Adverb & weasel word finder

When an adverb earns its place

A tool that flags every "-ly" word is annoying within a paragraph and useless by the third page. Most manner adverbs are cuttable, but not all of them are, and pretending otherwise is how these tools lose your trust.

It's load-bearing in an instruction

"Press Ctrl+C quickly to cancel the export" — "quickly" is telling the reader something: act before the window closes. Delete it and the sentence loses information, not just length. Compare that to "the process quickly fails," where "quickly" adds nothing a reader needed.

It's a discourse marker, not a manner adverb

"Finally, restart the service." Here "finally" isn't describing how the restart happens — it's signposting where you are in a sequence of steps. That's structural work, and this tool leaves sentence-initial adverbs like it alone.

The honest limitation

Without a part-of-speech tagger, this tool can't always tell an adverb modifying a verb ("answered quickly") from one modifying an adjective ("quickly obsolete cameras" — rarer, but it happens). It also can't tell you whether a specific instance is the weak kind or the load-bearing kind — that judgment call is yours. What it can do is point out every candidate so you don't have to hunt for them by eye, and say which ones sit next to the classic tell (a be-verb or a reporting verb, as in "was quickly dismissed" or "she said angrily") that usually means the adverb is doing the work a stronger verb should do instead.

Hedges and intensifiers are a different problem

Intensifiers like "very" and "extremely" add emphasis without adding information — "very comprehensive" rarely tells a reader more than "comprehensive" does. Hedges like "somewhat," "arguably," and "some experts say" do the opposite: they let a sentence avoid committing to a claim a reader could actually check. Both are worth a second look for the same reason adverbs are, which is why this tool reports them together, but they aren't graded against some invented "ideal density" — there isn't a defensible threshold, and making one up would contradict everything on why readability formulas fail.