Not bleeding across "all that much" might well be true, though I disagree since it's kinda the whole strength of the human brain applying learning from one area to another, that's why we're more intelligent than animals and have been more adaptable to new problems. You're also thinking about this in terms of conscious decision making when in our day to day lives we rarely sit down with an open mind and properly think about things.
But even if it is not all that much, it's still there. A million people having a 1% shift in their views due to language adds up. That's why we're seeing groups of individuals who aren't racist/sexist somehow reach answers that statistically do appear to be.
The specifics of which uses are neutral or not is the next debate. Step 1 is acknowledging that language is something that feeds into the "nurture" side of how we think, especially since language is such a large part of our learning during early brain development. With that in mind, it's something that should be debated and selection of words should be considered.
If there actually was even a 1% universal association of the color black with evil (instead of the more specific associations that happen in certain contexts), do you really think anyone in medieval times would have created representations of Jesus, Mary and other highly revered figures in black wood or metal?
It is up to anyone claiming such an effect exists to go and prove it with solid data, enough data to overturn hundreds of years of language use. And if such a link actually exists, and someone actually believes in it, then I would like to see them insist people don't wear black (or red, yellow, brown or whatever other color they think is racialized) to funerals.
Edit: I also want to comment on the idea that the brain's ability to generalize would apply to meaning bleeding between separate contexts that use the same words. I think this is a misunderstanding of how the brain works. Concepts used in our thought later get translated into (spoken/written) language, or back again when listening/reading. But, the concepts are not the words that they are represented by. The black of a funeral is not the same concept as the black used to refer to a black person, even though they use the same word or symbol. Just like I'm not applying anything I know about mice scurrying through my house to the mouse I'm holding in my hand while playing (even though the origin of the word is exactly that animal), there is no reason to think there is any transfer whatsoever between the various meanings of black, even if they are related.
If you're waiting for someone to find a case of someone joining the KKK because they once heard about a blacklist, of course that is never going to happen. This is about one facet of learning that happens over 20 years of someones childhood, and then trying to link it to statistical issues that emerge over large groups. If it were even possible to isolate all the different factors to just look at language, it would take generations to test.
This is where we're coming at it from different sides. I think changing language is a relatively cheap and easy thing to do, that might help fix an issue in some small way, and there's really not much reason not to. The cause and effect is plausible, if not provable on any reasonable timescale.
Metaphor, simile, pathetic fallacy are all powerful tools that really do change the way we interpret the substance of what we read. Colours too, it's not a coincidence the major tech brands have gone with various blues, those blues are deliberately designed to nudge people towards a particular emotional response that is completely unrelated to the brand itself. If word selection really did nothing to change the impression of what we read, then writers and poets would be out of a job.
But even if it is not all that much, it's still there. A million people having a 1% shift in their views due to language adds up. That's why we're seeing groups of individuals who aren't racist/sexist somehow reach answers that statistically do appear to be.
The specifics of which uses are neutral or not is the next debate. Step 1 is acknowledging that language is something that feeds into the "nurture" side of how we think, especially since language is such a large part of our learning during early brain development. With that in mind, it's something that should be debated and selection of words should be considered.