Thursday 20 October 2011

initial clusters

One of my favourite examples of metathesis is the Modern Greek verb βγάλω vghálo ˈvɣalo ‘I take out’ (aorist form).

To see how it came about we first have to dispose of one or two other sound changes en route from classical to modern. In Ancient Greek this stem took the form ἐκβαλ- ekbal- eɡˈbal-. The voicing assimilation of the consonant in the prefix ἐκ- ek-, making it voiced before a voiced consonant, appears to date from ancient times — see W. Sidney Allen’s Vox Graeca (CUP 1987) p. 18.

Ancient short unstressed vowels at the beginning of a word are lost (‘aphesis’) in Modern Greek. So for example the classical word ὄμμα ómma ‘eye’, or rather its diminutive ὀμμάτιον ommátion, stripped of its case ending -ον -on, loses its initial vowel to become Modern Greek μάτι máti ˈmati, still meaning ‘eye’. Classical ἐξερῶ ekserô ‘I will speak out’ (?) yields Modern aphetic ξερώ ξέρω kséro ˈksero ‘I know’.

Classical voiced plosives became fricatives in Modern Greek (‘spirantization’). Loss of the initial vowel in ἐκβαλ- ekbal- eɡˈbal-, the example we started with, left an initial cluster ɡb-. This duly became ɣv-. It was this cluster that then underwent metathesis to give the modern vɣ-. I do not know when the metathesis happened in popular speech. It was resisted in the katharevousa (puristic) form of modern Greek.

No parallel metathesis seems to have happened to γδ- from classical ἐκδ-. Homer’s ἐκδύνω ekdúnō ‘I undress’ (as in modern English zoological ecdysis and fanciful ecdysiast) yields Modern Greek γδύνω ghdhíno ˈɣðino with the same meaning and unmetathesized.

I used to find Modern Greek useful for widening my students’ appreciation of the phonotactic possibilities of language, and there was usually a native speaker conveniently to hand. Clusters such as word-initial are not difficult, once you have mastered ɣ, but can seem very strange to speakers of other languages — all except those familiar with French, where initial vr is found in such everyday words as vrai ‘true’. From you just need to move the uvular articulation forward to the velar position.

Greek has other interesting word-initial clusters involving a fricative + obstruent: φτάνω ftáno ˈftano or φθάνω ftháno ˈfθano ‘I arrive’, βδομάδα vdhomádha vðoˈmaða ‘week’, χτές khtes xtes or χθές khthes xθes ‘yesterday’, χτυπώ khtypó xtiˈpo ‘I knock’, σχολείο skholeío sxoˈlio ‘school’, σγουρός sghourós zɣuˈros ‘curly’.

For really complex consonant clusters, however, you need to go to the Caucasian or Salishan languages. When I was teaching my phonological analysis class we never seemed to have any native speakers of those languages around.

14 comments:

  1. Slavic languages have some complex onsets, too; Polish in particular... ;)

    ReplyDelete
  2. True. Don't start about chrząszcz brzmi w Szczebrzeszynie.

    ReplyDelete
  3. I find it interesting how this has obscured parallel words, especially in the pair μπαίνω |mpaíno| béno ˈbeno "go in, enter" - βγαίνω |bgaíno| vyéno ˈvʝeno "go out, leave".

    In the original, they were clearly parallel (ἐμβαίνω < ἐν- "in" + βαίνω "go", with assimilation of POA; ἐκβαίνω < ἐκ- "out" + βαίνω "go"), but subsequent aphaeresis and metathesis in only one of the two has obscured the relationship.

    Incidentally, modern Greek "to know" is stressed on the first syllable: ξέρω kséro ˈksero. And I always thought its etymology was from Ancient ἐξεύρω ekseúrō, the aorist of ἐξευρίσκω ekseurískō "find out, discover".

    ReplyDelete
  4. Are you positiv about the εξερώ thing? I looked it up in the two major greek dictionaries and both give the εξεύρω etymology.

    "No parallel metathesis seems to have happened to γδ- from classical ἐκδ-."

    A main reason should be that δγ- is ungrammatical in greek.

    ReplyDelete
  5. τκ- is also phonotactically impossible - as are τπ- and δβ-, though the reverse pairs, πτ- and βδ- are allowed.

    On the third hand, none of πκ- κπ- γβ- are allowed... even though βγ- is, with both pronunciations of gamma! (At least in Modern Greek.)

    ReplyDelete
  6. Philip and Ἀθανάσιο??

    How does the εξεύρω etymology account for its not being ξεύρω [ˈksevro] in Modern?

    ReplyDelete
  7. Sorry about ξέρω error: now corrected. On its etymology, you may well be right, but as mallamb says there is then the question of what happened to the /v/. It remains in εὕρηκα → ˈvrika 'I have found'.

    ReplyDelete
  8. Well, I'm no expert, I'm just copying what I read in the Babiniotis Dictionary: ξεύρω became ξέρω due to analogy with verbs like φέρω or χαίρω and because the type ξεύρω sounded archaic (not much, if you ask me). The Triantafyllidis Institute Dictionary gives another, rather awkward explanation: perhaps the v sound was rejected in phrases with consecutive v sounds, like ξεύρεις βρε "well, you know".

    Seems like they're stuck with the least bad of the possible etymologies.

    ReplyDelete
  9. I thought there might be some such Sprachgefühl, but analogy with verbs like φέρω or χαίρω is pretty convincing, isn't it? And it might be only by comparison with ξέρω that ξεύρω seems archaizing! A bit circular, what? Anyone feel uncomfortable with εύρω, apart from considerations of the Euro crisis? Has the etymology left any other consciousness-raising traces such as ξεὕρηκα?

    ReplyDelete
  10. There is the odd fact that the past is not έξερα but ήξερα. The eta reminds one of the one in εὕρηκα but it's on the wrong side of the xi.

    On the other hand, besides the regular βρήκα I believe I've also seen a form ήβρα, which also has the eta at the beginning.

    I don't have an explanation for those forms, though.

    (Another verb which has an eta at the beginning in the aorist is θέλω - ήθελα, but I believe this has its roots in Ancient Greek already: if I remember correctly, the verb was εθέλω and so the augment was eta.)

    ReplyDelete
  11. It rather looks as if you do have an explanation for those forms. What you have pointed out with θέλω - ήθελα reflects the case of augment ἐ + initial ἐ giving ἠ in spite of aphesis in other forms of the verb even in Classical Greek, where θέλω was already established alongside ἐθέλω, but the only perfect was ἠθέληκα, preserving the aphesis-prone ἐ. Even when it disappears completely in Modern Greek, and there is no *εξέρω, the η persists in ήξερα, preserving the ghostly ε- in ξέρω and it looks to me as if it's even on the right side of the xi! It also looks as if it outlived the υ lost by analogy with φέρω or χαίρω, which is a bit worrying. Perhaps ἐξερῶ ekserô ‘I will speak out’ might after all have been the etymology.

    Of course where they have to respell the aphetic form, as in βρήκα, there's no longer even the possibility of an archaizing ηύρα.

    I take it ἐξεύρηκα has not in fact survived in any form. (Always supposing you found ξεὕρηκα legible – that's what comes of pasting bits with illegible jots and tittles, but we all seem to have been playing fast and loose with the breathings. I have tried to keep them for classical forms and lose them for modern, whether real or imaginary!)

    ReplyDelete
  12. Speaking of funky consonant clusters, I'd like to throw in http://strcstskrzkrk.com or 'scvrnkls' - an actual Czech word - although they may not be that 'remarkable' phonotactically.

    ReplyDelete
  13. Clusters such as word-initial vɣ are not difficult, once you have mastered ɣ, but can seem very strange to speakers of other languages — all except those familiar with French, where initial vr vʁ is found in such everyday words as vrai ‘true’.

    I was totally like "WTF?" when I first saw the entry [ˈjɾʲamˠə] (Irish ghreamaigh ‘stuck’) in http://en.wikipedia.org/wiki/Palatal_approximant, but then again, Spanish gran is [ɰɾan] post-vocalically, isn't it?

    ReplyDelete
  14. Angr pointed out that Irish /j/ is usually a fricative before consonants (and replaced the example on Wikipedia) -- still, my point about being surprised by that thing even though I knew a very similar thing in Spanish still stands. (Some say that Spanish /g/ is a fricative too, but with some speakers I can barely hear the /g/ in negra at all, let alone hear friction.)

    ReplyDelete

Note: only a member of this blog may post a comment.