Indo-European linguistics: the Indo-Iranian vowel destroyers

After so much archaeology and genetics, we should speak of linguistics, which is the root of the concept of Indo-Europeans and has an important role in Indology. 

Actually, the evolution of modern Indo-European linguistics is quite similar to that of the question of the Indo-European homeland or Urheimat. Originally, Sanskrit was regarded as the most original Indo-European language, and India as the homeland, then Central Asia (particularly with Pictet) became the official candidate, and so Indo-Iranian had still the most authoritative position. But during the 19th century, some doubts arose about the fact that Sanskrit or Indo-Iranian were more original than European languages, starting from Greek. It is interesting that this change was contemporary with the passage from a long period of Orientalism (in a positive sense, summed up in the formula ex Oriente lux) to Eurocentrism in European culture. Europe was becoming the imperialistic master of the world, and had no more to learn something from Asia. In the 1820's, Hegel developed a philosophy of history which placed Europe at the summit of an evolution where the East was quite primitive, and Greece was the place of the great leap of freedom and philosophy. As he said in his Lectures:

The Orientals do not know that the spirit or man as such are free in themselves. And because they do not know that, they are not themselves free. They only know that One is free.... The consciousness of freedom first awoke among the Greeks, and they were accordingly free; but, like the Romans, they only knew that Some, and not all men as such, are free.... The Germanic nations, with the rise of Christianity, were the first to realize that All men are by nature free, and that freedom of spirit is his very essence. 

According to Hegel, all European science and art originated from Greece. He agreed that Greek origins should be traced in the Orient, but whatever the Greeks received from the Orientals, they 'reshaped', 'reformed' and 'rebirthed'. He also stated that the Greeks were the bridge from savagery to civilization and the inventors of history (see here).

Theodor Benfey

German culture at that time had a cult for Greece (as is seen also from Neoclassic architecture in Berlin and other German cities), which overcame the cult for India and the East. Is it a case that this is reflected also in linguistics? In 1837, Theodor Benfey, an important German philologist, suggested that the five Greek short vowels (α ε ι ο υ) could be regarded as more original than the 'poorer' Sanskrit system of three short vowels (a, i, u), rather than seeing e and o as a 'clouding' ('Trübung') of a (see here). So, Greek vowels were not a corruption of the Sanskrit ones, but the complete paradigm, impoverished by the Indo-Iranians. However, this view overturned the traditional view, held also by Hegel, that the Orient was the origin of Greek civilization, and that the evolution went from the simpler to the more complex.

In 1848 Curtius still replied to Benfey's suggestion observing that Sanskrit, Gothic and Old Persian all agreed in a system of three short vowels. But ten years later, in a work on Greek, he supported the idea that Sanskrit was given too great a prominence in comparative grammar. Then, the Neogrammarians arrived and 'discovered' the Law of Palatals, according to which two front vowels (e, i) palatalized the velars in Indo-Iranian: in the case of e, before becoming a. A famous example (see here) is the Sanskrit perfect of kṛ 'to do': cakāra. Why not kakāra? Because it came from *kekora, and Greek attests the original vowel with the  reduplication vowel for the perfect tense ε (e.g. γέγονε). But then, why do we never find cited the example of cukrodha, perfect of krudh, or cukopa from kup, or jugopa from gup? Where was there the front vowel e or i? Simply, in Sanskrit every velar is normally substituted by a palatal in reduplication, maybe for euphonic reasons of dissimilation.

The fact that Greek did not have such a reduplication is not a proof that it is not original in Sanskrit, and actually, the Sanskrit system has an inner logic, because roots have reduplication with the respective vowel of the root (a,  i or u), whereas Greek always uses e. As remarked by Kulikov (see here), "Old Indo-Aryan seems to preserve the original Proto-Indo-European system of verbal reduplication better than any other ancient Indo-European language". We can even suppose that Greek has e because the previous palatal palatalized the subsequent vowel, before becoming velar (in ancient Greek palatal sounds were unknown, I guess, because the substratum did not have them). About palatalization, it seems that for the law of palatals linguists had in mind the process that happened with Latin in Italy with the palatalization of the Latin velar c before e or i, but it is interesting that in French we have palatalization also before a, like chat from cattus 'cat' (see here).

Another important example often given is the Sanskrit particle ca compared with Latin que, Greek τε (Mycenaean qe), again explained as a palatalization due to the original front vowel e.

But if Indo-Iranians were so inclined to palatalization of the velar before e and i, why do we have many words with the syllables ki and ke (which however comes from *kai) or with ka from a supposed PIE *ke? Even if we admit that the neuter kim kept the velar because of analogy with masculine kas, why do we have a verb like ciketati, which is supposed to come from *k^u̯eit-? Or ciketi from ci 'to observe, perceive', supposed to come from *k^u̯ei-?  When does an exception stop confirming a rule and start confuting the existence of the rule?

However, my objection is more radical. I find difficult to believe that from a system of five short vowels was possible to arrive at a system of three, and particularly that all e and o became a, with a systematic will of simplification that would be an astonishing phenomenon in human history! In particular, while the passage from o to a is not unknown (for instance in Russian, and often in (American) English o is pronounced like a), the passage from e to a seems very unusual. On the contrary, the passage from a to e or o is attested in many languages, including Indo-Aryan and Iranian ones. Sanskrit itself changed as into o at the end of a word and into e within  a word. 

Also English is a good example: German Mann corresponds to English man (pron. men), German waschen to English wash (pron. wosh). If English was not written before the change, we would have now a language with a lot of e and o instead of a, and maybe some linguists would argue that the English vocalism could be more original...

As we read in this page on Indo-European vocalism:

The vocalic system appeared completely destroyed in Indo-Iranian, where two main IE vowels, *e and *o, coincided with *a. But the transition *e > a is not unique here: a few Anatolian languages, Luwian and Palaic, also suffered this feature.

How sad, that those barbarous races destroyed the beautiful European vocalic system!

It seems that the average linguist is not aware of the problems of this theory and generations of linguists did not find anything strange in the fact that 'Indo-Iranians' have transformed every e and o without exception into a, which is also not very useful for distinguishing words. The only justification that I can imagine for such an incredible theory is that the substrate language did not know e and o, like Classical Arabic. But this would imply that the Indo-Iranians were practically unmixed with the original Indo-European speakers and we also wonder why did they develop those sounds later in every Indo-Iranian language. And how is it possible that no Indo-Iranian dialect preserved them? The case of Arabic is also interesting, because a often becomes e in many dialects (see here), and in non-Arab Muslim countries like Turkey you can hear Arabic words pronounced with e, like Allahu ekber...

Moreover, as remarked in the quoted passage, you can find a for *e also in some Anatolian languages, which are regarded as the most archaic for some aspects. But strangely, it is not remarked that for supposed PIE *o we find a in Anatolian languages, except Lycian where we have e. Maybe because it refers to Proto-Anatolian, where linguists admit all five vowels as 'nearly intact' (see here). And obviously, the fact that Luwian has only the vowels a, i, u is seen as a 'reduction' (see here).
Anyway, looking at the list given on that page, we see that we have a for *o in Albanian, Illyrian, Thracian, Phrygian, Germanic and Baltic, besides Indo-Iranian. Three of these languages (Illyrian, Thracian, Phrygian) are ancient, and Baltic has preserved many ancient IE elements. Is it not a hint that a could be more original? With such a division among IE languages, how can we determine if one or the other is the original vowel?
The Spanish scholar F. Villar maintains that o was not present in PIE. And he also remarks that 'Old European' has often an a-vocalism. For instance, there is the common element in river names Var-, which is derived from the PIE root *(H)u̯eh1-r- like Vedic vār 'water'. In Luwian, we have u̯a-a-ar, in Tocharian B war. So, also Tocharian has a for *e in some cases, and also a for *ō̆, as in Toch. A śtwar 'four' (PIE *kʷetu̯óres), or vak 'voice' (PIE *u̯ṓkʷs). Meticulous linguists obviously say that all these are modifications of the original vowel, but I find this pure dogmatism. The common theory asserts that PIE *o became Proto-Tocharian *e or *æ, which became a in Toch. A, and e in Toch. B. Is it not more probable that all this comes from *a rather than *o? A change from a back to a front vowel is much more difficult than from the central or almost central vowel a. We should notice here that in the pronunciation of Sanskrit, short a is actually pronounced as a central vowel (see here), and not as a front vowel like a in many European languages. This central vowel could change easily either into the front vowel e or into the back vowel o.

A problem is that many linguists are fond of hypothetical proto-sounds, and they deal with them as if they were real entities. If you raise doubts about them, they react as a Catholic if you deny the Holy Trinity.
Laryngeals are one of these sacred entities. Linguists discuss about their number as theologians could discuss about the number of divine hypostases... or maybe they are more similar to mathematicians, and they think that linguistic facts are object of deduction like mathematical formulas. Platonism often emerges in Western science... Actually, Hermann Grassmann, a precursor of the Neo-Grammarians in searching for rigorous laws in phonetic change, was first of all a mathematician... And recently, as observed by Bernard Sergent, there is a sort of algebraic mathematization ("mathématisation algébrique") of comparative grammar, Benveniste's theory of the root has brought to a decomposition of words and phonemes, and the Indo-European words reconstructed nowadays resemble equations...

It is true that Anatolian languages give us the cuneiform sign for a sound transcribed as ḫ where other languages have no sound, like Hittite ḫants 'front, face' for Sanskrit and Greek anti, Luwian ḫawi- 'sheep' for Sanskrit avi, Latin ovis. But even if we accept that this is not an adding but an original feature, and that ḫ and ḫḫ (often found in Anatolian texts) represent two different sounds, where is the proof of three or four (not to say 6 or 8 or more) different laryngeals, outside the metaphysical reconstructions (often implying pronunciations that transcend human abilities)?

Ferdinand de Saussure

What is significant, is that the laryngeal theory has its roots in the coéfficients sonantiques proposed by de Saussure in order to explain why the supposed original *e became a or o. But if we admit that a was original, and that it changed easily into e and o also without necessity of any special coéfficient, we do not need three laryngeals. Particularly the 'a-colouring' laryngeal *h₂ would be useless. And about the o-colouring one (*h₃), let us take the case of English. The fact that ball is pronounced bol is not due to any mysterious laryngeal, but rather to the subsequent l, like in wall and fall. And also the change from a to o in Bengali does not require disappeared sounds, for instance nômoshkar for namaskar.
So, also Brugmann's law, which states that "Proto-Indo-European *o (the ablaut alternant of *e) in non-final syllables became *ā in open syllables (syllables ending in a vowel) in Indo-Iranian" can be reversed. There is another significant fact about the pronunciation of ā in Sanskrit: it is an open back, rather than central, vowel. And the same is true for Persian long a, whereas short a is pronounced as the front vowel æ (see here).
Since Greek is particularly involved in Brugmann's law (Skt. jajāna, Greek γέγονε), we can suppose that *ā in non-final syllables in Proto-Greek was pronounced as a back vowel, whereas short a was often pronounced as a front vowel in a way similar to Persian, then the quantitative opposition *a/ā was substituted by a qualitative opposition *e/o. In the case of final syllables, instead, final *ā was retained in Greek (and Latin too), and interestingly it became easily ē (η) in Ionian dialects.
As regards Tocharian, the passage from Late PIE ā to Proto-Tocharian o is already generally admitted, except in final position (as I have found in an article by Ronald Kim, cp. here).

Another interesting aspect in Greek (and Latin) are the nouns with a stem alternating -es-/-os- like genos-/genes-. If we assume that Indo-Iranian -as- is original, this alternation can be understood as due to two modifications, in different cases, from the same sound. And the same can be said for all the cases where some IE languages have e, others o, for instance Latin pēs, pedis 'foot', Greek πούς, ποδός, or, inversely, Latin novus 'new', Greek νεός.

Here I stop. I have not given a full systematic theory, but I think these hints, that I have collected in some years of observation, should be object of reflection, and maybe of new paths of linguistic research. I dedicate this post to Satya Swarup Misra, whose writings have given the first impulse to these considerations.


Giacomo Benedetti, Kyoto, Japan, 1 August 2013