Transitions

If you can already say what a transition does that a comma cannot, start at “The method” below; nothing later in this lesson depends on this block. It is here because Transitions is the one Reading & Writing skill where a student can understand every sentence in the passage, know every word in every choice, and still miss the question — and understanding how that is possible is most of the fix.

What the item looks like. A short passage — usually forty to ninety words, drawn from science, history, the humanities, or a student’s research notes — carries one blank, almost always at the start of a sentence. Four choices offer connective words or phrases: However, Therefore, For example, In addition. The stem never varies: which choice completes the text with the most logical transition. Note the word logical. Not smoothest, not most formal, not most academic-sounding. The test names its criterion in every single instance of the question, and the criterion is logic.

The fact that changes what you do: every one of the four choices produces a grammatical English sentence. That is not sloppy item-writing, it is the design. Grammar cannot decide these, and neither can register — all four options sit at the same level of formality. Once grammar and register are off the table, the only thing left that can separate four choices is whether the relationship each one asserts is actually true of the two ideas in front of you.

So: what a transition IS. Not a vocabulary item with a meaning that gets dropped into a slot. A transition is an ASSERTION about how two ideas relate. “However” asserts that what follows differs from what came before. “Therefore” asserts that what follows was produced by what came before. “For example” asserts that what follows is one member of a class the previous sentence described. If the assertion is false about the passage in front of you, the word is wrong — however naturally it reads, and it will read naturally.

The anatomy, and where the points go. Every one of these items has three parts: IDEA A, the material before the blank; the CONNECTIVE, which is the blank; and IDEA B, the material after it. Because the blank almost always opens a sentence, idea B is the whole sentence containing it, full stop to full stop — including every clause after a comma. Reading idea B only as far as the first comma is the most expensive habit on this question type, because the digital SAT routinely puts the reversal in the second clause.

Six words the answer choices depend on, worth having straight before the taxonomy. A CLAIM is something asserted. A REASON is offered in support of a claim. An INSTANCE is one member of a class — a particular drawn from a general. A RESULT is produced by something else. A CONCESSION grants a point that works against your own line, usually just before you turn back to it. A QUALIFICATION keeps a claim but narrows where it applies. On the hardest items, the entire difference between two surviving choices is which one of these six the passage actually contains.

How much this is worth, stated honestly. College Board publishes the domain figure: Expression of Ideas is about 20% of Reading & Writing, which the official framework puts at roughly eight to twelve questions across the two modules. It does not publish a split between the domain’s two skills, Transitions and Rhetorical Synthesis. Every “transitions appears N times per test” number you will encounter, including any you see from us, is a prep-industry estimate reverse-engineered from released forms. Treat it as several questions per test, reliably — enough that the gap between reliable and unreliable here is worth real points, and not enough to justify memorising the taxonomy below rather than using it as a lookup table.

Step 1 — cover the choices. On screen that means looking away from the options, or putting a hand over that part of the display, and it is worth doing literally for the first fifty items you practise, until the habit no longer needs the hand. The reason this is not merely a discipline exercise is in the mechanism block below.

Step 2 — compress idea A. Read the material before the blank and reduce it to a few words you could say out loud. Not a summary of the passage — just what the sentence immediately before the blank claims. “Lidar found tens of thousands of buildings.” “The code has been rewritten after every quake.”

Step 3 — read idea B to the full stop, then compress it too. All of it. If the sentence has a clause after a comma, that clause is part of idea B, and it is where the turn lives more often than anywhere else on this question type. A prediction built from the first half of idea B is worse than no prediction, because it will be held just as confidently as a complete one.

Step 4 — name the relationship in plain English, before any choice is visible. Use ordinary words. “B is what A caused.” “B is one case of A.” “B goes against A.” “B is A said harder.” “B explains why A.” “B replaces A.” Crude is fine; the prediction does not have to be in the register of the answer choices. It has to be committed to before they are visible.

Step 5 — uncover, then sort rather than select. Assign each of the four choices to a family from the taxonomy below. On most items three of the four families are irrelevant and can be discarded without any further thought about the passage at all — that is the speed this method buys, and on a timed module the speed matters as much as the accuracy.

Step 6 — if two survivors share a family, split them on the extra constraint the stronger word carries. “However” says only that two things differ. “Nevertheless” says the second holds despite the first. “On the contrary” says the first is being denied and the second is its correction. Each constraint is a fact about the passage, so it gets settled by going back to the passage, never by ear.

Step 7 — verify. Read A, the connective, and B as one joined stretch, and check that the pair now asserts exactly the relationship named at Step 4. Four seconds, and it catches the two errors elimination cannot: a survivor never positively checked, and a prediction built from half of idea B.

ADDITION — B adds a further point in the same direction as A. Members: also, in addition, additionally, moreover, furthermore, besides, what is more. Inside the family: also / in addition / additionally are flat, adding without weighting; moreover / furthermore add a further and usually heavier point, carrying a faint “and this one matters more”; besides is dismissive-additive (“and anyway”) and is rare on this test.

PARALLEL — B is a distinct case that matches A. Members: similarly, likewise, in the same way, by the same token. The constraint that decides this family, and the one most often missed: a parallel needs two cases. If A and B are about the same subject, “similarly” is wrong however alike the two statements sound — what you have there is addition.

CONTRAST — A and B differ, and nothing stronger is claimed. Members: however, in contrast, by contrast, on the other hand, alternatively, whereas, meanwhile (weakly). This is the default of the opposition families. If the most that can be said is “these two are not the same,” this is the family, and there is no credit for reaching into a neighbouring one for a more impressive word.

CONCESSIVE CONTRAST — B holds DESPITE A. Members: nevertheless, nonetheless, even so, still, all the same, that said, at the same time. The constraint: A must give a reason to expect B not to happen. “The bridge was built in 1846. Nevertheless, it carries freight trains.” Age predicts frailty; the freight defies the prediction. Remove that expectation and the word has nothing to overcome — which is precisely when it becomes a wrong answer that still sounds distinguished.

CORRECTION — A is denied, and B is the corrected positive version. Members: on the contrary, rather, instead. The constraint: A has to contain a negation or a rejection. “The delay was not a mistake. On the contrary, it was the only part of the plan that worked.” With no “not” in A, “on the contrary” has nothing to correct. “Instead” adds substitution — one thing put in another’s place — which makes it the standard credited answer when a passage discards one explanation and supplies a replacement.

MIRROR — B is A with its terms exchanged. Member: conversely. The constraint is strict, and failing it is almost always why “conversely” is wrong: it requires a genuine inversion, not merely a difference. “Adding salt lowers the freezing point; conversely, removing it raises the freezing point.” Two things that simply differ are a contrast, not a mirror.

CAUSE → EFFECT — B is produced by A. Members: therefore, thus, consequently, as a result, hence, accordingly, so, for this reason. “Accordingly” carries an extra shade — acting in conformity with what preceded — which makes it the right word when B is a decision or a conclusion rather than a physical outcome. The whole family is DIRECTIONAL, and the direction is the most common error on this skill: the connective belongs in the sentence that states the effect.

EFFECT → CAUSE — B explains A. Members: after all, because, since, for. Fewer of these open a sentence comfortably, so they appear as choices less often — but they do appear, and the check is the same one run backwards: does the blank’s sentence give the reason for what came before, or the outcome of it?

EXAMPLE and SPECIFICATION — two families that look like one. FOR EXAMPLE, for instance, to illustrate: B is one member of the class A describes, which requires A to be a generalisation and B to be a case of it. SPECIFICALLY, in particular, notably, more precisely: B narrows A to a definite case or detail without having to be an instance of a class — it sharpens rather than illustrates, and it can follow a claim that is not a generalisation at all. “Notably” adds “and this one is worth singling out.”

INTENSIFICATION — B is A, stronger. Members: in fact, indeed, more than that, above all. B points the same way as A and goes further; nothing is contradicted and nothing is exemplified. This is the family most often lost to “for example,” and the test relies on that. “In fact” has a second use worth knowing: after a stated expectation or a common belief, it introduces the reality that overturns it — “The species was assumed extinct. In fact, a population survived.” Which of the two uses is running is decided by whether A is a belief or a claim.

RESTATEMENT — B is A reworded, at no informational profit. Members: in other words, that is, put differently. The constraint is hard and easy to check: if B contains a single fact that is not already in A, this family is out.

CONCESSION — granting a point before turning against it. Members: admittedly, granted, to be sure, of course, certainly, it is true that. These almost always come in pairs, the concession sentence followed by a contrast that returns to the writer’s line. When the blank sits in the concession slot, the tell is the sentence AFTER the blank, not the one before it — the one case where idea B alone decides the answer.

SEQUENCE and TIME — first, second, next, then, subsequently, meanwhile, previously, earlier, later, eventually, finally, since then. The trap is that a passage containing dates does not thereby have a temporal relationship at the blank. Time words assert order or simultaneity and nothing else, so choosing one always means declining to assert whatever logical link the passage has just built.

CONDITION and ALTERNATIVE — otherwise, if so, in that case, alternatively. “Otherwise” means “if not that, then this” and requires A to state a condition. SUMMARY — in short, in sum, overall, on the whole, ultimately, in conclusion: B has to consolidate what preceded and introduce nothing new, which is the restatement test applied to a stretch of text rather than to one sentence.

Dropping one transitions question a module is not a vocabulary problem — it is not caused by not knowing what “furthermore” means. It happens on within-family splits, where two choices assert the same direction and differ only in an extra constraint, and on execution: naming the relationship correctly and then answering about the wrong pair of sentences. This block is only about that margin.

The contrast quartet, which accounts for more 800-level loss than anything else here. HOWEVER / in contrast: A and B differ, full stop. NEVERTHELESS / nonetheless / even so / still: B holds despite A, which requires A to predict that B would not happen — if the defeated expectation cannot be named, this is not the word. ON THE CONTRARY: A is a denial and B is its corrected positive form, which requires an actual negation in A — no “not,” no “failed to,” no “did not support,” no “on the contrary.” CONVERSELY: B is A with its terms exchanged, a real inversion rather than a difference. Four different claims about the world; the passage decides which is true, and only one of them can be.

The example / specification / intensification triangle, which is the other main site of loss. FOR EXAMPLE requires membership: A general, B a case of it. SPECIFICALLY / in particular requires narrowing, not membership — B sharpens A to a definite case or detail, and A does not have to be a generalisation at all. IN FACT / indeed requires same direction and greater force, with nothing exemplified. Run them as three separate tests, in that order, and the item that felt like a coin flip usually stops being one.

Addition has its own splits and they are tested. “Also / in addition” is flat. “Moreover / furthermore” adds a further and heavier point, so it is the better answer when B is the stronger of two reasons for one conclusion. “Similarly / likewise” asserts a parallel and therefore requires two distinct subjects — when a passage is about one thing throughout, “similarly” is a false claim about the passage, not a stylistic near-miss.

A structural rule worth more than it looks: the test cannot credit two answers, so it never puts two genuinely interchangeable connectives among the four. “Therefore” and “consequently” will not appear together; nor will “for example” and “for instance.” If two survivors look interchangeable, that is not an ambiguous item — it is information that one of them carries a constraint that has not been noticed yet. Go find the constraint instead of guessing between them.

Watch for double-marking. If the blank’s own sentence already contains “although,” “but,” “despite,” “though,” or “even though,” a contrast connective in the blank is usually marking the same turn twice, which means the actual relationship to the previous sentence is something else — very often addition or result. Read what the sentence’s internal connective is already doing before assigning the blank a job.

Not every blank opens a sentence. Some items place it after the subject — “The result, ______, was not what anyone had predicted” — or after a conjunction. Nothing about the method changes: idea B is still the whole sentence, read with the parenthetical lifted out, and idea A is still whatever precedes it. The only thing the position changes is which commas belong to the transition and which belong to the sentence.

The concession-slot blank is the one case where the sentence AFTER the blank decides the answer. A passage running “______, the sample was small. But the effect survived replication at three other sites” wants a concession word — admittedly, granted, to be sure — and that is visible only by reading forward past the blank’s own sentence to the “But.” If a blank refuses to resolve against idea A, read one sentence further before starting to guess.

The execution errors, in order of what they cost. Stopping at the comma inside idea B when the reversal is in the second clause. Anchoring on the wrong idea A — when the sentence before the blank is itself a parenthetical or a concession, the relationship may run back to the sentence before that. Answering the relationship predicted rather than the one the passage has, which is what happens when a prediction from half the evidence meets a choice that matches it. Eliminating three and never positively verifying the survivor by rereading the joined pair. And, on a timed module, spending forty seconds cycling through four choices when ten seconds of naming the relationship would have settled it — that one costs no points on the item itself and several on the questions it forces you to rush.