My suggestion is that you ask your colleague to read current papers with attention to language style. If they are actually paying attention to matters of language, they should be able to notice the difference. However, as an essentially native English speaker (I moved back to the US for 3rd grade) who has been told his sentences are too long and complicated with too many dependent clauses and phrases (like I'm writing in German), I should also say this might be somewhat a matter of personal style.

@AlexanderWoo thanks, as discussed in this comment that is great advice for others in similar situation, but for this colleague nothing conventional has or will work. I need something statistical, analytical, non-opinion-based. What I've described in my question.

Is his style of writing what is still used academically in his native country? That may complicate your desiring him to change.

I don't know how to motivate them and from what you say it sounds like it's going to be difficult, but I'd really recommend trying AI---not to do their writing for them, but as a coach. Have them enter a paragraph, tell the AI this is for a scientific journal and ask for a simpler version, and then ask the AI why this is better. Have them go back and forth. That should give them the opportunity to learn without taking too much of your time.

@JohannesBauer I've never (knowingly, willingly) gone near anything with AI in it. I didn't know there was something to which one could ask "why this is better?" That's news to me! And in fact if that's really a thing and works well, then that may be the actual answer I'm looking for here. I was thinking a statistical analysis of grammar (and its errors), vocabulary, and sentence lengths was all that one could get from electronic resources.

I understand and share your skepticism regarding AI, but I'll recommend checking out ChatGPT or any of the others; the usage pattern I described above is perfectly possible with any of them. They're chat bots, after all. YMMV and you shouldn't take anything they'll tell you at face value, but that's the same as with any other chat partner. Except a human chat partner will not be as patient.

@uhoh Frankly, were you just assigned to these other researchers to "help them with their writing" ? It sure seems more like that than if you were an active researcher on the topic written about. If so, you were unwise to undertake it. If you were an active researcher on the paper's topic, take the responsibility to lead the writing of the paper yourself. The other authors will learn how to write correctly a lot faster this way.

I've used Hemingway Editor to allow style checking on the go. It was for casual writing, not scientific writing, so I can't comment on its applicability to scientific writing, so I'm leaving it as a comment.

The origin/native language of these researchers may be relevant. If they are from a germanic language country, then they may recognise the base of English as German. I believe science use to only be in Latin (or possibly Greek), so maybe that has an influence. Are they from a face-saving country (e.g. middle east or asian), if so it might be the way you approach teaching them (particularly if they view you as lower in some hierarchy). Have you seen them write/speak in more natural English? If they can, then maybe they view it as their style (e.g. make the reader ponder on the work more).

Are you sure that the style MUST change? Clearly it is readable enough for you to be able to interpret it without too much difficulty.

@MikeB if you are uncertain of the importance of a clear, smooth writing style in getting scientific papers through the sometimes contentious review process and accepted by the higher impact factor journals, and then widely read, you can consider posting a new question about it.

@ROIMaison The "classic, no AI" desktop app sounds exactly like what I'm looking for! I will give it a try soon and let you know.

@NAMcMahon You've hit the nail dead center on the head. If they could write like they speak, their papers would really shine. They speak and write from two very different places, as if they are two totally separate 2nd languages. In speech they are quite compelling and passionate about what they want the paper to say, but once the door closes and the computer is fired up, it all disappears. It's very hard to un-learn habits like this, I suspect if there was fMRI we'd see different areas light up between talking and writing, more-so than for the average person.

@NAMcMahon I think bridging that gap will take quite a lot of time and effort, but it's exactly what needs to be done. I may encourage them to turn on some voice memo app on their phone then talk to me about the major points they want to make, then go back to the "writing room" and listen to it. Or even let me (temporarily) record and transcribe (I can do that quicker). Yes there are apps for that as well. I will definitely give this some thought!

@uhoh, I hope it suits your needs. Please report back, I'm curious to see how it does for scientific writing.

Is it factual that "Things have changed, scientific writing is now much less formal and more like natural speech" (to the point of allowing comma splices), or is that opinion? Is the reading ability of the scientific readers lower than it used to be?

w.r.t. the use of AI, keep in mind that just because idiots get terrible results when they use language models as search engines or believe they're self aware, that doesn't mean they're useless. Answering the question, "Does this sentence sound like something a native speaker would say?" is exactly the task these things are actually trained for. (Note that this is not exactly the same as "How would a native speaker phrase this exact sentence while preserving the meaning", which is why you need to combine it with other components (like manually checking the result). But it's a valid tool here

@Ray thanks for that. I have a powerful uncanny-valley feeling about using natural language to interact with things that are not humans or dogs. I may be forced to give it a try. :-)

@uhoh There are probably options that don't have a natural language interface. Well, not in the sense of "Please rephrase this sentence for me", anyway. Just "natural language" the sense that you input a natural language sentence and get a (hopefully) equivalent one out again. Kind of like google translate, but with the source and target languages being the same. (Actually, you might try that: translate their English into their native language, and then back again. The first transform might preserve the idiosyncrasies, and the second remove them? I'm guessing here, but give it a try.)

Unrelated to the question, but how can you get coauthorship just for English editing?

@knzhou Why would you come to such an absurd conclusion?

That's the impression I got from your question -- you said that you were mainly tasked with English editing, didn't even see the manuscript as it was written, and didn't mention doing anything else. I can't imagine ever taking part in a collaboration where I can't even see the manuscript.

@knzhou Go back and read my post more carefully. I indicate that it is mainly me that is tasked with editing/proofreading because I'm the only native speaker, not that I am mainly tasked with editing.