III. THE LANGUAGE OF THE SARMATIANS
2. Proto-Iranian and Ossetian
If we wish to clarify the problems connected with the language of the Sarmatian tribes of Southern Russia and its relation to Ossetian, we have to bear in mind, first of all, two considerations. We have seen above that, in judging this question, the great majority of scholars, e. g. Miller, Vasmer, Lommel, Kretschmer, Sköld and Abaev, started from the theory of the family-tree of languages. This manifested itself chiefly in the fact that, the earlier the stage of language they examined, the less inclined they became (often flying in the face of practically palpable linguistic facts) to assume even a slight degree of linguistic differentiation. The result was that they regarded the language of the Scythians and Sarmatians as uniform, and considered even the present Ossetian dialectal differentiation to be an entirely new development. Since the family-tree theory has thus exercised a decisive influence on research concerned with Ossetian and the language of the Iranian tribes of South Russia we have to raise the question whether it is right to accept this theory as a basis of our investigations. In order to answer it, we will examine the application of the family-tree theory in some examples taken from linguistic history.
One of the chief aims of comparative linguistics, based on the family-tree
theory, was to try and reconstruct the homogeneous linguistic status or
parent language from which later dialects and languages were to develop.
Says Edgar Sturtevant in "An introduction to linguistic science", 154:
"Comparative grammar reconstructs certain features of the language spoken
by the original, unseparated community, on the basis of corresponding features
of the descendent languages." In order to attain this objective, scholars
used to compare the different languages belonging to the same group or
family of languages, noting their identical features and regarding these
as characteristic of the ancient, homogeneous linguistic status. Thus in
reconstructing the Proto-Iranian linguistic condition which, in its turn,
was preceded by the Aryan linguistic condition, Bartholomae utilized those
correspondences existing between Old Persian and the language of the Avesta
as well as those existing between the language of the Avesta and some modern
Iranian language, chiefly Modern Persian [39]. But the
adequacy of this method is very questionable. Following a critical hint
by J. Schmidt, Kretschmer has pointed out long ago that certain linguistic
phenomena, though present in all separate languages, must not, in every
case, be regarded as characteristic of the fundamental language, while
conversely, it is sometimes only one language that preserves ancient linguistic
traits. [40] But it is not only the linguistics methods
of the family-tree theory that have aroused grave doubts: its historical
assumptions, too, have proved untenable. There is no doubt that one cannot
assume the existence of populous societies possessing a unitary organization
and speaking a homogeneous language in the early periods of history [41]
— though this assumption is implicit in the family-tree theory. There is
an increasing body of evidence, derived especially from archaeological
research, which shows that the idea of homogeneous linguistic communities,
and of corresponding homogeneous peoples, has
39. Grundriß der iranischen Philologie, I, 1, Straßburg 1895—1901, 3.
40. Einleitung in die Geschichte der griechischen Sprache, Göttingen 1895, 7 foll. Recent criticism of the family-tree theory is contained in Bonfante: Language XXIII (1947), 350 where he expounds the neolinguist position with regard to the family-tree theory. Recent pronouncements in favour of the family-tree theory are by Sturtevant: Language XXIII (1947), 376 foll. and Lane: Language XXV (1949), 333 foll.
41. See Altheim's telling remarks in Italien und Rom, Amsterdam—Leipzig 1941, 152 foll. and Literatur und Gesellschaft im ausgehenden Altertum, II, Halle/Saale 1950, 113 foll.
69
to be dropped entirely. [42] But even if we refrain from discussing the whole problem of the family-tree theory, and do not go beyond the reconstruction of the Proto-Iranian linguistic state, the deficiencies of this method are obvious.
We must raise, first of all, the problem of sources. By setting the
two Old Iranian languages (Old Persian and the language of the Avesta)
against the Middle Iranian and Modern Iranian languages, one may easily
create an impression that seems to be in perfect harmony with the family-tree
theory. The Modern Iranian languages are, undoubtedly, much more numerous
than the Middle Iranian ones, while the latter considerably exceed in number
the two Old Iranian languages. This temporal distribution of independent
languages and dialects is apt to rouse, at first sight, the idea of a progressive
linguistic differentiation in the mind of the spectator. One must not forget,
however, that this idea of progressive differentiation is due only to the
scantiness of material. We have a certain amount of data about practically
all the Modern Iranian languages and dialects; of the medieval Iranian
languages (in spite of the splendid discoveries of recent decades) there
exist data of only a few, while of the Old Iranian languages only two are
known to us. We must also bear in mind that there is a qualitative difference
between these data. Those dating from the present age derive in part form
languages or dialects that are not written down, while the languages known
to us from the Middle Ages, or from antiquity, are almost entirely of a
literary or written character. It we take these facts into consideration,
we have to admit that there is absolutely no evidence to show that linguistic
differences, among the tribes and peoples speaking Iranian languages, were
considerably less in antiquity than in the Middle Ages, or at the present
time. Thus it would be an entirely unwarrantable assumption to regard,
for instance, the language of the Old Persian inscriptions as the homogeneous
language of the Persians, taken as a body of people. Herodotus enumerates
in his work ten Persian tribes (I 125) which lived scattered over a wide
area and showed considerable differences in their material culture. [43]
Bearing this in mind, one would certainly hesitate to identify the language
of the Old Persian inscriptions, let us say, with the language of the nomadic
Persian Asagartiya tribe. The same applies also to the Medes. Herodotus
enumerates six different Median tribes (I 101): even if one of these names
denotes a social stratum rather than a tribe, there can be no doubt as
to
42. See Paret, WaG VIII (1942), 53 foll., Kühn, IPEK XV (1941—42), 256 foll. Especially characteristic is Pittioni's statement in Erasmus II (1949), 296: "Die archäologische Forschung der letzten Jahre hat uns eben zum Umlernen gezwungen. Noch vor kurzer Zeit der Meinung verfallen, daß die einzelnen indogermanischen Völker wie Zweige gleichzeitig aus dem Stamme sprießen, wobei die Wurzeln dieses Stammes im norddeutsch-skandinavischen Raum gelegen sein sollen, lernen wir nun immer deutlicher, daß nicht die Filiation uns das Werden der indogermanischen Einzelvölker erschließt, sondern nur die Agglutination oder die Substrattheorie, also die Tatsache, daß von den wichtigen oberpaläolitischen Kulturen aufwärts Schicht auf Schicht gelegt wird, wobei diese über weite Strecken hin gemeinsamen Schichten Verwandschaften und Beziehungen erzeugen, die in ihrer Abfolge Gleichzeitiges und Aufeinanderfolgendes verbinden und damit ein mehr als kompliziertes Bild einer Kultur- und Völkerentfaltung erweisen". — Recently, even the adherents of the family-tree theory have started admitting that the parent language or fundamental language could not have been homogeneous. See e. g. Sturievant's following words "We must admit the existence of dialectic differences within Proto-Indo-European. At present we cannot do very much about such features; but it is important to recognize their existence". (An Introduction to Linguistic Science3, New Haven 1948, 167.) This would mean, of course, giving up the idea of the parent language and the attempts at its reconstruction; so Sturtevant hastens to add: "In theory at least, a period of dialectic differentiation preceded the final separation of the Indo-European languages from the parent stock". Thus he succeeds in finding a formula combining the idea of a parent language with dialectal differentiation. But the only concrete basis of the whole theory is the actual existence of dialectal differentiation.
43. See Christensen, Die Iranier, 236.
70
the tribal divisions of the Medes. [44] Thus, there is no ground whatever for assuming the existence of a homogeneous Median language at the time. On the contrary, there are certain features in the investigations conducted hitherto which lend full support to the view that in the case of both Persians and Medes we have to count, at the very outset of their appearance in history, with a linguistic differentiation that accords with their division into tribes. Already, Geiger hit upon the idea of a dialectal differentiation among the Medes, when asking the question whether the Old Persian word farnah- is not borrowed from one of the Median dialects. [45] The investigations of Andreas, Lentz, Tedesco, and Herzfeld have contributed to the development of this suggestion. Following a hint by Andreas, Lentz has pointed out that the Old Persian farnah- is certainly an old loan-word from Median; but among the present dialects it is only in Sīwandī that we find the correspondence of f- to initial χv-, while in the northern dialects the usual corresponding group of phonemes is vχ-, vh-. Since, according to the testimony of the Old Persian word, the development χv- > f- must have taken place in Median as early as the sixth century B. C., while over the larger part of the linguistic area the initial χv- has been preserved, there can be no doubt that there already existed a considerable dialectal differentiation in Median at this time [46].
Tedesco's investigations concerning the dialectology of the West Iranian
Turfan texts led to the same result. Tedesco has demonstrated that in the
northwestern Turfan texts some phonemes and groups of phonemes have a double
correspondence: thus e. g. intervocalic d (δ)
is usually preserved but in some words it has a corresponding -h-,
etc. Since in the southwestern Turfan texts it is the phoneme y that corresponds
to intervocalic d, this double correspondence may be explained only
by assuming that the language of the northwestern Turfan texts is based,
not on one dialect, but on several northwestern dialects. [47]
This dialectal differentiation must reach back into far antiquity, as is
clearly shown by the juxtaposition of two data: the name
Fraδāta
was transcribed in ancient sources, as early as the second century B. C.,
in the forms Phraates or Phrahates which reflect already
an Iranian form Frahāta;
at the same time, Ptolemy gives the name of Isfāhan
in the form
as late as the second century A. D. [48] Thus in the
northwestern territory dialectal differentiation may be traced back to
the sixth century B. C., if no further, and the same is true also of the
Persian territories. It is again Tedesco's investigations which have demonstrated
that the language of the Old Persian inscriptions could not have been the
direct antecedent of Middle and Modern Persian dialects: 'Altechtpersisch',
i.
e. the Old Iranian antecedent of the south-western Turfan texts, must
have been a different dialect. [49] Accordingly, there
are indubitable linguistic facts indicating that, in the case of both Medes
and Persians, one has to deal with different dialects right at the outset
of their historical career; it is extremely likely that this dialectal
differentiation was connected with a division into tribes.
In this case, however, the reconstruction of the Proto-Iranian linguistic
stage have to be subjected to a thorough revision. According to Bartholomae's
theory, the Aryan Parent Language split up into two essentially homogeneous
languages
44. See Christensen, Die Iranier, 233. H. S. Nyberg, in Die Religionen des alten Iran, Leipzig 1938, 335, regards the Boudioi, too, as a caste of priests; with regard to the other data, however, he himself bears witness to the authenticity of Herodotus' account.
45. Grundriß der iranischen Philologie, l 2, 423.
46. Lentz, ZII IV (19:6), 288. See also Herzfeld, AMI VII (1935), 40 foll.
47. Tedesco, MO XV (1921), 195, 205 foll., 246, 253.
48. Tedesco, MO XV (1921), 185; Herzfeld, AMI VII (1935), 15.
49. Tedesco, MO XV (1921), 248.
71
one of which he simply called 'Proto-Iranian', This 'Proto-Iranian language' was, however, a purely formal linguistic concept, the contents of which were determined by the changes which took place in 'Proto-Iranian' from the time of its separation from the Aryan parent language until its disintegration. On these premises Bartholomae acted quite logically when he utilized, in reconstructing the Proto-Iranian linguistic state, those changes which he found both in Old Persian and in the language of the Avesta, since, according to his theory these common changes must have occurred in Proto-Iranian while changes peculiar to one of them must have taken place in the separate Old Iranian languages. [50] This theory is entirely logical: yet historically — even apart from its unproved and unsubstantiated premises — it is extremely unlikely. As we have pointed out above, only two of the Old Iranian languages supply us with a fair number of linguistic remains; of these, the language of the Avesta has undergone considerable distortion during the process of transmission, so that its value as a source for the history of phonemes is frequently open to doubt; while the language of the Old Persian inscriptions only gives us some insight into the language of a single Persian tribe. It follows that, actually, we have only data about an insignificant proportion of Old Iranian languages or dialects; this circumstance makes the reconstruction of a 'Proto-Iranian language' an ardous and rather hopeless task. There is no evidence whatever to show the changes, common to the Old Persian inscriptions and the language of the Avesta, took place also in the numerous other Old Iranian languages and dialects unknown to us; consequently, the changes determining the 'Proto-Iranian language' necessarily elude our grasp. Similarly, there are no indications whatever to show whether some, or even a considerable part, of the changes peculiar to one language alone, do not go back to Proto-Iranian times.
To take only one example: one of the most definite features of the reconstructed
'Proto-Iranian language' is the change of initial s-
> h
> χv-;
see e. g. Old Indian svarṇara-
~ Avestan χvarənah-.
But it is precisely this word which we find already in Old Persian, in
the form farnah-, as an Old Median loanword. Thus the development
of the Aryan initial group of phonemes s
-
was already different in the dialects of Median, one of the Old Iranian
languages; there is no evidence whatever as to the date when these differences
developed. There is no evidence, either, to show that the form farnah-
developed, through an intermediate form *χvarnah-,
from Aryan *svarnas-: one may easily suppose that in one part of
the Median linguistic territory there was a direct phonemic development
s
-
> f-. This would naturally imply that the phonemic development s
-
> χv- was not characteristic
of the whole Proto-Iranian language, i. e. that 'Proto-Iranian'
was not a homogeneous language, but was divided into different languages
or dialects. This idea leads to the obliteration of boundaries between
Proto-Iranian and Old Iranian.
There is no doubt that, from the angle of the family-tree theory, the
chief distinctive mark of the 'Proto-Iranian language' was precisely its
homogeneity, the uniformity of the changes separating it from Aryan; owing
to the emergence of features peculiar to some languages only, this uniformity
gave place to the diversity of the Old Iranian languages. If the existence
of such a homogeneous state of the language is not capable of demonstration,
there is, in fact, no need to adhere to the concept of a 'Proto-Iranian
language'. We have to point out, too, the essential difference that exists
between the concept of 'Proto-Iranian', on the one hand, and the designations
of 'Old Iranian', 'Middle Iranian', and 'Modern Iranian', on the other.
'Proto-Iranian', together with 'Aryan' and 'Indo-European', is a purely
formal lin-
50. See Grundriß der iranischen Philologie, I 1, 1 foll.
72
guistic concept denoting a homogeneous unit. 'Old Iranian', 'Middle Iranian', and 'Modern Iranian' are, on the other hand, historical concepts which do not admit of a clear linguistic definition. It has never yet occurred to anybody to reconstruct a homogeneous Old Iranian or Middle Iranian language which would possess common distinctive marks: these expressions are used to denote the Iranian languages known to us from different periods of history. Hence, even within one and the same Iranian language, these expressions do not usually denote stages of development admitting of clear linguistic delimitation. It follows that there is no road leading from the historical concept of the 'Old Iranian languages' to the formal linguistic concept of the 'Proto-Iranian language'.
The data supplied by Herodotus about the Persian and Median tribes date
from the fifth century B. C. There is no doubt, however, that the formation
of both the Median and the Persian tribes must be assigned to a considerably
earlier period. Hence it is obvious, too, that the linguistic differences,
observable among the Median and Persian tribes in the sixth and fifth centuries,
may be traced back into earlier times. An examination of the Iranian names
found in the cuneiform sources leads us to the conclusion that a considerable
part of the linguistic differences observable in the sixth century may
be traced back to the ninth. As early as 712 we come across a prince called
Auarparna;
from the beginning of the seventh century we know the names of princes
Šitirparna
and Ēparna:
in these names the element -parna is a transcription of the word
farnah-
which we have discussed above. [51] Hence the double
correspondence of χv- ~ f-
to Aryan s-
goes back to the eighth century. During the reign of Salmanassar there
is mention in 854 of a prince called Kundašpi, about 740 of a prince
bearing the name of Kuštašpi. The corresponding forms of these two
names in other parts of the Old Iranian linguistic territory would be *Vindāspa-
and *Vištāspa-;
hence the phonemic change of initial vi- > gu- ~ ku-
may be traced back, in this case, to the ninth century. [52]
All this combines to show that the differences in the Old Iranian languages
or dialects reach back in fact, into the Proto-Iranian period, i. e.
into the age preceding the historical appearance of the Iranian tribes
[53].
If we wish to continue employing the concept of 'Proto-Iranian' as a historical
designation in linguistics, it is most apposite to our purpose to mean
by it the linguistic facts, languages, dialects, and linguistic condition
of the period preceding the historical emergence of the Medes and Persians
and the foundation of a state by them. The only question is whether the
time limit of this historical period may be clearly defined, and whether
it may be organically connected with the period of the 'Aryan language',
a concept known to us
Additional Notes
To p. 73. It is hardly correct to trace back the names Kustašpi and Kundašpi to Old Iranian *Vištāspa- and *Vindāspa- as Kretschmer and Nyberg did. I now presume the Old Iranian forms *Kuštāspa- and *Kundāspa- as Old Iranian prototypes. |
51. See e. g. Nyberg, Die Religionen des alten Iran, 334; Herzfeld, AMI VII (1935). 28 foll. Herzfeld's contention that the p of the Akkad script stands in these names for vh is without any foundation. First, the example quoted by him — Iranian Gundofarr ~ Indian Guduvhara — illustrates quite a different point (here f is transcribed as vh, not vh as p); secondly, the word farnah- is transcribed even in later cuneiform texts as parna-: *Frādafarnah- = mIp-ra-(a-)du-par-na-’, *Dādafarnah- = mDa-da-par-na-’ (see W. Eilers, Iranische Beamtennamen in der keilschriftlichen Überlieferung, I, Leipzig 1940, 97).
52. See Kretschmer, KZ IV (1928): Nyberg, Die Religionen des alten Iran, 333.
53. This was clearly realized, with regard to Median, by Herzfeld, AMI VII (1035), 23 foll.: "In einem so großen gebiet kann von anfang an nicht nur ein dialekt gesprochen sein: medisch bedeutet eine ganze gruppe. Die assyrischen, babylonischen, elamischen, aramaeischen und griechischen Umschreibungen medischer orts- und personennamen lassen davon allerhand erkennen... Die aufgäbe ist, was da zu erkennen ist und was die heute noch lebenden dialektreste bewahrt haben, mit den großen unterteilen Mediens in verbindung zu bringen, in denen sich uralte stammesunterschiede ausprägen".
73
from linguistics. According to the testimony of archaeology [54] and of historical sources, the migration of the Medes and Persians to the territory of Iran may be assigned to the end of the second millennium, or the beginning of the first millenium B. C. [55] This is also the conjectural date established by Herzfeld for the occurrence of those changes which separate Iranian (or 'Proto-Iranian' in Bartholomae's terminology) from Aryan. [56] Herzfeld's argument is supported, for the time being, by a single linguistic fact. [57] Additional corroboration may be derived from the Aryan names found among the Hurrians. [58] According to the testimony of these names, on the western borders of later Iran, there lived in the 14th century B. C. certain ethnic dements whose language did not yet show those phonemic changes which distinguish the Old Iranian languages from Old Indian. Thus, historically, one may speak of Proto-Iranian languages during the period that extends roughly from the migration of the Iranian tribes into the territory of Iran, to the formation of the Persian state. But this Proto-Iranian period of history cannot be organically ccnrected with the period of the 'Aryan language'. In this case 'Proto-Iranian', precisely like 'Old Indian', would be a historical-geographical concept, while 'Aryan' is a formal linguistic designation. This qualitative difference in the ideas is reflected also in the attempts to determine the Aryan linguistic elements found among the Hurrians. There were some scholars who, on the basis of their phonemic characteristics, regarded them as Old Indian. [59] Others [60], feeling that 'Old Indian' is essentially a historical and geographical concept which can hardly be applied to linguistic remains from the Near East, consider them to be Aryan. [61]
But even if the idea and reconstruction of a homogeneous Proto-Iranian language must be given up, this naturally does not mean giving up, at the same time, the idea of a Proto-Iranian stage in the history of Iranian phonemes, or the reconstruction, in general, of Old Iranian and Proto-Iranian forms. Nevertheless we have to realize that one may reconstruct, with a greater or lesser degree of probability, the Old Iranian or Proto-Iranian forms of only such words that are attested in certain definite, individual languages: the reconstruction of forms pretending to be of universal validity, equally applicable to all Iranian languages, is a hopeless task; therefore, in the course of this essay, phonemic reconstruction will be employed only in the sense outlined above; conjectural Old Iranian or Proto-Iranian forms will mean only forms that may be supposed to have existed in the Old Iranian or Proto-Iranian stage of a particular language, or in general, forms that may have existed in one Old Iranian or Proto-Iranian language.
Thus, in investigating the language of the Iranian tribes in South Russia,
as well as their relation to Ossetian, we have to point out that these
languages lend support in two respects to the criticism of attempts at
reconstructing the 'Proto-
Additional Notes
To p. 74. On the Indo-Aryan linguistic elements in the Ancient
Near East cf. now M. Mayrhofer: Die Indo-Arier im Alten Vorderasien.
Wiesbaden 1966, with almost complete bibliography, A. Kammenhuber: Die
Arier im Vorderen Orient. Heidelberg 1968., E. A. Grantovskiy: |
54. See Ghirshman, Fouilles de Sialk, II. Paris 1939: the results are summed up by Altheim: Saeculum I (1950), 294 foll.
55. See e. g. Herzfeld AMI VIII (1937), 46 foll., AMI IX (1938), 164 foll., Archaeological History of Iran, London 1935, 9 foll.
56. AMI VIII (1937), 46 foll., Altpersische Inschriften. Berlin 1938, 183 foll.
57. This is obviously the reason why Herzfeld's conclusions are regarded as premature by Eilers, Iranische Beamtennamen in der keilschriftlichen Überlieferung, 117.
58. See e. g. Christensen, Die Iranier, 209 foll., Nyberg, Die Religionen des alten Iran, 330 foll.
59. See e. g. A. Götze, Kleinasien: Handbuch der Altertumswissenschaft, III. Abt. I. Teil. III. Bd. 3. Abschn. 1. Lief., München 1933, 59, where fuller bibliography is given.
60. See e. g. Christensen, Die Iranier, 210; Nyberg, Die Religionen des alten Iran, 332. — Herzfeld himself felt that he was 'anticipating' when applying the names 'Old Indian' and 'Indo-Iranian' to the Hurrian Aryans: see Archaeological History of Iran. 9.
61. Cf. also Oldenberg, Die Religion des Veda3–4, Stuttgard—Berlin 1923, 24 foll.
74
Iranian language.' First, there is Bartholomae's suggestion that the
initial group of phonemes s-
in Proto-Iranian lost the element
through an intermediate grade
-,
and was ultimately reduced to the phoneme
s-.; e. g. Avestan saēn
~ Old Indian
yenas,
Avestan sāmahe
'des schwarzen' ~ Old Indian
yāmas,
Ossetian sau ~ Old Indian
iyāvas
[62].
Since the formulation of Bartholomae's theory, the Middle Iranian linguistic
material at our disposal has been considerably enriched and partly supports
Bartholomae's point of view. Thus e. g. in Sogdian we find the form š’w
'black' which seems to justify the intermediate grade
-
conjectured by Bartholomae. Of course, even in this case, the utmost one
is ready to admit is that the first part of the supposed process s
-
>
- > s-
was accomplished in Proto-Iranian, while the second part was a development
in individual languages. But the Iranian names in the inscriptions of South
Russia have established it beyond doubt that Bartholomae's suggestion is
untenable even in this form. Among these names we find the following forms:
Panticapaeum,
Olbia,
Tanais [63]. These names which date from the second
century A. D. reflect the forms syāv,
syuvāγ,
syāvak
[64].
This makes it clear that the language of some of the Iranian tribes in
South Russia retained the initial group of phonemes s
-
until late historical times; actually, even in the late Middle Ages, the
word syāv
was taken over from Ossetian in the form šau by the Balkars [65].
Thus, these Iranian linguistic data from South Russia show clearly, together
with Avestan syāva-
and Modern Persian siyāh,
that even the first part of the phonemic change s
-
>
- > s-
cannot be regarded as going back to Proto-Iranian times.
Another important point in Bartholomae's Proto-Iranian reconstruction
was the thesis that Aryan palatal -
developed into
-
before
, and
that this group of phonemes
-
was simplified to
-;
see e. g. Avestan šavaitē
— Old Indian cyavati, Avestan š
~ Modern Persian šād,
etc [66]. The Sogdian data do not contradict this conjecture;
see. e. g. šw- 'gehen'. In Ossetian, however, there is a very interesting
correspondence to the Proto-Iranian conjectural initial group of phonemes
-:
Avestan šavaitē
~ Ossetian cäuən,
Avestan šāiti-
> Ossetian äncad. Thus in Ossetian we find c- instead
of s to be expected, or more precisely, in Western and Eastern Ossetian
we find c-, while in the Javian dialect of Southern Ossetian, we
find
- or
-
[67].
According to Abaev's suggestion, however, it is not the Old Iranian palatal
affricate that has been preserved in the phoneme
-
of the Javian dialect (actually, such a conjecture was not put forward,
as even in Bartholomae's view the Aryan k- had developed, already in Proto-Iranian,
into
- if followed
by
), but
the development of this phoneme is a secondary phenomenon.
[68]
We will disregard, for the time being, the first part of this suggestion,
viz.
the question of the phoneme corresponding in Old or Proto-Iranian to present
Javian -, and
will discuss the second part. There are several serious objections to regarding
the phoneme
-,
in the Javian dialect, as a secondary phenomenon. First of all, Abaev himself
points out that, according to Yalguzidse's data, deriving from 1802, the
phoneme
- was
still extant at that time in the Javian dialect, though to-day it
Additional Notes
To p. 75. On the Javian dialect cf. also G. Akhvlediani: |
62. Grundriß der iranischen Philologie, I 1, 37.
63. See Vasmer, Die Iranier in Südrußland, 5l foll.
64. See also Schaeder, Iranica, Berlin 1934, 51.
65. See Miller, Die Sprache des Osseten, 4. Cp. also Georgian šav-i < Ossetian *šau.
66. Grundriß der iranischen Philologie, I. 1, 7, 38, Reichelt, Awestisches Elementarbuch, Heidelberg 1909, 45.
67. With regard to South Ossetian
see ,
89.
75
has already been replaced by the palatal spirant -
[69].
If we were to accept Abaev's theory of the Javian
-
being a secondary phenomenon, we would have to regard the palatal spirant
-
as its predecessor, a phoneme which is actually found in most Iranian languages.
In this case, however, phonemic development would run along the line
-
>
- >
-
which does not seem probable at all. Moreover, one must not examine the
Javian dialect by itself, in complete isolation. According to Yalguzidze's
data, around 1800 the phoneme
-
was still pronounced in Javian, instead of the c- in the western
and northern dialects. There is no doubt, however, that in the latter dialects,
too, the phoneme c- had developed from an earlier
-.
This is shown by the fact that in the place-names of the territory once
inhabited by Ossetes, present Western and Eastern Ossetian c- is represented
by
-; see e.
g. šauf
ik
= Ossetian sau-äfcäg 'Schwarzer Paß' [70]. According
to Munkácsi's notes, this latter word has the following phonemic forms in the
various Ossetian dialects: Eastern Ossetian äfcäg, Southern Ossetian əf
äg, əf
äk,
Western Ossetian afcäk 'zur Sommerweide geeignete, bergige
Gegend' [71].
Since, according to the testimony of place-names, the Western and Eastern
Ossetian c- goes back to an earlier -,
and since in Southern Ossetian they still pronounced
-,
instead of the present
,
it is clear that the Javian
cannot
be secondary, unless the phoneme
,
which had originally existed in the phonemic system of Ossetian, was a
secondary development in the rest of the Ossetian dialects, too, i.
e. in the whole of the Ossetian linguistic territory. But such a supposition
lacks any foundation, because in Ossetian the palatal spirant
-,
whether old or of secondary origin, has been equally replaced by the dental
spirant s-. Consequently, if the predecessor of the present Eastern
and Western Ossetian c- ~ Javian
-
>
- had been
-
or
-
(whether in Proto- or Old Iranian, or at any other period) we would find
to-day the phoneme s- in its place. Thus we have to regard it as
certain that the predecessor of present Ossetian c- ~
-
>
- was
-
both in Old Iranian and in Proto-Iranian. This fact has a double consequence
for the reconstruction of Proto-Iranian. First, it is impossible to prove
that the group of phonemes
-
has already developed in Proto-Iranian into
-.
Secondly, the thesis that the Aryan palatal
-,
if followed by
,
developed into
-,
in Proto-Iranian, is also refuted. Ossetian offers clear testimony to the
effect that the Aryan palatal
might develop into
-
in some Proto-Iranian languages, even if followed by
.
Additional Notes
To p. 76. The problem of Ossetian c- < čy- was again treated by E. Benveniste. Études sur la langue ossète. 22 foll, with similar results. He omits, however, to refer to my above analysis. |
70. See ,
III, 8, Die Sprache der Osseten, 5.
[Previous] [Next]
[Back to Index]