I was reading in some publication, I think it was PLOS, but regrettably cannot find the link, that there was some debate as to whether Latin was the parent language of the Romance languages.

Although this was a side issue, and just mentioned the fact of the controversy in passing, I wonder if anyone knows about this?

I always thought it was more or less established that Romance languages came from Latin, although thinking it over I guess you could argue that the languages evolved from Vulgate Latin. And I do not know whether then these, Latin and Vulgar Latin,  could be interpreted as separate languages. My feeling would be that these were two coins of the same language, something like a literary and a spoken variants of the same language, but maybe they were farther apart that I supposed. Could this be what they meant?

Found the link http://www.sciencemag.org/news/2015/02/mysterious-indo-european-homeland-may-have-been-steppes-ukraine-and-russia.

The pertinent paragraph says, with the underlining in where they mention some linguist question the direct descent from Latin.

The second new paper to address PIE’s origin, in press at Language and due to be published online during the last week of February, uses linguistic data to focus on when PIE arose. A team led by University of California, Berkeley, linguists Andrew Garrett and Will Chang employed the language database and evolutionary methods previously used by Gray to create a family tree of the Indo-European languages from their first origins in PIE. But in certain cases, Garrett and Chang’s group declared that one language was directly ancestral to another and put that into their tree as a certainty. For example, they assumed that Latin was directly ancestral to Romance languages such as Spanish, French, and Italian—something that many but not all linguists agree on—and that Vedic Sanskrit was directly ancestral to the Indo-Aryan languages spoken on the Indian subcontinent.

I would be interested if someone knows who these linguists might be and if their case could have some validity, though a priori I cannot think what could it be. In retrospect I find it more troubling, since in this context, it would seem as if it did not matter whether it was Vulgar Latin or literary Latin.

No serious linguist would suggest that the Romance languages aren't related to Latin.

But the case you show here is actually one where a more subtle distinction is very important. They're using statistical methods to relate sound changes over time. The exact varieties and assumptions about relationships are important. We wouldn't want to use Italian (even though it's a direct descendant of Latin in some sense!) as a proxy for the whole Romance family. Likewise, assuming Latin is a DIRECT ancestor of the Romance languages is difficult because that doesn't explain some sound changes (etc.) shared throughout the Romance languages.

It would be like trying to run DNA tests on your family tree while not being certain about classifying one of your ancestors as your grandfather, your great-grandfather, or your great-uncle. Obviously for DNA, or the statistical methods they're using in that study, there would be an important effect.

Greek and Latin do share a common ancestor (a few thousand years earlier) but Latin is not descended from Greek. The relationship is more like sisters, or even cousins, than parent-child.

It is necessary to distinguish between Vulgar Latin and Proto-Romance.
... Labelling points on a continuum is arbitrary,

These two statements contradict each other. Certainly we can define them differently, and even arbitrarily make a distinction between the two at some point in time (or in society or in space, etc.), but since neither one is well-defined (e.g., by large amounts of texts), the distinction is relatively insignificant. That was my point above. I don't disagree with your points, but they are, as you say, arbitrary. Languages don't have historical boundaries, that's just our interpretation.

but when the point was reached that what people were speaking was not what from today’s perspective what would be considered Latin, it cannot be considered to be Vulgar Latin.

There is no such point. It doesn't happen one day that a language becomes something else. Latin didn't one day become Italian (or Proto-Romance, or whatever). And more importantly in this case, the timing doesn't entirely match up: Proto-Romance must have been roughly contemporaneous with the Latin or Ancient Rome, because that's how far back the roots go.

Since we do not know what this speech was like we have to reconstruct it and the hypothetical language is called Proto-Romance.

By definition "Proto-" languages are not attested. So in that very technical sense, we can only apply the term to something we don't have documented. But that would mean that we can never discover a proto-language (e.g., a new tablet somewhere) which is sort of an odd implication.

Vulgar Latin was a real language with dialects and a history; Proto-Romance, which is reconstructed by working backwards from attested forms of Romance languages, is a hypothesis outside time and space and was spoken by nobody.

This is a misrepresentation. Proto-languages as reconstructed probably never existed as such, but there still was some real-world language that corresponds to the hypothesis (and varies in some details). Proto-Romance (in one form or another) certainly existed, in order for the daughter languages to share so many features. Various reconstructions have been proposed, and they may or may not be accurate-- but that accuracy is based on real history, whether or not we're able to actually ever check them against real data. For Proto-Romance it's actually relatively easy to reconstruct because we have so much data. But if you look at another situation, like with Proto-Indo-European, there is a lot more controversy (and less data), and that's a situation where we also still know there really was some language spoken at some time, at some place, by some people, but we don't know the details, neither sociolinguistic nor linguistic. Thus no one ever spoke our reconstruction of PIE exactly as reconstructed, but certainly someone spoke a language that was the ancestor to the related Indo-European languages. So the wording here must be very careful not to overstate things, in both directions.

Now, there is one way in which "Proto-Romance" probably does not hold up as a concept. That is because it did not develop at a particular place in time and space, and then "split" into all of the modern languages, in the exact way that reconstruction would suggest. There were some changes already in place for Proto-Romance, according to the reconstructions, that may have not been in place for Vulgar Latin. In that case, true Proto-Romance would have split earlier (e.g., during the time of Roman conquests) and not yet had all of the reconstructed shared features. How then did those shared features arise? Mutual drift. That is, some changes were already "going to happen" or "in progress" (in some sense) when Proto-Romance split, such that when we reconstruct it, it looks like they happened before the split, because it was consistent across the varieties. Regardless, this just means that our reconstruction is challenged by complex data, not that there was no such language. Reconstructions will never give us a perfect image of how languages sounded, but they do a reasonable job of telling us that there is a common ancestor and some things about it.

However, when it is remembered that Latin was just one Italic language spoken in Italy in antiquity, it is interesting to speculate whether and to what extent the languages of Italy classified as Romance have descended from or been influenced by non-Latin Italic languages.

Maybe. But Latin itself was influenced by its neighbors, and most of the Romance languages developed outside the domain of the Italic languages, and after they were gone. So you might legitimately find some influences in the regional dialects/languages of Italy, but this does not seem to be the right way to explain Portuguese, for example (although other substrates elsewhere did have effects too).

These languages are very poorly attested,  but it seems that some of them were closely related to Latin.

There are descriptive grammars available for several of them, so the situation is not that dire.

Obviously as Rome expanded Latin came to dominate the non-Latin Italic languages.

But again, Latin spread beyond that region as well, and before (all of) the non-Latin Italic languages had disappeared. They're really separate things, neighbors, not just a situation of replacement.

It can be imagined though that varieties formed which were some sort of amalgam of Latin and the local language from which present day languages of Italy are descended.

Yes, to some degree, as I mentioned above. However, standardization at the time meant that Latin mostly replaced the other languages there, and then during the development of the Romance languages they were sort of on their own course. Thus I don't think there was a substantial amount of influence, certainly not something like 50-50 mixing. I'm sure there's some research into the influence of the ancient Venetic language on Venetian (Italian dialect). But again overall this influence was not so strong. For example, the Romance languages outside of Italy certainly differ more (and indeed due to substrate influence!) than those within Italy, for the most part.

It is often said that the languages of Italy classed as Romance differ more from each other than standard Italian (a partly created language) does from, say, Spanish.

Source? There may very well be some standard-influenced features shared in standard Romance varieties not found in dialects, but to say the Italian languages/dialects are more distinct than say Italian vs. French seems extreme to me.

This may be due to the languages being clustered round a centre of origin, but the possibility of non-Latin Italic languages being involved cannot be ruled out.

Yes, it makes sense that, setting aside later contact and merging, dialectal variation within a nearby area (Italy) would be higher than variation farther away. Think of concentric rings of distance, then dialects clustering around social situations.

So I don't disagree that there may have been some influence, but your arguments here seem a bit too strong to me. I suppose one argument in your favor could be that the modern attestations of Italian languages/dialects are heavily mixed with standard Italian, obscuring and thus under-representing the regional influences. Yet there are cases where there is clear influence, such as Greek in southern Italy. There would need to be some explanation for why very clearly Greek features are preserved there, while there is less strong evidence for a direct effect of other non-Latin Italic languages. Perhaps the explanation would be that more distinct (e.g. Greek) features could be preserved while less distinct (e.g. non-Latin Italic) features might merge.

Unless archaeologists come up with significant finds we shall never know.

Certainly some details remain mysterious, but have you looked into the better attested cases? We really do know something about many of the non-Latin Italic languages, and I'm certain that someone has looked into this influence. I haven't read that research myself, but I'm sure it's out there.

The point I want to make is that Proto-Romance and Vulgar Latin are two different concepts.

The "Evening Star" and the "Morning Star" (=Venus) are different 'concepts' but still point to the same thing. That could also be the case here.

Proto-Romance is a “language” constructed by linguists based on what is known of Romance languages.

No. Proto-Romance, as a concept, is the ancestor to the Romance languages, nothing more or less than that. There are additionally many (sub-)hypotheses giving various levels of detail about what it might have looked like as a language. Those two different levels must be distinguished, because the detail-oriented hypotheses can be falsified by details, whereas the more general hypotheses that some language existed can only be falsified by showing that the alleged daughter languages are not related. In the case of Proto-Romance, our understanding of the details certainly has changed and will continue to change, but the likelihood that there was no common ancestor to the languages is basically zero. I think of "proto-languages" like variables, say "X", in algebra: it's there, to be found, and by studying it, we can learn more about it. One particular proposed value may not end up being the right solution, but there still is, in principle, some value that will fill in that blank.

It is incomplete because the written record of some of the languages does not go as far back as it does for others.  If there is an intermediate stage between something called Latin and the things variously called French, Spanish, Italian etc then calling it Proto-Romance means you are giving the same name to two different things, one hypothetical and the other real. If you call the intermediate stage Vulgar Latin you have the problem that this is also what you call the language spoken in antiquity. There is of course no reason why the same label should not be applied until we have significant differentiation, so long as it is understood that the Vulgar Latin well back in B.C. is not the same as the Vulgar Latin well into A.D. The problem seems to be that there is no name for the immediate precursor of the Romance languages, that is or something which would not be recognised as anything we label Latin, but not yet sufficiently differentiated to justify applying more than one label.

You're right that the names are not precise (and I'd say cannot be, even in principle).
At the same time, the real problem here is that Latin "split" in various Roman dialects dispersed across a large area well before any of those dialects were mutually unintelligible with each other. Trying to actually pinpoint an individual thing of "Proto-Romance" doesn't make sense, because it was already at the time a dispersed, complicated entity. "Vulgar Latin" can apply to all of those varieties (and as you say, across time as well as space), but "Proto-Romance" does not hold up as a simplistic reconstruction because it wasn't a single point in space/time to begin with. Yet there still was an ancestor to the Romance languages, which we can refer to as "Proto-Romance", and in effect is exactly the same as "Vulgar Latin" over a large span of time.

Your point about time is crucial: there was once a heated debate about whether Hittite was daughter of PIE or sister to it. And to me, that question makes no sense. PIE was a span of time, not a point, and in that sense, Hittite (Anatolian) simply was the earliest daughter, and the rest followed later. By adding in and recognizing time as a factor in all of this, it begins to make more sense, but also becomes more difficult to describe. To be expected, I suppose.

There is a bit of a conundrum with labelling points on a continuum where the points at each end have different names. If you say that Latin and French are two different things then there has to be a point where Latin stopped being Latin and became French even if you cannot identify where it is.

Yes.

If you say there is no such point because you cannot identify it then you are saying that Latin and French are the same thing.

Yes, indeed. This is the only possible conclusion. French is newer Latin, and Latin is older French. We can crudely say things like "people speak French today, and spoke Latin back then", but that's just a macroscopic view, and simplification. Old English didn't die out, it just changed into Modern English. And similarly Latin never died-- it just became Italian (and French, and Spanish, etc.). That's a fundamental truth, and any question or answer not assuming it will inevitably be misleading and end up in the contradictions you describe.

If Latin had only developed into one language there would be no problem with labelling, apart from deciding where the boundaries are between Ancient, Middle and Modern Latin.

Languages have no inherent boundaries, especially not in cases like Romance where there is a clear dialect continuum-- similar at close ranges, more distinct farther. Neighbors always understand each other. There are social constructs of standard languages which can be reasonably separated from each other, but indeed Modern Latin is spoken in Spain, in France, in Italy, etc. It's a broad dialect continuum, not (entirely) mutually intelligible at the extremes of course. Recognizing this makes a lot more sense than applying arbitrary labels, like imagining that Latin died once French and Italian became mutually unintelligible. There's no such point (certainly not a specific one). The labels are just the boxes we'd like everything to fit in neatly today-- but they don't really hold up.

Yes, exactly right. The premise of the question presupposes that our analytical categories refer to entities out there in the world (or history) to be observed and classified (and presumably sharply distinguished). But it really doesn't work out well.

The first is that “proto-” when preceeding a language implies two different things. One is that you are looking at a lot of reconstructed forms arrived at by using the comparative method. The other is that you are referring to an undocumented language which immediately preceeded a known language or languages. Whilst it is possible that the reconstructed form is close or even identical to the unknown language, conceptually the two are different things. One is a real thing which is unknown and the other is a reconstruction. The real thing is fixed while the reconstruction is liable to change if new information becomes available and would be different if any of the information we do have had not been available.

Yes, that's correct. And "Proto-X" should always refer to the real, unknown entity, only approximated by specific hypotheses/reconstructions about its structure, just like we're not sure exactly who spoke it where, and various other things. The proto-language is very real (assuming the languages actually are related!), but the details remain a mystery, and even a good approximation via reconstruction won't really settle that.

The second is that Proto-Romance, whichever of the above you mean, is not as far back as we can go. It occupies a gap in knowledge as we have a huge corpus of Classical Latin which cannot be ignored, even if it is acknowledged that it is Vulgar Latin which is an ancestor of Romance languages and not Classical Latin.

OK, I guess I agree in a sense, at least for terminology. But what about comparing this to a number sequence, like "1,2,_,4,5"? We know there is a "proto-number" (gap in the sequence) in there, and we also can try to reconstruct it. I don't see why that is fundamentally different from having a gap at the beginning or end of the sequence. In that sense, "Proto-Romance" shouldn't be a problem at all. (A remaining problem is that it never existed as such in a single place, because it was already in place, all around the Roman Empire, before it became something "different" and drifted in the same direction, in those somewhat separated parts, fueled partly by continued contact, as I wrote about above.)

Vulgar Latin is an ancestor of the Romance languages, but is has a very long history. Accordingly, if Proto-Romance is the immediate ancestor of the Romance languages it is confusing to call it Vulgar Latin.

I guess. I just don't see this as much of a problem, because "Proto-Romance" also can be said to have a long history, even all the way back to PIE (why not?). The idea of "oldness" of languages is meaningless, so instead what you're talking about here is traditional assumptions about spans of time. And that's true, but that's like saying Old English isn't the ancestor to Modern English, because there was Middle English in the middle. More importantly, there's no clear reason to think that "Proto-Romance" and "Vulgar Latin" did not substantially overlap, even if they were not coextensive temporally.

Anyway, terminology aside, it seems like we mostly agree.