Reading & Writing

54 questions across two 32-minute modules. Each question is a short passage (25–150 words) followed by a single multiple-choice question. The four content domains repeat in roughly fixed proportions in every test.

The four content domains

Reading vs. Writing. The reading questions always come first in each module, then the grammar (conventions) and rhetoric questions. This page covers the reading question types and the reading side of Expression of Ideas. Punctuation, agreement, modifiers, and the other conventions rules live on the separate Grammar page — don't hunt for them here.

The eight reading question types — in order

The reading questions appear in a fixed sequence, and within each type the difficulty climbs from easy to hard. Knowing the order means you can recognize a type the instant you read the question stem, settle into the right routine, and skip whole clusters you'd rather save for last.

Mix note. Retrieval and Main-Idea questions are often interleaved (they test the same "what did the text say" skill), and Claims and Charts questions are likewise interleaved. Don't be thrown when the type seems to flip back and forth.

The universal five-step approach

Every reading question yields to the same skeleton. The order matters: you don't read the passage until you know what you're hunting for.

• 1. Read the question first. One passage, one question — so read the stem before the passage. It tells you the type and what to look for, letting you read with a purpose instead of memorizing.
• 2. Identify the type. A keyword in the stem ("purpose", "main idea", "most nearly mean", "support the claim") fixes the routine. This is also your decision point: a type you find slow? Flag and come back.
• 3. Read the passage actively. Read the blurb/citation first if there is one (it dates and frames the text). Then read for the gist — you do not need every detail. Adjust by genre (see below).
• 4. Find & mark the evidence; predict in your own words. Highlight the one sentence that answers the question. For Vocabulary, Purpose, Cross-Text, and Conclusions, also annotate a short prediction (your own word / the purpose / the relationship). Authors rarely state their purpose outright — your note bridges the gap.
• 5. Use Process of Elimination (POE). Ask of each choice: "Is this consistent with what I marked — yes, no, maybe, no idea?" Don't chase the answer you like; eliminate the three with flaws and what survives is correct, even if it's worded nothing like your prediction. The right answer matches the passage idea-for-idea, not word-for-word.

Open-book, single right answer. Treat reading like the weirdest open-book scavenger hunt: you need no outside knowledge, the answer is 100% supported by the text, and there's exactly one right choice. Go back to the text as often as you like. Don't reward an answer for "matching wording" — meaning beats matching.

Read by genre

• Fiction: Who's speaking? What's the sequence of events / any flashback? What emotion is shown?
• Poetry: Who's the audience? Any symbolic language? How does the narrator feel? (Re-read poems — meaning is subtle.)
• Science (natural / social): Is it an argument or a report of findings? What's the thesis or the main discovery? Is the author voicing an objection they later reject? Don't confuse the author's view with views they cite.
• Historical document: What kind — speech, letter, editorial? What's its point? What period assumptions color it?
• Text 1 / Text 2: What's each thesis? Where do they share assumptions, and where would each author push back on the other?

The six wrong-answer traps

When POE leaves you with two choices, you're almost always staring at the right answer and one trap. Nearly every distractor is built from one of these six patterns. Naming the flaw is how you cut it.

• ① Opposite. A single word flips the tone, viewpoint, or meaning — a "not", or a negative word where the passage was positive. (Passage shows distress; choice says "values".)
• ② Extreme Language. Looks perfect except for a word that overshoots the text: always, only, best, sole, impossible, never. Also flags answers that are insulting or sweeping.
• ③ Recycled Language. Reuses exact words/names from the passage but bolts them together into a claim the passage never made — a relationship that doesn't exist. The classic word-matcher's trap.
• ④ Right Answer, Wrong Question. True about the passage, but it doesn't answer this question — describes what the author said when asked why, or names a detail when asked the main idea.
• ⑤ Beyond the Text. Sounds logical or true in the real world, but the passage offers no support. Inference ≠ "true in general"; it means "supported by these lines."
• ⑥ Half-Right. Nails part of the task and quietly breaks the rest. Read to the end of every choice — the trap usually hides in the second half, often built from one of the traps above.

Don't force the label. Not every wrong answer is a tidy category — sometimes it's just unsupported. The taxonomy is a tool for the final 50/50, not a checklist for all four choices.

Craft & Structure (Types 1–3)

1 · Words / Vocabulary in Context

What it asks: Fill a blank with the most logical, precise word — or, in the second format, "what does X most nearly mean?" Either way you're choosing the sense that fits this context, not the word's headline dictionary meaning.

• 1. Cover the choices. Read the sentence with the word/blank hidden (plus the sentence before if needed).
• 2. Find the clue & predict. Highlight the phrase that constrains the meaning, then write your own one-word substitute — and its charge (+ or −).
• 3. Match charge first. Drop choices with the wrong charge before splitting hairs over shades.
• 4. Plug & test. Substitute survivors; keep the one that fits 100%. Never drop a word just because it's unfamiliar — eliminate the ones that only partly fit.

Strategy: On the "most nearly mean" format, treat the given word as a blank and predict your own — it's the foolproof guard against the most common decoy.

Trap: ⑤ Beyond the Text — a word merely associated with your prediction ("scared" when you wrote "cautious"; "outdated/bored" for the slang sense of "lame"). Also the most-common-meaning decoy: the answer is often the word's 3rd or 4th sense (e.g. "compromised" = weakened). And ② Extreme — "inventing" when the text only says "working with" data.

2 · Structure & Purpose

What it asks: Three sub-types — main purpose (why the whole text was written), function (what one underlined sentence does for the text), or overall structure (how the text is organized). The stem word — purpose, function, structure — tells you which.

• Main purpose: find two things — the central idea/topic, and any attitude expressed toward it. A structural clue ("As a result…", "This evidence…") usually flags the key sentence. Annotate "explain X + [tone]".
• Function: read the sentences right before and after the underlined one. A pronoun or transition ("That…", "However…") points to the part it connects with — highlight that, since the link is what reveals the role (example, evidence, qualification, counterpoint).
• Overall structure: locate the pivot where the focus switches (general→example, view→opposing view, cause→effect, beginning→ending). A time marker ("In 2015…") often marks it. Annotate the before/after in order.

Strategy: Purpose answers why, never just what — reject any choice that merely summarizes content. Authors don't announce their purpose, so your annotation does the bridging.

Trap: ④ Right Answer, Wrong Question — true of one sentence but not the whole text (main-purpose / structure), or describes a different sentence's job (function). ⑥ Half-Right multi-part choices ("appeals to economics, leisure, and religion"): one wrong element kills the whole choice. ③ Recycled Language reuses a real name/phrase in a false relationship.

3 · Cross-Text / Dual Texts

What it asks: Two short passages on one topic. Either "how would [author of one] respond to [a claim in the other]?" or "both would agree with which statement?" Whatever the wording, you're finding the single idea both texts touch.

• 1. Read the question first and highlight the named people/ideas, or the word "agree".
• 2. Read the claim-bearing text first. For a response question, read the text being responded to first; for an agreement question, read Text 1. Highlight its thesis in one sentence.
• 3. Read the other text and highlight where it touches the same point.
• 4. Annotate the relationship — agree / disagree and the reason, or the shared point (usually a single detail, not a sweeping claim).

Strategy: The relationship is often nuanced, not flat pro/con — one text may explain, qualify, or add evidence to the other. Agreement points tend to be narrow (a single piece of evidence both accept), so distrust grand "both believe…" statements.

Trap: ① Opposite — says one author would approve when in fact they disagree (or pins the disagreement on the wrong point). Mistaking an author's concession for their real view — they may raise an objection only to reject it. ② Extreme ("the sole component"), ③ Recycled Language stitching names together falsely. On tone, words like "antagonistic/flattering" overshoot a merely "ambivalent" stance.

Information & Ideas (Types 4–8)

4a · Central Ideas / Main Idea

What it asks: The single idea that every other sentence builds onto or off of — not a detail, the whole point.

• 1. Read the passage for the through-line. Ask "which sentence do all the others support?" It's often first or last, but can sit anywhere.
• 2. Don't over-weight one detail. If one line names a specific work but the rest cover a whole career, the career is the main idea.
• 3. Annotate a one-line summary when each sentence contributes roughly equally, then POE against it.

Trap: ④ Right Answer, Wrong Question — accurately states a real detail but misses the central focus. Also ⑤ Beyond the Text (it "becomes political" when politics was never claimed) and ③ Recycled Language.

4b · Detail / Retrieval

What it asks: "According to the text…" / "Based on the text, why did X happen?" Pure fact-finding — no why-the-author, no inference. The passage hands you the answer.

• 1. Highlight the name/detail in the stem.
• 2. Read for that thing and highlight what the passage actually says about it. Often you can go straight to POE.
• 3. Don't over-deduce — straightforward, but read carefully and don't grab an irrelevant detail.

Trap: ③ Recycled Language is the headliner here — wrong answers lift exact terms (a date, a name) and rebuild them into a statement the passage never made (e.g. tags the find to the "Jurassic" when the text said "Cretaceous"). ⑤ Beyond the Text ("migrated to improve survival" — true of animals generally, never stated).

5 · Command of Evidence — Textual (Claims)

What it asks: Which quotation/finding best illustrates, supports, or weakens a stated claim. The answer need not itself be supported by the passage — it just has to do the job to the claim, if true.

• 1. Read the stem for the verb — illustrate vs. support vs. weaken. Circle it; they read alike but ask opposite things.
• 2. Find & paraphrase the claim. Usually the last sentence (or, on a quotation-illustrate item, the phrase right before the blank). Frame it "X → Y, in setting Z."
• 3. Expect new info. The answer is fresh evidence applied to the claim, not a phrase pulled from the passage.
• 4. Test the variables. Reject any choice that shifts the variable, timeframe, or field — "right topic, wrong axis."

Trap: ⑥ Half-Right — when the claim has two parts, a choice satisfies one and ignores the other. ① Opposite — a "support" choice that quietly weakens. "Right topic, wrong axis" — same subject, wrong variable or wrong time period.

6 · Command of Evidence — Quantitative (Charts)

What it asks: Passage + a graph or table. Use the data to complete or support/weaken a statement. Same skill as Claims, plus reading a figure.

• 1. Read the stem — complete / illustrate / support / weaken?
• 2. Orient on the figure first: title, key/legend, variables, units, where each axis starts, and each trend's direction. Do this before the passage so the data doesn't bias you.
• 3. Find the claim in the passage. Most of the text is just setup; hunt for the sentence that references the same variables and makes a statement (often near the end). Highlight it.
• 4. POE against the passage FIRST, then the figure. Several choices will describe the chart accurately — only one also serves the claim.

Trap: ⑥ Half-Right is the signature trap — a choice reads the graph perfectly but is irrelevant to the claim (compares the wrong two groups, focuses on the wrong row, or describes a real trend the passage never tied to the claim). Also watch choices citing data the figure doesn't contain at all.

7 · Inferences / Conclusions

What it asks: "Which choice most logically completes the text?" The blank ends a short argument; the right fill-in is a synthesis of points already made — not a new fact.

• 1. Spot the topic (first sentence or two), then find the 2–3 key points made about it.
• 2. Mind the transition before the blank — "However / thus / yet" tells you whether the ending reverses or extends the prior idea. Structural clues drive the logic.
• 3. Synthesize & annotate. Combine the points into one broad, flexible prediction; don't lock onto rigid phrasing.
• 4. POE against highlighting + annotation.

Trap: ⑤ Beyond the Text is the headliner — the blank tempts you to invent any plausible conclusion, but it must be a summary of what's already there, not new. Also ① Opposite (polarity flip — "harming" vs. "helping"), ② Extreme ("impossible"), and ③ Recycled Language forging a cause-effect link the text never drew.

Expression of Ideas — the reading side

These rhetoric questions come after the reading block. They reward the same evidence-and-elimination habits, so they belong here. (The punctuation, agreement, verb-form, and modifier rules — Standard English Conventions — live on the Grammar page; this page doesn't repeat them.)

Rhetorical synthesis

What it asks: You're given bullet-point notes about a topic and must pick the sentence that best accomplishes a stated goal (e.g. "introduce the study to an audience unfamiliar with X", "emphasize a difference between the two methods").

• 1. Read the goal first and pin its exact demand — who's the audience, what must be emphasized, what relationship shown.
• 2. Treat the goal like a question stem. The right answer does the goal; true-but-off-goal choices are the standard distractor.
• 3. POE on goal-fit, not truth. Every choice may be factually accurate — accuracy isn't the test.

Trap: ④ Right Answer, Wrong Question — a true sentence drawn from the notes that simply doesn't meet the specified goal.

Transitions

What it asks: A blank between two sentences; choose the connector (However, Therefore, For example, Similarly, In contrast, As a result, Specifically, Nevertheless…).

• 1. Cover the choices. Read both sentences and name the relationship yourself — contrast, cause→effect, example, restatement, addition, sequence.
• 2. Then match. Pick the transition in that family. Don't let a smooth-sounding word override the logical relationship.
• 3. Beware "same-family" decoys. Several choices may all signal contrast; check direction and emphasis (e.g. "however" vs. "in fact" vs. "for instance").

Trap: ① Opposite — a contrast word where the second sentence actually continues the first (or vice versa). The most-common-sounding transition is frequently wrong.

Quick-reference recipes

The fixed routines for the types that most reward a drilled habit — condensed for last-minute review.

Words in Context

• 1. Cover the choices and read the sentence (plus the one before if needed).
• 2. Predict your own one-word substitute and its charge (+ / −).
• 3. Match charge first, then shades of meaning.
• 4. Plug & test; keep the 100% fit. Never drop an unknown word — cut the partial fits instead.

Purpose / Function

• 1. Identify the sub-type from the stem word (purpose / function / structure).
• 2. For function, read the neighboring sentences; a pronoun or transition points to the part it connects to.
• 3. Annotate why (idea + tone) — never settle for a choice that only says what.
• 4. POE; watch Right-Answer-Wrong-Question and Half-Right multi-part choices.

Textual Evidence / Claims

• 1. Circle the verb — illustrate / support / weaken.
• 2. Find the claim (last sentence, or before the blank); frame "X → Y, in setting Z."
• 3. Expect new info — the answer is evidence applied to the claim, not a recycled phrase.
• 4. Test variables; reject right-topic-wrong-axis and any choice that flips support↔weaken.

Quantitative / Charts

• 1. Orient on the figure first — title, legend, variables, units, axis starts, trend directions.
• 2. Find the claim in the passage (the sentence that references the same variables).
• 3. POE against the passage FIRST, then the chart — many choices fit the chart but not the claim.
• 4. Stay inside the data; reject true-but-irrelevant trends and data the figure doesn't contain.

Cross-Text / Dual Texts

• 1. Read the question first — track agreement, disagreement, or a subtler link.
• 2. One sentence per text for each thesis.
• 3. Separate view from concession — an objection an author rejects isn't their position.
• 4. Pin the precise relationship; avoid tone words that overshoot a mixed stance.

Conclusions

• 1. Find the 2–3 key points about the main topic; mind the transition before the blank.
• 2. Synthesize them into one broad prediction — a summary, not a new fact.
• 3. POE; Beyond-the-Text is the chief trap — no inventing plausible-sounding conclusions.

Section-wide strategy

• Pace. 32 minutes ÷ 27 questions ≈ 71 seconds each. Don't camp on any one question past 90 seconds — flag and move.
• Eliminate first. Two wrong answers are easier to spot than the right one. Cut to 50/50, then commit.
• Trust the passage. The SAT does not require outside knowledge. If your answer relies on something the passage doesn't say, you're wrong.
• Use the mark-for-review. Better to flag a hard question and come back than to spend 3 minutes on it and lose 2 others.
• Don't change answers without a reason. First-instinct answers are right more often than second-guessed ones — change only if you spot a concrete reason.
• Read at a talking pace (~100–120 wpm). No skimming needed — passages are under 150 words. There's no prize for finishing early; spend leftover time re-reading the hardest texts (especially poetry).
• Pick your battles. If time is short, prioritize the quick types (Vocabulary, Inferences) and guess on the rest — there's no penalty for wrong answers, so never leave a blank.
• Order is your friend. Types come in fixed clusters, each ramping easy→hard. Start with the cluster you're fastest at; you can do the grammar block before the reading block if reading is your weak spot.
• Ignore the experimental questions. A few RW questions are unscored pretest items. If one feels strange, answer it and move on — don't let it rattle you. The second module's difficulty adapts to your first; stay in the moment.