The real issue with the problem space is that it's underconstrained. We don't know how the brain works at the relevant level of detail. We don't know how genes coordinate to control brain development with the precision needed to understand how selection of alleles and novel mutations will impact cognitive capacities relevant for language production and understanding. Although we're not clueless, our understanding is far more basic, far more crude than needed.

Given that we're therefore massively underconstrained in what we can speculate upon, and we know we shouldn't just wildly make things up, the constraints tend to be sourced from gut instinct. But the workings of biology aren't particularly instinctive even to professors of biology, let alone anyone else.

So, very likely, anyone who says they have a good idea of how language evolved is wrong--assuming they go beyond what can be pretty directly read out from comparative biology. But also, very likely, anyone who says they have reason to think language can't have evolved is also wrong. (Full stop. Neurons can implement a Turing Machine to an arbitrary degree of precision, which is equivalent to arbitrary computation. Because the sky is the limit, you necessarily have to know the details to know when something is impossible.)