Grammar, Usage and Vocabulary — In Context, Because That Is How CAT Asks

CAT does not ask a grammar question. There is no spotting-errors item, no sentence improvement, no fill in the blank, and there has not been one since the section was rebuilt in 2015 — so a candidate who prepares grammar as a rule sheet is preparing for a paper that is not the one they will sit. Grammar and vocabulary still decide marks here, and they do it inside the four question types that exist: they let you find the main clause of a sixty-word sentence, resolve what this refers to two sentences later, read the connector that fixes a para-jumble's order, and catch the one qualifier — all, only, necessarily — that makes a plausible inference option false. This chapter teaches those four uses and nothing else. If you want the rule sheet, it belongs to the papers that ask for one, and this section is not among them.

The Four Places a Grammatical Fact Decides a Mark

Section I gives you 24 questions in 40 minutes, of which about sixteen are reading comprehension and the rest are para summary, odd sentence out and para-jumbles. Not one of them names a grammatical term. But every one of them is a sentence-level reading task, and sentence-level reading is grammar being used rather than recited. The table below is the whole of what this chapter claims: four recurring decisions, each of which a candidate makes faster and more reliably if they can see the structure of a sentence.

Where grammar and vocabulary are actually tested in Section I
The decisionThe grammar it usesWhere it appears
What is this sentence actually claiming?Finding the main subject and main verb under the modifiers.Every RC passage, and the source sentences of a para summary.
What does this / these / it refer to?Pronoun and demonstrative reference, and number agreement with the antecedent.RC inference options, and the ordering of a para-jumble.
Which sentence must come after which?Connectors and their logical direction — contrast, addition, consequence, concession.Para-jumbles, para-completion, odd sentence out.
Is this option too strong for the passage?Quantifiers and modals — all, most, some, only, must, may, necessarily.Inference, assumption and "which of the following can be concluded" options.
🎯 Why the rule sheet is the wrong preparation here
A rule sheet answers "is this sentence correct?" — a question CAT stopped asking a decade ago. The four decisions above answer "what does this sentence mean, and what follows from it", which is every question the section does ask. The practical consequence is a reallocation: the hours a candidate would have spent on subject-verb agreement drills are better spent reading dense non-fiction and, on each hard sentence, naming its main clause aloud.

Stripping a Long Sentence to Its Claim

A CAT passage sentence is long because it is loaded with modifiers, not because its claim is complicated. The drill is mechanical: cross out every prepositional phrase, every participial phrase, and every clause set off by commas or dashes, and see what is left. What is left is a subject, a verb and usually an object, and that is the claim. Take: Although the committee, which had been convened in haste after the first audit, recommended sweeping changes to the procurement rules, the ministry, citing the cost of retraining, adopted only two of them. Strike the concessive clause, the relative clause and the participial phrase and you have the ministry adopted only two — which is the sentence, and which is also the half an inference question will test.

  1. Find the finite verb that is not inside a comma-bracket or a relative clause. That is the main verb.
  2. Ask who or what performs it. That is the main subject — and it is often several words away from the verb.
  3. Read the concessive word — although, while, despite, notwithstanding. Whatever follows it is the part the author is conceding, not asserting.
  4. Note every quantifier left standing in the core — only two, most, at least — because those are the words an option will quietly change.
🧠 The concession is not the claim
In Although X, Y, the author is committed to Y and is merely granting X. Options that state X as the passage's position are the most frequently chosen wrong answers in inference sets, because X was in the text and the candidate remembers reading it. The same asymmetry runs through while, admittedly, no doubt and it is true that — each one marks the half you are allowed to disagree with.

Connectors and Reference: the Two Things That Order a Jumble

A para-jumble is not solved by finding the "topic sentence". It is solved by links: a connector tells you what kind of sentence must precede it, and a reference tells you which specific sentence must precede it. A sentence beginning That said must follow a sentence making the point being softened; a sentence containing these two failures must follow the sentences that named two failures; a sentence whose subject is a bare pronoun cannot open the paragraph, because there is nothing for it to refer back to.

What each connector tells you about the sentence before it
ConnectorRelationSo the previous sentence must…
However, by contrast, yet, on the other handContrast…state the opposite side, and state it as a claim rather than as a question.
Moreover, furthermore, in additionAddition in the same direction…already make a point of the same sign; these never reverse a direction.
Therefore, hence, as a result, consequentlyConsequence…contain the cause or the evidence, not the conclusion.
That said, admittedly, to be sure, no doubtConcession before a qualification…make the strong claim that is about to be softened.
For instance, in particularIllustration…state the generalisation the example is an example of.
⚠️ A demonstrative is a stronger link than a connector
Connectors constrain the type of the preceding sentence; a demonstrative plus a noun — this reversal, such schemes, these two objections — names it. When both are available, order by the demonstrative first: several sentences may qualify as "a contrast", but usually only one names two objections. And a plural demonstrative is a counting instrument: these two rules out every arrangement in which only one has been mentioned so far.

Qualifiers, and Vocabulary You Are Not Meant to Recall

Most wrong options in an inference set are wrong by one word. The passage says some, the option says all; the passage says contributed to, the option says caused; the passage says may, the option says will. Read every option twice — once for its content and once for its quantifier and its modal — and the set gets faster rather than slower, because a quantifier failure is decidable without rereading the passage. Vocabulary works the same way: CAT does not ask you to recall a definition, it embeds an unfamiliar word in a sentence whose structure gives it away.

  • Contrast gives the opposite. Far from being desultory, her reading was systematic — whatever desultory means, it is the opposite of systematic, and that is enough to answer.
  • Parallel gives the synonym. The scheme was moribund, its offices unstaffed and its funds unspent — the two clauses after the comma define the word before it.
  • A prefix is evidence, not proof. In-, ir- and un- usually negate, so ineluctable is unlikely to mean avoidable; but invaluable does not mean worthless, so check the sentence before trusting the prefix.
  • Tone words carry direction. Concede, purport, ostensibly, so-called signal the author's distance from what follows, and a purpose-and-tone question is decided by three or four such words.
🧠 The one-word test, in ten seconds
Before comparing two surviving options on content, compare their strongest words. If one says only, all, never, must or proves and the other says among, many, rarely, may or suggests, the weaker option is right far more often than not — because a passage that hedges cannot support a claim that does not. The exception is a passage that itself argues for a universal, which is rare and always signposted.

Key takeaways

  • CAT has asked no standalone grammar question since 2015 — no spotting errors, no sentence improvement, no fill in the blank. Grammar earns marks only inside the four question types the section does have.
  • Strip a long sentence by crossing out prepositional and participial phrases and comma-bracketed clauses; what remains is subject, verb, object — the claim. In "Although X, Y", the author asserts Y and merely concedes X.
  • A connector constrains the type of the preceding sentence; a demonstrative plus noun names it. Order a jumble by the demonstrative first — "these two objections" rules out every arrangement in which only one has been mentioned.
  • Most wrong inference options are wrong by one word: some becomes all, contributed becomes caused, may becomes will. Between two survivors, the more hedged option is usually right.

Practice questions (8)

Attempt each one before opening the answer. Every explanation names the tempting wrong option as well as the right one, because that is where marks are lost.

  1. How does the CAT VARC section test grammar?

    1. Through spotting-errors and sentence-improvement questions, two or three a paper
    2. Only inside the existing question types — parsing a sentence, resolving a reference, reading a connector, weighing a qualifier
    3. Through fill-in-the-blank questions on prepositions and articles
    4. It does not test grammar at all, so grammar has no bearing on the score
    Show answer

    Answer: B — Only inside the existing question types — parsing a sentence, resolving a reference, reading a connector, weighing a qualifier

    Since the section was rebuilt in 2015 there has been no standalone grammar item, but grammar is used in every question that remains: finding the main clause of a long sentence, deciding what a demonstrative refers to, ordering a jumble by its connectors, and rejecting an option whose quantifier is stronger than the passage. The first and third options describe the format of other objective English papers — SSC-style spotting errors, bank-style blanks — and they are the reason candidates mis-prepare. The last option is the opposite error and the more expensive one — grammar is not asked as a question type, which is different from having no bearing, and a candidate who cannot parse a sixty-word sentence loses marks across the whole section rather than on two items.
  2. "Although the panel, appointed after two decades of litigation, urged the state to compensate every displaced family, the government, pleading a fiscal crunch, paid only the landholders." The author is committed to which claim?

    1. That every displaced family should be compensated
    2. That the government paid only the landholders
    3. That the litigation lasted two decades because of the panel
    4. That the fiscal crunch was genuine
    Show answer

    Answer: B — That the government paid only the landholders

    Strip the modifiers and the main clause is "the government paid only the landholders" — that is the assertion. The panel's recommendation sits inside the concessive clause introduced by "Although", so the author reports it without endorsing it, and stating it as the author's position is the classic concession error. The litigation detail is inside a participial phrase and carries no causal claim; reading one in is an inference the sentence does not license. And "pleading a fiscal crunch" attributes the reason to the government rather than asserting it — "pleading" is precisely the verb an author uses to keep their distance from a stated reason.
  3. In a para-jumble, one sentence begins "These two exemptions, however, were never notified." What does that sentence tell you about its position?

    1. It can open the paragraph, since "however" often begins a paragraph
    2. It must follow sentences in which two exemptions have already been named, and it reverses their direction
    3. It must be the last sentence, because it reports a failure
    4. It must immediately follow the sentence containing the word "exemption", wherever that is
    Show answer

    Answer: B — It must follow sentences in which two exemptions have already been named, and it reverses their direction

    The sentence carries two links at once and they agree: the plural demonstrative "These two exemptions" requires that two exemptions have been introduced before it, and "however" requires that what came before ran in the opposite direction — the exemptions were announced or welcomed, and this sentence undercuts that. A sentence with a backward-pointing demonstrative cannot open a paragraph, whatever its connector, which disposes of the first option. Nothing makes it final; a paragraph can continue after an undercutting sentence. And the fourth option overreaches: the demonstrative requires two exemptions to have been named, not adjacency to any sentence that happens to use the word.
  4. A passage says: "Rainfall variability contributed to the decline of several settlements in the region, though the timing differs from site to site." Which option is too strong to be an inference?

    1. Rainfall variability was one factor in the decline of some settlements in the region
    2. Rainfall variability caused the simultaneous collapse of all settlements in the region
    3. The settlements in the region did not all decline at the same time
    4. Factors other than rainfall may also have been involved
    Show answer

    Answer: B — Rainfall variability caused the simultaneous collapse of all settlements in the region

    The option fails on three words at once, and any one of them would be enough: "caused" upgrades "contributed to", "all" upgrades "several", and "simultaneous" contradicts the clause saying the timing differs from site to site. The first option is a faithful restatement with both hedges intact. The third is licensed directly by "the timing differs from site to site". The fourth follows from "contributed to", which by its own meaning leaves room for other factors — a word that would be wrong if the passage meant sole causation. This is the one-word test doing all the work: none of these options needs the passage reread, only its quantifiers and its causal verb compared.
  5. "Far from being desultory, her reading was systematic: she worked through each author chronologically." In this sentence, desultory most nearly means:

    1. Thorough and disciplined
    2. Unmethodical, moving without plan from one thing to another
    3. Rapid and impatient
    4. Confined to a single subject
    Show answer

    Answer: B — Unmethodical, moving without plan from one thing to another

    "Far from being X, her reading was systematic" sets X against systematic, so X means unsystematic — and the colon then shows what systematic amounted to, which confirms the axis. No recall is needed, and that is the point: CAT embeds the unfamiliar word in a structure that gives it away. "Thorough and disciplined" is the trap for a candidate who reads the sentence as elaborating rather than contrasting, which "far from" forbids. "Rapid and impatient" borrows a plausible connotation from the sound of the word but is not the opposite of systematic. "Confined to a single subject" is on a different axis altogether — breadth rather than method — and the sentence says nothing about it.
  6. A sentence in a jumble begins with "Therefore," Which sentence must precede it?

    1. One stating the conclusion, which this sentence then explains
    2. One stating the cause or the evidence from which this sentence follows
    3. One stating the opposite position
    4. Any sentence at all, since "therefore" is a weak link
    Show answer

    Answer: B — One stating the cause or the evidence from which this sentence follows

    "Therefore" marks a consequence, so what precedes it must be the ground the consequence rests on — the cause, the evidence or the premise. The first option inverts the direction and is the commonest error, because a conclusion feels like the important sentence and candidates place it first out of habit. "The opposite position" is what a contrast connector such as "however" would require, and swapping the two is the single most frequent jumble mistake. And a consequence connector is one of the strongest links available: it constrains the previous sentence to a particular logical role, which is exactly what makes jumbles solvable without guessing.
  7. "The report, whose authors the ministry had itself nominated, and which drew on surveys conducted in four states over three years, was tabled but not discussed." Which is the main subject and main verb?

    1. authors / had nominated
    2. The report / was tabled
    3. surveys / conducted
    4. ministry / had nominated
    Show answer

    Answer: B — The report / was tabled

    Both "had nominated" and "drew on" sit inside relative clauses introduced by "whose" and "which", so neither can be the main verb however prominent it looks; "conducted" is a participle modifying surveys and is not finite at all. Removing all three leaves "The report … was tabled but not discussed", which is the sentence. The ministry option is the strongest distractor because the ministry is the grammatical subject of a verb and is the only agent in the sentence that does anything deliberate — but it is the subject of a subordinate clause, and mistaking a subordinate agent for the main subject is exactly how a candidate ends up answering an inference question about the ministry when the passage was about the report being ignored.
  8. Two inference options survive your first pass. One reads "the policy was the only reason exports recovered"; the other reads "the policy was among the factors in the recovery of exports". The passage said the policy "played a part in" the recovery. Which do you mark, and why?

    1. The first, because a passage that credits a policy is asserting its importance
    2. The second, because "played a part in" leaves room for other factors and "only" removes it
    3. The first, because it is more specific and specificity is rewarded
    4. Neither, because the passage supports no inference about causes
    Show answer

    Answer: B — The second, because "played a part in" leaves room for other factors and "only" removes it

    "Played a part in" is a hedge with a precise meaning: one contributing factor among possibly several. The second option preserves exactly that; the first replaces it with sole causation, which the passage neither states nor implies, and a single word is enough to make an option unsupported. The first and third options share the same underlying mistake — treating strength or specificity as a virtue in an inference option, when an inference must be weaker than or equal to what the text says. And the fourth is over-correction: the passage does support a causal inference, just a modest one, and refusing every causal option costs marks the section hands out for reading a hedge accurately.