Until recently, he negation pattern ("Not X, But Y") was a useful literary construct that could emphasize the contrast between two (or more) concepts or states.

Then LLMs started using it _everywhere._ Sometimes in the same scene. Sometimes in successive paragraphs—or even the same paragraph! (Yeah, rule of three. I get it.)

Sometimes, we can get the AI to realize and confess its prosaic sins. But it would be helpful if we can find a solution to identify negation pattern in text and somehow highlight it.

One option, of course, is to use a text replacement prompt with a very low temperature and explicit instructions to wrap any detected negation pattern spans in whatever NC-native XML element triggers highlighting. I might try this first anyway.

If at some point you can consider a way to detect it via pattern matching/regex, that would be incredibly helpful. That said, I know that it's not a simple request and that there are complexities that might make the task difficult to achieve.

But hey, it doesn't hurt to ask.