Why do some edited lines show as one removed line and one added line instead of a single modification?

Because the tool pairs a removed line with an added line as a modification only when they share at least half of their words. Then it shows them as a pair and highlights the words that changed. A line you rewrote from scratch has too little in common with the old one, so it is shown honestly as one line removed and another added.

Published · 3 min read

How the overlap is measured

The tool takes the set of distinct words in each line, ignoring repeats and punctuation, and counts the words the two have in common. It doubles that count and divides by the total number of words in both lines. If the result is 0.5 or more, the two lines become a modified pair.

Worked example

'Free shipping on orders over 200 SAR' against 'Free shipping on orders over 300 SAR': 6 shared words out of 7 in each line, so 12 ÷ 14, a score of 0.857. Result: Added 1, Removed 1, of which modified 1, with only the number highlighted.

The same line against 'Delivery within two days inside Riyadh': no shared word, a score of 0. Result: Added 1, Removed 1, of which modified 0.

The same line against 'Free delivery inside Riyadh only': one shared word, a score of 0.167, well under half, so it too shows as a removal and an addition.

Why this threshold?

If the tool paired any removed line with any added line, it would show 'modifications' between sentences that have nothing to do with each other and highlight almost every word, which misleads more than it helps. The half rule means that what you see as a pair really is the same line after an edit, and what you see as a removal and an addition really is new wording.

Word matching is literal

A word counts as shared only if it matches exactly. 'shipping' and 'ship' are different words for this score, and so are 'order' and 'orders'. That is why 'Free shipping inside Riyadh only' against our original line scores only 0.333, even though the meaning is close. Letter case never affects the score: it is always computed after folding capitals to lower case.

Does this change the result?

It does not change what changed, only how it is displayed. The added and removed counts in the summary always include modified pairs, and the 'of which modified lines' counter tells you how many of them are pairs. In the copied plain-text result both look the same way: a line starting with a minus followed by a line starting with a plus.

  1. Open Text comparison at the Lines level and paste both versions.
  2. Look at 'of which modified lines': those lines were partly edited.
  3. The remaining removed and added lines were rewritten, added or dropped entirely, so read them in full.

Still want to see which words survived?

If a line was rephrased and you still want to see which words stayed, switch the level to Words. There is no pairing rule there: every removed and added piece is marked inside the running text. Comparing our original line with 'Free shipping inside Riyadh only' that way gives: Free shipping [-on-]{+inside+} [-orders-]{+Riyadh+} [-over 200 SAR-]{+only+}, so you can see that 'Free shipping' stayed and the rest of the sentence changed.

Common questions

Can I change the half threshold?

No, it is fixed in the tool.

Do numbers count as words?

Yes, a number is a word, which is why 200 becoming 300 lowers the score slightly.

Does punctuation count?

Not for the score, but a changed punctuation mark still shows as a difference inside the pair.