make disagreements easier to diagnose.
Description: What Is Here?
Description moves from an object to a representation of that object. We identify a formal feature, summarize a pattern, count a term, characterize a scene, or model a corpus.
Description is never perfectly neutral: deciding what to notice already involves selection. But a descriptive claim is still different from a claim about what the observed feature means.
Interpretation: What Might This Mean?
Interpretation moves from evidence toward meaning or conclusion. If A is a textual feature and B is an interpretation, the basic move resembles A → B.
The arrow matters more than the notation. Evidence can support an interpretation without making that interpretation the only possible conclusion. The reader’s theoretical and historical lens participates in the move.
Explanation: Why Did This Happen?
Explanation reverses the emphasis. Instead of asking what A means, we ask what conditions, causes, structures, or mechanisms help account for A.
If B helps explain A, then B → A does not automatically mean A → B. A historical condition may help explain a textual feature without every instance of that feature proving the same historical cause.
Evaluation: What Value Are We Assigning?
Evaluation assigns value according to criteria. Calling a work innovative, canonical, successful, representative, prestigious, or politically useful is not merely describing it.
The implicit structure is closer to A = B, but the equality depends on the standards being applied. Those standards may be aesthetic, institutional, historical, political, statistical, or disciplinary—and they need to be made visible.
Why the Distinctions Matter
Weak arguments often become clearer when we ask which operation is actually being performed. A description may be presented as if it proves an interpretation. An evaluation may masquerade as neutral description. An explanation may identify a plausible cause and then quietly treat it as the only cause.
The distinction is also useful when computational and traditional humanities methods meet. A model may describe a pattern across thousands of texts. It does not automatically interpret that pattern. Close reading may propose a meaning for an anomaly. It does not automatically establish how common that anomaly is.
These methods become stronger when they answer different questions rather than compete for the same territory.
Reading as Method
The mathematical language here is not an attempt to turn literature into proof. It is a way of making the direction of a claim visible.
What evidence are we starting with? What are we claiming from it? Are we assigning meaning, cause, value, or representation? What assumptions make that movement possible?
Those questions make reading more—not less—interpretive, because they force us to articulate what we are actually doing when we say that a text means something.