Revising conventions: an error log beats a rule list

The English Language Skills subtest passes 68% of candidates at their first attempt, which is lower than Reading at 78% and lower than Mathematics at 71%. Most people expect to walk into it. That mismatch between expectation and outcome is the most interesting thing about this subtest, and conventions is where it happens.

How the subtest is built

On that estimate, conventions is roughly two thirds of the subtest. The cut is at most 63% correct, which on our arithmetic is at most 19 of 30, and because the published percentage is a maximum a harder form may need slightly fewer.

Those two nineteens are a coincidence and we are pointing it out so nobody builds a plan on it. Getting every conventions question right would not, in itself, be a strategy, because the estimate could be wrong by several questions in either direction. What the arithmetic does say is unambiguous: if conventions is weak, this subtest is not passable, whatever your vocabulary is like.

Why a rule list is the wrong tool

Open a grammar guide and you get the rules in the order a grammarian would teach them. Nouns, then verbs, then agreement, then punctuation. Work through it and you will spend most of your hours on rules you have applied correctly since you were fifteen.

An adult who reads and writes for a living has an idiosyncratic set of blind spots, usually three or four of them, and they are not the same as anyone else's. Someone consistently loses commas around non-restrictive clauses. Someone else writes a dangling modifier every time they open a sentence with a participle. Neither of them needs a chapter on what a noun is.

So find your own list first. That is the whole method, and it is why the plan above starts with a hundred questions rather than a book.

The clock is the reason this has to be automatic

30 questions in 40 minutes. That works out at 1.33 minutes each, against 1.83 for Reading and 2.86 for Mathematics, and those figures are ours from the published counts and limits. English Language Skills has the tightest clock on the test by a clear margin.

At that pace there is no room to reason. You cannot recall the rule about subject-verb agreement across an intervening prepositional phrase, apply it carefully, and check your work in eighty seconds while also reading four options. Either the error jumps out or the question is gone.

This is why timed sets of ten matter more than untimed sets of fifty. You are training a reflex, and a reflex only forms under something resembling the real condition.

The same work pays twice

The essay competency has 11 skills and is scored holistically by at least two raters. Conventions is part of what those raters are looking at, and it is the only part of the essay that overlaps exactly with a multiple-choice subtest.

So every conventions session is essay preparation, provided you do the writing step. Spotting a comma splice in someone else's sentence and never committing one in your own are related but not identical, and the bridge between them is generating your own examples. It takes four minutes at the end of a session and it is the reason we would sit these two subtests within a couple of weeks of each other.

Where this method does not work

If English is not your first language, an error log is not enough on its own and we would not pretend otherwise. The method assumes a large base of correct usage with a few gaps in it, which is what a lifelong English speaker has. A second-language candidate may have systematic gaps rather than idiosyncratic ones, and for systematic gaps a rule list is genuinely the right tool. Use both: the guide for structure, the log to find out which parts of the guide you actually need.

The subtest improves more on a return than any other except the essay, from 68% at first attempt to 83% at best attempt. Whatever it is that people fix between attempts, it is fixable.