Passive Voice Detector

What is passive voice?

A sentence is passive when its subject receives the action instead of performing it. Grammatically, English builds the passive from a form of “be” (am, is, are, was, were, be, been, being) or “get”, followed by a past participle: “was written”, “is being repaired”, “got fired”. The active version of the same idea puts the doer first: “The committee wrote the report” is active; “The report was written by the committee” is passive. Both are grammatically correct English — the difference is which noun the sentence puts in front.

Most writers can spot the textbook case but miss the variations, which is exactly where a passive voice checker earns its keep. “The bridge must be inspected annually” is a modal passive. “The results have been reviewed” is present perfect. “He got promoted” is a get-passive, common in speech and increasingly in print. A passive voice detector that only catches “was verb-ed” will miss most of these, and this one is built specifically to catch the rest.

This is not a grammar checker, and it does not pretend to be one. It has no opinion on your commas, your spelling, or your subject-verb agreement — it does one thing, which is to find every be- or get-passive in the text you give it and show you exactly what it matched.

How to check your text for passive voice

1. Paste or type your text into the box. There is no button to press — the sentence count, the passive constructions, and the long-sentence list all recalculate on every keystroke.
2. Set what counts as a long sentence. A slider controls the word-count threshold, starting at 25. Raise it for technical writing where longer sentences are normal, or lower it if you are checking copy meant to be read quickly.
3. Read the two lists below the statistics row. One shows every passive construction found, quoting the exact matched phrase and noting whether an agent phrase follows it. The other lists the long sentences with their word counts. Neither list is a to-do list — they are just what is there.

When the passive is the right choice

Every other tool in this category treats a passive construction as a defect to be stamped out. That advice is wrong often enough that following it blindly makes writing worse. The passive is the correct choice when you do not know who performed the action — “The window was broken overnight” is exactly right when you have no idea who broke it, and rewriting it to invent an agent would be dishonest. It is also correct when the agent is irrelevant: “The samples were refrigerated for two hours” does not need to say who put them in the fridge, because nobody reading a lab report cares.

There is a third case that has nothing to do with hiding an agent: sometimes the object of the action is what the sentence is actually about. “The suspect was arrested at 6 a.m.” keeps the suspect as the topic across a paragraph that is about the suspect, not the arresting officer. Forcing every sentence into active voice to satisfy a style checker breaks that thread and produces prose that reads like it was edited by a rule rather than a person. Scientific papers, legal documents, and news reporting all use the passive constantly and correctly, for exactly this reason.

So this tool counts passive constructions and shows you where they are; it does not grade you on how many you have, and it has no target number for you to hit. Deciding which ones to keep and which to rewrite is a judgment call about what the sentence is for, and only the person who wrote it can make that call. A tool that told you to remove all of them would be giving you bad advice, so this one does not.

How the detector works, and where it gets things wrong

The rule is the grammatical one: a form of “be” or “get”, followed by a past participle, counts as a passive. Because real sentences put words in between — “was never properly examined”, “is routinely marketed” — the detector walks over up to three adverbs or negators between the auxiliary and the participle rather than requiring them to sit next to each other. It also looks several words past the participle for an agent phrase starting with “by”, so “was built in 1890 by a Scottish firm” is correctly marked as naming its agent, not just as a bare passive.

This approach was tested against a hand-built set of 54 sentences covering every passive construction in English — canonical, agentless, progressive, present and past perfect, future perfect, modal, modal perfect, get-passive, adverb-separated, negated, irregular participles, and infinitival — plus 24 deliberately tricky active sentences designed to look passive: active present perfect, active progressive, “be” plus an adjective, adjectival “-ed” forms, “get” plus an adjective, and participles used as modifiers with no auxiliary at all. This heuristic scored 54 out of 54. Run through the same test, compromise, a real part-of-speech tagger, scored 42 out of 54, missing modal passives, get-passives, and every adverb-separated case. write-good, a widely used prose linter, scored 43 out of 54, missing the same adverb-separated cases and get-passives, and additionally flagging four adjectival sentences — “He is tired”, “The students were bored”, “I am interested in the offer”, “She was worried about the result” — as passive when they are not. On a sample of real prose, all three tools still disagree with each other on about 5% of sentences, which is the honest state of the art: three independent implementations of one grammatical rule, and they do not fully agree.

Some disagreement is unavoidable because English itself is ambiguous, not because any detector is careless. “The window was broken” is genuinely two different sentences with identical words: it can mean someone broke the window, or that the window was already in a broken state, describing it the way an adjective would. Nothing in the sentence tells you which, so this tool flags it and lets you decide. A few other results are flagged honestly as debatable rather than hidden: “He was gone by morning” is stative, but “gone” is an irregular participle; “I was used to the noise” is adjectival, but “used” ends in -ed; “She was supposed to call” is idiomatic; and “The office is located on the third floor” could be read as an agentless passive or as simply describing a location. This is a heuristic over the shape of words, not a tool that understands meaning, and every result here is worth a second look rather than a verdict to obey.

Why is it free?

Everything above happens in your browser as you type. There is no model call, so no text you paste ever leaves your machine or reaches a server — this tool has no idea what you wrote, and neither do we. That also means no per-use cost for us to recover, which is why there is no account, no usage limit, and no watermark on anything it produces. It runs the same way whether you check one sentence or paste in a whole manuscript.