One word swapped
- Input
- Original: the cat sat — Revised: the dog sat
- Working
- The shared sequence is “the … sat”. “cat” appears only in the original (removed) and “dog” only in the revision (added).
- Result
- 1 added · 1 removed · 2 unchanged
Compare two versions of any text side by side and see exactly which words were added, removed or left unchanged, with counts for each kind of change. Runs entirely in your browser — your numbers never leave this device.
Paste the original text in one pane and the revised text in the other, and the checker aligns the two at word level. The alignment is a longest-common-subsequence computation: it finds the longest run of words the two versions share in the same order, marks words that appear only in the first version as removed, and words that appear only in the second as added. Because the alignment works on words rather than characters, a swapped word shows as one clean removal plus one addition instead of a spray of changed letters.
The summary counts words in each class — added, removed and unchanged — so a contract revision or an edited paragraph can be assessed at a glance before you read the highlights. One honest consequence of the LCS method: if a whole sentence moves to a different place, it will show as removed at the old position and added at the new one, because the tool compares sequence, not meaning. It also cannot tell you which version is better — only where they differ.
Both texts are compared entirely in your browser and are never uploaded or stored. That makes this the safe way to compare drafts containing unpublished, personal or client material: close the tab and both versions are gone.
Granularity. A character-level diff highlights every changed letter, which turns a single swapped word into several fragments and is hard to read in prose. Word-level comparison treats each word as a unit, so edits read the way a human editor describes them: this word was removed, that word was added. For code or data where a single character matters, a character diff is stricter; for documents, word level is clearer.
Because the comparison is sequential. The longest common subsequence keeps only words that appear in the same order in both texts, so a paragraph moved elsewhere breaks the shared sequence at both spots: it is reported as removed from its old position and added at its new one. The counts still balance — nothing is reported lost that is not also reported gained.
Yes. If both panes hold the same text, every word is unchanged and the added and removed counts are both zero — a quick way to confirm that a copied version really is identical to its source.
The LCS alignment builds a table sized by the two word counts, so very large documents (hundreds of thousands of words pasted at once) can be slow or run into device memory limits. Typical documents — articles, contracts, chapters, long emails — compare comfortably. For whole books, compare chapter by chapter.
No. The comparison runs 100% in your browser. Neither version is sent to a server, stored or tracked, which is precisely why it is safe for unpublished and confidential drafts.
More in Text Tools · Browse all tools