A NOTE ON THE THREEFOLD REPRESENTATION OF "K" IN ARCHAIC ETRUSCAN WRITING
Central Research Institute for Physics, H-1525 Bp. 114. Pf. 49., Budapest, Hungary
In Etruscan script the sound [k] seems to be triply represented, by C, K and Q; or there were different laryngeal stops in Old Etruscan. (In New Etruscan only C remains.) I discuss the question of [K] sounds by using formant/noise band data for modern K+Vowel connections asc well as the representations of K sounds in diverse alphabets.
The biggest language of Hungary, an Ugric idiom in no direct connection to the most languages of present Europe is called in English both Hungarian and Magyar, but more frequently the first. So here I take this convention, although historically it is misleading.
As it is well known, the Archaic Etruscan alphabet contained 3 letters for "k"-sound(s), transcribable as C, K and Q, respectively. In the Neo-Etruscan they tend to coincide in one, C. It is an interesting, and nontrivial, question if they were originally different phonemes, converging in evolution, or they represented the same phoneme even originally. There are analogons for both possibilities. First, note that Archaic Latin had the same 3 "k" letters, of which K gradually went out of use (except for calendar words as "Kalenda" or “Interkalaris”). Second, in recent English and French C (if not before a front vowel), K and Q represent the same phoneme, and the distinction is only traditional, inherited from orthographies of other languages, while, e.g., Polish RZ and Ż were different sounds several centuries ago and have converged into the latter, transcribed as “zh”, afterwards.
The answer is not easy, since the lexical problems with Etruscan make it unclear if two words of slightly different orthographies are identical or not for meaning. However physical speech acoustics may help, even if it is impossible to analyse Etruscan sounds.
We start from a valuable note of Pallotino (1975; Chap. XII). There footnote 5 tells that in the Archaic Etruscan period there probably was some slight difference among the pronunciations of letters c, k and q, as shown by their connections with vowels, since "c" is generally followed by "e" or "i", "k" by "a" and "q" by "u". (These 4 are the only Etruscan vowels.)
As we shall see, this observation is appropriate for a comparison with several other languages of rather various families. In some cases the acoustic properties of the "k-sounds" are known, and the data will suggest something for the pronunciation of the 3 "k-letters" of Archaic Etruscan too. With caution this suggestion can clarify the situation to an extent.
Before the methodical comparison I mention briefly something about Greek and Latin "q". They are the nearest kins of Etruscan writing.
According to common opinion, all dialects of Classical Greek lacked the [q] sound (Mycenian Greek had had it, but then it changed to [p]), but the alphabet borrowed from Phoenicia in cca. 800 BC had both the letters K, “kappa” and Q, "koppa", rather “qoppa”. Attic used Q only for the number ""90", but e.g. originally the name of the city Kolophon was written as "QOLOFON", as you can see the inscription of the Greek mercenaries of Psammetich II at Abu Simbel about 590 BC, where a Pabis of Kolophon (or the scribe) uses clearly koppa. But this koppa=Q does not seem to be the descendant of Mycenian Q.
In Latin (see later in details) Q remained in use only before [u]. But this is not true backwards: there are "cu-" syllables too, and look at the Pronomen relativum "qui" (NSingM), whose Genitive is "cuius" and whose dative is "cui".
Chapter 2 is an overview of the languages/scripts used for comparisons, Chapter 3 gives the speech acoustics data for some modern KV syllables, and Chap. 4 is the discussion.
2. ABOUT THE LANGUAGES FOR COMPARISON
As told above, we are going to compare the written representatives of "k-sounds" of various languages with those of Archaic and New Etruscan. We would like to emphasize that the inclusion of a language here does not indicate any genetic relation to Etruscan; indeed, no unequivocal genetic kin of Etruscan has yet been found while the list of candidates is long. (For recent developments see Alieni (2003).) The comparison is performed in order to find common acoustic feature; anyways, human speech channels vary less than languages do. Although the physical properties of sounds of speech can be language-dependent, still they are so in moderate extent. One expects that the acoustic properties in a CV connection, where C is an unvoiced k-stop, are qualitatively rather similar in many languages.
First let us note that many modern European languages use 3 letters for the [k] phoneme, C (before back vowels), K and Q (before U, but not exclusively, see French). However this may be simply an inheritance from Latin; for my knowledge no acoustic difference has been reported between syllables of C+back V and K+back V. In Latin originally C might have differed from Q; or might not. Classical Latin, as told above, used C, K and Q, K had gradually become exceptional.
Modern Hungarian uses only K. Q appears only in foreign names and C stands for the affricate [ts]. However this is a new development; in medieval Hungarian ortography generally C stood before back vowels and K before front ones (just oppositely as in archaic Etruscan); [c] was written by the digraphs TZ or CZ.
There was a non-Latin alphabetic script for Hungarian (still in some use), the Szekler "rhunic" script (see Appendix) of Old Turkish origin. Its orthography is almost homologous with the Latin alphabet of Hungarian, with two exceptions: i) digraphs of specific Hungarian phonemes are single letters there, and ii) there are two letters for [k], one when [k] is preceded or followed by a front vowel, one when it is so by a back one.
Modern Ottoman Turkish uses only K; C stands for an affricate. However Old Turkish (the language of the Orkhon scripts) used a peculiar alphabet not indicating directly the back/front opposition on vowels, but doing that on many consonants. Because of the so called vowel harmony common in Uralic and Altaic languages this special notation was viable. 9 consonants were doubly represented ([b], [d], [g], [l], [n], [r], [s], [t] and [y]), while there were five letters for [k] (Thomsen, 1896) according to the neighbouring vowel.
The Etruscan (and likewise Latin) alphabet comes from Old Semitic through some Western Greek one. So the multiple representation may have come from the Old Semitic. In Modern Arabic there are 2 "k" phonemes transcribed as K and Q, and by any chance it was likewise in progenitor languages too. Phoenician had two such letters; the ancestor of Etruscan C was used for the [g] phoneme. The two Semitic "k"-sounds are continued in the kappa/qoppa duality in old Greek scripts.
3. ON THE ACOUSTIC STRUCTURE OF "K" SOUNDS
Vowels have clear acoustic structures, moderately language-dependent. Generally, except possibly for some Far East languages, the main discriminating factors axre the first two formant frequencies called here F1 and F2, i.e. the frequencies at the first two peaks of the Fourier spectrum of the sound (see e.g. Peterson & Barney 1952 and citations therein). Higher formants belong to the individual timbre, and amplitude ratios determine not the vowel but the intonation.
The actual formant frequencies are very language-dependent. However it seems a general rule that F1 is lower for high vowels than for low ones; and F2 is lower for back vowels than for front ones. This is true for such a variety of languages as English (Peterson & Barney 1952), Hungarian (Tarnóczy 1966), Frisian (de Graaf 1982), Ottoman Turkish (Lotz, 1975) or Buryat Mongolian (Matheev 1984). We do not know the quality of Etruscan vowels, of course, but the Latin alphabet is a direct descendant of the Etruscan one, and practically in all Neo-Latin languages A is a back low vowel, E is front low or mid, U is back high and I is front high one, respectively. Therefore the original Latin formant frequencies can be approximately reconstructed and the Etruscan ones cannot have been too far because of the application of identical letters.
For consonants the situation is more difficult. For some ones, as e.g. liquidae, formants are recognisable, for others noise maxima are observed. For K sounds only the noise maxima, or bands, called here K1, K2, &c., are seen, the actual values will come later.
"Assimilation" effects are often seen. We mention two cases. First, in some languages there is a so called "vowel harmony", when in a word only one type of vowels appear. In today's Europe Finno-Ugric and Turkish languages represent this group: vowels harmonise according to F2. Anything may be the opinion about Etruscan-Uralo-Altaic connections; still at least a tendency for a sort of "vowel harmony" is conjectured in Etruscan (Pallotino, 1975), although not quite the same as in the previously mentioned languages. The other "assimilation" happens between two neighbouring sounds.
Quite clearly the subsequent vowel did influence the preceding consonant in Latin, because in the Late and Neo-Latin evolution C before back vowels (a, o and u; low F2) remained a K-sound while before front ones (e and i; high F2) changed into spirant or affricate. Of course we do not know anything about this directly in Etruscan.
Now, there was a Hungarian study for the CV interaction in open syllables (Olaszy & al., 1982) for artificial speech, which demonstrated the serious influence of V on C. We admit that Hungarian is rather peculiar among the present European languages; however Etruscan was also not closely related to most present European languages, so it is as justifiable to compare Etruscan frequencies to Hungarian as to English ones. The Hungarian result suggests something for the reason of the multiple representation of [k].
The average frequency of K2 of Hungarian [k] is invariably 4500 Hz. However the frequency of K1 is highly dependent on and strongly correlates with F2 of the subsequent vowel (Borbély & Lukács, 1986). The data are seen on Fig. 1. The four Hungarian vowels possibly nearest to the four Etruscan vowels are indicated above the points.
What is seen is an almost linear connection between K1 of [k] and F2 of the vowel. The dashed line is the best power fit, and the exponent is 0.984. So the actual pronunciation of the only K phoneme of Hungarian is very diverse.
Now, let us see if this is a general property of consonants. According to the measurements (Olaszy & al., 1982) the vowel influences the acoustics of many preceding consonants, but in this extent the only other such Hungarian consonant is [g], and voiced stops, nonexistent in Etruscan, are irrelevant here. Fig. 2 show the Hungarian data for [t], another unvoiced stop, and the curve is quite different. There is a dependence on F2, but moderate, with two horizontal asymptotics at the ends.
Fig. 3 is the liquida [r]. There is a central horizontal region, so there is no dependence on F2 of V in the middle parts.
Fig. 4 shows the data for the nasal [n]. The horizontal asymptotics are long, with a short transient zone, as if [n] participated in the Hungarian "vowel harmony".
Finally we mention an example when in CV C is not influenced. The consonant is the nasal written by the NG digraph in Hungarian. It is questionable if this sound represents an existing phoneme at all or it automatically appears in N+G situation as a variant of [n]; however the study was extended to it. Formants were recognised with the result that F1=250 Hz, F2=1300 Hz, F3=2400 Hz and F4=3400 Hz, invariably.
Now, in the Discussion, we return to the Etruscan K's.
Let us return to the analysis of writing systems. Vowel harmony is not reflected on the consonants at all in present Hungarian, Finnish or (Ottoman) Turkish orthographies. Consonantal phonemes have unique alphabetic representations and the lettering of back and front vowels are different. As told above, the Szekler rhunic orthography of Hungarian has double representation for [k] and only for [k]; one before front vowels, one before back ones. Now, Hungarian linguistics is certain that there is only one [k] phoneme. It is very probable that the double representation of [k] is superfluous and redundant in the Szekler rhunic script.
The old rhunic script for (Old) Turkish, whose longest texts are the Orkhon Incriptions (Thomsen, 1896) was mentioned above. There, as mentioned above, 9 consonants were doubly represented, and the Hungarian study suggests that more or less that ones which "participate in the vowel harmony" (Olaszy, 1982; Borbély & Lukács, 1986). It was necessary in the Old Turkish script to indicate high/low frequency on consonants, because front/back quality was not indicated on vowels. However [k] was quintuply represented in the alphabet, which was redundant even in that orthography. It seems as there were a tendency to overrepresent [k] in orthographies. Our guess is that this is caused by the great variance in the acoustic data of [k] (obviously in the first noise maximum).
And now we can finally return to Etruscan. We do not have, of course, Etruscan frequencies, but Fig. 1 shows the Hungarian ones. Let us see only the [k]'s before the 4 vowels which, possibly, approximate the four Etruscan vowels. Then
Archaic Etruscan script
K1 of K, Hz in Hungarian
Table 1: The locations of first noise maxima of [k] in KV syllables: Hungarian data
As seen, there are 3 levels separated by two 620 Hz steps, then the last one is only a 300 Hz step, while the total variances of other consonants are no more than 300-400 Hz. If one makes orthography after ears, then it is possible that he writes all other consonant with a single letter, but [k] with 3. And the rule mentioned by Pallotino (1975) is quite conform with the (unfortunately, Hungarian) data: Etruscan Q seems to mean something with K1~900 Hz, K something with K1~1500 Hz and C something with K1~2100-2400 Hz.
So we guess that Q, K and C of Archaic Etruscan script were 3 representations of one phoneme, [k]. The frequency data confirm Pallotino's opinion that there were slight differences among the pronunciations of C, K and Q; indeed the difference is not even slight in Hungarian, in the different K1 values. In addition we think that this difference may have even survived in Neo-Etruscan, even if the orthography ignored the difference, as Modern Hungarian and Turkish ones ignore them, while Old Hungarian and Turkish ones did not. Observe that the data are from Modern Hungarian, so it is quite possible that an orthography ignores acoustic differences.
The probable explanation is that alphabetic scripts must distinguish phonemes, but does not have to distinguish pronouncial differences. If the differences of K1 are automatically caused by the next vowel, then one consonantal letter suffices. In Old Turkish doublets were needed for consonants because the orthography did not distinguish back and front vowels (that is, F2's), but the quintuplet was not needed. Modern Turkish eliminated the doublets by indicating F2 on the vowels, and in the same time eliminated the quintuplet of [k], showing that even in Old Turkish the quintuplet was redundant. (In between the Old and Modern Turkish Turks used the Arabic script not indicating vowels at all. However Arabic has two "k-sounds", so experts of Medievqal Ottoman Turkish may look for KV/QV correlations. Otherwise that milleneum is irrelevant here.) When Turks adopted the Latin script, they eliminated the doublet for [k] as redundant. Probably that happened between Archaic and Neo-Etruscan too.
As briefly mentioned above, the Etruscan script originated from a preclassic Greek alphabet, still with remnants of Old Semitic scripts. In the original system there were more than one "K" phonemes. It seems that at the beginnings of literacy Etruscans did not yet draw the conclusion that the [k]'s of very different pronunciation belong the same phoneme, and they did not eliminate the redundant representation, but after some evolution two of the letters seemed unnecessary and went out of use.
However, as Hungarian frequency dates show, the extinction of C and Q does not indicate that the pronunciations coincided in Neo-Etruscan times. They may have remained as different as in Archaic Etruscan, although it would be rather hard to prove the differences. We only can use the argument that it would be inconveniently troublesome to produce the same K1 before vowels of very different F2's.
APPENDIX: ON SZEKLER RHUNES
Szeklers are the Eastermost Magyar/Hungarian speakers, in Eastern Transylvania. Their origin is disputed. Hungarian historians either believe them to have descended from the conqueror Magyar tribes not much after 896 AD, or conjecture some Turkish origin (either from Kabars of the Khazarian Empire or from the Eskils/Esekhels of the Bulgarian Turks). Szekler historians either accept the Turkish connection, or try with surviving Avars. Finally, the Szekler tradition is firm that Szeklers remained in (relatively) the West when "sons of Attila" returned "to the East". (The event is clearly the migration of Dengizik & Irnak, after the unsuccessful Nedao Battle in 454 AD, from the Carpathian Basin to the Pontus.)
Now, Szeklers use a rhunic alphabet. Since it was in the past applied mainly on wood, really old surviving texts are rather rare; from Renaissance times there were preseved alphabets, but that is already the time of decline; the literate people know Latin alphabet too, and the rhunes are used rather for fancy. For this reason, the alphabets fairly vary; therefore for foreigners it would be an esoteric lore to compare different Hungarian sources. Instead I give one Austrian author (Doblhofer, 1957), who gives a "good, average" alphabet; that is enough. The longest extant and still traditional text is 109 character long (details come soon), and it uses the "-ka/-ak" of Doblhofer always in the neighbourhood of a back vowel, and his other "k" with front ones.
Now, this rhunic alphabet is in a clear connection with Turkish ones, including the Orkhon Old Turkish in Mongolia. The Szekler rhunes are sometimes called Magyar or Hungarian, but there are only 3 inscriptions outside the Szekler territories. They are as follows.
1) The first King of Hungary was Stephen I (1000-1038). Now, the very first Hungarian coins (silver obols/half-denars) have indeed the inscription "STEPHANUS REX", and are similar to the (earlier) half-denars of Stephen's father-in-law Henry II, Duke of Bavaria, but two of the letters (T & P) are rather unusual, and look like rhunes. However they do not belong to the Szekler alphabet.
2) There is a single rhunic inscription from present Slovakia (Püspöki Nagy, 1977). The orthography is somewhat peculiar, so it is possible that the scribe learned the writing poorly.
3) The Constantinople inscription of 109 characters comes originally from humanist and economist Derschwam on an embassy to Turkey in 1553-5 (Babinger, 1923). He spent some time in the "House of Envoys" in Contantimople/Konstantiniyya/Istambul, and observed a piece of marmor stone inserted into the stable wall with an inscription of unknown characters. Obviously he took it antique, so copied 109 characters, but later the text turned to be Magyar, in Szekler alphabet (Babinger, 1914). The author of the script is named in the inscription, it is Barnabas Bélaji, envoy from King Wladislaw II of Hungary in the 1510's, who had to spend 2 years idly there.
The Szekler texts are always in Magyar language, and in Magyar there is/was always a single [k] sound. So there is no more reason for 2 "K" letters in the rhunic scripts than for 2 in Greek, for 3 in Latin, or for 5 in Old Turkish. Still, there are all these redundancies.
Alieni M. 2003: Etrusco: Una forma arcaica di ungherese. Il Mulino, Bologna
Babinger F. 1914: Ungar. Rundschau 3, 44
Babinger F. (ed.) 1923: Hans Dernschwam's Tagebuch einer Reise nach Konstantinopel und Kleinasien, 1553-1555. Studien zur Fugger-Geschichte, Vol. XIV, München-Leipzig
Borbély I. & Lukács B. 1986: Acustica 63, 129
de Graaf T. 1982: Proc. 8th Colloq. on Speech Acoust., ed. T. Tarnóczy, Budapest, p. 211
Doblhofer E. 1957: Zeichen und Wunder. Paul Neff, Wien-Berlin-Stuttgart
Lotz J. 1984: Research in Altaic Languages, ed.L. Ligeti, Budapest, p. 135
Matheev B. V. 1984: Issledovaniya zvukovyh sistem yazykov Sibirii, ed. V. M. Nedelyaev, Nauka, Novosibirsk p. 56
Olaszy G. & al. 1982: 8th Colloq. on Speech Acoust., Budapest, p. 204
Pallotino M. 1975: Etruscologia. Hoepli, Milan
Peterson G. E. & Barney H. L. 1952: J. Acoust. Soc. Amer. 24, 175
Püspöki Nagy P. 1977: Magyar Nyelv, 78, 303
Tarnóczy T. 1966: Általános nyelvészeti tanulmányok X, eds. Zs. Telegdi & Gy. Szépe, Akadémiai Kiadó, Budapest, p. 181
Thomsen V. 1896: Inscriptions de l'Orkhon déchiffrées. Mem. Soc. Finno-Ougrienne, Helsingfors, Vol. 5
My HomePage, with some other studies, if you are curious.