OriginanddevelopmentofSanskrityy∗
MASATOKOBAYASHI
1
Introduction
Introduction
InA !.6.1.81to83,Pā)initeachesexceptional-ya-gerundiveformsofrootsendingin
-. The root vowel -, which is in the gu)a grade (-e) by A !. 7.3.84
sārvadȹātukārdȹadȹātukayo5(82 gu)a5),isreplacedwith -aybeforethe y-ofanaffix
onlyinthewordslistedthere.
A !.6.1.81k ayya-jayyauśakyārtȹe‘Ink ayyaundjayyafindetdieSubstitutionvon
ay für e statt in der Bedeutung „zu — möglich“’ (Böhtlingk 1887: 293). Exx.
k ayya-‘destructible’:TB3.11.7.4k ayyá-,1JB1.241k ayya-,MS1.8.6:124.4,KS
31.15:18.8,ŚB,TB,JBetc. a-k ayyá-. jayya-‘conquerable’:ŚB1.6.2.3,11.2.7.9,
13–20,28,29,31,14.4.3.24(K2.5.4.2,3.2.8.2) jayya- /jáyya-/,MS3.2.1:15.8,10,
TS1.7.5.4,5.2.1.1an-apa-jayyá-::JB1.3262jeya-.
A !.6.1.82 krayyastadartȹe‘In krayya- inderBedeutung„zumKaufausgestellt,
verkäuflich“’ (Böhtlingk). Ex. krayya- ‘for sale, available for purchasing’: MS
3.7.4:78.14 ékahāyanyākráyyā(Mss. ékahāyavyā-),23.7.4:79.1 apsúkráyyā[-a5]
ó adȹayo vái sómā [-a5] (Mss krayy), MS 3.7.7:85.10 vatsataré)a s)Sena
kráyyasyéndram áha5 krī)āti, ŚB 3.3.3.12 (K 4.3.3.12) somavikrayin krayyas te
somorājā3itikrayyaityāha.
A !.6.1.83 bȹayya-pravayyecaccȹandasi(Vārttika+hradayyā)‘Hierhergehören
auchdievedischen bȹayyaund pravayyā’(Böhtlingk).Exx. bȹayya-‘thatwhich
isfeared’:KS33.4:30.7,PB10.5.16,23.6.6.pra-vayyā-‘tobeimpregnated’or‘to
∗
IwishtoexpressmysinceregratitudetoGeorgeCardonaforhisadviceonnativegrammar,Werner
Knobl for reading the drafts of this paper painstakingly and for giving me many helpful suggestions,
especially on the interpretation of the cited texts, and to Alexander Lubotsky for his advice on the
historicalphonologyofIndo-Aryan.Theerrorsandmistakesareminealone,ofcourse.
1
TheoxytoneofTB k ayyá-insteadofexpected ×k áyya-isprobablybasedonthatof a-k ayyá-which
followsitinthesamesection(Knobl,p.c.).
2
Sincekráyyāhereisnotfollowedbyanudāttavowel,itshouldbeemendedtokráyyo(Knobl,p.c.).
bedriven’:Notattested.
TheseformsareofspecialinterestforthephonologyofSanskrit,firstlybecause
Pā)ini teaches them in A !ādhyāyī 6.1, which is a section mainly on phonology
(exceptionalalternation,sandhiandwordaccent),andnotin3.1,asectiononverbal
morphologywhereexceptionalgerundiveformsin -āyya-suchas ā-nāyya-(anameof
the ritual fire Anvāhāryapacana) are taught (A !. 3.1.127 to 131); and secondly,
because the alternation -i- : -ay- is taught here as special compared to the regular
pattern -i- : -e- as in √jay/ji ‘win, conquer’ : jeya- ‘to be conquered’, etc. In other
words,theforms jayya-, krayya-, bȹayya-etc.with/yy/ standincontrast withregular
jeya-,kreya-‘tobebought’,bȹeya-‘tobefeared’,formingminimalpairs3belongingto
thesamegrammaticalcategory.
From a diachronic point of view, the forms in -ayya- are actually no mystery.
Whitney (1889: 345 §963a) already mentions rightly that ‘[t]he original value of the
suffix is ia, and as such it has to be read in the very great majority of its Vedic
occurrences.’ Based on the metrical scansion of -ya- gerundives in the `gveda,
Seebold(1972:221)demonstratesthatthesuffixisactually-iya-insimplexformseven
afteralightsyllable,whilemonosyllabic -ya-isfoundinitsplacewhencompounded.
Since Sievers’s Law is not evoked after a light syllable, it can safely be argued that
gerundiveformslike jayya-wereoriginallyformedwiththesuffix -iya-.Thissuffixis
attachedtothefull-graderootsendingin -ay,and -y-iya-isthenshortenedto -yya-in
asyncopeprocessi>Ø/C(Edgerton1943:87‘theconverseofSievers’sLaw’).4
Verbal adjectives in *-io- or traces of them are found in other branches of IndoEuropeanlanguages,forexampleinAvestan(YAv. zaoiia-,OAv. zәuuīmacc.sg.‘to
becalled’),Latin(eximius‘distinguished’, genius‘tutelarydeity’)andGreek(hágios
‘holy’).Thissuffixisreconstructedeitheras*-iHo-or*-i-o-,astheoriginofthe*in
itcannotbeuniquelydetermined(Schindler1977:58,Lubotsky1997:142).5
3
Providedthateisphonemicallyrepresented/ai/.
4
Inthe`gveda,Icouldnotfindanydoubtlessexamplewherewhatiswritteniyshouldactuallyberead
/y/(cf.Edgerton1943:89),buttherearecaseswherey,whoseunderlyingformis/iy/,ismetricallytobe
read/y/,aswewillseein§22.
5
AccordingtoMayrhofer (1986:165f.),thesequences*-o-and*-i()o-arereconstructedas distinct
morphemesalreadyinPIE,forexampleinthedelocativesuffix.
Animportantfacttobenotedinthisconnectionisthatageminate-yy-isintheory
impossibleinSanskrit,for,ifthereisaProto-Indo-Iraniansequencelike*-aiia-(the
nucleusstatusof*iisleftunspecified),thefirst*iisannexedtothe*atoitsleftand
formsadiphthong*/ai/whilethesecond*ibecomesanonsetofthefollowingsyllable
(*/a/), creating the final output -eya-. However, -yy- is an acceptable and not
uncommonsequenceinPā)ini’sgrammar.Firstly,Pā)initeachesformswhichcontain
/yy/bothbyderivationalrulesandbydirectlisting(nipātana).Forexample,causative
stems form compounded absolutives in -ayya according to A !. 6.4.56 lyapi lagȹu-
pūrvasya (51 )er, 55 ay), and /yy/ here seems to make position because this suffix
occurs only when preceded by a light syllable (lagȹu-pūrvasya) and is apparently
conditionedbysomerhythmicfactorasinthecaseofthereduplicatedaorist;forms
with the sequence yy are explicitly given in many sūtras, such as A !. 6.1.81 k ayya-
jayyau śakyārtȹe, etc. mentioned above. Secondly, -ayy- and -aiy- must have been
distinct, as vhddhi forms like vaiyākara)a- ‘grammarian’ from vyākara)a- ‘grammar’
are provided for by A !. 7.3.3 na yvābȹyām padāntābȹyām pūrvau tu tābȹyām aic.6
Finally,theoptionaldoublingofconsonantsincludingsemivowelsafter/r/or/h/and
beforeavowel,taughtinA !.8.4.46 acorahābȹyāndve(45 yara5, vā),e.g. aryyá-for
aryá-m.‘lord’,showsthatthegeminationof/y/isasynchronicallyactivephonological
process. Based on these grounds, Pā)ini must have conceived of yy as a genuine
geminate, and there was a change in the phonological status of /y/ between ProtoIndo-IranianandPā)ini.
Thispaperaddressesthequestion,how/yy/emergedandhowitwasincorporated
into Sanskrit phonology as an acceptable geminate, from the viewpoint of historical
andsynchronicphonology.
2
OccurrencesofyyinVedic
inVedic
Wewillfirsttakeabriefoverviewofthe formscontaining yy,theirfirst occurrence,
andwhichofthemarenot/yiy/butreal/yy/onmetricaloretymologicalgrounds.
6
The `gveda already has one form with -aiy-, i.e. the patronymic váiyaśva- from the proper name
vyàśva-,butasthereareonlyafewcasesof-ayy-there(see§22),-aiy-hardlymakesaphonemiccontrast
with-ayy-.
a. [`gveda]: Gerundives in -yya- (Bartholomae 1907: 321ff., AiGr. II-2, 285
§173).Metrically-āyiya-except`V1.129.2bdak yya-‘whohastobepleased’
(-yiya-intheotherfiveoccurrencesoftheword),6.18.6dvitantasyya-‘tobe
contendedfor’(-yiya-intheotherthreeoccurrences),10.122.7c mahayyya-
‘aggrandizement’.
b. [`gveda]Gerundivesofrootsendingin -.PB9.1.21,22 a-saj-hāyya-‘notto
becollected[again]’vs.JB,JUB a-saj-heya-from√hā: jíhīte‘startup’.TS
2.2.10.2 dȹāyy-(A !.3.1.129)‘supplementaryverse’.`V4.29.5d ā-kāyyà-‘to
bewishedfor’.
c. [`gveda+]: -ya-derivativesin-āyya-(AiGr.II-2,794§642ef,795§643bα,286
§173c). While -ya- forms of roots in -ā usually end in -eya-, there are also
formsin -āyya-(-yya-/-yiya-/or -āyyà-/-āyíya-/).`V nh-pyya-‘protecting
men’, `V 5.66.6c and 8.27.22b bahu-pyya- ‘protecting many’. `V 8.34.5b
pūrva-pyya-‘firstdrink’(vs.`V1.135.4e,7.92.1d pūrva-péya-),`V8.17.13c
ku)Sa-pyya-nameofaritual(A !.3.1.130).
d. [Atharvaveda+]: -āyya-gerundivesofrootsendingin-orderivativesthereof
(AiGr.II-2793§642d,796§643bβ).AVŚ15.3.10pra-hāyyà-‘messenger’from
√hay/hi‘drive,impel’(vs.AVŚ,ŚB pra-héya-‘to bedispatched’inebelow),
AVŚ6.105.1dpra-vāyyà-‘forth-driving’(Whitney).Pā)initeachessuchforms
in A !. 3.1.123 and 127-131, i.e. A !. 3.1.123 upa-cāyya- ‘to be piled’ (?)
(cȹandasi),A !.3.1.127 ā-nāyya-anameoftheritualfireAnvāhāryapacana,
A !. 3.1.128 pra-)āyya- ‘thief’ (?) (Cf. ChU 3.11.5 prā)āyya- ‘trustworthy’, a
variant for pra)āya-), A !. 3.1.129 sājnāyya- a kind of dairy offering (TS
2.5.3.3 sājnāyyá-),A !.3.1.130 saj-cāyyà-akindofSomaritual,A !.3.1.131
pari-cāyya-‘sacrificialfirepiledincircle’(TS5.4.11.3pari-cāyyà-,KS21.4:42.7
pari-cāyya-),upa-cāyya-akindofritualfire(cf.Agrawala1953:370ff.).
e. [Yajurveda+]:Above-mentioned-ăyya-gerundivesofrootsendingin-(A !.
6.1.81-83,AiGr.II.2791§642cα,796§643bβ).kráyya-(forMSā-kráyya-,see
§11, footnote 2), bȹáyya- , jáyya- , a-k ayyá- (see §11 for reference). Of the
types jáyya- ‘conquerable’ and jeya- ‘to be conquered’, the former is
generally older, for it is attested in the Brāhma)as and the Brāhma)a
portionsoftheBlackYajurvedaSajhitās.Butthereisalsoanold-eya-form
from a root ending in -i, i.e. AVŚ 5.17.3c, ŚB 5.3.1.11 (ŚBK 7.1.4.12) pra-
héya-‘tobedispatched’(Bartholomae1907:322)vs.AVŚ15.3.10 pra-hāyyà-
indabove.
f. [`gveda, Atharvaveda]: Derivatives in -éyya- (Bartholomae 1907:320, AiGr.
II-2, 512 §344). `V 10.10.7b saha-śéyiya- ‘lying together’ (=AVŚ 18.1.8b),
stu éyiya-‘worthyofpraise’in`V10.120.6astu éyyampuruvárpasambȹvam
(Tri !ubh), and AVŚ 5.31.12b śapatȹeyiyà- /-yíya-/ ‘worthy of cursing’
(Whitney).
g. [`gveda+]:Whentaddhita-ya-isattachedtoastemendingin-ya,-yya-isthe
outcome in older texts (AiGr. II-2 807 §651a). RV 10.151.4c hhdayyà-
‘cordial’(3σ)from hdaya-n.‘heart’.TS4.5.9.1g nive yyà-‘belongingtothe
whirlpool’ from nive yá- m. ‘whirlpool’, and possibly TS 4.5.9.1g hradayyà-
‘belonging to the lake’ as if from an unattested (and unlikely) ×hradaya- n.
‘lake’.
h. [`gveda+]:Compoundedabsolutiveofcausativeandotherstems(AiGr.II2,781§635a).`V1.105.18c,3.26.1a ni-cyiyā.Cf.ŚB paly-áyya, ā-kramáyya,
palyya,anu-gamáyya,pra-janáyya,uj-jvaláyya(AiGr.II-2785§638a).
i.
[Brāhma)a+]:śayyā-f.‘bed’(A !.3.3.99sajjñāyāj...śīw-bȹhñi)a5,AiGr.II2 247 §142b), (-)śayya- absolutive, etc. (A !. 7.4.22 ayaw yi kwiti [21 śīwa5],
AiGr.II-2785§638a),from√śay(i)/śi(śī)‘belying’.
j.
[`gveda]: Patronymics of unclear etymology. vayyà-, always (6×) occurring
after caesura and to be read vayiyà- /vayíya-/ (another form in -ayyà- is g.
hhdayyà-).`V8.68.10dpuru-myiya-.
k. [`gveda+]: Irregular inflectional forms of rayí- m. ‘wealth’ (AiGr. III, 215
§120aβ, Lubotsky 1995). `V 10.19.7d (2σ), AVŚ2, MS, TS rayy inst.sg., TS
rayym loc.sg., ŚB rayyái dat.sg. These forms were created secondarily from
nom.sg. rayí5afterthemodeloffeminine -i-stemslike matí-:inst.sg. maty
andpu !í-:dat.sg.pu !yái.
l.
[`gveda+]: Inflectional forms of vhk-stems in -yī. `V 2.37.5a yayiyàm /yíyam/ acc. of yay- ‘running’, `V 10.155.2c arāyyàm acc.sg. of arāy-
‘blackguardwoman’.Cf.`V10.78.7cyayíy[a5nom.pl.
m. [`gveda+]: -yy resulting from external sandhi. Always to be read /-yi V-/ in
the`gvedaandtheAtharvavedaexceptin`V10.128.3b máyyāśrastumáyi
deváhūti5andAVŚ1.1.2d,1.1.3d máyyevstumáyiśrutámand4.25.3b tváyy
úditeprératecitrabhāno.
Inmostcasesof yyinthe`gveda,itismetrically/yiy/,butreal/yy/alsooccursat
least four times in the tenth book of the `gveda, (namely 10.19.7d rayy, 10.122.7c
mahayyyāya, 10.128.3b máyy āśr astu and 10.151.4c hhdayyàyā), once in the first
book (namely 1.129.2b dak yya5), and also once in the sixth book (6.18.6d
vitantasyya5).Sincebooks10,andpartlyalsobook1,belongtonewerlayersofthe
`gvedatextcorpus,wecanarguethattherewasnoreal yyintheoldeststageofthe
Vediclanguage,exceptin vitantasyya-ofthesixthbook,which,however,islikelyto
beoneoftheyoungerfamilybooks.Althoughreal yystartedtooccuronlyinnewer
layers of the `gveda, Śākalya’s redaction generalized yy to the older books and
changedall/yiy/toyy.
In the Śaunaka Sajhitā of the Atharvaveda, yy is usually to be read /yiy/,
exceptinafewformssuchas mayyV-, tvayyV-or rayy,andthegeminatestatusof
/yy/ does not seem to be established yet. In the Black Yajurveda Sajhitās and the
Brāhma)as, which do not contain many metrical mantras, it is difficult to judge
whether/yy/isageminateornot.
3
Developmentof*Developmentof*-ai-andrelativechronology
andrelativechronology
In order to understand how the gemination of /y/, which cannot be of Proto-IndoIranian origin, arose in Sanskrit phonology, it will be useful to review the origins of
Sanskritintervocalic/y/.HereIlistafewexamplesof/y/oflaryngealorglideoriginin
apostvocaliccontext,especiallybefore*i.
i.
Non-laryngealoffglide(*Ø>y/iV):
*ia> iya?:*i-á-dȹa>`V6.20.8d iyá-dȹyaidat.inf.of√ay/i‘go’.Thisform
probably does not serveas an example, for *a of the suffix *-ádȹyai is later
thanProto-Indo-Iranian,andalso,somethingirregularseemstohavetaken
placeinthereflectionoftheinitial*iasasyllablenucleus.
The non-laryngeal y in īyivájs- (`V 3.9.4a īyivjsam, 10.14.1a
pareyivjsam), pf.ppl.ofthesameroot,isnotusableeither,forconnective
-i-justseemstobeaddedtotheweakstemīy-.
ii.
Laryngealoffglidebefore*i(*H>y/Vi):
*iHi>iyi:nil.
*iHi>ī:*Hi-Hi-á>`Vījépf.1,3sg.mid.of√yaj/ij‘offer,worship’.
iii.
Laryngealoffglidebeforevowelsotherthan*i(*H>y/iV¬i):
*iHa> iya:*Hí-Har-ti> íyartipres.ind.3sg.act.of√ar/h‘startmoving’.*HiHy-a>iyāyapf.3sg.act.of√ay/i‘go’.PIIr.*priH-á->priyá-‘dear’.
Asidefrom*,alaryngealoffglideasiniii.íyartiisasourceofSanskritintervocalicy.
iv.
Laryngealonglidebetweenavowelotherthan*iand*i(*H>y/V¬ii):
Lubotsky (1995: 214ff.) discusses the development of *aHi in great detail.
Accordingtohim,thelaryngealinthiscontextisusuallylostandcontraction
of *a and *i to e takes place subsequently, unless morphological pressure
causesthelaryngealtoberestored.
*aHi> ai?:Inaugmentedformslike*á-Hisć-a-s>`V áiccȹa5ipf.2sg.act.
of √e /i ‘seek’ (trisyllabic in `V 10.108.5a), the augment is restored
accordingtoLubotsky(1995:223).
*aHi>ayi:*raHí-s>rayí5nom.sg.ofrayí-m.‘wealth’.Restitutionof*Hby
analogyisunlikelyaccordingtoLubotsky(1995:222),andsothisisprobably
theonlyexampleofuncontracted*aHí.
*āHi>āyi:*á-dȹāH-i>ádȹāyiaor.pass.3sg.of√dȹā‘put’.Itispossiblethat
ywasinsertedasaresultofanalogytootheraoristpassiveforms,like ákāri
from√kar/kh‘do’(Lubotsky1995:220).
v.
PIIr.*before*i(*>y/Vi):
*á-ćrā-i > áśrāyi aor.pass.3sg. of √śray/śri ‘lean’. Here also, * might have
beenrestoredbyanalogytootheraoristpassiveforms.
*ai > ayi : This is the usual outcome when a suffix beginning with *i is
attached to a causative stem, e.g. `V 7.81.6c codayitr f. ‘inciter’, 9.101.1b
mādayitnú-‘intoxicating’,10.162.6b mohayitvabsol.‘havingbefuddled’,etc.
Contraction no longer applies to productive formation patterns like these
(Lubotsky1995:219).
vi.
PIIr.*Hafter*:
*aHa>aya:*aH-as->váyas-n.‘vigor’.
*aHa > eya?: them.pres.opt.1sg.mid. such as *sač-aH-a > `V saceya of
√sac‘accompany’.Optativeformsofathematicstemhave ethroughoutthe
paradigm,andeinthefirstsingularmightbearesultofleveling.
*aHi> e/aï/:*ćráH-ištȹa->*ćráØ-ištȹa-> śré !ȹa-‘mostsplendid’(3σin
8outof31simplexoccurrencesinthe`gvedaaccordingtoLubotsky1995:
217.Cf.śréyas-2σ).
*aHi> e/a/?:*a-naH-iš-ta> ane ataaor.3pl.mid.(`V10.155.5a,4σ).As
we have only one occurrence of this form in the tenth book of the `gveda,
thepossibilitytoscaneinthisformasdisyllabicremainsopen.
*aHi> ayi ?:*kraH-ia->*kraØia->MS+ kráyiya-;*náH-ištȹa->`V
10.126.3c náyi !ȹa-. áyi in the latter form might be due to analogy to other
superlativeformslikejávi !ȹa-‘quickest’from√javi/jū‘bequick’.
*aH > ey: *ćráH-as- > *ćráØ-as- > śréyas- ‘more splendid’ (2σ. See
below).
*aHC > eC /aïC/: *náH-tar- > nétar- ‘leader’ (possibly 3σ in 4 out of 18
occurrencesinthe`gveda),*praH-tár-> pretár-‘friend,lover’(possibly3σ
in `V 1.148.5d, 2σ in `V 4.41.5b). It is not clear whether the disyllabic
scansionofereflectsvocalizationofthe*H(Lubotsky1995:218,Tichy1995:
36).
*aHC> ayi ?:*náH-tum>AB,TB -náyituminf.of√nayi/nī‘lead’(Werba
1997: 300). This is a late form and may not represent the original
developmentofthesequence*aHC.
*āHC> aiC :*(a-)nāH-s-ta> nai !a i -aorist of√nayi/nī(`V8.30.3d,2σ.
Narten1964:169.Cf.Narten1964:52‘istdieGruppe -āi-š-vielleichtschon
vorindischzu-āiš-kontrahiertworden’).
Whathappenedto*(<PIIr.*or*H)between*and*iinpre-VedicIndo-Aryan
is not an easy question, for we need to understand the relative chronology of the
developmentsoflaryngealsandintervocalic*atthesametime.Althoughwecould
not find any good example of non-laryngeal *i free from morphological pressure,
regulardevelopmentsof*aHiand*aHi(andpossibly*HCaswell)seemtobee/a/
ande/aï/,asweseein*daH-iHa->-déya-‘giving’asinvasudéya-n.‘treasure-giving’
(Lubotsky1995:215),and*ćráH-ištȹa->*ćráØ-ištȹa-> śré !ȹa-(3σ),respectively.
Onlyrayí5<*raHí-s,forwhichnoformseemstoserveasamodelofanalogy7,shows
ayíinsteadofexpectede.
The chronological order of the sound changes involving Proto-Indo-Iranian nonnucleus * can be summarized as follows. In pre-Vedic Indo-Aryan, non-nucleus *
was lost between *a and *i, including *i of laryngeal origin (if a laryngeal ever
becomes*ibetween*andaconsonant,cf.Lubotsky1995:218f.).Sincediaeresisis
stronglydisfavored,resulting/aï/wasthencontractedto e.Thischangewasstillgoing
onatthetimeofthe`gvedapoets,andwehavecasesofdisyllabicscansionof easin
śré !ȹa- or pré !ȹa-. On the other hand, gerundive forms like kráyya- came into the
derivational cycle rather late, when the loss of non-nucleus * between vowels and
subsequentcontractionwasalreadycomplete.Soitwassubjectonlyto i-syncope,and
resultedinkráyya-.
PIIr.
*ćráH-ištȹa-
post-PIIr. dissolutionof*H
*ćráØ-ištȹa-
post-PIIr. *-lossbefore*i
*ćráØištȹa-
upto`V
contractionof*aïtoe
>śré !ȹa-
post-`V
i/u-syncope
*kráy(H)-iya-
>kráyya-
Otherformswithunderlyingorsurface/yy/areexplainablewithin,oratleastare
notinconflictwith,theframeworkofthisrelativechronology:
•
śréyas-, comparative of śr- f. ‘splendor’, was discussed by Pinault (1982: 268),
whoproposesaruleoflaryngeallossbefore*.If*-aH-becomesdisyllabic e/aï/
beforeaconsonantasthe ein nétar- <*náH-tar-possiblydid,wewouldexpect
toseeadisyllabicscansionofthe ein śréyas-too.However,thefirstsyllable śré-
ofthiswordismonosyllabicinallofitsfouroccurrencesinthe`gveda (3.8.4b,
5.60.4c, 6.41.4b, 10.31.2d). This fact is more naturally explained by assuming a
lossofthelaryngealthatwouldotherwisebecomeavowel*i.
7
Analogywiththeobliquecases(rāy,etc.)doesnotprovideanexplanationfortheshort –ă-ofrayí5,
andsoitispossiblethatrayí5isaresultofregularsoundchange.
•
pra-héya- (AVŚ+) in §22e might be a form old enough to undergo *-loss and
crasis,butitisalsopossiblethatitwasformedwith*-a-insteadof*-ia-(Seebold
1972:222).
•
-pyya-and dȹāyy-in§22b:Wedonotknowwhatexactlytheseformsare,butif
theycomefrom*-āH-ia-andif*āHi>āyiisaregularsoundchange,theycanbe
explainedbyi-syncope.
•
-nāyya- and -cāyya- in §22d: These formations are not clear either, and if they
come from *-āH-ia-, we are not sure what would be the regular outcome. As
theyarenotattestedinoldliterature,theymightalsohavebeenformedafter*lossandcontractionwerecomplete.
•
Productivegerundiveformssuchas kreya-in§11areformedinanew,remodeled
derivationalpatternwiththesuffix-ya-.
4
Changeinthepronunciationof*
Changeinthepronunciationof*
pronunciationof*~y
* is a conditioned allophone of *i in Proto-Indo-Iranian, and it was probably not
distinct from *i except in the nucleus status. The fact that y appears after *i as a
Hiatustilgerintheplaceofadisappearedlaryngeal,asinSanskrit priyá-orAvestan
friia-<PIIr.*priH-á-,supportstheviewthat*wasoriginallyaglideequivalentof*i.
The current text of the `gveda, which was orthoepically normalized by the
redactor Śākalya, does not give any direct evidence for the change in the
pronunciationof/y/(*),exceptthatitisreportedtohavebeenweakattheendofa
word: According to Pā)ini, A !. 8.3.19 lopa5śākalyasya (17 apūrvasyaaśi, 18 vyo5),
word-final/y/and/v/arelostbeforeawordbeginningwithavowelinŚākalya’sview
({y,v}>Ø/-{a,ā}]wdX[voiced]-/),e.g./yásmāyarkám/→ yásm⨠arkám.8Whileno
other major phonetic change is visible, the phonemic status of /y/ (*) might have
started changing by Śākalya’s time. Medial -yi- is already common in productive
formationssuchas māday-itnú-(§33v)or patay-i )ú-‘flying’,and yyhasmultipliedin
hiseditionofthe`gvedabysyncopeandsandhi(§22m),suggestingthat/y/canstand
onitsownindependentofandunaffectedbyadjacent/i/or/y/.Phonemicizationof/y/
seemstobecompletebythetimewhenminimalpairmorphemesof i-vs. yi-,asthe
8
AlsoinTPr.10.19 lupyetetvavar)apūrvauyavakārau,VPr.4.127 yavayo5padāntayo5svaramadhye
lopa5.
reduplicantsofiyāyapf.3sg.act.of√ay/i‘go’(`V+)vs.a-yiyasasdesid.impf.3pl.of
√yam‘hold’(KS23.5:80.16),starttoappearinlaterVedictexts.
AstoPā)ini,wesawin§11that yyoccursinformstaughtinhisgrammarand
that consonants including /y/ are optionally geminated after /r/ and /h/ according to
A !.8.4.46,whereas/y/afterotherconsonantsisnot(A !.8.4.47).Wearguedthathis
yyshouldbetakenatfacevalue,namelyasarealgeminatewithnoepentheticvowel
inbetween.AccordingtoVennemannandMurray(1983:520f.),asimilargemination
foundinWestGermanic,suchasOldSaxon settianvs.Gothic satjan‘toset’orOld
Saxonkunniesvs.Gothickunjis‘race’(gen.sg.),occurredasarepairprocesswhenthe
segmentontheleftofasyllableboundaryhaslowersonoritythantheoneontheright
oftheboundary(sat.janand kun.jisinthiscase),thesonorityscalebeingglides>>
liquids>>nasals>>voicedfricatives>>voicelessfricativesandvoicedstops>>
voicelessstops.Ifthesameargumentcanbeappliedtoourcaseof/-ry-,-hy-/>-ryy-,
-hyy-, the sonority of Sanskrit /y/ would be placed between /r, h/ and /l, nasals,
obstruents/.ItsuggeststhatsomespeakerscontemporarywithPā)inipronounced/y/
withconsiderablystrongerconstrictionthanthatofaglide,possiblysomethingclose
totheIPApalatalapproximant[j]orpalatalvoicedfricative[ȭ].
In the Prātiśākhyas, ŚCĀ 3.2.8 rephahakārau paraj tābhyām and VPr. 4.102
paraj tu rephahakārābhyām provide gemination of consonants including /y/ in the
same context as in A !. 8.4.46. Mention in the Vājasaneyi-Prātiśākhya is especially
interestingfromadialectalpointofview,becausethePratijñā-Sūtra,atextassociated
withit,teachesthatword-initial/y/and/v/andthoseinthecontextofgeminationare
pronounced with occlusion.9 It might be a precursor of the fortition of Old IndoAryan initial /y/ to /j/ inlater Indo-Aryan (cf. Varma 1929: 127f.). Furthermore, the
Prātiśākhyas describe semivowels as ī at-sph !a- ‘slightly contacted’ (ŚCĀ 1.1.30) or
du5-sph !a-‘poorlycontacted’(`Pr.13.10),whereasŚCĀ1.1.31ū mā)āmvivhtajca,
althoughinterpretedinvariousways(cf.Deshpande1997:148f.),referstofricatives
9
Pratijñā-Sūtra 9–10. atȹānta(5)stȹānām ādyasya padādistȹasyānyahalasajyuktasya, sajyuktasyāpi
repȹo māntyābȹyām, hkāre)a cāviśe e)ādimadȹyāvasāne ūccāra)e jakāroccāra)am/ dvirbȹāve ’py
evam/‘DererstederHalbvokale,d.i. y,istinfolgendenFällenals jzusprechen: a.initiales y,welches
nicht mit einem andern Consonanten verbunden ist (...); b. jegliches y, das mit r, h oder hi direkt
verbundenist(...)c.wennesdoppeltsteht’(Weber1872:78f.).
as open. If those terms are to be taken literally, occlusion of /y/ might have already
beenunderwaywhenthosetextswerecomposed.
5
Discussion
Discussion
5.1
5.1
Asymmetricaldevelopmentsof*
Asymmetricaldevelopmentsof*and*
and*inIndoinIndo-Aryan
Aryan
InSanskrit,/y/and/v/aregroupedassemivowels(anta5stȹā-)togetherwith/r/and/l/,
andithasbeenarguedthatSanskrit/v/wasoriginallyabilabialglidejustas/y/wasa
palatal glide.10 On the other hand, the Taittirīya-Prātiśākhya and the VājasaneyiPrātiśākhyadescribethattheteethareinvolvedinthearticulationof/v/,11soSanskrit
/v/ has become a labiodental fricative by the late Vedic period, at least for some
speakers. Furthermore, there is significant asymmetry between the phonological
alternationsof/y/and/v/eveninearliestSanskrit,anditmakesussuspectthatProtoIndo-Iraniannon-nucleus*wasstrengthenedfromveryearlytimeson:
a. Phonotactically,vyisacommonsequence,whilethereisnoyvinearlyVedictexts.
b. /y/ can be geminated in Sanskrit while /v/ cannot, except in the postlexical
gemination(§5.2
5.2).
5.2
c. Thereisnovoriginatingfromalaryngealonglide(*V¬uHu)whileyinrayí5might
beaphonologicaldevelopmentof*raHí-s(§33iv).
d. The first *h2 in PIE *pth2ih2 develops into Sanskrit i in phtȹiv- f. ‘earth’ as a
laryngealtrappedbetweentwoconsonantsnormallydoes,whereas*h2betweena
consonant and * is lost as in PIE *-dh2é-ti > (ava-)dyáti pres.3sg.act. of √dā
‘divide’(Pinault1982:266ff.).
e. We discussed in §33 above that PIIr. * between * and *H(i) is lost and
contraction takes place subsequently,12 while * is retained as v in the same
environment.
*áH-ti>véti‘seeksfor’
vs.*mlaH-ti>brávīti‘says’
*naH-tar->nétar-
vs.*taH-s-iH>távi ī-f.‘power’
*(a-)nāH-išta>nai !a 10
vs.*á-pāH-iš-hs>apāvi ur(Narten1964:
169,52)
Whitney(1868:75)‘itsoriginalsoundwasthatofour w’,Allen(1953:57)‘itsearlierpronunciation
wasdoubtlessasabilabial[w]’.
11
TPr.2.43o !hāntābhyājdantairvakāre,VPr.1.81vodantāgrai5.
12
Inthecaseof*aH,itmightbethelaryngealthatislost(see§3
3vi).
3 *bhH-uka->bhvuka-‘becoming’(YV)
a)andb)mightbecasesofaccidentalgap,andthereareonlysingleexamples(rayí5
and phtȹiv-)eachforc)andd);buttheasymmetryinthedevelopmentofalaryngeal
after a glide in e) is hard to understand unless we consider that * had stronger
constrictionthan*alreadyinpre-VedicIndo-Aryan.
Anotherfactpossiblyrelevanttotheasymmetricalfortitionof*and*isthe
completeabsenceofvoicedcoronal(includingpalatal)fricativephonemessuchas[z]
[Ƞ()] [ʒ] [ȡ] or [ȭ] in Sanskrit phonology. Even where voiced coronal fricatives are
posited in intermediate forms in words like rīSȹá- vbl.adj. of √reh/rih ‘lick’ < *rihShá-<PIIr.*lih-tá-or mīSȹá-n.‘prize’<*miShá-<PIIr.*misdhá-,orinthesandhi
/s/ →[]→ r / -V¬a X{[voiced],¬r}-, they are systematically eliminated in the surface
representation,whilethereisnoknownrestrictionagainstthespirantizationof/v/.In
Middle Indo-Aryan, /j/ starts appearing in the place of Old Indo-Aryan /y/ as in
Ardha-Māgadhī jakkha- corresponding to Skt. yak a- (Pischel, Gr. §252), but we do
notseeanygradualfortitionof/y/goingoninSanskrit,letaloneanexplicitindication
of occlusion as in Holtzmann’s Law in Germanic (§5.3
5.3),
5.3 according to which
intervocalic*isdoubledandthenstrengthenedtoapalatalstopinGothicandOld
Norse.Justas rīSȹá-or mīSȹá-canbeexplainedbypositingvoicedcoronalfricatives
inintermediatestages,thedevelopmentof/y/to/j/wouldalsobebetterunderstoodif
weconsiderthatSanskrit/y/wasactuallyundergoingfortition,butduetothegapin
the phonemic system of Sanskrit, it remained subphonemic until it has completely
becomethestop/j/.
5.2
Changeinpreferredsyllablestructure
Changeinpreferredsyllablestructure
Syllabificationisatermcoveringbothassignmentofnucleusstatustoanunderlying
stringandadjustmentprocesses;however,theseprocessesaremotivatedbydifferent
principles sometimes contradicting each other, for example Sievers’ Law which
eliminatesoverlengthvs.postlexicalgeminationlike-ry->-ryy-whichcreatesclusters
at the coda. In our understanding, syllabification of Sanskrit involves the following
three processes. They partly correspond to chronological layers, but can also be
viewed as a synchronic set of rules whose priority may vary in different periods (cf.
Kobayashi2004:27).
•
Process 1. Nucleus placement (Kobayashi 2004: 21ff.): In Indo-European
languages, high vowels and sonorants (and sometimes even laryngeals) can
equallybecomeasyllablenucleus.Whenthereareoneormoresuchsegmentsin
arowunsyllabified,thenucleusisassignediterativelyfromrighttoleft,e.g.*imtá-‘reined’>*-tá-> yatá-,*HiuHn-ás gen.sg.ofm.‘youth’>*HuHn-ás>
yūná5, *HiuHn-tí- f. > *HuH-tí- > yuvatí- (Meillet 1934: 134–136, Schindler
1977: 56). This process, by which the rightmost segment becomes nucleus first,
reflectsapreferencethatthecodaofthesyllableshouldbeassmallaspossible.
ThisruleappliedcyclicallyinProto-Indo-Iranian,as śúna5<*ćún-asgen.sg.vs.
śvabhi5 (śvábhi5 in `gveda Khila 5.15.7b) < *ć-bhis inst.pl. of śván- m. ‘dog’
shows,butitisgrammaticalizedandisnolongeranactivephonologicalprocess
inSanskrit.
•
Process2.Prosodicwell-formednessandrepairrules:Whenstrings,includingthe
outputofProcess1,areprosodicallyparsed,ill-formedsyllablesorsequencesof
syllables are dissolved. Process 2a: According to the metrical scansion of the
`gveda, superheavy syllable rhymes are avoided and dissolved by Sievers’ Law,
e.g. srya-/sriya-/.Process2b:Also inthe`gveda,light syllableswith/i/or/u/
followed by /y/ or /v/ respectively are sometimes dissolved by syncope (Seebold
1972: 218ff.). E.g. hávya- /háviya-/ ‘to be called’ in `V 3.49.3c bhágo ná kāré
háviyo matīnm vs. 5.33.5d bhágo ná hávya5 prabhhthé u cru5, or anu-mdya-
/anu-mdiya-/‘tobeacclaimed’in`V9.24.4c sásniryóanumdiya5vs9.107.11c
anumdya5 pávamāno manī íbhi5. Such syncope is contradictory to Process 2a,
for superheavy syllables, like /md/ in the latter case, are created. In Śākalya’s
orthoepic normalization of the `gveda, Process 2b is generalized. Note that
Germanicalsoshowssyncopeof*-iji-/-ija-to*-ji-/-ja-afterlightsyllablesinClass
IweakpresentssuchasGothic satjan,OldSaxon settian‘set’fromPIE*sed-ée-
(Jasanoff1994:275).
•
Process 3. Syllable boundary rule: In Indo-Aryan geminates or clusters of
consonants of the same aperture are preferred across a syllable boundary. An
examplefromanoldperiodistheverbaladjectivevittá-‘found’vs.Avestanvista-
< *id-s-tá-, where interconsonantal *s is deleted (Mayrhofer 1986: 111,
Kobayashi 2004: 37). In the times of Pā)ini and the Prātiśākhyas, postlexical
gemination occurs in consonant clusters, e.g. mádhya-, márta- → máddhya-,
mártta-.Thisisthenewestoftherulesmentionedhere,forPā)inireportsinA !.
8.4.51 sarvatra śākalyasya that Śākalya does not accept such gemination in any
context,anditisonlyoptionalforPā)inihimself(A !.8.4.46,47).
Ifwetake hávya-/háviya-/asanexample,theProto-Indo-Iranianstring*ȹáuHiia- is syllabified as *ȹá.i.a- by Process 1. Although it passes the check for
superheavyrhymes,thesequence/iy/isnotfavoredinŚākalya’slanguageandpartly
inthe`gvedaitself,soitbecomes/háv.ya-/byProcess2b.Finally,the/v/inthecluster
isgeminatedas/háv.vya-/accordingtosomegrammariansbyProcess3.
Oftheseprocesses,syncopetakesplacein2b,andthatiswherethegeminate
yyiscreated.Since2bwasalreadyactiveinŚākalya’sdayswhenthepronunciationof
/y/waspresumablystillnotsodifferentfromaglide(§44),fortitionof/y/startedonly
afterthesyncopeof/i/and/u/before/y/and/v/hadtakenplace.
5.3
Polarizeddevelopmentsof*Polarizeddevelopmentsof*-i-outsideSanskrit
outsideSanskrit
InAvestan,Proto-Indo-Iranianglide*remainedassuchlongerthaninIndo-Aryan
(§5.1),atleastmedially,e.g.Av.gaoiia-vs.Skt.gávya-(Meillet1950:71ff.),andthere
was no need for resyllabification as in Sanskrit, where * did not form a diphthong
withthepreceding*a(×góya-)butmadeaclusterwiththefollowing*.13
Itisdifficulttoidentifydoubleglide*inAvestan,mainlybecausebothsingle
anddoubleglidesarewritten iior uuinthe‘Sasanianarchetype’spelling(Hoffmann
and Narten 1989: 43ff.). And even where double glide * is expected, Avestan
simplifies almost all gemination, and there seems to be no doubtless case of
geminated glide * there. So for example, sraiiah-< *ćráH-as- might represent
doubleglide*,butitmightalsohaveundergonedegeminationinIranian(Hoffmann
1976:615).14
13
Inotherwords,sonorityrankingwasdifferent:Indo-Aryany>{v,r}vs.Avestan{y,v}>r.Note
thatasemivowelsequenceisalsodissolvedbyanaptyxisasin daēuuaiifor*daēuuii (Reichelt1909:
72,Skjærvø1997:117f.,deVaan2003:541).
14
Pinault(1982)considersthatPIIr.*ćráH-as-,afterlossofthelaryngeal,wasdegeminatedalreadyin
PIIr.,andbecame לráyas-inIndo-Aryan,andthat ewasthenreintroducedintothefirstsyllablefrom
otherformslike śré !ha-.PositingdegeminationinPIIr.wouldbeunnecessaryif לráyas-,whichmust
OutsideIndo-Iranian,singleintervocalic*islost(asitiswhenthefollowing
vowel was *i in pre-Vedic Indo-Aryan) while *i that ended up between vowels is
pronouncedasambisyllabic/yy/inAtticGreek,e.g. treîs‘three’, āeí‘always’, pleîston
‘most’ vs. Atȹēnaíos ‘Athenian’ (Allen 1987: 81f.), and Latin, e.g. trēs < *trées vs.
Pompeiānus/ĕyy/<*-aā-(SommerandPfister1977:87,Allen1978:38f.).However,
there is a cross-linguistic tendency that geminates are more resistant to changes
(Kenstowicz and Pyle 1973), and ambisyllabic development of /i/ to /yy/ between
vowelsmightnotbelimitedtoIndo-Europeanlanguages.InGermanic,intervocalic*
is often lost as in Greek or Latin, but * after a short vowel is geminated in several
wordsbyHoltzmann’sLaw,andfurtherundergoesfortitioninGothicandOldNorse,
e.g. Gothic twaddjē gen.pl. ‘two’, Old High German zweiio (Krahe 1948: 87f.).
Although the principle behind the gemination is not clear, it can be compared to
Indo-Aryan developments such as OIA -aya- > MIA -e- vs. OIA -īya- > MIA -ijja-
(PischelGr.§153,§91).
6.
6.
Summary
Summary
In this paper, we have discussed the origin of the sequence yy in Sanskrit, viewing
problems such as intervocalic loss of * and subsequent contraction, syllable-related
alternationsofsemivowels,andpronunciationof*~/y/.Wehavearguedthat*was
originally a glide, which was at first lost between *a and *i; but it was gradually
phonemicized,underwentfortition,andbecameanapproximantorpossiblyavoiced
fricativeinitsdevelopmentinIndo-Aryan.
It is not fortition or phonemicization of /y/ but syncope that is directly
responsible for the creation of Sanskrit /yy/. When syncope of /i/ before /y/ first
occurredinthe`gveda,Sanskrit/y/wasintheprocessofacquiringphonemicstatus,
and /yy/ resulting from syncope survived without being resyllabified and dissolved.
Thesyncopeof/i/inturnresultedfromachangeinprosodicpreferencethatstarted
inthe`gvedaperiodandwascompletebythetimeofŚākalya’sredaction.Originally,
superheavysyllablestendedtobeavoidedasSievers’Lawindicates;butwhenσσ(y-/v-)
haveservedasthebasisofthevhddhiformśrāyasa-taughtbyPā)iniinA !.7.3.1(WackernagelKlSchr:
504)couldbeexplainedotherwise,forexampleasaneuter-as-stemfromtherootśr(Knobl,p.c.).
cametobepreferredtoσσ(i/u)σ(y-/v-),andσσ(y-/v-)toσσ(i/u)σ(y-/v-)evenifthefirstsyllable
would end up superheavy, syncope of the medial /i/ and /u/ occurred and created
manynewclusterswithsemivowels,including/yy/./yy/wasinitiallyaconcomitantof
syncope,butitbecamearealgeminateas/y/underwentfortitionandbecamefirstan
approximant,andpossiblyamoreconstrictiveconsonantsuchasavoicedfricative.In
thatsense,Sanskrit yyisaco-productofchangeinprosodicpreferenceandfortition
of /y/. Śākalya’s time is when word-final /y/ was lost between vowels while syncope
createdyy;acomparablesituationisfoundinGermanic,wherelossofintervocalic*,
syncopeof*ibefore*and(spontaneous)geminationofintervocalic*tookplace.
References
References
Agrawala,VasudevaSharana.1953.IndiaasknowntoPā)ini.Lucknow:Universityof
Lucknow.
AiGr. = Wackernagel, Jacob and Albert Debrunner, Altindische Grammatik.
Göttingen:VandenhoeckundRuprecht,1896–1930.
Allen,WilliamSidney.1953. PhoneticsinAncientIndia.LondonOrientalSeriesvol.
1.London:OxfordUniversityPress.
Allen,WilliamSidney.1978. VoxLatina.2nded.Cambridge:CambridgeUniversity
Press.
Allen,WilliamSidney.1987. VoxGraeca.3rded.Cambridge:CambridgeUniversity
Press.
Bartholomae,Christian.1907.‘ZurGerundivbildungimArischen’,KZ41,319–335.
Böhtlingk, Otto. 1887. Panini’s Grammatik. Leipzig (Reprinted in 1977 in
Hildesheim:Olms).
Deshpande, Madhav M. 1997. Śaunakīyā-Caturādhyāyikā. Harvard Oriental Series
52.Cambridge,Mass.:DepartmentofSanskritandIndianStudies.
Edgerton,Franklin.1943.‘TheIndo-Europeansemivowels’,Language.19,83–124.
Hoffmann,Karl.1976.AufsätzezurIndoiranistik,Bd.2.Wiesbaden:Reichert.
Hoffmann, Karl and Johanna Narten. 1989. Der sasanidische Archetypus.
Wiesbaden:Reichert.
Jasanoff, Jay. 1994. ‘Germanic’, in Bader, Françoise (ed.), Langues indo-
européennes.Paris:CNRSEditions.251–280.
Kenstowicz, Michael and Charles Pyle. 1973. ‘On the phonological integrity of
geminate clusters.’ in Kenstowicz, Michael and Charles Kisseberth (eds.), Issues in
PhonologicalTheory.TheHague:Mouton.27–43.
Kobayashi, Masato. 2004. Historical Phonology of Old Indo-Aryan Consonants.
Tokyo:ILCAA,TokyoUniversityofForeignStudies.
Krahe,Hans.1948.GermanischeSprachwissenschaft.Berlin:DeGruyter.
Lubotsky, Alexander. 1995. ‘Reflexes of intervocalic laryngeals in Sanskrit’, in
Smoczyński, Wojciech (ed.), Kuryłowicz Memorial Volume. Pt.1. Cracow:
Universitas.213–233.
Lubotsky, Alexander. 1997. ‘The Indo-Iranian Reflexes of PIE *CRHUV’, in
Lubotsky, Alexander (ed.), Sound Law and Analogy (Festschrift R.S.P Beekes).
AmsterdamandAtlanta:Rodopi.139–154.
Mayrhofer, Manfred. 1986. Indogermanische Grammatik, Band 1.2 Lautlehre.
Heidelberg:Winter.
Meillet, Antoine. 1934. Introduction à l’étude comparative des langues indo-
européennes.7meéd.Paris:Hachette.
Meillet,Antoine.1950.Lesdialectesindo-européens.Paris:Champion.
Narten,Johanna.1964.DiesigmatischenAoristeimVeda.Wiesbaden:Harrassowitz.
Pinault, Georges-Jean. 1982. ‘A neglected phonetic law: the reduction of the IndoEuropeanlaryngealsininternalsyllablesbeforeyod’,inAhlqvist,Anders(ed.)Papers
from the 5th International Conference on Historical Linguistics. Amsterdam: John
Benjamins.265–272.
Pischel, Richard. Gr. = Grammatik der Prakrit-Sprachen. Grundriss der IndoArischen Philologie und Altertumskunde I, 8. Strassburg: Trübner, 1900. (English
translation by Jha, Subhadra, A Grammar of the Prākrit Languages, Varanasi:
MotilalBanarsidass,1957).
Reichelt,Hans.1909.AwestischesElementarbuch.Heidelberg:Winter.
Schindler,Jochem.1977.‘NotizenzumSieverschenGesetz’,DieSprache23,56–65.
Seebold, Elmar. 1972. Das System der indogermanischen Halbvokale. Heidelberg:
Winter.
Skjærvø, Prods Oktor. 1997. ‘Avestica II. Yokes and Spades and Remnants of the
“TripartiteIdeology”’,MünchenerStudienzurSprachwissenschaft57,115–128.
Sommer, Ferdinand and Raimund Pfister. 1977. Handbuch der lateinischen Laut-
undFormenlehre,BandI:EinleitungundLautlehre.Heidelberg:Winter.
Tichy,Eva.1995.Dienominaagentisauf-tar-imVedischen.Heidelberg:Winter.
deVaan,Michiel.2003.TheAvestanVowels.AmsterdamandNewYork:Rodopi.
Varma, Siddheshwar. 1929. Critical Studies in the Phonetic Observations of Indian
Grammarians.London:RoyalAsiaticSociety.
Vennemann, Theo and Robert W. Murray. 1983. ‘Sound change and syllable
structureinGermanicphonology’,Language59,514–528.
Wackernagel, Jacob. KlSchr. = Kleine Schriften. Göttingen: Vandenhoeck und
Ruprecht,1953–1979.
Weber, Albrecht. 1872. Über ein zum weissen Yajus gehöriges phonetisches
Compendium,DasPratijnâsûtra.Berlin:KöniglicheAkademiederWissenschaften.
Werba, Chlodwig H. 1997.VerbaIndoarica.DieprimärenundsekundärenWurzeln
derSanskrit-Sprache. ParsI: RadicesPrimariae.Wien:VerlagderÖsterreichischen
AkademiederWissenschaften.
Whitney, William Dwight. 1889. Sanskrit Grammar. 2nd ed. Leipzig: Breitkopf und
Härtel.
Abbreviations
AbbreviationsandNotation
andNotation
*: Reconstructed form (Proto-Indo-Iranian unless otherwise specified); ×: Not
attested; >>: higher than; AB: Aitareya-Brāhma)a; A !.: A !ādhyāyī; AVŚ:
Atharvaveda, Śaunaka recension; Br.: Brāhma)a; JB: Jaiminīya-Brāhma)a; KS:
Ka!ha-Sajhitā; MIA: Middle Indo-Aryan; MS: Maitrāya)ī Sajhitā; PIE: ProtoIndo-European; PIIr.: Proto-Indo-Iranian; `V: `gveda; `Pr.: `k-Prātiśākhya; σ:
syllable(s);ŚB:Śatapatha-Brāhma)a;ŚCĀ:ŚaunakīyāCaturādhyāyikā;Skt:Sanskrit;
TB: Taittirīya-Brāhma)a; TPr.: Taittirīya-Prātiśākhya; TS: Taittirīya-Sajhitā; VPr.:
Vājasaneyi-Prātiśākhya; VS: Vājasaneyi-Sajhitā; YV: Yajurveda (comprising KS,
MS, TS and VS); yiy etc.: Spelled yy but pronounced yiy according to metrical
scansion.
© Copyright 2026 Paperzz