The aim of this paper is to examine the position of Balochi among the Western Iranian languages by looking at certain features of its historical phonology. These features will be examined for their validity, and the Balochi data will be assessed to determine its relationship to the traditional notion of North-Western Iranian languages.
1.1 Balochi as a North-Western Iranian language
From a historical point of view, Balochi has been regarded as a so-called North- Western Iranian language. This means that with regard to certain linguistic features (called isoglosses), Balochi shares a set of characteristics with e.g. contemporary Kurdish and Zazaki and the Middle Iranian Parthian, whereas Persian, Tajiki and some other languages – called South-Western Iranian – show something different under the same circumstances. The table shows examples of such features from the field of the development of the consonants from prehistoric times to the languages of today. The Balochi data is compared with that of two other North-Western Iranian languages, viz. Zazaki (chosen for its strong North-Western traits), and Kurdish (chosen for its similarities with Balochi, for which see below), and the South-Western Iranian Persian.1
z: B z¯an-, Z zonen-, K zan-
B z¯am¯at, Z zama, K zava
s: B ¯asin, Z asın, K hesin
d: NP d¯an-
h: NP ¯ahan
Such relationships between languages have often been visualised in the form of a
Old Iranian languages
Avestan Median Old Persian
Middle Iranian languages
Sogdian, Saka etc. Parthian Middle Persian
New Iranian languages
Ossetic, Pashto etc. Zazaki, Balochi, New Persian etc.
Figure 1: simplified family tree of Iranian languages3
1.2 Balochi as a southern North-Western Iranian language
It has long been observed, however, that the relationships between Iranian languages are more complex than the family-tree would suggest: In some cases, the Balochi outcome of some Proto-Iranian sounds or combinations of sounds is not the one we might expect in a North-Western Iranian language (judging from the Parthian evidence), but rather the South-Western one as in Persian. The same may be said for Kurdish, but not for Zazaki which usually shows the features expected in a NWIr. language. Examples are:
*´s ˘ u “dog”
hr: Z hire
sp: South Z espe
s: B sai, K sê, NP se
s: B sag, K seg, NP sag
The word for “three” is the only word with PIr. *\r4 which is found in all the fourlanguages, but it is not a good example, since the Balochi and Kurdish words for “three” are perhaps borrowed from Persian (PAUL 1998:1668). Better examples which show that Balochi has s for PIr. *\r are: ¯as “fire” (from *¯a\r-, obl. of *¯atar-, NP ¯azar), dialectal pis “father”, m¯as “mother”, br¯as “brother” (< obl. *pi\r- etc., GEIGER 1891:430, MORGENSTIERNE 1948:257), duks¯ıˇc/dusk¯ıˇc “sister-in-law” ( B s: s¯ah “breath, life” which (GEIGER 1891:430f.) might belong to OInd. ´sv¯asá-, unless it was (together with Pashto s¯ah, MORGENSTIERNE 1927:66) borrowed from some Indic language; s¯ıy- / s¯ıt7 “swell” is a cognate of OInd. ´sváya- according to GEIGER 1891:430 and BAILEY 1979:476a; šiš8 “louse” (NP sopoš, šepeš, Av. spiš) might go back to *´s ˘u- (via *siš, cf. Wakhi šiš, MORGENSTIERNE 1927:69, STEBLIN-KAMENSKIJ 1999:30, 330). Conversely, there is no good evidence for PIr. *´s ˘ u > B sp since the only examples which might be adduced here (B (i)sp¯et “white”9 and asp “horse”, may be Persian loanwords. Thus, one might think that a model which ranks the New Iranian languages according to their specific number of differences from Persian (i.e. the number of shared characteristics with Parthian) is more adequate for the description of relationships among the Western Iranian languages than a family-tree. Such a “scale of northernness” has been suggested by PAUL 1998:170, from which one may conclude that Balochi (and Kurdish even more) is a comparatively “southern” North-Western Iranian language:
PIE PIr. Pth. Gor. A”z. Zaz. Tal. Semn. Casp. CD Bal. Kd. Pers.
*´k/*´g *´s/´z10 s/z s/z s/z s/z s/z s/z s/z s/z s/z s/z h/d
*k( ˘u) pal *ˇc -ž- -ž- -ž- -ˇj- -ž- ˇj,ž -ˇj- ˇj,ž,z -ˇc- -ž- -z-
*g( ˘u)(h) pal *ˇj ž ž ž (y-) ˇj ž ˇj,ž ˇj (z-) ˇj,ž,z ˇj -ž- z
*´k ˘u *´s ˘ u 11 ? sip isb esp asb esp s esb s? s s
*tr/*tl *\r12 hr (ya)r (h)r (h¯ı)r h(*r) (h)r r r s s s
˘u *d ˘ u b b b b b b b b b(?) d d
(OIr.) *rd/*rz *rd/*r´z r(d)/rz ł,r/rz r/rz ¯r/rz r/rz l/l(rz) l/l l/l(rz) rd/rz ł/ł l/l
*s ˘u- *h ˘u- wx w h w h x(u) x(u) x(u), f w x(w) x(u)
*t ˘u *\ ˘u f u u w h h h h(u) h? h h
* ˘i- * ˘i- y- y- y- ˇj- ˇj- ˇj- ˇj- ˇj-(y-) ˇj- ˇj- ˇj-
Table 1: Western Iranian isoglosses (modified from PAUL 1998:170)
– PIr. intervocalic *ˇc (TEDESCO 1921 no. 4): The Balochi product is not ˇi as PAUL 1998:170 assumes, but ˇc (GEIGER 1891:423), e.g. PIr. * ˘uaiˇc- “sieve” > B g¯eˇc- (Z viˇi-, K bêžing (noun), NP b¯ız-); PIr. *ra ˘uˇcah- “light” > B r¯oˇc “day” (Z roˇi/rodz/roz/rož, K rož, NP r¯uz).14 – PIr. *ˇi (TEDESCO 1921 no. 3): e.g. PIr. *ˇian- “woman” > Z ˇiınêk, B ˇian, K žın, NP zan. – PIr. *d ˘ u (TEDESCO 1921 no. 12): Judging from do “two” and dar “door”,15 it has generally been assumed (thus also PAUL 1998:170) that Balochi shows PIr. *d ˘ u > d as do Persian and Kurdish (Zazaki has b: kê-ber/ˇcê-ver “(house-)door”). However, both words can also be explained as Persian loanwords. In this case, ipt¯ı/pit¯ı “other” B zird (Z ze¯rî, NP del) “heart”, PIr. *b ˙r´z- > B burz (Z berz, NP boland) “high”.16 – PIr. *h ˘u- (TEDESCO 1921 no. 17) gives B w (PAUL 1998:170 has v), e.g. war- “eat” (Z wer-, K xwar-, NP xwor-) or h, e.g. h¯ed “sweat” (K xwêh, NP xwei). – PIr. *\ ˘ u (TEDESCO 1921 no. 11): It is not clear if the outcome in Balochi, Kurdish and some other languages is h as PAUL 1998:170 assumes since the only examples are the numerals “four” and “forty” which are identical with (Early New) Pers. ˇcah¯ar/ˇc¯ar, ˇcihil/ˇcil and thus could be loans. Zazaki preserves w in ˇcewres/tsewres “fourty”.
– PIr. * ˘i- (TEDESCO 1921 no. 5): e.g. NP ˇiod¯a “separate”, Z ˇiıya, B ˇiit¯a, K ˇiihê.
2. Significance of the isoglosses
One might therefore ask if it is adequate to call Balochi a North-Western Iranian language at all, or rather, in which respect Balochi is North-Western Iranian. To answer this question, I will examine the isoglosses once more, focussing on two points which have not been taken into account so much in previous discussions about the Western Iranian isoglosses: the chronology of the sound changes in question and the distinction between archaism and innovation.
Some of the sound changes producing the characteristic differences between South- Western and North-Western Iranian languages date back to Old Iranian times, i.e. they already distinguished Old Persian (South-Western Iranian) from Avestan (Non-South- Western Iranian). Other sound changes only came about in Middle Iranian times, i.e. they can be seen in Middle Persian and Parthian, but not in Old Persian and Avestan. It is important to consider the date of a feature since a common characteristic of e.g. Balochi and Middle Persian might tell us something different than one which Balochi shares already with Old Persian.
The approach presented in table 1 above suggests that in the beginning, all so-called North-Western Iranian languages showed some common traits and then one by one came under the influence of Persian, with Kurdish and Balochi being the first affected and Zazaki among the last. It will be seen, however, that this cannot be the case.
Table 2 shows the sound changes (those mentioned above as well as others often cited in works dealing with the grouping of WIr. languages) arranged according to their date of appearance from the oldest (top of the table) to the most recent ones (bottom).17
Proto-Iranian Old Iranian:
Avestan (Non-SW) Old Persian (SW)
*´s s \
(> Middle Persian h)
*´z z d
*pas-ˇca “behind” pas-ˇca pas¯a
*´s ˘ u sp s
*\r \r ç
(> Parthian hr) (> Middle Persian s)
Parthian (NW) Middle Persian (SW)
*ˇc /V_V ž (OP ˇc >) z
*ˇI ž (OP ˇi > ) z
* ˘i- y (OP y >) ˇi
*r´ z rz (OP rd > ) l
*rd rd (OP rd > ) l
*d ˘ u b (OP duv >) d
*h ˘u- wx (OP uv >) xw
*\ ˘ u f (OP \uv >) h
*g, d /V_V g, d (OP g, d >) y
* ˘u- w (OP w >) w, g
*m /V_V m, w (OP m >) m
Z, B, K New Persian
*šm18 m (> K v) (OP šm >) šm
Table 2: isoglosses in chronological order (bold letters: innovations)
– PIr. *pas-ˇca is no. 21 in TEDESCO 1921.
– PIr. *ˇi: It is not excluded that the OP signs ˇia, ˇii already stood for ža, ži (HOFFMANN
– PIr. *g, d (TEDESCO 1921 no. 8 and 6 respectively) disappear between vowels in Zazaki
and Kurdish, e.g. pê “foot” vs. B p¯ad, NP p¯a(y).
– PIr. * ˘u- (TEDESCO 1921 no. 18): e.g. B giˇcin- / NP guz¯ın- “collect”, Z vin-en-, vên-en- / B gind- / K bîn- / NP b¯ın- “see” (pres.), Z va / B gw¯at / K ba / NP b¯ad “wind”.
– PIr. *m: TEDESCO 1921 no. 19.
From these findings, we may conclude that table 1 is a synchronic one, summarizing the data from contemporary Western Iranian languages without reference to their history, and that the “scale of northernness” cannot be taken to hint at historical developments. Contrary to their position in table 1, e.g. the isogloss NWIr. ˇi, ž etc. vs.
SWIr. z (from PIr. *-ˇc, ˇi) is of a more recent date than SWIr. (OP) ç, later s vs. Av. \r, Pth. hr (PIr. *\r), and Persian ˇi- from PIr. * ˘i- cannot be a young phenomenon, but conversely, has to be even older than the voicing of the intervocalic stops as may be seen from Persian loanwords in Armenian, e.g. ˇiatuk “sorcerer” (HÜBSCHMANN 1897:232, cf. NP ˇi¯ad¯u, OInd. y¯atú- “sorcery”).
2.2 Archaism versus innovation
The next interesting observation is that in the majority of the cases, the (Old and Middle) Persian outcome of a certain PIr. sound (or sound combination) can be interpreted as an innovation (printed in bold letters in table 2) whereas Avestan and Parthian preserve the PIr. state of affairs. With regard to the features studied here, the North-Western Iranian languages show much fewer innovations. These are: Pth. b from *d ˘ u (where Persian has an innovation in the other direction, i.e. d), Pth. f from *\ ˘ u (Pers. h), the development of PIr. *h ˘u- to Pth. wx19 (MP xw), occasional w for postvocalic m and the reduction of šm to m. 19 Written wx- might be interpreted as the result of a metathesis *hw- > wh- or as a “device (…) to represent a new sound, viz. devoiced [ ° w]”, cf. English wh- from Old English hw- (MACKENZIE 1967:2629). In the latter case, the Pth. result would be already quite near to that of Zazaki, Balochi
Most of the NWIr. characteristics of contemporary languages are of the type “shared archaism”, i.e. these languages preserve the development already shown by Parthian, e.g. B, Z, K z from PIr. *´z (see 1.1). Shared archaisms tell us that from a historical point of view, Balochi, Zazaki etc. belong to the group of North-Western Iranian languages (i.e. come from a common ancestor). Therefore, this does not tell us anything about the history of a given language after the time the traits in question came about.
The isoglosses which may be regarded as innovations will be examined next, taking into account data from Zazaki and Kurdish again. Common innovations may come about either independently or through contact of the languages in question. For each feature, it has to be decided into which category it most likely belongs. There are four sound changes which are likely to be independent innovations.
One change which is extremely common in languages all over the world is that of word-initial * ˘i- > ˇi which also occurs e.g. in Urdu, several Turkic languages and Low German. Moreover, this change is quite an ancient one in Persian (it must have happened in an old stage of Middle Persian, see p. 55). It is not likely that at that time,
Persian was in contact with the predecessors of all the Western Iranian languages which show ˇi from * ˘i- (even including Talyshi and Zazaki, see table 1), so this change is probably not the result of language contact. The weakening of intervocalic Old Ir. b, g and d to the corresponding fricatives b, g and d is common to Parthian and (some early stage of) Middle Persian (SUNDERMANN 1989:108).20 The further weakening of g and d to y (Persian) or zero (Zazaki, Kurdish) has probably occurred independently: the loss of postvocalic consonants, especially postvocalic voiced fricatives, is such a common phenomenon among the languages of the world that it seems safer to assume (also taking into account the peripheral position of Zazaki) that it has occurred independently in all three languages.
20 This sound change “is common throughout Iranian” (SIMS-WILLIAMS 1996:650). Note that if one considers the change of intervocalic b, d, g to fricatives as shared by all (Western) Middle Iranian languages, this implies that in some stage of Balochi, b, d, g must have reemerged from Middle Ir. b, d, g. In this case, the common assumption that the Balochi stops preserve the Old Iranian state of affairs would be subject to modification.
Something similar applies to the development of postvocalic m to w. This appears to be a Middle Iranian phenomenon since it has been claimed to be an optional development in Parthian already (TEDESCO 1921:208). On the other hand, this change seems to be a rather recent phenomenon in Kurdish and operates as a sound law for all instances of postvocalic m (e.g. nav “name”), including modern loanwords, e.g. ˇiiv¯at (Arabic ˇiam¯a‘at) “assembly” (cf. MACKENZIE 1961:70), and also including the outcome of the reduction šm to m (e.g. çav “eye” (cf. Z çım, B ˇcam(m)) vs. NP ˇcešm). As there is no trace of šm > m in Parthian yet, it seems that Parthian instances of m > w can have nothing to do with the Kurdish sound law and require some other In Balochi, the instances of w for m seem to be the result of a still less ancient process: m > w occurs in the Eastern dialects only, and only sporadically even there, and the Eastern dialects show a number of (quite recent, it seems) sound changes separating them from the other As the change of m > w is a very common one in the languages of the world (cf. e.g. Hungarian név “name” < Proto-Uralic *nime, cf. Finnish nimi), it is quite probable that it has come about independently in several Iranian languages. In Balochi, it might even have happened under the influence of Indic
The reduction of šm to m would represent the only common innovation of North- Western Iranian languages from post-Parthian times. However, it is also shared by a number of other Iranian languages as far away from each other as SW Iran (Bashkardi) and the Afghanistan-Pakistan border (Ormuri, an Eastern Iranian language). Since a phenomenon of language contact between all these is most unlikely for post-Middle Iranian times, it is safer to regard this change as an independent development, too. As in the case of common archaisms, sound changes which are the result of independent innovations do not tell us anything about the question of how the languages are historically related to one another, and which languages have been in contact with one another. Such sound changes are merely due to chance. They should therefore be deleted from the list of isoglosses. 21 In fact, the Parthian examples do not show the presumed change of m > w: angawan (otherwise anˇiaman) “assembly” does not exist, and abg¯aw¯ah belongs to abgaw- “increase” (Desmond Durkin- Meisterernst, personal communication). Parthian h¯aws¯ar “equal, similar” does not contain a variant of h¯am° “same”, but *h¯awat- “similar”, cf. MP h¯awand “like, similar” (SIMS-WILLIAMS 1998a:85). 22 The sound system of the Western and Southern dialects is very similar to that of Common Balochi (the ancestor language of the Balochi dialects), cf. GEIGER 1891:403. 23 Cf. Hindi/Urdu gã¯ v “village” from Old Ind. gra¯ma- (HINÜBER 1986:107).
Innovations which Balochi shares with Persian
All features discussed above are of the type where Persian shows some innovativedevelopment. This also includes those features which Balochi shares with Persian, the ancient developments *´s ˘ u and *\r > s (if this is indeed the Balochi outcome, see 1.2 above) as well as the younger ones *\ ˘ u > h (if so, see 1.2) and * ˘u- > g(w)/b. Kurdish, in addition, shares with Persian the developments of *d ˘ u > d, *h ˘u- > xw, and perhaps also *r´z, *rd > l. Zazaki shows the “true” NWIr. development in all these
Those languages which have been in contact with Persian show a number of similarities with Persian whereas Zazaki which is spoken farther off to the North-West has not been influenced by Persian very much. For the cases mentioned above, we may thus conclude that Balochi has taken over some innovations of Persian (and Kurdish even more).
Innovations which Balochi shares with Parthian
In addition, the evidence suggests that Balochi – in contrast to Kurdish – probably shares two innovations with Parthian: the development of b from *d ˘ u and of w/h from *h ˘u-. This might indicate that in Middle Iranian times, the ancestor of present-day Balochi has still been in contact with Parthian, but the ancestor of present-day Kurdish is likely not to have belonged within this sphere.
2.3 Summary of Western Iranian isoglosses
The table of potentially significant sound changes in WIr. languages thus looks as follows (the single line separates the NWIr. development to the left from the SWIr. One to the right): 24 Zaz. ˇi- < * ˘i- is probably not due to contact with Persian, but rather an independent innovation (see above).
PIr. Av. Parthian Zazaki Balochi Kurdish NP MP OP
*´s s s s s s h h \
*´z z z z z z d d d
*pas-ˇca “behind” pas-ˇca paš ? paš paš pas pas pas¯a
*´s ˘ u sp sp sp s(?) s(?) s s s
*\r \r hr hr s s s s ç
*ˇc /V_V ˇc ž ˇi ˇc ž z z ˇc
*ˇi ˇi ž ˇi ˇi ž z z ˇi
*r´z rz rz rz rz ? l l rd
*rd rd rd ¯r rd ? l l rd
*d ˘ u duu b b b(?) d d d duv
*h ˘u- xw wx w w,h xw xw xw uv
*\ ˘ u \b f w h? h? h h \uv
* ˘u- v w v g(w) b g, b g, w g, d
Table 3: Western Iranian isoglosses in Old, Middle and New Iranian times (in bold: innovations)
The findings from some characteristic sound changes thus show that the relationship among Western Iranian languages can neither be adequately represented by a family tree alone nor in the form of a “scale of northernness”.
The fact that some of the innovations which have occurred in the various stages of Persian have been taken over by Balochi (and some more by Kurdish) indicates that Balochi and Kurdish have been influenced by Persian since Old Iranian times. A small number of innovations that have occurred in Parthian are probably shared by Balochi, but not by Kurdish. This means that in contrast to Kurdish, Balochi might still have been in contact with other North-Western Iranian languages in Middle Iranian times.
This would mean that Balochi has been a North-Western Iranian language not only at the outset but remained so in Middle Iranian times.
Az. Azari OInd. Old Indic
B(al.) Balochi OP Old Persian
Casp. Caspian dialects Pers. Persian
CD Central dialects PIE Proto-Indo-European
Gor. Gorani PIr. Proto-Iranian
Ir. Iranian Pth. Parthian
K(d.) Kurdish Semn. Semnani
MP Middle Persian SW South-Western
NP New Persian Tal. Talyshi
NW North-Western W Western
obl. Oblique case Z(az.) Zazaki
1. A list of abbreviations is at the end of the article. For better comparison, the method of writing Zazaki and Kurdish consonants used here follows the tradition of Iranian studies and not the orthography used in Zazaki and Kurdish publications (e.g. š for ¸s). “Kurdish” as used here denotes Kurmanji.
2. Proto-Iranian (a language reconstructed by linguists) is the language from which all Iranian languages are thought to be descended in a way similar to members of a family from a common ancestor. The isoglosses shown in this table are no. 1 and 2 in Tedesco’s classical work on the internal grouping of the Western Iranian languages (TEDESCO 1921).
3. Avestan is grouped under Eastern Iranian here, although in Old Iranian times, the distinction was rather one between South-Western (i.e. mainly Persian) and the rest of the Iranian languages, so Avestan should more correctly be termed a “Non-South-Western” Iranian language (cf. also SIMSWILLIAMS1996:649f.).
4. This is isogloss no. 10 in TEDESCO 1921.
5. For Kurdish s from *\r, see MACKENZIE 1961:76f.
6. This isogloss is not found in TEDESCO 1921, but cf. e.g. PAUL 1998:166ff.
7. This word is only found in DAMES 1881, HITTU RAM 1881 and ELFENBEIN 1990/II.
8. šiš is only reported by ELFENBEIN 1989:635.
9. Balochi (i)sp¯et might have been borrowed from MP sp¯ed or Early NP saf¯ed. Devoicing of word-final consonants is comparatively common in loanwords in Balochi, cf. KORN 2001:319.
10. For references and examples, see above.
11. For references and examples (and s as the possible outcome in Balochi), see above. It should be kept in mind that the Kurdish outcome is far from certain. The Parthian outcome seems to be sp (as one would expect), cf. asp “horse”, isp¯ed “white”.
12. For references and examples, see above.
13. The entries for Balochi have been modified (see the notes below the table), a column for Proto- Iranian has been added; the symbols used for writing Proto-Indoeuropean (the ancestor language of the Iranian as well as several other branches of languages) have been adjusted to the system used here, otherwise the table is the one shown in PAUL 1998:170.
14. Parthian loanwords in Armenian and Georgian show that the development of intervocalic ˇc > ˇi > ž occurs within Parthian (i.e. early Parthian preserves -ˇc-) and that PIr. *ˇi gives ž in all stages of Parthian (GIPPERT 2000:2). The products of both thus merge in Middle Iranian times.
15. B dw¯azdah “twelve” and digar (etc.) “other” are Persian borrowings in any case as is shown by the -h of the former and the -g- of the latter.
16.According to MACKENZIE 1961:78, it is “not (…) possible to be certain which [i.e., l/ł or the preservation of rd, rz] is the true Kurdish development”.
17. The order in each section of the table hints at assumptions (not to be detailed here) concerning the relative age of the developments, but much of that remains speculative.
18. On the change of PIr. *šm, see below.
19. Written wx- might be interpreted as the result of a metathesis *hw- > wh- or as a “device (…) to represent a new sound, viz. devoiced [ ° w]”, cf. English wh- from Old English hw- (MACKENZIE 1967:2629). In the latter case, the Pth. result would be already quite near to that of Zazaki, Balochi
20. This sound change “is common throughout Iranian” (SIMS-WILLIAMS 1996:650). Note that if one considers the change of intervocalic b, d, g to fricatives as shared by all (Western) Middle Iranian languages, this implies that in some stage of Balochi, b, d, g must have reemerged from Middle Ir. b, d, g. In this case, the common assumption that the Balochi stops preserve the Old Iranian state of affairs would be subject to modification.
21. In fact, the Parthian examples do not show the presumed change of m > w: angawan (otherwise anˇiaman) “assembly” does not exist, and abg¯aw¯ah belongs to abgaw- “increase” (Desmond Durkin- Meisterernst, personal communication). Parthian h¯aws¯ar “equal, similar” does not contain a variant of h¯am° “same”, but *h¯awat- “similar”, cf. MP h¯awand “like, similar” (SIMS-WILLIAMS 1998a:85).
22. The sound system of the Western and Southern dialects is very similar to that of Common Balochi (the ancestor language of the Balochi dialects), cf. GEIGER 1891:403.
23. Cf. Hindi/Urdu gã¯ v “village” from Old Ind. gra¯ma- (HINÜBER 1986:107).
24. Zaz. ˇi- < * ˘i- is probably not due to contact with Persian, but rather an independent innovation (see above).