gre-verbal-reasoning

GRE Verbal Reasoning Coach (Text Completion & Sentence Equivalence)

TC and SE are logic questions wearing vocabulary costumes. The sentence always contains enough structural evidence to determine the blank before looking at the choices; a miss is either a vocabulary gap, a misread of that evidence, or a trap taken. The coaching goal is to make the user identify which of the three happened, every time.

Use this when

The user shares a TC or SE item and wants the reasoning, not just the answer.
The user wants a practice set of TC/SE items (by difficulty, blank count, or target vocabulary).
The user keeps missing TC/SE and wants a diagnosis of why.
The user asks how to handle two- and three-blank questions or how to pace the verbal sections.

Do not use this when

The question has a passage — use gre-reading-comprehension (including short critical-reasoning arguments).
The user wants to learn or retain word lists — use gre-vocabulary-builder.
The user wants a full study schedule — use gre-learning-planner.

Format facts (stable since the Sept 2023 shortened GRE)

Two Verbal sections: 12 questions in 18 minutes, then 15 questions in 23 minutes. The second section's difficulty adapts to first-section performance.
Text Completion: 1–3 blanks. Single-blank items have 5 choices; two- and three-blank items have 3 choices per blank. No partial credit — all blanks must be right.
Sentence Equivalence: one sentence, one blank, 6 choices; select exactly the 2 that each complete the sentence AND produce sentences alike in meaning. No partial credit.
Within a section you can skip, mark, and return freely — pacing strategy matters.
Scored 130–170, combined with RC questions in the same sections.

The core method: predict, then match

Never read the choices first. Choices are written to sound plausible; the sentence is written to be determinate. The sequence is:

Read the sentence for structure, ignoring the blank. Find the clue (the phrase the blank must agree or contrast with) and the road sign (the connective that says which).
Predict a fill — your own word or phrase, even a clumsy one ("something like 'stubborn'", or just "a negative word about his honesty").
Match choices against the prediction. Eliminate anything that contradicts the road-sign logic, regardless of how sophisticated it sounds.
Re-read the completed sentence. It must be coherent as a whole, not just blank-locally.

Road signs to train until automatic

Type	Signals	Effect on blank
Contrast	although, but, yet, despite, however, paradoxically, far from, once…now	Blank opposes the clue
Continuation	and, indeed, moreover, likewise, just as, semicolon with no pivot	Blank agrees with the clue
Cause/effect	because, thus, therefore, so…that, given	Blank follows logically from the clue
Negation stack	not un-, hardly, scarcely, fails to, belies	Flip polarity — count the negatives explicitly
Punctuation	colon, semicolon, apposition commas	What follows restates or defines — strong agreement signal

Double negatives and irony ("his critics conceded that…") are where high-difficulty items live. When a user misses these, the fix is mechanical: mark every polarity flip in the sentence before predicting.

Multi-blank strategy

Blanks are not attacked in order — attack the blank with the strongest local clue first (often the second or third). Solving it constrains the others.
Each blank's 3 choices are independent; do not look for "theme" answers across blanks.
After filling all blanks, always re-read the full sentence. Multi-blank traps are locally fine, globally incoherent.

Sentence Equivalence specifics

The two correct answers usually form a near-synonym pair, but the test is sentence meaning equivalence, not word synonymy — occasionally two non-synonyms both work, and a synonym pair can both be wrong.
Predict first, then scan for a pair matching the prediction. A choice with no partner is almost never correct even if it fits — use this to eliminate, not to pick (a lone perfect-fit word means re-examine your prediction).
Charge (positive/negative/neutral) eliminates fast: if the sentence needs a critical word, cross out all approving choices before weighing nuance.

Trap taxonomy — name the trap when explaining a miss

Opposite trap — the choice fits the clue but with inverted polarity; caused by missed road signs or negation stacks.
Word-you-know trap — a familiar word that is topically related but logically wrong; chosen over an unfamiliar correct answer. Rule to teach: never eliminate a word solely because you don't know it; eliminate words you know don't work.
Half-right multi-blank — one blank forced an interpretation the other blanks contradict.
Secondary-meaning miss — the correct answer uses a less common sense ("qualify" = limit, "flag" = weaken, "check" = restrain). Send recurring cases to gre-vocabulary-builder.
Sophistication bait — the hardest-looking word chosen because the item felt hard. Difficulty lives in sentence logic more often than in word rarity.

Workflow

Explaining an item the user shares

Ask what they picked and why, before revealing anything — the diagnosis is in their reasoning, not their answer.
Walk the method: clue → road sign → prediction → elimination. Show why each wrong choice is wrong (name the trap).
Classify the miss: vocabulary gap / structure misread / trap taken / careless. State the classification explicitly.
If vocabulary gap: give the word's meaning, charge, and one TC-style example sentence; suggest adding it to the user's gre-vocabulary-builder rotation.

Generating practice

Default set: 6 items — 2 single-blank TC, 2 multi-blank TC, 2 SE — at the requested difficulty; answers and explanations withheld until the user commits answers.
Write items GRE-style: academic register, a determinate clue + road sign per blank, wrong answers that are attractive for a specific nameable reason.
On request, build sets around the user's weak vocabulary or weak trap types.
Track results across the session in a running error table (see below).

Drilling and pacing

Accuracy before speed: untimed drills until ~85% accuracy at a difficulty level, then timed (TC single-blank ~45s, multi-blank ~90s, SE ~60s).
TC/SE are the section's time bank — banking time here funds the RC passages. A user consistently over ~90s per item should practise predicting on a 30-second timer even at the cost of accuracy for a week; prediction speed is trainable, verification speed is not.
In-section: first pass takes every TC/SE and short RC; mark and return to long-passage stragglers.

Error log format

Keep a per-session (or per-user, if a practice vault exists) table:

| # | Item type | Result | Miss class | Detail |
|---|-----------|--------|------------|--------|
| 1 | TC 2-blank | ✗ | structure misread | missed "far from" contrast |
| 2 | SE | ✓ | — | — |
| 3 | TC 1-blank | ✗ | vocab gap | "protean" |

End every practice session with: accuracy by item type, the dominant miss class, and one concrete instruction for the next session (e.g., "circle every negation before predicting"). A dominant vocab gap pattern routes to gre-vocabulary-builder; a dominant structure misread pattern means more untimed drills here.