Gains:
- Ability to carry out close reading step by step and have artificial intelligence mark figures of speech, themes and structure with quotes from the text
- Ability to prevent artificial intelligence from going beyond the text and producing clichés and linking each claim to the evidence in the text
- Ability to understand that even though it receives the raw material from artificial intelligence, it must always establish the holistic meaning, aesthetic judgment and final interpretation itself.
At the heart of the study of literature is text analysis: analyzing word by word, image by image, what a text says, how it says it, and why it says it that way. The most intense form of this is close reading - the method of carefully analyzing a text only through its own words, voice, images and structure, without resorting to external information. What is important in close reading is not the gossip outside the text, but the evidence within the text: which word was chosen, which sound was repeated, which image clashed with which image.
In this unit we will learn to use AI as a close reading “aid”: an assistant that outlines figures of speech, lists themes, maps structure. But let's underline from the very beginning: AI can mark the image, but it is you, the reader, who decides what that image does in the poem and what emotion it creates. The AI collects the raw material for the assay; You create the meaning.
How does close reading work step by step?
Consider close reading with AI in five steps:
- Give the text completely and accurately. The poem, story or section to be analyzed must be complete and accurate. Ask the AI to rely on the text you give it, not the text it remembers from memory.
- Remove surface layer. What happens at the word level? Sound (alliteration, internal rhyme), word choice, time, person, place. Alliteration is the repetition of the same consonant in words that are close to each other.
- Have the images and figures of speech marked. Simile, metaphor (referring to something by the name of another), oxymoron, repetition, symbol. Ask for each with a quote from the text.
- Map structure and tension. Where is the breaking point of the poem? How is tension built in the story, where does it reach its climax?
- You create the comment. You take the raw material that the AI has marked and ask "what does all this say together?" You answer the question. This step is non-delegable.
Tip: Don't tell the AI to "interpret"; Say "mark and cite". AI's forte is showing patterns; His weakness is establishing the meaning of that pattern without resorting to cliché.
Prevent escaping from text
The AI's most frequent mistake is to stray outside the text while seeking close reading: into the poet's life, into "generally" known comments, into memorized clichés. However, the rule of close reading is that the evidence is within the text. Explicitly forbid this in the prompt: "Base only on the text I give; do not make assumptions about the poet's biography, period, or other works."
Caution: AI is prone to tag a poem with ready-made phrases such as "sad" or "expresses patriotism". Don't accept these patterns without evidence; "From which word of the text did you infer this?" ask. An interpretation without evidence is a guess, not an analysis.
three mini cases
Case 1 — The rhetorical sketch saved time. A teacher had AI mark figures of speech as quotations before teaching a 40-line poem in class. The list came out in 5 minutes; The teacher saw that 3 of them were forced and eliminated them. He approved 9 of them and constructed the meaning of each one himself in the lesson. Preparation time has been reduced by half.
Case 2 — Building map sparked discussion. A reading group had the AI map the tension curve of a 12-page story. AI put the climax on page 8; When the group returned to the text, they realized that the real break was in a single sentence on page 6. AI's "wrong" map started the right discussion.
Case 3 — Fabricated quote caught. A student saw a quote in YZ's analysis saying "as in this line". When he looked at the poem, that line was not there; AI had made up a verse that didn't exist. The student adopted the rule of marking each quote from the text.
Four copyable templates
1) Close reading framework:
Your role: a close-reading literary analyst. Analyze the following text based on your own words ONLY. Mark the following under SEPARATE headings, quoting the text for each: (1) voice/consonance, (2) word choice, (3) imagery and figures of speech, (4) structure and breaking point. DO NOT comment, just mark and quote. Don't go beyond text.Text: [paste text here]
2) Rhetorical marking:
Find the figures of speech in the poem below. For each: the name of the art, the verse in which it is used (verbatim quotation), short justification. Mark the part you are not sure of as "disputable". No art that has no equivalent in the text is fabricated. Poem: [insert the poem here]
3) Theme map:
List the recurring themes in the story below. For each theme, give at least two quotes from the text and tell where it occurs. Highlight the evidence in the text, not your own interpretation. At the end, add a heading "Open points for my comment". Text: [paste text here]
4) Counter reading request:
Suggest an alternative reading to my analysis below. Show me what detail in the text I may have missed that supports a different interpretation. Just rely on the text.My analysis: [write your analysis]Text: [paste text]
Weak prompt / Strong prompt
Weak: "Tell me the theme of this poem."
Güçlü: "List the themes in this poem, quoting at least two lines for each. Rely solely on the words in the text; do not make assumptions about the poet's life or period. Do not use made-up quotes to support a theme; mark 'disputable' where you are not sure."
A strong prompt imposes a burden of proof and prevents AI from falling into cliché and fabrication. In the weak prompt, the AI gives a one-sentence memorized "theme" and you can't verify it.
AI reliability table with rhetorical art
item
What does AI do?
reliability
What does one do?
Alliteration/sound repetition
signs
high
Approvals
word frequency
counts
high
Makes sense
simile
Nominates
medium
Facts, comments
Izlek
Makes a list
medium
Evaluates with evidence
breaking point
guesses
Low-Medium
Verifies from the text
holistic meaning
Produces cliché
low
Creates original comment
Close reading in novels and short stories
Close reading is not unique to poetry; It is also applied to short and intense sections in novels and stories. It is not possible to read an entire novel line by line; but you can close-read an opening paragraph, a turning point, or a closing. The elements to pay attention to in the narrative are slightly different: the narrator (whose mouth are we hearing about the event - first person or an all-knowing outside voice), point of view (through whose eyes we see the events), use of tense (flashback, jump), and the balance of showing/telling (does the author say an emotion directly or hint it with a detail). It's much more efficient to have the AI mark these elements individually when giving a narrative section than getting a rough "summary".
For example, in an opening paragraph, you can tell the AI to "show who the narrator is, his point of view, and what expectation the first sentence sets in the reader, with quotes from the text for each." Thus, instead of repeating the plot, AI produces raw material that shows how the narrative is constructed. Still, you have the final say: your reading experience is inalienable, deciding whether a narrator is trustworthy, what a silence says, and why a scene jars.
Tip: In a long work, first have AI make a chapter summary and identify two or three intense scenes that are "worth a close read"; Then analyze it in depth by giving only those scenes from the text. This saves time and links the analysis to concrete evidence.
Common mistakes
- It means "with comment". AI gives cliché; Instead, say "mark and quote" and you make the comment.
- Letting it escape outside the text. Ban biographies and common knowledge.
- Not verifying quotes. See how each quote actually occurs in the text.
- Trusting the text given by the AI. AI may remember incorrect text from memory; You give the text to be analyzed.
- Absolutizing single reading. Test the interpretation by asking for a counter-reading.
In summary
In close reading, the AI is a raw material collector, quickly marking figures of speech, themes, and structure; But you are the reader who establishes the meaning, eliminates the cliché, and makes the aesthetic judgment. Prohibit going beyond the text, quote every claim from the text, verify the quotations, and always construct the holistic interpretation yourself.
Application task
Choose a poem. First, make three observations yourself in five minutes. Then give the same poem to the AI with the "Close reading skeleton" template. Compare what the AI flags with your own observations: which are real patterns you don't see, which are clichés or made-up? Confirm each quote from the text and write the complete interpretation yourself in one paragraph.
checklist
- [ ] I gave the text to be analyzed, I did not leave it to the AI to memorize it.
- [ ] I asked for a quote from the text for each figure of speech and theme.
- [ ] I verified that the quotes are actually in the text.
- [ ] I have banned extra-textual assumptions.
- [ ] I made the holistic interpretation and aesthetic judgment myself.