Plagiarism Risk Checker

You wrote a section that pulls heavily from research, and now you're worried it reads too close to the source even though you paraphrased. A real plagiarism scanner like Turnitin compares against a massive corpus you don't have access to. What you can check on your own is whether the text you produced has the structural fingerprints that often correlate with paraphrased-but-not-rewritten passages.

The risk score uses signals like sentence-length uniformity, formality patterns, vocabulary repetition, and syntactic templates that show up when someone substitutes synonyms over a copied skeleton. A high score doesn't prove plagiarism — it suggests the prose has the shape of derivative writing. Use it as a self-edit prompt, not as a verdict, and never as a substitute for actually citing your sources.

The heuristic checks are independent of any source database. Sentence-length uniformity: original writing tends to have high variance in sentence lengths (some short, some long, irregular rhythm). Paraphrased writing inherits the rhythm of the original, so when sentences cluster around a similar word count for many lines in a row, that's a signal. Vocabulary density: original prose mixes common and rare words irregularly; paraphrasing-by-synonym-swap often produces unnatural rare-word density because the original used a vocabulary the paraphraser doesn't naturally choose. Syntactic templates: certain sentence structures (e.g., "It is X that Y") show up at higher rates in patchwriting because the writer kept the structure and replaced content.

A worked example: paste in three paragraphs you wrote yourself versus three you patchwrote from a textbook by replacing every fifth word. The original paragraphs typically score in the 10-30 range; the patchwritten ones score in the 50-80 range. The score isn't ground truth — it's a structural signal. The numbers can vary based on subject matter (academic writing in a tightly-templated field naturally scores higher than personal essay).

The limitations are significant. A genuine database scanner like Turnitin or Copyscape compares your text to billions of existing documents and flags exact or near-exact matches. This tool can't do that — there's no database, no matching, no source identification. False positives happen on tightly-templated genres (legal writing, lab reports, business memos) where the structure is conventional rather than derivative. False negatives happen when the paraphrasing is genuinely thorough — if you understood the source and rewrote in your own voice, the structural fingerprints are gone and the score doesn't catch the paraphrase. Treat the score as a self-edit prompt: high score means re-read with skeptical eye, ask whether the prose sounds like you, and check that any borrowed ideas have proper citations regardless of how original the wording is. Citing well is the actual answer to plagiarism concerns; structural analysis is the warning siren.