LSAT Language Strength: How Strong Can an Answer Be?

By Jeremy LaRusso · Updated September 2026

There is a whole category of LSAT question you can understand perfectly and still get wrong. You read the stimulus correctly. You find the conclusion. You understand the evidence supporting it. Then you choose an answer that says slightly more than the argument allows, and you lose the point because of a single word.

Language strength is one of the most reliable eliminators on the Logical Reasoning section, and it takes students a surprisingly long time to trust, because it sounds at first like grammar rather than logic.

It is logic. Some is different from most. May is different from will. Helped cause is different from caused. More effective makes a comparison that effective does not.

This is not a difficult idea, only an unfamiliar one. The phrase gets used long before anyone defines it — students are told to watch the strength of the language without being told what that means. So this article starts from nothing.

First: what does “language strength” mean?

Language strength is how much a sentence commits itself to. Compare these:

A stronger statement tells you more. It also needs more support. And the relationship only runs one way: if every member of a group has a trait, then most do and some do. If some do, you know nothing about whether most do. If most do, you know nothing about whether all do.

One thing trips almost everyone at first. In conversation, some implies “not all” and most implies “not quite all.” On the LSAT neither carries that upper limit. “Most students passed” does not tell you that anybody failed — if all 100 passed, it is still true that most passed. “Some students passed” does not tell you that anybody failed either. Most permits all, and some permits all. Students lose points assuming the conversational limit is there.

The language bank

These are the working LSAT meanings and minimum commitments — not impressions about how the words sound in ordinary conversation. What matters is the minimum each expression commits the author to. Skim this, then come back to it as the examples below use it.

The categories matter as much as the entries, because these words fail in different ways. A quantity that was never established, a comparison where the stimulus only ever discussed one thing, and a probability where the stimulus only said something was possible are three different wrong answers — and all three are extremely common.

Conditionals are the one category where labeling a word “strong” and moving on will actively hurt you. If, only and only if all establish rules, but they point in opposite directions. Diagram them; do not just note that they are forceful.

Comparatives are dangerous for a different reason. An answer can invent an entirely new comparison while every noun in it appeared in the stimulus. If the stimulus told you experienced attorneys are effective negotiators, it did not tell you they are more effective than inexperienced ones. The second sentence is not a more confident version of the first. It introduced a second group.

The two modes

Now that you know what the words mean, the question is how strong the correct answer is allowed to be. That depends entirely on which of the two modes you are in.

How you learn this matters. Most students learn it as two lists of question types, memorize the lists, and then stall the first time they meet a stem worded in a way their list did not anticipate. The test is not the list. The test is the rule: if the answer choice is not entering the argument as a new fact, you are in Mode 1.

The lists follow from the rule rather than replacing it. Mode 2 is the short one:

• Strengthen and Weaken
• Sufficient Assumption
• Resolve the Paradox / Explain the Discrepancy
• Principle questions where you supply the principle that justifies the conclusion
• Principle questions where you supply new information to justify an application of a principle
• Most Logically Completes, but only when the word before the blank is a support indicator (since, because, for, as)

Everything else is Mode 1: Must Be True and Inference in all its stem variants, Most Strongly Supported, Most Strongly Suggested, Must Be False and Cannot Be True, Most Logically Completes when the word before the blank is a conclusion indicator, Necessary Assumption, Flaw, Main Conclusion, Role of a Statement, Method of Reasoning, Point at Issue, Parallel Reasoning, Flawed Parallel, the three principle tasks where a principle is already in play, Misinterpretation, Evaluate the Argument, Illustration and Generalization — and in Reading Comprehension, Main Point, Author’s Opinion, another party’s opinion, According to the Passage, Most Strongly Suggests, Definition, Purpose, Value Most Highly, and the hypothetical application questions. Strengthen and Weaken appear in Reading Comprehension too, and they are Mode 2 there exactly as they are in Logical Reasoning.

EXCEPT and LEAST variants do not change any of this. A Strengthen EXCEPT question is still in the Strengthen family; the stem is simply asking which answer fails to do the normal job.

Mode 1: the stimulus is the ceiling

In Mode 1 the answer cannot make a claim stronger than the stimulus supports. I tell students to hold one phrase in their head:

The upper bound does the eliminating. If the stimulus says something may occur, you cannot infer that it will. If the stimulus says some members of a group have a trait, you cannot infer that most do. If the stimulus never compared A with B, you cannot infer that A is more effective, more common, or more anything than B. Those answers are gone before you weigh anything else.

An important qualification: judge the whole claim, not the loudest word in it

This is where a mechanical reading of the rule breaks, and it is worth getting right, because a careful student will otherwise think the system contradicts itself.

Language strength applies to what the answer choice asserts as a whole — not to every strong-looking word sitting inside it. Consider a Flaw answer:

So you cannot do language strength by circling forceful words. You have to see the scope they sit inside. Overlooks the possibility that, fails to consider whether, does not establish that — each of those wraps whatever follows and weakens it.

This is also the reason Flaw is a Mode 1 question type even though its answers sound so absolute. A flaw answer is a claim about the argument, and the argument is something you were handed in full. Either it addressed the sample size or it did not; you settle that by reading the page, not by adding a fact about dentists. The same holds for Method of Reasoning, Role of a Statement, Point at Issue and Parallel Reasoning — all of them describe the argument rather than the world, and all of them can carry forceful language without breaking the ceiling.

The cleanest way to see the difference is to write the same underlying problem in each mode.

Why Necessary Assumption is Mode 1

This is one of the classifications I most often see taught incorrectly.

A necessary assumption is usually not stated in the stimulus, which makes it tempting to think you are adding it. You are not. A necessary assumption is something the argument was already relying on for its reasoning to work. Your job is to identify that commitment, not to supply it — which is exactly why the Negation Test works. If denying an answer destroys the argument, that tells you the argument needed it all along.

The strength behavior settles it.

Sufficient Assumption runs the opposite way: your job there is to add enough to guarantee the conclusion, so strength costs you nothing and is often exactly what you need. Those two question types look like siblings and behave like opposites, and the direction of the strength is what tells them apart. A ceiling set by what you were handed is the defining behavior of Mode 1.

It matters because the wrong version is teachable. A national prep course I took taught that every Sufficient Assumption question is a mismatched concept. It is not — a great many are enabling assumptions, where the gap is not a missing link between two terms but an unstated condition the argument needs in order to work at all. A student taught the false version will hunt for a mismatch, fail to find one, and conclude they are bad at the question type. They are not. They were taught something wrong.

This article also revises a rule of my own that I had been stating more simply than it deserved. Getting it right first time is not the standard worth holding. Not leaving a rule half-reasoned because the simple version is easier to teach is.

Mode 2: use exactly as much strength as the job requires

The standard advice for Mode 2 is “look for strong language.” That is a useful beginner heuristic. It is not the rule.

Often the job does require real force.

But strong is not automatically better. Sometimes one very narrow fact is devastating.

That is what it takes to contradict a claim about every case: one case going the other way. Universal conclusions are the most fragile things on the test for exactly that reason. An author who writes never has handed you a target a single counterexample destroys.

Move the quantity word and the requirement changes

Be careful about what that does and does not mean. It does not mean a Weakener has to contain a majority claim of its own. Weakening is making a conclusion less likely, not proving it false — an answer showing the sample was drawn from one unusual coastline could weaken the conclusion without counting a single species. The narrow point is this: one exception logically destroys an all or none claim. One exception does not logically destroy a most claim.

So before you evaluate a single answer choice on a Strengthener or Weakener, read the conclusion and name exactly what it claims. Is it all? None? Most? Some? Probably? A comparison with something else? A cause? Merely one contributing factor? Different conclusions need different kinds and amounts of evidence to move them, and that question is far more reliable than asking whether an answer “sounds strong.”

This replaces a rule of mine that was too blunt. I have told students that an answer beginning with some is a red flag on a Strengthener or Weakener. As a rough statistical observation about what tends to appear in answer choices, that is not useless. As a rule it is wrong, and I would rather you had the real one: ask whether that amount of information is enough to move this argument. Sometimes one narrow fact is decisive. Sometimes it is irrelevant. The word at the front of the answer does not tell you which.

The Floor Test

For vague words, use what I call the Floor Test. The idea is simple:

Run the same check on anything:

• some → at least one
• may → possible, and nothing more
• helped → made some contribution
• up to 40 percent → a ceiling, not a number near 40
• consistent with → compatible with the evidence, not established by it

The Floor Test stops your own brain from strengthening the author’s sentence on their behalf, which is what it will do if you let it.

The two four-letter words

If you take one habit from this article, train yourself to notice two words on sight: MOST and MORE.

MOST means more than 50 percent. Students read past it because it sounds casual in ordinary English, but on the LSAT it changes the logical burden of the sentence. And remember that it permits all: “most students passed” does not establish that anyone failed.

MORE creates a comparison, and it is one of the easiest ways for a wrong answer to smuggle in information that was never established. If a passage tells you experienced attorneys are highly effective negotiators, you cannot infer they are more effective than inexperienced ones. Maybe they are. The passage said nothing about inexperienced attorneys. That sentence did not restate the first one more confidently — it introduced a second group. The trap appears constantly on Inference, in Reading Comprehension, and on opinion questions.

What this looks like at the desk

• Identify the mode. Is the answer choice being accepted as a new fact? If no, Mode 1. If yes, Mode 2.
• In Mode 1, ask what the stimulus actually supports, and hold the answer to strength ≤ support.
• In Mode 2, ask what the answer has to accomplish. Do not reach for the strongest answer by reflex. Ask whether it is strong enough to strengthen, weaken, guarantee or explain what the stem requires.
• Read the conclusion carefully before the answers, especially on Strengthen and Weaken. Note its quantity, its certainty, any comparison, any causal claim, and its scope.
• Apply the Floor Test to flexible words. Judge the minimum the sentence guarantees, not the stronger reading your brain supplied.
• Watch for comparisons that arrived without a second thing to compare to. It is one of the most common unforced errors on Inference.
• Evaluate the whole proposition, not isolated vocabulary. A forceful word inside “may,” “fails to rule out,” or “overlooks the possibility that” does not make the answer itself strong.

Language strength is not a vocabulary trick. It is the practice of tracking exactly what each statement commits you to — and refusing to read more into it than it actually says.