Another word for "which" depends on the job it is doing in the sentence. In a restrictive clause, "that" is usually the better choice. In a nonrestrictive clause, the alternatives are "who" for people, "where" for places, "when" for times, and "whose" for possession. For the interrogative sense, "what" or a rephrased question replaces it. The choice is governed by grammar, not by taste.

English has no single drop-in synonym for "which," and that surprises people. The reason is structural. "Which" occupies several distinct grammatical slots, and each slot has its own replacement set. Treating the word as one meaning with one synonym is the root of most errors.

Mechanism: what the word actually does

"Which" is a relative pronoun and an interrogative determiner. Those are two different machines. In "the report which arrived late," it introduces a relative clause and stands in for "the report." In "which report arrived late?" it selects one item from a set, so it behaves like a determiner.

The relative-pronoun job splits again into restrictive and nonrestrictive clauses. A restrictive clause narrows the noun and is essential to the meaning. A nonrestrictive clause adds a comment and is set off by commas. American style guides generally prefer "that" for restrictive clauses and "which" for nonrestrictive ones. This is the single most consequential rule in the whole topic.

The mechanism is about information packaging. When you write "the car that hit the pole," the clause identifies which car, so it restricts. When you write "the car, which hit the pole," the clause assumes you already know the car, so it comments. Swapping the words changes what the reader assumes is already known. That is why the choice is not cosmetic.

A second mechanism is the "which" that refers to an entire clause rather than a noun. In "the experiment failed, which surprised the team," "which" points to the whole event. No single noun can replace it. Here the only clean alternatives are phrases like "a result that" or a full new sentence.

Quantified context: how often the choice matters

Corpus studies of edited American prose consistently find "that" dominating restrictive clauses and "which" dominating nonrestrictive ones. Journalism stylebooks codify the split; scientific and legal writing often does not. That divergence explains why the same sentence can look correct in one publication and wrong in another.

The practical consequence is measurable in editing work. In a typical 1,000-word draft, restrictive "which" appears a handful of times and is usually swappable for "that" without meaning change. Nonrestrictive "which" appears less often but is far more likely to be a deliberate, comma-marked comment. So the high-value edit is converting restrictive "which" to "that," not the reverse.

Interrogative "which" is rarer in running prose than in dialogue. It shows up in instructions and forms, where it selects among known options. That selection function is why "what" is a poor substitute: "what report" asks for identification, while "which report" asks for a choice from a known set.

Replacement table: matching the word to the slot

Grammatical slot Best replacement Example Caution
Restrictive clause, thing that the file that opened Do not add commas
Nonrestrictive clause, thing which (keep) or a new sentence the file, which opened Commas are required
Person, restrictive that or who the pilot that flew "Who" is fine but less common
Person, nonrestrictive who the pilot, who flew Never "which" for people
Place where the lab where it happened Drop "in which" where possible
Time when the day when it failed "On which" is formal
Possession whose the team whose data leaked Avoid "of which" for people
Clause reference a result that / new sentence it failed, a result that shocked No single-word swap
Interrogative selection what or rephrase what report came? Meaning shifts

The table rows are not interchangeable. Each row is a different grammatical environment, and a replacement valid in one row is wrong in another. That is the core reason a thesaurus fails here.

Edge cases: where the obvious swap breaks

Prepositional relatives are a common trap. "The method by which we measured" cannot become "the method by that we measured." When a preposition precedes the pronoun, "which" and "whom" are the only options. The fix is to move the preposition to the end, as in "the method that we measured by," or to restructure entirely.

The "of which" construction is another edge. "A process the steps of which are unclear" is clumsy but sometimes unavoidable. Most editors rewrite it as "a process whose steps are unclear," even for nonhuman nouns. "Whose" for things is acceptable in modern usage and far cleaner.

Clause-reference "which" has no pronoun substitute at all. You must convert it to a noun phrase or split the sentence. This is the case most likely to produce a fragment if handled carelessly.

Dialogue and informal writing tolerate more "which" in restrictive clauses than formal writing does. That is a register difference, not an error. The problem arises when a draft mixes registers without intent.

Bullet list of red flags:

A restrictive clause wrapped in commas, which turns essential information into a parenthetical. "Which" pointing at a person, which reads as an objectification error to most editors. A clause-reference "which" with no clear antecedent noun, which forces the reader to reconstruct your meaning. "In which," "by which," and "of which" stacked in one sentence, which buries the verb. A restrictive "which" in a document with a house style that mandates "that," which will be flagged in review.

Misconceptions: what the rule is not

The most common misconception is that "which" is simply the formal version of "that." It is not. Formality is a side effect; the real distinction is restrictive versus nonrestrictive information. A formal restrictive clause still takes "that" in American style.

A second misconception is that the that/which rule is universal. British usage is more permissive with restrictive "which," and many respected authors use it. If you are writing for a British publication, follow its style guide. If you are writing American English, the split is the safer default.

A third misconception is that "that" can never follow a comma. It can, when the comma belongs to a different structure, as in an introductory phrase. The rule is about the clause, not about punctuation proximity.

The counterintuitive detail: converting restrictive "which" to "that" often improves readability more than any other single edit, yet the change is invisible to most readers. Readers do not notice "that" doing its job. They notice only when the wrong word makes them pause and re-parse. Good relative-pronoun choice is the grammar equivalent of good plumbing.

Practical actions: a revision sequence

Start by locating every instance of "which" in the draft. Most word processors can highlight them in seconds. Then classify each one by slot using the table above. Do not skip the classification step, because the replacement depends entirely on it.

For each restrictive clause, test whether removing the clause changes the core meaning. If it does, the clause is restrictive, so use "that" and remove any commas. If it does not, the clause is nonrestrictive, so keep "which" and keep the commas. This test resolves the majority of cases.

For person references, replace "which" with "who" or "that." For places, prefer "where" over "in which." For times, prefer "when" over "on which." These swaps shorten sentences and remove prepositional clutter.

For clause-reference "which," rewrite as a noun phrase or a new sentence. This is the highest-effort fix and the one most likely to clarify meaning.

Finally, reconcile your thresholds. House style, national variety, and register all set different defaults. Pick one standard for the document, apply it consistently, and note any deliberate exceptions. Mixed standards are what readers perceive as error.

Limitations: what this guidance cannot settle

Relative-pronoun preference is a style convention, not a law of grammar. Multiple standard varieties coexist, and a construction acceptable in one may be flagged in another. No rule here overrides a publisher's style guide or a legal drafting convention.

Corpus frequency tells you what edited prose tends to do; it does not tell you what your specific audience expects. Disciplinary writing in law and science often tolerates constructions that journalism rejects. Confirm the convention for your venue before applying a blanket rule.

And the deeper limit: the choice between "that" and "which" cannot fix an unclear antecedent. If the reader cannot tell what the pronoun refers to, no replacement word will rescue the sentence. Restructure instead. That is the one case where the synonym question is the wrong question entirely.