I agree with your position (and with Dennett's). "Talking" is not like an independent module with a lede you can attach to a lion's brain to get it to talk. It must be integrated with the lion's brain, and that necessarily changes the brain to the point that it's no longer a "lion." This is because even simple statements like "I am hungry" require the ability for sophisticated abstraction. "Talking" is more than mere vocalization, it's also the thought processes behind the vocalization. Once you've changed a brain so that it is capable of those sophisticated abstractions, it's no longer the same.

Is a person missing half a brain, in a vegetable state with little hope for recovery not human?

Except they're missing half a human brain. A talking lion would not have a lion brain.

In the context of, say, figuring out what makes humans act the way they do: most certainly not.