And in Chinese – which is different to word level differences since doesn’t have Chinese identifiable spaces like other European/all? languages.

WinMerge, Meld Merge and Kdiff3 can all identify character level differences, but are all unable to merge individual instances found in the same line (or even in adjacent lines for some weird reason).

Beyond Compare seems to be the only one that can, although its user interface requires a bit of getting used to, perhaps because its so feature-rich.

Cost money though.

Update: I was too quick to claim Beyond Compare has this ability. It works for the “adjacent lines” case, and I jumped the gun; multiple differences within the same line cannot be merged individually 🙁