25 Sep 2010

Adapting the rule of Cyprian Syncope

Recap: What is Cyprian Syncope?

Cyprian Syncope is a sound rule that I noticed on my own several years ago when first pondering on the language origins of Etruscan. Having recognized like many others that Minoan must fall under a Proto-Aegean language family, distinct from Indo-European or Semitic, I then reasoned that Etruscan phonotactics must have been simpler in its more recent past, aligning more with the much-stricter phonotactics of Minoan which only appears to have allowed syllables of a (C)V(C)-shape.

Cleaving Proto-Aegean into two branches, Minoan and Cyprian, I noticed that some Minoan vowels were being deleted in later Cyprian tongues due to some sort of very early stress accent, sometimes creating new word-initial consonant clusters that couldn't have been possible in Minoan. Etruscan, Lemnian and Rhaetic all have word-initial consonant clusters, showing that if they were created from vowel deletion, this must have occurred when they were once a single idiom back around 1000 BCE (ie. when these languages first arrived in Italy). This rule of syncope is unrelated to a later second syncope in Etruscan which has already been widely remarked by past Etruscanists and which took place around 500 BCE. As far as I've read, no Etruscanist has published a word on this first Syncope that I'm exploring openly here, as I have in the past online.

A slight change

This past week, reviewing my research, a new corollary on Cyprian Syncope came to me. Vowel deletion isn't always guaranteed, it seems, and I've been striving to understand why. Certainly I long ago saw this in derivational suffixes of a CV shape, eg. Proto-Aegean *-na [pertinentive] becomes both Minoan and Etruscan -na without vowel deletion. I also noticed later that a word-final structure of -CCV within a word also blocks vowel deletion. Thus the original structure of Proto-Aegean *tʰaura 'bull' (> Greek ταῦρος) is likewise preserved in Etruscan θaura. Recently though, I've been grappling with other notorious wanderworts like 'apple' and 'bee' in Western Europe, seeking Aegean solutions to these riddles, only to find that there is a new implication that some trisyllabic words with initial accent fail to delete the word-final vowel.

Without going into details about reconstructions I haven't yet detailed on this blog, I think I've arrived at a very phonetically plausible revision of the general vowel deletion rule by noting a preceding accent shift in specific cases. Thus:
1. Euphonic Accent Shift: Word-initial *CəCV́- where both consonants (C) are plosives attracts stress to the first syllable: *CəCV́-*CV́Cə-

2. The Cyprian Syncope Rule: Any vowel in a syllable immediately preceding or following a stressed syllable is deleted.
The reason for the initial accent shift prevents consonant clusters like those perfectly valid in Greek (eg. κτεατίζω 'to gain' or χθών 'earth') from ever forming in Cyprian, thereby explaining why they are completely absent in Etruscan despite having several Greek loans.

The following table shows the regular patterns in correspondence I witness that are emerging from the attested and substratal data and will hopefully illustrate how the above rules can explain them:

Proto-AegeanCyprian
(before Syncope)
Cyprian
(after Syncope)
*aléli 'lily'*əlélə*lel
*ápia 'bee'*ápiə*ápi
*apísa 'pear'*əpísə*pis
*árapo 'sprout'*árəpu*árpu
*talóza 'sea'*təlúzə*tlus
*ṭapúri 'village'*zəpúrə*spur
*ṭínau 'moulded'*zínəu*zinu
*tʰáura 'bull'*tʰáura*tʰáura

18 comments:

  1. Just for clarification: *ṭ = [θ], right? What does *z represent, a voiced alveolar fricative, or [ts]?

    Am I right, then, in assuming the following phonemes for Proto-Aegean:

    p t k
    pʰ tʰ kʰ
    - tθ (/z/)
    - θ s (/z/) h
    m n
    - l
    - r
    j w

    i u
    e -
    a

    Does that sound about right? Did I leave anything out?

    ReplyDelete
  2. No, I think *ṭ was a pharyngeal stop /tˤ/ like in neighbouring Semitic languages. So in Cyprian, *ṭ spirantizes and depharyngealizes to *z (ie. /tˤ/ > /tsˤ/ > /tʃ/) while becoming d /tθ/ in Minoan. I see only one sibilant in Proto-Aegean, *s.

    The stop system in Proto-Aegean is then:

    *p *t *k
    *pʰ *tʰ *kʰ
    - *tˤ *kˤ

    ReplyDelete
  3. I like your theory of "early Aegean syncope". Not just because it can explain the evolution of words you cited, but even more! Remember, that in Beekes' collection of "Pre-Greek" words, there was evidence presented for some really strange initial consonantal clusters, such as *pkt- or *kdn-. Now these can be easily explained, if they were borowed into Greek after a syncope.

    For example (that needs to be refined further), the later Greek words for 'night': knephas / dnophos / zophos / psephas could all be derived from a (collapsed) stem like *tknupa or *tnkupa (somewhat - and perhaps superficially - similar to the Hittite dankui- = 'night').

    ReplyDelete
  4. Shouldn't that rather be *apia > *apa according to the rules?

    Also are you reconstructing accent just on basis of the retain'd Etruscan vowel? If so, I don't see any reason why (some) words of the shape /TaT/ should be presumed to come from *TaTá- with any shift in the accent. For the lack of TT- onsets in Etruscan, an equally possible reason could be that these were secondarily simplified or lenited (as seems to be happening in *ṭapúri anyway).

    ReplyDelete
  5. Bayndor, I strongly doubt any triconsonantal clusters developed in Cyprian languages. In Etruscan, they're very rare. Some examples are streteθ (LL 6.iii) and Stramenas (ET Vs 1.40) but these are surely Italic loans since Indo-European languages have nothing against an onset like str-.

    I theorize that the accent in Aegean was placed only ever on the first or second syllable, making word-initial triconsonantal clusters via this syncope impossible.

    ReplyDelete
  6. Tropylium: "Shouldn't that rather be *apia > *apa according to the rules?"

    Yes I admit that I need to reevaluate some of my reconstructions to solve these new paradoxes. If we say that Aegean *-Cia becomes Cyprian *-Ci, then *amária 'face' which I mentioned many blogposts ago would need to be reworked to avoid incorrect **amári.

    Let's then say *amáira so that diphthong *ai, forming a closed syllable, blocks the omission of the final vowel. This diphthong is already known to become e(i) in later Etruscan anyway and would still explain the reflexes I identify.

    "[...] an equally possible reason could be that these were secondarily simplified or lenited (as seems to be happening in *ṭapúri anyway)."

    Lenition alone isn't a very plausible solution overall and it looks to me that PAeg *ṭ becomes Cyprian *z no matter what. It wouldn't explain why, say, *kt- or *pk is avoided after syncope.

    This is why I think that an accent shift towards such "overly heavy" syllables would easily and naturally explain the absence in Etruscan, Lemnian and Rhaetic.

    ReplyDelete
  7. Lenition alone isn't a very plausible solution overall and it looks to me that PAeg *ṭ becomes Cyprian *z no matter what. It wouldn't explain why, say, *kt- or *pk is avoided after syncope.

    By "lenition" I'm not pointing at the development from *ṭ to *z, but the deaffrication *zp > *sp. This can also be consider'd loss of the occlusion, which in the case of a non-affricate would be equal to complete loss.

    I'm sure you've noticed onset clusters of two stops are quite rare in the world's languages - even modern Greek has πτ κτ > φt χτ. For loss we have the example of English treatment of Greek loans. Do you have any parallels for stops attracting stress?

    Of course what we'd really need to settle this is either some independant verification for PAe stress, or words where stop cluster simplification could actually be observed.

    ReplyDelete
  8. Tropylium: "By 'lenition' I'm not pointing at the development from *ṭ to *z, but the deaffrication *zp > *sp"

    Despite your imprecision, what language has undergone syncope combined with word-initial lenition in order to avoid word-initial stop-stop clusters, as you propose?

    "I'm sure you've noticed onset clusters of two stops are quite rare in the world's languages [...]"

    Not rare enough for your argument to have weight:

    * Polish ptak 'bird'
    * Russian кто 'who?'
    * PIE *dʰǵʰōm 'earth'
    * Ancient Greek χθών kʰtʰōn 'earth'
    * Georgian თქვენ tkven 'you'
    * Old Tibetan bdun 'seven'
    * Klallam pqʷaʔčáyəs 'brown'

    ReplyDelete
  9. what language has undergone syncope combined with word-initial lenition in order to avoid word-initial stop-stop clusters, as you propose?

    Think about your own list of examples for a moment and you'll have a few. Several IE languages show loss of the first stop in the onset *dʰgʰ- (*dʰǵʰōmos > Latin humus, Lithuanian žeme etc.) Or possibly the second (depending on how this dates wrt. the metathesis TK > KT), but that would explain a systemic lack of two-stop onset clusters just as well. (Again, actual examples would be needed here.)

    Modern Tibetan is also rife with this: "7" is [dʰun], for example. (Whether this results from syncope is irrelevant.)

    Also Etruscan, according to your proposal. Again, (1) affricates are a type of stops; (2) deaffrication is a type of lenition.

    Now your turn, what language has undergone a shift of stress to a vowel between two stops?

    Not rare enough for your argument to have weight:

    My argument isn't that lenition is necessarily the preferrable solution; it's that you haven't shown sufficient evidence for the conclusion that "Euphonic Accent Shift" is the preferrable solution.

    ReplyDelete
  10. Tropylium: "[...] you haven't shown sufficient evidence for the conclusion that 'Euphonic Accent Shift' is the preferrable solution."

    I call spades 'spades' when necessary. Your consistent dogmatic skepticism (ie. skepticism unmotivated by facts) is, let's face it, flamebaiting and it's unconstructive.

    Occam's Razor and Onus are enough to reject a priori your baseless, ad hoc pile of *interdependent* sound changes which none of the prior examples discussed have validated. Your theory is comparatively more complex and assumptive than mine which makes your theory beneath further debate until firm facts validate your views and invalidate my own.

    Conveniently you also don't clearly specify what proof is enough... more opinionative flamebaiting.

    Finding a 'preferable solution' to your already invalid pet theory is merely topic hijacking. All I can tell you is that the dialectal change of *dʰǵʰ to *ǵʰ is deletion, not affrication, and ergo irrelevant as always to what we're really discussing.

    ReplyDelete
  11. "Now your turn, what language has undergone a shift of stress to a vowel between two stops?"

    Perhaps I owe you an answer to this question at least and I'll seek direct examples of this.

    On the other hand, it's just as relevant to ask why you even feel it's important to question a suggestion so bland? Again, it wreaks of a toxic level of skepticism that wastes time.

    ReplyDelete
  12. Even the blandest suggestions, bland as they may be, have merit. The fact very fact that they seem bland suggests, ironically, that they may be of the greatest importance.
    Consider this - WHY are they bland? Are they uninteresting, or seemingly obvious? If the former, then chances are that they are under-researched. If the latter, then why are they obvious?
    (And not to be a spelling NAZI, but it should be "reeks of", not "wreaks of". Just saying. ^.~)

    ReplyDelete
  13. Axiom: Rational debate demands *concrete reasons* for any of our otherwise meaningless beliefs or disbeliefs, out of respect for others and self.

    Seadog's snarky spelling corrections and Tropylium's aimless denials of everything, much like a bratty kid who says "Why?" to everything, exhibit the kind of "intellectual blandness" that tempts permanent ban.

    Back to topic, phonotactically-motivated accent shift is found in Tiberian Hebrew and syncope is far too commonplace to be skeptical of.

    ReplyDelete
  14. I was afraid my comment would be seen as snarky. It wasn't meant to be so. As pathetic as it sounds, I tried to ameliorate my comment by adding "^.~", the "innocent wink" pictograph.
    Also, I put NAZI in caps because it's an acronym - I wanted to be proper, especially with a baggage-loaded term.
    As well, I have this thing when the wrong word is used, to try and correct it. I mean, I don't bother commenting on the "its-it's" or the "to-too-two" and such (that'd take a lifetime, and would accomplish nothing), and I admit I make typos too, but some rarely-used homonyms just cry out to me for correction.
    Add to that, that I have enjoyed the word "reek" from the day I first heard it in the Rankin & Bass version of "The Hobbit", in describing Smaug's lair.

    Back on topic, however - these "reconstructions I haven't yet detailed on this blog" have my curiosity chomping at the bit. Any chance you could give a sneak preview of one or two of 'em?

    ReplyDelete
  15. Yes, back to topic. When I mentioned in the post above of "reconstructions I haven't yet detailed on this blog", I meant to say that I haven't yet explained the evidence behind some reconstructions in the table above.

    I haven't, for example, explained the evidence behind Aegean *apísa 'pear', *talóza 'sea' or *ṭapúri 'village', although it's probably apparent already to Greek speakers.

    ReplyDelete
  16. Have you explained somewhere where Proto-Aegean *talóza is related to Greek θάλασσα (thálassa)?

    ReplyDelete
  17. Stephen,

    No, I don't believe so but this is indeed part of the evidence for this etymon. A variant stem δαλάγχα- exists as well. Note also tluscv., an abbreviation for the god Tluscva 'Depths'), written on the Piacenza Liver.

    And while we're at it...

    For *ṭapúri: Note Greek word πόλις 'city' with its curious variant πτόλις with metathesized onset. Perhaps also Linear A du-pu₂-re for Minoan *θapúre and declined in the locative case in *-e.

    For *apísa: Compare Greek ἄπιον 'pear' and ἄπιος 'pear tree', as well as Latin pirus 'pear tree'.

    ReplyDelete