The subject of etymology as far back as the Roman period is one that generates a lot of heated debate without much light. In my view, having been subject to this “debate”, it is safe to say that most etymologies get accepted, not through strength of evidence, but by the literal FORCE of argument. The result is that this area attracts those with forceful views and it seems the ability to argue that, black is derived linguistically from white, often by creating the illusion of some rigour by introducing linguistic rules akin to:
Obviously, I am oversimplifying (a bit) and venting my frustration at being drawn into numerous pointless arguments. But, trying to work out which if any ancient etymologies have any merit and giving due to considering to all potential languages is proving a massive nightmare.
For a start we simply have no idea what language was being used largely because of the Celtic Myth. We have all the indigenous languages and all their dialects, then Latin, then all the potential languages that may have been spoken by Roman solders. Another is that the area is beset with nationalistic sensibilities with people trying to build historic “empires” that “demonstrate” Britain was theirs in the past.
Thus, because etymology discussions tend to have all the scientific kudos of a bun fight, most reasonable people leave the subject alone.
The result is that in the 19th century almost all attempts to divine meaning from the Roman place names were undertaken by those with an interest in the minority nationalistic languages like Gaelic and Welsh. This was encouraged by a movement that had started in the 18th century, which sought to show that Britain was “Celtic”: a rather absurd idea even at the time, given that not a single ancient writer even suggests such a thing and that the home of the Celts was clearly located in continental Europe in Northern France.
In fact rather than showing Britain was Celtic, the evidence is clearly the reverse: at every point that ancient writers had an opportunity to describe Britons and Gauls as one race, they clearly and unequivocally referred to Britons and Gauls as two distinct groups. The closest, we have is that some belgique Gauls invaded shortly before the Romans and inhabited the “nearer shores” to France (easily identified as those regions in the SE England with Gaulic coins).
The rational used to bring the “Celts” to Britain, is rather akin to that used previously to give Britain a Trojan origin. In the Celtic area of France there are stone monuments. In Britain (and many many countries worldwide), there are stone monuments. But a Welsh nationalist decided that because Britain and a very small part of Celtic France had not totally dissimilar monuments (dating from the Neolithic) that Britain and France must both be Celtic, several thousand years later. That is a bit like arguing that because Britain and Turkey both have Roman remains today, that Britain today is Turkish.
There are many such vague ancient likenesses, some real, and most not. For example, there are very similarly looking round buildings in Sardinia (nuraghe) and Scotland (brochs). Using the same argumentative style, Scotland would have become “sardinians”. Others have used the same argumentative style to suggest we are a lost tribe of Hebrews, Scythians and even extra terrestrials. None are impossible, but all are improbable. So, given that such arguments were common, it is just a matter of historical accident that we in Britain were reinvented as “Celts” whereas during the same period Germans using the same bogus historical “rational” decided they were Arians. To put it bluntly the myth of the British Celt is now an enormous embarrassment to British academia. It is a dead horse standing, that would have been allowed to fall long ago, if it were not for the nationalistic academics for whom it has become an icon of their (modern) culture.
The British Celts are a myth, but that does not mean Welsh and Gaelic are not strong contenders for Roman period names, but it does mean that any etymologies that start from a presumption that only Welsh and Gaelic (and members of each family group) can be considered as ill conceived.
And even if you are one who takes the view that "celt" is now in such common use that, although incorrect historically, "we all know what it means" ... do you really?
Do you mean:
The Celts as identified very explicitly by Caesar: a subgroup within the Gaul (there were three: Belgae in the NE, Celtae in the center and the Aquitani. There was also a former Greek colony on the Mediterranean).
The Gauls (of Gaul = France)
The areas that had been at some point occupied by Celts/Gauls which was loosely referred to as "Celt" or "Gaul" by Roman/Greek Writings. (Britain was never included in this area)
Anyone anywhere near the above who were similarly "barbaric".
The Modern day mythical "British Celts". Which now means anyone who was never identified as a Celt by anyone in the ancient world, and who doesn't live in the "nearer" shores to France (SE England) where we were told there were Belgae Gauls and so only became "Celt" after the 18th century reinvention)
I suggest the main reason "Celt" has proven so popular, is that it is now used so loosely, that it can, and does, include anyone who the writer wants it to include. Thus if you want to include in a book artefacts from as far apart as Ireland, Russia (or even China based on Tocharins) you can fabricate one universal and mythical pseudo "culture" to bring all these disparate elements together under the fake name of "Celt" which allows you to fill your populist books with historical nonsense. But in terms of understanding the past, it does nothing at all but lead to misunderstanding and confusion. As such, in my view "Celtic" is now a hallmark of poor quality and/or biased research.
Anyone who has ever studied etymologies with a known providence to an ancient language will be struck by several things:
That many place names share common elements such as “Ford” in Old English.
That most place names are relatively mundane. So we get a lot of names based on people's names like “Edwin's town”, Harold's field or common elements with a simple but not very helpful addition: Ashford, Red-burn.
That except for personal names, it is often impossible to understand why a word was chosen for a place. This seems to be because features used in place named tend to be relatively local ones that are usually not specific to one locality and have usually changed with time beyond recognition.
In contrast, you will find that so many "Celtic" place name “etymologies” for Roman place names tend to be exotic, often citing a name based on some god and are used to "prove" the location of the site. As such they are not credible. They lack the common garden elements (ton=town, leah=field) that tend to get repeated in real names and somehow, and whereas real names are hard to locate, "Celtic" Roman period etymologies always seem to be cited as part of a “convincing” argument to locate the name at some place.
Another way that people are misled by etymologies is the common way that a single word in the original language will be translated as a whole sentence in English. Oxford: “a river crossing used by Oxen", is in reality nothing of the sort. Instead, as we can all see, it is an "Ox – ford". Yes, as suggested, it could be a ford used by Oxen. But it could also be a ford that was only fit for Oxen (perhaps deep and muddy). Or it could be a ford that had the character of an Oxen (big and stubborn?). It could also be a local name derived from another name, like a field used for Oxen (and the oxen may never have habitually crossed the river). It could even be a totally different name. For example, in Oxford there is a place called “Osney”. Ney is a common suffix meaning island. So this was likely Os-island. But wouldn't there need to be a crossing to get to the island? If so, wouldn't this be called “Os-ford”, a name that as it became used by outsiders unfamiliar with the local dialect, might easily have been transformed to Oxford. So it is possible the name Oxford had nothing to do with Oxen.
As the current guardian of Roman Britain, it is my role to try to improve the quality of material. In many areas the interpretation can be justified by evidence This is the preferred position. Even in areas of literature when we know the original language, both the words used to derive the name and its meaning are often uncertain. So the situation is all the more problematic for Roman Britain when we don't even know the original language(s) which were present at the time of Rome. This problem is worsened by people (usually believing the Celtic myth is real) who will dogmatically and persistently argue one specific etymology from one specific language out of the numerous possibilities available.
Overwhelmingly people are far too confident that “their” etymology or “their” particular language choice is right, when it is clear that most are not.
The British Celts are a fiction
Welsh and Gaelic speakers are not fiction, but nor are English ones.
There is good evidence that Roman period names can survive in local place names but that they can be very distorted. For example “Vente” and “Winchester”. There are a bewildering array of changes that have occurred to create modern place names from Roman ones but in general:
Some consonant forms should be considered as equivalent such as U ↔ V ↔ W.
Consonants are retained better than vowels.
Except for obvious prefixes, the initial part of a Roman period name retains its form better than later parts.
The ending (assumed to be a grammatical element in the native language + Latin/Greek) has been lost in almost all Roman names.
In addition, in many instances there were copying errors such as r ↔ n or ci ↔ a.
Whilst many etymologies cite various “linguistic rules” to justify their derivation, it is noticeable that it is difficult to find any list of such rules or find them being applied with rigour.
Very few real places can be located based solely on the derivation of their name (unless referring to another place like Walton-on-Thames). As such it is very unlikely that Roman place names can be located using their etymology even if this were known for certain. A typical example of this form of argument is the supposed etymology of BEGESSE taken to mean “ridge”. From the Ravenna Cosmography, we know this is somewhere on the Antonine wall, which largely runs on a ridge to the South of two river valleys. Indeed, the name “Kirkintilloch”, a town on the wall, is believed to derive from a word meaning “end of the ridge”. But despite "a ridge" being the general form of the landscape along the wall, this supposed “etymology” has been used as the sole justification locating it by some at a specific site on the wall at Castlecary: a site which is unique along the wall, not for being a ridge (which is not the obvious description here), but because of the steep sided river valley which concentrates road, rail and canal into a very small area. The name could have meant ridge. It is not helpful to say the etymology justifies its use for Castlecary, when most sites fit this description and there are many better contenders on the Antonine wall.
Likewise, many etymologies cite numerous great authorities who back the etymology, but are strangely reluctant to give the actual evidence in terms of specific words and forms used to justify the etymology. When I have checked such etymologies, I have found that most "authorities" are just regurgitating the same information, and without exception where there is an credible original source, their etymology is stated in far more uncertain terms than those citing them.
"Indo-European" is a set of "words" created by removing all distinct features of the individually unique words in each language except a few apparent similarities. As such it creates a dictionary of "words" which are far from distinct, have no features unique to any language and this makes them much easier to "match" than real words. They are the "fast food" of linguistics: easy to fit into any place and culture but with little substance. As a result it is very easy to find a supposed "match" where none exists and the result is a very high, and unacceptable (in my view), probability of a false etymology when using "Indo-European".
As such, I am going to take the following position. (Unless or until I am able to better understand the geographical reach of the various languages of ancient Britain).
Current pages with etymologies that assume the original language can only be Welsh or Gaelic and/or which give only one possible derivation for the name and/or which “prove” a location based on etymology should be considered unreliable until they can be verified.
Welsh (Cornish, etc.), Gaelic (Irish, etc.) and English (Scots, etc.) will be considered equally as potentially indigenous languages groups in all areas of Britain unless or until evidence (not opinion) clearly shows otherwise. Other languages will also be considered possible for etymologies in Britain particularly Latin but also various ancient forms of Germanic and Romance languages.
The aim (over time) will be to have etymologies that allow for a range of possible languages amongst which there should be Welsh, Old English and Gaelic as well as Latin and other languages that realistically could have been spoken by Roman troops.
I will attempt over time to make etymologies reflect the large degree of uncertainty that exists not only about the original language, but also the meaning.
Location is best ascertained using geolocated information such as the Ptolemy map or Antonine Itinerary.
There is good evidence that Roman period names can survive in local place names, as such, whilst far from ideal, linguistic “closeness” can be used to help decide place name locations.
As (non-Latin) place names are unlikely to be located solely by their etymology, locations based on etymologies will be considered much less reliable than phonetic similarities with modern places.
I have removed all reference to "Celt" and changed them to "Iron-age" or another appropriate term and will not include this term unless referring explicitly to the people the Romans knew as Celts (a subset of the the people in Gaul) or (as here) referring to the Celtic myth.
I have removed all references to a “Celtic” language and replaced it with meaningful terms such as “Welsh”, “Gaelic” or similar.
Likewise, given the very high probability of false etymologies when using "Indo-European", I intend to remove all references to "Indo-European" and all supposed etymologies based on "Indo-European" and replace them with specific words from specific languages.
Unless strong evidence is available to the contrary, linguistic “rules” will be considered to be statements of the likely "closeness" of words. That is the similarity in particular phonetics between two words from different sources.
I will not entertain “arguments from authority” on supposed etymologies. A single source with good evidence will be considered superior to agreement by all the authorities in the world, if they cannot back their views by evidence. I will only consider etymological arguments based on specific words & word forms from recorded languages.