21 Nov 2010

Caper of the three kays

In Subtle truths about Etruscan letter-names, I explained why minimal pairs such as and among Latin letter-names were impossible in Etruscan because the language lacked these voicing contrasts. Surely then such pairs could only be distinguished in Etruscan by Semitic names, similar to those of Greek. There's yet another piece of proof.

Regular patterns emerge in the naming of Latin letters:
  • Vowels are named entirely by their phoneme (eg. ā, ē, ū, etc.)
  • Plosive letters methodically terminate in ē (eg. , , , etc.)
  • Letter-names of fricatives & affricates begin with short e (eg. ef, el, em, etc.).
  • Most recent Greek borrowings, hy and zēta, disobey the more ancient pattern.
Missing in the above are the outlier letters , and whose exact motivations are obscure.


Ecce cē, kā, qū que...

Note the 'three kays' of the Latin alphabet which represented the same sound /k/: , and . What's more, the Roman 'q' was restricted to positions before 'u'. This habit was borrowed from Etruscan which in its oldest stages chose 'k' before 'a', 'q' before 'u' and 'c' everywhere else. Many take these arcane rules for granted[1] but in attempting to solve this mystery, we should also be aware that these same spelling rules were even in effect among early Greeks who used the equivalents gamma, kappa and koppa respectively. The earliest Greeks likewise restricted koppa to positions before /o/ and /u/, kappa elsewhere.[2]

After much thought, I realized that the rule must be motivated by the very names of these letters. Look again at the Greek pair kappa and koppa. Since Semitic /q/ was an exotic sound to Greeks, the functionality of koppa (= Semitic qoph) was modified to convey non-aspirated /k/ just like kappa. To justify the usage of both however, the first syllable of the letter-name koppa must have inspired its eventual restriction to positions neighbouring back vowels in Greek before fading into oblivion.

Yet for Etruscan scribes, the merger of phonemic contrasts seen in the Semitic alphabets extended further since voiced /g/ was also a foreign sound to them. A confusing trifold representation of Etruscan /k/ by the three former Semitic letters (gimel, kaph and qoph) was the result. Like Greek but more extensive, a spelling rule seems likely to have been based rather trivially on the first syllable of each of these letter-names. This would help us reconstruct the native Etruscan letter-names for these 'kay' letters.


Refining the reconstruction of Etruscan letter-names

Using Greek and Semitic letter-names as guide, together with this odd spelling rule, I now find myself reasoning that Etruscans had originally called 'c', 'k' and 'q': *cimla[3], *capa and *cupa. In this way, the first syllables of each (ie. ci-, ca- and cu-) serve as fine basis for the attested spelling rules of Old Etruscan while simultaneously providing an elegant etymology for Latin , and . In other words, when the Romans simplified the Etruscan alphabet names, they simply clipped them down to their first syllable and used them doubly as spelling mnemonics.

This devilishly implies separate sources of the Etruscan and Greek alphabets since the required vocalism in *cimla is less like Greek gamma and more faithful to its Modern Hebrew counterpart, gimel.[4] This suggests that Etruscans didn't adopt their alphabet from the Greeks but instead gained the alphabet more directly somehow, directly from West Semites in Asia Minor rather than Euboea perhaps.[5]


NOTES
[1] Bonfante/Bonfante, The Etruscan language: An introduction (2002), p.75 (see link). Here, the spelling rule is mentioned but the authors leave out any further explanation.
[2] Woodard,Greek writing from Knossos to Homer: A linguistic interpretation of the origin of the Greek Alphabet and the Continuity of Ancient Greek Literacy (1996), p.161 (see link); Threatte, The Grammar of Attic Inscriptions: Phonology (1980), p.21 (see link).
[3] I had reconstructed *camla in previous entries but I reckon that the spelling-rule argument will justify *cimla in its stead.
[4] Hamilton, The origins of the West Semitic alphabet in Egyptian scripts (2006), pp.57 & 283 admits to variants for this Semitic letter name: *gaml-/*giml-. (see link).
[5] Bonfante, The Etruscan language: An introduction, 2nd ed. (2002), p.52: "On the other hand, the Etruscan alphabet also seems to preserve the traces of a very early Greek alphabet, older in part than the split between 'Western' and 'Eastern' Greek alphabets, since it preserves all three Phoenician sibilants, the signs samekh, sade, and Sin [...], which neither 'Western' nor 'Eastern' Greek alphabet possesses any longer [...]." (see link).

9 comments:

  1. It seems unclear to me why you reconstruct *cimla, or why you previously reconstructed *camla for gamma.

    Maybe the implicit reasoning behind *cimla is as follows:

    As we expect the vocalism to give an indication how the distribution of the spelling works, *ci is more likely that say, *ca. Greek has gamma, so the name can't have a greek origin, therefore it's Semitic as it does have an *i in the original letter name so it's *cimla.


    Is that what you were going for? That makes sense.

    ReplyDelete
  2. My reasoning starts right back to Herodotus who tells us that the Etruscans are from Asia Minor. If they're from Asia Minor, why should they borrow a Semitic alphabet from Euboea when they could borrow one in their own land where Semites flourished?

    It's also no secret that Greek double -mm- in gamma can only come from -ml- (see γάμμα). While this change is motivated in Greek, it's not in Etruscan which can not only tolerate word-medial -ml- but also word initial ml- (eg. mlaχ, Mlaχuχ, mler, mliθu, etc.).

    Since Euboean origin is suspect vis-à-vis Herodotus (and a mountain of other links between Etruria and Asia Minor), a Greek borrowing is unreasonable. If not from Greek, it's directly from a Semitic language. Yet if from the Semites, there's absolutely no phonotactic motivation to reduce -ml- as has occurred in Greek. Ergo, I expected nothing less than *camla.

    However, I realize now that this spelling rule motivates *cimla, which just so happens to be a valid alternative form for this letter-name. It seems that Semites hadn't collectively decided on whether to call it *giml- or *gaml- and so we see variants between regions.

    ReplyDelete
  3. when the Romans simplified the Etruscan alphabet names, they simply clipped them down to their first syllable

    But 'C' isn't *cī (nor for that matter, 'T' *tā). Generalization of the ē-system has occur'd here, so that means the Roman alphabet tells us nothing for if 'C' came from a name beginning with /ka/ or /ki/.

    ReplyDelete
  4. Tropyllium: "Generalization of the ē-system has occur'd here, so that means the Roman alphabet tells us nothing for if 'C' came from a name beginning with /ka/ or /ki/."

    The association of Latin & with Greek equivalents kappa & koppa is self-evident. So your pseudo-intellectual arguments are ignorant of the facts and address nothing. Nice try.

    ReplyDelete
  5. Tropylium also seems irritatingly unaware of the original Greek alphabet which sports ei, xei, pei, phei, chei and psei, and which were closer to the Semitic equivalents than modern Greek.

    If someone had thought through what I said, they'd see that I imply that the Etruscan names likewise ended in diphthong *ei, from which we get Latin's penchant for letters ending in . The implied reduction of Etruscan *tau and the first "kay", *cimla, to Latin and respectively follows the normal pattern of simplification already explained, a pattern that *replaces* the original values. Subsequent "kays" in the alphabet however preserve the original Etruscan vocalism to avoid a merger which evidently didn't happen and which demands the explanation I offer.

    So after all that, to expect Latin *cī or *tā out of this is disinterested nonsense.

    ReplyDelete
  6. The association of Latin kā & qū with Greek equivalents kappa & koppa is self-evident

    Yes, and I didn't deny that. I only said, as you also repeat, that and are extensions of the pattern of etc. (the origin of it is quite evident, than you very much) — and therefore cannot serve as evidence that the Etruscan name of the letter would have been *cimla and not *camla.

    ReplyDelete
  7. Tropylium: "[...] and therefore cē cannot serve as evidence that the Etruscan name of the letter would have been *cimla and not *camla."

    You're only absorbing a piece of my entire reasoning at a time, causing an unfortunate brain failure. Let's try again another way.

    If Etruscan was truly the original innovator of the Latin system, we certainly could theorize Etruscan *ce-*ca-*cu to explain Latin cē-kā-qū. But the Latin bē-pē contrast is impossible in Etruscan. Fail.

    If Etruscan took names directly from Greek, resultant *cama-*capa-*cupa leaves the un-Greek & purely Etruscan rule of placing its "gamma" before front high vowels, e and i, unexplained. Fail.

    Etruscan *cimla-*capa-*cupa, directly borrowed from Semitic languages in Asia Minor, finally explains the spelling rule while giving a rational source for Latin cē-kā-qū with optimal parsimony.

    (Now, if one wants to ignore parsimony, this is the wrong blog for them.)

    ReplyDelete
  8. If Etruscan took names directly from Greek, resultant *cama-*capa-*cupa leaves the un-Greek & purely Etruscan rule of placing its "gamma" before front high vowels, e and i, unexplained.

    Not quite. 'C' is not "the front vowel letter for /k/"; it is the default letter for /k/, also used preconsonantally and word-finally. We must consider the possibility that it is this "default" usage which causes it to appear before /i/ and /e/ (I would even say that as long as we adhere to parsimony, we must conclude this to be the case; to treat the distribution of 'C' as three independant rules is a multiplication of hypotheses.)

    As for where its default status comes from, for a first gess I would point to it being the alphabetically first of the three. I do not see how to parsimoniously derive the default status from an original association with /i/. We could suppose a system where 'C' is first extended to the environment /ke/ on account of /i/ and /e/ being a natural class (the front vowels), and then elsewhere on account that it can occur in multiple environments unlike 'K' and 'Q', but the cinch is - this requires assuming intermediate stages which AFAIK aren't attested. (If they are, I'm sure you'll tell me.)

    ReplyDelete
  9. Tropylium: "We must consider the possibility that it is this 'default' usage which causes it to appear before /i/ and /e/"

    If true, we'd expect Etruscan c to be as unrestricted in usage as its counterparts, Greek gamma and Semitic *gaml-/*giml- from the very beginning. We would NOT expect the three letters to be so well preserved into Latin with similar spelling restrictions.

    Please stop bickering.

    ReplyDelete