You sort of hit the nail on the head when you wrote about the unification of the language in the 19th Century. National languages are 'compromise' languages - though more often than not a particular dialect (usually for social/power reasons) will have 'preference'. The languages of Italy haven't devolved away from Italian, but are moving toward a more unified Italian. What is a Vulgar Latin? In the way you seem to talk about it here it would appear that the Italian peninsula took on some formal 'Latin' - in other words it is treated as if everybody became Latin speakers and then devolved from that. That isn't the case. What we consider Latin (Classical Latin) was itself a particular dialect (or socialect) within even Rome itself. Because of the economic and political importance of Latin then, depending upon an individual's/group's social standing, different levels of Latinisation will have affected the already divergent dialects/languages of the Italian peninsula. It's not 400 years of regionalism,. not even 2000 years but 18,000 years of shifting regionalism and shifting contacts.

It is precisely the 'Linear' approach of the language tree that is a problem. You talk of mixing and consequent change but....the model upon which language change is formulated to have occurred categorically deny the opportunity for 'mixed' languages. they have to be classified as one or another. French is argued as being a 'Romance' language, but it can be argued as a 'Germanic' language also. Within the tree model such cannot be argued; it has to fit into the structure upon one branch or another.

It is a model, and it is utterly incoherent with how we see languages actually mutate. It is also misleading in terms (as you argue it here) of the idea of 'peoples', and their migrations. There is a distinct hole in the archaeological record of the kinds of mass movements that the Indo-European 'expansion' is based upon. When Koch says that what we call 'Celtic' might be a much deeper root within European languages I think he is suggesting deeper (older) than current ideas of an Indo-European/Neolithic language migration. There is a 15,000 year gap between the ice age refuges (and consequent language isolations) and the first written languages. There is no Indo-European refuge; whatever language might have existed within a Ukrainian or any other refuge will have been mixed up with the others by 15,000 years of contact.

There is needless talk of language branches, with the concurrent idea that each language is derived from only one other language; and along with that goes the idea of genetic groups or 'peoples' - so that we see blue eyes, red hair, tallness etc. being linked with these 'peoples' - whereas there are blond haired, red-haired, blue-eyed, tall Romans and Greeks - equally there will have been dark-haired, dark-eyed, short Celts and Germans. There are clines within both genetic and language structures in Europe, but there are not clear demarcations; there is a great mix up within the structure of both.

Of the two language (when addressed outside of the constraints of linear genealogy) might offer an insight into the contacts and interactions made with other societies. In my opinion we have for too long pre-judged those relationships on false notions invented in the 18th Century; we are still looking for migrational 'take-overs', for long-lost (and falsely conflated) shared languages and genealogically self-sufficient 'peoples'. The first of those notions does not show up in the archaeological data, and the other two are narratives super-imposed upon that basic idea, with no evidential (or scientifically causal or common-sense) basis.

I have seen arguments made defending historical linguistics as being a serious science. I beg to differ. Serious science generally publishes material openly; getting hold of linguistic papers is near impossible unless you operate within the tight confines of historical linguistics. More pressingly, when sciences find that their models are insufficient for purpose they drop the model; they don't carry on with obsolete formalisms. As an example; one would not find any serious chemist referring to Carbon Dioxide as a form of phlogiston.