Irish clann "plant; offspring; child" (the source of English clan) is borrowed from Welsh plant with the same meanings, which is itself a borrowing of Latin planta. Why did Irish change the initial p into c?

PIE p was lost in Celtic, so the explanation I've seen given is that Irish did not have initial p at the time of the borrowing. But this leads to other questions:

  1. There are in fact numerous Irish words beginning p-. These, as far as I can tell, are mostly loanwords. Why the p in these words but not in clann? Do the loanwords in p date from a later period, and if so when did Irish acquire a p and begin to retain p in loanwords?

  2. Why the specific change to c? Other than being voiceless stops, the two sounds are not particularly close; one might have expected the p to be lost completely, or maybe to be borrowed as b. A possible factor is that there are numerous cognate sets where Irish c corresponds to Welsh p (from PIE * kw), e.g. cethair : pedwar 'four'. If speakers were aware of these correspondences, this could explain the choice of p. Is there any evidence that this might have been the case?

Finally, are there other Irish loanwords from Welsh (or other languages) which show substitution of c for p?


1 Answer 1


You have to a large degree answered the question yourself, really: clann is indeed a very early loan word.

Like Common Celtic, Common Insular Celtic had no /p/, but it had /kʷ/.

The Brythonic branch changed /kʷ/ to /p/ (keeping the labiality, losing the velarity) quite early on, whereas in the Goidelic branch, /kʷ/ merged with /k/ (retaining the velarity, losing the labiality) somewhat later on.

In the earliest Ogham inscriptions, there are still separate letters for /k/ and /kʷ/, so we know that the change took place fairly close to the ‘proper’ attested stage of Early Irish. Sadly, we don't have any direct evidence that dates the Brythonic change, at least not that I know of. But unless I'm misremembering (and I don't have access to my works on Brythonic sound change at the moment), it can be positively shown to have predated certain other changes that definitely happened quite early on.

At the time when clann was borrowed, Brythonic /kʷ/ => /p/ had already taken place, but Goidelic /kʷ/ => /k/ had not. In other words, Brythonic *pland was borrowed with the closest equivalent available (the only unvoiced labial plosive in the language), as *kʷland, which later lost its labiality and became kland (orthographically since Old Irish cland, later on clann).

Once /kʷ/ was lost in Irish, however, there was no good, existing unvoiced plosive to use as a substitute for /p/, and it was around that time that they started using this hitherto foreign phoneme in loan words, reintroducing the long-lost /p/ back into the Irish language. This was still fairly early on, of course, predating the earliest ‘proper’ Old Irish sources (the Milano and Würzburg Glosses)—early enough, also, that this new /p/ became a natural part of the system of initial mutations on a par with /k/ and /t/, all three being subject to both lenition and nasalisation.

I am certain there are other words (either directly from Latin or through the medium of Brythonic) that show a similar route, but I cannot think of any off the top of my head at the moment.

Edit: Not true, there is one other well-known example I can think of: the Irish word for Easter, Cáisc, is derived via Latin pascha and Greek πάσχα ultimately from Hebrew (see etymonline for details). That shows the same development.

As Colin Fine mentions in the comments, another parallel example is corcair ‘purple dye, lichen’ and its derivative corcra ‘purple’. This word is perhaps not borrowed directly from Latin purpura (itself borrowed from Greek πορφύρα ‘purple (colour, dye, clothes, etc.)’), but it definitely has the same source and shows the same development of *p being borrowed as *kʷ > k.

  • Thanks, this is very informative! I guess the only part of the question that remains unanswered is whether the kw for p substitution is simply a case of replacement by the phonetically nearest phoneme or whether it was influenced by some metalinguistic awareness of the kw : p correspondence in Irish and Welsh, but that may be impossible to answer.
    – TKR
    Commented Sep 20, 2015 at 17:03
  • 1
    Sadly, I think that is impossible to answer, yes. I’m sure the speakers on both sides of the Irish Sea back then were aware that there were clear correspondences if they had any dealings with each other—the languages would have been close enough to each other still that they were more or less mutually intelligible, and a /kʷ : p/ correspondence should be an easy one to learn and internalise—but whether or not that played a part in how they naturalised loan words… you’d have to be there to get an inkling of that. Commented Sep 20, 2015 at 17:10
  • 2
    @user8017 There are different views, that is true. That particular one is not much of a counterargument (/kʷ/ → /p/ is far too common a development to be in any way conclusive; within Italic, it happened independently in Oscan, Umbrian, and Romanian, for example), but there are somewhat more persuasive arguments. Overall, though, the evidence for a Common Insular Celtic (especially the syntactic changes present in all the Insular Celtic languages and demonstrably absent from all non-Insular Celtic languages) are—to me—far more convincing than the arguments against it. Commented Sep 20, 2015 at 20:25
  • 3
    Some evidence that the British Celtic change * kW > p predated contact with Latin: Welsh rhelyw "remainder, remnant" may be from Latin reliquus, and Welsh cysgu "sleeping" may be from Lat. quiescere "to rest" (although I've also seen a proposal that cysgu comes from IE * kub-sk-, reflecting the same root as Latin cubo "I lie down"). Also, Breton koarell "shoe sole" may come from an unattested Latin form *quadrellum, as reflected in French carreau "pane, tile".
    – user8017
    Commented Sep 20, 2015 at 21:50
  • 1
    Irish corcra ("purple") is surely another example.
    – Colin Fine
    Commented Aug 9, 2023 at 15:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.