Grammar curriculum

Almost every agreement error is really a subject-identification error; the number rule itself is rarely in doubt. Strip the sentence to its bare subject and verb and read that two-word skeleton aloud. If the skeleton sounds wrong, the full sentence is wrong.

Cover the phrase with a finger and the choice becomes obvious, which is why the test is mechanical rather than a matter of ear. The distraction is usually a prepositional phrase whose object is the opposite number from the subject, sitting there to pull the verb toward it.

Decide by what the sentence says the group is doing: acting as a body takes the singular, members acting separately takes the plural. Once chosen, stay with it. A sentence that switches number partway reads as an error even when each half is defensible alone.

This is proximity, not logic, so the verb can end up singular beside a plural idea. The same rule governs or, either/or and not only/but also. It reaches person as well as number, which is why “neither she nor I am ready” is correct and still sounds strange.

When the two subjects differ in number, the proximity rule leaves you free to choose either verb by reordering. Prefer the arrangement that puts the plural second: a plural verb after a plural noun draws no attention, while a correct singular after a plural often reads as a slip.

Test by asking how many things exist, not how many nouns appear. Two nouns naming one referent — “my colleague and friend” when one person is meant — stay singular. A following each or every also pulls the compound back to singular.

These pairs have fused into one idea, so the test is whether the halves can be separated without changing what is meant. If you can insert a modifier before the second noun and still have the same referent, the pair is a genuine compound and takes the plural.

These are prepositional rather than coordinating: they add information about the subject instead of joining a second one. The tell is that such a phrase can be moved or fenced with commas, while a real and-compound cannot be moved anywhere.

Form beats sense here. The phrase distributes over individuals one at a time, so everything downstream — the verb, and any pronoun referring back — stays singular. It is formal and slightly archaic, and a plural rewrite is usually better in ordinary prose.

The verb follows the singular noun that ends the phrase, not the arithmetic. Reverse the wording to “more applicants than one” and the plural verb returns: same meaning, different grammatical head. Sense has no vote once the head noun is fixed.

Here the intervening phrase does decide the verb, which is why it needs stating separately. A fraction has no number of its own, so it borrows the number of whatever it measures. The rest of, the majority of and a lot of all behave the same way.

The claim that it must be singular rests on a false etymology — that it means “not one” — and it has been used both ways for a thousand years. Choose the singular to stress individuals, the plural for a group; only “not one of them” forces the singular.

These stay singular however plural the following of-phrase looks, because the pronoun is the subject and the phrase only names the set it selects from. The pronoun referring back is the harder half: a singular verb does not settle what is available later.

Unlike the some/all/most group, these carry their own number rather than taking it from a referent, so no of-phrase can change them. They also cannot head a mass noun — “several water” fails — which is a quick test for whether a quantifier is count-only.

These are the quantifiers that genuinely look through to the noun: a mass referent takes the singular, a count referent the plural. That makes them the exception to the habit of ignoring the of-phrase, and the reason one flat rule about intervening phrases misleads.

These nouns end in -s for historical reasons and name one thing, so the -s is not a plural marker at all. Some sit in both classes with different senses: statistics as a discipline is singular, statistics as a set of figures are plural.

The title names one object — a book, a film, an album — so its internal grammar is frozen and cannot reach the verb. The same freezing explains why a title keeps its own capitalization and punctuation inside a sentence that would otherwise change them.

There is a placeholder rather than the subject, so the real subject sits after the verb and you have to read ahead before choosing. With a compound following, proximity usually governs, which is why a mismatched opener passes in speech and is worth rewriting on the page.

Inversion follows a fronted place phrase, a question, and a negative opener. The reliable move is to unwind the sentence into normal order, choose the verb there, then invert again. Judging by ear while the sentence is still inverted is what produces the error.

Find the noun the relative pronoun renames and give the clause that noun's number. The pronoun is invariable, so it offers no clue of its own, and the nearest preceding noun is often not the antecedent — which is the whole difficulty of the next entry.

Ask what the relative clause is actually describing. In “one of those who succeed” the clause describes all of them, so the verb is plural; adding “the only” makes the clause describe just the one, and it turns singular. Two words change and the referent changes with them.

The test is whether the sentence treats the quantity as one amount or as separate items. Ten years is a long time measures a single span; ten years were spent on separate projects counts them. Distances, weights and scores behave the same way.

A linking verb has a noun on each side and only the one before it is the subject. When the two differ in number the sentence sounds wrong whichever verb you choose, so the honest fix is usually to reverse the two nouns and let the plural lead.

Position decides. Before the subject, each makes it singular; after it, each is an appositive that distributes the action without touching the grammar. Both “each of the players was” and “the players each were” are correct, for that reason.

Every distributes over the elements one at a time, so it overrides the plural that and would otherwise create, and each does the same in this position. The effect stops at the verb: a pronoun later in the sentence often needs the plural to avoid awkwardness.

What can mean the thing that or the things that, and nothing inside the clause settles which, so the noun after the linking verb decides. What we need is patience and what we need are volunteers are both correct, and neither is ambiguous once the complement arrives.

English marks case on pronouns and almost nowhere else, which is why these choices feel arbitrary rather than structural. The subject slot is easy in isolation; nearly every real error comes from a compound, a comparison, or a linking verb, each with its own entry here.

One form serves direct objects, indirect objects and objects of prepositions, so a single test covers all three: if the pronoun receives the action or follows a preposition, it takes this form. Hypercorrection is the commoner fault — the subject form inserted to sound careful.

The error survives because it sounds refined, which is the signature of hypercorrection: a rule half-remembered from childhood, applied where it does not belong. The preposition does not care how many objects follow it or in what order they come.

Isolating the pronoun restores the intuition the compound suppresses, because native ears handle a single pronoun reliably and a paired one unreliably. The test works in both directions and settles the majority of case errors on its own.

Answer the question with a statement and see which form fits: a subject answer gives who, an object answer gives whom. The final -m pairs with him and them, which makes the test easy to recall. Speech is dropping the distinction; edited prose still marks it.

The preposition or verb in front governs the whole clause rather than the pronoun, so a subject pronoun is correct even directly after a preposition. Look only at what happens after the pronoun; everything before it is a distraction planted by the sentence.

Phrases such as we believed, they said and I think are parenthetical and can be lifted out entirely. Remove one and the pronoun's real job appears — subject of the following verb — which is why the subject form survives an object-looking position.

Delete the noun and read the pronoun in its place; the correct case is the one that still works alone. The noun sitting after the pronoun often makes the wrong case sound acceptable, which is exactly why the deletion test is needed here.

Finish the clause and the case chooses itself: one completion makes the pronoun the subject of an implied verb, the other makes it an object. Both are grammatical and they mean different things, so this is a content decision rather than a correctness one.

A linking verb equates rather than acts on, so the noun after it is a complement rather than an object. Speech has settled on the object form and edited prose has not fully followed, so the safe move in formal writing is to recast until the question does not arise.

A reflexive points back at something already named nearby, so it can never be the first mention of a person. Its two honest jobs are marking an action that returns to its doer, and emphasis; everything else is the misuse described in the next entry.

The substitution is a hedge — it feels less blunt than either plain form — which is why it spreads through business writing and formal introductions. With no antecedent in the clause there is nothing to reflect, and the reader hears evasion rather than courtesy.

Removability separates the two uses: an intensive can be deleted with the sentence still meaning the same thing, while a true reflexive cannot. The intensive can also slide to the end of the clause, which a reflexive object can never do.

The hard cases are the indefinites, where the verb is settled and the pronoun is not. Singular they has been in continuous use for centuries and is accepted by most style guides, though some readers still object; a plural antecedent sidesteps the question entirely.

Proximity is not a rule, so the reader cannot resolve this by taking the nearest noun — they have to guess, and will sometimes guess wrong. Repeating the noun is not inelegant here; it is the cheapest fix and the only one that is certain.

The writer knows which idea is meant because they have just thought it; the reader has several sentences to choose between. Adding a noun after the pronoun costs one word and forces the writer to name the thing they were pointing at.

It is common in speech and generally understood, so the fault is precision rather than grammar. The fix matches the one for a vague this: supply a noun summing up the clause, which also tells the reader how the writer wants the preceding idea labeled.

Often the noun exists only inside a possessive or an adjective, where it is not available to be referred back to. A possessive modifies rather than names, so careful prose does not let a pronoun reach back to it even when the meaning is obvious.

The drift happens because generalizing invites the generic second person partway through a paragraph. Pick one footing and hold it: third person for analysis, second for instructions. The shift is most visible when the reader is not the person being described.

These are already possessive, so they need no marker, exactly as his and her need none. An apostrophe on any of these forms signals a contraction instead, which is what makes the its/it's confusion a collision between two different words.

Expand the contraction and read the sentence back: if the expansion fits, the apostrophe belongs. This is the most common apostrophe error in edited copy, and a spellchecker cannot catch it because both forms are real words correctly spelled.

Same expansion test as its and it's, and the same reason the error survives proofreading. Whose is also the possessive for things, not only people — a claim whose basis is unclear — which is standard despite a long-running superstition against it.

The distinction is real: seeing him leaving reports the person, resenting his leaving reports the act. Where the act is what matters, the possessive is the precise form. Many guides relax this with full nouns; it survives firmly with pronouns.

The prohibition is a preference rather than a rule, and that has been used of persons continuously in English. Save the real attention for the restrictive and nonrestrictive choice, which changes what the sentence claims, unlike this one.

Switching to you or he partway through is the visible fault. The invisible one is that this pronoun, repeated often, sounds stilted in American prose, so choosing a concrete subject at the outset usually beats sustaining it well.

The label misleads: this tense rarely describes the present moment, which belongs to the progressive. It reports what is generally the case. Its scheduled-future use is restricted to fixed timetables, so a train leaves at six and a decision does not.

It also carries a temporary reading — true now but not permanently — and an arranged-future one when a time is given. One form doing three jobs is why it gets overused; the test is whether the situation has visible edges in time.

Definite is the load-bearing word. The time need not be stated, but it must be recoverable from context, which is what separates this from the present perfect. Once a past time is fixed, later verbs in that stretch of narrative inherit it.

It supplies background rather than events, so a narrative built only from it never moves. Paired with a simple past it sets up the interruption structure: the longer action takes the progressive, the one cutting across it takes the simple form.

The incompatibility is the usable test: if a finished time expression can be attached, the tense is wrong. It reports the present consequence of something past, which is why it answers whether something has happened and never when it happened.

It stresses the activity where the plain perfect stresses the result, so choosing between them is choosing what to foreground. Stative verbs resist it, and with those the plain perfect carries the duration on its own.

Overuse is the commoner fault. Where a conjunction, a date, or plain narrative sequence already fixes the order, the simple past is enough and the extra auxiliary only slows the sentence down. Reserve it for the genuine flashback.

English has no future tense, only these strategies, and they differ in what they imply: a decision made now, evidence already visible, an arrangement fixed, a timetable published. The choice reports the speaker's grounds, not the timing.

It needs that reference point to mean anything, so a sentence using it without one is reaching for emphasis it has not earned. The point normally arrives in a by-phrase or a time clause, and the time clause itself stays in the present.

Backshift is a convention rather than a logical requirement, which is why it can be suspended. When the reported statement is still true at the moment of reporting, keeping the original tense is standard and often clearer than shifting it.

The errors cluster where the past tense and the participle differ but sound close, so the auxiliary is the thing to watch: after have, or after a passive be, the participle is required. With no auxiliary present, the past tense form belongs.

The trap is that the past of the intransitive verb is spelled the same as the present of the transitive one. Ask whether something is being put down: an object means the transitive verb, no object means the other, whatever the tense turns out to be.

The same structural split as lie and lay without the overlapping forms, so the test is easier here and identical in principle. If you can name what is being lifted, the transitive verb is required; if the subject goes up by itself, it is not.

The third member of the same family. Set keeps one form throughout, which removes the tense confusion but hides the transitivity question underneath it; ask what is being placed, and where nothing is, the intransitive verb is the answer.

Only conditions that are false or hypothetical qualify. An if-clause reporting something that may well be true takes the ordinary past, so not every if calls for this form, and substituting one for the other misstates how likely the speaker thinks it is.

The base form appears regardless of subject, so no third-person -s turns up and the clause looks ungrammatical until you know the pattern. British usage often inserts should instead, which is why the two varieties differ visibly here.

The if-clause stays present even though it refers to the future, and doubling the modal into it is the standard error. This pattern says the speaker treats the condition as genuinely open, which is what separates it from the unreal patterns that follow.

The past form here marks distance from reality rather than past time, which is the single idea that makes the whole conditional system coherent. Read it as a signal about likelihood and the tense that looked wrong stops looking wrong.

This one is always counterfactual, because the past is settled: it states what did not happen and what therefore did not follow. That makes it the natural form for regret and for blame, and the reason it can sound accusatory when none is meant.

The two halves are chosen independently, each clause taking the form that matches its own time. Recognizing that dissolves the apparent extra rule, because a mixed conditional is only the ordinary patterns applied to two different moments.

Strength is one axis and source is the other: some of these carry the speaker's own authority while others point at an external requirement. The negatives diverge sharply, one forbidding an action and the other merely releasing you from it.

These report confidence rather than obligation, and the same words serve both duties, so context does the disambiguating. Two of them sit at the ends as the confident poles, with the rest covering everything uncertain in between.

The passive is a legitimate construction rather than a fault, so a blanket rule against it is bad advice. What makes it a problem is using it to hide an actor who should be named; ask whether the reader needs to know who did this, and answer honestly.

Method sections and incident reports rely on it for exactly this reason. The same omission turns evasive when the actor is known and accountable, which is why a dropped agent is worth noticing rather than automatically correcting.

The pattern is largely lexical and has to be learned verb by verb, but one tendency is worth knowing: infinitives lean toward the potential and unrealized, gerunds toward the actual and ongoing. It predicts the meaning shifts in the next entry.

With a gerund the complement names something already happening or already done; with an infinitive it names what comes next. That single distinction accounts for every verb in this small set, including try, forget and go on.

Many verbs are both and shift meaning as they shift class. The category matters beyond correctness because only a transitive verb can be made passive, so an attempted passive that will not form usually means the verb was intransitive all along.

With a separable verb a noun object may sit on either side of the particle, but a pronoun object must come between the two. That pronoun rule is the one that catches learners, and it is absolute rather than a preference.

The resistance breaks where the verb shifts to a dynamic sense, so one of these in the progressive usually signals a different meaning rather than a mistake. Verbs of thinking, possessing and perceiving make up most of the class.

Unmotivated is the qualifier that matters: a shift reflecting a real change in time is required, not an error. Establish a base tense for the passage and change it only where the timeline changes, then check that each change earns itself.

Adjectives modify nouns and adverbs modify everything else. The -ly test fails in both directions: friendly and costly are adjectives, while fast, hard and late serve as both without changing form. Ask what the word modifies, not how it ends.

A linking verb describes the subject rather than the action, so the word after it reaches back to the subject and must be an adjective. Several of these verbs work both ways, taking an adverb when they report an action instead of a state.

Well is also an adjective, chiefly in the health sense, which is what makes this pair confusing rather than simply irregular. Feel good and feel well are both correct and mean different things, so neither one is a correction of the other.

The choice is set by how many items are in the comparison set, which often goes unstated and has to be inferred. Where only two exist the superlative overstates; speech has made that overstatement idiomatic and edited prose still resists it.

Marking a comparison twice adds nothing, because the ending and the preceding word do the same job. The second half of this fault is a different thing: applying degree to a word that has no degrees, which the absolutes entry covers.

These name an endpoint, so strictly there is nothing more of them to have. Usage has softened this for several, and approach words such as nearly are accepted where degree words are not — which is the defensible line to hold.

A participial opener attaches to whatever noun follows the comma, so the fix is to make that noun the actor. Reading the sentence literally is the diagnostic, and the result is often absurd in a way the writer cannot see, knowing the intent.

The same structure as the dangling participle with a different opener, so the same test applies: whoever performs the infinitive must be the subject of the main clause. Passive main clauses invite this, because the passive removes the actor.

The clause omits its own subject and verb, expecting the reader to copy them down from the main clause. When the borrowed subject turns out to be the wrong one, restoring the omitted words exposes it immediately, and that is the whole test.

English attaches modifiers by position, so the sentence means what the arrangement says rather than what the writer intended. The repair is to move the phrase beside the noun it belongs to; rewording it in place leaves the error standing.

Both readings are grammatical, so no rule resolves it and the reader has to guess. Moving the modifier to one end of the sentence settles it; punctuation can settle it too, but only for a reader who takes the punctuation as a boundary.

In speech, stress does this work, which is why writers place these words loosely and hear no problem in them. On the page, position is the only signal available, so the limiter goes directly before the element it limits.

Each slot in a short sentence yields a different claim and every one of them is grammatical, so nothing flags the wrong choice. That is what makes this the standard demonstration that word order carries meaning rather than merely style.

It has its own subject, which is precisely why it cannot dangle: the actor is stated inside the phrase. That makes it a structural repair for many dangling participles, and a way to attach circumstance without reaching for a subordinator.

Removing it changes which thing is meant, so it is not removable at all. The absent comma is the real signal and it is doing work: a reader who sees no comma knows the clause narrows the noun rather than commenting on it.

The commas must be paired — opening one and omitting the other is the commonest form of this error. Deleting the clause should leave a sentence still true and still identifying the same thing; where it does not, the clause was restrictive.

The same test as for clauses, applied to a renaming noun. One consequence is worth noticing: my brother Tom implies more than one brother while my brother, Tom, implies exactly one, so the commas make a claim about the world.

The hyphen groups the words so the reader knows they act together, which is why it is needed before the noun and not after it. Once the phrase follows the noun the grouping is already unambiguous, and the hyphen is dropped.

The hanging hyphen holds the place of the repeated element so the reader supplies it without the phrase being written out twice. Dropping it changes what the first number modifies, which is a meaning error rather than a typographic one.

Here the hyphen is the only thing separating two different claims, one about the business and one about the owner. That is the argument for hyphenating consistently: a rule applied only when ambiguity is noticed misses the cases nobody notices.

The ending already shows the word modifies what follows, so a hyphen would add nothing. Adverbs without that ending are different — well, best, ill — and they do take the hyphen, because they could otherwise be read as adjectives.

Two tests agree here: if the pair can be reversed, or joined with and, they are coordinate and take the comma. Both are reliable, and a phrase that passes one while failing the other usually signals an adjective doing some other job.

Each one modifies everything to its right rather than the noun alone, so the order is fixed and no comma belongs. The reversal test is how you tell: a phrase that turns strange when reversed is cumulative rather than coordinate.

Native speakers follow this order without being taught it and notice violations instantly, while being unable to state the rule. It applies to cumulative adjectives, which is why those cannot be reversed, and it explains that entry's test.

Each noun could be modifying any of the ones after it, so the reader holds several parses open until the chain ends. Unpacking one link into a prepositional phrase resolves it, and past about three nouns is where the cost starts to show.

Read the introducing words once against each item in turn; the element that will not attach is the one out of form. Listing the items vertically makes the mismatch visible immediately, which is why the fault survives so much longer in running prose.

Whatever follows the first half sets the form the second half must match, so the error is almost always in where the first half sits rather than in the second. Move the opening word and the two arms usually fall into line without rewriting.

The word tends to drift leftward toward the verb, where it brackets more than the writer intends. Put it immediately before the first genuine alternative, and the second one after or will then match it without further adjustment.

Both forms are correct on their own, so nothing in the sentence is ungrammatical and the fault is one of pattern. A reader who has settled into one form has to re-parse at the switch, and that re-parse is the cost being paid.

Dropping the repeated subordinator is fine in a short pair and breaks down over longer ones, where the reader loses track of where each clause begins. Repeat it once the clauses are long enough that the boundary stops being obvious.

Test by reading the preposition separately against each item in the list. Where one item takes a different preposition, the shared one has to become two, and no amount of rearranging the list will make a single one fit both.

The two comparisons take different completions, so collapsing them drops one of the pair. Written out in full it reads heavily, which is why the usual repair is to move the second comparison to the end of the sentence rather than nest them.

Advertising relies on the omission because an unstated basis cannot be checked. In prose the missing term is sometimes genuinely recoverable from the previous sentence; where it is not, the claim is not merely incomplete but unfalsifiable.

The error comes from a possessive left implicit, so the sentence ends up comparing a thing with a person. Repairing it needs a possessive form or the word those, and the meaning was never in doubt, which is why writers read past it.

Without the extra word the sentence asks a member to exceed itself, which it cannot do. The same correction applies to anyone else constructions, and the fault stays invisible until you notice the compared thing sits inside the set.

Both readings are available because the comparison is elliptical, so the missing verb is what creates the ambiguity in the first place. Finishing the clause out loud resolves it, and in writing that completion is the cheapest fix available.

Many dialects and older English use two negatives to intensify rather than cancel, so this is a convention of one variety rather than a law of logic. The deliberate version — not unwilling — is a recognized figure that weakens a claim on purpose.

Joining with and tells the reader the two ideas rank equally, which is a claim the writer usually does not intend to make. Subordinating the lesser one marks the ranking explicitly and leaves the main clause easy to find.

Both arrangements are grammatical and they emphasize different things, so no rule chooses between them. The test is what the paragraph is about: the main clause should carry the idea the surrounding sentences are developing.

The sentence parses cleanly, so automated checkers pass it and the fault is semantic. It usually comes from a compressed chain where an intermediate noun dropped out; restoring that noun repairs the sentence and names what was really acting.

A definition needs a noun on both sides of the linking verb, and these words supply a time or a place instead. It is common in speech; the repair is to name the category the term belongs to, which usually improves the definition anyway.

It happens most in long sentences where the writer revises the opening and never re-reads the whole. Reading the subject straight through to its verb, skipping everything between, exposes it — the same diagnostic used for agreement errors.

Scope is what these pairs express, so a mismatch makes a claim about a different set of things than the writer intended. Check by asking what each half of the pair reaches over; where the answers differ in kind, the bracketing is wrong.

Delete the list and read what remains: a complete sentence licenses the colon, a fragment does not. That is why including and such as almost never take one — the clause was not finished at the point where the list began.

Consistent form lets the reader predict each item's shape and spend attention on content instead. Where one item genuinely resists the pattern, that is usually a signal the item does not belong at that level of the list.

Test each side for a subject and a finite verb; two complete sentences need more than a comma between them. It is a deliberate device in some fiction, which is why it appears in respected prose, but expository readers take it as an oversight.

The same structural fault as the comma splice with nothing at all at the join, so the same four repairs apply. It is rarer because the missing pause is audible when read aloud, which makes reading aloud the quickest way to catch it.

The comma is required only when both sides could stand alone as sentences. A compound predicate — one subject with two verbs — takes none, and the tell is simply whether a subject appears after the conjunction.

It marks where the introduction ends and the main clause begins, so it matters most when the opener is long or could be misread as running on into the subject. Very short openers may drop it where no misreading is possible.

House style decides it and both conventions have serious adherents, so the honest advice is consistency rather than correctness. It is worth knowing which style you are in, because the cases where it changes meaning are the next entry.

Without it, the last two items can be read as renaming the first, which produces the famously absurd dedications. That is where the choice stops being style; reordering the list fixes it too, if house style forbids the comma.

Both commas are required, and dropping the closing one is the usual failure, because the writer hears the opening pause and not the closing one. Treat them as brackets: an unclosed pair leaves the reader without a boundary.

Single is the operative word. A long subject invites a pause, and the pause is real in speech, but on the page the comma must be paired or absent. A pair fencing an interruption is correct; one comma standing alone never is.

Its absence changes the meaning rather than merely the rhythm, which is the point of the well-known examples about eating. The name takes commas on both sides mid-sentence, following the same pairing rule as any fenced element.

The closing comma is the one that goes missing, because the pause feels finished once the year or the state has been written. The day-month-year order takes no commas at all, so documents mixing both formats look inconsistent.

It signals that two complete thoughts are more closely related than a period would suggest, so it is a meaning choice rather than a stronger comma. Both sides must be complete; where one is a fragment, a colon or dash is the mark.

These words connect meaning without doing grammatical work, so they cannot join clauses the way a coordinator can. Treating one as a coordinator produces a comma splice, which makes this the commonest route into that error.

The semicolon becomes the higher-level divider, letting the reader see where each item begins. Once the device is in use it must appear at every item boundary in the list; mixing the two marks reintroduces the ambiguity it removed.

Cover everything after the mark and read what is left: a complete sentence licenses it. That single test rules out a colon after a linking verb or after a preposition, which are the two commonest places writers reach for one.

It is a forward-pointing mark, so what follows has to deliver what the first clause set up. That expectation is why a colon adds emphasis where a period would not: the reader is already leaning toward the second half.

Commas, dashes and parentheses are one tool at three volumes, so the choice is about how much attention the interruption should attract. Dashes are also the only pair that stays readable when the fenced element contains commas.

In a range it stands for the word to, which is why pairing it with from is redundant. Its compound use marks a relation between two equal things rather than a modifier, so it differs from the hyphen in meaning, not only in length.

Both fence removable material while signaling opposite intentions about its importance, so the choice tells the reader how to weigh what is inside. Parentheses also nest more comfortably, which matters in a sentence carrying two asides.

Styles genuinely differ here and several allow the bare apostrophe after a final s, so this is a house decision rather than a correctness one. Pick one and hold it; the visible fault is switching between them inside one document.

Form the plural first, then add the possessive marker, and the placement follows automatically. Working in the other order is what produces a misplaced apostrophe, because the writer is pluralizing a possessive rather than possessing a plural.

These plurals do not end in s, so the bare-apostrophe rule never reaches them. The same plural-first procedure gives the right answer here too, which is why it is worth learning as a procedure rather than as a list of cases.

Marking only the last name claims shared ownership; marking both claims separate ownership. That is a factual difference rather than a stylistic one, and it is also why the noun that follows changes number between the two.

The time expression owns the thing it measures, which is why the apostrophe belongs even though nothing is literally possessed. Singular periods take an apostrophe and s, and the whole class is widely written with no mark at all, in error.

An apostrophe marks possession or omission, and a plural is neither of those. The family-name case shows up on doormats and holiday cards more than in prose; the decade case is the one most likely to survive a proofread.

It is a typographic convention rather than a logical one, which is why British style places these marks by sense instead. Colons and semicolons go outside in both systems, so the rule is narrower than it is usually remembered as being.

This is the one mark American style also places by logic, so it is an exception to the previous entry rather than an inconsistency in it. Exclamation points follow the same reasoning, and only one mark appears at the join.

British style typically reverses the pair, so quoted material moved between the two conventions has to have its marks swapped. Past two levels, alternate again — though three levels is usually a sign to paraphrase instead.

The indentation does the work the marks would, so keeping both is redundant. Styles set the threshold differently by line or word count, and they also differ on where the citation sits relative to the final period.

The obligation is in the second half of that: an omission reversing or narrowing the original is a misquotation whatever marks surround it. A sentence-ending period is kept alongside the points when a full sentence has been dropped.

They keep the boundary visible between what the source said and what the quoter added for grammar or clarity. The same brackets carry sic, which flags an error as belonging to the source rather than to the person quoting it.

One joins parts of a word, one spans a range, one fences an interruption. Word processors substitute silently between them, so the mark on the page is not always the one that was typed — which is worth checking in final copy.

It compresses a relation the writer has not specified, leaving the reader to decide whether it means and, or, or both. Spelling the relation out is usually one word longer and removes the ambiguity entirely.

Its force comes from scarcity, so each one after the first in a passage weakens the rest. Where emphasis is genuinely needed, a stronger verb or a shorter sentence carries it without spending the mark at all.

Styles differ on whether a complete sentence must be capitalized after the mark, and they agree that a fragment is not. The safe procedure is to check which style governs the document rather than apply a remembered rule.

The distinction is container against contained: a book and its chapter, an album and its track, a journal and its article. Deciding which category a given work falls into is the real difficulty, not the marks themselves.

Finite is the word carrying the weight here: a participle or an infinitive will not serve, which is why a phrase built on one is a fragment however long it runs. Testing for a finite verb finds most fragments faster than reading for completeness.

It has a subject and a finite verb, so it looks complete until the subordinator is noticed, and that word is the only thing keeping it from standing alone. Deleting the subordinator is the quickest repair when such a clause is left stranded.

Each type is missing something different, so the repair differs too: attach the clause, supply a finite verb, or fold the appositive into the sentence before it. Naming the type first saves rewriting the sentence unnecessarily.

A fragment is a register choice in advertising, fiction and speech, and a fault in academic and professional prose. What the reader detects is whether it looks chosen: one in a paragraph reads as emphasis, several read as carelessness.

The classification is a description rather than a target, and its real use is diagnostic: a paragraph built from one type only is what produces monotony. Counting types across a paragraph is a fast check on sentence variety.

The pronoun may be dropped only when it is the object of its own clause, never when it is the subject of that clause. That single condition accounts for why some omissions read naturally while others collapse the sentence.

The reduction leaves a participle behind, which can then be read as attaching to the wrong noun — the dangling problem arriving by another route. Reductions are safe when the noun is adjacent and risky when anything intervenes.

It occupies a noun slot, so the test is whether something or it could be substituted for the whole clause. As a subject it can grow long enough to strain the sentence, which is where a cleft or an expletive opener earns its place.

In front it takes a comma; at the end it usually does not, unless it contrasts with the main clause. Placement is also an emphasis choice, since a clause at the end lands on the reader last and therefore carries more weight.

The inversion holds the subject back, which is why the construction is used to introduce something new into a scene. It does not occur with a pronoun subject, so the test of whether it is available is whether the subject is a full noun.

It splits one clause into two so a single element can be spotlighted, which is how writing supplies emphasis that speech would give with stress. The cost is words, so it earns its place only where the contrast is genuinely the point.

The opening spends two words before the sentence says anything, and it postpones the subject past the verb. It is not a fault — it is the standard way to introduce existence — but a paragraph leaning on it repeatedly reads as flat.

The action moves into a noun and a weak verb takes its place, which also makes it easy to drop the actor entirely. Find the buried verb and promote it back; the sentence usually shortens and acquires a subject at the same time.

These are recognizable shapes rather than a matter of taste, which is what makes them editable: due to the fact that, in order to, each and every. Cutting them is safe, because none of them carries meaning the sentence loses.

The end is where the reader's attention naturally lands, so putting already-known material there wastes the position. This is the principle behind the passive's legitimate uses: it exists partly to move the important element into that slot.

Applied across consecutive sentences it produces cohesion, because each sentence begins where the last one ended. A paragraph that feels disjointed despite correct sentences is usually violating this rather than anything grammatical.

Variety is a means rather than an end: a short sentence among long ones emphasizes, and repeated openings flatten. Read the first three words of each sentence in a paragraph, which exposes repetition faster than reading the prose itself.

The four repairs are not interchangeable, because each states a different relation between the clauses. Subordination is the only one that ranks them, so it is the right choice whenever the two ideas are not of equal weight.

Three things move at once, and the deictic words are the ones most often missed: here, now, this and tomorrow all have to be re-anchored to the reporting moment. A shifted tense with unshifted deixis is the usual half-done result.

The tag mirrors the main clause's auxiliary and reverses its polarity, supplying do where no auxiliary exists. Same-polarity tags also occur and carry a different tone — challenge or surprise rather than a request for confirmation.

Say the next word aloud: a consonant sound takes one form, a vowel sound the other. That is why an hour and a university are both correct, and why an abbreviation depends on whether it is read out as letters or as a word.

The choice tells the reader whether they are meant to know already which one, so the definite article used too early implies a reference they cannot find. Uniqueness in context is enough — the door, in a room that has one.

Adding an article narrows a general statement into a specific one, which changes the claim rather than only the register. That is the difference between a remark about a whole class and a remark about particular members of it.

The pattern is largely conventional, but one regularity helps: plural and collective names take the article while singular ones usually do not. Lakes and single mountains follow the bare pattern, which is why the exceptions stand out.

Many nouns sit in both classes with different senses, so the category belongs to the use rather than to the word. A noncount noun needs a unit phrase before it can be counted, and choosing that phrase is often where the precision lives.

Money, distance and time take the mass form even though they can be counted, because they are treated as single quantities. The rule is also younger than it sounds, having been floated as a preference and later hardened into a rule.

The same distinction as fewer and less, one word class over, so a single test covers both pairs: ask whether the noun can be counted individually. This pair attracts less correction, which is why the error survives longer in print.

Countability splits the first two, and register splits them from the third: much in particular sounds formal or negative in affirmative statements, where a lot of is the ordinary choice. That is a usage fact rather than a correctness one.

Without the article the phrase means almost none; with it, some. One word therefore flips the sentence's attitude toward the quantity, and the same contrast holds for few and a few over countable nouns.

The basic split is by sentence type, and the exception is pragmatic: an offer or a request uses some, because it presumes a yes. Any also appears in affirmatives with a free-choice sense, meaning that it does not matter which.

Both are singular, so the difference is viewpoint rather than agreement: one looks at members individually, the other at the set as a whole. Each works with two items while every needs more, which is the practical test between them.

All three are restricted to exactly two things, and they differ in number, one being plural and the other two singular. Neither is also the one that pairs with nor, which is where the proximity rule for agreement comes back in.

The series runs from most precise to least, which makes it easier to recall than three separate rules. The exceptions are the fixed phrases — at night, in the morning — and those have to be learned rather than derived.

The same three prepositions ordered by dimension: zero, two, three. The choice can be a genuine claim, since it says how the speaker conceives the place, which is why one location can take different prepositions in different sentences.

One reports where something is, the other how it got there, so the verb usually settles which is wanted. Informal English uses the first for both, and the distinction is worth keeping wherever the sentence turns on a boundary being crossed.

The restriction of between to two items is a myth: it is correct for any number, so long as the relationships are one-to-one. Among is for an undifferentiated mass rather than a larger number, which is the distinction actually being drawn.

These pairings are lexical, so no rule generates them and a nearly-right preposition reads as an error. They are also where a sentence with two verbs sharing one preposition usually breaks, since each verb may demand a different one.

Compared with sets two things side by side while compared to likens them, so the pair carries a real distinction rather than a stylistic preference. Different than is standard before a clause and contested before a bare noun.

This is the part of the system with no logic to appeal to, which is why it is the last thing a fluent second-language speaker gets right. Reading widely is the only reliable route, and a usage dictionary settles the individual cases.

A single preposition can govern several objects only where it genuinely fits each one. Repeating it also disambiguates a long list, by marking each object clearly rather than leaving the reader to judge how far the preposition reaches.

They are different parts of speech, so the slot in the sentence settles it: a verb position takes one, a position excluding something takes the other. Except also works as a verb meaning to exclude, and that is the case that catches people.

The usual rule holds most of the time, and both words cross over: effect is a verb meaning to bring about, and affect is a noun in psychology. Ask whether the sentence needs an action or a thing, then check for the crossovers.

The two are unrelated in meaning, so this is a spelling confusion driven by pronunciation rather than a grammar problem. The comparison word always has something on both sides of it, which is the fastest check available.