The example is a cognate of flee:


Descendants[edit] Old English: flēon English: flee Old Frisian: fliā Old Saxon: fliohan Old Dutch: *flion Middle Dutch: vlien Dutch: vlieden Old High German: fliohan German: fliehen Old Norse: flýja Icelandic: flýja Faroese: flýggja Norwegian: fly Swedish: fly Danish: fly

Gothic: 𐌸𐌻𐌹𐌿𐌷𐌰𐌽 (þliuhan)

It just seems so peculiar, and I wonder whether it is a rule of sound change or an exception.

PS: the PGmc root comes from PIE *pleuk-, enlargement of *pleu- (“flow”).

Here are four Gothic words with cognates of PIE etymons starting with *pl-

  • Do you happen to know how frequent initial /fl/ and /þl/ are in Gothic? If this is a one-off, it's clearly an error, and there are many occasions for error in Gothic. But if there are other cases of *pl > *fl > þl, things could get more complex.
    – jlawler
    Commented Jun 2, 2013 at 15:46
  • FWIW: Lehmann 1986 PGmc *fl- => Go thl- before h, hs, kw (Matzel 1962, Anlautendes thl- und fl- im Gotischen.)
    – Alex B.
    Commented Jun 2, 2013 at 19:18
  • @jlawler Hi! I find four Gothic words with such a phenomenon, please see jstor.org/stable/408893?seq=4
    – archenoo
    Commented Jun 3, 2013 at 6:30
  • So it looks like before roots containing velars /h, hs, kw/ the cluster lost labiality and became purely dental. That's a back gesture. Anticipating the velar? That depends on what happens to other velars. It's probably worth mentioning, too, that /f/ and /þ/ are awfully similar sounds. They both have the same upper articulator (apex of upper incisors), and only differ in the lower articulators (lower lip vs tongue apex), which are independently controllable. This makes such slips more likely as errors. Plus Gothic is known only from a religious translation and errors happen.
    – jlawler
    Commented Jun 3, 2013 at 13:58
  • 1
    To add to @jlawler's answer: changes the other way round (from /þ/ to /f/) are common - see various dialects of English. It may seem odd that if a change is common, it can occur the other way round, but that is clearly the case. /kw/ changed to /p/ at some time in (at least) Ancient Greek, p-Celtic and Romanian, but /p/ changed to /kw/ before a following /k/ in several words in Latin (quinque, quercus, coqui)
    – Colin Fine
    Commented Jun 4, 2013 at 14:54

2 Answers 2


This is an old problem within comparative Germanic linguistics. The main ideas out there have been:

(1) Only Gothic has both þl- and fl-. The other Germanic languages must therefore have undergone a sound change þl- > fl-.

(2) Gothic underwent a sound change fl- > þl-. The Gothic words with fl- are newer loanwords (or other ad hoc solutions).

(3) Gothic underwent a conditional sound change fl- > þl-, but it's not clear what the condition was.

It's not clear to me whether there is any consensus in the field here, to the extent that this topic has been sufficiently treated to allow a consensus to arise at all.

  • I think the third idea is the most currently accepted, after Lehmann 1986, although the condition involved (some sort of following velar consonant or cluster) is IMO questionable. Commented Oct 15, 2013 at 5:33

I've seen two competing theories about this (both of which are mentioned by other commentors) - that either fl > þl in Gothic or þl > fl in everything but Gothic, but I think I can expand on these a bit.

Most favor the former theory, particularly because at least the word þliuhan can be traced back to IE pleuk-, which would give a PGmc fleuh-. If this is the case, as archenoo mentioned before, the rule is fl > þl / ## ___ V {h(s),kw}

The second theory, however, is mentioned by Voyles in Early Germanic Grammar. He posits that there was no change in Gothic (and presumably pl > þl somehow in PGmc, though he doesn't mention it), and that there was a later NWGmc change of þl > fl, which he relates to the NWGmc change of medial -þl- > -hl-.

FWIW I personally favor the former theory, because it seems more natural and a little easier to explain. The jstor article linked in the original question favors the latter.

Ultimately, though, the difficulty is that there are only four attested words with þl- and, surprisingly, only four attested words with fl-: þlahsjan, þlaqus, þlaúhs, and þliuhan for the former, and flautjan, flauts, flōdus, and flōkan for the latter. There's also one mysterious occurrence of flahtō 'braid', which violates the rules established above, but since there is only one instance of it, I am tempted to attribute this to scribal error (whereas there are numerous consistent examples of þliuhan, and at least a few of flōdus). It could also be a later borrowing, I suppose.

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.