Empires of the Word - Nicholas Ostler [162]
While Lepontic, Gaulish and Brythonic (P-Celtic) all usually convert old kw to p, Celtiberian and Goidelic (Q-Celtic) retain the k element. It would be possible, then, to see Q-Celtic as the original form, spread to the shores of Gaul by effective users of iron, and then, through the establishment of exchange relationships and trade, beyond to the south and north across the sea. Subsequently, the Celts in Gaul and the Alps innovated in converting kw to p, followed by their close associates in Britain, while the peripheral ones, Celtiberian and Goidelic, retained the kw, those in the north, Iernē, later simplifying it to k. *
* In fact, few linguists today take this P/Q criterion as a very strong discriminant. The change could happen anywhere: indeed it has, in modern Romanian, and quite independently in the Italic dialects (e.g. Oscan changed to P, Latin didn’t). And even in the centre of the P-dialects, not all Qs changed to P: on the Coligny calendar in the Rhône valley we find EQVOS, EQVI, ‘horse’ (even though the usual Gaulish name for the horse goddess is Epona), and the ‘Sequani’, living on the river ‘Sequana’ (Seine) in northern Gaul, seem unaffected. But P-Celtic and Q-Celtic are such a chestnut in the tradition that it seems deceptive to leave it out of the discussion.
In fact, some strange changes came over Celtic in the British Isles, as nowhere else: verb-subject-object as basic word order, mutation of initial consonants, conjugated prepositions, strange locutions to express status and activity (’I am in my student’, ‘I am at reading of my book’), and much else. There are those who believe that these strangenesses are really inherited from the lost previous languages of the old inhabitants, perhaps spoken by the civilisations that raised megalithic monuments. Failing to learn the incoming language fully, they simply continued with many features of their old languages. This is the substrate hypothesis; interesting, but it explains little since we know nothing of the languages of the British Isles prior to Celtic.
Another hypothesis is language mixing, or creolisation. It too can be brought into the theory of Celtic spread by navigation along the Atlantic coast, by noting that major partners in this network, for most of the first millennium BC, were the Phoenicians, many of them (specifically the Carthaginians) based in North Africa, and quite capable of maintaining links along the whole Mediterranean. Now it so happens that in the North African language families, Egyptian, Semitic and Berber, there are direct parallels for at least seventeen of these curious characteristics of British and Irish Celtic, characteristics that are quite unparalleled in any Indo-European language, let alone their Celtic cousins, and which are indeed extremely rare globally.30 If Celtic was indeed spread as a coastal lingua franca, these North Africans, in trade and exchange, would have been among its speakers, and effective in moulding it.
But there is no direct linguistic evidence for any of this at the moment: as to the spread of Gaulish across most of Europe, and the origins of Celtiberian, and the Celtic languages of the British Isles, we are in the realm of speculation and reconstruction. By contrast, we have direct testimony on the advent of Celtic speakers in Italy and the eastern Mediterranean.
The Gauls’ advances in the historic record
It is clear that the ideal of the raid, whereby parties of young men would seek to cover themselves with glory and booty, never ceased to be current in Celtic societies that remained independent. Successful raids, especially if perpetrated by younger sons without prospects at home, could turn into de facto invasions. And we also encounter examples of deliberate decisions by Celtic tribes to seek new land in a mass migration: