Subjects · Literature & Language Arts
Literature & Language Arts: the methods that matter most here
Know which methods actually build the skill this subject assesses, the reading, not the summary, and which popular habits quietly do the assessed work for you.
What you'll be able to do: Know which methods actually build the skill this subject assesses, the reading, not the summary, and which popular habits quietly do the assessed work for you.
What makes this subject hard, specifically
A reading and a summary look similar and are not. A summary restates what happens. A reading makes a claim about what the text does and why, and defends it with evidence. Students who can produce fluent summaries are routinely marked down for "not enough analysis" without being told the difference is structural, not a matter of trying harder.
Evidence here means the exact words, not the gist. A misquoted line, a detail from the wrong chapter, a paraphrase presented as quotation, these aren't rounding errors the way an approximate number might be elsewhere. The evidence is the words, and if the words are wrong the evidence has evaporated.
The interpretation is the one thing being assessed, and it's also the fastest thing the tool produces. This is the subject's specific danger, spelled out in article 00. Nowhere else in this corpus does the model's raw output land so close to the deliverable.
Close reading is trainable, not a talent you either have or don't. Noticing that a line break falls mid-word, that a verb tense shifts, that a narrator withholds a name, these are learnable habits of attention, built by repeated, guided noticing. Students told they're "just not good at analysis" have usually never been shown the moves.
"I liked it" is not yet an argument. A response is data. An argument specifies what in the text produced the response and what evidence would count against it. The gap between the two is most of what separates a personal-response paragraph from an essay.
Linguistics runs on a different failure entirely: intuition standing in for evidence. "That sentence sounds wrong" is a claim about the speaker's own dialect and register, not a fact about the language. The discipline's method is descriptive, collect what speakers actually produce and judge, not prescriptive, assert what a rule says they should produce. Confusing the two is the linguistics equivalent of "descriptive" in the history margin: a habit of thought that feels like analysis and isn't.
The ten
| # | Method | Why it earns its place here |
|---|---|---|
| 1 | D16 Primary source interrogation | Turns the text from content to be summarised into evidence to be questioned |
| 2 | C22 Forced operationalisation | Converts "I liked it" into a claim specific enough to be wrong |
| 3 | C9 Counterexample hunting | Close reading is the discipline of finding the line that doesn't fit your claim |
| 4 | C7 Assumption surfacing | Thematic claims smuggle premises the text may not actually support |
| 5 | C5 Steelmanning | A reading is only as strong as the rival reading it has survived |
| 6 | F12 Style analysis | Technique-spotting is the tool's genuine strength; naming it and then attempting it makes it a skill rather than a label |
| 7 | K1 Verification drill | The subject's largest specific risk: quotations and plot details the tool did not actually check |
| 8 | F9 Draft-and-critique loops | Essays improve through structured critique cycles, not by being generated well once |
| 9 | F10 Reverse outlining | Shows you wrote a summary when you meant to argue |
| 10 | H9 Standard-setting comparison | The gap between a competent and a strong reading is one specific, learnable move |
The methods that work badly here
"What does this poem/scene/chapter mean?"
The single worst question to put to the tool, because the answer to it is the exact thing the assignment is testing. Asking it directly is not a shortcut to the work, it is skipping the work and keeping the label. If you use AI on an interpretive question at all, ask it for questions, not answers (article 01).
"Summarise this chapter so I don't have to read it."
You cannot close-read a summary. The techniques that create meaning, the specific diction, the sentence rhythm, the exact placement of a detail, are precisely what a summary strips out to compress. A student who works from AI-generated summaries can discuss plot fluently and has nothing to say about how the text produces its effects, because the raw material for that discussion was never in front of them.
Trusting its quotations and plot details as though it had the book open.
It doesn't have the book open. It has a statistical impression of the book, which is excellent for famous passages and increasingly unreliable the further you get from them. Every quotation, every "the scene where X happens", every "chapter fourteen" needs checking against a copy you have actually opened (article 07) before it goes anywhere near submitted work.
The composed workflow
- Interrogate the text before forming a view (D16). If you can state a reading without having reread the passage, you've skipped the evidence.
- Turn your first reaction into an operational claim (C22).
- Hunt for the counterexample that complicates it (C9), and surface what it's quietly assuming (C7).
- Build the strongest rival reading and answer its best point (C5).
- Write. Then reverse-outline to find out what you actually argued (F10), and verify every quotation before it enters the final draft (K1).
That sequence covers evidence, argument, rival positions and accuracy, which is most of what a literature or composition rubric actually rewards.
The ten articles in this subject
01 Primary source interrogation · 02 Forced operationalisation · 03 Counterexample hunting · 04 Assumption surfacing · 05 Steelmanning rival readings · 06 Style analysis · 07 Verification drill · 08 Draft-and-critique loops · 09 Reverse outlining · 10 Standard-setting comparison