TOEIC Link Reading Paraphrase Recognition: The Five Paraphrase Families ETS Reuses Across Every Part 7 Form

The single most reliable Part 7 pattern across every released TOEIC Link form is this: the correct answer is never a verbatim repeat of the passage, and at least one wrong answer almost always is. This pattern alone — if you can recognize it in real time — eliminates one distractor per question on roughly 80% of Part 7 items. The remaining three choices then split into a correct paraphrase and two near-paraphrases that fail on one specific swap.

This article is the five-family classification system that names the paraphrase patterns ETS uses repeatedly. Every paraphrase in the correct-answer position belongs to one of five families: lexical substitution, syntactic transformation, scope generalization, scope specification, and inference compression. Every wrong answer that survives the verbatim-repeat filter belongs to one of three trap families: scope mismatch, polarity flip, or out-of-passage detail. Naming the family lets you process each choice in two to three seconds instead of the eight to ten seconds candidates spend re-reading the passage.

Why paraphrase recognition is the Part 7 bottleneck

Three structural facts make Part 7 a paraphrase test rather than a reading test in the conventional sense.

Fact 1 — the passage is almost always longer than your working memory. A single Part 7 passage runs 150 to 300 words. By the time you reach the questions, you cannot hold the exact wording of every sentence. What you can hold is the meaning. ETS exploits this gap by writing answer choices that share meaning with the passage but use different wording, alongside distractors that share wording but distort the meaning.

Fact 2 — the question stems themselves are paraphrases. "What is suggested about the company?" is a paraphrase of "What does the passage imply about the company?" which is a paraphrase of "Based on the article, what can be inferred about the company?" The question itself is a translation of a generic stem into the passage's domain. Recognizing the stem family — fact-retrieval, inference, vocabulary-in-context, purpose, attitude — is the first paraphrase recognition you do.

Fact 3 — the score band where Part 7 weight is heaviest is the B2-to-C1 region. ETS targets paraphrase recognition specifically because it discriminates sharply between candidates who can decode English at a sentence level and candidates who can compress meaning across sentences. A B1 candidate matches words. A C1 candidate matches meaning. Paraphrase recognition is the single most direct test of which one you are.

For the broader Part 7 framework these items sit inside, the per-section playbook in our TOEIC Link reading strategies by question type guide covers timing targets and the order to attack the question types. This article goes deeper on the paraphrase mechanic that sits inside each Part 7 question.

The five correct-answer paraphrase families

Family 1 — Lexical substitution

The simplest paraphrase. A content word from the passage is replaced by a synonym or near-synonym in the answer. "The company announced layoffs" becomes "the firm reduced its workforce." "Quarterly revenue grew 12%" becomes "sales increased by roughly one-eighth in the quarter."

The diagnostic is that the sentence structure and the truth conditions are unchanged. Only the lexical items have moved. If the answer choice has the same skeleton as a passage sentence with synonyms in the slot positions, it is a Family 1 paraphrase and almost certainly correct.

This family accounts for roughly 25% of correct answers across our analyzed sample. It is the easiest family to recognize and the most common starting point for B2 candidates.

Family 2 — Syntactic transformation

A passage sentence is restructured grammatically without changing truth conditions. Active voice becomes passive. A relative clause becomes a separate sentence. A coordinated noun phrase becomes two clauses. A "because" clause becomes a participial phrase.

"The CEO will announce the merger next Tuesday" becomes "The merger announcement is scheduled for next Tuesday." "Because demand grew, the team expanded" becomes "Growing demand led to team expansion."

The diagnostic is that the content words are mostly unchanged but the grammatical relations between them have shifted. Candidates who read for word-level overlap miss this family because the choice does not look similar enough. Candidates who read for meaning catch it instantly.

This family accounts for roughly 20% of correct answers. It is the family that B1 candidates miss most often.

Family 3 — Scope generalization

A specific detail from the passage is restated at a higher level of abstraction. "Three employees were promoted to senior roles" becomes "Several staff members advanced in the organization." "The report identifies five key risks" becomes "The report covers multiple risk factors."

The diagnostic is that the answer choice would still be true even if the passage's specific number or instance were slightly different. The generalization is consistent with the specific, so the truth condition is preserved.

This family accounts for roughly 20% of correct answers. The trap that lives next to this family is over-generalization — a wrong-answer that generalizes past what the passage supports. The discipline is to ask whether the generalization stays within the passage's logical scope.

Family 4 — Scope specification

The reverse direction. The passage states something at a higher level, and the answer choice specifies one instance. "The company offers multiple payment options" becomes "Customers can pay by credit card." "The training program covers core compliance topics" becomes "Employees learn about data-protection rules."

The diagnostic is that the answer choice is one valid instance of the passage's general statement. This family is rarer than Family 3 — roughly 10% of correct answers — because ETS uses it more often in distractors than in correct answers. The trap version specifies an instance that the passage does not actually mention.

The discipline is to verify the specific instance is one the passage explicitly covered, not one you could plausibly infer from the topic.

Family 5 — Inference compression

The hardest family. The answer compresses multiple passage sentences into a single conclusion that no individual sentence stated. "The product launched on schedule. Sales exceeded forecasts. The board approved a follow-on release." compresses into "The launch was successful." The conclusion is not stated; it is what the three sentences together support.

The diagnostic is that the answer cannot be matched to any single sentence but is consistent with the combination of two or more. This family appears mostly in inference and main-idea questions, which together account for roughly 25% of Part 7 items.

The trap version of this family is the plausible inference — an answer that sounds like it should follow from the passage but actually requires information the passage does not provide. The discipline is to ask whether every step of the inference chain has explicit support in the text.

For the underlying skill of identifying inference questions versus fact-retrieval questions in the first place, the question-type taxonomy in our TOEIC Link reading strategies by question type playbook is the upstream reference.

The three trap families

Recognizing the correct paraphrase is half the work. The other half is recognizing the traps that ETS plants alongside.

Trap 1 — Scope mismatch

The wrong answer reuses passage wording but changes the scope. The passage says "the marketing team adopted the new tool"; the trap says "the entire company adopted the new tool." The passage says "during the third quarter"; the trap says "throughout the fiscal year." The lexical overlap is high; the truth condition is wrong.

The diagnostic is to check the quantifiers and time expressions explicitly. Any answer that swaps "some" for "all," "team" for "department," "quarter" for "year," or "trial" for "permanent" should be challenged against the passage's exact phrasing.

Trap 2 — Polarity flip

The wrong answer matches the passage exactly except for a negation, an opposite adjective, or a contrast conjunction. The passage says "sales improved despite the headwind"; the trap says "sales declined because of the headwind."

The diagnostic is to read the answer choice through twice, the second time specifically looking for negators, "not," contrast conjunctions, and antonym swaps. Polarity flips are the easiest traps to miss when reading fast and the easiest to catch when reading slowly. The fix is to slow down on the second read of each surviving choice.

Trap 3 — Out-of-passage detail

The wrong answer states a detail that is plausible for the topic but is not in the passage. The passage is about a company's product launch; the trap mentions the company's stock price, which sounds related but is not in the text.

The diagnostic is the verifiability check. For every surviving answer choice, ask which sentence in the passage supports it. If you cannot point to a specific sentence, the answer is out-of-passage even if it sounds reasonable.

This trap is the most common cause of point loss on inference questions, because B2 candidates extrapolate from the topic instead of staying anchored to the text. The fix is the verifiability check on every inference answer before selecting.

The two-pass reading protocol

Apply the families above in two passes per question.

Pass 1 — eliminate. Read the four choices. Eliminate any verbatim repeat of passage wording. Eliminate any choice that uses passage wording with a quantifier or polarity swap. Eliminate any choice that mentions something not in the passage. This pass alone typically reduces the field from four to two.

Pass 2 — verify. For the remaining two choices, identify which paraphrase family each belongs to. The family that matches the question type — fact-retrieval pairs with Families 1 and 2, inference pairs with Family 5, main-idea pairs with Family 3 — is almost always the correct answer.

Total time per question, after twenty hours of drill on this method, runs forty to fifty seconds. Without the method, candidates spend seventy to ninety seconds on the same question and still miss roughly one in three.

How to drill

Practice with ten Part 7 questions per session, three sessions per week, for three weeks. After each session, classify your wrong answers by trap family — scope mismatch, polarity flip, or out-of-passage. The classification will concentrate in one trap family, and that family's diagnostic check is what the next session focuses on.

After three weeks, your Part 7 accuracy on practice forms should rise from roughly 60% to roughly 80%. If it does not, the bottleneck is upstream of paraphrase recognition — usually vocabulary depth in the passage topics. See the TOEIC Link vocabulary essentials overview for the parent vocabulary framework, or the business-domain clusters in TOEIC Link business email vocabulary cluster for the most-tested topic area.