plik

Iran and the Caucasus 15 (2011) 169-205



Koninklijke Brill NV, Leiden, 2011

DOI: 10.1163/157338411X12870596615557

Selected Pashto Problems II.

Historical Phonology 1: On Vocalism and

Etyma

Johnny Cheung

Leiden University

Abstract

This study presents a new, systematic treatment of the Pashto continuations of the
Old Iranian vocalism. The analysis is accompanied by an assessment of the etymo-
logy of the Pashto forms mentioned in George Morgenstierne’s Etymological Voca-
bulary of Pashto (EVP) and its postumously revised New Etymological Vocabulary of
Pashto (NEVP). This contribution has also implications for our understanding of the
historical morphology, notably the case system, of Pashto.

Keywords

Pashto, Historical Phonology, Iranian Etymology

NTRODUCTION

One of the greatest iranologists, Georg Morgenstierne, has contributed,
through  his  numerous  articles  and  publications,  much  to  our  under-
standing  of  the  history  of  the  Pashto  language  and,  indeed,  of  the
Pashto  speakers  themselves.  His  Etymological  Vocabulary  of  Pashto  (EVP)
and the New Etymological Vocabulary of Pashto (NEVP), reworked by three
other eminent iranisants, remain our indispensable handbooks for any
research  on  Pashto  history.  His  contributions  have  remained  basically
unchallenged,  cf.  Skjærvø,  CLI:  384-410;  Grjunberg/Ėdel’man,  Osn.:  44-
153.  While  I  was  working  on  Ossetic  and  most  recently,  on  the  Pashto
accent,  I  discovered  that  several  of  his  assumptions  on  Pashto  I  came
across by accident, may be rather tenuous, either because his followed
approach  to  phonology  or  etymology  was  inconsistent,  or  simply,  be-

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

170

cause no explanation at all was offered to many of the exceptional in-
stances or their distribution.

Some of his statements have also made their impact immediately

outside  Pashto  scholarship.  His  assumption  of  the  archaic  character  of
carmə́n f. ʻskin, hideʼ and lamə́n f. ʻhem, border, skirtʼ is a point in case.
For  these  two  forms  Morgenstierne  postulated  an  original  pl.  neuter
ending  *-ani,  which  was  subsequently  accepted  as  fact  by  the  great
Dutch Indo-Europeanist and Sanskritist, F. B. J. Kuiper (Kuiper 1978). As
for  the  latter  example,  the  exact  origin  and  morphological  analysis  of
this lamə́n is uncertain. Although this form is well attested in the later
Iranian dialects,

the stem may contain almost any of the Ir. “dā-“ roots

‘to place, put’, ‘to bind’, ‘to distribute, divide’ (from *dā- ‘to divide’ ac-
cording to Rastorgueva/Ėdel’man 2: 438 ff.), or “dam-” forms ‘to build’,
‘dwelling’, each of which still requires some semantic juggling.

Also

carmə́n is more likely a recent formation, with carm- from Pers. čarm-.

It is for this reason that, first, we may want to take another look at

Pashto  phonology  in  a  historical  perspective,  also  in  light  of  recent
studies on the other East Iranian languages. We need to get the phono-
logical  developments  on  a  more  sound  footing,  which  should  also  in-
clude an assessment of the reliability of the quoted examples and their
etymology.

One of the intricate problems of Pashto phonology is the treatment

of the old vocalism that has been affected by stress or the surrounding

For instance, the frequent appearance of x̌ for Persian borrowings with š, e.g.

pox̌awǝ́l ʻto wear, dress, clotheʼ (Pers. pōšīdan), x̌ād ʻjoyous, merryʼ (Pers. šād), x̌ār
ʻtownʼ (Pers. šahr), x̌kār ʻchaseʼ (Pers. šikār).

Khw. δʼmny, MP, NP dāman, Kurd. dāw f. / dāwēn ʻid.ʼ (Cabolov 2001: 299 ff.), Siv.

dāmen ‘down’, dāmene ‘underskirt’ (Lecocq 1979: 174a).

The formally similar Vedic form dhā́man- ntr., pl. dhā́māni, is of little help here.

Gondaʼs  exhaustive  philological  analysis  of  this  formation  (Gonda  1966)  seems  to
point  to  a  meaning  ʻabode,  seat  of  divine  power,  essence  or  lightʼ.  The  formation
dhā́man- is usually considered to contain the root dhā- ʻto put, placeʼ (cf. Mayrhofer,
KEWAia  1:  783  ff.).  If  the  Iranian  forms  were,  indeed,  cognate,  they  might  reflect  a
locative *dā́mani.

One has to wonder why the plural would have been generalised in the first

place for carmə́n, cf. lemə́ pl. tant. ʻeyesʼ (from an originally ntr. n-stem *daiman-, Av.
daēman-).  Also  the  stress  on  °ə́n  has  not  been  explained  either;  if  carmə́n  was,  in-
deed,  an  archaic  formation,  surely,  we  would  expect  initial  stressing,  in  line  with
notably num ‘name’ < ntr. nom.-acc. *nā́ma, cf. Skt. nā́ma. I would rather analyse car-
mə́n differently, considering it a formation with the suffix -mə́n. This relational (ad-
jectival)  suffix  is  quite  productive  and  is  always  stressed  (Grjunberg/Ėdel’man,  l.c.:
149): e.g. barxamə́n ‘having a part, partial’ (bárxa f. ‘part’), rimə́n ‘purulent, unclean’
(rim ‘pus’), sobmə́n ‘victorious, conquering’ (sóba f. ‘victory’).

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

171

consonants.  The  present  study  is  primarily  dedicated  to  the  changes
that have taken place in the vocalism of the Pashto forms. The outline
of  the  historical  phonology  of  Pashto,  given  by  Skjærvø,  CLI:  398  ff.,  is
the starting point of the present study.

TRESS

One of the obvious effects of the accent is either the “strengthening” of
the vowel, when stressed, or the weakening and even total disappear-
ance of the vowel, when unstressed.

The effect on these short vowels seems to have taken place after the

Pashto Stress Rule (here abbreviated as PSR), which I postulated most re-
cently (Cheung 2010

). This rule entails the shift of the accent from a

light syllable (i.e. containing a short vowel in open position) to the pre-
ceding or following heavy syllable (i.e. containing a long vowel, or being
closed

), which is most noticeable in disyllabic forms. The accent in

these  disyllabic  forms  would  not  shift  if  both  syllables  were  of  equal
weight  (either  “light”  or  “heavy”).  In  trisyllabic  forms  the  accent
tended to be on the middle syllable, provided that the stressing was not
on a heavy syllable already.

1. >

(

)

ā in closed and also stressed position

- lā́rγa f. ʻdelayʼ < *lārγā́ < f. *dar(H)gā́- (< PIIr./Pre-Ir. *dṛHg

á-, f. *dṛHg

ā́-),

Old Av. darǝga-, Parth. drg, Skt. dīrghá- ʻlong (also of time)ʼ.
- lās m. ‘hand’ < *d(z)ásta-, MP, NP dast, Av. zasta-, Skt. hásta- ‘id.’, etc.
- γā́ṛa f. ʻneckʼ < *gár

ā-, Sogd. γrδʼkh, Khw. γrδk, MP (Man.) grdn, NP gar-

dan ʻid.ʼ.
- mā́lga f. ʻsaltʼ < *namáδgā- (PSR) < *námadkā- or *namádkā-, Sogd. nmʼδkh,
Khw. nmθk, M. namalgo, Parth. nmydk, MP (Man.) nmyhk, NP namak ʻid.ʼ.
- māt m. ʻbrokenʼ < *mášta- (PSR) < *maštá-, past ptc. Orm. maṣt-, Parth.
ʼmšt ʻbrokenʼ.

I am also taking the opportunity here to correct a number of erroneous as-

sumptions encountered in that article.

A syllable generally remains open before *Cy in prehistoric Pashto, both con-

sonants would be part of the adjacent syllable. In the case of tyš ʻemptyʼ we have to
reconstruct a preform *tussyá-, which is supported by the Ossetic cognate tyssæg ʻid.ʼ
(Cheung 2000: 67 ff.), i.e. the initial syllable is closed and therefore heavy, triggering
the Pashto stress shift (PSR).

Whether a syllable becomes open or closed obviously depends on the develop-

ment of specific consonant groups that would have been simplified at different
stages in the history of Pashto. This will be discussed in a forthcoming article, Pashto
Problems IV.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

172

- skām  <  skámba-  (PSR)  <  *skambá-,  Av.  fraskǝmba-  m.  ʻfront  hallʼ,  Skt.
skambhá- m. ‘prop, pillar’.
- wā́wra f. ʻsnowʼ < f. *wáfrā-, Sogd.  wfrʼ f., Yi.  wárfo f., Kurd. berf f., Zaz.
vewr f., NP barf ʻid.ʼ.
- zā́ma, (Eastern dialects) žā́ma f. ʻjaw; molar toothʼ < *zámbā-, Khot. ysi-
mä ʻteethʼ, Skt. jámbha-.
- zā́ṇa-, Waz. zōṇyē, Mahs. zū́ṇiyē f., zā́ṇay m. ʻcrane, storkʼ < *zarnā-, Khw.
znwk m., Oss. zærnyg/zærnug ‘id.’ (cf. Gr. géranos ‘crane’).

This rule also gives us a clue of the origin of plār ‘father’ (Av. pitar-, OP
nom. sg. pitā, Sogd. ’ptr-, MP, NP pidar, Skt. pitár- ‘id.’, etc.). Rather than
deriving  it  from  the  acc.  ending  *pitáram  (Av.  pitarǝm)  (cf.  Morgen-
stierne 1942: 95), the alternative solution, from the vocative *pítar, sug-
gested  by  Skjærvø,  CLI:  406,  is  more  preferable:  *pítar  >  *pidár  (PSR)  >
plār,  Wan.  p(i)yār.

The acc. ending *pitáram would have resulted in

Pashto †plar (Wan. †p(i)yar), with short a.

In unstressed, closed position *a has remained a in Pashto, which ap-

plies to saṛáy m. ʻmanʼ < *sardáka- ʻfellow ?; the strong one ?ʼ, cf. Av. sa-
rǝidiia- m. ʻfellow ?, rival, challenger ?ʼ, Skt. bāhu-śardhín- ʻhaving strong
armsʼ, śárdha- m. ʻhost, troopʼ, cf. EWaia II: 619 f.

The form γārmə́ ʻheatʼ is also remarkable for its long ā. The stem

vowel of γārmə́ may have been imported from sāṛə́, both forms are or-
iginally masculine plural adjectives in Pashto.

The stem vowel in the case of čāṛə́, čāṛá f. ‘large knife, dagger’ (*kart-

yā-, Khw. krc f., Kurd. kēr f. ‘knife’, MP, NP kārd ‘knife’, etc.) is odd, being
long in the singular but short in the plural, čaṛé. A plausible explanation
is currently wanting: the long ā in the sg. is perhaps due to the al-
literation with the related form čāqú, čākú ‘penknife’ ?

A kind of vowel alternation can be observed in the forms ās, as m., pl.

asúna, āsā́n ʻhorseʼ (< *áswa-, Av. aspa-, Sogd. ʼsp, Khot. aśśa-, etc.) and ás-
pa, (Bellew: 4a) ā́spa f. ʻmareʼ (< f. *áswā-, Av. aspā- f., Skt. áśvā- f.). Evi-
dently, the vocalism of ās / as and áspa / ā́spa shows mutual contamina-
tion. The regular forms cannot be established with certainty.

Consequently, the other kinship terms may also derive from the vocative end-

ing: mor ‘mother’ < *mā́lr (loss of unstressed *a) < *mā́tar, wror < *wrolr < *brā́tar. In
the case of lur ‘daughter’ any other ending than the vocative seems phonologically
unlikely: *dúxtar (or *dúgdar) > *lúgdr > lur. Accusative *duxtáram/ dugdáram (cf. Skt.
duhitáram) would have yielded Pashto †l(ǝ)wár, whereas the other dominant ending,
the genitive *duxθráh would have become †lǝrə́. Of course, the terms that indicate a
more distant relationship do not necessarily go back to the vocative.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

173

2. > *ā́ > ó before bilabial *u,

(

)

w in closed position:

- owə́ ʻsevenʼ < *(h)ā́wda (PSR) + -ə́ < < *haftá , Av. hapta, Sogd. ʼβd, Oss. avd,
MP, NP haft, Skt. saptá ʻid.ʼ, etc.
- óx̌a, úx̌ka f. ʻtearʼ < f. or pl. n. *ásrukā(-), Khot. āṣka-

, Abyāne’i asl f., NP

ašk, Av. asrū, Skt. áśru- n. ʻid.ʼ.
- pox  m.  (f.  paxá),  pl.  pāxə́  ʻripe,  cookedʼ  <  *pā́xwa-  (PSR)  <  *paxwá-,  Av.
Khot. paha-, Oss. fyx/ funx, Skt. pakvá- ʻid.ʼ.
- tod m. (f. tawdá) ʻhotʼ < *tā́wda (PSR) < past ptc. *taftá-, Av. tafta-, Khot.
ttauda-, Skt. taptá- ʻheatedʼ, etc.

This rule does not apply if there is already another w present, i.e. wā́wra
ʻsnowʼ:  a  very  labial  form  **wówra  may  have  been  difficult  to  pro-
nounce.

3. > ó, in closed position, featuring the adjectives:
- corb, Waz. cörb m. (pl. cārbə́, f. carbá) ʻfatʼ < *cā́rba (PSR) < *čarpá-, Khot.
tcārba-

, NP čarb ʻid.ʼ.

- koṇ, kuṇ (f. kaṇá-) ʻdeafʼ < *kā́rna- (PSR) < *karná-, Av. karǝna-, Khw. kn
ʻid.ʼ, Wakhi kĭr ʻcrop-eared (sheep)ʼ.
- soṛ m. (f. saṛá), pl. sāṛə́ ʻcoldʼ < *sā́rda- (PSR) < PIr. *sar(H)tá- (< PIIr./Pre-
Ir. *śṛHtá-), Av. sarǝta-, Wakhi sĭr, Oss. sald, MP, NP sard, Kurd. sār ʻcoldʼ.
- xoǧ, f. xwaǧá ʻsweet, pleasant, dearʼ < *xwā́rza- (PSR), f. *xwarzā́ < *hwar-
zá-, f. *hwarzā́-, Av. x

arǝz°-, Khw. xž, Oss. xorz/xwarz, NP xuš ʻgood,

agreableʼ.
- zoṛ m. (f. zaṛá), pl. zāṛə́ ʻoldʼ < *zā́rda- (PSR) < PIr. *zar(H)tá- (< PIIr./Pre-
Ir. *źṛHtá-), Av. zarǝta- ʻoldʼ, NP zāl ʻan old man, womanʼ (borrowed).

This category consists solely of adjectives, which may indicates that the
outcome o in the masculine direct case ending does not stem from a
regular sound development. This is also shown by the oblique singular
and plural of the masculine forms, which have long ā. This suggests that
also the dir. sg. should have originally contain this ā as well. This secon-
dary “ablaut” pattern

may have risen on the basis of other adjectival

forms, where o has developed regularly, e.g. in cat. 2., and in adjectives
with original *ā.

Both Khot. forms, tcārba- and āṣka-, have a secondarily lengthened *a > ā before

certain consonant clusters, thus resulting in a closed syllable (cf. Gercenberg 1965:
56; Emmerick, CLI: 211).

This model has also affected the form noǧ, f. naǧá ʻpure, unmixedʼ, which

seems to be an early loanword from Pers. (MP / Parth. ?) nōš ʻsweet, pleasant; water
of immortalityʼ.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

174

4. > a before n, in closed position:
- wand m. ʻdam, dyke, bundʼ < *bandá-, Av. baṇda-, MP, NP band, Skt. ban-
dhá- m. ʻchain, bondʼ (derivational formations, e.g. wandáy m. ʻropeʼ, wan-
danáy m. ʻgrass band of a sheafʼ).
- wandə́l : wand- ʻto scold, insultʼ < *wi-wánda-, cf. Av. vaṇd-, MP (Man.)
wnd- ʻpraiseʼ, Khot. vąn- ʻto honourʼ, Skt. vand

ʻto praise, honourʼ (deri-

vational formations, e.g. wandə́na ʻscoldingʼ

- cang ʻside, flankʼ, cāng m ʻwingʼ, f. cā́nga, Waz. conga ʻbranch, twigʼ
< *čánga-, Oss. cong ʻarm, twigʼ, NP čang ʻclawʼ, kang ʻwing, branchʼ.
- γanə́m, Wan. γandə́m m. pl. ʻwheatʼ < *gantúma-, Av. gaṇtuma-, MP, NP
gandum, Sogd. γntm ʻid.ʼ, etc.

The vowel generally remains short. Two other forms that have been
considered inherited by Morgenstierne, may be explained differently.

The verb γandə́l

: γand-, γānd- ʻto blame, condamn, criticiseʼ

(γan-

də́na  f.  ‘blame,  probation,  condamnation’)  has  been  connected  to  NP
gand  ʻstenchʼ,  Bal.  gandag  ʻbadʼ  (Av.  gaṇti-  f.  ‘bad  smell’,  OP  gasta-  ‘evil,
repugnant’, etc.). The connection is awkward, especially considering the
odd,  presumed  shift  in  meaning,  despite  Morgenstierne’s  insistence
(NEVP:  31).  Perhaps  a  better  etymology  is  a  connection  with  *ǰan-/gan-
‘to slay, strike’, cf.  Oss.  qæn/ǧænæ ‘wound, fracture; shortcoming; (also
in Dig.) guilt, transgression’, Skt. ghaná- (m.) ‘slaying’ (Av. ǰan- ‘to slay’,
etc.).  Alternatively,  it  may  be  an  expressive  formation,  cf.  Sariq.  γind
‘mumble,  mutter’,  Khufi  γinj  ‘brief  neighing  of  horses’  (Morgenstierne,
EVS:  36),  Wa.  γinḍ-γónḍ  car  ‘to  slur,  hesitate’,  γóṇd  ‘snuffling’  (Steblin-
Kamenskij 1999: 182). As for the pres. stem γānd-, the long -ā- may have
been imported from semantically similar verbs, notably traṭə́l ‘to reject,
punish’, pres. trāṭ- (traṭə́na f. ‘telling-off, rebuke; banishment’)

The form spā́nda (also ispánd) f. ʻwild rueʼ is more likely borrowed

(cf. Steblin-Kamenskij 1999: 314 f.), than inherited (*spantā, NEVP: 75).

NEVP: 88, s.v.

wand- also mentions wand (from Aslanov 1966: 956b) ʻreason, ar-

gumentʼ (and the verb wandaw- ʻto prove, give reasonsʼ), but this word is evidently a
loan translation of Pers. band, whose wide range of meanings includes ʻidea, agree-
mentʼ.

Also γəndə́l ? (cf. Aslanov 1966: 611).

The meaning ‘to dislike’ is only found in older dictionaries compiled by Raver-

ty and Bellews, perhaps due to the confusion with γāndə́ ‘disagreable; loathsome’,
which is rather a Persian loanword (gandah ‘rotten, stinking’), contaminated with
the Arabic form γā’ít ‘faeces’.

Cf. Rosh. sepān, Wa. spandr, Yazgh. spandān, MP spandān, NP ispand, sipand, Bal.

spantā́n ʻid.ʼ. Possibly from Middle Persian / Parthian, cf. Arm. LW spand ‘rue’, MP
spandān ‘mustard seed’.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

175

5. > –i-, before n, in closed position, with i-umlaut (q.v.):
- pinjə́  ʻfiveʼ  <  *pánča  +  pl.  *-ah,  Av.  paṇca,  Khot.  paṃjsa,  Sogd.  pnc,  Oss.
fonj, Parth., MP, NP panč ‘id.’, etc.
- skindá f. ʻsplintʼ < skandyā́- ? (NEVP: 73, with initial stress skínda: wrong
?), Av. skǝṇda- ʻfractureʼ.
- wínja- f. ʻslave-girlʼ < *bánda-čī- f. of *bándaka-, OP bandaka-, MP bandag,
Khw. βndyk, NP bandah ‘servant, subject’.

6. > ǝ before n, in open stressed position, with i-umlaut (q.v.):
- ǧdǝn m. (pl.), Wan. eždə́n ʻmilletʼ < *harzana-, NP arzan, Yi. yurzun, Khot.
āʼysaṃ ʻid.ʼ.
- ǰǝn f. ʻyoung girlʼ < *kanin-, Av. kainīn-, Skt. kanyā̀- ʻyoung womanʼ, ka-
nī́na- ʻyoungʼ.
- plǝn, (Waz.) plan ʻbroad, wideʼ < *paθána- (PSR), Av. paθana-, Oss. fætæn,
NP pahn ʻid.ʼ.
- stǝn f., Bang stana, Wan. sǝnǰə́n ʻneedleʼ < *sučánya-, Khot. suṃjsañu, Sh.
sij,  Oss.  suʒin/suʒinæ,  MP,  NP  sōzan,  Bal.  sōčin,  sūčin,  etc.,  cf.  Skt.  sūcī́-  f.
‘id.’.
- wə́na, wúna f. ʻtreeʼ < *wanā- f., Av. vanā-, Sog. wn-, MP wan ʻid.ʼ, Sh. wān
f. ʻwillowʼ; also in compounds °wan m. < ntr. *wána-, Skt. vána- ntr. ʻtree,
wood(s)ʼ.
- ž(ǝ)n- ʻto chop up, minceʼ < *ǰan-, MP, NP zan-/zadan ʻto strike, beatʼ.
- zə́na, Waz. zan(i)ye, Wan. zení f. ʻchinʼ < f. *zánu- + f. *-ā or *-yā- (?), Av.
du. zanauua, Khot. zanuva, Sh. zingůn, Wakhi zǝngúl, Parth. znx, NP zanax,
Skt. hánu- f. ʻchinʼ.

This also includes (old) borrowings:
- from Persian: dux̌mə́n, dux̌mán ‘enemy’ (Pers. dušman), qahrmə́n ‘cham-
pion, hero’ (Pers. qahremān, ultim. < Turkic).
- from Greek (one single example ?): mečə́n f. ‘hand-mill, quern’ (ultim.
mēkanḗ).

The origin of several forms is somewhat uncertain:
- čə́na  f.  ʻthe  outside  of  a  house  wallʼ  <  borrowing  ?,  cf.  Pers.  čīnah  ʻa
course of bricks or stone in a wallʼ, Munj. čin- ʻto build a wallʼ (NEVP: 20:
from *kanī- ?).
- ǧmǝnj, (Wan.) wženz, wženj, uženz, žmenz < *pašána-čī ?

The preform *fšan-čī- suggested in EVP: 106 (NEVP: 29; also Skjærvø, CLI:
404)  is  hardly  satisfactory,  without  any  morphological  analysis  being
provided. It does not seem to have exact correspondences within Indo-
Iranian, only somewhat similar forms have been cited: M. šfīn (Morgen-

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

176

stierne,  IIFL  II:  250b),  Sangl.  afṣ̌ūn,  Ishk.  šofūn,  NP  šānah.  Perhaps,  the
Pashto  form  (and  its  variants)  goes  back  to  the  *ana-formation  of  the
root  *paš  ‘to  comb’  (Oss.  fasyn/fasun,  Yaghn.  nĭpóš,  Yi.  nuvā́š  ‘to  comb’,
Wa. nəbə́sn ‘comb’, Cheung, EDIV: 299

The centralisation of the vowel *a > ə́ / ə, especially in the final sylla-

ble is almost without exception. This development suggests that it must
be  due  to  the  result  of  some  sort  of  i-  and/or  u-epenthesis/umlaut.  It
can  be  noted  that  normally  old  *u,  *ū,  *i,  *ī  would  have  become  ə.  We
can  surmise  that  prior  to  the  distorting  effect  of  the  strong  stress,
Pashto had the (thematic) case endings, nom. *i (< PIr. *-ah), gen. *-iya (<
PIr. *-ahya), and acc. *-u (< PIr. *-am), a development that is similar to
other  East  Iranian  languages  (Sogd.,  Khot.,  Oss.).  The  corresponding
feminine endings would have been nom.

(

)

-a (< PIr. *-ā) and gen. *-(a)ya

or -(a)yi (< PIr. *-āyāh), *-o (< PIr. *- ām)

. Of course, it is also possible

that the feminine has simply adopted (most of) the masculine endings.
Regardless what has taken place in the feminine, we no longer need to
postulate the presence of an ad hoc relational suffix *-ya / -yā.

Forms such as plən /plan ‘wide’, dux̌mə́n/dux̌mán ‘enemy’ can be ac-

counted  for  if  we  assume  that  that  the  variant  °an  (often  preserved  in
the  dialects)  goes  back  to  the  acc.  *-u  (<  *-am),  whereas  the  more  ubi-
quitous °ən derives from the nom. and/or gen. On the effect of i- and/or
u-epenthesis/umlaut, see further down below.

7. > a in open, stressed position:
- γar m. ʻmountainʼ < *gári-, cf. Av. gairi- m., khot. gara-, ggari-, Sogd. γr-,
Skt. girí- m. ‘mountain, hill’.
- las ʻtenʼ < *dása, Av. dasa, Khot., dasau, Sh. δīs, Yi. los, Wakhi δas, MP, NP
dah, Skt. dáśa ʻid.ʼ, etc. (cf. lās < *dásta-).
- xar  m.,  pl.  xrə  ‘ass,  donkey’  <  *xára-,  Av.  xara-,  Khot.  khara-,  Sogd.  xr-,
Parth., MP, NP xar ‘id.’, etc.

8. *a > ø in open, unstressed position:
- bə́n f. ‘co-wife’ < *hapáθnī-, Av. hapaθnī-, Skt. sapátni- f. ‘id.’.
- swa f. ʻhoofʼ < *safā́-, Av. safa- m., Skt. śaphá- m. ʻid.ʼ.
- zǝ ʻIʼ < *azám, Av. azǝm, Khot. aysu, Sogd. zw, Kurd. az, Skt. ahám ʻid.ʼ
passim.

The meanings, ‘to shave’ and ‘to shear’ assigned to the (only East Iranian ?)

root *paš in this work are evidently erroneous, as all the attested Iranian cor-
respondences point to ‘to comb’.

The change to *o is assumed on the basis of the Khot. evidence, but this would

no doubt be replaced later on in Pashto, as it would coincide with the oblique pl. –o.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

177

*Ā

1. > o in stressed position:
- bob  ʻpure,  unmixed,  unalloyedʼ  <  *abā́ba-  (PSR)  <  *apā́pa-  or  *apāpá-
(*apa-āpa- ʻdevoid of waterʼ), cf. Bact. ababgo ʻwaterlessʼ.
- calór ʻfourʼ < *čaθwā́ra-, Khot. tcahora-, Sogd. ctβ’r, Oss. cyppar / cuppar,
Parth. cf’r, MP, NP čahār, Skt. catvā́raḥ, etc.
- dwólas ‘twelve’ < *dwā́dasa, Khot. dvāsä, Parth. dw’dys, NP dawāzdah, Av.
duuadasa, Skt. dvā́daśa.
- lor  m.  ‘sickle’  <  ntr.  *dā́θram,  Wa.  δĭtr,  dǝtr,  MP,  NP  dās,  Skt.  n.  dā́tram
‘id.’, etc.
- óspǝna, óspina f. ʻironʼ < *ā́swanyā- (initial stressing is original ?), Khw.
ʼspny,  Sh.  sipin,  MP,  NP  āhan  ʻid.ʼ  (also  zerograde  *āsuna-  in  MP  (Man.)
’’hwn, Parth. ’’swn, Kurd. āsin m.).
- por m. ʻloan, debtʼ < *pā́ra- (PSR) < *pārá- or *pā́ra-, Av. pāra- ʻguilt, debtʼ
- póre ʻacross, beyondʼ < *pā́r° (PSR) < *pārá-, Av. pāra-, Khot. pāra- ʻbor-
derʼ, Skt. pārá- ʻgoing beyond; opposite bankʼ.
- stóray  m.  ʻstarʼ  <  *stā́raka-,  Khot.  stāraa-,  Khw.  stʼryk  m.,  M.  storǝ́y,  MP
stārag, Skt. tā́rakā- f. ʻid.ʼ.
- wróγ m. ʻcrowʼ < *warā́γa- (PSR), cf. Khw. wrʼγ, MP warāγ, Bal. gurāg ʻid.ʼ.
- wo m. ‘wind’ < * wā́ta-, MP wād, NP bād, Skt. vā́ta- m. ‘id.’, etc.

2. > u before N:
- lúma

f. ‘snare, noose’ < *dā́ma + f. -a, Sogd. (Man.) δʼmʼ, Parth. dʼmg,

MP, NP dām ʻsnare, trapʼ, Skt. dā́man- ʻchain, bondʼ.
- mlúna  f.,  Waz.  vlī́na,  Wan.  awlūn  ʻbridleʼ  <  f.  *abi-dānā-,  Av.  °aiβiδāna-,
Sogd. βδʼʼnh, Khot. byāna-, M. avlān, Sariq. viδun, Yazgh. avδen, Oss. (w)i-
don / jidonæ ʻid.ʼ.
- num m. ‘name’ < *nā́ma, Av. nāman- ntr., OP nāman-, Khot. nāma, Sogd.
nʼm, Sh. nům, Rosh. nom, Sariq. num, Oss. nom, MP, NP nām, Skt. nā́man-
ʻid.ʼ, etc.
- patún  m.  ʻthighʼ  <  *pati-štāna-,  Av.  paitištāna-  ntr.,  MP  padištān  ʻlegʼ,
Sogd. ptštʼn, Sh. bix̌ůn, M. píšken ʻthighʼ.
- šum- ʻto drink, gulpʼ < *čyā́m-, Av. ṧam-, Khot. tsām- ʻto swallowʼ, Sogd.
šʼm  ʻto  swallow,  gulpʼ,  Oss.  cymyn/cumun  ʻto  gulp,  sipʼ,  NP  āšāmidan  ʻto
drinkʼ, Skt. ā-cām- ʻto gulpʼ, etc.
- um,  om  (Khl.),  (y)īm  (Waz.)  ‘raw,  unripe’  <  *ā́ma-  (PSR)  <  PIr.  *Hāmá-,
Khot. hāma-, Oss. xom, NP xām, Skt. āmá- ‘id.’.

The synonymous femin. forms taláka and kuṛakǝ́y may have given rise to the

different variants of lúma: lumǝ́ka, lumǝ́y.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

178

- zum m. ʻson-in-lawʼ < *zā́m°, Av. zāmātar-, Sogd. (Man.) z’mt’yty, MP, NP
dāmād, Abyāne’i zūmōy (m.), Bal. zāmāt, Skt. jā́mātar-, Askhun zamā ‘id.’,
etc.

A problematic instance is obǝ́ f. pl. ‘water’ (cf. Skt. áp- f., sg. nom. āp,

gen. apáḥ, etc.), which is a plurale tantum. Morgenstierne 1983: 171 con-
sidered the form a direct continuation of the old plural of *ap-: “Nom.
Plur. Mask. auf -ə́ darf man wohl direkt auf *-āh zurückführen. Vgl. auch
obə́ f. plur. “Wasser” < *āpāh (für -ah)”. Rather, the original form may
simply be *ob from older *ā́p, to which the (stressed) -ǝ́ has been added
secondarily and interpreted as a pl. ending.

3. > a in open, unstressed position:
- aγúnd-: aγustə́l ʻto put on, donʼ < *ā-gúnda-, Sogd. ’’γwnt, (Man.) ’’γwnd/
’’γwst ‘to cover, dress’ with different prefix, Par. āγun- / āγust- ‘to dress,
put on’, Oss. aγ

yndyn/aγ

yst ‘to cover a building’.

- kará ʻin/from the houseʼ < abl.

*kārā́t, OP kāra- ʻarmy, peopleʼ.

Also the frequently unstressed final feminine *-ā may have become
shortened to -a in Pashto. This outcome would have been generalised, at
the expense of stressed f. *-ā́ > Pashto -ó.

4. > ā in closed, unstressed position:
- wrārǝ́ ‘nephew, brother’s son’ < *βrālrúya- (PSR) < *brāθruyá-, Av. brāt-
ruiia- m. ‘brother’s son’, Skt. bhrā́tṛvya- ‘id.’.

*U AND *Ū

According to Skjærvø, CLI: 400, PIr. *u would have yielded:
- a, in hask ʻtall, highʼ (*uskah), cf. Geiger 1893: 10.
- stressed ə́, in wrə́ǧa f. ʻfleaʼ (*frušā-).
- ø, in nǧor ʻdaughter-in-lawʼ (*(s)nušā- + -or).
- with secondary lengthening *ū > Pash. u, in sur ʻredʼ (*suxrá-).

In final position, *-u would have resulted in:
- *uw > Pash. -ǝ, in psǝ ʻsheepʼ (*pasú(š)).

Comparable (liquid) mass nouns are, as a rule, pluralia tantum in Pashto, such

as gǝnā́ški f. pl. ʻslow, flowing waterʼ, gwaṛí m. pl. ʻgheeʼ, māstǝ́ m. pl. ʻcurdsʼ (NEVP:
52), mǝtiā́zi, mǝtiā́ze f. pl. ʻurineʼ (NEVP: 54, s.v. mež- : mit-), mu m. pl. ʻcongealed fatʼ
(NEVP: 48), orǝ́ m. pl. ʻflourʼ (NEVP: 10), šawdǝ́, šodǝ́ m. pl., šudé f. pl., (Western dial.)
šidé f. pl. ʻmilkʼ (NEVP: 80). Even recent borrowings such as ā́yl m. ʻoil, greaseʼ (< Engl.
oil), tel m. ʻbutter, oilʼ (< Lahnda, Punjabi tel) have been re-interpreted as a plurale
tantum.

The form kor ʻhouseʼ, on the other hand, reflects nom./acc. *kāráh/kārám (>

*kā́r°, PSR).

The most likely reason is to avoid convergence with the oblique pl. ending -o.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

179

This rather awkward outcome may be simplified as follows.

1. > ǝ (generally) in stressed, open position:
- γanə́m, Wan. γandə́m m. pl. ʻwheatʼ < *gantúma-, Av. gaṇtuma-, MP, NP
gandum, Sogd. γntm ʻid.ʼ, etc.
- lǝm m. ʻ(sheep) tailʼ < *dúma-

, Av. duma-, Sogd. (Man.) δwm, Khot. du-

maa-, Oss. dymæg/dumæg, NP dum(b) ʻtailʼ, etc.
- psǝ ʻsheepʼ < acc. *pasúm, Av. pasu-, Khot. pasa-, Oss. fys/fus, Skt. paśú-,
etc.
- sə́ǧay  m.  ʻlungsʼ  <  du.  *suši  +  -ay,  Av.  suši,  Khot.  suvʼä,  MP,  NP  šuš  ʻid.ʼ,
etc.
- təš,  Wan.  təs  ‘empty’  <  *túsya-  (PSR)  <  *tussyá-

, cf. Khot. tuśśaa-, Oss.

(Iron) tyssæg, Parth. twsyk, MP tuhig, NP tuhī, Skt. tucchyá- ʻid.ʼ.
- wrārǝ́ ‘nephew, brother’s son’ < *βrālrúya- (PSR) < *brāθruyá-, Av. brāt-
ruiia- m. ‘brother’s son’, Skt. bhrā́tṛvya- ‘id.’.
- wrə́ǧa f. ʻfleaʼ <*frúši- + f. *-ā

, Sogd. ʼβšʼʼh f., Yi. friγo, Skt. plúṣi- m. ʻid.ʼ.

- wǝz m. ʻgoatʼ < *búza- m., Av. būza-, MP, NP buz ʻid.ʼ.

The delabialisation of short *u is of recent date, which can be found

in recent loanwords as well, e.g. ulə́s ʻpeople, folksʼ (< Turkic ūlus), sǝstí
ʻweakness, impotenceʼ (< Pers. sustī).

2. > ø in unstressed, open position:
- bar ʻtop, summitʼ< *ubára- (PSR) < *úpara-, Av. upara- ʻhigherʼ, MP, NP
bar ʻonʼ, Skt. úpara- ʻ superiorʼ, etc.
- sra f. ʻredʼ < f. *suxrā́-, Skt. f. śukrā́-, see also sur m. (q.v.).
- w(u)zá f. ʻnanny-goatʼ < *βuzā́ (PSR) < *búzā- m., Av. būza-, MP, NP buz
ʻid.ʼ.

3. > u in closed position:
- aγúnd-: aγustə́l ʻto put on, donʼ < *ā-gúnda-, Sogd. ’’γwnt, (Man.) ’’γwnd /
’’γwst ‘to cover, dress’ with different prefix, Par. āγun- / āγust- ‘to dress,
put on’, Oss. aγ

yndyn/aγ

yst ‘to cover a building’.

- búr, f. búra ‘having lost a son, childless’ < *apúθra-, cf. Skt. apútra- ʻwith-
out sonʼ.

Final stressed **dumá- would have become Pash. **l(ə)mə́, cf. psə́ ‘sheep’ < *pa-

sú-.

On the preform *tussya-(ka-) see Cheung 2000: 69 ff. Obviously, the consonant

cluster *ssy would have seen further simplification prior to *u > ə.

The accent of wrə́ǧa is unexpected, as we would expect **w(ǝ)rǧá, but it can be

noticed that designations for ʻsmall verminʼ, such as mə́ǧa ʻratʼ, ǧγā́nja, káyka ʻfleaʼ,
ričá ʻnitʼ are generally feminine. Similarly wrə́ǧa may have been feminised after the
PSR stage, parallel to the Sogd. cognate form.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

180

- gúta f., Wan. nəgút ʻfinger’ < *angúštā-, Av. angušta- m. ʻtoeʼ, MP, NP an-
gušt, Skt. aṅgúṣṭha-, aṅgúli- f. ‘finger, toe’.
- lur  f.  ʻdaughterʼ  <  voc.  *dúxtar,  cf.  Av.  dugǝdar-,  duγdar-,  Khot.  dutar-,
Waxi δǝγ̌d, Yazg. δoγd, plur. δǝdar, M. lə́γda, Yi. lúγdo, MP duxtar, duxt, NP
duxtar ʻid.ʼ, Skt. voc. dúhitar ʻoh daughter!ʼ, etc.
- púča, pə́ča f., Wan. pukē ʻdroppings of sheep, goat and camelʼ < *puškā-
(?),  M.  puška,  Sh.  paxč,  Wa.  pǝṣ̌k  ʻdroppings  of  sheep  or  camelʼ,  NP  pušk
ʻsheep  or  camelʼs  dungʼ,  cf.  Khot.  pulske  ʻexcrementsʼ,  etc.  (DKS:  246,
Steblin-Kamenskij 1999: 280).
- ruǧd ʻhabituated, accustomedʼ < *fra-(y)ux(š)ta- ʻtrained, used toʼ, Sogd.
yxs-  ‘to  acquire,  contract  a  habit’,  ywxs-,  ywxs-  ‘to  learn,  study,  be
taught’,  Yi.  yůxs-/yůxt-,  M.  yúxs-/yuxt-  (inch.)  ‘to  learn’,  Yaghn.  yūxs-/
yūxta- ‘to get used, accustomed to, to contract an (annoying) habit’, Wa.
yǝx̌k (ppp.) ‘learned’.

- sur m. ʻredʼ < *súxra- (PSR) < *suxrá-, Av. suxra-, Oss. syrx/surx, MP suxr,
NP surx ʻredʼ, Skt. śukrá- ʻshining (white)ʼ, etc.
- ux̌  m.  ʻcamelʼ  <  *úštra-  m.,  Av.  uštra-,  Sogd.  xwštr,  NP  šutur,  Skt.  úṣṭra-
m., etc.
- wuč  ʻdryʼ  <  *huška-,  Av.  huška-,  Khot.  huṣka-,  MP,  NP  xušk,  Skt.  śúṣka-
ʻid.ʼ, etc.

The outcome a is rarely found, notably in hask, and should not be con-
sidered in our survey. It is more likely that hask is merely a contracted
form  of  deictic  ha-  ʻthatʼ  and  *usk  (<  *uska-,  Av.  uskāt̰,  Khot.  uska,  Khw.
ʼsk, etc.).

Another putative example, páča (see pə́ča, α.), cannot be used here

either, as the form has probably been translitterated inaccurately (cited
by Geiger 1893: 18). The standard form, as found in Aslanov and Kabir/
Wardak 1999 is either pə́ča or púča (cf. Aslanov 1966: 160b, Bellew 1901).

4. PIr. *-am > *u > ə́ in stressed position:
- o(w)rə́, warə́ ʻa cloudʼ < *aβrú < ntr. *abrám, Av. aβram ntr., MP, NP abr,
Skt. abhrám ntr. ‘id.’, also Oss. arv ʻskyʼ (*ha-abra-, cf. Cheung 2002: 154).
- zǝ ʻIʼ < *azú < *azám, Av. azǝm, OP adam, Sogd. ’zw, Khot. aysu, Oss. æz,
Skt. ahám ʻIʼ, etc.

The development PIr. *-am > *u can be found in several East Iranian lan-
guages, probably also including Pashto.

Not from *fra-wrzda-, Skt. pravṛddha- ʻgrown up, fully developedʼ (NEVP: 69).

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

181

5. Old *ū may have merged with short *u, but only after PSR:
- mə́ǧa f. ʻratʼ < *mū́š- + f. *-ā, cf. Khw. mwf f., Oss. myst/mistæ ʻmouseʼ, MP,
NP mūš, Skt. mū́ṣ- m., f. ʻmouse, ratʼ.
- nən ‘today’ < *nū́nu (PSR) < *nūnám, cf. MP, NP nūn, Skt. nūnám ʻnowʼ.
- stǝn f. ʻpost, pillarʼ < *stū́ni-, Av. stuna-, stūna-, Khot. stūnā-, OP stūnā-,
MP stūn, NP sitūn, Skt. sthū́ṇā- ʻid.ʼ, etc.
- stər ‘big, large’ < *stū́ra- (PSR) < *stūrá-, cf. Av. °stūra-, Khot. stura-, Oss.
æstyr/ æstur, Skt. sthūrá- ‘id.’.
- wrə́ja, wrúja

f. ʻeyebrowʼ < *brū-čī- + f. *-ā, Ishk. vric, cf. Av. dat. pl. f.

bruuat̰.biiąm, Sogd. βrʼwkh, Khw. βrwc, Wakhi vrǝw, MP brūg, NP abrū, Skt.
bhrū́- f. ʻid.ʼ, etc.

There are two instances that would show final -u in Pashto. The form

lu  pl.  tant.  m.  ‘smoke’  <  *dūta-  (Sh.  δūd,  Sariq.  δůd,  Yazgh.  δod,  MP,  NP
dūd ʻsmokeʼ) usually occurs as the diminutive formation lugáy. This for-
mation  may  go  back  to  earlier  <  *lúdgay  or  *lúlgay.  As  the  stem  vowel
was  in  front  of  the  medial  consonant  cluster  (and  hence,  closed  sylla-
ble),  it  would  have  become  Pashto  u.  Subsequently,  a  simple  form  lu
would have been abstracted from this dimin. formation.

The other form mu m. pl. tant. ‘congealed fat’ is more likely to be a

borrowing from Indo-Aryan (cf. Waig. muī ʻmarrowʼ, Phal. mī m. ʻmar-
rowʼ, Shina. mī͂ f. ʻfatʼ), rather than inherited (~Av. mūθra- ‘excrements’,
NEVP: 48).

An interesting borrowing can be mentioned here as well:

- həg m. ‘pig’ < MP/Parth. hwg [hūg] ‘id.’ (Sundermann 1981: 162), cf. NP
xūg ‘id.’.

(not in EVP / NEVP).

6. > wə, (w)u after velar:
- γwəl, γwul m. pl. ‘shit’ < *gūθa-, Av. gūθa-, guθa- ‘faeces’, Khot. gū-, Khw.
γwθ, Sh. γaθ, gi (contaminated with NP gūh ?, Steblin-Kamenskij 1999:
177), MP, NP gūh, Skt. (late) gūtha- m. / ntr.
- kwə́na f. ‘backside’ < f. *kū́nā-, Kurd. kūn f., Abuzeydabadi, Abyanei kǖn
f., MP, NP kūn ‘id.’, etc.

Pashto kwə́na f. ‘backside’ has been omitted in NEVP, presumably be-

cause it might be a loanword from Persian (EVP: 32). The feminine char-
acter of the form, which is also attested in the West Iranian languages

Wrúja is the more common variant, as wrə́ja is also homonymous with wrə́ja

ʻ(sheep) tickʼ: the vocalism of wrúja may come from the NP borrowing abrū or reflect
a pl. form *bruwah ?

A genuine Pashto form would have been †wəg or †wəy. Pashto has also bor-

rowed NP xūg.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

182

that distinguish gender, may suggest that it is inherited rather than
borrowed.

*I AND *Ī

The vowels *i and *ī would have merged, similar to *u and *ū.

1. > ǝ in an open, stressed position:
- bǝl, Wan. bī (f. bə́la, Wan. biyā) ʻsecond, otherʼ < *dwitiya-, cf. Av. bitiia-,
Sogd. δβtyk ‘second’, etc., or < *dwita-, cf. Khot. śäta- ‘second’, Parth. byd
ʻagain’, Skt. (adv.) dvitā́ ʻfor the second time, as beforeʼ.
- šə́ga, Wan. sǝga f. ʻsand, gravelʼ < *sikā- ?, OP θika°, Khot. siyatā-, Orm.
sigo, saga ʻsandʼ, Oss. syǵyt/sigit ʻsoilʼ, also Phalura šíga ‘sand’ (from older
Pashto).
- x̌əl, also m. x̌al (Aslanov 1966: 566b) ‘stairway of stones and earth’ <
*srita-, *sriti-, Yazgh. x̌ad ‘ladder’, Khw. ’šc f. ‘ladder’, Skt. śritá- ‘situated,
depended on’, etc.

Absolutely unambiguous instances of short *i becoming ǝ are difficult to
find in Pashto, as *i is often in an unstressed syllable, and therefore lia-
ble to loss. Even the forms quoted here are problematic, as it cannot be
excluded that ǝ in some of these instances is anatyptic, serving to allevi-
ate the pronounciation of an awkward cluster.

The preform of bǝl, Wan. bī (f. bə́la, Wan. biyā) ʻsecond, otherʼ is

somewhat  unclear.  It  is  derived  from  *dwitiya-  or  *dwita-.  Both  recon-
structions  *dwitīya-  and  *dwitá-  would  have  been  acceptable,  in  my
opinion: * dwitī́ya- (PSR) could have yielded Pashto *blǝ (= Wan. bī), but
this cluster would have been metathesized to bǝl, cf. sǝl ‘hundred’ < *slǝ <
*satám (Cheung 2010: 115).

The outcome, the initial b-, has not been explained sofar: the old *dw

appears to have yielded w- in war ʻdoorʼ (*dwar-) and dw- in dwa ʻtwoʼ
(*duwā), besides b-. Perhaps, it may be imported from byā ‘again, anew’
(< Indo-Aryan, cf. Punjabi, Lahnda biā ‘again’) ?

The form x̌əl and its apparent variant x̌al are derived from a ntr. or

masc. form *sritá-, which would have become †x̌lə, which again has an
inadmissible initial cluster group *x̌l.

Finally, the preform *sikā-, with the assumed accent on long ˚ā́ (PSR),

should have yielded †šgá. Even with an anatyptic vowel, the form
should have remained final-stressed **šǝgá. As the ‘sand’ forms are typi-
cal Wanderwörter, also Pashto šə́ga could be an old borrowing, for in-

Evidently the derivative formation kwəní ‘(homosexual) passive’ is a loan

translation of Pers. kūnī.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

183

stance  from  Parth.  or  MP,  which  tend  to  have  initial  stress,  cf.  MP,
Parth.  sygd  [sígad],  MP  sygyn  [sígēn],  [ségēn]  ‘stony,  of  stone’  (Durkin-
Meisterernst  2004:  312).  The  loss  of  final  ˚d  would  have  occurred  at  a
later stage, resulting in the feminine assignment of šə́ga.

2. > ø in unstressed, open position:
- mlúna  f.,  Waz.  vlī́na,  Wan.  awlūn  ʻbridleʼ  <  f.  *abi-dānā-,  Av.  °aiβiδāna-,
Sogd. βδʼʼnh, Khot. byāna-, M. avlān, Sariq. viδun, Yazgh. avδen, Oss. (w)i-
don/jidonæ ʻid.ʼ.
- patún  m.  ʻthighʼ  <  *pati-štāna-,  Av.  paitištāna-  ntr.,  MP  padištān  ʻlegʼ,
Sogd. ptštʼn, Sh. bix̌ůn, M. píšken ʻthighʼ.
- weǧd m. ‘pillow’ < nom. sg. *βárzi (PSR) < *barzíš-, Av. barəziš- ntr. ‘pil-
low, cushion’, Wakhi vorz, Oss. baz ‘cushion’, MP, NP bāliš ‘pillow, cush-
ion’, etc., Skt. barhíṣ- ntr. ‘sacrificial straw’.
- wlešt,  Wan.  lwast  f.  ‘span  between  the  thumb  and  little  finger’  <  *wi-
tá

sti- (PSR) < *wí-tasti-, Av. vītasti-, Oss. wydisn(y) / uʒesnæ, MP widest, NP

bidast, Skt. vítasti- f. ‘id.’.

3. > i in a closed position:
- writ  m.,  f.  writá  ʻroasted,  friedʼ  <  *βríγta-  (PSR),  f.  *βriγtá-  <  *brixtá-,  f.
*brixtā́-, Wa. vrǝšt-, Bal. brēxta, ‘id.’, Khw. βryγ ʻkebabʼ, MP brištan, NP bi-
rištan ‘to roast, fry’, Skt. bhṛṣṭá- ‘roasted’, etc.
- wit  ‘open,  wide,  ajar’  <  *wíšta-  (PSR)  <  *wištá-,  OP  višta-  ‘unbound’  (in
°aspa-  personal  name),  cf.  MP  wišā-,  NP  gušāy-  ‘to  release’,  Sh.  wix̌i(y)-  :
wix̌id,  Rosh.  wix̌ay-  :  wix̌id,  Bart.  wixī-  :  wixīd  ‘to  open,  unlock’,  etc.
(Cheung, EDIV: 135 f.).

4. *ī > ǝ:
- rǝ́ma f. (usually pl. rǝ́me) ʻfluid excrement, diarrhoeaʼ < f. / pl. *rī́mā,
Khot. rrīma- ʻfaeces, filth, dirtʼ, Sogd. (Bud.) rym, rymh, NP. rīm ʻpusʼ, Av.
irimaṇt- ʻfull of filthʼ.
- x̌ǝ́ja f. ʻwomanʼ < *strī-čī- + f. *-ā, Sogd. ʼstryc (*strī-čī-ā-, Sims-Williams,
CLI: 190, Sangl. ṣǝc, Av. strī- f., Skt. strī́- f. ʻid.ʼ.
- žər, zər ’quick’ < *ǰī́ra- (PSR) < *ǰīrá-, Av. jira- ‘quick, quick witted’, NP,
MP zīrak ‘clever, astute’, Skt. jīrá-‘quick, quick witted’.

Also in borrowings from Persian, e.g.:
- enjǝ́r m. ‘fig’ (variant anjír m.) < Pers. anǰīr ‘id.’

5. Final PIr. *-ah > *ī:
- meṛǝ́ m. (voc. máṛa) ‘husband; manly, brave, generous’ < *mārdī́ < nom.
*martáh, cf. Av. marəta- m. (< *martá-), MP, NP mard ʻmanʼ.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

184

- udə́, Kandahari bidə́ (Waz. wewd, Mahs. wȫwd)

‘asleep’ < *(h)uβdī́ < nom.

*huftáh, Skt. suptá- ‘id.’.
- zǝr ʻthousandʼ < *(h)azī́rī < *hazáhrah, Av. pl. hazaŋrō, Sogd. zʼr, MP, NP
hazār, Skt. sahásram ʻthousandʼ, etc.

Final PIr. *-ah may have resulted in *ī, which is a common East Iranian
development. Subsequently, it would have become ə́ in Pashto, if
stressed (see further i-umlaut, q.v.).

In the case of marγə́, murγə́, mərγə́ m. ‘bird’ < *mṛgá-, cf. Skt. mṛgá- m.

‘wild animal’, the variants may reflect the generalisation of different
case endings, on which see further below.

*AU

1. > wa
- γwaǧ, dial. γweǧ m. ʻearʼ < *gauša-, Av. gaoša-, Sogd. γwš, MP, NP gōš ʻid.ʼ,
etc., Skt. ghóṣa- m. ʻnoise, soundʼ.
- lwaγ, (Khushhal Khan) lweγ ʻmilkingʼ < *dáuga-, Sh. δůγ, NP dōγ ʻbutter-
milk, a kind of yoghurt drinkʼ, Skt. dóha- m. ʻmilkingʼ.
- lwaš-, dial. (a)lweš- ʻto milkʼ < *dáučya- (not *dauxšaya-, Skærvø, CLI:
400).
- rwaj, Waz. wrǝz, Wan. wrez, f. ʻdayʼ < *raučah-, Av. raocah- ntr. ‘light’, OP
raučah-, MP, NP rōz, Kurd. rōj ‘day’, Skt. rócas- ntr. ʻlightʼ.

The diphthongs may have gone first to monophthongised *ō, and sub-
sequently, the diphthongisation wa, wǝ, we would have occurred in the
different Pashto dialects.

The forms xol m., Waz. xēl ʻhelmetʼ, xwála f. ʻhatʼ are unclear: they are

either from *xauda- or a borrowing from another Ir. language (e.g.
Bactrian), which seems all the more likely in view of the variety of the
forms. The form šómle f. pl. ʻbuttermilkʼ can hardly derive from *xšau-
dah-, Av. xšaodah- ʻstream, currentʼ.

2. > u
- γúna f. ʻcolour, complexion; body hairʼ < ntr. pl. ? *gáunā, Av. gaona-,
Khot. ggūna-, Khw. γwn ʻcolourʼ, MP, NP gūn ʻcolour, kind, mannerʼ.
- ruṇ, f. rúṇa ʻbrightʼ < *ráuxšna-, Av. raoxšna-, Sog. (Man.) rwxšn, MP, NP
rōšan, Wa. ruxn ‘id.’.

Some dialect variants that appear to show i-umlaut, may reflect the old nomi-

native: *(h)uβdī́ > *

iwdí > common Pash. *

əwdə́. Other variants appear to be based

on the accusative: *(h)úβdu > common Pash. *

uwd.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

185

- yun m. ʻmovement; gait, stepʼ < *yauna-, Av. yaona-, Khot. gyūna-, jūna-
ʻmovement, gaitʼ.

In front of -n, the monophthongised *ō appears to have joined *ā́ and
has thus become u.

Some forms are difficult to explain. The form x̌na f. ʻleg, hipboneʼ is

usually  connected  to  Av.  sraoni,  Khot.  ṣṣūñi  ʻloinsʼ,  Sogd.  (Buddh.)  šwn,
Sh. x̌ūn, Skt. śróṇi- f. (usually du.) ʻhipʼ, cf. NEVP: 98. We would expect a
form like †x̌ún(a). In Cheung 2010: 114, I have assumed a contamination
with  px̌a  ʻfootʼ,  but  see  further  below.  There  is  also  an  apparent  com-
pounded  formation  x̌ánga  ʻlegʼ,  with  °anga  from  Indo-Aryan  ?,  cf.  Skt.
áṅga- ntr. ʻlimbʼ.

Also unclear are:
- niz, nyuz, nyoz m. ʻflood, torrentʼ < *ni-yauza- ?
- zwǝǧ m., Waz. (Raverty) zwaǧ ʻbitterness, distress; gallʼ < *a-zauša-?, cf.
Pers. zōš ʻwicked, violentʼ, Arm. LW zoš ʻuglyʼ (rather ~ Pers. zišt ʻid.ʼ ?).

*AI

1. > e:
- γelé f. pl. ‘flocks’ < pl. *gaiθā́h, Av. pl. instr. pl. gaēθābiš ‘living beings’,
OP gaēθā- f. ‘cattle, flocks’, MP gēhān, NP ǰihān ‘world’, cf. Skt. gehá- ntr.
‘dwelling, house’.
- lemə́ m. pl. ʻeyesʼ < ntr. *daima + pl. -ə́, cf. Av. daēman- ntr. ʻeyeʼ, MP, NP
dēm ʻface, countenance, cheekʼ.
- lewár  m.  ‘husband’s  brother’  <  *daiwár-,  Yaghn.  séwir,  siwir,  Ishk.  sew,
Oss. tiw/tew, Skt. devár-, devara- ‘id.’, etc.
- lewə́ m. ‘beast of prey’ < *daiwíya- or *daiwáya- ‘daevic animal’.
- me encl. pron. ʻme, of meʼ < *mai, Av. mē, mōi, OP -maiy, Khot. -mä, MP,
NP -m, Skt. -me ‘id.’.
- meǧ,  Afr.  myag  (influenced  by  m.  ?),  Wan.  myež  f.  ʻeweʼ  <  *maišī́-,  Av.
maēšī-, Abyane’i meš f., Skt. meṣī́- ʻid.ʼ.
- melmá m. ‘guest’ < *maiθmán-, Parth. myhm’n, MP, NP mehmān, Yazgh.
miθmen ‘id.’.
- welá

‘advantage, profit, gain’ < pl. ntr. *wáidā, Av. vaēδah- ntr. ‘posses-

sions’, Skt. védas- ntr. ‘possessions’.
- wex m. ‘root, base’ < *waixa-, Sogd. wyx, Parth. wyx, NP bēx ‘id.’.
- xež- : xat- ‘to rise’, *xaiza-, Khw. m|xyz- ‘to rise, get up’, Oss. xizyn/xezun
‘to climb’, NP xēz- ‘to rise, get up’, cf. MP Man. (+ *awa-) ’wxyz- ‘to des-
cend’, Parth. (+ *ā-) ’xyz- ‘to rise’.

The accent of welá is from bahrá f. (< Arab.-Pers.), sarfá (< Arab.-Pers.), x̌a f. ‘ad-

vantage, profit, gain’ ?

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

186

The exceptions to this development are:
- maǧ, Wan. myaž m. ʻramʼ < *máiša- (PSR) < *maišá- m., Av. maēša- m.
ʻsheepʼ, Khw. ʼmh f. ʻeweʼ, MP, NP mēš ʻsheep, ramʼ, Skt. meṣá- m. ʻramʼ.
- wə́la, wála, wúla f. ʻwillowʼ < *waiti- + f. -a, Av. vaēti-, M. wī́ya, NP bēd, Bal.
gēt ʻid.ʼ.
- xwalá f., Waz. xēla, Bang. xolyá, Ghilz. xwolyé, Sher. xwǝlye (usually pl.
xwalé) ‘sweat’ < *hwaidā-, Kurd. xwih f. (< f. *hwidā- ?), cf. Av. x

aēδa- m.,

Sh. xēδ, Skt. svéda- m. ʻid.ʼ, etc.
- yaw, yow m., f. yawá, Afr. iwá, Bang. ewá ʻoneʼ < *aiwá-, *aiwā́-, Av. aēuua-
m., aēuuā- f., Khot. śśau, Sogd. ʼyw, Oss. iw/(j)ew, MP, NP yek ʻid.ʼ, etc.

In the case of yaw, it can be noted that in many Ir. languages, most pro-
minently in Persian, a prothetic y- has developed for which there is no
doubt a phonetic reason: the unique, monosyllabic *ēw may have been
quite challenging to pronounce, resulting into a diphthongisation of *ew
> *yew > *yǝw or yaw.

As noted by Morgenstierne (NEVP: 48; EVP: 49) the vocalism of maǧ is

difficult to account for. It may have been influenced by similar forms,
such as warg m. ‘3-year old ram’.

The wə́la, wála, wúla forms are semantically and formally very similar

to wána, wúna f. ‘tree’, to which they may have been contaminated.

Finally, xwalá and all its divergent variants, which are difficult to re-

concile, are perhaps borrowings from certain Pamir languages, cf. Orm.
xola, Yidya xul, Yazg. xwiδ ʻid.ʼ. The suggested preform *xwaidiyā- (NEVP:
97) has no further correspondence in (Indo-)Iranian.

2. > i before -n:
- spin  m.,  f.  spína  ʻwhiteʼ  <  (<)  *spáila  (PSR)  <  *swaitá-,  Av.  spaēta-,  Khot.
śśita-/śśiiya-, Sogd. spytk, Yi. spī, Parth. ʻspyd, MP spēd, NP (LW) sipēd ʻid.ʼ,
etc.
- šin m. ʻgreen, blueʼ < *axšáina- (PSR) < *axšainá-, Av. axšaēna-, Khot. āṣ-
ṣeina-, Oss. æxsin, Kurd. šīn ‘id.ʼ.
- win- ʻto seeʼ < *wáina-, Av. vaēna-, OP vain-, Sogd. wyn, Parth., MP wyn-,
NP bīn- ‘id.’, etc.

The  connection  of  the  Pashto  form  spin  m.,  f.  spína  with  the  rare  Rig-
vedic formation śvítna- ʻwhiteʼ (attested only twice ) that was suggested
by Morgenstierne (NEVP: 75; EVP: 68), is rather spurious, cf. Mayrhofer,
KEWAia II: 678. The Pashto form is more likely cognate with Av. spaēta-
etc., which is well attested in (Indo-)Iranian. The expected Pashto out-
come  †spel  may  have  been  influenced  by  the  ʻblue,  greenʼ  word  sin:  >>
spin.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

187

3. > ø in unstressed position:
- f. šna ʻgreen, blueʼ < f. *axšainā́-

, cf. šin m.

*Ṛ

Old vocalic *ṛ is extremely susceptible to its surroundings.

1. > ǝṛ in stressed syllable (before *t),
> ǝṇ (before *n),
> ǝ (before sibilant):
- mǝṛ ʻdeadʼ < *mṛtá-, Sogd. mrt-, Oss. mard, NP murdah, Skt. mṛtá- ʻid.ʼ, etc.
- pǝṇ ʻfullʼ < *pṛná-, Av. pǝrǝna-, OP paru-, Oss. fyr / fur, MP, NP purr, Skt.
pūrṇá- ʻid.ʼ, etc.
- pǝṛ ʻbeaten, defeated, worsted, found guiltyʼ < *pṛtá-, Av. pǝṣ̌a- ʻguiltyʼ
- stə́ṛay (orig. past ptc.) ʻtired, wearyʼ

< *stṛta- ʻthrownʼ + -ay, Av. stǝrǝta-

ʻthrown downʼ (Hoffmann/Forssman 1996: 90 f.), Parachi astar- ‘to rub,
wipe away’.
- wə́ṛay (past ptc. of wṛ-) ʻcarriedʼ < *bṛta- + -ay, Av. bǝrǝta-, Khot. buḍa-,
Bal. burta, MP, NP burd, Skt. bhṛtá- ʻid.ʼ, etc.
- yǝǧ m, f. yə́ǧa ʻbearʼ < *

ṛ́ža (PSR) < *Hṛ́θsa-, Av. arša-, Sogd. ’ššh, Khw. hrs,

Khot. arrä, Sh. yūrš, Yi. yērš, Oss. ars, NP xirs, Kurd. hirč, Skt. ṛ́kṣa- ʻid.ʼ,
etc.

2. > ø in unstressed syllable:
- zṛǝ m. ʻheartʼ < ntr. *zṛdáya-, Av. zǝrǝδaiia-, Khw. zrz(y), Khot. ysära-,
Parth. zyrd, MP, NP dil, Skt. hṛ́daya- ntr. ʻid.ʼ.

3. > i before nasal, in closed syllable, with i-umlaut:
- činjáy m. ʻwormʼ < *kṛmi-čī- + -ay, Sogd. (Man.) kyrm ʻsnakeʼ, Munj. kǝrm,
Yaghn. kirm ʻwormʼ, Oss. kalm ʻsnake, wormʼ, MP, NP kirm ʻwormʼ, Skt.
kṛ́mi- m. ʻworm, maggotʼ.

4. in closed syllable, with u-umlaut

> ur,
> u before sibilant:

- murγə́, marγə́ m. ‘bird’ < *mṛγī́ < nom. *mṛgáh, Av. mǝrǝγa- m., Sogd. mrγ-,
Khw. ’mγ-, Oss. marǧ, Parth. mwrg, MP murw ʻbirdʼ, cf. Skt. mṛgá- m. ‘wild
animal’ (see below).

A better explanation than I assumed in Cheung 2010: 114, is that the accent

was originally on the last syllable. The accent on this syllable would not shift in the
f. form, having a heavy syllable (*°nā-), whereas the accent in the m., with its light
final syllable, would retract to the second syllable.

Semantically, perhaps influenced by Pers. x

astah ʻbeat up; tired, boredʼ.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

188

- pux̌t- ʻto askʼ, pux̌tə́na f. ʻquestionʼ < *pṛsa-, Av. pərəsa-, OP prs-, Khot.
puls-, Sogd. ’prs-, Sh. pēx̌c-, Sariq. pars-, Yazgh. pis-, Parth., MP, NP purs-,
etc. Skt. praś, pres. pṛcchá- ‘id.’.
- pux̌tə́y f. ʻribʼ < *pṛsú-kā-

, Khot. pālsuā- ‘rib’, Yi. pərsəγë, M. pūsäγä, Oss.

færsk/færsk’æ ‘rib’, MP pahlūg, NP pahlū ‘side, rib’, cf. Av. pərəsu- m. ‘rib’,
Wa. pĭrs, Skt. pṛṣṭí-, párśu- f. ‘rib’.
- úǧa, Wan. murža f. ʻgarlicʼ < *bṛžn°, Sogd. ʼβzn-, Yi. wεẓ̌nu, ʻid.ʼ.
- úǧd, f. úǧdá, Waz. wīžd, f. wužda ‘long’ < nom. sg. * bṛzáh, Av. bərəzaṇt-
‘great, high’, Khot. bulysa-, Sogd. (Buddh.) βrz, Sh. vūγ̌j, f. vōγ̌j, Bart. vūz,
Yazgh. vəz ‘long’, Oss. bærzond ‘high’, MP, NP buland ‘high, loud’, Skt.
bṛhánt- ‘big, large, vast’.

In conclusion, vocalic *ṛ may have been vocalised to *ir or *ur, both

of which would have become normally Pashto ǝr.

In the case of yǝǧ m, f. yə́ǧa, the previous stage must have been *ir, in

order to explain the glide y-.

As for činjáy, the vocalisation of *ṛ to *ir triggered the palatalisation

of  the  initial  velar  *k-  >  č-,  simultaneously,  the  i-element  has  been
ʻʻstrengthenedʼʼ  to  -i-  in  Pashto  before  the  consonant  group  °nj°  (ob-
viously simplified from *°rmj° < *°rmi-č°).

The variants murγə́, mǝrγə́ (eastern dial.) marγə́, mārγə́ ʻbirdʼ may be

explained as follows. In the first place the form murγə́ may have derived
from the accusative: *mṛgám > *mṛ́γu (PSR) > *múrγu (u-umlaut) > *murγ
(strengthening  of  *u  in  front  of  CC).  The  variant  mǝrγə́  on  the  other
hand,  may  reflect  the  genitive  or  the  nominative,  thus  resulting  in  i-
umlaut, and subsequent centralisation of this (unstressed) secondary *i
> ǝ. The Eastern forms may be abstracted from the pl., in which *ṛ would
have  developed  into  *ar,  due  to  the  long,  stressed  ā  in  the  following
syllable: *mṛgā́nām > *marγā́nām. This form has become the stem marγ°
in the Eastern dialects.

I-UMLAUT AND EPENTHESIS

As shown, the development of old *a > ə́ in front of *n may be due to the
effect of i-/u-epenthesis. This effect is not limited to the nasal, the other
liquid sounds have also been affected.

On *rs > x̌t, see Morgenstierne 1940: 168 ff., Skjærvø, CLI: 404.

The vocalisation of *ṛ > *ar appears only before a consonant group *HC ?

The implication of this explanation is that Proto-Pashto, prior to its breakup

into the numerous dialects, had retained grosso modo the Old Iranian case-endings, a
feature shared with the other Eastern Iranian languages (notably Khotanese, Sog-
dian, Khwarezmian and Ossetic). These languages still had a full-fledged inflexional
system, albeit somewhat simplified, until the later stages of their development.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

189

1. > *i > ǝ
- əní Wan. ‘other’ < *in

í < *anyá-, Av. aniia-, etc.

- bən f. ‘co-wife’ < *bin < *hapáθnī-, Av. hapaθnī- ‘id.’.
- nər, nar ‘man, manly’ < *nir < *nári < nom. nárah.

- stən f. ‘needle’ < *sojín < *saučánya- f., NP sōzan, Khot. saujsaña-, Oss. su-
ʒin/soʒinæ ‘id.’.
- yə́we f. pl. ʻplough, ploughingʼ < *yáwyā- ʻrelated to grainʼ.

A peculiar form is zyaṛ, Wan. z(i)yaṛ ‘yellow; pale’, which is compared to
Av. zairita-, MP, NP zard ʻyellowʼ, Skt. hárita-. This form appears to show
some sort of dipthongisation of an intermediary *e > ya. Interesting en-
ough, the zeṛ-variant is also found in derivational formations: zeṛə́ka, zéṛ-
ka  f. ʻ red-headed  bunting  (Emberiza  bruniceps)ʼ,  zeṛáy  ʻjaundiceʼ.  Other
etymologically related forms or variants are žaṛ ʻbrass; yellowʼ and žyaṛ
ʻbrass; yellow, paleʼ. Obviously, we have to establish the regular form(s).
The ʻʻvocalismʼʼ -iya- or ya is reminiscent of one of its antonyms s(i)yāh
ʻblackʼ  (<  Pers.  siyāh),  so  it  is  conceivable  that  zyaṛ  and  žyaṛ  may  show
some kind of influence from this antonym. On the other hand, the wide
range of phonetically irreconcilable variants, zyaṛ, Wan. z(i)yaṛ, žaṛ, žyaṛ,
zeṛ,  suggests  independent  borrowing  in  the  different  Pashto  dialects,
perhaps from Nuristani *źa(i)rita- ?, cf. Kati zəŕə, Bashgalī zĩr and Prasun
žiərä  ʻyellowʼ.  The  forms  zǝrj  ʻhazel-brown,  chestnut,  mauveʼ,  zə́rka  f.
ʻGreek partridgeʼ may be inherited, from *zari-čī- (Av. zairicī- ʻname of a
womanʼ, NEVP: 103) and *zari (+ dimin. ka-suff.) respectively.

Another odd form is w(ə)ryaj, w(ə)ryej f. ‘cloud(s)’, for which a pre-

form *abra-čī- is given (NEVP: 10). Actually, the preform may be rather
*abr(i)ya-čī (cf. Skt. abhríya-, abhriyá- ʻpertaining to cloudsʼ, which would
have developed into *iwryéj > common Pash. *əwryéj.

2. > i before n, in closed position:
- pinjə́ ʻfiveʼ < *pánča + *ī (< pl. *-ah), cf. Cheung 2010: 114, fn. 21.
- skindá f. ʻsplintʼ < skandyā́- ? (NEVP: 73, with initial stress skínda: wrong
?), Av. skǝṇda- ʻfractureʼ.
- wínja- f. ʻslave-girlʼ < *bánda-čī- f. of *bándaka-, OP bandaka-, MP bandag,
Khw. βndyk, NP bandah ‘servant, subject’.

3. > e before sibilant, in closed position:
- lex̌- ʻto load (a beast of burden)ʼ < *darzaya-, Av. darǝzaiia-, ptc. dṛšta- ʻto
attachʼ, Khw. δžy- ‘to (up)loadʼ.

The variant nar probably goes back to the accusative: *náru < *náram.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

190

- calwéx̌t  ʻfourtyʼ  <  *čaθwársadi-  (PSR)  <  *čaθwarsáti-,  Av.  caθβarǝsat-  f.,
Khot. tcahaulsä, MP, NP čihil, Skt. catvāriṃśát- f. ʻid.ʼ.
- meǧā́na, Wan. merǰā́ne ʻcayenne pepperʼ < *marži-° < Indo-Aryan, cf. ma-
rīca- ʻpepperʼ.
- reǧd- ʻto tremble, quiverʼ < denomin. *rarzaya-, Yazgh. riwz-, MP, NP lar-
zīdan,  BSogd.  wyr’rz,  CSogd.  wlrz,  CSogd.  wdrz  ‘to  tremble,  shakeʼ,  Khw.
βržyk ‘shaking fever, agueʼ.
- špetə́  ʻsixtyʼ  <  *xwášti-  (PSR)  <  *xwaští-,  Av.  xuuašti,  Khot.  kṣaṣṭä,  Sogd.
xwššty, MP, NP šast, Skt. ṣaṣtí- ‘id.’, etc.
- tex̌t-eǧ-,  Waz.  tašt,  Wan.  tešt-  (pres.  st.  terš-)  ʻto  fleeʼ  <  *t(a)rš(a)ya-  ?,
(Waz.: *t(a)rsa- ?), Sogd. trš, Sogd. (Chr.) tš ‘to fleeʼ, Oss. tærs-/tærs- ‘to be
afraidʼ, MP, NP tarsīdan, Kurd. tirsīn : tirs- ʻto be afraidʼ, etc.
- weš m. ʻsharing, distributionʼ, weš- ʻto divide, distributeʼ < *baxšya-, Av.
baxš-  ‘to  divide,  have  a  share’,  Sogd.  βxš-,  (Buddh.)  βxš-,  (Man.)  βxš  ‘to
give, distributeʼ, Khot. būṣṣ- ʻto giveʼ, NP baxšūdan/baxšāy- ‘to grant, be-
stow’, etc.
- weǧd m. ‘pillow’ < *barzíš-, Av. barəziš- ntr. ‘pillow, cushion’, Wakhi vorz,
Oss.  baz  ‘cushion’,  MP,  NP  bāliš  ‘pillow,  cushion’,  etc.,  Skt.  barhíṣ-  ntr.
‘sacrificial straw’.
- wlešt,  Wan.  lwast  f.  ‘span  between  the  thumb  and  little  finger’  <  *wi-
tá

sti- < *wí-tasti-, Av. vītasti-, Oss. wydisn(y) / uʒesnæ, MP widest, NP bidast,

Skt. vítasti- f. ‘id.’.

4. > ǝ in other instances:
- kišə́p, (Waz.) kašáp

m. ʻtortoiseʼ < *kasyápa-, Av. kasiiapa-, Sogd. kyšph,

NP kašaf (old LW ?), Skt. kaśyápa- m. ʻid.ʼ.
- kə́š(ǝ)r  ʻyounger  (brother)ʼ,  f.  kə́šra  ʻyounger  (sister,  daughter)ʼ  <  *kas-
yah-  +  comparative/  contrastive  *-tara,  Av.  kasiiah-  ‘small’,  MP  kēh,  NP
kih  ‘small(er)’,  cf.  Oss.  kæstær  ʻjunior  personʼ,  Bal.  kastir  ʻsmallerʼ  (in
some Pers. dialects kester, keser ʻyoungerʼ).
- kašáy m. ʻonly sonʼ, f. kašə́y ʻonly daughterʼ < *kasyá-ka-, f. *kasyákī- or
*kaswáka-, f. *kaswakī - ?
- mǝǧ- : mǝx̌-, ʻto rub, wipe, smearʼ, Wan. murš-en- : murš- ʻto rubʼ < *m(a)r-
ša-  (with  -š-  from  *mṛštá-),  pass.  ptc.  *mṛštá-  <  *Hmarza-  ʻto  wipe,  rubʼ

The form cited here is according to Aslanov 1966: 713b and is the (almost) ex-

pected standard Pashto outcome of *kasyápa-: > *kešíb > kišə́p. However, the indi-
genous Pashto forms or variants, such as kišáb, kašáp, may almost certainly have
been influenced by (obsolete) Pers. kašaf, in which final –f is regularly adapted as -p
in Pashto (the pronounciation of f is a sign of educated speech among the Pashtuns).
Pers. kašaf is quite often pronounced as kašaw in colloquial Dari / Eastern Persian, cf.
āwγānistān for Afghanistan.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

191

and *Hmars- ʻto rub, touchʼ (cf. Cheung, EDIV: 180 f.), Av. marǝz- ‘to wipe,
rub’, Oss. mærz- ‘to wipe, brushʼ, MP, NP māl- ʻto rubʼ and Khot. ggumäls-
(gūmaly-) ‘to besmear’, Wa. mĭrs-/morst ‘to feel; touch’, Skt. marś ‘to
touch, to handle’.
- mə́š(ə)r, mə́š(ə)ra f. ʻelder, senior (person)ʼ < *masiiah- + comp. *-tara.
- mǝšáy m. ʻ(little) fishʼ < *masyáka-

, Parth. mʼsyg, MP māhīg, NP māhī, cf.

Av. masiia- m., Skt. mátsya- m. ʻid.ʼ, mātsyá- ʻbelonging to fishʼ, etc.

The outcome differs in these instances, (generally) resulting in ǝ. It can
be noticed that consonant group *sy has yielded š in Pashto. The i-vo-
calism in kišə́p is unexplicable.

The forms kə́š(ǝ)r and its antonym mə́š(ə)r ʻelder, senior (person)ʼ

have a strange (stressed) ə́. This ǝ can hardly go back to *i (after i-um-
laut from *a), which would subsequently have developed into e in front
of a sibilant (not to mention *k- would have been subjected to palatali-
sation, becoming č-). Instead, this central vowel may reflect the effect of
u-umlaut (in the presence of the initial labial *m-): *masya° > *mušša° >
məša°. This vowel has been exported to kə́š°.

The effect of u-umlaut can also be noted in the verb mǝǧ-: mǝx̌-, ʻto

rub,  wipe,  smearʼ,  where  the  archaic  Wanetsi  cognate  murš-en-  :  murš-
confirms  the  former  existence  of  the  umlauted  vowel  u.  We  can  also
note that it occurred in a closed syllable.

5. > i before final *-ti:
- -i  3sg.  pres.  <  thematic  *-ati,  cf.  Av.  3sg.  pres.  them.  -aiti,  MP,  NP  -ad,
Skt. -ati ʻid.ʼ, etc.
- nawí, nǝwí

ʻninetyʼ < *nawád < *navatí-, cf. Av. nauuaiti-, MP, NP navad,

Skt. navatí- ʻid.ʼ, etc.

It seems that the umlauted vowel has developed differently when

the dental has not developed into l, as it is the case with šǝl ʻtwentyʼ
< *winsati- (q.v.).

6. secondary *i > ø in unstressed position:
- mla f. ʻwaistʼ < (Jaji Khatak) mǝl

ā́ < *mil

ā́- (PSR) < *mádyā-, cf.

Av. maiδiia-, Sh. mīδ, Rosh. mēδ, Skt. mádhya- ʻmiddleʼ, etc.

The Pashto and the WIr. forms may rather go back to a relational formation

with long stem vowel *ā, cognate with Skt. mātsyá-.

The forms nǝwe, nawē, as cited by Morgenstierne with final -e, are far less com-

mon (if attested at all), they are no doubt influenced by Geigerʼs preferred transcrip-
tion of this form (nave): the final vowel might reflect a more archaic stage

(

)

-ey, cf.

nawey or nawi in Raverty 1901.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

192

- pla ʻsinew, tendon, nerveʼ < pǝl

ā́ < *pil

ā́ (PSR) < *pádyā-, Av. paiδiiā- f.,

Khw. pzy ʻtendonʼ, Skt. pádya- ʻrelated to the footʼ.

Apparently, a short syllable does not become closed (and thus

“heavy”) before a consonant group *Cy.

The form px̌a, (Northern dial.) xpa, Wan. špa f. ʻfootʼ was already con-

nected by Darmesteter to Av. pāšna- ʻheelʼ, Skt. pā́rṣṇi- ʻid.ʼ, despite the
phonologically difficulties (cf. NEVP: 67). The IIr. ʻheelʼ form may indeed
have been continued by púnda, Waz. pīnda f. ʻheelʼ < *pāršni-tā (*-tā < ?),
cf. Skjærvø, CLI: 405. The form px̌a, on the other hand, is actually more
likely to be an old borrowing, from Pers. pušt-i pā́ ‘upper part of the foot’
(lit. ‘back foot’), an expression also borrowed in Sanglechi with the
meaning ‘instep, upper part of the foot’ (Morgenstierne, IIFL II: 409a).

*Ā

1. > e
- ména f. ‘house, habitation’ < f. *dmā́nyā-, cf. Av. nmāniiya- m. ‘belonging
to the house’, OP māniya- m. ‘household slave’, Oss. moj / mojnæ ‘hus-
band’.
- wéra f. ‘fear’ < *awaharyā-, Parth. whyrd, MP (Man.) whwr- ‘to confuse,
disturb’, Arm. (LW) veher ‘vacillating; unsturdy; fearing’.
- wreγ

m. ‘dye, paint’ < nom. *abi-rā́gī (<PIr. *abi-rāgah), cf. Khot. rrāṣa-

‘dark coloured’, Rosh. rēz-, Yzgh. rǝž- ‘to colour the eyes’, NP razīdan/ ‘to
colour,  dye,  taint,  tinge’,  NP  rang  ‘colour’,  Skt.  raj-  ‘to  colour,  become
red’,  rāgin-  ‘coloured’,  abhi-rañj-  caus.  ‘to  colour’  (late).  vi-rañj-  ‘to  dis-
colour, fade’ (late), vi-rāga- m. ‘loss or fading of colour’ (late).
- wrerá  f.  ‘niece’  <  *  βrā(l)ryā́  (syncope)  <  f.  *brāθruyā́-,  Av.  brātruiiā-  f.
‘brother’s  daughter’,  Skt.  bhrā́tṛvya-  ‘(father’s)  brother’s  son,  nephew;
rival’,

a-bhrātṛvyá - ‘having no rival’.

This obviates the need to postulate a special, common development (Cheung

2010: 114), for x̌na ʻleg, hipboneʼ, x̌pa ʻfootʼ, sna f. ʻblue, greenʼ and also pinjə́ ʻfiveʼ. In
the case of pinjə́, I still think we need to postulate a pl. *-ah > *-ī > -ǝ (from pl. *čaθwā́-
rah > *calwā́rī > calór ʻfourʼ), in order to explain the preceding -i- in pinjə́. But the ac-
cent shift is purely for rhythmic reasons, the accent of the numerals 4-8 usually falls
on the final syllable: calór, pinjə́, špaǧ, owə́ (cf. Skt. saptá), atə́ (cf. Skt. aṣṭā́).

No correspondences are given in NEVP: 90, only a totally hypothetical preform

**w/b/fraiga- ? Actually, the presence of several derivative formations of the Skt.
cognate raj- provides us with the ansatz of a plausible preform.

The original accent in Skt. remains uncertain, bhrā́tṛvya- may have been influ-

enced by bhrā́tar- ‘brother’. For its Pashto correspondence wrārə we have to assume
stress in the middle syllable during the PSR stage, therefore accounting for the lack
of i-umlaut: *brāθruyá- > *βrālrúya- > * wrārú > wrārə́. Pashto has better preserved the
older Indo-Iranian stressing ?: *b

rātruyá-, f. *b

rātruyā́-.The pl. form obviously goes

back to the gen. pl.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

193

- wrerúna dir. pl. (sg. wrārǝ́) ‘nephews’ < *βrā(l)ryā́nā (syncope) < *brāθru-
yā́nām, Av. brātruiia- m. ‘brother’s son’, etc.

Compare also the causative verbs :
- ceṛ- ‘to discuss, study’ < *(wi-)čā́raya-, MP w(y)c’r- ‘to separate, inter-
pret’.
- tej- ‘to cause to run, hasten’ < *tā́čaya-.

The stressed obl. pl. ending –ó has apparently not been affected by i-

umlaut, if we accept a derivation from *-ā́biš (cf. Morgenstierne 1942:
95).

A special case is meṛə́, Wan. māṛǝ, voc. mǝ́ṛa,

pl. maṛúna, meṛúna ʻman,

husbandʼ < nom. sg. m. *martáh, Av. marǝta- (< *martá-), Parth. mrd, MP,
NP mard ʻmanʼ, etc. Skt. márta- m. ʻman, mortalʼ.

The suggested preform *martiya- (NEVP: 52) for meṛə́ would not ex-

plain the vocative mǝ́ṛa and the pl. variant maṛúna: there is no com-
pelling reason why the old IIr. formations *marta- and *mart(i)ya- would
have been enmeshed in one single paradigm in the history of Pashto.

It is possible that the stem meṛ- came from the singular nominative

*-ī (< *-ah) and also genitive *-īya (< *-ahya), whereas maṛ- would reflect
the accusative, vocative and the genitive plural. The variant forms with
the  a-vowel  suggest  that  they  have  developed  from  the  lengthened
(stressed) *ā before *rC, e.g. from the accus. sg. *martám > márdu (PSR) >
*mā́rdu.  The  long  vowel  would  have  been  generalised  in  the  sg.,  thus
also in the nomin. and gen. (retained in the Wanetsi dialect). This nor-
malized  nomin.  and  form  mārdī́  would  then  be  subjected  to  i-umlaut  >
meṛə́.

In the plural on the other hand, this lengthening may not have taken

place,  on  account  of  a  different  accentuation:  the  gen.  *martā́nām  has
developed  regularly  into  Pashto  dir.  pl.  maṛúna  (meṛúna,  with  -e-  from
the sg.). The vocative form mǝ́ṛa ʻoh man!ʼ may derive from the original
nominative form, thereby replacing the original vocative *márta: nomi-
native  *martáh  >  *mirdī́  >>  vocative mírda  (with  added  or  restored  -a)  >
mǝ́ṛa.  The  alternative  form  maṛá  ʻoh  pal,  boy!ʼ  (Aslanov  1966:  817a)
seems to derive from the plural form, which would also explain the final
stressing.

The initial i- in irá f., pl. iré ʻashesʼ < pl. ntr. *āθryā́, Av. ātriia- ntr., Sh.

θīr ‘id.’ is exceptional, it may be a contracted form of initial ya˚, yǝ˚ < *e˚

As cited by Kabir/Wardak 1999. The form maṛa mentioned by Morgenstierne

appears to be cited from Bellews, who did not distinguish between ǝ and a in his
transcription of Pashto words though.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

194

(cf. yaw m. ‘one’ < *ew < *aiwa-). Note also the dialectal Wanetsi variant
aré.

2. > yā before (pretonic ?) *t, *s:
- tiyārá,  t(i)yārǝ́  f.  (adj.  tor  m.,  tóra  f.  ʻblackʼ),  Wan.  tārā  ʻdarknessʼ,  cf.
Khot. ttāḍe ‘darkness’.
- atyā́, Wan. atā́ ʻeightyʼ < *aštāti, cf. Av. aštāiti, Sogd. ’št’, Oss. æstaj, MP,
NP haštād.
- awyā́, Wan. awā́ ʻseventyʼ < *haftāti, Av. haptāti, Khot. haudātä, MP, NP
haftād, Skt. saptatí- ‘id.’, etc.
- myāšt,  Wan.  māst  f.  ʻmonthʼ  <  *māsti-,  cf.  Khot.  māstä,  Sh.,  Rosh.  mēst,
Sariq.,  Bart.  most,  Yazgh.  māst  ʻidʼ  (not  in  WIr.  nor  in  other,  closely  re-
lated EIr. languages, e.g. Wa. mĭy).

The  numerals  go  back  to  preforms  containing  a  final  segment  *°tāti,
which  is  also  identical  to  the  abstract-collective  suffix  *-tāti.  Precisely
this suffix would have become the Pashto abstract suffix -tyā́ according
to  Morgenstierne  1942:  93,  presumably  as  follows:  *tā́ti>  *tā́

di > *t

ā́d >

*tyā́.

This would also explain the development of the numerals. In the

case of myāšt (and also mā́šay ?, see below) we have to assume a similar
development before *s as well.

A remarkable case is tiyārá, t(i)yārǝ́, which may have been suffixed

with *-tāti, which would be a precise corresponce of Khotanese ttāḍe:
*tār + -t

ā́ > t

ār-tā́ (anticipation of the palatal feature) > tiyārá (dis-

similation  of  t  …  t).  As  the  outcome  is  no  longer  analysable  the  final
vowel can be adapted, e.g. re-interpreted as a nom. pl. –ǝ. The antonym
raṇá,  Kak.  roṇyá,  Wan.  rúṇa  ‘brightness’  would  also  have  been  suffixed
with  *-tāti,  the  outcome  of  which  seems  to  have  been  influenced  by
tiyārá, t(i)yārǝ́ and/or the adj. counterpart rúṇ, f. raṇá.

Finally, although mā́šay, myā́ša, Waz. myāsay m., Kak. māsa, Wan.

mésa, f. myā́ša ‘mosquito’ appears to have participated in this develop-
ment, the form and its dialectal variants may actually suggest borrow-
ing from Indo-Aryan, cf. Skt. maśáka- m. ‘fly, mosquito’. For a genuine,
inherited form, we would rather expect the outcome †méšay etc.

*AU

1. > (*o >) e:
- rebǝ́j m. Waz. rēbǝz, rēbuz f. ʻbroomʼ < *raupī-čī- (?), Yaghn. rōpč ʻbroomʼ,
Yazgh. rǝbág ʻ'wooden shovelʼ.

The conditions of *t >

(

)

d or l will be discussed in a forthcoming article.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

195

- meǧáy m., Wan. merža ‘ant’ < *maurijaka- (methathesis of *rw > *wr)
< *marwičaka-, Khot. muṃjaka-, Sh. mūrjak, NP mōrčah ‘id.’.

2. > (*u >) i:
- wína (pl. wíne) f. ʻbloodʼ < *wáhuni-, Av. vohuni-, Sangl. wēn, NP xūn.

The intervocalic *h would have disappeared at a very early stage, in

line with other East Iranian languages.

3. secondary *i > ø:
- x̌na f. ʻleg, hipboneʼ < *x̌iná << *x̌ína < *x̌uni- + f. *-ā < *sráuni- + f. *-ā
< *šuní-ā (cf. γúna < ntr. pl. *gáunā).

The form may have borrowed its final accentuation from the desig-

nations of the other bodyparts, such as px̌a ‘foot’, patún m. ‘thigh’, mla f.
‘waist’, pla f. ‘sinew’.

PALATALISATION

Another effect of the i-umlaut is the palatalisation of certain con-
sonants.

1. *s > š:
The effect on *s > š has been noted above, notably in kišə́p m. ʻtortoiseʼ
(*kasyápa-), kə́š(ǝ)r ʻyoungerʼ (*kásyah-).

Other examples include:
- mə́š(ə)r, mə́š(ə)ra f. ʻelder, senior (person)ʼ < *masiiah- + comp. *-tara.
- mǝšáy m. ʻ(little) fishʼ < *masyáka-, Parth. mʼsyg, MP māhīg, NP māhī, cf.
Av. masiia- m., Skt. mátsya- m. ʻid.ʼ, mātsyá- ʻbelonging to fishʼ, etc.
- (w)šəl, Wan. šwī ‘twenty’ < *winsati-, Av. vīsaiti, Sogd. (Chr.) wyst°, Khot.
bistä, Oss. (y)ssæʒ/insæj, NP, bīst, Skt. viṃśatí- f. ʻid.ʼ, etc.
- šə́ga, Wan. sə́ga

f. ʻsand, gravelʼ < *sikā-, OP θika°, Khot. siyatā-, MP

(Man.), Parth. sygd, Orm. sigo, saga ʻsandʼ, Oss. syǵyt/sigit ʻsoilʼ, Skt. síkatā-
f. ‘sand, grind’.
- təš, Wan. təs ‘empty’ < *tussyá-, cf. Khot. tuśśaa-, Parth. twsyk, MP tuhig,
NP tuhī, Skt. tucchyá- ʻid.ʼ.

2. *z > ž:
- niždé, nəždé

‘near’ < comparative *nazdyah-, Av. nazdiiō ‘nearer’, MP,

NP nazdīk ‘near’.

This form does not show the expected Pashto Stress Rule of the initial stressing

to the following heavy syllable in *ā: †š(ə)gá. As the ‘sand’ form is a typical “Wander-
wort”, Pashto šə́ga is possibly a loanword, rather than inherited ?

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

196

- žay m. ʻleather bag, mussuckʼ < *izyaka- ?, Av. izaēna- ʻleather-ʼ, Yi. ízë,
M. yij(i)ya ʻgoatskin bagʼ.

3. *k > č (ǰ after voiced consonant):
- činjáy m. ʻwormʼ < *kṛmi-čī- + -ay, cf. Sogd. (Man.) kyrm ʻsnakeʼ, M. kǝrm
ʻwormʼ, Oss. kalm ʻsnake, wormʼ, MP, NP kirm ʻsnake, wormʼ, Skt. kṛ́mi- m.
ʻwormʼ , etc.

- čāṛə́,  čāṛá  f.  ‘large  knife,  dagger’  <  *kartyā-,  Khw.  krc  f.,  Sh.  čǟd  f.,  Ab-
yane’i kārd f., Kurd. kēr f. ‘knife’, MP, NP kārd ‘knife’.
- če  rel.  pcl.  ‘that’  <  *káhya,  cf.  Pers.  kih  relative  particle  (not  borrowed
from Pers.

, NEVP: 20).

- čə́rg m. ‘cock’ < *kargī < nom. *karkah, Av. kahrka-, Ishk. kǝrk ʻhenʼ, Sh.
čux̌, f. čax̌, Yazgh. čix̌, f. karj ʻcock, henʼ, Oss. kark ʻhenʼ, MP, NP kark
ʻdomestic fowlʼ, etc.
- ǰǝn f. ʻyoung girlʼ (< *nǰǝn < *čnǝn) < *kanin-, Av. kainīn-, Skt. kanyā̀-
ʻyoung womanʼ, kanī́na- ʻyoungʼ.
- mečə́n f. ‘hand-mill, quern’ < ultimately Gr. mēkhanḗ.

For kišə́p, kašáp m. ʻtortoiseʼ and kə́š(ǝ)r ʻyoungerʼ, see above.

U-UMLAUT

Far  less  noticeable  is  the  effect  of  u-umlaut,  as  in  many  instances  the
umlauted  *u  has  become  the  central  vowel  ǝ,  making  it  indistingui-
shable from ǝ < *i.

*A

1. > ǝ, after m:
- mǝǧ- : mǝx̌- ʻto rub, wipe, smearʼ, Wan. murš-en- : murš- ʻto rubʼ < *m(a)r-
ša-  (with  -š-  from  *mṛštá-),  pass.  ptc.  *mṛštá-  <  *Hmarza-  ʻto  wipe,  rubʼ
and *Hmars- ʻto rub, touchʼ (cf. Cheung, EDIV: 180 f.), Av. marǝz- ‘to wipe,
rub’, Oss. mærz- ‘to wipe, brushʼ, MP, NP māl- ʻto rubʼ.
- mə́š(ə)r, mə́š(ə)ra f. ʻelder, senior (person)ʼ < *masiiah- + comp. *-tara.
- mǝšáy m. ʻ(little) fishʼ < *masyáka-, Parth. mʼsyg, MP māhīg, NP māhī, cf.
Av. masiia- m., Skt. mátsya- m. ʻid.ʼ, mātsyá- ʻbelonging to fishʼ, etc.

The variants with -z-, for instance nizdé, is no doubt influenced by the (near-)

synonymous nazd, which could be Persian borrowing.

According to Morgenstierne (EVP: 21), the lizard form f. čarmǝx̌(k)ə́y might

contain kṛmi-: ʻʻa derivation from *kṛmi-wṛsu-kī- ʻworm-greedyʼ is possibleʼʼ. This sug-
gested preform is rather too fanciful to me.

One would rather expect the ubiquitous Persian relative particle kih (modern

Persian ke) to have been borrowed: > Turk. ki, Urdu ke, etc.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

197

As the i-umlauted vowel *a would have yielded e in Pashto, the out-
come ǝ in these instances should be ascribed to a different development,
here due to a form of u-umlaut, probably caused by the presence of the
initial m.

This kind of u-umlaut would also have affected some (old) loanwords

as well, e.g.
- mǝč m. ‘fly’ < Indo-Aryan, cf. RV mákṣ- (probably m.), mákṣā- f. ‘fly, bee’.

2. > *ā́ > ó before bilabial *u,

(

)

w (< PIr. < *f or *w), in closed position:

- owə́ ʻsevenʼ < *(h)ā́wda (PSR) + -ə́ < < *haftá, Av. hapta, Sogd. ʼβd, Oss. avd,
MP, NP haft, Skt. saptá ʻid.ʼ, etc.
- óx̌a, úx̌ka f. ʻtearʼ, < f. or pl. n. *asrukā, Khot. āṣka-, Wa. yāṣ̌k, Sariq.
yux̌(k), Yi. yāšk, Sh. yūx̌k, Oss, syg/sug, NP ašk ʻid.ʼ, etc.
- pox m. (f. paxá), pl. pāxə́ ʻripe, cookedʼ < *pā́xwa- (PSR) < *paxwá-, Khot.
paha-, Sh. pēx, Yazgh. půx, Oss. fyx/ funx, Skt. pakvá- ʻid.ʼ.
- tod m. (f. tawdá) ʻhotʼ < *tā́wda (PSR) < past ptc. *taftá-, Av. tafta-, Khot.
ttauda- ʻhotʼ, Oss. tævd/tævdæ ʻhot, heat, Skt. taptá- ʻheatedʼ, etc.

The noun óx̌a, úx̌ka f. ʻtearʼ (*asrukā, Khot. āṣka-, Abyane’i asl f., Pers. ašk
‘id.’) is somewhat ambiguous. The variant forms óx̌a, óx̌ka appear to be
Eastern dialect forms. In the absence of variants with a or ā the labial
vowel is best to be explained as the result of u-umlaut.

*Ṛ

1. > u:
- (w)úǧa, Wan. múrža f. ʻgarlicʼ < *bṛžn°, Sogd. ʼβzn-, Yi. wεẓ̌nu, ʻid.ʼ.
- úǧd, f. úǧdá, Waz. wīžd, f. wužda ‘long’ < nom. sg. * bṛzáh, Av. bərəzaṇt-
‘great, high’, Khot. bulysa-, Sogd. (Buddh.) βrz, Sh. vūγ̌j, f. vōγ̌j, Bart. vūz,
Yazgh. vəz ‘long’, MP, NP buland ‘high, loud’, Skt. bṛhánt- ‘big, large, vast’.
- pux̌t- ʻto askʼ, pux̌tə́na f. ʻquestionʼ < *pṛsa-, Av. pərəsa-, OP prs-, Khot.
puls-, Sogd. ’prs-, Sh. pēx̌c-, Sariq. pars-, Yazgh. pis-, Parth., MP, NP purs-,
etc. Skt. praś, pres. pṛcchá- ‘id.’.
- pux̌tə́y f. ʻribʼ < *pṛsú-kā,

Khot. pālsuā- ‘rib’, Yi. pərsəγë, M. pūsäγä, Oss.

færsk/færsk’æ ‘rib’, MP pahlūg, NP pahlū ‘side, rib’, cf. Av. pərəsu- m. ‘rib’,
Wa. pĭrs, Skt. pṛṣṭí-, párśu- f. ‘rib’.

The vocalisation of *ṛ would have been *ur: the *u vowel has strength-
ened before a consonant group, resulting in Pashto u.

In the case of (w)úǧa, the preform may have been *bṛžnā- > *βurǧā- >

Pashto (w)úǧa (Wan. múrža).

On *rs > x̌t, see Morgenstierne 1940: 168 ff., Skjærvø, CLI: 404.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

198

SUMMARY AND CONCLUSIONS

On  the  whole,  the  vowels  of  Pashto  are  quite  susceptible  to  the
influence  of  the  accent  or  the  nature  of  the  adjacent  consonant  or
consonant  clusters.  A  few  general  observations  can  be  made  here.
Unstressed short vowels would quite naturally have disappeared, while
unstressed long vowels would have been shortened. A vowel under the
stress or before a consonant cluster tended to be lengthened.

*A

The vowel *a has become in Pashto:

- ø in an unstressed, open syllable, in bə́n f. ‘co-wife’ (*hapáθnī-), mā́lga f.
ʻsaltʼ (*namáδgā- PSR), zǝ ʻIʼ (*azám), etc.

- a in a stressed, open syllable, γar m. ʻmountainʼ (*gári-), las ʻtenʼ (*dása),
xar m. ‘ass, donkey’ (*xára-).

- ā in a stressed, closed syllable, in lās m. ‘hand’ (*(d(z)ásta-), γā́ṛa f. ʻneckʼ
(*gár

ā-), mā́lga (*namáδgā- PSR), plār ‘father’ (*pidár < *pítar), etc.

> o (in front of a labial), in corb ʻfatʼ (*cā́rba PSR), óx̌a f. ʻtearʼ

(*ásrukā-), pox m.ʻripe, cookedʼ (*pā́xwa- PSR) etc.

> a (in front of a nasal, irrespective of stress), in cang ʻside,

flank’ (*čánga-), γan(d)ə́m ʻwheatʼ (*gantúma-), wand m. ʻdam, dykeʼ
(*banda-), wand- ʻto scold, insultʼ (*wi-wánda-).
- ə in a stressed syllable with umlaut, see below.

*Ā

The vowel *ā has become in Pashto:

- a in an unstressed, open syllable, in aγúnd- ‘to put on, don’ (*ā-gúnda-),
kará ʻin/from the houseʼ (*kārā́t).

- ā in an unstressed, closed syllable, in wrārǝ́ ‘nephew, brother’s son’
(*brāθruyá-).
- o in a stressed syllable, in calór ʻfourʼ (*čaθwā́ra-), dwólas ‘twelve’ (*dwā́-
dasa), lor m. ‘sickle’ (*dā́θram), wo m. ‘wind’ (* wā́ta-), etc.

> u (in front of a nasal), in lúma f. ‘snare, noose’ (*dā́ma + f. -a),

num m. ‘name’ (*nā́ma), šum- ʻto drink, gulpʼ (*čyā́m-), zum m. ʻson-in-
lawʼ (*zā́m°), etc.

*Ṛ

> ø in unstressed syllable, in zṛǝ m. ʻheartʼ (*zṛdáya-).

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

199

> ǝṛ before *t, in mǝṛ ʻdeadʼ (*mṛta-), pǝṛ ʻbeaten, defeated, worsted,
found guiltyʼ (*pṛta-), stə́ṛay ʻtired, wearyʼ (*stṛta- + ay), wə́ṛay ʻcarriedʼ
(*bṛta- + -ay);
> ǝṇ before *n, in pǝṇ ʻfullʼ (*pṛna-);
> ǝ before sibilant, in yǝǧ m ʻbearʼ (*

ṛ́ža < *Hṛ́θsa-).

> i before nasal, in closed syllable, with i-umlaut, in činjáy m. ʻwormʼ
(*kṛmi-čī- + -ay).

> u in closed syllable, with u-umlaut, before sibilant, in úǧa, Wan. murža
f. ʻgarlicʼ (*bṛžn°), úǧd ‘long’ (* bṛzah), pux̌t- ʻto ask’ (*pṛsa-), pux̌tə́y f. ʻribʼ
(*pṛsú-kā-).
> ur before velar, in murγə́ m. ‘bird’ < *murγú (< *mṛgám).

*U / Ū AND *I / Ī

The old high vowels *i / ī and *u / ū have generally merged into an in-
distinct ə́, a development also found in notably Kurdish and Iron Ossetic.
Well before this, the vowels *ī and *ū had lost their long pronunciation.
Similar to many other Iranian languages, these vowels have frequently
affected the preceding vowel or consonant(s). The vowels *i / ī would
have triggered i-umlaut and palatalisation respectively. The effect of
the labial vowels *u / ū is more limited in scope. We can summarize the
developments as follows.

The vowels *u / ū have become in Pashto:

- ø in an unstressed, open syllable, in bar ʻtop, summitʼ (*ubára- PSR), sra
f. ʻredʼ (*suxrā́-).

- ə́ in a stressed, open syllable, in γanə́m, Wan. γandə́m (*gantúma-), lǝm
m. ʻ(sheep) tailʼ (*dúma-), psǝ ʻsheepʼ (*pasúm), wrə́ǧa f. ʻfleaʼ (*frúši- + f. *-
ā), etc.

from *-am, in o(w)rə́, warə́ ʻa cloudʼ (*abrám), zǝ ʻIʼ (*azám).

from *ū, in mə́ǧa f. ʻratʼ (*mū́š- + f. *-ā), nən ‘today’ (*nū́nu PSR <

*nūnám), stǝn f. ʻpost, pillarʼ (*stū́ni-), stər ‘big, large’ (*stū́ra- PSR), wrə́ja,
wrúja f. ʻeyebrowʼ (*brū-čī- + f. *-ā).

- u in a closed syllable, in aγúnd- (*ā-gúnda-), búr ‘having lost a son,
childless’ (*apúθra-), gúta f., Wan. nəgút ʻfinger’ (*angúštā-), lur f.
ʻdaughterʼ (*dúxtar), etc.

The vowels *i / ī have become in Pashto:

- ø in an unstressed, open syllable, in mlúna f., Waz. vlī́na, Wan. awlūn
ʻbridleʼ (*abi-dānā-), wlešt, Wan. lwast f. ‘span between the thumb and
little finger’ (*witá

sti- PSR).

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

200

- ə́ in a stressed syllable, in bǝl, Wan. bī ʻsecond, otherʼ (*dwitiya-,*dwita-),
rǝ́ma f. ʻfluid excrement, diarrhoeaʼ (*rī́mā), x̌ǝ́ja f. ʻwomanʼ (*strī-čī- + f.
*-ā), žər, zər ’quick’ (*ǰī́ra- PSR).

- i in a closed syllable, in writ m. ʻroasted, friedʼ ( *brixta-), wit ‘open,
wide, ajar’ (*wišta-).

The effects of i-umlaut are most noticeable in:

- *a > *i, which would be further treated as old *i, see above;

- *a > i before *-ti, in 3sg. pres. –i (*-ati), nawí ‘ninety’ (*nawati-);

- *a > *ā in stressed, closed syllable, treated as old *ā, see below;

- *ā > e (stressed), i (unstressed);

- *ā before *(s)ti > yā, in atyā́ ʻeightyʼ (*aštāti), awyā́ ʻseventyʼ (*haftāti),
myāšt f. ʻmonthʼ (*māsti-), abstr. suff. -tyā (*-tāti);

- *au > e, in rebǝ́j m. ʻbroomʼ (*raupī-čī-), meǧáy m. ‘ant’ (< *maurijaka-);

- *au > *i > ø, in unstressed syllable, x̌na f. ʻleg, hipboneʼ (*srauni- + f. *-ā);

- *ahu > i in front of n, in wína f. ʻbloodʼ (*wáhuni-);

- *ṛ > *ir, the vowel would be treated as old *i, see above.

Palatalisation has affected the consonants:

- *k > č, in činjáy m. ʻwormʼ (*kṛmi-čī- + -ay), čə́rg m. ‘cock’ (*kargī < nom.
*karkah), mečə́n f. ‘hand-mill, quern’ (Gr. mēkhanḗ), etc.

- *s > š, in kišə́p m. ʻtortoiseʼ (*kasyápa-), mǝšáy m. ʻ(little) fishʼ (*mas-
yáka-), (w)šəl ‘twenty’ (*winsati-), etc.

- *z > ž, in niždé, nəždé ‘near’ (*nazdyah-), žay m. ʻleather bag, mussuckʼ
(*izyaka-).

The effect of u-umlaut is most noticeable in:

- *a > *ā́ in stressed, closed position > ó (treated as old stressed *ā́, see
above).

- *ṛ >

(

)

ur, the vowel would be treated as old *u, see above.

A different kind of u-umlaut may have been caused by the presence of
m, in view of circumstantial evidence:
- *a in closed syllable > *u > ə, in mǝǧ-, Wan. murš- ʻto rubʼ (*m(a)rša-),
mə́š(ə)r ʻelder, senior (person)ʼ (*masiiah- + *-tara), mǝšáy m. ʻ(little) fishʼ
(*masyáka-).

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

201

*AI AND *AU

As a rule, the diphthongs *ai and *au would have gone first to mono-
phthongisation to e and *o respectively.

*ai has become in Pashto:

- ø in unstressed position, in šna f. ‘green, blue’ (*axšainā́-).

- e, e.g. in γelé f. pl. ‘flocks’, lemə́ m. pl. ʻeyesʼ (ntr. *daima + -ə), lewár m.
‘husband’s brother’ (*daiwár-), wex m. ‘root, base’ (*waixa-), etc.

- i in front of n, in šin m. ‘green, blue’ (*axšaína-), win- ‘to see’ (*waina-).

*au has become in Pashto:

- wa, in γwaǧ m. ʻearʼ (*gauša-), lwaγ ʻmilkingʼ (*dáuga-), lwaš-ʻto milkʼ
(*dáučya-), rwaj f. ʻdayʼ (*raučah-).

- u before n, in γúna f. ʻcolour, complexionʼ (*gáunā), ruṇ ʻbrightʼ (*ráux-
šna-), yun m. ʻmovement; gait, stepʼ (*yauna-).

- ya in initial position, yaw m., f. yawá ʻoneʼ (*aiwá-).

ON THE PROTO-PASHTO INFLECTIONAL CASE SYSTEM: SOME OBSERVATIONS

From all these phonological developments it can be deduced that Pashto
at an earlier stage retained the inflectional case system inherited from
Proto-Iranian, of course with some simplifications. Proto-Pashto pro-
bably possessed the following (singular) case endings:

- vocative, as inferred from lur f. ʻdaughterʼ < *dúxtar, plār ‘father’ < *pi-
dár (PSR) < *pítar, and máṛa ‘oh man’ < *márta + voc. pcl. *ā.

- nominative masculine *-ī (< PIr. *-ah), cf. meṛǝ́ m. ‘husband; manly,
brave, generous’ < *mārdī́ (< *martáh), udə́, bidə́ ‘asleep’ < *(h)uβdī́ (< *huf-
táh), nər < *nárī (< *nárah).

- nominative feminine -a < unstressed *- ā.

- accusative masculine / neuter *-u (< PIr. *-am), cf. o(w)rə́, warə́ ʻa cloudʼ
< *aβrú (< ntr. *abrám), murγə́ m. ‘bird’ < *murγú (< *mṛgám).

- genitive > Pashto obl. ending -ə < *īya (< *-ahya). On non-final *-ah- >
Proto-Pashto *ī, cf. zǝr ʻthousandʼ < *(h)azī́rī (< *hazáhrah).

- ablative, cf. kará ʻin/from the houseʼ < *kārā́t.

The dative may have disappeared at an earlier stage, having merged with the

genitive (cf. Khot. and OP).

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

202

ETYMA REVISED

The phonological rules formulated here also allow us to reconsider the
etymologies of several Pashto forms proposed by Geiger 1893 and Mor-
genstierne (EVP; NEVP).

The following form(ation)s are probably neither old nor inherited:
- spā́nda f. ‘wild rue’ < early Persian / Parthian ?
- čə́na f. ʻthe outside of a house wallʼ < borrowing ?, cf. Pers. čīnah.
- carmə́n f. ‘skin, hide’ < Pers. čarm + suff. -mə́n.
- mu m. pl. ‘fat’ < Indo-Ary., cf. Waig. muī, Phal. mī ‘marrow’.
- mā́šay, myā́ša ‘mosquito’ < Indo-Aryan ?, cf. Skt. maśáka- m. ‘fly, mos-
quito’.
- px̌a f. ‘foot’ < Pers. pušt-i pā.
- šə́ga f. ‘sand’ < MP, Parth. sygd [sígad].
- hask ʻtall, highʼ < ha- ‘that’ + *uska- (rather than solely *uskah).

An old borrowing, not noticed previously, is
- həg m. ‘pig’ < MP/Parth. hwg [hūg] ‘id.’.

The following forms may have a different origin than suggested previ-
ously:
- γand- ‘to blame, criticise, condamn’ (< *gana- ‘slaying’ or expressive ?);
- ǧmǝnj, (Wan.) wženz, wženj, uženz, žmenz ‘comb’(< *pašána-čī ?);
- kwə́na f. ‘backside’ (< f. *kū́nā-, not borrowed from Pers.).

RELATIVE CHRONOLOGY OF THE SOUND DEVELOPMENTS: PRELIMINARY
OBSERVATIONS

Pending a full investigation in the historical consonantism of Pashto, we
may already suggest a relative chronological order for the following
selected developments.

- Pashto Stress Rule (PSR)

- au > *o > *wa > Pashto wa, wǝ

- stressed *ā́ > Pashto o

- merger of long *ī and *ū with their short counterpart *i and *u
respectively

- u-umlaut, notably *a > *u after m

- i-umlaut

- lengthening of (short) *a, *i, *u in closed (often) stressed position

- centralisation of *i and *u > ǝ

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

203

Abbreviations
abl. = ablative, acc. = accusative, adv. = adverb, Afgh. = Afghan, Afr. = Afridi
(Pashto  dialect),  Av.  =  Avestan,  Bal.  =  Balochi,  Bang.  =  Bangash  (Pashto
dialect), Buddh. = Buddhist, f(em). = feminine, comp. = comparative, Ghilz.
= Ghilzay (Pashto dialect), Gr. = Greek, Ir. = Iranian, Kak. = Kakari (Pashto
dialect), Khot. = Khotanese, Khw. = Khwarezmian, Kurd. = Kurdish, m(asc).
= masculine, M. = Munji, Mahs. = Mahsudi (Pashto dialect), Man. = Mani-
chaean, MP = Middle Persian, nom. = nominative, NP = New Persian, ntr. =
neuter, OP = Old Persian, Par. = Parachi, Parth. = Parthian, pcl. = particle,
Pers. = (especially classical) Persian, PIIr. = Proto-Indo-Iranian, pl. = plu-
ral,  Psh.  =  Pashto,  Oss.  =  Ossetic,  Sariq.  =  Sariqoli,  sg.  =  singular,  Sher  =
(Pashto dialect), Shugh. = Shughni, Skt. = Sanskrit, Sogd. = Sogdian, Wa. =
Wakhi,  Wan.  =  Wanetsi  (Pashto  dialect),  Waz.  =  Waziri  (Pashto  dialect),
Yaghn. = Yaghnobi, Yazgh. = Yazghulami, Yi. = Yidgha, Zaz. = Zazaki.

IBLIOGRAPHY

Aslanov, M. G. (1966), Afgansko-russkij slovar’ [Pashto-Russian Dictionary].

Moskva.

Bellew, Henry Walter (1901), A Dictionary of the Puḵḵẖto or Puḵs̱ẖto Lan-

guage in Which the Words are Traced to Their Sources in the Indian and
Persian languages, Lahore.

Cabolov, Ruslan Lazarevič (2001), Ėtimologičeskij slovar’ kurdskogo jazyka

[Etymological Dictionary of the Kurdish Language]. vol. 1: A-M,
Moskva.

Cheung, Johnny (2000), ʻʻSome Remarks on Gemination in Ossetic. The

Ossetic Reflexes of PIr. Intervocalic *-Ci̯-ʼʼ, Indoarisch, Iranisch und
die Indogermanistik. Arbeitstagung der Indogermanischen Gesellschaft
vom 2. bis 5. Oktober 1997 in Erlangen (eds. Bernhard Forssman/
Robert Plath): 69-74.

———— (2002), Studies in the Historical Development of the Ossetic Vocalism

(Beitrage zur Iranistik, Band 21), Wiesbaden.

———— (2007), EDIV: Etymological Dictionary of the Iranian Verb, Leiden-Bos-

ton.

———— (2010), “Selected Pashto Problems I: the Accent in Pashto”, Persi-

ca 23: 109-121.

CLI: Compendium Linguarum Iranicarum, Rüdiger Schmitt (ed.), Wiesba-

den, 1989.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

204

Durkin-Meisterernst, Desmond (2004), Dictionary of Manichaean Middle

Persian and Parthian, Turnhout.

Emmerick, Ronald, “Khotanese and Tumshuqese”, CLI: 204-229.
Geiger, Wilhelm (1893), Etymologie und Lautlehre des Afghānischen, Mün-

chen.

Gercenberg, Leonard Georgievič (1965), Xotanosakskij Jazyk [The Khotan

Saka language], Moskva.

Gonda, Jan (1966), The Meaning of the Sanskrit Term Dhāman-, Amsterdam.
Grjunberg, A. L.; Ė del’man, Dž. I. (1987), Osn.: Osnovy iranskogo jazyko-

znanija [Fundamentals of Iranian Linguistics]. Vol. IV: Novo-iran-
skie jazyki: vostočnaja gruppa [The New Iranian languages: the
Eastern Group], (ed. V.S. Rastorgueva), Moskva.

Hoffmann, Karl; Forssman, Bernhard (1996), Avestische Laut- und Flexi-

onslehre, Innsbruck.

Kabir, Habib; Wardak, Akbar (1999), Dictionnaire pashto-français, Paris.
Kuiper, F. B. J. (1978), ʻʻOld East Iranian *nāmani ʻnamesʼ, etc.ʼʼ, IIJ 20: 83-

94.

Lecoq, Pierre (1979), Le dialecte de Sivand, Wiesbaden.
Lorimer, John Gordon (1902), Grammar and Vocabulary of Wazari Pashto,

Calcutta.

Mayrhofer, Manfred, KEWAia: Etymologisches Wörterbuch des Altindo-

arischen. Vol. I: ADH, 1992; Vol. II: N-H, 1996; Vol. III: Vorbemer-
kungen, Jüngere Sprache, Register, 2001, Heidelberg.

Morgenstierne, Georg (1927), EVP: An Etymological Vocabulary of Pashto.

Oslo: Dybwad.

———— (1938), IIFL II: Indo-Iranian Frontier Languages, Vol. II. Iranian Pamir

Languages, Oslo.

———— (1942), “Archaisms and Innovation in Pashto Morphology”, NTS

12: 88-114.

———— (1974), EVS: Etymological Vocabulary of the Shughni Group, Wiesba-

den.

———— (1983), “Bemerkungen zum Wort-Akzent in den Gathas und im

Paschto”, MSS 42: 167-175.

———— (2003), NEVP: A New Etymological Vocabulary of Pashto, Compiled

and edited by J. Elfenbein, D. N. MacKenzie and Nicholas Sims-
Williams (Beiträge zur Iranistik, Band 23), Wiesbaden.

J. Cheung / Iran and the Caucasus 15 (2011) 169-205

205

Rastorgueva, V. S.; Ėdel’man, Dž. I. (2003), Ėtimologičeskij slovar’ iranskix

jazykov [Etymological Dictionary of Iranian languages], Vol. 2: b-d,
Moskva.

Raverty, Henry George

(1901), A Dictionary of the Puḵẖto, Pus̱'ẖto, or Lan-

guage of the Afg̱ẖāns; With Remarks on the Originality of the Language
and its Affinity to the Semitic and Other Oriental Tongues, etc., Lahore
(several repr.).

Sims-Williams, Nicholas, “Sogdian”, CLI: 173-192.
Skjærvø, Prods Oktor, “Pashto”, CLI: 384-410.
Steblin-Kamenskij, I. M. (1999), Ėtimologičeskij slovar’ vaxanskogo jazyka

[Etymological Dictionary of the Wakhi Language], Sankt-Peter-
burg.

Sundermann, Werner (1981), Mitteliranische manichäische Texte kirchen-

geschichtlichen Inhalts, Berlin.