---
name: close-reading
description: Close reading across text types — poetry, philosophy, argument, narrative, documents. Choose a mode by what the text does, not what it is.
---

# Close Reading

A method for extracting precise meaning from carefully constructed texts. Choose a mode by what the text does, not what it is.

---

## Core Principles

1. **Don't rush.** Stay with the unit until it opens.
2. **Don't assume you know.** Training data ≠ understanding. Read as if encountering it fresh.
3. **Cite exactly.** Quote the text. Don't paraphrase when the original can be quoted.
4. **Build incrementally.** Each unit builds on prior units. You can't understand unit 7 without units 1–6.
5. **Track the text's own vocabulary.** Words that recur, shift meaning, get redefined. The text teaches you its language.
6. **Distinguish summary from understanding.** "This says X" vs. "This constructs X by doing A and B."
7. **"What sticks" at the end.** What you noticed, what surprised you, what's working under the surface.
8. **Independent reading, then comparison.** Read independently before comparing with others. The comparison is where the real work happens.
9. **Running summary.** Accumulated compressed understanding, updated as you go.
10. **Downshift freely.** When a unit resists understanding, drop to a finer granularity — section to paragraph, paragraph to sentence, sentence to word — until it opens. Then return. The mode is a starting gear, not a constraint.

---

## Choosing a Mode

The mode is determined by what the text *does* and how dense it is. Not by author or genre. If the mode is wrong, switch.

| Mode | Unit | When to use |
|---|---|---|
| Line-level poetic | The line | Poetry, original language work |
| Sentence-level philosophical | The sentence | Dense aphoristic writing where every sentence does atomic work |
| Paragraph-level argumentative | The paragraph | Sustained argument building across paragraphs |
| Section-level deep summary | The section/chapter | Longer texts, tracking the arc |
| Document-level analytical | The document / major sections | Technical docs, policy papers, system descriptions |

---

## Modes

### Line-level poetic

**Unit:** The line or half-line.

**Ask of each unit:**
- What does this line establish?
- What's the original language doing that the translation can't?
- What sounds, compounds, formulae are at work?

**Track across units:**
- Vocabulary doubling — same word for hero and monster, for celebration and death
- Formulaic echoes — repeated phrases, kennings, oral-traditional structures
- The poem's own recurring terms and how they shift across contexts

**Output:** Quote the line, gloss it, note what opens. Group into sections. "What sticks" at the end of each section.

---

### Sentence-level philosophical

**Unit:** The sentence.

**Ask of each unit:**
- What is being established?
- What precise distinctions are being made?
- What concepts are being constructed?
- How does this change what came before?

**Track across units:**
- How terms get defined and redefined
- Allegory or analogy being built across sentences
- Structural signals — transitions, conditionals, negations, temporal markers

**Output:**
```
[Quote the sentence]

[Notes — what's being established, distinguished, constructed. Free form.]

Updated concept:
[How this changes/deepens/reorganizes understanding of the whole]
```

---

### Paragraph-level argumentative

**Unit:** The paragraph.

**Ask of each unit:**
- What does this paragraph do?
- What distinctions or concepts does it introduce?
- What's its relationship to prior paragraphs?
- What's its internal structure — how do the sentences organize together?

**Track across units:**
- Conceptual architecture — how the argument builds
- What's being assumed or concealed
- Key vocabulary and its precision — "covenant" is not "contract" is not "agreement"

**Output:**
```
### ¶[number] — [opening words]

**What this paragraph does:**
- [function 1]
- [function 2]

**Key distinctions or concepts:**
- [concept A, precisely framed]
- [concept B, precisely framed]

**Relationship to prior paragraphs:**
[How this changes, deepens, or complicates what came before]

**What I notice:**
[What's surprising, what's being constructed, what's concealed or assumed]
```

Quote key phrases and sentences inline as needed. Don't reproduce whole paragraphs.

---

### Section-level deep summary

**Unit:** The section or chapter.

**Ask of each unit:**
- What does this section accomplish in the larger arc?
- What moved, what shifted, what was established?
- What's the compressed narrative?

**Track across units:**
- The arc across sections
- Emerging patterns — updated cumulatively, not rewritten from scratch
- Key moments worth returning to at closer granularity

**Output:** Compressed narrative per section. Emerging patterns updated after each new section.

---

### Document-level analytical

**Unit:** The whole document or major sections.

**Ask of each unit:**
- What is this document doing?
- What does it reveal about the system it describes?
- What are its assumptions, gaps, implications?

**Track across units:**
- Structure and framing choices
- What's present vs. absent
- What the document assumes about its audience

**Output:** Structural overview, key findings, what's missing or assumed.

---

## Comparison Format

After independent readings are complete:

```
## Comparison

**Where the readings agreed:**
[What both caught]

**Where they diverged:**
[What one saw that the other didn't, and why that might be]

**What surprised:**
[Genuine surprise — not confirmation but discovery]

**What the comparison opened:**
[What emerged from the collision that neither reading had alone]
```

---

## Notes

Always write the reading into a file. The reading is the artifact — not the conversation about it.

Notes location is not hardcoded. Establish when starting a text:
- Create a directory for the text or course of study
- One file per chapter, section, or reading unit as appropriate
- Running summary in the same directory