How it works
The detector looks for a specific pattern: a form of the verb “to be” — am, is, are, was, were, be, been, being — optionally followed by an adverb, then a past participle. It recognizes regular participles by their -ed ending and irregular ones from a built-in list (“written,” “taken,” “shown,” “found,” and so on). Constructions with “get” such as “got rejected” are matched too.
This is pattern matching, not grammar parsing, and it produces false positives. “The results were consistent” and “the sample was small” match nothing, but “the finding is well established” and “the data are limited” can be flagged even though the participle is working as an adjective. Sentences with an intervening clause between the auxiliary and the participle can be missed in the other direction.
Because of that, the tool highlights candidates rather than issuing verdicts. Each highlighted sentence is shown in full so you can decide in a second whether it is genuinely passive, and the percentage is calculated over sentences that contain at least one match.
Detection runs entirely in your browser on the text in the box. Nothing is sent anywhere, and the highlights update as you edit.
Why it matters for academic writing
The passive is not an error, and in methods sections it is often the right choice because the procedure matters more than who performed it. The problem is unintentional passive: an introduction where nobody ever acts, or a discussion where findings “were interpreted” with no interpreter. Seeing the density lets you choose the voice section by section instead of by habit.
