I am not per se ruling out the existence of a non-Indo-European substrate (though I think that the vocabulary is much smaller than past authors thought), but such vocabulary would have to have entered before Grimm's Law came into effect (see below).
Oh, I have read that. But you were making the impression to me that you were just focusing on the occurence of any *p that you could find without any care of where and how it comes about.
Now let's bring some order into this:
first, cases of *kʷ > *p:
I never said it was originally PIE. If you read my previous post, I explicitly grouped it with *kʷ > *p.
compare Irish "ceithre", Welsh "pedair", Latin "quattuor", Lithuanian "keturi"
I never doubted that PIE *kʷ > *p in both Lusitanian and Gaulish. But again, the situation is not homogenous in Iberia. Because, we have cases where PIE *kʷ is retained in Hispania:
- ethnic name "Querquerni"
- ethnic name "Equaesi"
- Celtiberian "nekue" and "-kue" (occurs multiple times in Botorrita I)
It should also be pointed out that both Celtic and Italic share the assimilation of *p > *kʷ before *kʷ (e.g. Latin "quercus" vs. Lithuanian "perkuna", Latin "quinque" vs. Lithuanian "penki"). Without this common innovation, we would expect Gaulish *etro- and Welsh *edair.
...and now, cases of PIE *p > *p:
Old Irish "orc", British "Orca" (Diodorus 5.21), "Orcas", "Orcades" (Ptolemy 2.2). Also note that Latin has "porcus", the above names might be plain and simply Latin. Would that be surprising? The Celtiberians are known to have possessed Greek personal names (Botorrita III "Antiokos").
Yes, I acknowledge the existence of the above forms, but are you aware of the consequences of that (loanwords, maybe?)? To go ahead and say "okay, so Celtic retained PIE *p and let's be done with it" is not the way to go. Because, at the same time, we have the following attested:
- Gaulish place name "Ar(e)morica" (not *Paremorica or *Paramorica)
- Gaulish ethnic name "Arverni" (not *Parverni)
- Gaulish personal name "Vercingetorix" (not *Upercingetorix, I'm sure Julius Caesar would have told us!)
- Lepontic "uerkalai" (TI 36.1)
- Lepontic "uvamokozis" (CO 48)
- Celtiberian "uerzoniti" (Botorrita I)
- Celtiberian "uerzaiokum" (Botorrita III)
- Celtiberian place name "Uxama"
- British "Uxella", "Uxellum" (Ptolemy 2.2)
- Gaulish place name "Uxellodunum" (Bello Gallico 8.32, 8.40)
- Gaulish "sextametos" (Old Irish "secht", compare Latin "septem")
I'd also like to add, again, from my previous:
- Celtiberian place name "Aregrada" / "Arekoratikos" ( < *pare-)
- Celtiberian place name "Clunia" / "Kolounioku" (< *klepn-)
... and then...
We have potentially the above (in your own words, it's plausible). So in my opinion, we still have the loss of *p attested in all branches of Celtic...
... but, let's ignore these inconsistencies for now:
If you say Proto-Celtic retained *p (and PIE *p was only lost at a much later point), you are creating an unsolvable dilemma. As you know, we have the devlopment of *kʷ > *p in Brittonic:
Welsh "pedair", "pump"
Breton "pevar, "pemp"
(cf. Irish "ceithre", "cúig")
if *p > Ø happened after *kʷ > *p, the above would be obviously impossible. We'd expect something like *edair and *um in Welsh, and *evar and *em in Breton which clearly differ from the above observed forms.
Gaulish, Old British, or whatever language, throughout their evolution, have no memory of their previous state. Hence when the *p was lost in the respective language, all instance in which a *p stemmed from a previous *kʷ, it should have been lost as well (because the language obviously has no memory wether a *p came from PIE *p or from *kʷ). Because of this, saying that Proto-Celtic retained *p creates only problems.
(I might remind here of the analogy with the Japonic languages: if the development was via the intermediate stages of *p > *φ > *h > Ø, we would infer that at an earliest, *kʷ > *p occured after *p > *φ. In this scenario, we may imagine that the development *φ > *h > Ø occured subsequently while the Proto-Celtic language was already in the process of fragmentation. This would explain Lepontic "uvamokozis", and possibly occurences of *h- in Gaulish.)
Also, because of the above, the development of *kʷ > *p in Britanno-Gallic must be independent from the similar development in Lusitanian (since Lusitanian retains PIE *p).
I do not know where this leaves Lepontic words like "pala", but the only possibility is that these words were introduced later as loanwords, after the loss of PIE *p. As an analogy, Old Irish has plenty of loanwords with *p (but of course, many of them are Latin).
I don't necessarily say that Euler is correct with his late date for Grimm's Law, but it is elegant in so far as that it also offers an explanation for the linguistic identity of the Cimbri (which, otherwise, must be argued to be Celtic). Both Euler's late model and the traditional (early iron age) model are compatible with the corpus of Celtic loanwords into Proto-Germanic.
If Grimm's Law occured at the very basis of Proto-Germanic (adoption of an Indo-European language by non-Indo-Europeans, why is it that Celtic *dūno- ("fort", "fortress") was borrowed into Germanic (English "town", Dutch "tuin", Icelandic "tún", German "Zaun") and via Germanic mediation into Slavic (cf. Ukrainian "тин")?
I never, ever stated that I used his work exclusively.
As Untermannn wrote already back in 1987 "Ich fürchte, eines Tages werden die Keltisten lernen müssen, mit dem p zu leben" ("I fear, one day the Celti(ci)sts will have to learn to love with the p"): it is the desire to include Lusitanian (as well as the various "Celtic" forms in the NW of Iberia that retain *p) under the umbrella of the Celtic languages. But as I mentioned, that is effectively an arbitrary decision, since it still requires that *p > Ø (with the intermediate steps) holds true as a unifying feature of all Celtic languages besides Lusitanian.
(at this point, you might also want to check out this
list of Proto-Celtic roots by the University of Wales, which also is based on the assumption that *p was lost, via the intermediate step of *φ as I described)
Okay, I have to confess that I haven't read Yocum's work yet, but I can read it, get back to this, and give you my opinion of it at a later point.
- Notwithstanding the above that I do not know yet, I have personally been unconvinced by Koch's assessment that Tartessian is a Celtic language (I've describe the reasons before, you can also find most of them in
Zeidler's criticism of Koch's work). I concede that Tartessian looks at a cursory glance to be "Indo-European-ish", but I think that until we find a bilingual inscription I say this is purely hypothetical. Then there's the fact that Celtic place names are suspiciously scarce in what archaeologists deem the Tartessian core zone in Andalusia. For cross reference, you can compare this with
Alexander Falileyev's map.
- You must be confusing something here: the "700 or 500 years before Lepontic" statement is clearly false. The earliest Lepontic inscriptions are dated to around 500 BC, which would make the Tartessian stelae date to 1000 to 1200 BC. If you consider all the consequences, that would be a immense sensation in itself, but it is not true: the dates that Koch (2009, 2010, ff.) actually gives for the Tartessian stelae is ca. 700-500 BC. This would make the stelae either roughly contemporary to earliest Lepontic, or a few centuries older. A lot people talk erroneously about the supposed "ancientness" of the Celtic evidence on the Iberian peninsula (which in some aspects is certainly true), but for the greater part this is plain and simply false: the main corpus of Celtiberian is effectively contemporary to the corpus of Gaulish (in fact, the earliest attestations of Gaulish in northern Italy are actually older than Celtiberian). Any evidence for Lusitanian or the "Para-Celtic" elements in western Hispania dates
exclusively from the Roman period. If we disregard the possible Tartessian corpus, the ancientness of the Celtic presence in Iberia is
merely indirectly infered from the circumstances, rather than directly attested. I agree that the conclusion is logical (see below), but is only a conclusion.
- On the flip side, I do not mind Koch's proposal that the Atlantic Bronze Age was (partially) Celtic-speaking. This actually makes very much sense and solves a lot of problems that arise in the traditional models. What I cannot agree on is the conconclusions that many people (not necessarily him) draw from this. I would summarize my (current, anyways) opinion on this as follows:
- the genetic evidence of R1b (especially the combined pattern that emerges when combining the phylogenetic tree of R1b with the distribution patterns of it's R1b subclades) is not particularly well-compatible with the supposed origin of the Beaker-Bell Culture on the Iberian peninsula.
- I always struck me as vastly stretching my suspension of disbelief that the Celtic languages are somehow supposed to have derictly arrived in SW-Iberia from the Pontic steppe (via a land route or maritime route is irrelevant here). Also, where does this leave the Italic languages?
- the spreading pattern of R1b's subclades is more compatible with a central point of origin in southeastern France (from where it spread to the whole of western and central europe). I find it more conceivable to assume that R1b arrived in SE France via either a Central European land route or a maritime route, than to assume that it arrived "out of the blue" from the Pontic steppe in Iberia.
- Given the ancientness and vast extend of the Beaker-Bell Culture into areas that were never Celtic, I find it doubtful that the Beakers spoke Proto-Celtic. Instead, it's much more sensible to assume that the Beaker-Bell people spoke a more undifferentiated western (Centum) Indo-European dialect. Even then, I seriously doubt that the Beaker-Bell Culture was uniformously Indo-European.
- I think that the "Para-Celtic" elements may indeed be an older, autochthonous development that arose independently in the west of the Iberian peninsula (I believe this to be fairly consistent with the archaeological evidence, however).
- I would (very tentatively) place the actual development of a Proto-Celtic into northwestern Europe, into the very broad trade networks that interacted between the Wessex, Armorique and Tumulus Cultures. I know that this sounds vague, but it is most compatible with the available evidence.
- As a result, I see the Celtic languages as being introduced from the north (from the British Isles/Armorica) across the Bay of Biscaya via the bronze age trade networks, rather than the other way around. This is also much more consistent with genetic evidence (in particular the distribution of
R1b-L21 and it's subclades - which is not particularly compatible with a "Mil Espáine"-type scenario that the west-iberian-origin essentially requires). Note, again, if you take a look at Falileyev's map, that the distribution of Celtic names on the Iberian peninsula is much more compatible with such an introduction from the north via sea.
- The subsequent spreads and movements of Urnfield, Hallstatt and La-Tene explain the later spread of P-Celtic languages in Central Europe, Gaul and Britain (Koch himself also conceives this possibility, he describes the movement of these cultures as "a matter of intra-celtic dialectology").
Now, I concede some of you may not like the hypothesis as I have layed it out above, but make of that whatever you want.