c-AFRASIAN-3_schwa.htm

Tlazoltéotl

PROTO-LANGUAGE PHONEMES

in IE and Afrasian

Roman Marble. circa 100 PE


    Preliminary to discussing the nature of schwa indogermanicum, let us take a closer look at the characteristics of the series of long vowels in Indo-European.

    Indo-European e should be considered to be a mid-low front vowel based on its reflexes in the derived languages; Indo-European o is should be considered to be a mid-low back vowel also based on its reflexes in the derived languages. Based on the reflexes in the derived languages, Indo-European e: should be considered to be a mid-high front vowel; and Indo-European o: should be considered to be a mid-high back vowel. We observe that lengthening, on the basis of the reflexes in derived languages, has a correlation with degree of closeness as well as duration of articulation. The third vowel reconstructed for Indo-European, a, should be considered to be a mid-high/mid-low central vowel, which is the articulatory position of the vowel which is often called schwa. If we assume a degree of closeness analogous to that which we believe is demonstrated by reflexes of e: and o:, then we would expect that a: would be a mid(-high) central vowel. Further evidence for this interpretation of the articulatory position of a: is that in Lithuanian and Germanic, its reflex is o:, a mid-high back vowel, and in Ionic-Attic Greek e:, a mid-high front vowel. These two reflexes have in common the mid-high degree of closeness; and it is suggestive that the Indo-European articulatory position of a: was also mid-high but central. However, the majority of reflexes to low central vowel for a:. Which of these is original is difficult to say but the picture that emerges is that in the central position, length created the opportunity for additional openness as opposed to additional closeness in the front and back positions. These considerations have a bearing on the nature of the so-called schwa indogermanicum, which we will indicate as 6 in this essay.

    Up to 1912, the traditional view of Indo-Europeanists was that there was an Indo-European sound, called schwa indogermanicum, that represented a weakened form of e:, a:, and o:; and was realized as a "Murmelvokal". Thus, in a root like *p6te/ér, 'father', according to traditional teaching, the initial syllable represented a reduced grade of *pe:/a:/o: occasioned by the stress-accented element (*-té/é:r) which followed it.

    In view of the phonological relationship between schwa and what we may reasonably assume to have been the phonetic realization of a:, we can imagine that, if the root were *pa:-, a reduction due stress-accent in the following syllable would produce *pa-; and that is exactly what we find for 'father' in all branches of Indo-European but Indo-Iranian: Greek paté:r, Latin pater, Gothic fadar, Old Irish athir but Old Indian pitá: and Avestan pitar.

    Thus, on the basis of all of the Indo-European branches except Indo-Iranian, it would be completely reasonable to reconstruct *paté/é:r. But, on the strength of Indo-Iranian -i-, Indo-Europeanists of the past reconstructed this strange entity which supposedly yields two incompatible results.

    In the case of *p6té/é:r, the very fact of the schwa in the stress-unaccented root syllable indicates that it is a reduction of some other vowel because Indo-European only permits e, o (the apophonic pair), and the long vowels e:, a:, and o: in stress-accented root syllables. This means that *p6- is a stress-accent-caused reduction of *pa:-, *pe:-, or *po:.

    Since -*té/é:r is an agentive suffix, we can investigate what its combination with these Indo-Europeans roots would produce — presuming that the root has not been lost, which seems unlikely for such an important concept.

    Indo-European pe:(i)- is indeed listed, and means 'to hurt, damage, revile'. To characterize a 'father' as a 'one who hurts, damages, or reviles' seems absolutely impossible. Indo-European pa:- has the main meanings of 'to feed, nourish, herd'. This also seems inappropriate to characterize a 'father' but, under this listing, we find Tocharian A pa:s-, 'to protect', and 'protector' might well be a term appropriate for 'father'. Indo-European 2. po:(i)- means 'to drink'; while 'father' as 'drinker' might be accurate in certain cases, it is unlikely as a source for the honorific appellation. Indo-European 1. po:(i)-, rather similarly to pa:-, means 'to herd cattle, preserve, protect, cover'. Obviously, in view of Tocharian A pa:s-, there seems to be overlap between the semantic ranges of the two roots. Now although 'he protects' is Old Indian pa:ti, the Avestan equivalent is pa:iti; and there is also Old Indian pa:yú-, 'protector'. As is well-known, Indo-Iranian -a:- represents both Indo-European -a:- and -o:- but, in the non-Indo-Iranian cognates, we find exclusively -o/o:-; and they refer overwhelmingly to 'herding' and 'covering'.

    These data lead me to the following tentative conclusions:

      1) there were two Indo-European roots of similar shape: one meaning 'to feed', and another meaning 'to protect';

      2) I assign pa:- to the meaning 'protect' and po:- to the meaning 'feed';

      3) several forms meaning 'to protect' seem to have the additional root extension -i.

    If I assume that the non-Indo-Iranian forms were based on pa:-, their short -a- is easily understandable without reference to a creation like schwa. If I further assume that the common Indo-Iranian form of 'protect' had the form pa:i-, then a zero-grade form could be pi-, which would account for Old Indian pitá: and Avestan pitar. Thus, it is not a differential response but a different stem which produces the difference between Indo-Iranian and non-Indo-Iranian cognates of IE pa(i)té/é:r.

    Now this remains only a perhaps slightly interesting suggestion unless more evidence can be brought to bear on the issue. Krahe (1966:57) cites another standard bearer of the schwa indogermanicum: Indo-European *st(h)a:-, 'stand', to which Old Indian sthitáH, 'standing', opposed to Greek statós, Latin status, Gothic staþs (OHG stat), Lithuanian statau[~], and finally Old Bulgarian stojon which is explained by the fact the the Old Bulgarian reflex of Indo-European -a- is -o-.

    What does sthitáH actually prove when we look at the entry for 'stand' and see that a common form with the root extension -i is sta:i-?

    Now these are the two examples that are always cited in the textbooks as examples of why schwa indogermanicum was reconstructed; presumably, they are the best examples, and we have seen that the best is not very good at all. I conclude, on the basis of these examples and analogous examples, that there is no justification for creation of a schwa indogermanicum to explain the data which are actually attested.

    But there are a number of other roots which show -i- in Old Indian but e / a / o in other languages. One example is *me:-, 'to measure', which has ma:ti-, 'measurement', and miti-, 'measurement', while Greek has métron; in addition, Avestan attests two participles: mi/i:ta- and -ma:ta. This looks very much like a root, *me:-, which always occurs with the root extension -i-; and Tocharian B mai-, 'measure', substantiates that such a stem did actually exist. Another similar root is *2. se:(i)-, which, of course is listed with the -i- extension. In an Old Indian form like prásita-, 'shooting off', we have our choice of regarding -sita as deriving from *-s6-tó (from se:-tó-) or *-si-tó (from se:i-tó-). In view of words like sé:na:, 'throwing spear', it appears likely that Old Indian is utilizing the extended root *se:i-. An additional root we might look at is g[^]en(6)-, 'to produce'. According to the general tenets of the application of the schwa indogermanicum, *g[^]N:-tó- (for form, see below) should arise from this stem + -tó, which would, in turn, occasion and does actually occasion Old Indian ja:tá-H, 'born'. What happened to the i:(Resonant) and u:(Resonant)? I see, what I believe, is a simpler explanation: *g[^]énV + *-a: = *g[^]o-né-a: + *-tó = *g[^]N-a:-tó = *g[^]a-a:-tó = g[^]a:-tó.

    Having made this assertion, we can now look at other evidence which bears on the question. On page 101 (Brugmann 1972, I) discusses some interesting forms derived from Indo-European *sta:- in Old Indian. The reconstructed form of the 1st person optative of that verb would be *st6-yé:m if we assume the current theoretical teaching. But, on the basis of the attested Old Indian ste:ya:m, Brugmann tells us: "On the other hand 6 became a before i-vowels." The root syllable e: is the normal Old Indian reflex of Indo-European *-ai-. Though Brugmann had another explanation, I think the simplest one is simply that, in the optative, Old Indian utilized a stem derived from *sta:-i-, namely ste:-, to which the optative terminations were added.

    So far, we have suggested that there is no need and little justification for reconstructing 6, the schwa indogermanicum, as the root-vowel. In a number of instances, it has also been reconstructed as an extension added to the root-vowel.

    One of the important bases for the "laryngeal theory" is the how derivations from roots like *ster6-, 'to spread out', develop in various environments. Lindeman (1970:23-28) describes the theory put forth by A. Cuny in 1912, which he believed was a justification for regarding the previously reconstructed long vowels of Indo-European as having a "weniger leicht vokalisierbares Element", suggesting, of course, the consonantal 'laryngeal', currently indicated by Hsubscript.

    Cuny's argument, as described by Lindeman, is as follows. The previously mentioned stem *ster-6-, when combined with -*nó, is reduced to *str6-, which, according to the theory that IE 6 should yield Old Indian i, must result in Old Indian **strin[.]á, which, however, does not exist. Now we have seen above reason to think that **strin[.]á is not the form we should expect in Old Indian. But instead of questioning the status of the schwa indogermanicum as I have done above when its theoretical predictions are not realized in fact, Cuny looked at the attested Old Indian form, sti:rn[.]á, which theoretically derived from an Indo-European *stR:-nó, and explained, in Lindeman's words: "Der lange silbische Sonant in *stR:- erklärt sich nach Cuny nur (emphasis added) dadurch, daß das, was von dem langen auslautenden Vokal in *stro:- nach der Ablautsreduktion übriggeblieben war (d. h. das Element *6) ein 'weniger leicht vokalisierbares Element' als die eigentlichen Sonanten (**y, *w, *r, *l, *m, *n) gewesen sei: Die Form stR:-nó sei demnach als *stR6-nó mit einem konsonantischen Element *6 zu analysieren, denn nur das ursprüngliche Vorhandensein eines konsonantischen Elementes könne in diesem Fall die vokalische Form des Sonanten *r erklären".

    Lindeman comments: "Wir finden diese Analyse zwingend." Well, I do not find it compelling at all; in fact, I find it very unlikely.

    If the plain stem were *stér6-, and 6 were just an obtuse way of notating the stress-unaccented development from a:, which, in my opinion, would be a, then we could rewrite the stem as *stéra- (from *stéra:-). If it, in turn, is placed before a stress-accented suffix like -*nó, i.e. according to present theory, *stér6- becomes *stR:-. The long syllabic r appears in Greek and Latin both as ra:, which would, of course, not be surprising or remarkable at all if derived from *stéra:-; in fact, the Greek and Latin reflexes would not necessitate the formulation of a long syllabic r at all. But according to the theorists, Old Indian requires the theoreticization of R: because of words like sti:rn[.]á in spite of the fact that we also have stRtá! It is so plainly obvious that, if Indo-European R: occasioned Old Indian i:r in sti:rn[.]á, then we should also find **sti:rtá not stRtá, which is what we do actually find — but the Neo-Grammarians simply invent a new rule to explain the exceptions.

    I firmly believe that anyone who objectively approaches the facts described above will agree that the correct Old Indian response to ster6-nó would be **stRn[.]á and that sti:rn[.]á cannot reasonably be derived from it. If that is so, from what Indo-European form may we derive it?

    The solution is easily at hand. Old Indian words like stári:man, 'spreading out', and Avestan urvaro:-straya, 'cutting down plantings', show that, in addition to *ster6-, another stem-form, *ster6i-, should be reconstructed. Thus, the situation is analogous to that which we found above. It seems that Indo-Iranian preferred a stem-form of root + i for some forms which the other branches combined directly with the root.

    If we assume an Indo-European form of *stera(:)i-, the Old Indian developments are simplified to an instance of metathesis: *stra(:)i-nó -> *stri:-nó -> *sti:r-nó -> Old Indian sti:rn[.]á.

    This, of course, opens a new can of worms. How accurate is the formulation of branch responses to Indo-European long sonant liquids and nasals?

    Not very, I am sad to say. For Indo-European R:, L:, current theory expects Old Indian i:r or u:r, and i:r or u:r. I think it should be unavoidable to anticipate that the correctly to be anticipated response is R; and that i:r or u:r are the results of -ra(:)i and -ra(:)u in stress-unaccented stems. Today, any linguist with normal phonological knowledge would question the development of a syllable liquid or nasal to two divergent forms, for which the conditioning factor has been offered that a preceding labial or labiovelar yields u:(Resonant) while a consonant of other classification yields i:(Resonant), especially in view of the fact that an Indo-European root like *pel6- has pra:tá- and pra:n[.]á-, both 'full', in addition to pu:rtá-, 'full'. Latin has the clearest and most straightforward responses to the "long syllabic liquids and nasals": ra:, la:, na:; and, in fact, it should be noted that it is not at all necessary to assume "syllabic liquids and nasals" to explain the Latin reflexes. But the Greek reflexes are suspect. For N:, Greek na: is proposed, and certainly is reasonable. However, for R: and L:, Greek ro: and lo: are proposed — in spite of the fact we have words like stratós.

    If we are correct in assuming that Old Indian u:r is best explained as representing a sequence of CVra(:)u, then the stem of a form like Old Indian mu:rdhán-, 'head', should be reconstructed as *mela(:)u-; if Greek blo:thrós, 'soaring upwards', is truly related, the o: would be explained as a Greek contraction of a:u, a type of contraction quite frequent in the world's languages. A substantiation of this reasoning is that the word for 'wool', which appears in Old Indian as ú:rn[.]a:, and is reconstructed by Lindeman as *wL:na:. If the Greek reflex were correctly lo:, then we should expect Greek **(w)lo:nos but what we actually find is lênos, with Attic ê representing proto-Greek â (I am using the circumflex accent to represent the Greek tilde). Now, what is extremely disturbing and, at the same time, disappointing, is that an Indo-Europeanist like Lindeman will blithely assert that "Idg. L: = . . . gr. lo:", and cite forms in Old Indian, Latin, Gothic, Old Bulgarian, and Lithuanian for 'wool' but neglect to include Greek lênos which makes Greek lo: seem a bit problematical.

    This short investigation reveals that the traditional reconstruction of schwa indogermanicum is so feebly based that it serves no useful purpose to assume it. The motivation for proposing it can only have been to include an element known from Hebrew as a small step towards linking Indo-European and Semitic, a link to which I subscribe also without feeling the urge to concoct spurious phonemes for Indo-European to support it.

    This essay also questions whether there is any justification for the reconstruction of long syllabic liquids and nasals, and concludes that there may not be any at all.









NOTATIONAL CONVENTIONS





For an explanation of the Proto-Language and Indo-European notational conventions used in these essays, press here.





Afrasian











the latest revision of this document can be found at



HTTP://WWW.GEOCITIES.COM/Athens/Forum/2803/c-AFRASI AN-3_schwa.htm

Patrick C. Ryan * 9115 West 34th Street - Little Rock, AR 72204-4441 * (501)227-9947 PROTO-LANGUAGE@email.msn.com








1. Starting with a theoretical Indo-European *ster6-, or better stéra(:)-, a reduced grade would lead to stra(:)- and then possibly to stR:- so Cuny is wrong to insist that only regarding 6 as consonantal could occasion a long syllabic liquid or nasal in Old Indian.