While I haven't run a stand-alone test for this idea, I do analyze backlinks for maybe several thousand sites over a year's time. I see this principle in action over and over again - it's the context of the linking page that best sends "theme" juice to the link target, and yes,that includes the Title element of the linking page quite naturally, as well as nearby text on that page.

Anchor text on its own is still helpful, most certainly, but only up to a point. Where that point is exactly is hard to say, but too much "keyword keyword" anchor text is a dead giveaway after a while that the link is not exactly a "freely given editorial citation" for the cite.

Do you know of anyone who has. Seems like a likely update that Google would be interested in.

Yes, overdoing "exactly the same keyword" each time can have negative consequences. Need to vary it up.

Its been theorized, and makes a whole lot of sense due to themeing, but it would be very difficult to test accurately.

I think it would be easy test. Just but two new domains, exact same websites. Then get two other new domains and create one link with a title tag like "people dog" and the other whatever. Have both anchor text be the same. Then watch who ranks higher for people dog.

The link title from my site B to site A is shown as the search result title, when I search for site A.