Sunday, September 12, 2010

An elegant explanation for the lack of R-L distinction in Linear A and B

I am now going back to one of the problems that have long plagued the research on ancient Aegean languages: namely, the baffling lack of r/l distinction in all Minoan writing systems. The only exception, Cypriot, was born outside Crete. Although I initially proposed the possibility that the Minoan language inherently lacked r/l distinction, like Egyptian (as some extant languages, e.g. Japanese also do), the sheer number of Minoan loan-words present in Greek, with good examples for both r (e.g. Rhethymnon, Rhadamanthys, Rhea) and l (Lyktos, lyre, lily, etc.) sounds, have shaken my theory. All other Aegean-born languages, such as Etruscan or Eteocypriot do have both l and r, and so do all Anatolian languages. Also, Eteocretan - a likely descendant of Minoan - apparently did possess both laterals (e.g. isalabre, *isal-awr-e [goat-cheese]) and trills (e.g. the common word irer). This is not consistent with my original theory, so I gradually decided to give it up entirely.

Recently, while I was collecting material and doing research for my future posts, a brilliant new idea came to my mind. In all Anatolian tongues, a word-initial r- never happens (it is forbidden), thus there are no initial Rv-type syllables either. What if Minoan was similar in that regard? All Minoan writing systems are believed to be largely acrophonic, right? Now what if a particular sound is forbidden in initial positions? That is where you need a work-around to the situation. The Cretan solution: without giving up the acrophonic character of the script, use the L-series signs as a substitution for Rv type syllables! Really simple, no?

Before going into details, we should return to the Egyptian script once more. Many experts of Egyptian linguistic history would immediately quarrel with my above view, because of two proven facts: 1) Egyptian also lacked an r/l distinction, and 2) The Middle-Kingdom Egyptian hieroglyphic system was the eventual basis of the Minoan script. But we have to keep in mind a number of facts: First, Minoan Hieroglyphs are not a simple copy of the Egyptian system. Apart from the fact that the objects depicted are different, and that the system is now a pure (open) syllabary, there are also many new phonetic values represented: for example, pure vowels. Minoans did not simply copy-paste the Egyptian system, like Mycenean Greeks did with Linear A. So the percieved similarities between the two are probably a non-issue. Second, we know pretty well from ancient authors, that Eteocretan language was confined to the eastern half of Crete: eastwards from the Knossos-Phaistos line. This is exactly where most relics of Minoan culture were found, also including the overwhelming majority of Linear A tablets. Thus even if the Western Cretan dialects were slightly different (the indigenous people were referred to with a different name. the Kydones). they would have had but a small impact on the development of the Minoan writing system. On a geographic territory as tiny as Eastern Crete, dialectal differences should have been negligible in the Middle Minoan era, as culture was otherwise very similar. And the uniformity of the Linear A records seems to testify this. Thus if Eteocretan was a descendant of Minoan, the latter should have had a good r/l distinction as well.

I have been studying the Anatolian-Aegean connections a lot lately. As you likely know all too well, both the Aegean and the (Indo-European) Anatolian languages were subject to an areal effect, from the early Bronze Age onwards. Not only the phonological characters have become similar, words and even complete grammatical structures were also exchanged. Though it is in my intention to write a complete post about the Anatolian loans in Minoan, I can cite a few examples beforehand. For example, take the Minoan theonym Alauta (A-RA-U-DA = classic Greek Eileithyia). Her name is undoubtedly derived from the Proto-Indo-European stem *h1leudh- ='free'. The meaning of the name makes perfect sense, since Eileithyia was mostly worshipped by pregnant women, in hope of an easy childbirth and less complications and labour. However, the name Alauta already shows developments specifically pointing to an Anatolian language as the donor of this phrase: These include the disappearence of initial h1 laryngeals with the consequential a-colouring of initial vowels (h1e->a), and the de-aspiration of stops with a subsequent conversion to a simple, voiceless stop (dh->d->t).

Phonetics could be borrowed too. The "Aegean" languages had many peculiar characteristics, largely shared with their Anatolian neighbours: For example, there was either no (Etruscan, Lemnian) or very little (Minoan, Hittite, Luwian) distinction between o and u vowels. Neither Minoan, nor Cypriot or Etruscan had any distinction between voiced or voiceless consonants: this is reflected by the lack of voiced consonants in writing. On the other hand, they certainly (Etruscan) or probably (Minoan) possessed a second series of consonants, not distinguished by voicing, but rather stress or aspiration. Linear A also used an additional series of stops - either stressed or (even more likely) aspirated. Thus the difference between -say- PA and PA3 syllables was that of *pa and *pha (or *ppa), while DA and TA likely represented *ta and *tha (or *tta). Of course, simple stops were likely commonly pronounced as voiced, while stressed stops were certainly voiceless (this explains the evolution of *ta and *tha/*tta sign-values to Linear B DA and TA), but voicing was originally not phonemic. Cuneiform Hittite also had this strange feature: it rejected the distinction between the original voiced and voiceless syllables: instead, a contrast system based on single/doubled consonants was used. This keeps hittitologists in uncertainty even up to this day: Did Hittite have a voiced-voiceless distinction at all? Although simple consonants mostly correspond to PIE voiced consonants (and conversely, the double ones are expected to be voiceless), the phonemic character of sounds is disputed. The problem is, there is no trace of voiced-voiceless distinction in Luwian Hieroglyphs either. It is certain,that even if Hittite did retain voiced-voiceless contrasting to some degree, it must have borrowed the Akkadian cuneiform script (and perhaps the Hieroglyphs, too) via a local substrate language that had no such contrasting at all, instead having lenis (simple) and fortis (double) consonants.

Recently I wondered if some phonological features also went in the opposite way. We know all too well, that in Hittite or Luwian, no initial r- sounds were allowed. This was a restriction inherited from Proto-Indo-European, and the ancient Anatolian languages preserved it faithfully. Although the Aegean languages were not Indo-European by any means, but - as many examples show - were subject to heavy IE, particularly Anatolian influence. What if some borrowed this feature too? If so, that would provide a brilliant answer to the question why the Minoan scripts had no separate signs for Rv-type syllables. Because it was an acrophonic system, it simply could not build any signs for syllables forbidden in initial positions!

Conforming the mentioned theory, members of the Linear A and B R-series all appear to be standing for L+vowel syllables. There are at least two, separate lines of evidence for this. The first one employs the Hieroglyphic counterparts of Linear A signs, and the meaning of their images. There are at least two signs in the R-series with a good etymology. The RE-sign, depicting sea lily flowers, possibly stands for an acrophonic abbreviation of *(a)leri ='lily' (disregarding my previous concerns about the conflicting origins of RE and RA3). The RU-sign, on the other hand, originally depicted a lyre, getting its phonetic value from *lura = 'lyre'. Though the stem-words are not attested in (the mainly accounting) Minoan texts, they very likely originate from or were transmitted via Crete, whence they were borrowed into a number of Mediterranian languages, such as Greek and Latin.

The next line of evidence comes from the evolution of Linear Minoan syllabaries: namely the way Cypro-Minoan and Cypriot Linear C were derived from Linear A. Interestingly, the Cypriot scripts do have a distinction between r and l consonants: they present a different series for each. However, if one looks at the way the signs were drawn, will quickly realize that it is the R-series that is novel. Most of the L-series signs are actually taken over from the Linear A R-series without a major design change. This is a further confirmation about the "true" phonetic value of Linear AB R-series signs.

At this point one could ask if Minoan scribes could not have found a simpler solution for this problem. Why did they not relax their rules for sign-generation, allowing a word with v1cv2- initial structure to be the basis of a cv2 sign? It is the most rational solution one could find to their problem. The Luwian Hieroglyphic script - that was less stringently acrophonic than Minoan - also made widespread use of mid-stem open syllables. Indeed, some evidence does suggest that Minoans used such workarounds, too: The Linear A and Hieroglyphic PI sign undoubtedly depicts a bee. However, the pi syllable is rather rare as a word-initial in Linear A (it is a rare sign altogether), and possible cognates (Latin apis = 'bee') suggest that the corresponding word also started with a vowel in Minoan (*api? *ipi?). With this example in mind, it somewhat harder to explain why the scribes did not create any Rv signs. Clearly, the mere lack of initial r- is not a good enough answer on its own.

This is not to say, that an r-sound was not allowed word-initially when in clusters. The solution to the above question probably lies in the way trills behaved in Minoan word-radicals. Unlike Hittite or Luwian, that admit a number (although a restricted number) of words with v1Rv2- initial structure, most Minoan words (reconstructed from Mycenean loan-words) seem to present the trills as part of a consonantal cluster (e.g. v1Rcv2). Minoan loan-words in Greek that begin with r- can mostly be traced back to stems with such initial consonantal clusters. For example, Rhadamanthys might continue *Artamantha, rhodon (= rose) *wrata or *urta. Words like rhétiné (O-RA2-DI-NE = resin) and the divine name Rhea (RA2-T?) actually stand with RA2 = rya in Minoan records, implying a consonantal cluster in the word-radical. Ariadné might have been as well *Aryatna. If true, this would have made it impossible for the scribes to find a suitable word with a word-initial pure v1Rv2- structure, because there were probably too few or maybe none.

A further, elegant proof for the existence of initial v1Rcv2- clusters we find in the Minoan sign RA2 = *rya. It clearly depicts a stream or small river. We know from Hittite, that the word-stem for 'to flow' (and also 'river') was arsa-. The Hittite word is derived from an IE stem *h1ers- = 'to flow', also attested in Sanscrit as arśa-. But we should not stop here. This is not the first time we see perfectly IE stems in purely Aegean context (a good example is PO-TO-KU-RO in Linear A from the IE stem *pot- = 'powerful') If we allowed a hypothetic "Proto-Aegean" language to borrow the same *arsa (to flow), that would nicely admit a Minoan word *arya = 'stream' (by lenition). Our theory can also explain a previously shunned connection: the phonetic value of the corresponding Linear C sign. Although it is clear, even to an untrained eye that the Cypriot ZO sign is almost identical in design to the Linear AB RA2 sign, no one has ever been able to give a consistent derivation of the Linear C value. Now we have one: If the Cypriot word for 'stream' was something like *azzo (by assibilation from *arsa), it would be more than meaningful to suggest a correction of phonetic value of the "river-sign" from rya to zo. And this is not the only case when such Cretan-Cypriot lingustic discordances have triggered a slight sign-value correction: For another good example, see the article of Miguel Valerio discussing the identity of the Linear A DU sign with the Linear C SU one. Unfortunately, we cannot establish such a nice etymology for the Linear AB RO2 = ryo sign, as we cannot determine the object it depicts. All I can say at the moment its that it likely also represents a "true" r-sound, like its RA2 counerpart (given that it alternates with the former in Linear A texts).

Given these problems with the Minoan syllabaries, it is no surprise that the scribes used both the L- and Ry- signs indiscriminately, for both r and l sounds. The L-series was probably used as a shorthand solution in place of open syllables with r-. While in the earliest Hieroglyphic documents, the scribes likely also experimented with the use of Ry- signs (e.g. RA2) at this position, the latter eventually remained constrained to the clusters ry- or ly-. And so was the clumsy Linear B ortography born.


  1. For example, take the Minoan theonym Alauta (A-RA-U-DA = classic Greek Eileithyia). Her name is undoubtedly derived from the Proto-Indo-European stem *h1leudh- ='free'. The meaning of the name makes perfect sense, since Eileithyia was mostly worshipped by pregnant women, in hope of an easy childbirth and less complications and labour.

    Hm. In what context were the relevant ancient words that shared the proto-Indo-European stem for "free" used? "Free" is a very overloaded word in modern English. Was "free from pain" a correct use back then? Could "free" imply no price? Was "free" a word reserved for non-slave peoples?

    In my ideal universe, I will have an etymological dictionary which takes pages for every stem word and tracks its entire history of usage, through multiple languages, spelling changes, etc., and would detail answers like this.

  2. Sorry for not citing references here. I shall remedy that by devoting an entire post to this theonym. Until that, I give the following examples (I hope this will be sufficient).

    Some of the better known derinatives of Proto-Indo-Eurpean stem *h1leudh- (also reconstructed as *leudho) are the following: Greek ελεύθερος = 'free', 'liberal', Latin liber = 'free', German Leute and Russian люди = 'people', even the non-Indo-European Etruscan has lautni = 'freedman'.

    Do not be fooled by the wikitionary list of PIE words: given the above distribution of meanings, the original one was likely the adjective 'free' and hence also 'free people' (neither slaves nor nobles).

    As for the presence of the stem in Proto-Anatolian, we have no direct proof. Hittite uses a different stem do denote the adjective 'free': arawa-. So the case is not closed yet...