{"course":{"slug":"ai-for-students","title":"AI for Students","url":"https://addaly.com/learn/ai-for-students","level":"beginner","track":"work"},"lesson":{"slug":"literature-and-close-reading","title":"Literature, and the text it half remembers","url":"https://addaly.com/learn/ai-for-students/literature-and-close-reading","position":34,"of":73,"minutes":8,"answer":"Generated quotations from literary works are usually near-misses, and an argument built on words that are not in the book is worth nothing: read and notice first, consult the critical consensus only afterwards.","keyIdea":"Generated quotations from literary works are usually near-misses, and an argument built on words that are not in the book is worth nothing: read and notice first, consult the critical consensus only afterwards."},"body":{"format":"markdown","text":"## It has read about the book\n\nA model has encountered enormous quantities of writing *about* literary works (reviews, essays, study guides, forum arguments) and, for older works, often the text itself. What it does not have is your reading of it, and it cannot do the one thing literary study is actually about: attending closely to particular words on a particular page.\n\nThis produces a specific failure. Ask about a well-known novel and you get a competent summary of the critical consensus, which is what a study guide would give you, and which is what a marker gives low marks to. Ask for a quotation and you will often get a plausible paraphrase presented as the text.\n\n## Misquotation is the characteristic error\n\nGenerated quotations from literary works are frequently near-misses: the right sense, the wrong words, or words assembled from different parts of the text. For a subject where the argument is built on the exact wording, this is fatal. A close reading of a sentence that is not in the book is worth nothing, and it is immediately visible to anyone who knows the work.\n\n**Every quotation gets checked against the text in front of you, with a page or line reference.** Not against a search result. Against your copy, or a reliable scholarly edition. For out-of-copyright works, **Project Gutenberg** and the **Internet Archive** are free and searchable; for verse, be careful about which edition you are using, since line numbering differs.\n\n## What close reading is, and why it cannot be delegated\n\nClose reading is noticing that a word is doing something unexpected, and building an argument from it. The noticing is the skill. It requires having the text in front of you, slowly, and being willing to sit with a sentence that will not resolve.\n\nA model can tell you what critics have noticed. That is useful *afterwards*: it tells you whether your observation is original, obvious, or contested. It is useless *before*, because reading the consensus first means you will notice what you were told to notice, and the resulting essay reads exactly like what it is.\n\nThe order that works: read, notice, write your observation, *then* consult. Never the reverse.\n\n## Uses that are good\n\n- **Historical and linguistic context.** What did this word mean in 1798? What would a contemporary reader have understood by this allusion? Strong, and verify the specifics.\n- **Prosody and form.** Scanning a line, identifying a form, naming a device. Reliable on standard forms and worth checking against the poem, since it is working from a possibly-misremembered text.\n- **Foreign-language and older texts.** Getting a foothold in Chaucer, or in a language you read imperfectly. A ladder to the original, not a replacement for it.\n- **Counterargument.** *What is the strongest objection to this reading?* Excellent, and it is where most literature essays are weakest.\n- **Structural feedback on your own draft.** Which paragraph does not advance the argument, where the evidence is thin. Ask for diagnosis, not rewriting.\n- **Finding secondary literature to look up.** Ask what critical debates surround a work, then find the actual articles yourself. Treat every named article as unverified until you hold it.\n\n## Study guides, old and new\n\nSparkNotes and its descendants have existed for decades and departments have always known what an essay written from one looks like. A model produces the same thing more fluently. The tell is identical: an essay that is *about* the novel's themes at a distance, with no sustained attention to any actual passage, and no observation that could only have come from having read it.\n\nThe strongest thing you can put in a literature essay remains a small, precise, surprising observation about a few words on the page, defended. No tool can supply that, because it depends on a reading nobody else has done.\n\n## For set texts and exams\n\nMany literature exams are closed-book or use a clean text. Memorising quotations is a real requirement, and it is one place where the model is a good drilling partner: paste twenty quotations you need, ask to be tested on them one at a time with the first three words as the prompt, and to be told which you keep getting wrong.\n\nThat is retrieval practice on material you have verified yourself, which is the safe and effective arrangement: the verification happened first, and the drilling happens on text you know is right."},"check":{"question":"Why does consulting a model about a novel's themes before doing your own close reading damage the resulting essay?","options":[{"text":"Because the model's account of the themes is usually wrong, and the misreading propagates into the essay.","correct":false,"tempting":"Overstates the accuracy problem; on well-known works the thematic summary is generally serviceable, which is precisely what makes it dangerous."},{"text":"Because markers can detect the stylistic signature of model-influenced argument even when the student writes every word themselves.","correct":false,"tempting":"Shifts to a detection claim that the course elsewhere shows is unreliable; the damage here is to the quality of the reading, not to how it looks."},{"text":"Because quotations supplied at that stage will be unverified and are likely to be misremembered.","correct":false,"tempting":"A real hazard covered elsewhere in the lesson, but it applies to quotations at any stage, not to the ordering problem being asked about."},{"text":"Because you will then notice what you were told to notice, and the essay becomes a version of the consensus.","correct":true}]},"prev":{"slug":"history-and-sources","title":"History, and the quotation that was never said","url":"https://addaly.com/learn/ai-for-students/history-and-sources"},"next":{"slug":"learning-a-language","title":"Languages, where it is genuinely excellent","url":"https://addaly.com/learn/ai-for-students/learning-a-language"},"authorship":"Founder and series editor: Dr Mustafa Mun. Written by AI models to a written standard. Exam keys checked blind by a second model: 3,665 of 3,666 agreed. Academic review invited.","review":{"status":"Draft, open for academic review"},"cite":"Addaly (2026) 'Literature, and the text it half remembers', in AI for Students. Edited by M. Mun. Addaly. Available at: https://addaly.com/learn/ai-for-students/literature-and-close-reading (Accessed: 13 October 2026).","licence":{"name":"CC BY-NC-SA 4.0","url":"https://creativecommons.org/licenses/by-nc-sa/4.0/"},"reader":{"schema":1,"title":"Literature, and the text it half remembers","courseTitle":"AI for Students","module":{"index":4,"slug":"subject-by-subject","title":"Subject by subject"},"answer":null,"keyIdea":"Generated quotations from literary works are usually near-misses, and an argument built on words that are not in the book is worth nothing: read and notice first, consult the critical consensus only afterwards.","hero":{"kind":"plate","image":null,"alt":null,"caption":null},"blocks":[{"t":"h2","id":"it-has-read-about-the-book","runs":[{"text":"It has read about the book"}]},{"t":"p","runs":[{"text":"A model has encountered enormous quantities of writing "},{"text":"about","i":true},{"text":" literary works (reviews, essays, study guides, forum arguments) and, for older works, often the text itself. What it does not have is your reading of it, and it cannot do the one thing literary study is actually about: attending closely to particular words on a particular page."}]},{"t":"p","runs":[{"text":"This produces a specific failure. Ask about a well-known novel and you get a competent summary of the critical consensus, which is what a study guide would give you, and which is what a marker gives low marks to. Ask for a quotation and you will often get a plausible paraphrase presented as the text."}]},{"t":"h2","id":"misquotation-is-the-characteristic-error","runs":[{"text":"Misquotation is the characteristic error"}]},{"t":"p","runs":[{"text":"Generated quotations from literary works are frequently near-misses: the right sense, the wrong words, or words assembled from different parts of the text. For a subject where the argument is built on the exact wording, this is fatal. A close reading of a sentence that is not in the book is worth nothing, and it is immediately visible to anyone who knows the work."}]},{"t":"p","runs":[{"text":"Every quotation gets checked against the text in front of you, with a page or line reference.","b":true},{"text":" Not against a search result. Against your copy, or a reliable scholarly edition. For out-of-copyright works, "},{"text":"Project Gutenberg","b":true},{"text":" and the "},{"text":"Internet Archive","b":true},{"text":" are free and searchable; for verse, be careful about which edition you are using, since line numbering differs."}]},{"t":"h2","id":"what-close-reading-is-and-why-it-cannot-be-delegated","runs":[{"text":"What close reading is, and why it cannot be delegated"}]},{"t":"p","runs":[{"text":"Close reading is noticing that a word is doing something unexpected, and building an argument from it. The noticing is the skill. It requires having the text in front of you, slowly, and being willing to sit with a sentence that will not resolve."}]},{"t":"p","runs":[{"text":"A model can tell you what critics have noticed. That is useful "},{"text":"afterwards","i":true},{"text":": it tells you whether your observation is original, obvious, or contested. It is useless "},{"text":"before","i":true},{"text":", because reading the consensus first means you will notice what you were told to notice, and the resulting essay reads exactly like what it is."}]},{"t":"p","runs":[{"text":"The order that works: read, notice, write your observation, "},{"text":"then","i":true},{"text":" consult. Never the reverse."}]},{"t":"h2","id":"uses-that-are-good","runs":[{"text":"Uses that are good"}]},{"t":"ul","items":[[{"text":"Historical and linguistic context.","b":true},{"text":" What did this word mean in 1798? What would a contemporary reader have understood by this allusion? Strong, and verify the specifics."}],[{"text":"Prosody and form.","b":true},{"text":" Scanning a line, identifying a form, naming a device. Reliable on standard forms and worth checking against the poem, since it is working from a possibly-misremembered text."}],[{"text":"Foreign-language and older texts.","b":true},{"text":" Getting a foothold in Chaucer, or in a language you read imperfectly. A ladder to the original, not a replacement for it."}],[{"text":"Counterargument.","b":true},{"text":" "},{"text":"What is the strongest objection to this reading?","i":true},{"text":" Excellent, and it is where most literature essays are weakest."}],[{"text":"Structural feedback on your own draft.","b":true},{"text":" Which paragraph does not advance the argument, where the evidence is thin. Ask for diagnosis, not rewriting."}],[{"text":"Finding secondary literature to look up.","b":true},{"text":" Ask what critical debates surround a work, then find the actual articles yourself. Treat every named article as unverified until you hold it."}]]},{"t":"h2","id":"study-guides-old-and-new","runs":[{"text":"Study guides, old and new"}]},{"t":"p","runs":[{"text":"SparkNotes and its descendants have existed for decades and departments have always known what an essay written from one looks like. A model produces the same thing more fluently. The tell is identical: an essay that is "},{"text":"about","i":true},{"text":" the novel’s themes at a distance, with no sustained attention to any actual passage, and no observation that could only have come from having read it."}]},{"t":"p","runs":[{"text":"The strongest thing you can put in a literature essay remains a small, precise, surprising observation about a few words on the page, defended. No tool can supply that, because it depends on a reading nobody else has done."}]},{"t":"h2","id":"for-set-texts-and-exams","runs":[{"text":"For set texts and exams"}]},{"t":"p","runs":[{"text":"Many literature exams are closed-book or use a clean text. Memorising quotations is a real requirement, and it is one place where the model is a good drilling partner: paste twenty quotations you need, ask to be tested on them one at a time with the first three words as the prompt, and to be told which you keep getting wrong."}]},{"t":"p","runs":[{"text":"That is retrieval practice on material you have verified yourself, which is the safe and effective arrangement: the verification happened first, and the drilling happens on text you know is right."}]}],"check":{"question":[{"text":"Why does consulting a model about a novel’s themes before doing your own close reading damage the resulting essay?"}],"options":[{"text":[{"text":"Because the model’s account of the themes is usually wrong, and the misreading propagates into the essay."}],"correct":false,"tempting":[{"text":"Overstates the accuracy problem; on well-known works the thematic summary is generally serviceable, which is precisely what makes it dangerous."}]},{"text":[{"text":"Because markers can detect the stylistic signature of model-influenced argument even when the student writes every word themselves."}],"correct":false,"tempting":[{"text":"Shifts to a detection claim that the course elsewhere shows is unreliable; the damage here is to the quality of the reading, not to how it looks."}]},{"text":[{"text":"Because quotations supplied at that stage will be unverified and are likely to be misremembered."}],"correct":false,"tempting":[{"text":"A real hazard covered elsewhere in the lesson, but it applies to quotations at any stage, not to the ordering problem being asked about."}]},{"text":[{"text":"Because you will then notice what you were told to notice, and the essay becomes a version of the consensus."}],"correct":true,"tempting":null}],"order":[1,0,2,3],"why":null},"prev":{"slug":"history-and-sources","title":"History, and the quotation that was never said","minutes":8},"next":{"slug":"learning-a-language","title":"Languages, where it is genuinely excellent","minutes":9}}}