Allen - Phonetics in Ancient India - OCR PDF

—-
COVEKNMENT of INDIA
department of archaeology
CENTRAL ARCHAEOLOGICAL
LIBRARY
ACC rj £
Call noJ U2i4fl3S4=fiU
D.C.A. 79,
SCHOOL OF
ORIENTAL AND AFRICAN STUDIES
UNIVERSITY OF LONDON
Lomtim Oriental Series

Volume I
LONDON ORIENTAL 9FIRIBB ■ VOLUME I
PHONETICS IN
ANCIENT INDIA
BY
W. a ALLEN
Lfchtrn m CnmpiiTiMvt tjnjuiitwi
at lA# Sthwi e>/
Oriented mu} tlfrifiBi Stiuhti
7855
i+qi. 2140^4
Alt
GEOFFREY CUMBERLEGE
OXFORD UNIVERSITY PRESS
LONDON NBW VOEK TOFONTO
*953
Oxford Umvenil\ Presi, Amen Homt, London E.C. 4
otiicnp nir mw TOJKHit* mtvvm mmson*
WUMMay Cii.^m MUMUJ UUMI £111 n^wjt ttuun
Groffrty Cumhttitge, PubSiihcr to tkt UmvtTtily
OENTit 1 ' i*'ilUAL

LIB* r ■ m jHI.
Ace V
Duty £ff 7 ff'-3

n„r $<!('% Ml
OINTTUL aH<:im »*»<'*! lUAIa

LIBRARY tfKW OlLiiL
.AOCL Nft ~7Sff
flO* ... _
'£Ji&.--,
«UI
MiTSTO W Cxut uBritaes

PREFACE
Tms book h intended as n guide to the appreciation of the earliest
phoneticians. Whitneys pioneer expositions of certain of our
sources some eighty years ago arc acknowledged in the introduc-
ton* chapter; but a general reinterpretation has now Song been
overdue,.and this fact bin itself suggestive of the remarkable quality
of the Indian texts. For it implies thacthey display a level of phonetic
discourse beyond the full comprehension of Whitney and lib con¬
temporaries, such as only the advances of the late nineteenth and
the twentieth centuries enable ns to appreciate today * The recogni¬
tion that analyses so advanced hi! their technique should have been
evolved at so early a dale may Well inspire a salutary scientific
humility B and it would he m once arrogant and pessimistic not io
expect that a remterpretation will again be necessary in another
eighty years -or even eight.
I ant grateful tu Professor J- R, Firth for the encouraging interest
he has shown at all stages in the progress of this work, and no less
for hia constructive suggestions; and i am happy to acknowledge
the researches of Dr. Siddhcshwar Varnia, the stimulus of whose
published work has been augmented for me by the background of
his personal association with Professor Firth over twenty years
ago—an association wliich has led. in the light of contemporary
linguistics, to a fuller realization of the wealth that lies in the
ancient treatises My thanks are also due to Professor J. Brough,
who read the work in manuscript and made .1 number of helpful
criticisms, md to Mr. C. A. Rytinds ami Mr. R. H. Robins, w ho
generously undertook to read the whole of the proofs during my
absence in India.
Finally l acknowledge the gemôsuy shown by the authorities
of the Sti&ool of Oriental arfti Afrit an Studies in providing a full
subvention for the publication of this work.
# ., ►. w
W. S. ALLEN
IJEPARTMtVr OP FHQNETECS Atttl LU*CTI£Tfra
SCHOOL UP OMENTAL XMlI AFRICAN STITCHES
UNIVERSITY OF LONDON
19SB
CONTENTS
PREFACE? v
EDITIONS OF TECHNICAL WORKS REFEAJtHD TO IX
INTRODUCTION
0*0. The Grammatical Achievement l

O.i. The Indian influence tm Western Phonetics 5
0-2.. The Sources +
0.3. The Sanskrit Alphabet 7
0 4. The Principles of Description: 8
0.40. Word and Sentence 9

0.4!. Phonetics and Phonology u
042, Terminology *3
0.43. Order of Analysis iS
Chart of Sanskrit Alphabet 20
PART 1. PROCESSES
t.o. Mental 31
1.1. Physiological: 23
1.10. Class in ration 22
j.ii, Tntm-buccal: 24
1,110, Vow els and Consonant* 24
■ .ill. Fricatives 26
1.112. Semivowels 37
‘Liquids* 3J
1,113. Retroflexion 32
1.11. Rxirt-biiccal: 33
t.iao. Glottal 33
i.lji* Pulmonic 37
1.133, Nasal 39
AfimpSta 40
1.13. The vargas 46
»lli CONTENTS
P*HT IL LETTERS
2.0, Consonants: *s
at.oo. Pulmonic and Glottal
48
3.01. Velar
51
2.02. Palatal
5*
3.03. Retroflex
52
2,04. Dental 56
2.05. Labial
57
a.i. Vowels:
57
2.10. a
57
3.11. 1, LI
6t
2-*2. | 61
2.13. e,u; ai. an
6*
Part HI, PROSODIES
3.0. Definition
*5
3.1. junction: 65
3.10. Word- and Morpheme-junction 66
3.11. Initialitv and finality (19
3.12. Letter-junction: 70
3. iso. Consonants stop (tibftimdh&m)
71 i
3i!2!- rSconsonant {târahhakli}
73
3.122. Stop-|-nasiI ( vwiii)
75
3-123- Fricative—nasal
78
3.124. Stop4-fricative
78
3-125- Gemmation
79
3.2. Syllable structure;
79
3.20. Vowel and Consonant
79
3.21. Syllabic Division
Hi
3.22. Length and Duration
83
3.23. Quantity
%
3.24. Tone
87
3.3. Tempo
93
INDEX
9S
I
EDITIONS OF TECHNICAL WORKS
REFERRED TO
N<B. Where more Utah otic edition s* noted* an asterisk signifies that the
edition in 0111=3don lias been adopted for puipoftcu of Icjce and numbering
{cjtceptioita fire duiy indicated).
Apiia!i~£ih}ti (Ap< S \
Ed. Raghu VklKjVS i. 2, 1934* PP- i25
/Jf/wrrta’/VdniyM>a (/if) ( = SatmaMyd CaturddhdyiM)
FaI. and 1ml- W. D, Vthitnty>JAQS vii, 333 iL
Atfutnw-Prdtif&dtya (See p. 5‘ n- *)
■Ed- jind imL 5, K. Shashi Lahore, 1*139
Ed. Vr B. V. ShsBtd. Ijfiort, 19*3
iHidro-di dpi -SiMsd
Ed. and tr*l E. Sick. BertuiB 1892
■Ed. V. R- R. Dikihim and P. S. S. Ayyw. Poona. i»«
Ed. Sadtui Pam, JFS ii. ». 1935- PP rog

MtitoVArfjVO
♦Ed. F. Kjc Shfim. Skamboy, 1892-1909
With Ptadlpa of Kaiv^tc and PradfpoddyoLina of AnnambhHffft, ed.
P. P S. Surtri (Pan lh JAirifair 1^4) Madina, 1948
TnL V. Trapp [Ahndtui i~sh Leipzig, 1933
Pdir/ni f Pffrt.)
♦Ed and trsJ. O. Bdhtiingfe. *887
Ed. and tml S. Chandra V®*u- Allahabad, 1891 “7
Ed. O. DtihtUngtcr Lionn, 1839 40
Tnd L. Renou (Ease. !, i-iii). Puny. *94&
P#P3ptf>0-^>^5d < P5)
Ed. and trsL A- Weber, IitdhtJir Studitn. Wr 1858, pp. 345~7i
■With the and
t-araiuT, cd. R- P. Shaima. Hennro, 1937
With the S^PaîM and -Proton, cd- and in I M. Ghj£*h_
Calcutta, 1938
C/ron^+T^ jfotra
Ed. and tnl. -1 Weber, Siadimw vni, 1863, pp, 209 ff.
■ \\p(|| the commoiiao' of Halityudlt*it «d V. N. Shastri (J/rV. iidfro].
Calcutta. 1874
Prufr/fld ^fiHrri
A, WVbiiFp Ohrt tin zi<m swiVnrtt PtijeJ gthfrigt* pfomiththu C^mptn-
diumt das PralijM S&W {Ahk d. kZn, Ah. d. Wits. z. Birtm. 1871)
* EDITIONS or TECHNICAL WORKS REFERRED TO
PX,PrtUUOkhy<i iftP)
With the crrtturantary of IJvaja
•Ed. and iral M- I> She**"- Allahabad, <93*. UJjok, (937
With nrtrtto from Uvufa,
Jul, anti irtL F. Mas Mftllcr = Rtg-ltda vn(, t, pp. 1 -cum*
Lrnftti*. 1856 „ , , ,
Ed, and trel Regnkir, Jeitmni Atwtujue, V »*nr. wol*. vn-su
MJtintr<try^karaêj / AT)
■Ed S ft. Shaatn. Lahnre, 1933
Ed. A C. Uuriiell. MangxWe. i8?o
^ ETT^^t»‘4hury,jra ti. -=. t9J5, PP- tqfjS.

Sonuu^tmaiaîkjil
Ed. and tnl. A. O. 1'ranke. Gatuagw. iHS6
Sikfd-Sttntgralut iSS)
A ttaltrrtiw ./ SrZbdJ by Ydjltuvolkya crnd athm, ed, Yugnlalutara
Vrfsa (Hemiff ' A’Af Srru\), Benares, 1893
Tuittiriya^Pratiiilhkfa 1 TP)
With die Ttibh&tyitratna
‘lid. Lind rid. W. a Whnney, JAOS Lx. 187»
Ed. ftiiji;iul™Ula Mitra (BMintheca fruit«rt. Ctleurm, to?*
With the Padahttsmatudan* of MMiijcya.
Ed. V. V. Shftima. Madras, 1930
With the commentBriia of Stnmnyiry» ■*<& GArgy* GopUejwwn.
Ed. K. RanffnchiirymndR.S, Saitri {RillSonsh 33). Mysore. 190b
UptilMta
Ed and tral. W. Fertwh Betlio, 1854
V&j(UMMyir>Pr&iMkkyn or Kdlj&ymfjo-JVdt. (VP)
With tin- cnrtitrtentarix* of lint* anti AnaiiinfcihatPi-
•Ed. V. V, Sim mm- Madras, 1934
With extracts from the commentaries,
Ed.mdtral.A.Weber, InfrA* Studi**, iv, 1858* PP’ ^-'7L 177-
3lr
Paniopii(altm
Ttu- Poritiftm 0/ the AttwrvMtda, xlvii, ed- BoLting end NcrcIcih.
LeipriK. lOOO-to'’
Fyim-iS/M (FS)
■ H. Llidcra, Hi* VydtorSihfiL fiamnarn, 1894
lid. V, V. Shnmui- Madras. 1tm
lid- K. V. £asm t Grant ha terth Tinnmdi, 1‘JoS
INTRODUCTION
o.o. The Grammatical Achievement

In the sphere of grammar it is a gratifying custom of present-day
linguists to pay lip-service to the greatest of descriptive gram¬
marians. the ancient Indian Panini: anti it was an eloquent tribute
to Ills achievement that one of the great linguists of our own time
should write:
Ipdo-Kumpcan nunpmavr gramnnr had (and hasi at it* **rvi*.«i
one complete description »f a lauipuee, du: etaminar t,i
P.mmi. For all
other I mi o-European Itineumjes it had only the traditional k ram tun ri of
Greek and Laiin. wilfully incomplete and unsystematic, , . . Far iu> lan¬
guage of the past have we a twrtil comparable to Paijini» it cord of his
mother tnnpue, nor i* it likely tliat any language spoken today will be so
perfectly reeuftkd.'
But in spite of the invaluable translations of his work by Riihtlingk
and now by Renou, unless the linguist is himself also a Sanakriliat
there are imuipcmble difficulties in the way of a full appreciation
of Pitting achievement; and even for the Sanskriiist a complete
understanding is not easily attained—again lu quote bloom held.
Even with tire ninny women tones that w possess .. srvenJ lifetime*
of work will have to be spent upon Pflrum before we have a conveniently
usable cupcsilion «f die linguae whiel. hr recorded for all time-
It » indeed in the extent of the interpretative material, some of
which Has itself attained to a canonical status, that we find striking
evidence of the honour accorded to the great grammarian in his
own land.7 But this profusion of commentaries also bears witness
to the difficulties of Pinini's technique; composed with an algebraic
condensation, his work is a linguist's and not a language-tocher^
grammar, and for the more pedestrian purpose of teaching Sansknt
rewriting was a practical necessity, thus giving rise to further
grammatical hierarchies descending to a miscellany of school-
grammare of recent dated It has been calculated that there arc in
existence over a thousand different Sanskrit works on Sanskrit
* U Bloomfield, r.onri«tfr, v. *7* ff , ■ _

* [ft the rrvJipoddyrt* of Nl^ji BfotW. '*««. same tm> thou«M! W
Afw |n„Lni’i Jf pLlkydyt, we have a mb-coinnicntiHy of nr> !«* thim «>e imitrit
tUnu«UidW» PrakriyjtuiumuK- Wwttojt SMh

rrttu!i> VaniJiraja’* La^kukaummiL
1 PH0HKTICS INANCIKNl INDIA
grammar,* all inspired, directly or indirectly, by Pimini's modej
beside such a concourse the thousand manuscripts of Priseian's
Latin Grammar,1 the pride of our western tradition, are hut a drop
in the grammatical ocean. We have also to remember that Pan ini
himself acknowledges a number of predeccsscra, whose work,
except for fragmentary citations, has been lost to us—Burntii has
listed by name no less than sixty-eight of these prs-Panincsn
grammarians;* well might the medieval philosopher Kumarila
remark,
We cannot think of any point of time totally devoid of some work or
nthrt dealing iiith the iiranirnntit jI ml* i treating of the different kind*
of rcotii unci affix t-$/
Httf commendable as the cause may be, the non-Sanskritist can

hardly lie expected to acquire die grammar of Sanskrit—for which
the Indian tradition prescribes twelve years’ study—to the sole
end ihat he may more fully appreciate the work of the 'linguistic
Homer",5 or of later 'grammatical sum a’,"
in phonetic*, we all too rarely look back lieyond the great names
of the nineteenth century—Henry Sweet, A. J. Ellis, Alexander
Melville BcJJ—except occasionally to honour n few lonely and half-
forgotten figures of the immediately preceding centuries./ We
justify some of Our more grotesque and inadequate terminology
(e,g* '(amts' and ‘medtaf by tracing it hack to the Latin gram¬
marians, sometimes as far os Dionysius Thras or even Aristotle:
but generally speaking the expressions of ancient phonetic thought
in the west have little to repay our attention or deserve our respect,
whereas Indian sources as ancient and even more ancient ate
infinitely more rewarding. And in this field the linguist is fortu¬
nate] y in a more advantageous position u> appreciate the ancient
achievement, in that the acquisition of a working knowledge of the
Sanskrit sound-system demands no very- considerable labour, and
in so far as there is a basis for general phonetic discussion which
there is not for 'general grammar"
1 Sytitmt of Sanikrit Gummarr p.

fidvitkuf.
1 L‘J Knlory ofObuW Sckofmkfp, i. 359,
T Tk* Aitultu StMpi c/ SamJvit Grcmrnttn*wtm pp, f.
* 7anfrbYrarrtikar tr*L Can&Sfltilha JhaF p, 3*6,
1 P. Ttuen^. Pdmm «ind th? Vtda^ p, 15.
* Ckjidwadw, rdnimt p. S3 (of I-ulyAvatia mnd
‘ CL D AtsmrnsmbEc, ‘Fotflirtren Phothf-tidim1. TPS pp, % *|
* Set further IkI*0 bdaw. * Cf- Vendryfc»< SSL xkr 8 f
introduction 2
0,1, The Indian Influence on Western Phonetics

Moreover the link between flic indent Indian ,nid the modem
Western schools of linguistics is cunskicrjbly closer in phonetics
than in grammar, For whilst Paninean technique* are only just
beginning to banish the incubus of Latin grammar, our phonetic
categories and terminology owe more titan is perhaps generally
realised to the influence of the Sanskrit phoneticians. The impact
of Sir William Jones's discovert' of Sanskrit on the course of
Western linguistics is well known ; Inn Jones, apart from hla know ¬
ledge of the Sanskrit language, was also acquainted with the tradi¬
tional statements of it* sound-system: in his ‘Dissertation on the
Orthography of Asiatic i-ouii* s:; Ivuhiei LettSO "ir 'vhule Or4ot
of treatment and descriptive technique is clearly based on Indian
models. In a paper on 'The English School of Phonetics'’ Pro¬
fessor J. R. Firth has said of this great orientalist.
Without the Indian grammitnans and phonetician* wh<no he introduced
■nd recommended to us, it it diffieoit to imagine our nineteenih century
school of phonfika *
The influence nf the Indian works on the phonetic views of

William Dwight Whitney may be dearly seen in the discussions
published in live Journal of the American Oriental Society during
the yean iS6a-6, subsequent upon the appearance of l.epsiua's
Standard Alphabet and we have the feeling that without their
teaching Whitney might nut have been in a position to express self-
righteous indignation against that oilier country irom whicti he
had learnt so much—
It is f atty nnturinK how some nf lh< most able ph}"*ioIogist4 anil philo-
|nCjiW «f that notion fi.c, Germany! have bluiuknci ovtr the simple wd
seemingly obvious distinction between no t and a s, an / nod 11 t , rs p ami
a h etc.*
The "seemingly obvious’ distinction nf voiced and voiceless here
1 TPS 1^461 pPr *>4 ff~

‘ It n remurlaWe jhiu o < uudy of the Enjlith S*h.xjl should :.ul to
my :tf«rrore wlisloVcf to thr iinicofl influence lK Hiudnrtzky, Dir ltd!.
Szrmtcht Stkut*: fin ftnfriOf jur Grttfoditr dn rnitlntht* t'kantlih, XLubuig
! ** vr sttmttwd Alphabetfir rfdmirv Umnittm Fi*He« Ompkit

SyUrmt Jr, a tjmjvim Qi digraph}/ fa Purtypran fattiw' {London, j£A$).
4 JAQSviL 313. Cf T>auj, Am- Phit. Altar,, 1877. pp. *1 ff.
4 PHONETICS IN ANCIENT IN Hi A
referred to was subsequently rcCOgnuwjd by Lepsius as 'derived

from die Sanskrit grammarians’.1
In England the Indian inline nee is evident in the work of A. J.
Ellin, especially in Fart IV uf his Early English Pronunciation,
where frequent references are made to Whitney's translations of
the ancient works and also tn bis own olwervations nn ihr: speech
of latter-day pandits. He displays on occasion a fuller appreciation
of the ancient statements than Whitney !iad done, and generously
remarks on their descriptions of ‘voiced h\*
The wonder u, not that they chnuld he indistinct, but that they should
have Item mi much mto* distinct than lire heat of European irrammmian*
aisd unl.iirpiii.s who nurtoKltd ihi-m
As yet. however, the linguist cannot survey die Indian phonetic
achievement without undertaking an extensive course of reading,
of which only a certain proportion will In: relevant to his purpose;
ami on ihe oilier hand, without having viewed the overall frame-
work uf the Indian analysis he can hardly assess individual descrip¬
tions The principal works have been translated -ind commented
upon (so far as the phonetic climate of the translators environment
permitted) by such Western scholars as Whitney. Weber. Regnicr,
and Mux Midler, and more recently by Indian scholars amongst
whom may tic especially mentioned M. D. Shastri, S, K, SI i asm,
and A!, Ghosh. An interesting selection of special problems lias
been discussed in detail by Siddtieshw.ii- Vanina in his Critical
Statin f in the Phonetic Observation; of Indian Grammarians, The
pro i-nt study limr .it presenting a systematic account of Indian
phonetic doctrine so far as it appears to possess more than purely
Sanshritic interest. Where Western antiquity provides any parti¬
cularly striking parallels or contrast*, some account of these has
been given with a view to the comparative evaluation of the Indian
statements, Occasional discussions related to later Indu-Aryan
developments have been inevitable in establishing a control for
the pronunciations described in the treatises.
o.z. The Sources

Of the works themselves it will lie sufficient to note that they
fall into two main categories, die PratiiUkhyas and the kikjas,
• JAOS iroL j+a.
’ Stt further (, 1:0. below.
INTRODUCTION J
The former are phonetic treatises relating to the pronunciation of

the four Vedas, namely:
Rg- Vrdei Rk-PrUliiaktiya
S&ma-Vtda — Rk-tantrts^vyakarsna
f Black Yajur-Vaht;-* Taittiriya-Pmiiidkhyit
\ White „ —VRjmamyi- or
Katyaymlya-iâtisakhya
A thareif- l'W<i — A thurru - PratiiBkhya,1
The Siksas on the other lurid tire, with some exceptions, less
specifically related to a particular Veda, but in many cases supple¬
ment the teaching of the PrSiiidkbyas/ Whilst it is likely that she
I'ritisathyas are based on an early Siksa (such as that referred to
m the Taittima Arunyakaj,' our extant texts of the latter appear
to he of later dale than the former: Liu: most important of them,
the so-called Pi&inlya-is sometimes claimed as the original
&iksi and in consequence put back u> a very' early date: but this,
as a!ho its attribution to Plnini, is highly doubtful/ Vanna pbces
the PrSti&ikhya* in the period 500-150 R.c, and the extinct biksa
literature between 8oo and 500 B.c.1 It is significant tlwt one at
least of the extant Eriksas contain* the admission,
If yik?i and lYtuf ikhva arc found at vumiice, the &k$S » sajd to bt
the less authanpUive, u tin; deer is wtsfcst than the lion*
Apart front these specifically phonetic works, numerous state¬
ments on phonetic mailers are to be found in the pramniptJcal
works, mure especially in Panin is Afi&tlhyayl and Patanjali st
1 A* Whitney himself hits admitted. the tern which he he. edited itrd tiaiiv
IsttU under tin-, title n pifihitily net tit* -iP, and *0 should rtfktty he known lw
tht title ivliich it ten, vb. ^KwiiLyi Cerw4iflli«iii. Nercnbrlf:’, ij f <
if tuch if lw (eit S, K. Shaaln. V. H. V Shnrtri), eunttum. almost T.vCrun£ pf
KCiieraJ irttenst, and fur ptwtot purfe-srs the title siP B»J be I*(*tO*“ +tthnu!
la irlti to Wbimey"« editinti
1 Cf. Kklis&tn, /but. V 141 4, i#-mo [esp P
1 Vn. iu 1, (nfl filffifrn ttydkky4iait!rru>h Tfoff fubjeeb of tht Shh»3 urt wvai y-
wr4 [■w.m(J-imiVt. *™ r«mtbJp -uilrd fqiUOlCl?) butem.- idma Hivli-MM
Savjhii ifLlerprftb wiih ai Fdc^rcc af bocedi fjoauiw' (cf, i .f t htlow- h *»*&* 11
"tempo’ *ind uTTfiJrtTTuA m 'junction"
Cf- Sim KjOfww, Uia Oriauaiiot m* jv, mh3, v W-
4 Cf- THIunfe* op. cit.h y Bfr jk
1 tip, diH (rural,
* Swv*unfiai*toT*£$fydr cdr Franks, *o-
fikjd tfj ptdtiUikkyum m nmdhytU Jk^«tp4iroffi
hkpaco iforbzlriy dhih tm$itiir:arta rtt/jtf yutkti.
« PHONETICS IN ANCIENT INDIA
Mtikabh&iy&i and it h uvidvtit that in India, unlike Europe, gram¬

matical writers availed themselves of Lhc hesi professional phonetic
advice.' In such undent and mm4ectmia] wjjts as tlic Bilh-
mana^p Aramiikis, and Upani&aiia we find a familiarity with
various phonetic categories, c,g. 'Articulator1* "place of articula¬
tion', 'stop1, 'fricative'* 'seinivowclv ‘votvcF, and 'voice1 (in its
technical connoiatidUjf and in the White Yajur-Yeda there appear
in sin unutomkal list various features which belong to the structure
of articulatory and not of general physiological analysis*1 e.g. Teeth-
nms\ ’alveoli'/ and parts of the tongue, of which the tip is appro¬
priate tv associated with Sarasvath the goddess of speech.f
I In aphorisms of the phonetic works are at unce prescriptive
and descriptive, " tTiuir avowed purpose is to preserve the >>ral
tradition of the Sacred texts: to this end the dirvai penalties are
threatened for mispronundatbn, including descent to the hell of
KumbiupiIU;* ihe competent pupil, on the other hand* b on-
Cduragcd by verses nuch as that which doses the TP-
Me who know* ihe tiiitinctifltn of t'Jiw :tnJ length mjy go ind til with
ihc professors/
-a felicity which the commentator interprets as applying nut only

in tins world hut iilso in ihe next, However, the authors of our
treaties were clearly phonetician'; nil her than priest*, alula sdenti-
fie curiosity, coupled with keen audition and an effective methodo¬
logy, led to description;- which must surely liave transcended their
original terms of reference*
\of are the accounts of the various authors identical; we find
considerable divergences of pronunciation as between one treatise
and lilt: next, and we may surmise that these disagreements in
man; cases reflect actual clhkccaJ feature; corresponding to the
locution of I lie several SaMiaS or Vedic schools* In such nutters
there is a singular tack of religious dogmatism, and the authors
1 Fatttfijajri-tM.i It I- OH hin i. iL^j, Kteliwni, i, loSiiuilue, iLcunportjint

F''!ril 1 sTUjLL‘ uf ifranimaii prr’HiiPpfrv * ijiJtejyuile phonetic tducdTi&fi
3 f8k fioputfM Hr. i. 2^ \itkitidxttprvdiUL2■ ft^ruhum}; ,4i/ Ji. m. jj, jd
n ml x-$ up^TiiT^
il-^rem. mtaiihd, morn, fthotGVfJt), Cf, Ucbjch-
tMtfiihnms iH die mi. fitihrimitcfts SpmthudufnjLehtfrt u VVcb^r* lmi. St iv
n
A
t.l LIjflJIilIcv, Otaaitrufe1 SpttijgtrQTWil: firujii/'.r^riir, p, jj,
On iHpfcc Icrmi *« further n,qj hchfth
ft VS *X*r
*■ * r*4m hS- {£$.> p. dj>,
x* iv, fa. nura- ?:*fttil-nhhtif '' jfio i'iui/t/d diiby'v*itjrn*(tJam.'
INTRODUCTION 7
refer to each other’s opinions in a coinmendnbly objective manner.

Certain pronunciations, however, are generally recognized as
faulty, and lists of such faults (e.g, in chap, xiv of the RP} are
hardly less interesting titan the details of the approved pro¬
nunciation.
The Pratisakhyas have received the attention* of various later
commentators In so far as tlitre are the bearers uf a continuous
tradition, they are able to augment and elucidate the laconic
brevity of the aphorisms: unfortunately, howev er, the main stream
of she tradition see me in many cases io have been lost, and the
commentaries that wl- possess have a habit of wrapping the obvious
in obscurity instead of casting light on the numerous difficulties.
Morcovcr. it is clear that the intellectual climate of phonetic study
lind undergone a marked deterioration between the time when the
treatises were composed and the lime of our commentaries. In
general wc may say that Henry Sweet takes over where the Indian
treatises leave off—though in some matters even Sweet could have
learnt from them: and a recent study of a modem Indo-Aryan
language has successfully shown that many of the ancient descrip¬
tive techniques can still be employed to advantage ! These early
phoneticians speak in fact to die twentieth century rather than to
the Middle Ages or even the mid-nineteenth century, and many a
statement which die commentators and even Whitney or Max
Muller have failed to comprehend makes immediate sense to the
phonetician today. The one outstanding exception to the general
mediocrity of the Indian commentators h Uvapi, whose interpreta¬
tions of the RP and of the VP reveal an enlightened ami enlighten¬
ing approach to a variety of phonetic topics
0.3. The Sanskrit Alphabet

Whilst the statements with which we shall be concerned are of
wide phonetic interest, even the most general of them are of course
based on the description of a particular language, namely Sanskrit.
On p. io a chart is therefore provided showing the basic system of
sound-units 33 generally assumed by our treatises: certain diver¬
gences from ihi* system will be considered in their appropriate
place. As regards the transcription, two conventions hove been
* ft. S, Prasad, A Pbmttif tituJ Pk'it<'lospf<d Study f t flfcnj'frtirfp '■*!»-

milted ft* the Ph.D. Deere? nE (He b'nivpndiy uf London.
g PHONETICS IN ANCIENT INDIA
employed: for purpose* of textual quotation h the atandnrd Roman

Transliteration of the Devanagarl text is used fin italic type)—this
vr ill not ftenendly concern the noii-Sanskritbr* as the texts will be
translated and the original Sanskrit unless it calls for special com¬
ment* rvkgftted to fLK>tnule&. Where, however, Sanskrit sound*
and smiml-ûenccs tire made a subject of phonetic discussion, a
irankerjption h nsed which departs in some tcopeck from fhe
standard synem* and which 1 have found convenient in the teach¬
ing of Sanskrit phonetics: such i ninscripi ion & fire primed in heavy
LP'Af type. In the chart on p. «o the two conventions are shown
side by side. Where narrower transcriptions arc required, these
ore indicated by the use of square brackets.
It should be stressed at this point that, except for ininacrip-
tionnl purposes, the representation of a complex structure by
category* labels based on a monosysEcmic analyse is an unaccept¬
able procedure, which ima nevertheless- been adopted by die many
modern linguist who favour an exclusively * phonemic' approach:
the reason for setting up such a system in our chart is that the
Indiana themselves have done so. It is true that the Devanagari
me thud of writing s& syllabi c, hut the analysis underlying it and
actually :net out, fur example, ill the Tunuiômimnilyn nr alphabet1
ai the loginning nf Fan ini's grammar, comes very near to that
which a modem ^phrocmicist' would evolve for Sanskrit by a
su bsritutinri td- d i stributional an ah1 sis of the ward-isolates.1 How¬
ever* we can hardly criticize our predecessors of soma fovn mi Tlmn fa
HjZO for a procedure which only a few linguist:?- in the lost two
decadra have begun to reject os inadequate;5 and we xliall see that
the Indians, unlike many ol their Webern successor, appreciated
thiEt this* Technique was a means to a limited practical end,, and by
no means the ultimate analysis.
0.4. The Principles of Description

Wc come now 10 a consideration of the fundamental principles
of analysis mid description as postulated and as observed by the
authors of our treat!sea.
' t’i M. n. Erne:.!*!*, "Hie N*«»l iTvifinnt, oi S/in^nt'. Lix^lgr, ttii

S* ff-- A, IJ Eiy. ’A Fhoimmic iiif(fpf(rtitwn «f Yv-wi'. Li^v. ivii. a 74 ff.
K I I H !': rt I fphnnnl-iffrca3 t«ftirc-i ■«: jmug Indisr, Eungu^gf*',
■■ PT-f Ifit. Piu/tt Si\. pp. it. | Vt h. TxîSkil'ldcijrtiikff llit- I'jmitcinr',
jli&wfncr^ iv Ft, especially i:-p 54 r
INTRODUCTION' 9
0,40. Word md Sentence

III lmHv Indian linguistic djsoi$sion \vc find a full awareness of
the view that the basic linguistic unit, upon which all other
analyses must he founded^ the sentence; a famous Couplet of
fihartrhari’s treatise on genera! linguistics* the V<ikyapudlyot
where the matter is debated at some length states the ease in the
following terns:
Within the tound-Unit ihr eoiripctterit feature* lmvc no iniLrpendent
exist vtmc, ftoir the aowtd-untis wiihiri the wtini; m>r luivc the wards my
separate existence apart from the sentence,*
Tor purposes of phonetic description the basic unit is ako somt>
times stated as the "breath-group' {rka-pfma-hhaui)fz correspond¬
ing in, the Vedic hymns to one line of veree. rFhe tendency to deny
independence to the word is further stressed by the Sanskrit system
of writing which (unlike, e.gn Old Persian) takes no particular
account of word-division,* Thus word-sequences such as tuan
eva, tat fuirtah are written together as tauneva, tntpuuali, the
sequences -ne- and -tpu* being rep resumed as single graphic
units (il, f^)r Tim elimination of the inter-word spaces indica¬
tions of junction b part Lilly compensated by the graphic representa¬
tion of such phonetic junction-features as the uvaikblc symbols
are capable of showing: ihus the junction of tut—bfruvati is
written as tadbhuvmt* uit-hfrutvaa as tacchrumnt, tat+hl as
tnddfif, man—udukuih as modakaih, and so forth. Certain
Other jutlCtion-featlires art not generally indicated, sequences of
the type -h-f k- or -h+p- being only sporadically written es -x k-
or -^JU4 since X and $ are outside the phonemic and hence the
genera! graphic system. Even rarer h the indication of the linking f
prescribed by' the phimtiidans in iiequences such as taaâbruvan
(for tanh+abruvatl), which h generally written with hiatus as
taa ubruvan.' Elsewhere junction-featufcs may be neither writ¬
ten nor prescribed, so that no distinction is recognized between,
for example, na tena Hkhito lekhah, "he did not write the letter',
1 s 73. Ci Lt £fvJ, pr 20.
flr>«rti+
J !!' V. I. Of. Si&fetp Primer nf Fkomtm, $ KoUvckii. Principe* J/
phttfwifiqur f ip&imintrJx, p. 971 fL* mol , , . n'tsiSTC nhii rattan qufl Vtvnl
t>ol£- \jc Hreupc mrmtmrt1 [so^Bide uiw iHdividiiatiii parptt); Thumb,
Handhwft J«tj Sam^rii, J 160.
1 Cf. Whitney, Ski. Or.. | 9 p; Bloch. Lmiitde-Ar) *n. ftp. ?$ f.
* Cf. Widwmagel, Al Ci Jm6.
" Cf, W,ekcmBU^. °IJ- dft* $ afiJ b <0)>
ig PHONETICS m ANCIENT INDIA
and antenn likliUo Ickhah, * bowing, he wrote the letter*; this
fact provides material for die construction of various types of
word-play and fiddle based on alternative division* of the piece,1 * *
and our treatises do not mention any phonetic criteria by which a
distinction might have been indicated in utterance.4 Whether in
fact there wcrtsufatle distinct ions of profninctvccanchasarc cjp.tble
in English of differentiating* for example, a nativn from an oetm1
we cannot tell—is would certainly be unwise to deny the possibility
on the basis of an argummtnm ex ukntm—hut it appears certain
that in Sanskrit a delimitation of the word by purely phonetic
criteria was even less of a possibility than in English.
The Vcdie texts in fact const- dtiwn to us in two principal forma;
the SamhitauT Compound" text, with the sentence or brauh-gruup
a& its basic unit, and the Pada or 'word' tew, hliving the word-
isolate u tia basil; the latter is generally recognised to be an arti¬
ficial analysis devised by grammarians and others for [Hirposes of
instruction: that of the Rguda i$ generally attributed to the ancient
grammarian Siikalya.1 In the AP we find,
Tbt itud> or ihr w nrd-kokini it designed tn teach thr. 5 hd lining and
ctid* of words, and Lhetr corteet Iwm, tone end m«ming.s
in which the commentator adds,

Without studying the word-rtolufci one might make errors in the con*
hmifHii h xt; it ]> forth* rcuatm thut the smdy ofihe isolutcs if mtteswy *
Some statements of the rcbtiotiship of Pada to Samhitfi, however,

seem to have left room for misinterpretation: the RP makes the
highly auibiûmjs observation, tauikitR padupTakrtsh' r which
according to the interpretation of 'ptida-pYtriifiih’ {where
prahtih ^'basis') might mean diher "The Samhittl is Lhe basis of
the Pada or 'The Satiihi£3 has the Pada as its bask1; the term
pruhrti is also regularly used of the word-isolate in contradistinc¬
tion to rihdm (■modification1, Variant'), the latter being applied to
1 Cf. Mbh. 1.1 i [Kidhom, L i4h

* Thtiutfi ^ long a* ihc tom\ wymmn. nimvd. thm muM v\ wmAjf cue* h*v«
provided 4 mww of dhtmcTkrcii
1 Cf- Ck Jon£fl, *TUt word IS ■ [’hotlehc entity \ MF igii, pp, ff; Bk-om*
fwldh (rtmfrv^rr. pp nj f* iB*.
* Cf. L^rhkh, op- dT.r pp. io If
1 iv. 107. pododkywyimam mltiM - fjti: fa -it&rdrf/ui-1 imnarihtW.
4 •.. iWfddtibt* Piwtiayrl; lumfid rbtdfr k.iraTt^
T il 1*
INTRODUCTION if
iht juncti{jfi“farrm. A convincing solution to the difficulty i& pro¬

vided by the VmJiitfibharfn&i a commentary on the TP+ which
points qqi that as a result of Statements *Uch as the above Vermin
slow-witted [arsons Imve made die mistake of thinking that the
Veda is constituted of the word-isolates’, whereas in fact "the word-
isolates are only treated as a basis for the purpose of facilitating
instruction/1
Ilcine also should be mentioned the Krarua-pafhti or ^repetitive*
text, in the simplest form of which a word-sequence 12:3.4:5 is
recited in pairs as follows—1 2.2 3:3 414 $h with the realisation of
the appropriate j unction-features between the members of each
pair.3 This device forms an instructional stepping-stone between
the /WfJ and S&mhita tests; it appears To be held in no very high
esteem H and the most that the RP can say for it \%
'Hie Kramn i* nf no use to one who know* both the PuJu mul the
. . , tt doe* neither nur ilk and ha* no tucrcd iratlitiofi/
The recognised fund inn of die Pratmkhys* appears to have
been instruction firstly in die pronunciation of the wurd-isolates
and secondly in die mode of their synthesis in the sentence. The
first of these duties, however, involves the teacher in fun her
analysis, below the ward-isolate level: and since all analysis must
be followed by synthesis/ die TP aptly observes that there are
various types* of synthesis5—of words* of syllables, and of sound-
unita, to which is added as a fourth category, if we follow’ the com¬
mentator's interpretation/ die reintegration of by tluhtc-structure.7
Whitney, Failing to understand this passage, can only remark licit
'these four rules have no rigmlfcucc whatever, being u mere hit
1 On lLL 1 ttibhakitf-niptrijm In prnbfiitvom lyuipiMum-tauk/B&frihim
diVfj-oi/,
* CL RPxi 44.
1 li. 6fr.
Icrtirtt l'PElH wLrtkilh puriu-tnrphit/i-1-idilk
m c&tttydpify^hdi * tta m intUtfi*
v Cl Swed> op* cm § 3 91.
1 aajv. 1-4. <ilba ciUafTah fupfafrdftl p.Lâî\îniAh^<t-^hUdratnnîj\h-
rdtkgfr-firmhini titi; «ditd - p iujit -1 tmuHtihia - its my utfn h prida-tundatelr Midhlyitlt:
yfflhfi-n'am iikgoTii^^mfutdJind/H apy n nm.
* Thii iiuvtprefiiti»n uf tatixm-satphittihtawed onsxi. 1 (vyt&pouw1 tvatd*%£tm 1:
cf bIid 3,20 below.
T t-<rc the fourth culinary dl nphecirttly SwMt, be dl-+ *Synth*Jiii, hnlly, deals
with the oixnhic xiid Bcmntie sermiptn? of sounds into *y iliiblc*, etc.* and du
daviiioTii between ihtiv group*/
M PHONETICS IN ANCIENT INDIA
of outside classification, in which someone has itmused himself by

indulging'. in one passage of the RP UvaSa notes ihal strictly
speaking the author fa exceeding his duties by giving rule* for tonal
synthesis within compounds; the basis of synthesis, he claims,
should be whole words as institutionalized' units; he is prepared to
be indulgent* however* for
.. just at a Howcr-picker may alto pick fruit, and b wood-gadwrermay

alio g&ihcr finney, tod* is the care,1
0 41, Phonetics and Phonology

There is, on the other hand, one respect in which our treatises
do not fulfil the functions which they claim. The first verse of the
vIP declares k
Ouf fuhjvcts are ihe (phonetic) attribute?- in junction and in wolnjicrn
of tin- four tvond-clasiSp vii, noun, irrh, prepmirion and p&rtkle,*
This grammatical enumeration ûgîs that we may expect some-
thing tike a phonological treatment, with grammar and pi nine ties
integrated into a functional whole; end l vala» commenting on the
VP, goes so far as to claim,
Thii freamr exdi aJI oihcr iftaiiiw in dust ii combine* the two db-
eiplinr* uX phone 11123 iuil! grgfnmir*
This* however, b in Lhe nature uf a pious aspiration which regret¬

tably does not see fulfilment: for our test makes ®amt reference to
any grammatical function,3 Particularly remarkable fa the failure
of the phonetician* to discuss one uf the outstanding phonological
process of Son^knc, dial of Vowel-gradation': for certain phono¬
logical purposes it is convenient to recognise a system of vocalic
dlenmtion of the type;
Grade r* 1 u f
,, a. e o ar
n 3. ai au aar
1 rnM*, lit
1 On a ti- sfi. rtjj.r±i£ puniJnJjTt tmkf&muft ksxrtacynm sidtlhcju Gj pndtju
ro^n (ii-pridr/tth: Maryam ynth:l puip&Mr&iy& ptmLlhurtintiffi *t£tT-dfiJrusyn
Hudhn-dh&amtin: fvvm Hat.
1 tuiwti.iift padn-jdtQmlm mSittd^ytiiQpuiwga-wpauiruini tamMy^pitdym
fvm* pfdlijrLtm.
* On L i&g.
' Cf. I.udertp p. 10*^ OolditiickcT^îrawc, pp, jy* ff
INTRODUCTION
The working of this ahemauon is seen in vetbd forms mth as?

(v-aty*t 'praise"} (-Jlqr-, ‘make*)
r.P, etura P.P. kft&-
inf, stAtum Inf. k&num
P: tni stauti Pf ImL aika&m
Though ignored hy the phoneticians, this alternation is duly noted
by Panini (tn hb opening aphorisms) and hb followers, who treat
Grade i (corresponding lo the indo-European 'reduced' grade} as
basic, giving It* Grade 3 {■= IE H no mid" grade, or 1 Volhiufe1) the
title of guna or "secondary quality \ and to Grade 3 the title of
v[dtfki or 4increaseY1 A further phonological process winch is
similarly disregarded is that of samprasarann (tit. ‘extension")pi
whereby a sequence of the type vat La* Y+sylkHtity, alternates
with P* i.e. ‘sylbbic v" (cf. Pr. Jnd sv&piti: P,P. supra-* &c.)>
Pliiini uses the term both for tile process and for the resultant
vowel ; but WL tend neither the terra nor any discussion of the
proce^ in the phonetic works. Nror sixain do we find ihere any use
of that great creation of Paninih& genius, the phonological Tenp*
This mention of pHonologksI omission^ however, is nut to lie
taken as in any way detracting from the value of the treatises from
a purely phonetic point of view,
0,42. Terminology
Before proceeding to the textual material some account must be
given of certain terminological features which run through the
whole system of description.
A particular problem is presented by the word tariût, which can
T /Vm. L t. 1 :F, irfddMr dd*etc: Ac guna-*jddhL The icnn nod
process of rwrTJi jre in fact flmt referred la bf the curly £tymnk£Ht Yaska
tlVi>. i. 17. di.-tLvum iWn from %î) Only pasd'Uf tthtztlttv >ur found in the
phnnmc WAriuffMtft m ftPiti i op f jrfrfkt in VP v* 1$, AP (cJ S. 3-; Shmri)
tiL 1-IJ.+.3). Vi Ed&erfftn., Skt ffiu. Pho**!*#}' (jAO$ Supp. 5, >134^
# 1*5 ff.
1 The fdmnor of the term t» fif-t A?*Tr Tht form is ib found,
and ul lined hy tbfl ^P(cd- Shajstri’j ill. i ( ] c to trfcf tn Hit replacement df -tab-
hyp- m rilp&uTi> dahh-. $nx further Edgcrten, 11 Kprirriiuiee;
caw?ftem (rowel)” \ JAGS Inn- m f,
1 * t* #S. tj? >j|Wj Cf vi. a, vi 1.13* vi. iv. in).
' t- i 16 ff. Cf. Huikkittl, TAr Frip&ft pp. 31 l! The modem u-ige
bejpjfli with 5w«i+ Afro- &wmw# pi. i, $ 7? CrafStive iiifloctkso'J:
ef. De Somme, CW*. pp, in, 354 ff,. Nidn. p. 46 ; JaJiob&onF
'Siiitie Zero'. AM. Bullyr pp. m$ IT.: Fret. *2*ro, vide et 2, /.
rArtw/fi‘,1 ly^n, 3/4, pp. ifu IT,
14 PHONETICS IN ANCIENT INDIA
be moat conveniently discussed by reference to the terminology of

the Latin doctrine of letters,1
The usual method of designating a particular Sanskrit consonant-
unit is to realize its 'potestas* by combining ii in a syllable with the
vowel -a-[c g, kfo), con^sponding m a 'figura* i|); the ‘nomen* is
then provided eilher by this syllabic alone {bc< ko) or, more usually
with the gufflX -karu/ 111 'making* (i.e. ko-karaj. In the case uf
the vowels the necessity of adding -Q does not of course arise (thus
t.%. figure f, patejfai L rumen j or i-funu). Alternatively the wmtti
may be manufactured by realising the potftim m combination with
die enclitic parityJc r#i4 (e.g, k(iti), giving a mmm kiti), which has
an eifect comparable with that r>f niir graphic device uf inverted
commas (i.e. kiti — "k'T
The nutstanding exception to this mode of designation is t\
which is given the nomfn 'rephaV variously interpreted as mean¬
ing ‘growl, snarl* {cf. the Latin 'liftern emuna') or 'tearing' (3* of
doth)**
To refer to a sound-unit in general, however, we encounter the
word varm, the use of which b concisely summed up by the VP—
What aw lined (He, in die mtnoômmnndyo) ate rarnoi*
But the term ritrrta is never used to form a mmett in the manner
of Its suffix] use is in fact strictly limited: it is found
appended only to the short vowels, with the function of designating
not only, for example, the short vowel I itself, but also the long ii (I)
and the rare protracted ill (ij): • to denote simply the short vowel,
i+kitra and not i-tarm is used, (t ihus becomes evident that tarna
primarily denoted not a sound-unit but a more comprehensive
sound-qua Jin (in fact always a vowcl-qualfty):* as PaEahjali ex¬
presses it. if iit a generic temi comprclicnding a whole Tamily of
wiund*’ (t arnfakEifoyil) This function of rhe term is consistent with
I S** *.%. DflrufrJ* (Keil, iv ^tSi.
II TP 1 31 uhtSift fiyrntjuntittani. Cf. FtnhH fP5 p. E1&.
1 TP L ifl-17. kdrottar* v-grijdkhyd; vpodffartt&ttim,
* VPL «frdA£tin&- 1 vk- TP*, iy tjrfuu tu m%yu.
1 i'fnVfaArfMb. an TFL to nphyatt i'awtrdrft-pdlmTtt+dhrfmirad jjivdr.
}Mm ifi rtphah
' 1 14- Ej/Wifpi wnidfi (t.vufci: i inrKi-iafrwirnn&y* knthiii} v47r7i1J.fi>
1 CL TP * ao. Arret* fwprtjtaftn tf*ty4$4wT on which TttbMfytitaiHd h*i
L'l-irfMkff^rc* hrest+1 9Fm&-dfyk^ptut#mm dkkyii bkot'afi
T A suliwiy eierption i* pttrttd«l hy Pdn, vrr. ht. - A
rT MAh 1. L i (Kit(horn, L 13V ti-v&Qdkrtit vpt«tiff4 fmvom a-earitti-Aidtnp
tûkijyuti Willi MF^Milb cf, tSjWCidty l> Jcme*, Tht P’^mrmw, } £2.
INTRODUCTION
its meaning in other contexts, namely 'colour —a band of the
vocalic spectrum. And although the word comes ultimately to
designate simply a "sound-unit-, it is noteworthy that Panini's
larnti-sumdmnuyu gives only the short vowel of each quality1—an
analysis which Panin i further supports by his treatment of the long
vowels as phonologically equivalent to two short vowels of the
same tnirnaJ Whilst we may note j.ho the graphic similarity of the
short and long vowels in the Dcva&Sgari script {W ^ 3 'Sh it
is unnecessary to suggest. a* Bkuld ha^ done, that Pimm's listing
may have a graphic basis:1 whither or not Panini knew an estab¬
lished system of writing is atilt a moot paint,4 but he was certainly
too great a linguist to allow orthography to affect hh phonological
analysis.'
Of some intercut sn this connexion is the indifference to vowd-
length in the writing of Latin and Greek; the attempts of earty
Latin orthographers to introduce double writing (onp &c.) for the
long vowels* had no lasting success, and a Greek grammatical
scholiast, noting that there are 24 letters but many more sounds*,
points out that the single letter 1 may hear various prosodies of
length, aspiration, and tone.7
We have now to consider the translation of r*ima. In the
^pecialkcdp suffixal use discussed above it may be adequately
represented by ‘quality' (‘i-qmlhy', &c.); it i* its wider usage Unit
presents some difficulty. Whilst it there has much in common w ith
the modem term ‘phoneme1* no phonemic theory' is implied by it,
and ft would be reading at once too much and too little into the
icrm thus to translate it. The noncommittal ‘sound-unit’; by
which wc l&ave thus far represented it, suffers from ihe disadvan¬
tage that, unlike varm, it ia restricted to technical u*age. A happier
tendering, and one which would fit into the Latin terminological
L £rtii-Sulr*i i-a.
' Sec Ktnhci 3.1a bsio™
1 Pi 1 pm ml Aim™, p. it.
J A (arnnus but ■rpauntHl in t-ivour (i thul of LiuTiiiiilckrr.
P+JTzrnt. pp 1 j f?r
3 Cf. Stcii Korun*, up, dt.F p- mb V. . . tin: tcrfxi VimJmtui}-* s iradi-
oonol ftratnunfion handed dm vtrbiilly from teacher i<i pupil, and not ■
written lint1).
‘ Ct RlfacM. opuitula PMoioitiv, av. 143 ff.; Lindsay, Tkt tjsim Languor,
PP* 91
f Schd on Dionfiiin Thtu (ed - Eiilfutd, p. 31 = Bdcket* J witota Cowed,
PP, 774
PHONETICS IN ANCIENT INDIA
tradition, is pletter1r— letters after all come very near to being un*
sdfconscioiis phonemes,* One objection that may be brought
against the term is that the tarna of our earlier authors may have
possessed mfigura:1 and it is inn* tliat if they were acquainted with
any system of writing, they do not allow it to obtrude itself on their
diiicossion of ptftifas—even at a time when writing was certainly
well known and widely employed, reading from a written ie\t was
condemned as one of the *sht vilest modes of recitation1.1 Howtvrr,
it does not seem that the absence or latency of this third attribute
need prevent m from using an otherwise convenient term; and wc
may support it by the fact that when the language does come to
be written, each patriot and fwmsn of tbc phoneticians' catalogues
receives its appropriate figura-lshd*
it remain* io mention in this connexion that certain dements
discussed by our author*, *omc of them included in their hui^ iJuaJ
versions of the alphabet, line excluded from the Pipootn iwjuj-
simfitfut&yn- outstanding amongst these are the fricatives, -h (visar-
jsntyrt), {uptidhmmfya), -Xijihvdmfiityti}¥ the nasal nj (flirjfjrdru}
and the faucal plosive* {ynma)—ill of which are bound to trmone
closely limited series of contexts than the other letter* and so are
given the litle of parMmya or 'dependent".* A further title
accorded to them by the Pdniniya- Siksd and certain other treatises
presents torae difficulty:
The fricative* -ti, -xp amuttfra jnd the ymnAr shat* the place of
grticukhori ut the sounds whereon they depend* and are la be known fit
*ay9$m-'dka* J
The term aypg&mJui is generally interpreted as moaning ^drawing
unyoked1 (a-ycga-% and is explained by Patanjali a* follows:
, . . hi.Lijtune they drew unyoked K Lt. art heard thmtgh net memicncd
tic. ihoti^h not in eluded in the PAj^lnom varna - jflHidmrtdjd) '
1 Cf, Aliercfwbk 'Whm u a ' letter”?'. Ltitgn*j, II. i 54 ti
* Of. TwwJdfll, Op. lit, p. fj|4 in Ptytkyltifit Ju iQi;
F Mali tern r FirU Grurnmatifu/ TrmftK (Stlflp. to La^. 36, 4I. p. fi.
j cr wWktp /hd. si. kP t4»
1 36,
Jfixi %Arl tiraA^juiijM Wtkti IMtUl-flulkitkak
'tpti-ktmSJutJ «r faqL pd^\ukdrfJ*rimdh
1 Cf. K«wn\ toe, rit*
* -Pi 5. (WttrtJre T mnjsfW' «| oljw Ikmlmoftw,
' O, - . . qyvi'acdhd tijfhyd d/raftt-tthtinfahhdgfmik
1 A0A. l- i- 3. no F*t 5^11 5 (Kir.IhumK i 2ft): Ar peaces TV&+
b*th*v putter diûjffu-
INTRODUCTION* 17
But * drawing unyoked1 seems hardly 10 l>c a natural metaphor for

"heard (though) excluded".1 Uvata. in hk commentary on the VPt
has an interesting alternative; he takes the initial a* of the term not
as privative but as referring to the letter a and standing for the
alphabet as a whole: his explanation then reads.
They jure called taee&use they draw* i.e, eitoin their renliâ-
non, <]fi|y when joined with n, &c+p i-c. with the tetters of the alphabet/
In view of this explanation, Weber reads limply yagtivalm in both
text and commentary;' hut appropriate as the term would be as
applying to the contextually l>ound nature of the demons in ques¬
tion.4 it is to he noted chat the j? 7* specifically distinguishes ayoga-
raha f= the con text daily dependent elements) from yogotSka
(= the other letters)/
An important terminological distinction underlying a large num¬
ber of tile ancient descriptions is that of ttfi&w and karana (Lit,
"place" and £organs)h which, generally speaking, denote the passive
and active organs of mtieulruion: as the commentary to the AP
explains.
The tthdna it that which ii approached. thr luirwui that which vtp-
proachw/
T he terms closely correspond to what Tike calk ‘Point of Artieutfl-
wMkrffa ■ yad uvjtktfl voh&nty Then? fallow* it iLwcuatimt
ci U> which pmlyafriira the could tw included wndrr, ending with
the [TURCfffisriQTi that ther my belong ucuirr dhat Aar*
*j/r ?_ an J Italyymtl itcconllxigly the title to frl’cf to thii loilk of .1
pratydthfira {qyukttifi prafydJubc^h^rui >
1 The explarutJini of the fto$j£d(ob PpS, ]oc. eir) ii eniiel bus h:in.Hy 2tnnro
acceptable (fUi vidyat* yQgfrfl tamyogo varlpUitmrQtt yrfdm}, Bchtlingt'* inlcT*
rttttaticm (Pdn. (iS4c) i|. 4ij> U fanciful—H'Tpmuiuiic lie^Dthrinflend, die
Vuealr van den Cciuonuaxm jthrjdcfid, JWKhcn BdJen b dtr Mittc jichcnd.1
Waekenjuspn'E utTribuEea the UritlgtWWflftht tfiJTft to its luivb^ bfcfi i/iTcîtnily
earned far the fetching of ehiyrrft {Ai. Gr.. i, p, k*l a 7= 'Did nod drr
humoriiiiBCh t Chnniktcr mother Termin i wio urora tltrkmiTi
«un dem ju^mhuttemcht’),,
1 On vii*. |Hr riMrttdjJTfJ trJJTW-jxwwh*M£J>f»W «thiVd^ rmust rtf twWnU> rfimO*
pM/taDi 4jnr> crtnfJtiiAjJA, 1'Of thft us# at yfchtfdi, 'a. &e.* u - ahe1
cf. SSvanp iin Tmtt. Jr. vn. ii.
: viii. 3? IT. CT on P&f luv- ciL (fnd St iv, JH* ufh3 m ftwftf** i-
* Mp jmaicr appropriate™ of chi» term wm evidently We hy Camirac
trarntaimnn* whrt adapted It jmreud o: (s«if LF S, S J>pt£iri, brtiw-tt
*m Pitumjdi't MMtohdfyrt, i, i4jti.h rVU Amarca S.(SS p. hui
Ml|^lKidAdl
1 Eotced., cd Bhaitri, p i,
1 On i uptikftiriyccic tat lihdnam; ytrtQpukramyat* t&t hmonam,
asm c
eft PHONETICS IN ANCIENT INDIA
Uan and 8Articulator; 1 In a large majority nf the articulator

is an acea of the tongue, vis, "root of the tongue*
'middle of the tongue" (jikva-m&dhyti) and 'up of the tongue"
(jiktdgm), whilst the opposing points of articulation are "root nf
the jaw' [hanu-ntHhij_ i.e. soft palate, "palate* (tfiiu) and “teeth"
{danta} or 8 teeth-roots' {dnnta-mula). The same classification is
extended to the lip*r that its the articulation of the bilabmls the
dP1 and tlie Tribhafyaratm* prescribe the lower lip as atid
the upper lip as ithana; and the APgoes far as to apply it, some-
whai artificially, m infra-buccal articulation in ihe cane of the
glottal sounds, for which the lower part of the glottis" is eon-
widened as the hiTrami*
The specification of minor distinctions of sthana in the alveolar
area is sometimes not as dear as we could wish, but in the absence*
so far as we know, of palatogmphic aids.* this as perhaps hardly
surprising. An attempt to apply the system to the feature of
nasality can only lead to confusion* the nose in such cases being
stated by seme treatises to be the articulator and by other? the
point of articulation.
Other terminological items of less wide application will be dis-
cussed under their appropriate headings The reader is also referred
to the excellent glossary now available in vol iii of Renan's TVr-
tmnvlogie grammatuaU du Samkrit*
0.43. Order 0/ Analysis

The treatment here adopted closely follows that of the Indian
analytical procedure, which recognizes three main stages:
L Analysis of the basic articulatory' ‘processes*,
ii. Segmental analysis of the speech-stream ('letters1)*
Hi Synthesis ('prosodic feature*1).'
1 pp. 12o If
* L 35, &fthy£nam MfJtL&aMjfJxm (sc- A^rcuii^).
1 On TP ri acr, rntmimtiffim itJkinam umiTiiTi a-Kimy&i tpzm HhilMttdmi
adhstfiiihait kmunxm
* L ig, kiinfhydnfit** od*vnfckaKthuht*
1 The Jim rtoordfd m^Toiicv Appear* fti he thul nt on tnirlixhmm, |, Qukkry*
Colta, rnhs> m 1K71, <n the Cfltuwt of phnnclie pkuikv, pamtrd thr racf ot hu
nmuh wilh n. mixture of finur *nd muciL^r IcL ftoiissctar. Pnndp*tr p. S3}.
4 Sc* alio K. <L ChAiuijf. Trchmad Tmut *ntd Tmh^tpm >1/ Sitmhrii
Gfdmrria*, pt, i (Calcutta, in+SX
1 Cf J H Kirth, 'Saiiiuf* and . TPS * #. pp. i»V fF . K |. A.
11 4
IlmdcrauEL, Fitnoditu in A Study in Synih*ditt\ A»ia Mujur, t. bL
|S* tr.
INTRODUCTION iff
In (ii). where the individual segments are made the basis of descrip¬
tion. tisc fragmented processes arc regarded, by a common but
[juestiorLahk technique of inversion, as 'distinctive features', sen -
in^, together with the various articulatory positions, to differentiate
one letter from another. Once the basic processes have been de¬
scribed, however, phonetic discussion under heading (ii) is largely
confined to the places of articulation.1
1 cf Tntlwtiktiy ‘ dtutmctmo of 1 ArnkutMifinutrlttgtntXliiS «d 'LokuMtt-
v*fi;*#rfrmiilzS iTCLP iv. io| I-).
i* PHONETICS IN ANCIENT INDIA
THE SANSKRIT ALPHABET
ComftMJLtfta rGUUaS' J'litr f-llltaf
1 f t p
Iruupiralrtf
E k c t t 1*
1a M d* rA ih
> A*p toned
I kh th Lit tb ph
s
« t / d 6
Umkoi rated
1 6 1 «s b
1 gh i& t* dh M
> Aifinttnl
fib m «lh dh bh
* « B IT PI
XftAUbfe
D i* ft n m
* r i V
Sdfijvdufil*
t t T
b 1 # f (A)
V nirrl^.*
-h -* 1
Fricatives f * ■+
h
Vowed
%
Vowels
« * f i m
Short
ft j r l u
4 f ? f 0 «
on M * tTT Urj w
w* ff-) l-TLl
DtphthcFiifpi
«i uu
* Abo oram ft}

'ommAtika*—m *
AW. Thf ntdef nf tprten m fitmnted m Fftytat1* vhcinn con.
■iderahk tfKrf£eE3*& from the above: thii (mi, however, r* rxpljtihk by [tie
fihwujbKkal, u opposed Ifr pHnneCrc, ipjirC4di fhvn: adopted. €f+ ThiKnic.
op dr., p. 1*4: "The arrangement of Pipini'i Uat af iijniidj, wJueh IE ftm
Jiôk* imther dLWjrJerly, ±1 explainable w due tu- The phoiKtk tSiUlojfUr of i-iitndp
hrnn* bisra adapted tn rh* praflkal TrqitircmeTPi* or the srmmmir, in which
ripiEii wBntftl to nfer to certain groups uf v^mdi by then eiptraiimu.'
PART I
PROCESSES
1,0* Mental
The Indian phcm£tiriatu spend but little time in discussing the
mental or neural bases of speech. The introductory stanzas of the
P& are representative:
The touJ, apprehending dungs with the uudltcr. inspires the mind
with a desure to speak; ihenund then excites the bodily file* which in it*
turn impeb the? breath. Vtw brewh, circulating in the lung*, h forced
upw ard * arid* i mpinginij upon the head, reftches the apectb-Qfffan* und
gives me to ip ted i boards. These ore dwHitrd jn five w*y* by mncfc
by length, by place ef arucuSniion,. by process of artivuktiun and by
secondary features Thin itw phoned riaira have ipotan: (ftbc careful
heed*1
The 'secondary features1 here referred to [Mtiptodftiui) arc inter¬
preted by the &ik$u-Ptok$id as 'antittjisikkdi\ 'nasality* etc/ (see
further t.xo below);
The musical treatises contain similar statements, though these
are less closely related to the actual speech-organism* The relevant
passage of the SatpgtU&atn&k&u reads as follows:
f]Tic nqut, desirous of aprsiion^ iratigsies the miml; the mind then
esdtca the bodily fire. which in ita turn impel* die breath Thin then .
movo gradually upvwudft mid produces Bound in ihc nave), die heart, ihc
throat* the head and the mourh*. - *
Parallels to such fllAtemnts arc not far to seek in die west*
notably in the doctrine of the Stoics. Zeno is quoted as defining
speech in terms of
a ttrniAi of aj r extending from the principal pan of the wml to th^ thfDtt
and tilt tongur nod the appropriate org&fiV
whilst Aristotle described it as
ihc sinking agaimt list w~eailed 1 arteryh (i-e* traches) of the air exhaled
by the foul*4
1
, , IrjrJrfl psliLMJkS jlflffflJj
Kunr4 kdlatab rikdnut ptapimlmtpndSmUtib
(If f rtMUl'TiM ptttfmr mptuwrfl tun mbodkuta,
* i+ nk j H.
1 PliwMch, IV PtM. PM. iv. SI. ijoje 9*w;». d*i *ts
p/g|H lapwpjn* rfai yVnTTjf tttl rur o{pf<isn* ojftwl-
* fo Ammu, « 430** ft* »tg^*ofJwv Up* inti rip* it Tt^rcu tuf*
IliJpiwi -iujfiff vp&% nj» jtuXwjpA'g. djinjû ^ Am*).
si PHONETICS IN ANCIENT INDIA
1*1. Physiological
A general statement such as that of the PS also introduces, in a
rather haphazard manner, some mention of the haste articulatory
processes (prayatfia) which are more systematically presented by
other treatises.
t.to* Classification
These processes arc divided into two main types, ahkyatitara.
1 mtcmiiT, and bttfiya, 'esternaJ'. The first type comprises processes
occurring within tire buccal cavity ('intra-buccal’) at id the second
those occurring cUcuhnrc (‘extra-buccal’). Fur the first type both
the I F1 and Panini* also use the term tiiyu-pruyalmi. *mouth-
prnccas': this is interpreted by Pataftjali as referring to the area
from the lips Id the ' kulutlaktithe latter being further identified
by Kaiyyafa as the thyroid cartilage or ‘Adam's Apple’.* The
Indian classification of the processes may be summarized as fol¬
lows:
1. tntra-buccal processes {Shhyantara-ptayattta)
(<j) Closure —associated with the class of stops.
(A) Opening— „ „ „ vowels.
(e) Constriction, of two degrees, associated with
(t) the class of fricatives,5
(ii) „ „ semivowels.
2,
Extra-buccal processes [bahya-prayatna]
(n) Glottal —associated with voice and nun-voice (breath).*
((c) Pulmonic— ,, „ aspiration and non-aspiration,7
(e) Nasal — nasality and nun-nasality.
4
Nut all our statements adhere vigidly tothis dewipfiv^ framework,
but it may be taken as a generalisation uf the various systems, anJ
i* set out in precisely the above term* by Patafijali* and by the
Aptiitli-Siksa" Departures from this system arise when* by the
inversion already mentioned* the basic processes are considered as
1 i. 43 febi-ed by Utoiuhs imhMa'prfymtia). i i. i, 9,
1 Mhh L t. an■ loc- til-iKidhcm, 611 vilhdt prMiytipnik kMaiakuI
* On Mbh., ioc, tit jcnvtSydm imiaJn-fr^J^,
1 CL TnibctzksyV <l« th.)
* CT Tmiwtekfiy j
* Of. Tr\lh*uk«y‘i 1 E*ip*riitixmiiwl\WTfialron1
9 Mhh. i_ i 4. on F4n i- L v-ie* Kifi^Lom i. 61 If,
■ lii. i ». 1 If
PROCESSES S3
distinctive features serving to differentiate ode letter from another.
A passage from the TP may be quoted in this connexion:
The distinction of ft tiers ii effected by isKoadmy femora, by combi na¬
tion, by place of iiticnksion, by ihc position of the articulator, and by
length/
The meaning of wmc of these terms is made dearer by Uvaja, who

quotes this passage in his commeniuiy on the RP\' as an instance
of‘secondary feature' (ampratlona) he mentions the voice-process
(aa in the above summary); as examples of 'combi nation1 (iowirsiigti)
ho gives aspiration and nasality (ata1); and he interprets ‘position of
tire articulator' {huntmi vinyaya) as referring to the intra-buccal
processes of closure, opening and constriction (tahe), which he
exemplifies its the statement,
Brtv,ltd letter^ hivirit! the totne place of articulation and secondary
fesi urea, e g. I , y. J, acoustic distinction is effected by the articulator,1
But it will be noted that in the TP's statement of 'distinctive

feature’ these processes arc t red Led on the same terms as the
places «f articulation uml a prosodic feature such as length, to
which other writers also add tone.4 This, however, is by nu means
only an ancient Indian failing; Twaddell, For instance, in his mono¬
graph On Defining the Phonrmr. i* prepared lo admit as parallel
‘component terms of articulatory differenced such various features
as places of articulation, duration, and the processes of voice,
aspiration, closure, and constriction/
A list of live resonators (prffftfrrifAa) is also given by the TP,*
1 11E 2:
iTirwpt mrfiisttgdi ithdHdi kar^ rinyir^tt
jtfvfff* rvrna-trmiftyam parirndyfir m
J aiii 13. UvatfiL rcfen txi 'drfiaam fean™' by the temp 'qurti?.
whicls b *Imj used by the Ap. i. (iv. 7> ihe Mbh- i. 61) to refer
piiTtcculuiy ro nudity- Belli Uvnla Seft RP iii VP ib iôi 111J the
Tnbhfyy&ilimHM TP f, 11 um. in COltfwndoh with toudx thr term dhafmu,
^property's dw Vdw«r it*df thm being nifci™3 lo an dkermim, 'pmmmt of ■
property1,
' miya. iikd^djhtpraddnd»Hm *pi ikdf a ■ja him >>'UAdm W m kararu^kflak .ruTi-
* lrVj.f to-mc hw Uvnû Oft I^P iJL iIO^gF. IS* It li ffjtcwd

by PatnidaUtiUt*. PdV. t... 9. KWfcwM t*» gwsind* of Ut
ItOhdutiilcuvthcM—ahhriiskd nijltddajrtlk,
1 P-
1 li 3 tdO-a prtcibulMM bhmanly unik A^nh ndiikt rli
(Tnhk.priitiitui fanATrffkmuh).
*4 PHG-NETICB IN AN C l K NT INDIA
of which the buccal, phafyngul and nasal may be justified the

further mention of *chest' and "head1 as resonators* however, is
probably taken over from the subjective terminology used in
India* a$ in the west, for the description of the various voice*
registers,1 *
Wc will now examine in detail -md in the order set out above,
the statement* on the individual processes,
1.11. Itttra-buctiii
Four degree* of doetut between stham ami karma are recog¬
nized. Maximal domfe is referred to a.* spffja, 1 touching1, and
minimal closure as ’opened*.
i. 11 o. Vmcth and CmunuinLs

'[lie process of minimal closure, or 'mm-contact1, (aspfffny pro¬
vides the phonetic criterion for the tlifruiuritun of vowels (ftwii)
from consonants {vyafijma}: the TP expresses this in the following
terms:
For i he vowels ibe 'pJuce of irtic-uLaUctn' iht pl^c? ?o ^hirh
appfoxiittftiioA ik m*dc+ jjnd the uricubtor' refers to the organ which
effect* the ppprKrimmon, For the rest the 'place of urtiL-iUncion1, refer*
to iht pjftcr where ccmimt b misde, and the 'artkulator^ refer* to rhe organ
which efTccEji the ecmtaet,4
Maximal closure, on the other hand, provides the criterion for

the category of stops (spuria)*
Thus far no problems arise. Hut the intermediate degrees of
1 Cf. f00i, .l-OMi/j* Fhcinrtit'i. pp_ 58 f . Forcftharnrnti, Thmri> imd

forkmk SSn$*m uvd Spmkrtur pp, 271 if. (Kach^mt^rudn^ Sawnduz,
Mimdrtttirifiitt).
2 TPuffi. to, PJS 36-37: VP i. to. 30 Pike (op. dl,r pp 17 ff.j grvei s
abort critique of ihe ' Station-Libel technique' Ukj4 in ; dm of
the inonjcuom quoted, viz. to JpUtc lhc tone between the ey»’ find* * rioac
parmilfl in the ahhu-msJkyah ef lb* rdev^nt p«wgc of the FP, Cf Pop&faara*
incr. op at, p. 37*6 C. . , mil Jem Beiorff drr Korfroacyuix veiksstnw da*
Urlitef drr bkiutujh^phyai^li-pgiwhm Kmihemunsiitii und tf£tB) wtif dm Gehirc
dCf K^rK»mtiprindutt||eti liber"),, p. aBf (■ Diti Bn m lfrîutiz mufl wahl dmnv^
*** dte Kopfrctoiwra, in die Heilws drr gcwgrocliiHfectart Vrnrrungtn
vfcrdeu'f
p Pi 3S.
1 fl- ynirûMmitAtm Utt tthAnum: yu4 bpntmphmati for
hnnwvm: uny#?#m iuyntra tpuriumirp tnt uMr^; jvnd tpatfoyaH im. kzrentm*
1 * tf. .IP i- fjfrsrirfnrim frorantozn.
PROCESSES a*
constriction arc designated by various terms. Tht Ap,S.' refers to

lint four intia-huccal processes ns
(i) Contact
(ii) Slight contact (lidi-rpnta)
(iii) Slight openness {isml-rivjfa)
(iv) Openness (wefta),
a classification which is reminiscent of our modem terminology for
describing degrees of vowel-closure. The PS employs a rather
different set of terms:-
(i) Contact
(ii) Slight ranker
(iii) Half contact (*mo~*prtt*y
(iv) Non*ctmlaet.
The statement of the AP provides Home difficulty ot interpretation.
l,ite ihe Ap.S. it mentions (i) contact, (ii) slight c<l,lldtl' an“ flvJ
openness; under (in), however. we find the word; Jtid openness
(viirtam «)s* Pataiyali, who quotes this staiement. is probably
ri-dit iii saying that we must here understand*slight.' (ffaf) the
preceding rule, tfau» bringing the statement into line with that of
the M' The AP commentator, however, suggeststhat the
whole term ifat-fpffltt is to be understood.* so that (iii) won
then read 'slight contact and openness'—a description which ts
more to the point than it might at first appear («e further i.m
l*6*01*)* , , , . ...
To (ii) the RP gives the further title of dtifi-ipiffa imperfect
Processes (ii) and (iii), like (i) ™d (>v)* provide dassificaUiry

criteria, (ii) for semivowel? and (iii) for fricatives. 1 lie application
i til.
I *|
MO ‘tpftfd >mwj rv ntmO^PTH^ teak wrjtes
iftfh ifiiftii hftlab prnktii wib»<tkJnup'wLi»<tt,th.
Usi arilho-ifinlo m YmtimtekyaS. Mf f.. V<fl nntm*a-i>rteipak<i &,
• i. ip/ihm tpa&mtp*kpruyffla- t<**antahnkAtem- utmost?
™*T Jt^*iKlS 5£ l. i. W iKtellwm,.. 44). Tir,t*m ip*tern i,te ir m

Matron*. Th* dewripilm ai<iii> » ‘ope*'(aw. P-
4 failure to obwrve (Kb oMutpm
1 CJn L it.
46 PHONETICS IN' ANCJRNT INDIA
or this descriptive framework may he exemplified hv the palatal

scries as follows:
(i) Contact —c
(a) Half contact —;y
(iii) Half openness—/ [9]
(iv) Openness —j
1.111, Fricatives
One of our treatises gives a more detailed account of the articula¬

tion of chc fricatives bv process (iii). Hie TP, having remarked
that the fricatives are articulated in the same places as the corre¬
sponding stops’,I goes on to say, ‘But the centre of the articulator
is open ,* a statement which [ends some support to tht view of the
AP commentator quoted alwvc. Whitney, commenting on this
doctrine, makes the criticism that,
Thu fitocripiiun bf an undosutr of the middle of the orsran in rather
hi amficttl device for savin* tlic credii of the general prescription of
Actual tonratzt in fill dap consentsnta.
fttatognum showing the articulation of the fricatives by modem

Indian speakers would tend to support the TP'h observation as
against W hitney’s uninformed scepticism.>la the case of the retro¬
fit^ fricative the AP gives a rather more graphic description by
referring to the tongue as ‘ trough-shaped‘<(cf. Gmmmont, ons.t,
Is Janguc sc dispose cn forme de guuttierc ci forme un canal
iris etroit. . . .*)
The general term for the fricatives Ufyfton, literally ‘hot, sicam-
ing\ perhaps because of their resemblance to the hiss of escaping
steam: it is gtacd by Una* as t Syu, 'wind'* The term I* applied
not only to the letters | g s but also to -$ -x -fa and W and to the
11. djmpirry,™
* it. 45, karatta-mtuthyiHft In ihjiam
Btugpml' voL ii, Palito^nmi Son, t, 6R, r<o {utiph. aifu:
■Lai' •- * H
-' - 3be'- “»" ihn-f tWmv dic z„wr ihn

, jE‘ limber K. C Qiilirr]iF op, cir.. pp ff
5)w« re Hnufy t. h but»,hc MbK
pta* mu 0*1l *t Ofi ot.t kwflusm, i, 27) thi* h nnly for caftvrai«Ke
is i- certain phQn-atiQgtrtH I nJ^-v
PROCESSES z7
breathy release cf the aspirated stops [tofman). There is no special

term corresponding ut 'sibilant*, though excessive sibilation is
referred to by the RP as kjvqknvm, 'whistling’.1
MIX Semivowels
A* regards process (ii), with which ts associated the class of semi¬
vowels, the validity of the analysis is not entirely beyond question,
involving as it does the postulation of a greater degree of contact
for this class than for the fricatives, In the case of the lateral l and
the rolled r the classification might be justified; but the case for
y and v ] w]1 is less clear,! We should expect the criteria for setting
up a category of semivowels to be phonological, and related to the
fact that they do not function as sonants* in the structure of the
syllable; from the phonetic point of view y and w might be de¬
scribed with the dose vowels i and u,* and we may suspect that in
erecting a separate phonetic category fur them the Indians have
been misled by their system of letters. As Pike points out.
Syllabic contextual function h rellscted m phonetic alphabet}- rounds
which in described by the same procedure but "'Inch ait used differently
in phonemic *y*rcms m *yHaines in contract t»> non-iyHabits arc civ>-n
different symbols, and at times arc given lums vuchas ‘scmlvowtl' and
the like *
The apparent failure of ihc Indians to recognize the phonological
(as opposed to phonetic) basis of this category of letters has the
result that whereas I is regularly classed as ‘open* or ‘lacking con¬
tact', the corresponding semivowel y U described as having 'slight
contact'. Regarding the nature of this contact a more specific state¬
ment is found iti the TPt
for y co ni net « mode on the paJale by dw ednjvs of the middle of d«
tongue,7
The accuracy of this particular statement would in fact be sup-
' niv, £9 {Uv*Es udkiko tfjifflinyd ttf/vpQ dhvtimfi). A tun hr; UtiH in iliiFir
pnmunclDf i<jU e* Riven The nnnu' rjf tammylit ihnigTJrc’i* , tl ** initraliu# tîa7
the irnie mrtiphof ii u*dl ift the gaier*! Greek term for the uspirata, va.
1 On the ulbcnttffrc Libitf-dcr^l artiCfttaUcn we Wmt
1 Stc further oNeivjiwJW oo p. 6% nr i bef'*'*-
H Cf- ), ll. Rrth, Tit* Smantiw of Lintttfftfc SciPficr , UnfM. i + '*48-
p. 4«. Bliwmfifl'L Lm$uB£f . pp tea n . 13 e if
1 Of- dp, cit* p- *4?-
* Op. cic, i pr 7*: cf. TngerF iVUi- aa If,
1 ii. 40. td£au yakdrt.
PHONETICS in ANCIENT INDIA
ported by palntographjc evidence'—it k the description of the

vowels that is really at fault, no distinction being generally made
between open and dose qualities;! jt should be mentioned, how¬
ever, that the TP, in discussing the e-vowel, prescribes some
degree of contact, an observation which could again lie supported:*
it is only strange that it should make no such statement uith
reward to I, where the contact is considerably greater.4 The TP
also mentions ■approximation of the lip" for the articulation of the
up-rounded vowel it.* The tradition of the TP is followed by the
T5t which refers to lip-protrusion in the case of ufl and, discussing
the process of openness genera Mv associated with the class of
vowels, points out that this do® not apply in the case of i and u,7
wlulst the next rule goes on to mention actual contact,” It is pre¬
sumably to isolated statements such as these that the AP is refer¬
ring when it gives as the opinion of some sources that contact is
involved in the vowels,on opinion which Whitney impatiently
dismisses ar \ . . loo obviously and grossly incorrect, one would
think, to be worth quoting'.
Against the foregoing criticisms of the Indian analysis it m*v be
argued that in certain contexts, more especially as initials, y mid v
were more tensely articulated than elsewhere, and involved greater
contact than tn the case of i and it; for this we have the specific
statement* of a number of the some of which even pre-
senbe for y a pronunciation as | in such cases"-™ observation
which is significant with regard to later developments.'- ( July on
f rm-uJ. tip cut., Pn)4iir>firmn .No. F,i> [rno.yai).

TTsi* OjOffComilii! iiruvijr * t hr i -,+rrv
ohunetkknti with one of that ftw
yssrpsiqffi ?r-“
‘ y™,' *4,1>P clt< P^toswtia No. <>e. It» ftthe:. heha:»)
I twL. W <)fK 07 (I s. Unarm}. vt
* 1,1 ^4* f> ffhnp gj/i mh dru ur m-nr
* M'XrQfrprAkfUr vffhmt tfirghav . , r ,
* *3£TS£3i„ to mm, * wS1 ^ "• -• *

•• yjr “«■ *•*»**,.» ■*«.
p&iiidtiu ra podddtm tzi itMfffyggAîjfrahffa
... - . _ J* ****lfj v'V*ry'-> >« m ya if I
i i ihfitfTO, 9-1 j.
(f „ .d„o • pteuiJar Iriuc tn,,p v- m

cert.® C,- , let tt» K-r.T,ujffio" r.f the k[oes Ym £,,S)
w.g. StT..v®.o>PkL jbro (> Hindi J*w>, Ac. Ct slw the dcvelopmrnt
PRUCESSES *9
such grounds could the doctrine of a speck! degtra of closure for
the semivowels !>e justified;' the earlier treatises, however, quote
no such evidence in their defence.
The Sanskrit term for the category of semivnweb is rntta^ytha,
lit, 'standing between'. It is tempting, and has tempted modem
commentators, to interpret this term as referring to the postulated
"intermediate' degree of contact discussed above,1 or, like our term
"semivowel*, to their phonological alternation.' The ending -sthat
1 standing1, bows er, is more readily applicable to ihe place which
these letters ootnpv in the alphabet, vis. between the stops and the
fricatives l* fctid it is doubtful whether the ancient sources provide
evidence for any other imcrprctatlotu
A comparison with ihe undent watern classification is here of
^mc interest. It will first he necessary, however, to mention i hat
the Indians do not set up their vowel-consonant distinction on
exclusively phonetic grounds it has also a phonological basis in
the structure of the syllable (see further 3.20 below); from this
point of view the vowel is defined by its ability to function as a
sonant or syllabic nucleus*—as the HP observes:
A vtmrl with Ji consonant or even by fisuJk forms a syllabic,''
and it k significant that Patanjali etymologizes the word 1vara

('vowel1} as <*svayam rSjalt — 'k uiitortommilh7
In Greece also both type* of criteria were employed. Pluto men¬
tions the classes of {lit, * having voice) find di^sa (lit.
backing voice1);* these categories, exemplified by Grech vowels
and consonants respectively, appear to be set up on a phonological
basis, and might he rendered by 'sonant1 and 1 non-sonant\
Aristotle goes on to relate thi* phono logical distinction to the
phonetic criteria of1 non-contact' (airi3 ; cf. Sfct. aspfstn)
and "contact' (ptra ttpoaôX^l cf. Ski spiffa), Plato further men-
ti. -\v* in Skf- eano 4 mrux ffuridi frrm, See. F*« further 5 K i halted'.
Onytrt Ite'rioti'nxnt oj ike Btngtalt Lanûm. i. $ I 3,1
* Cr ijio palllOgmiD of y (m Vi, Vi^y) freudfo ihnt of I j^vsn by C?nmim&Bt,
<Tf>. dt,, t>. 77, fijf- 9i.
1 Whoa*? on AP k ja 1 Cf. Renmi, G* .S\mtrr., J j.
* Cf UrPt* on JIT l V tpiirtvrmcntfm smt&mtMp* fttfkitntily miufmhiih.
f CL Pike, op- cit., pp L
* iviii. 33. ttivytlMjan&fi tritrundruS hddha vfifri scar* ’j^fw3^
1 I- ii. i, an Fin. : Li. iKieihum, L ich) it^ryAM rJ/rffir# weanI emttog
bkmtzii vyatftmnm The word i* in (net n> b<t ralifed t-> rh*? n»i nj-k ‘mud’.
1 Cf- alflo Eufrpiiinpi, Fritf . 57B <Pdfawm&ih 31, k f-
tions a sub-category of consonants which have "noise but
no voice* or 'no voice but some sound (^fldyyoc)1, and which he
elsewhere calls intermediate' foiwx). Aristotle refers to this class
as "half-sonant", and proceeds to define them by ,t com¬
bination of phonological and phonetic criteria; the 'sonants*, he
says, arc 'without contact and independently pronounceable', the
'non-sonant*' arc'with contact and not independently pronounce*
able , whilst the '’half-sonants' ate ’with contact and independently
pronounceable’.- The mly actual example of these "half-sonants'
given hy Plato is—rather surprisingly—s; to this Aristuth- adilsr,
and .1 hill list is given by Dionysius Thru, followed by Dionysius
of Halicarnassus, viz, s, z, I, r, m, n.* Thus the Greek ’hair-
sonants' turn out to be the fricatives, the liquids and the nasals in
the absence of I or r vowels and (in Attic-In nic) uf y nr w glides,
the question of a phonological category of ‘semivowels’ (the usual
translation of does not arise We are here in tact dealing
not with semivowel* but with 'continuants’ of various types, some
of which may have quasi-syllabic function outside the Greek
phonological system—ns Dionysius Thru* expresses it.
The) ot called Half-wiwni in that, when used m nninniirin^t nnd
hinuias, they arc only Iran sonorous than the ’wruiiitt'. *
And it is noteworthy than he Latin grammarians generally include

amongst their 'srmwocaLtf the Latin fricative t*- but not the semi¬
vowels y and W {/. r ) ■ Some Greek sources stem also to have
classified as ‘half-sonant* the h-element nf the voiceless aspirates
ph. th , kb t\ %) ' the inclusion of the aspirate h- is rejected by
1 Cf, the < irniiJn use of th<! nntl Tieiaupchbut' (e g. Dicth, Vmhtmkum irr
Phantuk. Kaooff.l; »ce *bo BkemrieM, Lang***, fi 9S.
* The relevant pSsiijjet from Plato urtsi Arutoilr o«: Pinto; CraS. 1t,.
PM.rtnUThwi. wjt. Aifantle: ***.«. .«*!•, i45T»i Hitt. Aw. I*. 9. *u«‘
&1. Uhl*, or ii f lldtker, p. fi3, * <. ,r
(( - **hwir \ Cf D. Hal, l>* Com^4 ^ f.b 7S1T.
C f. Mirciiifjii, Ltxiqup Jf iPijj finguii.'(• ;ar p
4 L<KL «L
***"* W &n wafib** TWF iftlfltar™* ir „
TPLI ptypsh «ai tnypaU
*v- Pnn-Lm csjarmlj ds.ia^m* wxtli ihi* mcln*
*i«n tjEflUa tia y, i t}m
* Ibkt, ijs
* a. Sestifl Empiricus, Adr. Grenv, fAfotfi, i!, tea. It m a|«0 to be noted
rhit in ttirtlm ufDwgmM Brtbykmfm iht*.pirate* *re not included Jtmon^t m-
iiO|a(cf Dio|f- UtJl vii. 57). A^rttiptu n been maiji- togyplum ih™ j-|pfflffrej-
i ion by E*iummR m tSrksteor fricurK* nzbinthm (cf. SauimnL op,
processes J'
Prisdiin,1 but appears again in the Old Icelandic grammatical
treatises, which also include the Icelandic dental fricatives.-
The Greco-Roman tradition of the 'temhwcalh' still find* expres¬
sion in the work of Graimnont:
Lcs acmi-voycllw mtu encore ^rninetnmtm del ipirajitea e: hums bien
d» fricative* rt de? cnaMrirtjvrt-1
There is in fact little common ground between the Indian ap¬

proach to the atitahttha and the Greek approach to the ^www,
The only mention in our Indian sources of a contrast between
instantaneous and continuous articulation is that of the RP’.
For the stop* there u momentary contact: for the vowels and fricatives
tlicu is continuous non-coniact;4
and the only Western statement of a special degree of contact is
that of Martus Victnrmus:
Semivocalei' in enuntt&uune propria ore semiring itrcpunl.1
Our own term J semivowel' has its origin in the Greek
through the medium of the Latin semhocalif, whilst its employ¬
ment, though not its justification, generally corresponds more
closely to that of the Sanskrit antahitha.
It is further to be noted that our terra ‘liquid’, a word more con¬
venient than descriptive, owe? its origin to the west rather than
the east. The Greek term $yp<k. lit* ‘moist’, ‘fluid’ (translated by
the Latin liquidW) is first used in a phonetic sense by Dionysius
Thru, who applies it to the Greek I, r. m.
n :* moat of hi* com¬
mentators interpret the word as meaning "slippery , i.e. 'unstable ,
with reference to the metrical effect ol these sounds as second
members of a group stop +- liquid, where a preceding syllable con¬
taining a short vowel is of 'doubtful’ quantity, a state also referred
to as C>yfnk' Terentianus Maunia. however, explains the term as
referring to their luhrica nature’, in that they may function either
Bit.,, Pirni. nj HnrtflU Grttk, ml Purtua, pf> W ft ): but tucll * praoiflieistKB*
owuiot bt jupperied » a* secnud rentmy b.c,
* J^dc. dt.
< CL Cadta Upahatdi, td D-hkiup & jcvnwn, pp, Oi, by*
* Op. Q&r, pr 77-
* sail djF tt. tpfUim asfldiam: tt'indnun'drwmwdm mpftfcm UfaUtm.
* KjclX vi. 33.

* Ed tJHNff, p. M= EWck«,*,*3* , - . „ „ „ „
3 Hiljctird, pp. +fL 34a; B<kkcr( pp- Sib tf- Cl Pn*ci*n. ResJ. II *s MIX.
V»£tonjii*tK KtitH wi, 11b,
^ vowd- or a* corswniMJ1—a remarkable interprets km as applied

m ilir phmicdogseid systems of Greek or Latin." Atilius Foriu-
natijiniib sets in the term a reference to bdt of tenseness
mittmxiwum kabtml);J and other interpreters of Dionysius Thrax
refer simply to ihdr * smooth and even articulation1.*
"Liquid1 is in fact one of those terms of which Gnimmont has
said,
'Ellen ncmi eonificntei par an lonsi empkoi* ffnicc auquel tc kcicur §ah
iinm^dtaftmiint dc quoi IVm veut pidcr; ik* iippellntHmij iHfUVdlcs pour-
rsiem tire |ilut i&d Iquia tri d&jr lc min it stvania^c’.'-
1.113. Retroflexion
Amongst the imra-buccal articulatory processes we might have
expected the Indians to have mentioned otic further feature,,
namely, retroflexion 1 This* however, is generally discussed by them
in connexion with the place? of articulation (see 3*03), and also in
relation to its prosodic function (see 3.10). To consider the retro-
ilex articulations on the same terms as the velars, palatals, dentals,
or labials is, even from the point i>( view of ihc Ludiaii descriptive
framework, not entirely jusiiRed,
In the TP we find a prescription regarding the position of the
utioibion in their quiescent or 'neutral1 state (a close j undid tu
Sieven1 'Rubdage'*' or 'IndiffcntnaLigc')^
.. 1 he tuntfor bevtcxukii ntd dcpreAwd, and th< Up* arc in the position
for a.*
1 Kell, vl 350.
*r ■ GVdzwui udtn mmtmi'
hdirita tii mtww in iUu JHfJhgii#- ft nUernm
miw rnf»fj poaiii mam. HUfli Ttnmttrtil fouujuof
1 Mm Jujliftalile 11 u-t of die trim (5^. Gr . p. 11, f 17. B | +
od Skt. h Ui f» \}—-"Vowth 'A'hkh atc Ruble tu be dunged ifkiu *emifowdj . .
liquid toifeU1- Cf Rfnpu, Or -SfijiHT,, | 5.
1 Ecil, vL 17^
* llitgsrd, p. 46; Betdicr. p £17: cf I'wllifnip hoivscrunte, Wrtsc, O. tiL J13*
and mLeci D* .Jitdti 8031>. C£ alw> Jikc-lmm. hOb»irri«Njm iut It
cljitaerneftr phooolt»giqu* dcs cotuoniut1 inf. Cam# PA*/pp. ^-Kp ô:
’ll *trnh!r qof f/rtt W tail du |fjjiimnrf oui of d^vmf pmir 1'imnfitrsiinn oemu-
tiqur tin contj-iitirt efi qutfitk®',
* Op. cvt., p. fi * O. <f /‘IwiH'fiV, § 5j,
T G/ J. LjmfpAîicibx*iV, p. ij; ti *k» Sweet"* *Qretnk UbuuEi"
§| i B4 ff.'i artd VicTor'i AmkiR*Twmih**5V {Rfcm. d, Plm.*, 128 fF. >
1 h. 215-JL p^nîrdji/nii. .jAdiir+-itri-r^j'A^v, mt uhaL'h Tjihh.,ymirm
, fc , amddtfwtutra . . .raYri rf - .. fri? rriwl ■ kliÛi bhuvnij (cf Sicvttl,'DJC ZungeUo£t
■chlaff in def MuilrShtihk')■
PROCESSES n
In litis condition the velar, palatal, denla]T and III!'ml articulators
a*e approximately opposite their respective places ol articulation,
and the utterance of the** series is effected simply by means of the
clrou re-process, already discussed.* This, however, is ttoE the
Case with the retroflex series, which is afticubttttd* as atir treatises
rceogm/x, 'hy rolling hack the tip of the tongue11—iLit is to say,
the place of articulation is not automatically determined by the
application of the dosure-processes to the apical articulator:1 there
is need of a further prayaitta, 'articubtory effort \ which might
with consistency liavc been included at this point/
i.i2. Extra-buccal
i .120. Glottal
In their recognition of the voicing process the Indian phoneti¬
cians make one of their grcatcM single contributions, The term for
* voiced' (ghoimat) is, as we have already seen, found in early non-
technical literature, and tlie specialist discover) is likely to have
been of even earlier date. To designate the glottis the Indians use
either the word tumfha, which in nofi-techtiicai usage means simply
' throat1, or more specifically khah (or hrfam) kanthmyu, 1 aperture
of the throat'. In the Indian musical literature we also find the
picturesque term Jftrlri vina, 'bodily lute’/ which sditu- authorities
have interpreted as referring to the vocal cords f in a recent paper,
however. Dr. A* A, Bake has pointed out that this interpretation is
unfounded, and that ‘strange to say, there k no trace of the know¬
ledge of tile existence of the vocal cords in the texts on the theory
of music'.3 The following are typical of ihe phoneric statements:
This air, rrapimdrojp or pulmonic endtikm, m tiroes of viâl tcfjvity,
! Cf Sinwri, l*kom.t 1 57 ■ Uif ftuJrtidgr ^rnhoipu m die imitirlidur

Baa it Fu? £;* emrclriK* Ankokio^^bc^tm^np wckhtf zm [lUdurj; von
is pro chi a uteri filhrvn 1
* r a. TP li prarirvjjfrflm&rdhm* APL ±x> murdhsm-
ydttdift jihi.'dgfmfi VP i. 78. miadhanvith ptntir^fyagrum,
i‘lw >~3 3June ihia baric drlr 1 ntlioii (aHH-g):
kmmg&Bpt fUioddt-wiadky&rb*f&e*rt wpun:
fiivtifgi mktra-mmihyrrM jihr^gr^ja yatfcu *prtf r.
Cf. LaiJr n» pp. oik ^4-
1 Nat* TndurLflcoy » Ittchiaioa under Pirtr. fmg rgnnu'ttzr \TCLP iv. 103 fcj.
1 r-jj SaÛmLsrpanat i. 4@ C'F. al»o Ait At, iil 2. $
Cf J GtVtSirr in I jYsui.iir* Enryxfopr.Jir Ji in tnuirtjufn i. ^3 5.
[n » paper 'Tur AnnLormâl Background ■: tndian Muaic'i rrad at I he
Leyden Coeigtmd Oncstkliiti, June jgje-
hedonic.' hfrirth <tuSui) nr vmi:e i imfo) according as ihe glnttii is open or
clmcdL*
Whfti the is closed, voice ** produced: when ft i* opm. hrt*ih.f
Hie vvorcb used to refer to the two poltr^ of this articuJbrmy pmcess
arc stitfnjtn, ‘contracted* elosed\ and vhjta, 'opened1: it will he
remembered that the hue* term was aho uscdjri the description
of the introduced proerat*.
As regards the rdadonahip of the voicing process to the venous
letters the j4P states,
lire mb it --Toitud in the east of ihe vokdm cowmanis, and voice m

the case if! the voiced o.hstsiuiiiii und the vqndft1
flic Allfh. aj&o nute* the effect of the voicing process on inter-
vocalic ‘voiceless* stops* saying of the c in a word Bitch as pucati,
11 ii overlaid by the wiring of the prvcvdinf’ wnd following vgrwck/
a feature which is further supported by the statements of the

Prakrit grammarians/
Apart from the two poles of "breath" and "voice*, a number of
treatises introduce a further factor into their descriptions; die RP
observes ihatf
When the gtartEt ts in nit imtrrnedklr condjikm (between closed and

op«i!i LkitIi brvtidi and votec are produced**
and goes on to relate thia feature to die individual Icum as follows,

limuh 1? erratra) Cor rhe vdeekss rtunds and voice for the odien,
except tor the voiced fricative fh| und the voiced aSpirnSrt, wheit bods
brtath and void- xrr cirtineif.*
HP xjti r-3 . rtymb pr^puft kutfhpom mmtpraddnam kantfutiya kj* tivrtt

tomtit* f d tipadyu lr ttaiitiUlwp ndJnldm i4 t ahtrih^y^m.
‘ TPli kanfki* tt&bfr teiyvU: hvtfnh Whjl*l
lire umr«Uy u«|d Cal Vtfkwd* VrfeW, ih. n«i4 tmn for voice u
ind not tfftopi; 1I1U fact te&dv cu pdiuridemjindmt em the ptm pC Jjstei
«ho lift Ktteo mtltkrw t* ^ntf ^c.g. 5uA^j Abjjjn. C3
‘ L. 12-1V, îdi? V^niiir
i n -f on Ptfrj fit eo-,1 i k iriiiurii. i. 3 J.5J piiri^puiviyn? pro*tA&/.
J4Jjfr ef. Ksivy^J. *r«Mrr jfcirflyu* ftwrfW /iOjm-flrr FT-.d
friAjywcj ity nr: huh
• ijitf ;cf.mocKlTMo-Ann
PP77*
* xiii. 3 tr&A^pujK cU«riixrÂAiflb
71 m. 4'tr hvl10 tu mMtih: Mwmuiiuxtim ykînd^ rrdifi*
mdrftiti
PROCESSES 33
Thk b supported by the statement* of the TP, viz

Whm the glntfia is in ms intemneJialc condition. ' h-soundr in produced '*
Fpr voweb ami voiced (tmaapiimted) oonaotianW the emkrion is voice,
{&t vakeleu coD5fltmnts brtudi, and for h atstl the voiced Mpintrt
ph-soundV
Further, the HP condemn* as a fault in the prununtiilttan of h

'excessive breath nr similarity to a voiceless sound",1
Regarding this third category of glottal 'Julf-dottin' or h-
smiiuT, Max M tiller remarks.
Dies isi cine mdUciie VoiaioUiiftg, vrelehc woM riteht m rcirhtfcrrigrn
mt/
and Whitney in a *erie^ of unsympathetic comments*
I cptifra my*df unable i > shrive my distinct idee Item this description,
km wine no iitermediflii' utirtunec liciween breath and auund- * . -
(The RI•*) dviikst* bftth bmth arid sound ta be pirariu in the fumani
jm;‘min'- mid in hr which could nos possibly he true of the latter, unless it
we re composed like the farmer, of iw o sep^rar* patts, n sonant and a « uni;
anil rhifi b impoîbh^
, . . The attempt to eatuMt&h ihist dutinciian in forced and futile. . .
Thai intonated .md unintonatcd breath should lie oniltnl frsiin the tame
throat at once t» physieaJly hnptosfblf 4
Needless to tty* the two western scholars were wrong. I he

modem In do-An an Languages bear ample evidence, if evidence
were needed, that the aspiration of The voiced aspirates (gfl. jhr
!cc.) is voiced asp i nation^ and there a re strong historical and phono¬
logical reasons- for believing the Sanskrit h to liave been Kvoiced
hp (fi] ;* the possibility of such an articulation is no longer a matter
of doubt—to quote one of many available descriptions:
A voiced h can be made. For ttufl totind the vocal cord* vibrate along
* iL 6, mtadhyf faiktirnh
* It H- id arida 'mrprjyaitafjp n tiira^h^fae0m; h&kdrv JiQ-tatuTfhfju.
Intieh. Cf. I. ta-14, (2}mttv«*arjorti h*asfr ^r/\U Ogh&f&fi: w? hak&raht
ifO p/f/j/r-p ifpr
■ iiV, jB. j'ivIpd itiûj:if i it hukfitt.
1 On ftp 710.
■ Oa WPi tj- *On7-P ii. (I.
' See e.g J. It Firth in Huiay, OJtajuufJ IJtnHuitam. p. TX1
' =.(f. Ii V/m» foW{hr.iije Ay aj>w, A* J With*Somklit cf .llerTuticiu
UlcH ... (Aounn.'tevfi, &t„ amt juDCtwm 'if the Type W -fa ■ tnddfa.
i’t Thumb flWIniLli rfrr SiTtrtfa-jf. f iw; Eill*rtun. Sfcf, Jhit- PWialu©-,
H 4J 1.
$6 PHONETICS IN ANCIENT INDIA
* comidrrnbk pari of ihdr length, while n rnangulir opening idJowi the
«ir TO escape with i. mr friction 1
The failure of early western phonetics to lake note of the voicing

process has already been suggested. Aristotle, in a passage of his
Historia Animation,' in fact comes nearer to its discovery than is
generally recognized: in making the distinction between vowels and
consonants he says that whereas the latter are produced by the
tongue and lips, the former are produced ‘by the voice and larynx*.
But the mailer is not further pursued by him or his successors, and
the western tradition is really that which begins with Dionysius
Thrax; the latter distinguishes she three classes of Greek stops
(voiceless, voiced, voiceless aspirate) by their degrees of 'aspira-
tion', viz. as smooth’, 'medium', and ‘rough* respectively,1 the
voiced stops being considered from this point of view as" inter¬
mediate between the voiceless unaspirated and the vniccl css aspirated
stops. It is difficult to set- how this classification can have been
justified: $tiirtevani lias now rightly abandoned an earlier theory
that the statement could refer to u Unis forth distinction;* and
another hypothesis, which would ro some extent lit Dionysius’
classification, namely, that the Greek jS, &, y represented voiced
aspirates, is supported by no positive evidence whatever.»
But whether or not true description was ever applicable to Greek,
it was evidently nut in the ease of lîtin, and with one notable
exception is not taken over by the otherwise ovine Latin gram¬
marians. The responsibility for transmitting to us the still familiar
Latin translation of (lie Greek terms—tenuir, media, and aspirate?
——muel be home by Priscian, who takes over the Greek classifica¬
tion in its entirely. Undaunted hy the fact that Latin possesses no
aipi/aUr, he applies rhe term to the Iamn fricatnx f, which he
' Wefitennum *m> WunJ, Pmnirat Phauiia Jar Studrm, *f AfritmI Lao.
£mq;rir £ digram},
335^ f *i<r u£r v A ?d M a&wm A
*ai frt Xt£\r}).
= Ed l'hhV' P l” * “ P, «‘.j i Trtn*, Me P, rpi*, W*
W +< Xi ^ y fh^t^ ^ Jpyfrm. fra ^
-***> '-»* & cr. Own. JinJ., n* G&mp. Si: Atiîdn
QuinfiEuiiiif i pp a<|. 14 Jihn
Tfl< haui'ni+iM dkliiiflkm « ifi Uck fiitiftd fJiW Lrs Dt Uirfi-
Ufihut Qfjtyl’,
* Op, rit.p p. M. n. i ^
1 The Uwi ritucaSaiifjm trmn and iiipiwta ffur Cic. Mi, Unla\ Atrtrmmrk*
■bit: iWd ttpti.1 trmi ansi (t|, Im-txprT fur rtviâ
PROCESSES 3?
identifier with the Greek ft (tht biter having by lib time probably
developed its present fricative value), The ambiguity which the
term thus developed was ulliMtdy to pnjvide Jacob Grimm with
a deceptive symmetry in Hiî famous statement of the /jwfwr-
tthiebuttg.1
As regards the nature of the distinction between the voiced and
voiceless Lai in stops, the ancient writers seem fo have had only
the vaguest impresaionij. With l and il there is the suggestion of a
different place of articulation*—a method ejF differentiation that
we find perpetuated in Ben Juftsan'* English Grammn The learn¬
ing of the distinction is recommended by Quintilian as m essential
item in a bay'* education/ but he guardedly umiti to discuss the
matter in detail- In a description by Tercmianus M aunts, dealing
with the distinction between b and p„ g and k* there is & sugges¬
tion, though obscurely expressed* dual the author had recognised
the extremely important knis?forth opposition (which Whitney
was rather too ready to dismiss)/ and Marius VictorimiB, para¬
phrasing this statement does in fact use the former term/ But any
good that may have resided in these descriptions quickly perished i
tile medieval grammarian Ilugutiu distinguishes uliqmiMdt* from
utiquu at a by the position of the stress*
licet ctiiiti jeti suit divmae littcrse, babcm lumen ndto aJfmmi wnwu,
<\um\ rx um& itetj pfrimt perpendi *hquu differentia/
and in the seventeenth century John Wallis states the distinctive
feature to be nasality/ Only in the latter part of the nineteenth
century, under the Influence of Indian teaching, does the recogni¬
tion of the vojicing process make headway.
M21. Puimmfc
It will he convenient to consider nest the process of aspiration,
in as much as one of our treatise*, the Paf links this with the
voicing process:
h anit the valued asp dines are voiced. llhc- icmfvcwelii and voiced slurps
1 Kg/. I! 5H>:" Inter r lint nnpiiaiJdtie tl cum (wpnalwnc ftrl *, inlcr t quOqus
et ih t'rt Jt rt mtrf p cl ph «ve / pit fr- Sunt sjfuur Ivat Ire. Sw en A, dw e,
naCiJjj.F. quilt tsic pcil-itui cflrexrl mspimtipfie ti£C caitl J>knjm jxundcnU "
1 For dun^.>ion cf. Jnppticn, p. 44
1 Ktil, vj. n. 4 1 iv tfc
! Cf "On thtr ReUtion tif Sunil ujkJ .HHuianr1' Ironr Am, Phil h^n. 1^77,
ppm 41 fT, * Fci irxti he Sfurtewii, np. tic* f i&>a.
H«‘r, Of *\t#dn A**i Studih Phitalitgim, f>, 34,
* Gnanmmtav Infu« -‘Ipyj'lftftttir. pp, ijf,
JB PHONETICS IH ANCIENT INDIA
»re partly voiced; the voicclew iiipirttc* hit brnnhcd. die voiceless jcop>
axe partly hneothed, Thi* is the lun gf speech.1
In other words, h ant! the voiced aspirates arc considered .is mote
felly voiced than ihe non-aspirites, and the voiceless aspirates
more fully breathed than the non-aspirates, In the case of the
voiceless pair the statement makes {rood sense when we consider
itata, ‘ breath’, as referring to 'force of voiceless breath' rather than
simply 'voicelessness1. 'Phis interpretation is supported by the
statement of the TP that.
More brcHtli ti emitted in the other voiceless consommes the. the
Mpirated <jt(3ps and the frie#tivc*iJ than m the mmspirated sropa/
and is in accordance with the grammatical and later phonetic

terminology of the distinction between aspirate anti non-aspirate,
namely ' mahu-prdrtn, lit. ■big-breath1. and 1 alpa-prUna', lit. ‘little-
breath'1.6
If we now torn to the P$’s statement regarding the voiced pair,
simibr considerations apply: nadat ‘voice1, being interpreted as
'force of voiced breath1, the statement implies greater breadi-forcc
on the release nf the aspirates than of the non-aspirates. The
justification for such a statement is clearly reflected in kymographs
tracings, where the voiced breath correlate* with a particularly high
ampiiHide in the vocalic wave-forms;3 from this (mint of view h
ami the release dements of the voiced aspirates may be considered
as an ‘overblowing1 of the following vowel6 (cf, also a.oo below).
The Indian treat men: of the aspiration-process provides little
else fur disnissuin, with the exception of a statement in the HP
that,
Some w) that ihe uspiration of the pirate* consults of n hoiflorganic
fricative/
* 30-îO.
™>'W>rn ha-)hiifdtt UHftiSfy
hun-midd yt/fi jaiai to hînut lu bb^phthtuynh
itat-th 1, ifofffj torf> ruty&J yt/r dhdmmm priuaktut*
, Flalu, Lwt. 4J-II I - - - feu hi roi Tr,i, ftri <m rnff •twin tin n>G fpjrj,
VTI TO yfMHîtara),
MOjirfa The l'\ isKo, Sin: t.lldert, p .)*J
Kiv« to till, tpcetf des«e ct bitaLhbnwj riic tide ul taltii.
’ e if, Mbh 1 i 41 fin ftla. I i. ij{Kielhom. i 6i);Ab j}, ir, jit, Cf A C.
!)», Frof- iwl Int. Gouf. Pkon. Sc., p.
■ Cf hmsii. op, at , Kymotnmi Sm, 8j. p* (flhpiahur, bitafchmti.
(. f. f’itw, op. «t, pp 71 t. Dokr. CfflTjP Study m Shona Phtmtiu, ». <u.
iiiL |6. lo/motdiji ta tOfmandm ififfifwHiuh uitthdnnui.
PROCESSES 3*
a view dial is repeated for the voiceless aspirates by the Ap+£.*

The breathy release of an aspirated stop inevitably has1 as Sweet
observed, ‘something of the diaracter of the preceding consonant' ;3
andt in the voiceless series at least, varying degrees of oHrication
are to Ik- heard from speakers of scans modern Indo-Aryan lan-
guages. Dialectal pronunciation* of this tripe arc likely to have
existed in the case of Sanskrit; the later development of ihe
aspirates, however* hardly supports Uwpa's assumption of 3
strongly aJfricated pronunciation (p$, is, kx) such as is attested
in, for example. High German,
i.c*2. Nasal
The nasalization process may occur in combination with various
mtra-buccal process, and die ancient statements regarding the
mode of combi nation for the most part present tto great difficulty
of interpretation. The nasal consonants are referred to either a*
nasikyn, +nasal'. nr ammOfika, having a nasal component", Of the
mechanism of the process the TPsays simply tint,
Natality h pitHlueeti by opening the nasal cavity.51
Our phonetic treatises, as also hlnini, realize that both nose and
mouth are involved,1 ami ihe TP further points Out that the articu¬
lator b as for the corresponding oral consonants.5 Applied to the
slop series (fparia, this process gives rise to the nasal con-
sonants jj* ju n* nt m: nasalized forms of three of the scraivowles,
p, 1, v, line also attebted aa junctional features in Vedic Sanskrit
and duly noied by the phoneticians ®
In connexion with the vowels the working of the process is
similarly quite dear. Here again the term mmneuikit is regularly
1 k\ I) faJihdnrrzji ufoi.
1 Primerf p, 59; d P^ce, op. let., p n j,
1 ii 5* -riwttd r*4J dmsx&dkyum-
* AP i. 27, mim.idkandm mukka-ndukitm
VP 17^ mukhii-nihikd-katm>' 'nwrJrjkxh i
Fib. J. i 8. mtiktui-'miitilukaH
1 it* jt wfiwitr i-dlfiE
1 CL TP v. lS. MmiatthA-parai fit ftftiu'rtflffl amituUtkatn
e,£. lor wmH >iidhl
lUfiirgul tokiUU Jjji r lokiiitl
ynjjmv vmitn far yajpiui}
Only 1 found is CIiuhJezI ftanfcjit (*nd *uly whim - phooftJoguai -n). t ii
nm nnciied f*e±* however, Kjgfwjw, op. c«.a p, 339); din u sn aecordiacc with
the non-occurrence oi r fcf. jtmrUnni of ihc type +■"" > -U r* for
Uprctd •-iiT-i
** PHONETICS IN ANCIENT INDIA
i is-exI, m opposed to the jWdfctf nr 'pure' nnti-nasaibcd yd web,*
Another icito, how^vef, is also used hy seme of the treatises,,
namely takta, 'coloured’, oasaJizatiun being referred to as waga or
ruAga. i.c. (nasal) colour 1 Some of our authors give picturesque
descriprions of the quality of these vowels: the account in ihe
Srtru(mtmm®ta*£ik£ii is as follows
i’he nsial t^yr akimld afrjrt fitsm ihe hnft,1 wllh u scumd like rhut of
Ixlls; juvt ii the milLimjjkl* of Suraffn cry *tafcrA&U1 rfauttcimilkP). so
should the naanhiy he rcaJistd,
The rusdiacd vowel# are not of frequent occurrence. They appear
in certain type# of junction (e.g. Iriir ekauda/aa iha for trim-I-
ckfliidajmin—iha), and ;is features of finality in the sentence or
breath-group/' Apart from such ease# there was a tendency, cen¬
sured by the RP but general in the modern Indo-Aryan language#,
for vowels to take on some degree of nasal ‘colour1 in contact with
nasal consonants."
Uut apart from the above, the accounts given bv our treatises,
as also the system of writing, present it# with a third sub-category
of the nasalisation process, The name which thU third feature
hears is 'amavira [t/i, tty), which might be literally translated
either as 'after-sound' or ‘subordinate-sound1, ‘rhe contexts in
which it may txjcur are dearly defined. It is restricted to post-
vocsllc push inn, and its primary context is before the fricatives J,
0* s (and also Is), in cases where historical and phonological evidence
pmni u* un alternation with m or (medially) tl? at an eariv date it
also made its apficurance under certain conditions before r,*J and
in Classical Sanskrit replaces the Vcdie f and S before y and v
1 AP iv, [ 31; r&%apa$aUm liL %
* *'€■ RPi- rukta-Knjijfo'mntlukvh
1 4s
WoW'AntKj- vmaw rhrdwdJvtlMtetnI hkmul
rwMj ntM'Utkr$j rty tthhtlMjur*
rntyijl prayakimydh, . . .
cr Vnfma, op dl.. pp, 1** t
* V:c Mrfyvrfy^&te bui, ‘note r^lmidUy. ndutt tdWptf* rmst/h .
t F aUn the ^called vtvtHUSP u. fej, c.e. satiii imJmli
fur i JCdki -findrah,
* L f fif11 lij Jj/ufr, wiwlnt f /rw/hyiln dcJrydd/mr anumhjjtJn ti&in.

Sait tm >n>» tla-SihJ 4b. tint* 'rttnwit faddatu • itkn ndfity* rtmzi. ttmjHakoh
’ nr xiv. (6. Tsktut MimtcSyt fr arJ(iijj|. Cf. *Lnj xiv, <».
* Lf, lam aarram btijdr tim apl, fa., huujsu betide Germ. Gant
Gk. jnvf, Bte. i*ihtmt.y
* VtJ vimnut trade MIQsut,faunions the pwteticrity nf ii,e seuuefic*
■mr- (c£ WmJcmiiKe!, <tp dt., | *Sjf j
PROCESSES 41
respectively (see above). Its optional use is further extended, even

35 early as Panmi, to word-firiaJ position preceding a stop.1 where
previous phonetic teaching had prescribed 3 homorgamc nasals
Inter treatises extend this practice to morpheme-junctions within
the word, and even to iatra-marphcmic position,1 being followed
in this lost extension by the graphic practice of manuscripts and of
some printers, In Prakrit it is further extended lu the position
before an initial vowel*4
The phonetic value of this feature* however, has provided a
problem for phoneticians and commentators whether ancient,
medieval, or modem. Some confusion may perhaps he avoided by
first considering the evidence for its pronunciation in the earlier
and more limited contests, namely; before the fricatives. Sequences
of the type -Yn)S' (where V — any vowel and S — any fricative)
were phonologicolly parallel to sequence* of the type -VEX- (where
L — any semivowel except r) or -YNT- (where T — any stop and
N = horn organic nasal): aan]*skrta, for example, is parallel to
sol day a and to san^taana, sutu-panta, tec. Thus the nasality
tn the types -VtL- and -VNT- conforms 10 both the sthana and
the dbhywtfaia^pray&ttut of ihe following con son an 1; by its paral¬
lel i^m with these sequence? one might also make the theoretical
supposition tlcit *VfqS- = -VZ3» (when: f, is a nasalised frica¬
tive), a farm of realization that may be heard, for example, in
Modern Icelandic, where in a phrase such as t&lin %kin fegurst the
word-junctions arc realized with some overlapping tif the nasaliza¬
tion and friction processes (in a segmental representation -z
I*),1 In only one Indian statement, however* is there any
implication that rrj is to be considered as a fricative*
In a number of languages the tendency is in fact for the nasality
in such contexts to he realized in the preceding vowel*- and for the
* VIII. iY, ^
; Cf TP ’i *.7, ttwtiifriiA ipitrin-piTrm mjTa tarthdmjm awtadMikm
p Cf. Satrîfpmata S. li-.tf 'U.uiJ-J Jliiinam. iiufljVp
* Cl. P^sditl, \ 34S [Hcniacandjii, L £4).
1 Cf- SLcfin EtojuTLion, lidanJif iirtimmti*, pp- IQ, JQ; jBtifrijgt ±Sif Ph>frrtik
tftr Iil3ndiitf\tn Spratht, p. i®, Sore dm Petruvriri, D* ia munhtt tn Hmmamz
Cr pi trulptiiJii If ku C art i end at ion qui dfi'knt yiratiquc 1 «lui dr La
catnaime iuivume, nuiia mini lc mode ârticulation/
* HP i. to* with l v afa'i unimnil. I In ihr (fmirfiil mrtttrrt of iU'^1 friiilim
cL Grunmont, Traiti, p. 95; Maninei. TCLP visi j#*; and on the tocom-
pJiibiEiTy of frirnon und v&bt m SkT, tf. p, 4+, n- 4
' Cf- efpeciflliv tha 4c\chjptumi in Aw&iMti (Barthelomaa, GV f #o£-
dtgjtugm bciiijff Ski. damtaht mq$r/m btaide Ski. mmtram.
a PHONETICS IN ANCIENT INDIA
syllabic quantitative pattern In be maintained by j lengthening of
th* vowel; as Sweet long ago pointed out with regard to Latin,
n before the htx*n mid semivowel* represented * nasal Ictiflihutitnn uf
the ji f sved i riff voweb1
Hie Same development is postulated for *->»- in Irish,- anti is

further Attested in Old Lithuanian by alLtTmtinns of the type
kJmlu(pTvs,)iM*iu (fut.J.> Amongst the modem languages parallel
alternations may he quoted from Polish' and Spanish/
On the basis of analogies in other languages it is therefore tempt¬
ing to assume that the value of itj was a nasalijtution and lengthen¬
ing of The vowel (if not already long); this view was adopted by
Whitney,* and has support in the fact dial the TP speaks of attui-
mint aa having precisely this value/ whilst (he AP mattes no
mention of it apart frtfmunwatika* Whitney further supports his
interpretation by the fact that the amtsv&rtt symbol is written nerr
the vnwel-symbol (c.g 'tin or ^ for unjju). The kngihening
of the vowel rs implied in the statement of die TP that metrically
a syllable which is nasalized is equivalent to a syllable containing
a long vowel.y
ihlt on (lie other hand some of our authorities quite certainly
distinguished the terms amnmika and tuam/ara, as thr following
passage from the VP indicates:
According to Aupasivi iuiwn,iiiha of • vowel occurs only before a fol-
liiwirs|Lf vowel lUvflm, Vie. maJiifl Lndeah-|, whilsr before n following
Consonant thi:ft is on iastmoti uf antuvostj {Uvula V.tf ^vBtJuutit
tvAjcre'j,1"
A viniil.ir view is taken even hv the which in most respects is
* 1*™- PM, £* (S&1-4. r XV. Cf. Stairtvnnf, DO, eii., 4 i74.
* CL tew#, Ktflurhr Gt, i, H i).iC, Ot J
i .ntirh, Ltlauittkit Lrcr&uth, | *5,3; note aim present feme formalinm jg
the fyj* ts.i/ti. The n^.lb-eJ vowd, tf Old I.iUt. (tj, Ac t an now pwmwncwi
ttti tifrti^fnwal Irtrijf fmvuli
1 Cf- Un«&. Shniifh* Pfaxtfik, J 1*4 ;-Sf>L-ritkikU-!zdicbikq, fotiJt Exfibi^
du Coxmmm \malm rtt Fctanm pp. o V. RtndmmJct
^ f^gtSpr^viMw 05 tntl rr. '
% Cf Sui-arro Tutus*, PrTitwwifcim Etpufirfi^ pp ml
1 On TP u 3cv, cf. CpfimrnrmT, \yp . ctl,,. p,
cn. tirmn artitirimd mth 1, T ■ 'mmdrik&L
1 Ci VsiTTm, up. tit pp fi, Ktnuu r.r
1 Liii. 14, Cf. modem dew^nmm tydi hi Hindi Mm<$kt. wa>|ei; «-
alK? K L TuflUT* tTkm IndoNMi m ljujmil1 {JliAS i9ij)p p.
u' lit. IjO-L ffttMfwInfrsfm
upadkd mira ,'Iu/jjj[tttA ■ mmmt&M&l
{Unfa: aynm up vdh t} ■ nakfa avv t cut Or A bhj^-ri:),
PROCESSES +3
closely related to the IP.1 The attempts to describe the phonetic
value of this inwrtian\ however, are remafkably unhelpful. The
P$ describes it as 'sounding like the fttte' j1 the RP classes it with
the vowels and fricatives as lacking contact.3 and elsewhere says
that it has either vocalic or consonantal qualities;1 and at least one
ancient grammarian could not be certain whether it was a nasaliza¬
tion of the vtnsid Or a separate nasal unit ■
It is thus extremely difficult tu determine die phonetic basis of
our descriptions. Translations of the difficulty such a* ^pirantical
to vowd-likc n (m)1* do not realty help; metrical consideration*
alone provide no clue since syllables of the types VNC, WC,
WC would in any cosc be metrically equipollent; and modem
Sanskrit pronunciations arc so various with regard to tmim Cira as
to he of little assistance/ It seems certain, however, that some of
the undent writer* had perceived something other than a simple
nasalisation of the vowel; the linear nature of this feature is indi¬
cated in the VP quotation above, and we have also the statement
of the TP that some authorities prescribe the insertion of a ‘nasal
mcrcmeiH* after the virmAJ In tile present state of our knowledge
it would perhaps be unwUc to say more than that these authorities
had observed same form of nasalized glide Lq the transition from
the vowel to the fricative.* Their difingreemcnls as to its vocalic or
con*ommiiil nature may perhaps have a dialectal basis; in this
connexion the following slight indications may he noted on either
side;
(i) In describing the mode of transition from a to i/u in the
1 VS i u*) (Ludrr*, |> H5I.
4 1} {iddh u-fif r r ■ I - rx ry h \t \ .iiv \
1 Jtiii. IE- mprtfm# itwtam
* i. 5 .ilTtrlTrtr.. :-VdTfijriH.jtTi «ffi f4f.
1 Cf. RF iiii. 17. Vydlir mHiikyum OTmrujukinn rtf.
' Ch Ilc2ii.tc.c3i*!, tip at-, pp, 146 F.
Cf. Giicfion, On l)lr SJacVm Vrm&vtiitirf, p lot r dfO
JR.4S Ctrl! Stfpp p. rrijffn ndKKlll, m JJ^vfnf crnirn wuntli K'uir
neeEutlid, sin* technic bandit* tifl &attl of îpcsiirrlEi U> cnfu-r-■= thsut
cemraeo i2|Hjai the pupiUT An csimaie cKampEt it provided ivy the te«JuiH
tirsii rtf ro m fifljamdk rteieatitm -.1 L£hum|, [u£hutnjtm-i4) for
unjjimw,
1 r. 3-3, uflAr?4î.' tMm ft! wnvtyamk {Tr^h ^mndt /w* "mm fm dtfcznc
bharali).
Cf, diieui4ieri Betg^Lgnt-WhftxMj in MSL ir. 31 if.. 1^4 rt.
* [foth Etnd; (ftjs. eil p p $?', Cimjtituiitnmr* \it Ffeolt & dtt JutittrJ
ttudet 1 ijli „ Parti* vritnltslr. p. 65.? oml CbiiltcrjL {op «!«. £ jjd) stem to Mijufoal
1 ! rngtheti m g oF the with niwllration of ibo ratio J ponton,
diphthong; oi/iui, the HP com|u«s it to the union of u with

tmusviiru1 tills comparison, with its implication that anj in some
way resembled a diphthong,J is <m the fate of ii reminiscent of the
Portuguese 'sonorization* of nasals (c&ttStK. Lat. ({&),'
Some support for such a realization is perhaps provided by Gnya
bau8a<Skt, van)fa, tec*
(ii) The treatises which insist on a consonantal mterpretstion of
anuTvUra (generally velar)* are of late composition,6 and this pro¬
nunciation is expressly condemned by Uvapt in his commentary
on the HP.7 North-western developments, however, an exemplified
by Smdhi vaujku < vumiu, harija<LfiQmjnt are suggestive of a transi¬
tion -NS->-NtS- (cf. the Sanskrit sandhi muhmintsnn for
mahnaitjsan.
The extension of attusTtlra in Classical Sanskrit to cases where
the follo w ing consonant is a semivowel provides no fresh difficulty,
Bui, ns already mentioned, Panin i allows the optional use of tt) in
certain cases where earlier treatises prescribe a homorânic nasal,
It hoa been suggested that this extension was purely graphic, and
based on the convenience of the simple symbol to represent
nasality f this seems Ices than fair to Panini's linguistic competence,
though the possibility of a phonological rather than phonetic basis
for the extension must not be overlooked,
Graphic Consideration* may well underlie the later extension of
f iiil- 41 jjf),
4 m m fart o>jptf)cipifd will! »/ci* uhcre iht quaBitiitu of a <md l/u hit
fuMti' ! SUL 4&. Mtiftxud): tf 1, 13 btW
4 Cf, Ufj(jjLj«, fiifatfnJi Jr lifTzuutufn: kam.-.n?' §333*; £, llr WiHiaiiw. From
U1 JWijFT+r^, § 157; A. R CuikiiKt* Vitmnn, F'-Ttugu-ii: if
phimtiiogit' $j 4i/-ja_
4 Of liitmst. though of doubtful hiitmiaLT isitnific&ncr, 1* the Maraihl pro-

iiimciation oi Skt, ry. aflj-11 ■ rilutfa f ■ miirtstiO,
r< ft&fpfrf}, "i#h# (■=- HTYj/w-) If thb ilftc* in fittl rdlcct nr: hjiioricA ftaiuir,
fhc rcsJJxfttkin *if might he »imply natn^l \n icmu pf homurfUnk
UtiOjhman fur every cue, including the frbujvvtfr—*_K. *-um t 9- > *-ai a.
fnctidv and Vine* being mutual V pnrompatibtv. ihc heavy • |U4ttTitr of the
lyllrihlc. if it contain n shim vowd. ii jiuuniamriJ hy a Vf-iypr v^d~cU»wuret
(cf on TiHir&fandiU' p, 6HP n rj; thru* *'01 B- &c.
* Cf Vantii, *pr cit.y fi. 15, and F& 170 {Ulster*, r £7). Note ut*o the
prrttcir-day pTonunsisibn of Sfci. it> hi p.
* Nolt* bom!W, Ihr VF\ rektttt in a Parity of .in-ijwcjfirt! lypcfi.
rifotafa^jflAnnrfrd /kMta’'MdfrHdh ef. jfjp. L lo.
? ii ii- ti-
1 Blrtch, np, ol_, p. 8S,
* £L K. ShfMitrt, RT 54i cf TlbchH op. cii.p p, +q.
PROCESSES *5
anumUra to replace horn organic nasal* within □ morpheme.1 Gf

such extensions Whitney has said:
Tu - p + write thi1 antitto&rd turn in the interior of a word fnr a nasal mute
which i» equally radical or ihcmaiic with the suwfidings ntin-nusal . , ,
ttcani on inclrfcmfbk practice md one wholly to be disapproved and
rejected-5
11sesc striemm arc directed against Max Mfilter's support of the

practice referred to*1 hut there is much to be said on the other side.
For although this 'slovenly and undesirable habit14 may well have
gTo^n out of mere graphic convenience it incidentally recognizes
an important phonological principle namely+ that the n or tn in
a sequence Vnr or Vmp is a very different functional unit from
that in VnV or VmV; for whereas jn the tatter ease n and m are
mutually contrastive, thta m not bo in the former ca&r.* The
homorganic nasals form a single phonological unit,, and a phono¬
logical Transcription will recognise this fact.* This is particularly
the esse in those modem Indo-Aryan languages where the only
purpose of certain nasal symbols taken over from Sanskrit {velar,
palatal/ retroflex) is to represent homorganic nasality before con¬
sonants of the appropriate series. In some dialects, moreover* we
find alternative pronunetalions of the type VNT/VflV i t: an
alternation of homorganic nasal (por^-im m™) with natality and
length of vowd (amnaiika)x m cuch cases ihe convenience and
phonological appropriateness of a single symbol for the two po*-
* Ulhutnutn he re pwridrt aimther parallel., in ihjii ihe mdigEmnu emenman

extend the naanl rmfrkipmbol (which k phcirwticdJ p juii fined before ihe frfeai-
iwtt) m a. graphic de vice for writing the homorân^c nasal before mop% (e.R. pfii
fot pi*u)* cf. l.c\ktcnP upr cit.„ } ad.
Note atao thil the Sxrr'as&Fwnto-£rW, in prescribing thi* extension (jij.
peculiar prOpcrtit/ {\ . , *lja rfrtdttwre
jHrkiniwbrLjg^ ihii >mujrdrrt heir sEado eh
ctfrwj rpi vidharm4ik^t%
* JAQS ix_ Oo it,
r Hit&padria, IfilfixL, pp i-xi,
‘ Whitney, Skt. Or. i 73b.
* Cf. J- R. Firth, Prur* Z.t*d Ini, Cnrtf High. p t&X
1 Note ylî ihr use of 3 ipecid aymbof (^T) for the homorgamc na=j.| in
Airman, whtr® there u.im hurdly b*s *ny qyctifrm Of graphic ^mvcnienrc frt - i,
[J = t, m ■■ C): cf. RaitbuEnmur. a63. 53.
1 Fvcri m Sanskrit: the psdaial Jl canm/t be junlincd on * distinctive baBLu {cf.
E:rieneuur Im**#. mil, H flX
a Cf. Prated. op. ext., pp. 4^7if [±^. lunt»:/lS;toi). Similar iltemiitfon.*
ire historically i-u^g^ted by djiubku in Marathi (cf. BIjjcIi, 1m Futmutitm dt im
Irtngut m*rathrY p. 8i>-
4* PHONETICS IK ANCIENT INDIA
sible rtulitcttiom are obvious.* Possibly similar alternations were

prevalent in more ancient times, and gave rise to at least some of
the confusion which besets the early phoneticians.
Hie problem of ajfUsu&ra has been considered at some length;
for whilst si is in itself disappointingly unrewarding, it serves to
demonstrate how little we might know if our sources had been
equally imprecise on other points. In view of their generally high
standard of compel cnee ii sccmi fair lo assume that the phonetic
problem in question was :± particularly difficult one> complicated
peritup^ by multiple coiiirvuiah disdccuh and personal Huctim-
tioiM- If we were to $mk an attested feature of a modern language
such as nikht give rb^ to uncertainties of a similar order, if not of
type* it might perhaps be found in the Japanese so-called "syllabic
nasn]T (q)f* which has so strangely received special recognition
amongst the otherwise general phonetic categories of the fnter-
nsnonal Phonetic Alphabet;
1,13. 77if Vargas

With regard to The three extra-buccal articulatory procesiea dis¬
cussed abov e (glottal, pulrnomc, nasal) it should finally be remarked
that their combination with the infra-buccal process of maximal
closure gives rise to a set of five spuria letters fur each of the five
r Cf. Enoch, Inc l’jt. l-fl d* «lte fihrmancc fajt one Ic irtiS iigne
J'jmwwinri fulfil * flWf f i!o -Jmi ilitW le C*s efi la PsycHc tfl brtvt, il
repr*»nt* U iiM&ta <.k mfcme ordtu ûc la eonDaniic qui mJt. ,it In vovcllc
lirfktfuc, Pamttedra 4 la mfiriK vikur qttf Ytmttudrika.k
* Refute cttEKfcmniitp* Ihc lrttliitiia for their disagree uimt* and
ihr- de-êripiEnn. At flnuiiifNi, 3* mflj ho mlatary Ed compare ihe diversity oi
ETm Iren Jr % 0f tLe Jnjuuiew feature rn quuuun, tit'1 folluwtn# ttwy he
referred; so:
H. Kftfi, ftitH- di /fl ^msnrt From'o-Ju£nmitii*„ Vlll t. 157,
It- E. Palmer, TA» c/p. m,
D. ]{Hie, îhmsw, p. 88, ft. j
E, tt. E-lward-H, fbodr pftzmdtvpu d* In longue fitpimmie, p, j 1.
M ii. 'Inn* Tfcr /Vn-ftTjnn.T^'ivri */jfapujtfWt $ 1 jrr5.
P- M ^ Lin kt. TAt Pkmeiitt ftf Japaviit LxtHjftiii#/, ppr 71 ff,
H. Obeli, 'Siudm in Celbquuil Japnnesr IV OtodftmiraP, Language kkvi.
i^S°i P■ i«-
S, E Martin, */ SfjWtfJri Citfoguuri Jopatfttr
ItinfrluriM ATu. /7>h |rji. it i
Whifrt nut impLymp advene critkHem orn[l ihe&c anurtsicstt**. we may waoder
■whefhr-f SiTiflUMts in *i lintiint future, nrufinur toids vmDUi fltecnjnlj "-mining
Dram îemi-WMomut mi iemi-VLWcMr' 10 * voiced friruoalretmdiuvtlar iprrun'
- will h;Lvc nny dearer an idefk feesrdiflfl the pltaattic Value of Ihf Japanese
pyUalNr lUxiP than have rvârtlinf ili.it nf tiae SaniLril rwttmfru
PROCESSES 4?
places of articulation—e.g. for the- vclflts k, Ich, g, gh, g, Each
such set h known as a ?iirgay ’class', tile members of which are
sometimes referred in by number as follows:
Voiceless non-aspirate—1 ret.' fprathanui \
t, aspirate —and.' (dtitlya)
Voiced non-aspirate -‘3rd.’ (ifltya)
„ aspirate —‘4 th/ {caturtha)
Nasal "‘5th.1 (pemcama) or 'lasi’(uf/ijjmi),
In the table on p. 20 the block of 5 by g vargtyxt letters ta enclosed

witiiin die heavy line.
PART II
LETTERS
2,0* Consonants
It is the Indian practice to describe the places or articubmon in the
reverse -order to that of the I PA, Quilc logically they begin with
those which arc nearest to the origin of the air-stream and work
progressively upwards and forwards towards the tip*,
2,0q, Pidmonk and Glottal

ITuifl the first organs to be considered arc the lungs, which are
treated as die place of arfctiSaliun far the Voiced h |B] and soicr-
kss This treatment* however* is optional: mofli of the treatises
alio allow these sounds lo he classed as 'glottal fricatives1 * *—a term
which U still commonly accepted today, though in need uf revision
(the Greek term 'breathing' (ttpetFjia) might he more appropriate).4
The following statements illustrate the alternative prescriptions*
The fncimvt^ h and ■ h *n? glottal lAwn/hjwlj or, as some *ay( imJrnmuc
(tiriuyd)**
h nmJ *h urn glutial; the litter muy alternatively be omiaitafvd m
pulmonic**
Certain of our authors allow the pulmonic alternative only in the
case of h followed by nasals or semivowelsi
h before nasals and semivawdi i* to be considered ta* pulmonic: other¬
wise it is gfotfoL4
1 E'urticuLiirSy in yf ita frtquJb£Ll pcvsddic hmctiuii (the Cleri -■ i~.. j h *

wpwtf&ia); ct |, k, firth. TPS, p. 131,
4 RP L pTalhnma-puniamciu t-j dfmopifoi <v Jtwtfipjcau) ■ haul ttrf
utaiyuu.
1 RT 1-3 htlfr kttfhtt imm rthirfonlya ta
4 PS tfc
kakfrtnp purbvmmr yukUim >mJuhiiki\hhii ra mmyviam
mifalyiff lam vjjdmydl kanffiytm dhxtt ninynyutnm
CL Saiv&mmmata-S. 42.
twklimm tidy4J unttiilMru pprOm ft*
utmiftffu jVrrjjn frttnr , . .
l*hc bid 1 *4 ihii d&iimrtian ii prnbLîFuUc f Kit it si ti^ be unfed iJul for nnc
torm ufjunction initial groups of the type h nuâl or wjfuvonel are option aljy
U^leJ *j il the h were pho-tumM 1 c_JISy imlrvyidt hyah. &c,—
further 3,1^2 bclo*)F Le, h u ctimulcml it ■ prt&odk, nuivtiifeir future <4
LETTERS 4V
Wc have already suggested that Sanskrit h [fl] might he con¬

sidered as an ôverblowing" of the following vowel; the close
relationship of both It and -h to their vocalic context It mentioned
by the TFi
For h and -h the glottic i& ihe place of artieulotictfu but in the opinion
til hvftTit authorities h h hnunontEtnic with the beginnmu of the following
vowel, md -h i* hnttiorp^nic with the «ul of the preceding vnwel.1
Whitney's observations are for once entirely sympathetic;

with reference to the pulmonic nature of these sounds he says,1
The authority who called the jjpimifmi L-he*t«4nund* may aim be
Commended for hit acuteness, wince in their production it limy even be
tun! that the throat has no part: it ii only, like itic motilh+ the avenue by
which the hmuh cxpdled frrrtn tlie dim Finds exit;1
elsewhere he shows himself to he in agreement with the view that

they are homorganic with their vocalic context:
Why* then, shall wr ptoumuiee the larynx the 'dwwtnktic place of
production1 uf h. any rticm- than of the vowels? T . An ft Ls a sound which
in pinduced m uny iff of these same pwttiûn of the mouih-organs
fee as im vim-Elil, hut with the vocal eorda m die larynx only aligmly
lippm-ichcd,*
rite RP rather surprisingly lists this homcifgafik real Marion

amongst the Tanks' in chap xiv;s but Uvata quotes another com¬
mentator who considers this statement to be out of place* and who
prefers ro treat it as a rule ntther than a prohibition—1'for not even
the gods could pronounce it m arty other manner*!4
This will be a convenient point to give some account of various
breath mesa (dFr It SO AS xlei.. 044 Lh tiic rmluMiiHi of which may opcm nmul-
tmuhitidy with T.h;it «f Ihe phtfUcltmuc unit*- It ii perhapt this peculiarity that
our i rathe* intend to indicate
1 IL Jfoffftfr- f rbfKSfc hukdr^trnrjuftly itti udtty*->xm&di' wi fAiirfcP /jaftt-Vi*
*h*vm: Timr/mdruft Tribhdfy&rntMi cwptci*m iip
Ab the
NEC l,
they hirer no ortteulutor of iheu own (fwi ay* ft ftonjwfftftotKift), Cf. bw«t*
? L>- Jnngj, Outlius vf Ef^thk Pharutk3\ ft 777 fL; Bradt. Statiidu Phone-
m
1 Up APL 1%
1 I CAnhot a^rcc with Fry (La Tig, xriw igg) ^hen ll? tlitm; The trac of the
Jtbcctjvt nanud nM j-prÂf to br mens than a htifiy turnip* to locals
I he Open ipiram* refirr*J*fUttiK ft in paima.P

* 7AOS vm 350- Cf, IX JojiL-fi, opP dt., |i ij, fi t; ii Ahratami, £/uifrr
Ptau&tiQm iw/fi Trndifii 1 dtj Oidurivrj |>. iox.
* SiV. 30,
* Cfrui-ifAJn* . * * dmu* apt *1*t iukya uftdntyiium.
■ UK £
S* PHONETICS IN ANCIENT INDIA 33^.
prubkms connected with *h* This voiceless breathing primarily

occurs only in final position in pdu$ur where kbtoriealiy it replaces
Ms {or less frequently V),1 Corresponding to -h in pnusa, there
appeared in junction with initial voiceless consonants the appro¬
priate iinmorgamc fricatives (vis:, -x. -f, *|P -s. -^).4 Three of
these fricatives, /, a, occur also in initial and medial positions,
where they arc in parallel disiribiHion and qualify as scpaLfntc
phonemntic unks^ which consequently find their place amongst the
other 'letters of the alphabet1, -x and -t$*« however as nlso -li, are
"(ajyo^naha (cf. 0,42 above), i.e. they are bound 10 final position,
and arc in complementary distribution (-X before velars* before
labials, and *h inpuiisu). This alternation h understandable enough,
and it is impossible to agree with Whitney's evaluation of -x
mud ;
tt muy bi' burly quirsttoDcd. perhaps, whcihtr tbrtc two snundi sur not
pure fsmnim-iucki] rtbaTtficcioini. ’
Since these variants me nor mdutlcd in the alphabet, special

names are devised far them, vhe, visorjaniya (or later visargti) for -h,
jihrfimutlya for -X, and npadhmanlya for The last two terms
provide ms difficulty; jihvwnuhyon lit. "formed at the root of the
tongue", is the general term for 'velar'* and upadimaniya meanji
literally 'blowing upon"—the consecrated description of the voice¬
less bilabial fricativ e.4 The term for *h is not so readily explicable
—a fact which is reflected in Monier-WilliamaTs dictionary':
fS Isa called Yltarjdiilya either from its liability eo be "rejected' or from
it* brine pneoiounect! with ji full 'eimssim/ of linsaih, or from ste usually
appûnnc at the end’ -it a word of mittnce.
I he verb From which the word it derived (w-jr/-) has meaning* af

the type translatable by4to discharge, relics, cast olF. We shall
1 Cf. c.lt. ghannah 3*taido Gk_ antah hrmidc L-»c inter
* AP H n Ufi^-T n iytuyi* pt-WV - r.v i ?#: .,tf .
HP 1\. 51 ■ j dghw* . , , £0:r.'wlit2im ipuj±<2 uM&j full!hfimiHi , fpfm
mi intern ti j jjMim
NT 17T* tlAtfnaw (ConarL Cjtna m par^taiihdnum Ap^Jyuu \
TP Lt. 1 ilgiwj a*pj:rat tuiya Jttti&ntli wnuititm.
l*F iii. is-u.
J Ski, Gr.. | 69, Cf. on AP iL 40, 'Hie Jivticno of U11* tmlhgfiiict *ml In-
drflnrlc wimd mlc ihrn kind a of pvgn i:»Bj;|y of oYe: itfsne-
ment of
For other GCepto #« nnEc by Fry, up- cft,p p. 194.
* c if. D JlW»> «jl. eii.* f 6&J. 'One form at £ b thr MUtruJ nude \n bkfwirjt
uut a catMflt+ QL Sppsr. LriL ^7 ff.
LETTERS 52
perhaps be giving ihe mast direct and phonetically appropriate
translation If we render st by fcoff-glide*. as referring to the breathy
transition from the vowel to silence.
In later, though still ancient, time* there appears to have been
a tendency for -Is to extend its usage tn contexts other than m
pmnaw The earliest of thcê extensions \\a$ 10 the position before
the initial fricatives f-, g** where it replaced the horn organic
final -j. -g, -s fin(1 raj jmirah >indrah fuurali, Sec)-1 Hiis
practice was then extended to the position before the velar and
tabid voiceless stops: in connexion with this innovation we find
mentioned the names of Agnive£yat VllmJkh £akalya* and the
MSdhyandma school, whilst the ancient grammarian Saknbiyami2
ts quoted as holding 10 the more conservative practice.1
These change* have been generally accepted so fur as the writing
of Sanskrit is concerned, and A. H- Fry in his article \4 Phonemic
Imcrpretafinn of Vi&srga’* has suggested that the spread of -h was
due to the writers of Classical Sanskrit 'operating with a phonemic
orthography1. ITiough the term 'orthography' once again begs the
vexed question of writing, it is possible that this extension had a
phonological rather than a phonetic bnsis; bin in this matter we
are faced with similar uncertainties to those which enshroud the
extension of urtufrara at the expense of the homorgamc nasals.
2.01 * Velar
lltc vdar scries (k-iwjfl) is most generally described .is being
produced at the phvfi-miiki, Voot of die tongue*, which ka iirictly
speaking, an articuktcir and not a pker of Urticutation : the tthatra
of this series is in fact the htmn-mula* or ‘root of the (upper) jaw'-—tl
tnthcr inadequate though intelligible designation of the eofi pakic—
In the A-serirt contact ia muik by ihe mol of the tongue &t the rooi of
the jaw.*
Amongst the velar consonants is ako mentioned the velar fricativ e
x (jikv&mullya)* to which reference has already been made.
1 C£ TP h-5 ; VP lib 10; Pd». m in.
1 Noi 10 be eonfuBfti with ihr lypiinsHmixy tothor of rhr - Tytt*
WiJStL
* Cf TP ut. 4: VP Hi. 1 i-ia. * Lorîogr, T&fh n>4 iE
CC 'IP L. io. ji Ari f Jim j-j P. 'jt hi ftp
TP n. j j* Jjjj nu-r?w!{ irV iU^W.,1 ipMutyali. Of- TP t flj.
: eft. Vp i. 65; ftp j, 41; H T 4 For the fttntraE icnr* Vetai* ihe Ap S. ujki
'IujjJUiT nf ih#i ihhjiI}Tb--Jmutiy4-
7855
$* PHONETICS IN ANCIENT INDIA
The Indian term 'nwt of the tongue* ha# found favour with a
number of Western phoneticians, amongst them Sweet' and Pike-"
It has to be mentioned tJun in the later Pamneau scheme, as
reflected, for example, in the SiJilhdntei-Knun:mil. the pulmonic
breathings (alias ‘glottal fricatives’) are '.lasted with the k-series,J
the whole group being referred to as 'glottal' {Amffiya)* The
unequivocal name of the friCstive/AtAidZ&ga preserves it from this
confusion.*
3,02, Palatal
No difficulty is provided by the descriptions of the c-serica,
which at the period described by our treatises appear still to have
been true palatal plosives rather than prepulmf affricates such a#
are general in modem Indian pronunciation*.* They are described
w» being articulated 'at the palate (Irf/w)':7 mote specifically,
In the C-itnci cntiim it miiit with the middle *>f the tungut upon the
palate,4
2.03. Retroflex
We haw seen thst the refroflejt series involves a special process
rather than a place of articulation. Since, however the Indian
scheme treats this series its parallel to the other tuzrgas, and next
in order after the palatals, it will he appropriate to consider it at
this point. Though the term for TetrailcxctT {fralkrffita) is well
attested in the descriptions given by our treatise*, the genera! term
for the retroflex series, employed by both grammarians and
phoneticians. \i. m&rdhanvtt, an adjective derived from murdixm.
'head'—e.g.
for the iwirtthiwyai thr urticuLtur b (he tip of the tongue r*iioitc*ed ;*
In tht t,'icTies contact i» mad* with the tip of the tongue rolltd buck
in the mUrJjMih"'
1 iff 1 Phnxttif}. ftp 1*0 f.

1 SK il1 * * 4. a*ku*hCl Ap. ,*?, jr ■*,
‘ In thi 1 imprri-nr ui^c to infkyn it perfcap* b«i rendered by the equally
impmrtif 'tfutmrjl*.
1 SK 3 5, jrhvtimttfTyatpiM riJ,rr\i-rniiiVrJn-
4 StTstlî Grirnrî, iflij, p|i. }9> fi
T e.a, ITi oc: RT 5.
1 TP it, ji. fifnTti riAl iJ"lniTî^rt?flil r J.rrTPjsTi‘. Cf AP l mm VPkn.
* AF l.TA. rtnlidKm^Hdrii praHviftiltimr Cl, VP t. 7^.
"* TP il- J7- likrdgrr$iT prultt fjfya nmrdkaM /a-t-vcir&tm
LETTERS SJ
Commcrning on the latter statement, the Tribhdsyarutnu says.

By tht word fiiftrdlum j* mcunr thu icpper pun uf the buccal cavity ;■
But there is no evidence that the word ever used in this special
sense, and comparisons by modem commematom with Greek
pTjpfji'dr- (lii. '{vi«k of) heaven1* thence applied to "roof of Lhe
mouth*) are hardly relevant, Murdhan means simply 'head* or
"summitV and the Indian terminology is reflected in the still tint
entirely obsolete term? ‘cerebral1 and 'cacuminal1/ The term b in
fact unusually imprecise* and Whitney is probably right In sug¬
gesting that it represents u traditional title surviving from a period
when phonetic science was 1cm well developed* (cL also the term
fiwitm for the fricatives.- t.tti above). From the historical stand-
point the retroflex sounds arc relative hammers into [into-Aryan
and they consequently occupy a peculiar place in the phonological
system / they are thus likely to have attracted attention even at a
period when specialist phonetic analysis was unknown K and the
terminology, like tlmt of Latin in the west, is likely to have per-
stated into a period of mure precise description.
In contusion with the role of the tongue in the retroflex series,
the Ap* makes lhe remarkably acute observation that the contact
is made not with the tip but "with the pan next to the tip, or the
underside of the tip* *
Functioning phonological ly us a member of the retroflex series
we have also rhe semivowel r: on the phonetic value of this letter,
however, widely diverse accounts arc given/ ultimately depending
perhaps on dialectal vartatioiL Hie same applies to the vowel f,
which will therefore be most conveniently considered in connexion
with the semivowel*
The retroflex pronunciation uf both semivowel ami vowel i* in
1 trtAfra<tta*vpim-&kd£v tjrtakât^
; Cf Max MiiUcf cut RP 44
* Tha PS (ijK 3 In-1 of rhe rfkfJTiii, itw uses she Kim iirm hi p!«u of
ifttiftftafl,
* <X ftp, of,, p, is^.
1 Oo At*La, TP\L17-
1 Cf. sho Jukubson. Ptot, 3rd Ini. Phan. St.* p. 40.
h, 6~7> jitrtvfdfrrpei wrirdhjtmytfndM: jihu^ddhahîiriî^ rtfk Cf, | U.
1H ittli |fi f Ijii3 ry, up. qLt p. in: ' The LidLin l is jtol nude îlh ihu tip iq tllc
sols manner, h»r with fhfi vtffy edge <ir rim US flir Op, which in JtllghlLy
coiled tudt !«> makr thi* pzrtiibkV See *l«o xia,
CL VnrmUfc up. dU, pp. G EL
S4 PHONETICS IN ANCIENT INDIA
fact prescribed by the P&1 but is exceptional elsewhere.- The

Pmiiakhyas generally require an, alveolar articulationj (which
agrees with the present pronunciation of Sanskrit and the general
practice of the modem I rub-Ary an languages)^
The AP, VP, and RT refer to the alveolar position by the ierm
dtifttn-f?[uh, lroor(s) of the iedh\* a name which has been employed
liIsd by Sweet.1 A slight difficulty \u caused by the fact that some

of the treaties refer to the dental aerie* by this same term; in andi
cases, however! the (reference is to the junction of the teeth with
the gums (Sweeps 'rim')/ and the alveolar position of r is then
clearly distinguished by 3 further description, eTg.
Fur r contact ts made by the centre of the tongue-tip behind l priaiyakh
the mots of the teeth/
the word prtifyuh being further interpreted by the TrBh^tymmtm

an mcaninu 'within and above1/ Certain authorities quoted by the
RPd1» refer to r as lnarteya\* a ha pax glossed by Uvrsfa a.- denot¬
ing 'the project ion behind the roots of the teeth1/5 i.e, the alveolar
arch11
Thr prescription of alveolar articulation corresponds well with
the name tcpfta interpreted as tearing sound' (see 0.4z above), in
that 3 rolled r such as this seems to imply could hardly be retro¬
flex /- excessive rolling, however, is listed as a fault by die RBJ and
U vata refers to this type of pronunciation as ‘indelicate/11 Two
treatbes, the RP and the R'l\ treat r m dental, but nuns ion the
alveolar pronunciation ^ an alternative.15
1 t-. UW rnttedhattyd f-ftt-FG'T&b
Ap. S. i. 13. r*$L-tQ-id muTdhimv&h, but to Jattija-wruiîi^frititH /JLrfJm.
1 Cf APi aH- JPh. 41; FFr 6&
* AP L 1$, fr/iitrja dwtt&‘m£tem; VP i.(jXr K7 &.
1 fVrmr*, p.
■ Ibid. (irf*wf# imnmolfjfy tlL*fmi(uithcn ihh 'rim' train thr
TPM 41 rrphfjikLâ-mtidhyrntipraiyijgdtftttot-muJcbliyuh.
* jrmtyctf it} rtbkytnilara upon -AAf^<] rJ> tirihith.
' i. 4li. lartiyiim ,h*.
I ‘ r rfiirtia*m£i4it epfmjfurf usThtmafi prûtfSa Utyair,
M If h if^cLul ttnw b requuml E*> twniibu? migftt be apf res-
prut*—rf hhe. ap* citlH p tas; \ nlvcQtar urch (which mi^ht with more
!■ t-alfc.J rltr tftribE-ilîtltntx the cWW* n m*dt fltfatmt Oi- gum, not
(hr Klt/lt . f l BlffwiQy, p.
II Ell I? fora Elk-. n^nsr-r nf th* rcttoJln T t* ckorly imJit-ilnJ by iu {’ijuivnJ fskc
fa 1^1 in list umdld 4iarvdif - ^Ltr^nih niirvMir |>n^:MhL m
ri Ih- z(i (f;i|pOTH? ffl Tfphf
U M«lr
H ff/‘ I. 44^4* ■ WT 7^1 (diailf/ U^j; mnl/t rJ),.
LETTERS 5$
'Hie- disagreements on the pronunciation of r arc duly noted by
Uvapi;
Sfrniv v:hito]n pionculficc r ,i *crrrbnd\ Indus m nn nlvroJarJ
As regards the vowd rf an alveolar prtmutitEJtEon is suggested

by the TPm a passage which reads:
Inf umi it,t, tJic lip of i he torque i* appratimM^d to the *fram,0j,.a
The TrtlrfMsyimdnti interprets the *6&rtms' a* referring to "the
elevations behind the row of teethV which h rcrninkceiii of ira
comment on the semivowel r (see above). Other tranucst however*
agree in allotting f to tiuc velar class:* this prescription is prob¬
lematic* and ts applied by at least one author also to (<- The latter
appears only in the single root k]p-, and ir has been suggested that
in such a phonetic contest } is likely t« have been articulated with
“dark' resonance* and that it is this w hich ha* caused it ti> be classed
as- velar® (cf. on consonantal L 2,04 below). But no such arguments
apply to i\ and indeed the Middle Indian development# point
rather i** a palatal refinance for both voweb.7 h is in any ease
difficult to believe that the Inclines would have classified these
sounds by iheir secondary rather than their primary articulations,
It h just possible that m connexion with f v^- should interpret
jtkrdmfihya as ‘uvular1 rather than "velar1; 11 ri only Grange that
we have no such description of the semivowel r* except in so far
as it if mentioned amongst a list of alternatives by the Vtmta-
pafftlmit*
In the retroflex series :herc renoins only a peculiarity connected
1 On RP i. J taiyurn t reptro m&dhmyttk koxytfqi itt.

1 it. \8. jihtu/grrxm /kitfiirk&rnihdt11 n fhif jr l-jKjpki>TjffthtiralL
1 kersrfjf txi dotita-fatikUT vpanjfcfri Uina-prmlnffr tty nr1ha$r
1 Cf- VP i bz. rfr&iw jikvd-mtilr; AT 4- }iht’tf'rTt\lU [tkp; RP L
RP E. 41 fh4Jirik<lr&l‘ a:ha f^h*i mwUipin dMsdliydfj pratkiiitm3 m 1 w&ih
Whitney <m AP 1, bo.
f hAitta < hipifft kisa < krw, mjtm C jrp The Prakrit srmmmnjriHxiB
preterit* 4 «iflvtl^¥pmcrrt to ah bul ihn iM lif^dy fl MrWiftrl peCiil larify k .g,
ffmit fikpiaL cf, Hemflcandm i. lift, am! FitehcL $$ 4?JI. )< Th# det*Snpiti«LL
"■o i4 1J11 * $ M, ?rrthua •' w&Ay/ji l» Eenc-ri-Ily uiddami with h tub ml cuntrst,
There Li, howrvcrT cutn*j4*n.bte ihicnurtica* uml Ubchr jt.r
Kmc* ^ Lj aa to roiurii, 't-a calqrttkm Jc Is VmUc o! impîiibsc * For
^1171 w« I'wve mi A Mg, f ot u 11'iiWh^L ^ 5ft}.
* ii. i.
A^mj.rrai^ Ui rrphilf 1 n^J ^nT^Rl^^y tvJ pirJirtA

1 J daHta-miiffbhp* murdHonyii iti r4pmr.
with the fricative g. In the FtaitjtiS Suiru we find the state¬
ment.
Alone or combined wnh ron-sonint* (other than rcmrtkn), 5 is to be
fimnfbunccd o> kh,‘
a statement which is repeated in a number of the later Siksis,1

This pronunciation nf 5 is common in Sanskrit loan-words in some
of the modern languages (e.g. Hindi doth beside Ski. do@a), and
b traditional in Ynjtmcdic recitation.' There is, however, no
mention of such a pronunciation in the earlier ireatisea^
3.04. Dftttai
T he dentals are unequivocally described as being produced 'at
tlie teeth ** or 'at the rim of the teeth'.* The VP and AP go on to
specify the tip of the tongue as articulator/ the latter work further
stating that the tip is lpraitirna\9 a word which is most naturally
tmtiJiLucd by ‘spread, flat’ {not. as Whitney, Thrust forward1)- a
characteristic feature of denial articulation." The KP, however,
disapprovesof«toessrve tongue-spreading in any uf the stop series.1'1
1 is (generally treated as the semivowel associated with the dental
class, and most nf the trcauses prescribe a dental realization. ’Hie
HP includes in its hsi of faults an articulation 'with both ends of
the tongue* ;n on this rather a (range statement L’vafa is not i Do¬
minating; and although the word tfw/oinay also mean 'side*,” Max
Muller's ‘mit den Seiten der Zunge' is phonetically unacceptable
unless the reference b to opening and not contact1* (Le. bilateral as
1 iK Jfho mSrdhat>yoirtul& tomyvluatya turn rK J.wruiitoVfl ™ fthnkfctc-
tJLPrS^nj,
1 ( f. Vimia, op. Cit p pp 34 f
7 fjftcn».»rs, ItidtÂry*th p jS.
' Yhett El xEiaw evidence in ih* wort of Mh*nm\ i ink wm, *ro,l fne ■ pr^
mitttiiMjn n a vcbr tnmXfYx \%h S*c Kmi. 11 ndo. Aiabttcht Sfijtlim' {Phi. V,
Hat. Ahh af. AMsl Ak*J J. Wm. =. Btrlitt, ittS}, pp i it. flnd AihrunTt
'ni SaOmUt »- JS* ^ il im); ter further S. K_ ChnitetjL Al-binmi
1 W,p fi, yl p Indrrtfi Ufigaiilifi. Fit, JJ. pr
1 * HP L .44; TP ii 38.
■ VP i 76. .^P t, 34
* jlftrJjfriHp praitFrmitn
f i i y H Firtk 1 ft CifnEiiini^Bailcy, TVnrJt FsKiTirfr Iftntjm taM* p. rvtli,
- d ■* produces] 1* rtli * tbt tanjfue Mprx*il mu
jay j 1 1 argau jifatirprathunaw Htfurpi (UwU: jihnludk prrfAdtlrtfr ftrintd
WliffuAL
57. ra itorfflurm ftjfafrr,
* Cf TP it, *&.
'1 Cf t3rc ticKTipciftn of (_-&j aj. ffJhîrAaio you i
LETTERS 37
jigdnsi unilateral articulation)* The RP*$ disapproval is perhaps
directed against the secondary back-rabing mentioned above,
ivhich would produce velar resonance or "dark r it]. the rypical
Indian I being nf dear quality.
The vowel ] is also generally said to be dental; a statement pre¬
scribing velar articulation has already been referred Loh and an
alveolar artieulatton is required by the TPJ
, . Labial
2 05
'The labial stops and the itpadhmnniya fricative are described as
being produced with the two lip-.- or simply m "at ihe lip1,1 he, at
the upper lip as ythaua. For the semivowel v, which is regularly
associated with the labial senes, and which has a close phonological
relationship with the lip-rounded vowel u, further observations
an? necessary. For whilst its earlier pronunciation was doubtless
as a bilabial [w], it had by the time of our treatises acquired, at
least in some dialects, the labio-de ratal articulation [uj which h
typical of many modem In do-Aryan languages—The learned1,
aays the p£t 1 declarev to be tabm-dcntal.1* For [v\ only the middle
of the lower lip is in contact with ihc teeth, and the observation
quoted by the Ap £p that it h formed "at the comers of the
mouth1.5 is hence noi inappropriate.*
3,1, Vozeels
z.io, a.
In the discussions of the vowel-system considerable interest
attaches to the firet kircrof die varna-samSmnifya, a. Mention has
already been made of die tendency to group the long and short
vowek into pairs designated by a single term, 1 and it* for example,
being referred to as r-t/rtma, 4-colour or quality". In the case of
S/ll and u/uu it would appear from the phonetician*’ accounts that
there was no great divergence of quality a* between the short and
the long vowels.7 Phonologicaily parallel to the pairs i/ll and u/uu
4 it. tS, J 7'P EL 39. i\tf }vih?i\um
: FPi, “a_ u^tvk-p4 wthr; RT^
|R. , . Jantyrttfhx# imftd bttdhtkA. Cf. TP iL 43; VP i. Bi.
1 i 1 ^ - i 7. t +T h i r a dint i frffhy Clh if JlTi j ■ f fjfcff nn m tkf.
Cf, D. J one v Ovf/jHf. I <39j: 'ItiihjjH. |{mt|y rtpliact v b;- a frictibn!^

etmhtiqant r m which the lower Up rouchei ihe ctntie front imh ffcfchrly and n
m> hdj iu to jlluw ihv air to crape chwfly at th* atcU*1 {with Fiu 0&).
*8 FHONtrriCS IN ANCIENT INDIA
h the palra/aa ; hut it h trident fram the ancient descriptions that

ihr member of this pair differed considerably not only in length
[kfilct-hfimna) hut also in their quality or 'degree of openness4
(riv&ra-bhirma}—in the modem languages in fact the distinction is
more generally maintained h> the qualitative than by the quantita¬
tive difference.1 @3 is generally recognized by our texts as the most
open (mfta) vowel, whilst o is referred to as relatively closer
{jamv/ta)* But the advantages of treating a and no m members of
a qualitative pair arc, as Fanini realised, considerable: on this treat¬
ment depend*, for example, tin* rule that.
When - '-hurt vowel n followed by * similar vowel the cunx* pending

long vowd hi auhadtuted for them,1
so that parallel to junctions of the type yadi-f-1 cellar = yadiiccbet

and snadtiu i-uktam sand hunk mm the nile h capable of
embracing, e.g, mi-nsti = uaasti, £c. Then, having based hh
phonological statements on the fiction of identical quality. Paninj
cancels out the error by the phonetic admission of his famous Iasi
aphorism, "a a\ i*c,
Thr bF which hu* fî phoitil ugical purposes been treated m Eif
identical qualify with fb* h*ng aa, is phonetically closer.4
Ill is si m pic chough ingenious device of Pan inf* has given rise
to a spate of involved and for the moat part unedifying controversy;
a lengthy discussion is found at the beginning of the Mih.t but
perhaps all that really need be mid has been said in the vwttika
with winch Lite discussion opens—
1 he trf:3irsiEfni t/f a ±s open hsa 2Jt iti purpose the iftcJiman nf H?4
{sc, 33 (hr long member of the
Even from the phonetic point of view, however* a and im sue

&tttlscscctilv similar to be considered under a single articulatory
1 See Ji R, Firib. Pit*, imi Juf Gwt> FAuw. p. mL cf &]«hf ett.F

P- J**
e.^. i. 55-î; LJ*i. #1- W (UV*|*: tfAkffttrVa ^7*
mairihitiya tetftha-tihdmttd uklii . . , dMj a prûtttal Du fiLhnU mm-.'} ,
Jmj it-pruwint* t*k4mh t if -pr^yulitif iWr fiarjA),
1 VI. i. ioi umtrrw dirghn^
4 bm Jfy tif i he rm&Liml fm^hl be preserved hv it rcttdmriff tuch

W
1 r i lt M FliIp , .'ll!*. i (Kidhom, L i^). aMrd-
grakanJiiknh.
LETTERS S9
heading, via. aa kanfhyti, "glottal*—3 term uhkli |m already born
used in connexion with tbe voiced and voiceless "brcitliings'—
U ift ^k.ELiil1
j Olid L etr* gfaltftl*
a, h, arid -h arc formed ar the Klmtk1
To class the: open to web as glottal' appear* at fir*t sight m
indefensible procedure It becomes Ices so when we perceive the
conceptual framework underlying these statements, It will be
remembered that the TP referred to a 'neutral1 position of the
aniailatoiy organs, in which
ihc longue m extended and depressed, and xhc lips aut in the portion
for a *
rile clarification of a as glottal begins to make sense if we assume

that it was viewed as a "neutral' vowel in the sense of involving no
special mtra-bticcal articulatory effort** Such :m ussumption is
fully supported by A Statement in the Mahdhhasya:
The place uf imkubiion tv( ihr a-vowrU n rvtr*-biJccil: ^r. a# -.nine
would Jinve it. it k the whole mû(h-fl
In other words a h&d no specific intra-buccal stkana or karuna\ as
with h and -h, h is a ease of (cf p 40, n, 1)
From this recognition wc may proceed 10 the peculiar doctrine
mentioned in/ the RP* in itself inexplicable/ that all die vowels arc
to lie pronounced with the * articulatory condition' (k&tas0smib^
of a.* This statement abo become phonetically meaningful if u
is interpreted as 'vocalic neutrality' or "unmodified voice", on which
are superimposed the vowel-articulations involving various degrees
of tongue-raising.11
1 RP i j.#, "hlraft. 1 P& 17- AfmfJijtft
* IT L 7*. 4 cxiUjiyJ/j JbfliAi. CL fimm cm JF i,
* Sec i. 113..
1 CL Sieve ra, G*\ d LwfjMgrtt&fcgjr, P ^ "Bttm* ilidcr MumksnaJ durull^
gthcmjU mlDic £c*£Fncl; die ZuiLge enlferaf iitfh nkbl vicE m* thrtff llsiliiTer-
mKlflfl-e.1
t L 4. cm pdn. 1. L 9 fKidbcm, L 61 > Agr &n<ic lihfimim mmimiyA:
tvnu~mtikha*ilhiin£im emurnam tka jicfujuU. Cf. Ap, £. L II.
r CL Mai M i^Htr Sij), . i*hr au urm *ch«cwn/
xlv. 65- ailwiji; tofufrarf^d^fffl fruyrtL CL 66,
The atntcmcn? i* erf cm lire nnmenniral rf t* t imply snJrFptetcd
■i nefcrrin4+ tu the ?^ti|;ue ■ jxni Mon i huj tHn u*p of iht Itim .w-.irsfod t.fiGt tihdna*..
* Wwd <Mhfcr*i*t* uiuEtmcd in I he pfroii eitc ineruturr. ti ionic KiuStftlCi of the

ipociid utillTt □! itu? frEemncr Cf M* SViKcwf, 27 / HJJ 3-
6o PHONETICS IN' ANCIENT INDIA
We are now in 3 petition to understand a third problematic

doctrine referred to the RPn via*
fvDniJL’ uy ih;Lt the voice of thL- voiced consontiiifi* consist* of aJ
This last statement enables tis to trace a consistent thread running
through die series of apparently eccentric aphorisms, and to relate
them precisely to Use descriptive framework of the other 'glotteT
anicnhition&:
(a) -h is considered as "pure brauh*, liable to modification by
the close vowcLef* and capable either of independent function
(i= vivaijanh'd} or of providing the appropriate air-stream
lor the vuicebs^ consonant^1
(A) li is considered as “hreatfi-f voice V liable to mr>dificaiion by
the dose vowels,- and capable either of irtdcpendehi fime-
lion or of providing the appropriate air-stream
for the voiced aspirates,1
(e) n u considered as 'pure voice, liable to modification by the
close vowels, and capable either of independent function
(■= Qvarnn) or of providing die appropriate air-stream for
the voiced coitsonams.
Artificial $$ such a descriptive basis may appear* it is in tact not
very remote from some statements of the most recent branch of
phonetic analysis, "acoustic phonetics1! the following may be
quoted for comparison:
We therefore discuss vowel productitm oft llie hypothesis that die
Ifkittii cmiti t ipc^Tum that it independent of tupra^glofial urticulonnn,
«ud that the ftUrcmg which determines fhr ulrirmttc vimtl vpectnitn it
independent of the el octal adjustments tha t b, the origin nl pruduruoii
and the articulatory modification of the glmtnl tone are entirely indepen¬
dent of each otlirr. Tlw ^«tmm of the vowel at it arisf* m the open
oil b to Ur reckoned. thenr «i the glottal spectrum multiplied for *acb
freîitincy hy the tnmunwion percentage of ihr tirticulutory fUter^
Two thousand years and more before the sound-spcctrogiapli.

"a-Bcund- wa* not an unreasonable substitute for the fiction of a
pure spectrum*.11
1 Itiii, 15, rfo. . . ,

fe C| TP ii 47^4It (hoc 3loo *l>Ot'r).
1 Cf. .4P i, u, RP \iis c-t^ rP ee 9--to(Kc t.zo above},
1 Sec l,B ibovtr 1 M. Jimb, Afnuitic Phomtu^ p. j4}
1 Fordjreia: ; 1 nn the izonurpt of u u the Jnulund vowdh or 'princtp* t*’<nthim'
Ît Skvro, PW+ fj igy§ ; cf *J*o Jakolnou in Tmb*t*kay+ jPfrtfr/^. p 376;
LETTERS *1
With regard to a ir remains only to mention that in Jatgrtraatisc*,

owing to the extension of the term kanfhytt {tice ^.01 above), a
(like h and -h) is grouped with the vekr series.1 thu* adding eon^
stdciubly to the symmetry of ihc va^m-umimnuyn at the expense
of phonetic precision.
2.1 L it U
The close front quality ! is appropriately classified as * palatal'
(£filmya)y and the TPsays more specifically,
F(tt l-quirfitv the middle of the tongue a approximated ia the pg]*teJ
The dose back quality n is classified by the lip- rather than the
tongue-posiibtiK via, as Tihiid1 (ofjhya)? the shape of the tips is
variously referred to as 'appro*immed\ Le. rounded,5 or as +1ong\
Le. protruded.*
2.12, ft l
A* to the pronunciation of the vocalic f and the ancient state¬

ments are perhaps not as clear us wc could wish, but rhdr general
trend is easily followed. Thctr places of articulation have already
been discussed above* In distinction from the other vowels they
are referred to as * mixed",7 Lc, combining features of vowel and
UmrLdjJ, JW. Jrdf /ftf. Ctwe Pben. S^t pp, 4, j If r TCLP vi *\z ft In the
A7fthri)p(*r-Luitf$chr\ft a it the r*ro/u imtifFrtnu' (yea M. 1 kepc. £i»ifcdcArfl.
P- <0l
Cf. further Miflftt'l afrtOTltobni mi thfl rule of The pharnijjiit ravoruSor mid ih
timbre—L'ArTfcjrfatiiTti cffjr nayWfci* p j 'Soup ccniidfr<nu h Itoutkip .
CQJttlljp He tjeiiralcur efficHCc, In IÛOIW ti^crmiiuiil du itt r^niiACe
vaoiliquo; h civil* phjirynjjLennt don at ion Timbre a III. vme uulrmnit qui
wrekppi rdui Je U cyclic'-j nml iJV la /ofMHtfew dvr
rovef/tr. p. 6S: *W pat de umbtt do vcyfiljo qui no «nii ud^ptumfe du
wtihrv 4e It voht * Allotting fur ibe Fact Thjat iht Indian* not to have
differentiated pharynx aiw\ larynx* Wt ftiny tflV that (heir canceptiim of ii with_
ifl Milkti tcrminofctiiy* that of a phuryTiffAl Timbre witfciutbnCcd dercrmmatuin.
We m*i' here p.jte ihnl ft uimilai1 device wAx adopted in Imim v-ith rtfritJtwr
10 HitmTi(vh tmuttdra bring Treated U 'pant rutulUy^ fanning I he heists of all
ftaul Mnuuii (IZP xtiL 15. ... tm*m§Ktm i™u*aiiAjjajifn>.
1 ILg. .îddn. Kmrjn, 10 (cf. 2.01 Above) Seme aulhuntirs even save to tt
the title which mrann ip*ciffcm(ljr 'lehr' Tp. i. to).
* PS i-f. t-eu-yn-ifit hiltrtr\ilh CF. Swrel* Primrf, fjj.
1 ii 22. Ittfau fihrn-fruhlhyf'im i-CHrrQ*.
* Cf- PS IJr tt-pll
4 TP it. %4- QtthopMQrphnra uMtHt.
* VS £34- Mtmim prm*Tt*r »tfhm 4i*ghm*
1 rs ii.
lb PHONETICS IN ANCIENT JNOIA
consnnant (r/l);T for this reason wrw writers even refused to

admit shem to the vowd^yttnii- There is general agreement that
their phonetic struct ure is of the type: consonantal element—vocalic
element—consonantal element;
f contains r im also does die hr»t half of iy j . . . ajid the r is in the
middle.1
The ehir^tcmtle of y i* that it 3 ccsqipomuled of four aegnwsm; of
these the first and \mt are vocalic. wMItl the central pair are con^miinlak
viz, panicle^ r.*
In this connexion it is of interest to compare such Avc&tan parallels
as pir.jdu beside Skc. pfthu* for.tp- beside kfp-.- As to the quality
of the vocalic dement, the FP state* that r and 1 consist of r and )
blended into one unit with the vowel a.n
Regarding die method of com inning the vocalic and consonantal
dements wi may note as an example of pitauresqueness rather than
illumination the statements quoted by the commentary on the AF\
which dec! sire that they are connected
like n nail 00 the Unger, or a pearl on a string ora warm m the grass *
2.13* et o; aiP ati

It will be convenient to consider in conjunction the gâ and
tfddhi* vowel* e/o and al/nu, The latter, 32s the transcription sug¬
gests, are diphthongs and are regularly so described: the former
also were historically diphthongal and continued 10 function as
auch for certain phonological purposes (c,g. rijpaaya-fldam =
vijjianyedam), But there arc indications that whereas the phono-
31
* ct AP L 37, ? tam*pnta-rtp}idm 71^™ . . , Mlakdram pmmim.
Ap Sr L lb, tirtfpk* fi'arûft
? Cf- Kaiyjr&t* dh Sfbh. 1. i. 4, on FAn. 1. I q. my* iu lît*rpy$^k£u^3imidJ
tffwrvnr rfbiTi\-{kdriiy^f rn rfrffttfttff fÂ>vM toyw flftatftt&n u/mh,
r RP tin 1 + fiphti tty fkilr* m partvyc1 ufrdhf ptex* . modify* wMrf
AP 1 3&. diryhii-plvtayab purt'A mftaf}.
■ S*rm*4m»vzt*i~Sik$tl id.
rkdrajyn TJOTupnrp hi pAda-cmu^yUm
p&irju Irjti tijfa-y&r iL/cfr rt nrAtmahau
upu ttphmyi* madhyt tu ixjfkrynu tmalutu,
1 t ■ flfm Ptr tiliittiT iqpra, anil nfekril-urj’, ZDMG hi, S15.
'■ iv. 14^- rfphii-Ukawin. ftrorfij/tfc rfd-TOrf™ For
the quality ol the vuwet in Meddle Indian *trve!op.mcn» Stt 1.53 tbow.
1 On AP 1 IT J«r£Ad rubid 4 Ixtshii rftn *™ir mtfy r*f 1711* t#W
rtrfi ra-
* Stt 0,41 dbott.
1 CT. Gk. Aw t aCdd besidt Ski. veda. &c.
BETTERS rt3
logical value of e/o was a+l/u* that pf al/au was once atH-i/u:
this distinction: may he illustrated by junctions of the type
naâre-j-iha = nugara ihu
beside strlyai -yktam = striynu uktnmJ
From the phonetic standpoint c/o are represented ai a atill com¬
paratively eady period by simple Jong voweta biicrmediate in
quality between tut and Ji/uu,
To consider now the ancient descriptions: the term for the djph-
tliongs (including e/o) is yumdhy-ahara, 'compound vowel \ in
contrast to uimnKtikmrah 1 simple vowed', til and an arc designated
respectively ‘glotto-palatal1 and 'glatto-tibia]11—as die rPeays,
I ri a I and tut she fir*t mora ja BtottnJ m\d die second patniskl or labi al
both tlic AP* and the VP* point out, however, that
AiihdUBh diphihon^ are ciunbuiniiutit of vtjwd*, they are treated ai
singFe letters.
As regards t and o. the P$ seems to preserve the tradition of a

diphthongal pronunciation {distinct from that of iii and ati):
In e and o the Îc^tsui, element hea * length of 4-mora nod in aJ and uu
i miira;*
the passage continues with the words 'tayor trkyta-Xitmpam',
literally 'in them there u openness and closeness", which Ghosh
interprets as referring to the fact that, in ni and an, ’their first half
or the tf-element is open and the second half or i* and w-dbncut is
cio>e‘ i hut it is more probable due the words refer to the open aa
which forms the first element of ai/nu and the closer a which
forms the first dement of the narrower diphthongs e/o+7 The
1 ^QDe, however, iJwt tltc tlfcltsd sandhi of hoLti 4 nnd Ail±\fu fo e/fc (c
baatmi-f-tllc^nti?' - tmstl^kgatev
3 P£lS. *-frt ns AjsjJwI - ftf/nvysf o*£tj kMtfhojfhtijnu tfftfUtUr Cf. Tp, & L I l-l 3.
' i 72 ■ jbmfAvd ££re-d sradfrd /Aiii *r mr^ra. Cf.
Rfr xcii, ijS-jp.toirtdhyam wfMy-mkfiptifo' tiku* ek* det-uk&&mtopt tathB"
btu^ntir ntmdhytff uAdttr fdhjm rkdr* utiure^ yujtir ilt iAh^fdyimttfc
* i- *o. wTiJhyrQktfirdm mwff'rstG-i HF&Ins; tJu^Mrpocuid ijtiih.
* rr. 145,
k Ghodi. BwVnWi Lir+*A ttUT tj,
ardha-mAird tu kaitfhy my a rkt^mkArayor
ui^raduûj'!.^ mdird . .
T Far the d-btifictioii of the two van dim of diphthong ane niay compare thti
Npjjdi TatEinir' ryp* lal/ituh where ihfi fim (ikmcnl i* ecntTdmtily ih* more
pttmtjivrnt, suid the ninnwer AJ/3W I [it ]/[aoJ), with cEnacr am! idiUVrly ff?i
phcimment «mrtrn^pDim—eL|. bStallt beside miylfl. FI‘hm ti also wanfl
Uitcr interpretation is supported by a passage in the MahitbhSfya.

wtierr the itnpvjta a of e/o ti specifically contrasted with the more
open -An ofut/au.1
The monophthongs! pronunciation of i*/o seems to be indicated
bv the RP when it says that they arc not, like ui/uu, heard as a
distinct sequence, because of the coalescence (iOBtwfd) of their
parish ,is l)va;a goes on to explain.
One dfK1? noi O^ncrrc where ihe ft ertdi and ihr I nr u begins bfciiue
the two coalesce like milk and water/
a type of cymhinmian which Kaiyyafa, cutiijuuinp: the traditional

simile, contrasts with tlic aiixture of sand and water,*
There h little in the way of detailed descriptions of the numoph-
thonga! articulation of *?/o* The TP. however, mention* that for e
the lips arc more spread and for o more rounded than in the case
of a / and the intermediate degree nf closure for e (between a and
1} is streused by the Tribhaiy/iratria:
In r ihi.1 racing of tb^ mniilk of the- tongue tn^nrdfi the paldtc ii !rs*
than in the «vs? of L froHiig to ihs feci ihm thv fqimrr in tfikcil with a.11
11 will he iwwd that even where the monophthong:!] value of e/o

is phonetically established, the feeling for its phonologic! equiva¬
lence to a-ei/u still prevails, and the isasiM of description h still
pmvided by the simple framework
a
ikUrmannfl of the Kfnftd type with a maftophtfestrat rculkatann ik—f ,g tlhrrt*
'nmdi, niaO'.-, vicry'. btfufe ftabay, 'nit*. whrfcf the fit us] -« mu I me mnrphn-
logically compiirphle,
4
fc Mh , nn. Er L ^(KsrSUvrn, ?■ Sji pndlit limit f\th *t*m 4
(* -, cm, wtjla*
MrJiyr«f’rt' ftmt (*C. OJ>>,
1 tait 4^ ifliifrJ-«s]rix.Trfd'rf Unit* ‘ffikidt-irulL
1 fcn'ur« />urtT r- o tiy mu Hwud'wajwflprdf: mtitrttyoh samjyvb kfirodakecrei
ramri^jjr if*
■ On . iIbh i L 4 IjhtpiApUtiiNtf)-
* fj.13-15.
1 I Mil. 43 mi AWt etiim tfcdr* jbip Ju
fiito MvtiTti if) arfAti/a: faftiJi*' ak&ra-imiriutiBtid tkdtmya
PART 111
PROSODIES
3.0. Definition
In the technique nf letter-abstraction various features of tie larger
units of utterance are left unaccounted for. ft is the reintegration
of these features that forms one of the tasks of synthesis, and it is
to them that the title of ’prosody' is here applied,1 'fire ancient
account* of these prosodic features w ill be considered under the
following headings;
1, Features of junction [intuiki).-
2. Features of syllable-structure.
3.1. Junction
'! he nature of our material makes it convenient to work with the
following sub-divisions:
(а) Word- and morpheme-junction.
(б) Letter-junction.
The treatment of word-junction and morpheme-junction under
the saint heading is justified by the close parallelism of the twn
classes of prosodies in Sanskrit,1 as also hy the stated principles of
our treatises, e.g.
Unless directed to the contrary, one should treat the pans nf a word
u word*,1
Marpholiîcal anniysH mmi ohurre the same rules of finality m apply
to word-italstn.1
JF - “"" 6 ‘
In both (a) and (A) certain of the prosodic features are re la table to
the basic processes considered in Pan t; this is only to he expected
in view of the fact that these processes hat! been arbitrarily seg¬
mented by the leticr-aiiulysis, amt have to be restored in the syn¬
thesis here considered.
1 further J. R Firth, "Soiiudft slfiJ FrojMftiin1. TPS, iu+8r pp. 157 if,
1 Ci VP iii. 3, ptiddnlG-fflridJyoti umthi*!.
1 Vvt diver^riicn. «d WfettUcy, Ski. O., § tpg£ Thumb, H&. d*l
§ ibtf.
H RP j. fn rtftruiydiTr «iv ca posfy4n
* VP *,153 , m'&eruhtib parffirttaodi*
fP
ih PHONETICS tK ANCIENT INDIA
3.10, Word- mid Morpheme-junction (cf. also 0.40)
In the synthesis of the word-isolates certain of the features
affecting initials anti finals are capable of description in terms of
letters: the fact, for example, that die junction-form of lat-i-ca is
taccu, may be anil is described in terms of a 'replacement' of the
isobte-hml -I by -c. Other features are dcseribabte, as we Jiave
just noted, bv reference to an extension or transference of certain
processes such as were designated ‘distinctive features’ of the
various letters; this is mure particularly the case with the extra-
buccal processes—e.g.
Voicing: usiit+ raujaa ■= aasildraajas

Aspiration: [abh—f>ta = iabdha
Nasalization: tank ■ msima = vangmama.
I he ancient methods of stating these features provide little of
interest, and since it is with the statements and nut with die phono-
logy itself that we are primarily concerned, they need not be re¬
peated here; the information is in any case readily available in the
tatidki section of any Sanskrit grammar Macdondl makes [hr
■J&eful classification into Changes of Position* (i.c. junctional pm-
idlest of sthQnu, such a# in tacca above) and 'Changes of Quality'
(i,e. junctional prciHHlies of pmyatna).
One exception is provided by the prosody of retroflexion,1 more
particularly in its application to morpheme-junction. Where the
process of retroflexion is indicated by a rum-tpmia letter (i.e. by
r, y, or g) the process continues within the word until an inter¬
fering articulation is initialed (ase.jp in pra-pubt-ana) or the pro-
ccsf: is ‘signed otT' by a retroflex slop (as e g, in pra-piit^-anu):
if, however, an apical nasal appears in die sequence, it is realized
with retroflexion and so closes the process; thus when to the verbal
rooi rabh- arc .iddctl the morphemes ~ya- and -manna, the
n-juli is rubh-yu-nimnû. The ancient phoneticians were aware
■>f the remarkable nature of tins prosody, and the RP and VP
employ for it the special term 'mtT* lit. 'bending, curvature'.
i he term 'interfering articulation’ referred to above is taken
from the HP’s, discussion of this prosody,* It refers, as Uvaja
* S« further tlSfKIS *m, W. gjgff.

J RP V $1, tt& natir ifnntya-mtitdhvnyii.hhtirah
* F L 4* fantysnye mfr4k*mytfpmtT muifa
f S* *ey*t*iam ni^ntht fttffwktam.
PROSODIES tn
points out, to the three middle tîrgui,1 to, palatal. dental, and the
retroflex series itself, I'hc full appropriateness of the term will
he at once appreciated by any phonetician who has been concerned
with paleography; in investigating by this method words which
illustrate a particular feature of articulation, one has tu restrict
one a examples to those cases where no interference t6 caused hy
other articulation* such as would involve a further wipe-off on the
artificial palate.-
Certain features of word-junction weft found by our author* to
require description in terms oilier than those of processes or of
phunematic units. This is outstandingly the ease where, so far as
the letters arc concerned, a 'hiatus’ (tftyKi) is left between a final
a-vowel and ,m initial vowel Examples nf this tvpc of junction
are—
te+au — ta an
prnhho-f ihl prahbn Nil
pnadnu-j-ucyete — pan dan u cycle
tasmui-f ukgii = tusmaa nljjsJi
sutanb imt- = stitaa iroe
khyuh ■ an = Ichya aa
vidvaati -aftnc - vldvcul ngne
Beside these forms, however, we also find the following alterna¬
tives written:1
te f na ■= tay an
prabtio -f ihl = prabhav ihl
rau-lmiraafinu = taav Indniugrril
(this last ts the tegular treatment before vmveb other than u).
In the above cases historical justification can he found for the y
and y (e/o< *at 'an. Sc. ;cf, 3,13 above}, and the distribution of y and
v is in accordance with the historical facts (thu* we du not find lav
na or probhay Ihl). But in the phonetic treatises we find similar
tfrt madhyammk IpaHu-fafgait t'jvn.v/rfHi.
* Sw iillv J. tt. Firth, 'Word- Palatupmmi ami Artiatlaiion1. B$OAS
Oil. Sj; ff, (ji. 85a: 'For the itudy «f imoilaUQIU to thi* way the •election of
UrtcnuiCEi r* lieftirniOtii by knowledge that lame inirulilioiu, bi-llbhli, foi
flam pie. give no pulatagrun, and utbttn Immoigstuc « mm-interfenng
ITtlCUbl iom1),
rVha imt ihn yw *% ngifimt f, fummoni di n^n-im^tfctinF« bicUentilljr ■
furrhrr Argument jpainir the Iwlwt doctrine of .1 gmilCl1 dtf^ttr Of rnlilnri
ihia wmiviiwtl thffl frtf 1 he E'ncamn; I, howrr, like *, function i *> an ihlcr-
i’rrinK ion (cf_ remarki on pp. 17-iti ahov^t.
1 U. WaJccTiiwBfL +*fi. C?r. ir f$ 373 ff.
™ I'no NET ICS IN ANCIENT INDIA
prescriptions for incise canes where no such hbiiHical evidence can
be adduced. Thus with reference in examples of tlic type stimuli
-fimB and khyuh ■ aa above, we find the statement that
-h Ihtfort mt initial mucI -y,*
BURfitsrine junction forms sutaav Ime, khyay aa, See.: and with
regard to the type vldvuan-afcoo.
-turn he fun- <tn iniriuj vowel -
i iijiiL-stiriy vidviuiv ugne, Sic.
Hut whether written or unwritten, mid whether historically
justifiable or not, tiie quality of this 'final' y or V was observed to
differ from that nf the phoncmatic letters y and V in initisij ant)
medial position. As the PraH/na Sutra says,
\ hm ducr fcaiijatifnLi, vii. Wrvuc in inithl jxwitkm, medium in
medial position, iiisd Im in final position
Tht liix Ankubiian of these sounds is mentioned by Pan ini m a

statement ascribed to Sifcatayana, where they arc described as
4tyku*praytitnatar*i\ i.il *havmg a lighter :micuiaiury prrjwas'*—a
Htsfenient interpreted a*» fedhim by the Siddhttnla-Kmimudi:
A light anitrulufirw] it wo in which dim is si rthecathh of the lip, the
rim. the middle, or ihc root of ihe twiguc.*
T TP U- m- iithii I- Cf, APu. 41. nary >a&4hi£. Thb

11 m foci ii rsiihrt nLEM h iding wty of ttarinc the mMiejt, and «mlhr oMtquitin
itill mar (he piwfltantm af WqrfMmdhl m modern firntmnan Tht farm*
ivith furml -h art ahitilute Hi mb only, and elucwhcfc the sunini^pmui fur the
•jLiiiSht :*■• •&«. ] U«- JiiidE *-l fnlloih a the f-risprsil landln lav.m
before vutired initial*—[I) **aa-i:> *-ju £_* py >
t.u: --••■(.• an-I fncljon hdn*t -FiuiLEaLJy :jnvjii]jalihtc in (cf, \>, 44,
n. a}, thii lyllablc m Tnurntjine%l whtit ricttîuy by the m 1 «f y nr W
pTOtodj? m îini.-h m Oj provide ft Yttcdir (£§010*0 rnftf&t&njfBg the hctil~v cjtuiitkv
■ the ■ 1 liable r-lLi &- g- - *« fc'J, md m (nt i. njTjdt of v©wd -
iCpamaon (*■■«*♦* ■ -aOT-y. Iliv h grnendked before rn:ii„ni,iiit i
And ih-J. y-i-nitwJy h*1"re rawrlu* though in fhr tVitim-r ciswr *Ljfm o: a y-dontK
are in. ‘fcwmlRtil u.uil ni LnitMui puditiuM ff .g siwrc ituhlioui
■fcHUOiH rdllLl^, pedllll *«ZLJ-uh; d 4^0 I hr Md^dliJ d^bp-
mfllt of -ah -ei In the t>£ a find rdfnflel fricative, the ptwxfy i. abn
lit ntArilcK type I *4\ -ir, 3CC.), and » mdiEratcd by the i^T0b&I for the
plurnemmic mmfha. ûntUTiumt r(cf, i.oj ±hv\*)—e,g *mitnu^ f nuuma >
msliLOrn^LUivj, 4 Jvu > ^^pirtvsi,
1 f ni. c 41 Jktifvfitiiih* y tdtfro?m
17- frntddi-nvi*Sfjy&itir*nhuy* tri+viifJMm gum-
maM}rnmi4i#hti*ufitfbhtr wtfbvQam* CL 1.1 ?a above.
* V[EL iiL 1 ,8 iqAir jm* 1 ahtainrofi Sdiuifdy&mQW.
ymyXviriTyr p^twop^^matfky^m^lhtdm imthifaim jdyxlf id
PROSODIES 65
Silk ut Ilya 11 j is elsewhere quoted as referring m this articulation m

'a reduced retlmtion or slight coo tact1.1 Various other views are
reported by the TP* which, having given the orthodox doctrine of
hiatus. ascribes to Ukhya the full art [rubiron of y/v* to Vatsapra
reduced realist ion, to Sainkfta the realization of v but not y, and
to MicakJya the Ins* pf y/v when followed by ti or o, The RT
refers 10 these Attire? as 1 half-el id. d v and v. whilst the RP giver,
to the v the peculiar title of *bhugna\ lit. 'hent’-*
What imr irettbes arc in feet describing is not tlu p hone-mat ic
y/v but the y/v junctiun*prosodies—a^ Uvula rightly says of the
'bhugna v, it is simply a mode of junction* âmJiwna)\k for, as he
dwwhcrcremarks of the y-prosody, "if it were not there, how could
the coalescence of the two vowda be avoided rlft The weak articub-
tion of these prosodic markers is closely paodlelvd bv the va-fruti,
■y-■sound"-' or ya~traf y-mW which senes ^ a syllable-divider
in Prakrit, and by the y/v of, fur example, the Hindi verbal forms
ayaf ovo (mfin. a-na; beridc e>g. bay^ha. boy [he* infin,
bay^h -tia).*
3 n* Init talify and Finality

Closely related to features of junction arc features of tniUalitv
and finality in the brenth-grpup. On the negative side, the RP
lists those artiailaiEQEiK which are excluded from the initial or final
position and so indicate tum-jmtUliry or non-finality.411 Bin posi¬
tive ^mcmenis arc few.
Whitney, in 4 criticism of Lepsjus, wrote*
We H4V not aware dun die Hindu gmriniJtrittnq thi'Juvdvi-i—acute 4nd
/IP if- i4- Ititf'-fltir iiilkiafvtriiiMi SdkBUfyamsjyiL

1 ' 31> i 31 htpyti* tT m'UFtrQrptiivau nukhtfity* vnk&f*is tu
ntMfA/liinu, tvpywlf mdteklyiiyu? fri-i t ditmpfttt}\rituynh.
rfii «fJiiwffj e4{i^: * ^k4m-jyckîityi>r tumfMntr)
" If It- t'tthyd-yvwyit* ifrAiu^r* *nf*rfhyt
kkujpuim ratios umdhdmeif\ hbcboiL
* ^ VP iii, yruft fel' r.-rtav/LK f-T,V,rJT|WJfti-tdryaifi r.j kuryrit kathiW* ihu
mwfras'eJi uiftJjiit m tytil {ft f, cl dirrc ait j fc^ initmictr? wkip ihr pro*- -.h ;
nmaitî, with pftultant coateseciKC. in thr fiV thsa n pardculiriy she ca#e with
1 he runouts *11 iMil ibf u tu, &e s,. iht RP rcientimu this pranuhritticn .=1 a
taiUr (uv. îo), Even when not indiCHtrci phonernaliiîlEy, the ifltfir-Wtifrf ipacc
or ihc ns3n-irâJeto;nw qf mdjtrrat twA TUIy be CfimMcted «- a murk tff ihe
imacciaQ-piHd%.
1 Hsmiicansirstp %, rile.
Cafija, Prakrlulakfami, Hi JJ; ct- TH&Adjj $aratPtE { MS, Bodf.)nn IP Cl. ifr.
* Cf. Filth; TPS¥ T948, p, 1+5. I# xii. r-*.
7° PHONETICS IN ANCIENT INDIA
hiurspEiitifti* bi they were in catching ,rnut noting the finest thaeJeA of

•fjiind. anti much, au they would hjjve been debchtoi with. mid. tnudc die
most of* jtasT sudi 3 nicety w ihk1—ever took my nodee of a smooth
brcAEhmg/
By 'smooth braihing* Whitney meant the gEuttal oneot of an
initial vowel, but this statement h prot?ahly nni entirely accurate,
for in the f\5 we in fact find the isolated and notable observation.
At the beginning of YoWela which □ tt not preceded by 1 conacnuml. thc
glumj n amd 10 be the place of omen! alien/
It seem* HOT unreasonable to imerpm these words as referring to
1 lit? particular feature of initklity in question
There arc also some remarks of interest in connexion with final
consonant*. The general rule is that of the non-rma] sparim only
the voicetess unaspirated slops may occur finally m pcfusa'J but on
this matter there seems to have been m$nt divergence of opinion.1
the HP quotes âkataynna for the orthodox view/ but attributes
to Gargya the doctrine that stops in Lhis position were voiced,* and
Plnini permits either pronunciation.7 Tha truth of the mailer may
well lie in the view somewhat abstrusely stated by the AP:
Words ending (sc, according 10 orthodox doctrine} in voiceless stop?,
ctid jKcafdmg to Smm&ka in voiced stupa, hut they arc nw to treated
{%er phoin>toj£ kally); they have in t eontact-*
This observation is perhaps to be interpreted as indicating a
realization as voiceless stops, but With ihc lax atiicuktion charac¬
teristic of the voiced Mops (i.e. as g. ^ ^ ty). G raphkaily They are
always represented by the voiceless series. For further observa¬
tion* on final consonants see below under 3,120 (Qbhmui/mtta)r
3,12. LeUer-pmctim
We arc here concerned with those features of transition from
letter to letter which may be considered without regard to the
question whether a word* or morphenieûnction is also involved.
1 JAQS MS
4 jSs (Uidcrs. p, 13). flf 'yufyiMa-nwr&mw ta dtl*u kanjtm iiinhik Cl
Vitmfi, or- tit., pp 1S5 i
* k R. ;[P t 4. tpayitifr pf*tU*MHttanidh
* Jp ment tirm-E ■imilir djjuiLfnmcm*. TcgArdinft Lbfcmniun find*. ui 10
be found io dsc auraumi of Sctrtwclttr (&., j>. *7?, Kurvdat (O , $ i7S>f
LeduQfL-wtwA, f 30)1 ' i, 16. jOvufAnifi|i jSskaiayuiu.k!
i ij. ... iftiyitm G4r%?itH tfmiam. 1 vi|l, iv. jo. itfvannv.
* i S-9. ;iratW«MM4w /ffryaaJJuffi' j$uiirufrijit-<i tut rfttifi:
•r.iifdipanafft fia .
PROSODIES 71
'Hiese are principally related to various types of eonsantmt-group.
The general term for a gTQiip h uxmyoga nr ‘conjunclionV but the
term piVu/tf, Hbdb block\ is atao used in liter treatises, llie various
types of pirtilii are classified according to their composition by the
i ujiun alhya-îksd under fanciful and largely irrelevant tides such
as "iron*, 'flame1, 'wool1, Stc,1 I t ta of some interest, however, to
toite that both the Y$ and the VurndmtnapmdiptjM Sifoa describe
the combination stop-5-semivowel as a 'block of wood1 (it. which
can easily he broken) on account of the ‘laxity of their junction'
{ihithfiHhrmikft);* in the Greek and lattin grammarian* we find
groups: of this type compared to a “mused yoke* (c.g. ass and oxp nr
ass and mutt), the tractive effort of which is less than that of two
simitar yoke-mates.1 In the Western examples the simile refens to
tilt fact that such a group optionally fails to 'make position’ in
verAc: nn such option generally prevails, however, in lilt cast of
Ye die or Classical Sanskrit/ and there is only doubtful evidence
from liuddhtai Sanskrit6 and the statements of late metrical writers
such m the "Prakrit Pihgata\7
3.120. Comonan t + stop (uhhxmdh&im)

One of the most important feature* noted by our treatises goen
by the title of fihkimdh&riat 'close contact*, 'Phis refers to the non-
release of a consonant, more particularly a tttopt when followed by
a stop, and parallels the French term "implosion1.* ITie signifi¬
cance of the term is indicated by the Indian statements, e.g.
AhhimdhJna ti the checking of a ccmaoiumT; making it (ibicurc*
1 Cf jIP i. qH. iryAnjmmifitf Qvytmtfri tvarath tnmf$£ah.
L 2at {3Sr p rtiftir itiflin-'.-idhilh ah* Viirm.
op dL+ V* f*Jr
1 ram. 3, 177-3 (S5, p 133).
Iptnui Upaiktftbl ft rdnim$fldtfni «tj
tf&tti-pipdttrn t€ tuiytih ifathit-t^mdhdh jWdAirOJ#.

€t y3 5toj. , , utnliislhfm
The term fiat ha'batidhd ii Ul»d of virion* other camb iim bo eu by the SmM*iya-
Srkid (73. 7^).
* CJ. %Jjfr. Yictjumuis, Kcilt v*. mi Hcpha^don^ EiuAetiWimi; ol.
Wwiplifll, pp. 10S f.
* Bu t sure Mciilft, 'Yaleur Ue* -TR- *n Sutakrit', M&L iviii. jn tT.r
•nd cf. 3. it
4 Cf. Vumrx h^+. dt.. pp- M wii j&j F . ;)VfO,S l*vi-
107 IT., KtifipuMtinum Autri C'aflrmim FWU pp 39 ff
7 Cf. Weber, Jnrt .Vr (nij- 224(00 I'inÊa, Chanda)1 -fftifrif I, $ lf+),
1 Cf. OnmjTxont. JV«iirr pp aft iT. For 11 full appreciitiun nee RoupdEy,
MSL x. 347 ff
7* PHONETICS IN ANCIENT INDIA
va-uimed, deprived 4if hroith *nd Voice; it liiLl->. piate when 3 ii
lullmwd by j uUip; ii i* also called (mMttfjNfij)*1
hi ihv of utop* umd tauivmvck (except r) tallowed by Hop*.
«WMaii£fc«J^ uikia pWcn l.e. a rent nun* <*r wbseufing of the totind; it also
occurs in pama *
The ex tendon of iht term ahhinidham to finals in puum is also

of int^mt; the description of them as ‘obseurc. weakened, de¬
prived of brcalh and voice" goes far to explain the divergent views
quoted above regarding their voicing or m>n-vcjicmgaJ The doc¬
trine of unreJeased final stops U ttrongty supported by the enm-
parativc evidence adduced by Gun t him in hi it La Fin dr ,\ht m /ff/
as also by later huio-Aryan developments*5
In the Rt* \\t also find a description of chc stop-elements of the
abhmdhMmi consonant* which distingxiiahe* between voiced and
voiceless:
■Jin- vfinml v-hich follnw* iht doiuirc fir an mi released (voiced) -ifop n
called 'J/jrmd'r *nd it cuntuiur* fur the duralIon of the *lop- rn the ca»r
of b ^ oict:lesa flop the dhruva in not a inti Me- and in the ea*c of a na*al n
u naaaliied^
*Dhrvm\ lit. *«iminuance * h parallel to the 't&nte' of French

phoneticians 7 The Indian observation compares favourably with
a modem statement such 35,
While the nTjpm* intkulaimg u pioxivq eonsenanr are actually in cun-
tael thi y farm what i* tailed tltr In ihc ease of Yofasle** ccnvuorumti
nothing whatever it, I .card during the *i:qp; m thr ca*^ nf uiieed semw-
nants bothc voice js heard during iIil: stop *
The stop-demem is, of course, not only a feature of the un-

rc teased stops: but in other positions am authorities speak of Uic
AF ; - 4-i 4K. am,’inTn ntai ffrrffafjOTjjcpitfiiah umr.ntizra hina-
n isa titfdith 1 tptsriiuyti tpurh 'hhittiMnah: djtihdpitas# tu,
NT i-'^rnnfAiTP/mi Affa-*uÂiirdfliin* ijwriilrtfiiiJfalmJiw upavddy.i
r*pkf*m Mmvwfiii*r iodifi ^ ^ rrftvntfnr Cf
I T it i^5. tnd Sarrrui. CrifuaS ShuStt im Ktitytij-artta'i SuMavjmnxdoprdj^
JilA.kytf, p_ jtj^
Li hfUfud, op. eft,, p. a*3, ‘The HTtxciiUuuti af the unmnee-ruiiils. b
filw-ipLiri » wy Lei *ml they ate jifmtipmfiKl with hreaifi nr voic^H (ef
K^Tiioiirem No. qj,
* 9“" v 1 Ct TJlwb*Arm. pp. 7, f.
m JV ir nd(^ jfKiJ-Fis Wujfh^jrJ.y Jhrv^4 rat auî to
tifkntrft ■ ndrihd-irhfiitfitoi tmvmiiikac frt
’ Cf JVinj*, 44,, n. Bln. fh* lenti Jtonm u*ol n iih refrmuw
» •lr: ■"■ I' ■'•■ .ta wmW n»e.«, ,
* t>. Jusk ji
PROSODIES n
cun!act a* only monic-rttirteous'1 'Kill ^.o not Jc&cr.'tiii; special

description; they take nuu\ however, of the specially brief and
light 11ap-ar!tailJlion of intervocalic voiced retroflex stops.1
3*T2i. r consonant (svarabhaktfy

A passage of the RPquoted above excluded r from the effect of
abhinitlhrma; this exclusion finds its explonsition in the following
statement:
After r njuiUinu behvctn n v uwcl and a ei-it^manf, a RmtUngmoit
I ttora^bhakti} o! ^îbiinv is [fuy+rtctLJ
I hi* dm trine of rvarMaAti, îimptyxfe", is elsewhere restricted to

the sequence r ■ fricative* (in so me treatises aim 1i fricative ).*
The most derailed stsicmeni is tliat of the AP*
Bciwrat r and aprevnualic fricatives wambluikii is pronounced horinf
tiit length td l nr J of u - before other am?- nunti fthim fricatives its
length ii l ot l of u/
Regarding the greater length of the svarahkakti in the sequence
r+fricative, ii n? to be noted that manuscripts of the AT' write
vocalic r for r in such eases the metre abo occasionally requires
the pronunciation of a vowel which h not noted in our texts
darfuta = 4 syllables),*
As to the quality of the nmmhhakti vowel our treatises show
wide variation; apart from r we find a* i. e, and u mentioned^ the
ft/1 ah * staler that it may Lake nn the colour of a preceding or fol¬
lowing vow cLat*
1 mtkit&m,
" r w £ « datprrt&i I rri rijnty. tfa-itfujj'oh nsra-madtiyayttfi (jn
ihc HP («e mo rihui'c) iu^wpffpj rz ifatiprpa *nd u u.wsi ihe
imcuitfktn of the wmiiowda), QtnboVbiv 14$ /is-/A«Jr rk?c:]m:
RF i. Fur ii nuiI* uf th» fwiure m Gujann T M, Dm,
BSOAS vi, &73 ft
HP vi. 1 r;{-!*up#hvJj'I J irjtnrjiSffio,inytld fkdfQ■ jvjt>lJ tt'firitfrAakiir
UI Utrii.
f fi IP to tj itphaîrnArtphn-nii*, A ri.V
v-K- f ^ “ 17 ir-nm i. f. Fivti-
/fr! Shfru r*-ijb iir/prfprtntitfrt!t4ô>n d■ hf-f//i tfrmyukimyy> i me-fkdnttr
itiim Iftiv jttfmi&ffiyu Jtraril.
P. loi T/tfoiU My?™** R ifChjrv/jj' irkdratydulham ^Huwihtim
tty *hf : tmymmin tyefftam fatmrtham tif jamum ttf.
1 Cf, WbLtflC}s tin AP L Ida.
WacfcpTO^.l, /fi. GV, i, j 3c. In I'niJim, dd ihc athtt hand, a written
n-irr ah ha ft it mu;. hu.\y ^ciu luu-diMJ vdllv fFridis], ^ e ji ; cf- 3,M bc;3aH.t i
M Cf, Vnmui.dp c\t B pp, f.: Reno it. TrrWi'rro/^-rV., iii. r3i
lJ ’Kj- 5Ji m
The statement of the APvti the lengths of 1 he irurubhokti vowels
goes «t u> mention as being of length | a an dement referred to as
*$pk#t*ntifrom a later passage we learn that this occurs in groups
where a stop i< fnllpwcd by another of a more back senes:' espe¬
cially a velar' (c.g in vuâ^^ftani.tud gaayntre). Uvafa defines
the term (which literally means ‘splitting’) as the (optional) separa¬
tion of a consonant duster.1 This would appear to indican a type
of warabhakti, whether voiced nr voiceless, tlie infinitesimal dura¬
tion of which is suggested by tile specification of a value 1 a, in fact
a minimal audible release. The mechanism of the feature referred:
to is perhaps the release of the front closure during I he formation
but before the completion of the back closure, resulting in the
momentary outflow nf an air-stmam attenuated by the back con¬
striction; it might then be identified with what Pike terms 3 'cross¬
ing glide":
When ihft rtlsusb* jtHdr nf itnr stricture is dmnhsKOts with die
nppAruetriri? alide of another, j trouiug/ttidr it devrliijttfd, . One inch
nudihle owting glide is the sound of open ttamition between tv,j ôicc-
lt >* .vr .n[i:ei] stop-, in [apia] j type of tiMslnu-frlidr ospiraiinn may he
hrard if the tips hi'j(in in optn while the tnngtir i» moving toward rhe
n\vtolar artli.1
Prasad m fact report*just such a transition fur Sanskrit loanwords

in Bhojpuif:
in the wc of "inch scunner* ini the talanmj loan-wot t_b _-li pninounrrd
by the educated fspraker*^ th? corn mer i* fim formed In oiw pcs i turn ami
jfter ih>ii the <irg&fl4 temj tu abmpily take xip ihir for the MtaneJ
con^nmit before die first is properly ftfaftd *
If the back closure were completed before the initiation uf the front
release, the result would be abftmdkitta; if the front release were
effected before the initiation of the back closure, the result would
he full ivaabhnkii?
\n exception to the occurrence of tpfipjona is mentioned in the
case of a retroflex followed by a palatal/ where a feature called
AorfitjKi, lit, ‘dragging, Ottoman", is said to take place; the term is
E jejj turf n -0 iph&fanaft
3 ii- .%& paFgwipatytyt ptm tm cr J
* ,
VP fa 165 rptiajmiam ret knk4rtf-€Wgr til tpufldi.
f*ri VPt Jfjc €ti tph^mum rjifnn vnpyvfuiVa fifthag-
ufftlranam 1 Op rim, p jit
* Op, di.. p. 1$^ t Cf+ me. fa* dt,
* Ettttttplti qtitHcd trufis the A V arc ^(ca, JUll Jilutiiia.
PROSODIES 71
further explained as kakiviprAhnrfQh ‘extension of duration*,1 In
tile absence of more precise descriptions we can Iiardly attempt to
interpret this isolated statement; we may only note that the author
evidently intends to distinguish the tfundtinn in question from
nnmuil nhhimdhuTUi nr w&rahkakn. The term as hi-re employed is
thus not to he confused with the t iprîor^j of Prakrit grammarians,
which refers simply to anaptpzis-
3.122, Stop+nasal (yam#)

Cloudy related to the preceding prosodies b the feature referred
in by the Indian phonetician* as yanuit lit. 4twin\ The Implication a
of the term will be discussed after a consider mi un of it* context*;
Within u wnrJ, when a rnrm-na^C iparla m followed by a nasal. it u
hepamted by ilir appmpnatc yitnutJ
N’on^ruod sp&tiar followed by n*»ab become (hdr own ytumat; the
yurpwi ait in conformity with ihe tfniritu *
After a nun-nasal sparia followed by a rmsd there are insertni ihe
uprrrripiidtc rtaaal tuiimLu tailed by some yomai.1
The feature here ob&crved Lv that generally described as nasal or
' fa vital* plosion of Ule oral slop;"1 the further statement of the VP
dial
Tl\e \mn*u m produced by the mot of the oum- (ndakS^mQia)1
may be brought into line with modern descriptions if we interpret

nmiha-mufo as 'velum’, in this type of transition there is a slight
nasalization of the oral stop, at least in its latter portion, and m a
Ilasal olf-glidc to the following nasal,* The Indian recognition
. If* li- mi t«s&argaQ'B kula-t?ipfakat{£ii ft! 4Urj _ r r {am
kdTftipu iti
# ^<He *Jw> tlw U‘<: of hit law la tht PuiptuAtm led, Srtnrwi, p. 5 |JS—1 Ziehen,
êrrbinffcm eiixr Silbe ttber dun* Stettdaucr uml Utwr ibren Haupnan hiram
ilatLh Eiji.Kikiebeu tma mdzm Mrr m^Knrcr uutotr Zw), and
similarly Uvapi in Cttmm. on VP i 114
* Cf. Vataruia, di. 60,
JiP i. 9^1’ f-tfJMrlrta-fwa^f tfiitUimtft Mt/eMf yurrurV ytithtl-i&n&fryttfn
(w Xiymwdiutontfl bhimtii}
HP Vi. 2U, JI. ip»Jr(ri tvinren *;pijeujhl ci? /uji.r/ij tpatMtt MffidnW *
1 rp Dcti ipvrliht nffiTiriiiprmTif itiHflhi 11, I Tij ndnk^St. ui*

yam4n fkt.
* Cl Outfmr. I jM; Pike, tip. cit.* p, 5*.
*■ Si. yamtfh.
* on the ^vlUbiOlKift uf ihe vjjjtlj (cf. Vantil, Of?, cit., pp, 79 f.
wouIlI iKUi ti* that there wu pome Icn^-thcziinc of the nSep,
7b PHONETICS IN ANCIENT END IA
of tliis feature has not always been duly appreciated; Whitney
writes.
Phonetic ataitytift dnea not, *1 it Accmi |u nu4 help to i coogniia the
yamd of ihe 1 lindti urjmimtrjjmi oA any ftcCcsuiirv jCCdmpfllliflicnt of the
ui (Crsifint of & mute tod nasal.1
Max MttUer refers to the Indian statements as 'vollkammcn kbr
tmd phyBiologiseh begndflicV; but it h only his own misinicr-
pretation of them that is so pellucid—
Sk wUrdu eb«n t inf.ich die lltiUJiche hervcuhebeti, diB wo etfl Nasal
as if cine Muta folift. drr Mtliu id but etn leicktr-r maaEoi iwsftifi *<Mrr-
gtki* [my itoJic&j.
Muller,, in fiwt, as aUo Relict, takes ihcr ywjifci to be a nasal
articubiion preceding the stop, But the TP speaks of the yvmia as
oceurnuE 'sparsad't Le. lifter the atopy1 and this is yet more clearly
brought out by the N&radf £iksii:
If ft nnti-natti rfiffifi h fallowed by a nasij, ihen a hoTnnr>'jnrr
with 13 i-t* firecmUrp bttar is tu be imwrttd Ixtwcrn them..1
The number of distinct yamas is tnven by the the JrttAo-
jyarafwfr and Uvata* as four* it, voiced ami vykekas* aspirate and
iiuo-aspinue; Uvn(a mention* and disagrees with a theory that
there arc. tvvenir {i.e, one for each id" the non-nasal letters!,
ami atates that only four types are dhstingniched 1 l>y their own
qualities*10—a mLcmcm of which the precise im plication m not
clear
We have no special in itni charts regarding the realization of the
aspirates in such contextsf and must assume that the faucal release
1 Oti AF i. iKi. For in ii|i|;ifc-ti.iSioii of thi KrrKO-doctnjie cf Itoupdk
AfÛi ]54« - ai*> I VU\\mwkf} AXFEukX*
h Du RP 405.
* Jawwtf AriaHitu*' V* Scr «f, pp. 333 &
h ct fUr PtSn. 1 j. 67—toftW&f tty wtrtrmyn.
' AS. p 4=3.
imrfniytd «tfl fcAunrt rurr,„ 'rtpgl *4 p*rmla rirJi
td#Mf fjtfjtff tm*arnnh purr
k 4. . fii^nJraf *w >omufft myftlii
' On jTP *.%; 11, JwXttAyjj
fifiixhujmd.ipirridt ftTulfhimu*narikynlr: Jt-utymt th itiyuh VftM myaltdpi,
* On VP L Ha. , , . Cf. nn i, 74, vtii. 54,
* On RP u rr.TJ^ riijiwtu- }mti baht tedruitp Mwvnutt and vL , f im.
i .mftvlJ ifttJmmJfn dirirA&Kfrrt yamnmfjp irupintitriu m# bhist:
atiurpdm mt rimdmiw prutiurndhpnlhamrtfft dziiiyti ch tsiyvm rmm j
(j/Hm/jYTjijni ity ufjirtt.
M Da ftiJ t. jo rt v/fiijpijjij fjantiifji rt.ii
PROSODIES 77
w*b followed by an expulsion of voiddess or voiced breath through
ihe noAc;1 the RP< however. objects, to the teaching of Gfiqgya that
ilic yd#td is then followed by si miaul fricative:" with the nasal
breadiing there is likely to be some degree of cavity friction* hue
the RP ii perhap* only referring to some objectionable form of
local friction^
Immediately after the discussum of tht yanrn the TP has the
rule;
After h followcd by n, rj,* nr m h inMrUtd u nasal tuund (iwIriAwi) "
A similar nlit b a bo found in the AP.' If, as Whitney assume, the

reference is simply to another yuma, it m strange that a special rule
should ht stated, and the Tribfta^ytrrtUna interprets the statement,
in spite of its wording, m referring to a nasalization of the h/ an
interpretation which is supported by the explicit statement of the
StiTvaiamrrmte-SiJqS* and, even more significantly, by the VS.h
Whatever tfic precise interpretation of the rule in question, it
evidently rceognho the fact the! the two processes of brtalb and
nasality may overlap, whether partially tar wholly; that this was the
ca_«c in Sanskrit h strongly suggested by the so-called metal he ?cs
in Pmkrit developments of the type tnmhann bi aahmuqa, &c,^
as also by PSnhi's optional prescription of junct ion-forms such as
kinjhrum beside kirg hunted"
As regards the figurative implications of the term jt&iwr, we per¬
haps have 3. hint m the statement of *hc VP that
Medially a lum-ouaal nop Wtaie * h*3d undrr^(H-> division {ftfctheida).u
1 Cf. k>r[id|>mii[H by A C« P*oc. 3#tjJ twl Cam' Phm, Sr. r p rî

’ *J. 37 3
- $^ flfWd" >079*090^ lUmii BtTfliAiJ-rtAAiaAy; mtjayrl tmn.
A a e,g;. l*iLcT r&wi#Tiirj, p. *4* . 'NHadw *jth loot] Jrklion at the vthc air
/nVdfviv *to*aJ* (very urt, limited ttlmwt to lypn tnuvli dnrin^ of the nose j".
f.7 Sweet. PitjrrrT. } 7S, ideJlinp fO ;i '* trough woneti' [«]# OCctiffmjf in 'ui not
liOmhimoti end Vrf: rtmigfrabk fmtrp of mlUtt^1.
' III. 14. AJnM HV-Htl'i'jU-Jh'JiJTII K.if, Jjky*j-1
H 1, ifl-iw. . koUtfrm* miakyimn fir.
* HLr-nn-4w-£fl*ttW AoldhflW ifruAyu PwfiMjmn an ■ Mtwiukve
A#lwuA i>vjV fJ_V ffrMdA,
1 4J, . . wfffftwju rt Htft wJn^’CfTvxm
* =MS, 301 OUldrfs, pp. t?St. oi, 101) The niiiriiiiitinn r>f H la, houevrr. prp-
lnI+sEpd by rhe CdrrJyiFtf> £ i«r VitiiliAp op, ch., p
" Pjêhrf, I iJt Ilh.isiElJ* Ndtypiditna. %tii it, i
krokmilifini.
■' viu, iii. J:f t4 {w. »twAJ: rui-jM^r ire^. C'f. VAn„ usi 1^.
jvma-Aj wiA^rt^wpih
H iv. 1^3, oreAisiijmt,
7# PHONETICS TN ANCIENT INDIA
Uta$i equates the term tnctkeda with ytitw* and it is understand'
able thai in a sequence with nasal plosion, e.g. [pad^maj, the
i>ni3 stop and its nasal release might have been considered as a
division of the d into two pans or "twins1 ([jiiitPma j}. fn ihe term
vkefttda we should then have a parallel to the German ,FSrtfcfoung\
a description which at least two scholars have hi fact applied to
the
3.123. Fricative + nasal

In the sequence fricative -f ruml one phonetician is reported a*
observing the insertion of a voiceless 5top humorganjc with the
nasal P thus kfâ* grflfma would be realized as [kpgbipj,
[griig'mn]/ the oral closure for the nasal taring completed before
ihe lowering of the velum. As both BEoch and Vartna have pointed
our/ thss type of transition is strongly supported by developments
in later Indo-Amm
3.124. Stop+Fricative
There are certain peculiarities connected with the sequence
slop Africa rive. Our treatise mention that the stop may lie aspi¬
rated/ari observation that is to some extern supported by Prakrit
developments of the type macrfuBra<nmtBsm9 kkaralcJiAm^
k p lira, &c/ This 1 realm cm* however, is restricted by âunaka
and Badahhlkara to those cases where the stop and fricative arc
hetetorgatiicr® by far the most frequent group of this type k kg,
and we may note the Swi$krrt alternation of kg- w ith khy- in the
root kg an-/khy aa- ;* an identification kg — khy U also required
by cty m olo gies in 1 he ôtapafka- Brii/wu 1 na and in Yaska > (\ j >«ftfaJli
1 Mfr'.Wb y^mL ill,

; Weber, /wrf. if. w taj: Tmpp, o*i Aflto, p, 152. n_ 150.
1 TPtly U. m id ^ ujmrwh fiaruh
pnitktimi) 'bhimdkinnh tp£tfia*pnrdl JuJihiin Ah).
1 Cf the inanition wimym in Engtiili {sen**] urn*)
* Bb$chr V{*4**Arymr p SS; Vumia, op, rit., pp, ff,
* AF ii- 6. rftitfpdk
ftp wi 54. $fmKkp*m ptaiMmnm ipmiitm *h* dvitiyam aiW.
TFiiw 11, tirrfui-jtarro dvitiyam,
a iV*. 4 t^s, p 4jj)
T Cf. PiaeheL f JI«: V<UWU. p 7). Nult ih.t, A V PS
* I 'P iv, t«. juntAJni> trailti Jcillytrp Himaakatya.
TP Kiv, ij. ijddatrfafcUfti>0tutriidna‘uh-
* Cf. fUoomfirii- Forte Vanantot it 99 d
f* CL Stlsdtdît3t IF*In. WZKM oa. : j5 f
PROSODIES 79
\\ c are also told of -i realization ns k j, going back as far as Gargya, *
ft thus seems likely that the complex which we transcribe as k£
may in fact have been in the nature of an affricate unit not precisely
describoble in terms of k and rtlmaiarmun, in his commentary
on the PratijM Sutra, points out that it is a tingle and indivisible
letter;* this statement simply refers to the fact that it Is written
in ftagari with a single symbol fa or ®r) which bcara no evident
relationship lo either dement fa;, tr); but the rwaJtsof a recent com^
pamtive study are of some phonological interest—Sanskrit kg, ^ is
well known, combines the reflexes of TE *ksand mkpr and Benveniste
IU-, now convincingly demonstrated that on structural grounds the
latter (*kp) must be ironed as a single phonological UfiiL1
3 -12 5. Getmnaiwn
We need not here consider in detail the doctrine of ktama or
‘doubling', which occur? in most of our (realises. This refers to
the lengthening of consonants in certain context* and would be of
considerable interest were it not for the fact that ides* regarding
the identity of thebe contexts arc so diverse. The matter has been
discussed at some length by Varma-1 The most general agreement
i* to Ih: found in the esse of a consonant preceded by Fi where both
manuscripts and printed works frequently show the double symbol.
I he fdoubling1 has no phonological consequences, arui Panin fs
observations arc as follows:
Afipr r or h prreoded by a vowd, und us the rim member a
■ coju.i»nanl ituy be doubled. .. h Bui according to SiLaly* therr i* no such
doubling, and til tqchm n%txc diet It hi laopcnaivc after a lonjf vowel.*
3.2. Syltabtf Structure

3.20. Voml and Consonant (see also 1,112 above)
Wh en considering the processes of anknbtkm wc encountered
a phonetic criterion for the class of vowels, namely, openness of ffic
* Wiv. 167; RP vj sj f,
' Cf- nS*o ibc wide mriftp at Llcvrlujtincuu in Kit£m {\f u* mr, £**-
Cve [/i,- Min'flft f (p Afghiirni ftm, pp 5S E)k
1 On iofrtL Wffbirr, P 3j>. * ESL niviik 130 C
‘ 1}V‘ efcp PP 99 C Cf. I'oMch^ £imU* d* Ph&ncfujui GSmbutr, p . 56.
VIII. iv. 4b, uw-ii Ki'k&bkpdTft tfo* (»*. tJ}
47- «Wtf Pff,
51 WNflnA (jc. mt)
5** dtrglÛ *L ijrytf mlw.
F^r diAcue*io£, tf. Wadtcnuagcj, op. di., $
8a PHONETICS IN ANCIENT INDIA
buccal a rticutot ore, By 1 his crilcrion h and «h might also be classed
as vowels, but such a possibility is excluded by the existence of
phonological criteria related to functiun rather than to mode of
articulation. This phonological distinction between vowel and
consonant is made in the following statement*:
A vowel farms u syllable ;*
A VOWtl with a conwmwu or even alone forms u syllable;*
A syllable « compoM.il uf ;t vowlJ, t wether with initial c^numnuO)
and, in pauut. a following coiwj unm;1
The conMituint is subordinate to the vowel;*
which the Tribhafyaratna augments by saving.
A wtiVftfljint i* mwpable r>f stinJinR alone. uiul »o » dependent, where ad
* vowel hi independent.1
Thus the vow'd is phonological!) defined by the fact that it form*

a syllabic, or (he nucleus of a syllable6—a criterion for which
western antiquity provides parallels;7 indeed the statement of the
Tri&h&iyaralm is almost exactly duplicated by that of Dionysius
Thrax.1' One mult of this phonological interdependence of syl*
lahk (dqttra) and vow el {tvara) i$ ilmt the term ukjara is frequently
extended to mean 'vowel'.*
I he distinction between the phonetic and phonological criteria
is further exemplified by the rase of ivarabfmliti. We have seen
that from the phonetic standpoint this feature is defined us vocalic
in character; our sources are cartful to add, however, that from the
phonological point of view it docs nut break up the con sun Lint-
group nor does it form an independent syllable."'
* AP L 93. m*rn ‘fcjrtnmt.
* RP*vui. 32, wtjanjamtfi . iuJdko td*ii ,ira 'hfuram.
* , P "■*»• '*>«*««■ mAlWvwr i jUft-anfn/i; uttoroit cdvuiitaiif
(Uvaiai e.#- ««***. vdkrpfm. * rp«|., T-ya^nuan
' 1‘imfyanav ti tritium aviUlhtltUm m ftlbrntl Mm lit lty h,™ Jiam, t*
* Cf- Tmbctxkoy, Ptinnpti dr pJwntî^ ml Cintincao* nq. , ^ rf ■

CrfHE, "ViryrJIt, OmiuTfif^ Ct $yibhe\ ANPE jevit. u it.
1 CL D, Tina*. <xl Uhli*. p. ifi. BrkXrr, P {,32. ‘\MaJ9* it ^
7™^;' i fruiîm,.. Pot I Jim definition, ct.
f toehde, ft, An!tatgs$riknlt Jt, r*w. GrvmamtH, m,
* Ed Uhb?> PP ‘J. in Sekktr, p. 6ji. W Vymw Sr,
4
** by*- i It, .M III, p.w* tnvTa *(* IW,,
mfrmtKripwt*. & ***£ #wi^f rr^p ^ errstcJLn,
- And iheiise al*», libr to ronn Ibnply letter’—cf MUL 1, L t, on
ftfti. SS&, (Kttbom, E_ jfe),
lB Cf vi jjt m wwwtnp n etrMmhtir vi&eafi; H*vp on VP Iv. 17. aa
PROSODIES iii
The etymology of the. term tikftna is a matter of some doubt.
Must usually it is derived from the negative prefix a- plus the
verbal root kfar-r perifih\ Lc* 'imperishable'1—a term which
is not Lnapptqpriatc as applied to its ntm-technical meaning of
'holy writ'. The VvidiMbhararta lias an interesting alternative;
taking akjaru as meaning 1 vowel* and accepting the derivation
from it interprets the v erb as meaning pto be moved as
subordinate to something else" (aitytmg{itiiy&)* The consonant ™
referred to in the TP itself as the * subordinate member1 fcr^r), and
the implication of the etymology ia thus that the consonant b
"moved' by the vmvcLJ
Etymologies of the word for 'consonant' {u\<injrmi) are various,
V vajaT deriving it from the verb ry-afjj- m the sense of "to minifesih
explains that the consonants arc so called 1 because they manifest
the meanings*,-1 This suggestion that the consonants rather than
the vowels are respomible for the differentiation uf meanings is to
some extent justifiable in a language where there are only three
baaic vmrd-clasra, and where otic of these {&) occurs approxi¬
mately twice as frequently as all the rest together;5 and in a notable
discussion on paradigmatic lines regarding the semantic function
ot individual letters the Mnhahhdfyu selects consnmntal values for
its examplesfrif/tf, kupti> mpit)*
3-Et. Syllabic Dwuim

Statements quoted above treat the vowel as the nucleus of the
taitizu turzuiM rtphadakStqpt* VpnuTL**! «< i<mtna ttapye^

ftuyrt lightiiam kuritivh AP i 104, mmym*ihriti)idia{ cuL (For
iii/ci iec jii[S| *bovr, and Vumu, pp. K41) C\ ilio Firrh, TPS, n34£F
P> 141, K BrursSului, R*rKi‘Litpprlk GrumnmliAh, pp. c 5 f.
1 Cf. Mbh, loc_ siic. ahfiifiiTfi kfwiirn vziiyfU—ms kriyatt na kfszrftiiti 1 4k-
pwflt/i The Nitukidda a drrivatioiL fnrai ufAjci, 'aiu*\ ii it 'ihi. 4* is
or »pycch" I nil ii HA kfiyaU IWim-AU tdiu 'hja ifi r 4 1
: On TP i a, nil Apmffnffrj Aparapm c^.trxmvi; Pad-
f r.uTSTT- di*t/j?r4ii-/dWo tttrtrf*.
1 For ihe UiluicoJrvfy if A rah mutethMnik (opposd to wish which
Cf. SkL I'Ufl'wul. KtFr full cliicuHaiaiL ot t)ir Arab termmu!^) cf. M. Jl_ A
El Surma. A CntuoS Study uj thr Phonrnt ObnivnUtm •>/ ttw Arab Grtim^
minium [The***, Ph-D-, l^cmticLci, 195 s >■ pp- 12S fT
t in RF t i> fr)v:#rt.jjiri pr&tatffifl hurmmty «wfMn Ifi' fynUjundm, Other-
ymdiftiShJiuwun^ mi TPu 6. parryci rtwfltw vyiqyuki iti tjoffiiniim.
See U'hitllcyp 5^. GV . ^ si, "j. Cf. aEÔ Scetron. Motor PHtmetia. p. *6;
Czwmuik, be rit.
1 1. i. a. tm PJ». SS& $ (Kidhom. L jo)_ Cf. 0. Strau**, ZDXtG kj. 6p
np.
b £2T4 o
B* PHONETICS IN ANCIENT INDIA
syllable: ami Lhey go on to describe the consonantal structure of

die syllable In terms of ike attachment of the consonant* to the
vocaljl nuclei. The general rule is tint an intervocalic consonant,
as aL-oan absolute initial or initial group* bdimgs with the follow¬
ing vowel, bur that the first consonant of a medial group Itdong*
with the preceding vowel, as also due* a final con&onatU in ptiusn*1
Some VLiriatkrn of doctrine is, however, found; the /£/J allows either
the division of a media] consonant-group or its attachment m tvio
tq the following vowel * and the TP attaches to the following vowel
a group consisting of conson ant-h semi vowel or atop-r fricative,1
The Indian smtements are partially paralleled by those of Greek
grammarians, more particularly of Ilcrodianu& Technical iti liis
Work rhpl JVimffems T&v Zrtuxi{ufv {"On the Combination of
Letters1.)* Tlw principal Greek rules art that jungle intervocalic
Consonants belong with the following vowel and that medial
groups may t:k divided between syllable* unless they are such as
Could NUnd at (he beginning of a word, in which vase the whole
group belong* to the follow ing syllabic (a treatment reminiscent of
the Slavonic languages),1 Geminates are specified as being always
divided, and the group stop+liquid as belonging to the follow ing
syllable,* The statements of the Latin grammarians are similar to
those of the Greek, stressing in particular the principle of possible
initial groups.7
It is to be noted, however, that these rules -if the Indian phoneti¬
cians arc nut put to any further phonological purpose, nor can their
basis in utterance be certainly established. Out we may fiurmiAe
*■£■ dP h 55*5 j- parevya n ^myti t yattiundm ■ tntpy &ldi ftikrrizty&i
t-«i> ttt :: \ : W \ 0'> I erf-.
* i, a?. 25. rt itrtiniKi? vyafiyutefaf nttmmya: . . un^pjgdi^r vd Cf. xviii 33,
3S‘ i^wn/smJflv Vtmmrysivu n iiftuyfimytm tn pitn^bAdh: tMtytfgihlu .a
; jfi7 JTB fcii. Ouitlrdi. wp. Cir , 5 114. tlwt ihe NS-yurE pry pine ipirm
(nay W Iwi.'tl on lUch .1 mvIImIik ruuctuir
1 J’”!)' wwdtigsm: mi parfi-ntxrttm: uriuAion /ni'rwuya:
âimlfdi nhitttbfxiktti -/j: mhnfiithdpîim uurrut^m
TpOTMi rofjrui'pawh, .
* Mm cf tlir fTtiimwaii ue id be fnuitd in vdI, ii, pp j,dj fr, of Leads’*
editiQn. The nuiornmn ur cciHuted in RUhner-Bka AutftihrL G*< d, Grr
Spty-.rh*. i ff., >« nW Hrtmann. »3tfArrArAfrfl|g im %n«thlU'hrir und m dfU
&mbw ftfc Sprnrht.tr, i*p. 123 ff.
6
* Cf n Hroth, V:fiurAr PfrrinrxiA, f 2I&,
* Cf. Slctwlri, i>|i. PJS, *4 £fc
l.f ktlrriiltfi, Dir J/J Li(ni|f pp. 140 EL: Htmumi, flp.
PP- AV* &. H_Lk, Sy sLihifK.-.Urjn :n Hunch %Khv, fforwtJ Stwftrr m Ctumcai
viL 24V ft.
FHOSOIJIES *3
that They were formulated with reference to the perceived termina¬
tion .tod onsfit of certain pri>sodic sylkbb-features mch as promi¬
nence nmj tone; the PS speaks of tone as a characteristic *->nIy of
the vowel/ but the I P remarks that tt may be shared by it eon-
sonant/ and the VaidiMbfiamna makes the acute observation,
which we have no p*rricul»r reason to doubcP that in certain cases
the final portion of a falling tone was borne by the consonant
closing the ay liable, which was accordingly lengthened.1 In any
case w e should hardly be justified in following the example of the
Western Sceptic, Sex tins Empiricu*, by referring to discussions on
syllabic division as la lot of empty linguistic IlQm«!^c6c,.^
3.22. Lajii and Duration

From the pmni of view of syllabic structure the prosody of
vowel-length Ss of great importance*3, The device adopted by the
Indians for purposes of phonological description is that of the
maim or 'ntotap 3
A uhon Vowel 1 mdtrti; a Ions? vowel = ; m&tr&r*
A piufa vowd ■» 3 m&f&tJ
A vowel having ihr. 3^ncdi of a h short, and its value m 1 mfitrd; a vowel
twice as Ion* h long, and a vowel duct tmu& 43 loniE is piuta*
The mfitrB device has an evident utility in a system where die basic
Vgwd-Unks arc considered as members of qualitatively similar
pairs each comprising a short and a long member, and where the
1 i*.
1 i* 105". mannta iawitwm. Of Piic, Tont LuM^w^wi^ p. 7:
'Although the ftnemt* am ufiuitly placed wi the vowel 01 xht »ÎLat>l« a
nmy rod aO « canjorymt m thv ante *.y]Lililc" £ wc *ho pp. 11 , 30 ft., Jl j,
Koih yiizwn are imattiTtd in the. no mmc-nla rr on AF ill. 7*. *m rrrdtn vyaH-
jtmOm: imMj
1 On TF i. 37, , . (/idtitwMlVl-MftuitVAifi ■fcrfr/nl JrUnix*;t nrrfu titf AfefUlfi
HiJ-^nâ-bKQirfit r yn■njfnrtfv #ra: fifrd jrtmlfrfprf/i rtfity ucyan:*. tffdfft
kdtdj*iiliytn*t ttit ^hritair. Of, Vurmt p_ |6j8.
* Liirrfu* Gnimmittkut. J « f vf mf ir&pa relf ypOfjb.ii-
3r<ji . [c Kj* sn fefcrt bwn WKftfUd that the bscm of
ihe Gri^sif Atticnirali ji ortii aphic , cf, Havel* R^vutr Crffifur, xvi tifi. for
lUMriptirnuS iUiRKJtj^ uc 1U0-SstHtfl-Jianfe. Cfaifs J?r, it 3ohfF., HcupaRn,
<ni. dt,, pp 132 ft. - K F A Schmid c. fintt* 1. Gtich. ii. Giam. d. G*. v Lut.
PP* ija ff
Cf. RF L ... . dfrghdft . . uhfmy* ft- iikjsirdm
* -IP i- r±*S'-rui\ra hrxrmh: ifri-mtlt*
^ Rf* 1 J7t 2^-yj. mdltd itfti?fu,h: 4j * diryiuih: pfota Ktyuft nxsmk+
F/1 1 55-58. tf-iTttf/ra-n;nrv firm* dt: mdtrd at: dvh Ididtr dlwhah: ptutat
tr& ct TF i 11-36; AW^HiÂKfff 4&; Upalthha. t £ ft
»4 PHONETICS IN ANCIENT INDIA
junction of two similar abort vowels results in list: corresponding

long vowel (e.g_ dlvl-f-h u — tlivllva).1
Ths/dWa or protracted Vowel mentioned above is of rare occur¬
rence ami is bound to a very limited scries of contexts; it represents
the over-lengthening of the final vowd of a word or phrase and is
used in eases yf quest inning, especially of a balancing between two
alternatives, and akn of calling to a distance or urgently'-’ (e.g,
idum bhuuyaaa idaaam itl, *h this more, or that?’)* For most
phonological purposes, however, the pluli vowel may be ignored;
as tlit HP points out, it occurs but three limes in ihr R V and it is
in all cases related to the special type of speech-function,5
The statements on vowel-length are followed bv a reference to
the length of consonants, die general prescription "being fnritru*
though die AJ* says i mniru'P tijc VP is even more specific in its
detail:
A consonant bos u value of ) mulrd: | mdira is known to an n»u
t'mmnmt’t and 1 miw it known ss ■ parattufritt.'
But the mat™ concept has no justification in connexion with con¬

sonants; certainly it might have been used in conjunction with the
rules of syllable-division for purposes of stating syllable-quantity
(see lie low), but in Fact our treatises do not so employ it. Their
statements are thus only of value as generalired prescriptions of
relative duration; this unphonological approach undergoes a
further degeneration in statements such as the following (from
the P$):
i m>ltru is equivalent t:> the cry of die blue jay. ; wan,it to that of the
crow, 3 ntiitrA, to ihat of the peacock, mil i rmllni to that of die roon-
umxtc,T
Httt vve havc an attempt to define phonological Length in terms of
absolute phonetic duration, a practice comparable with & modem
statement in terms of ccotheconds, It is gratifying, however, to
find that these attempts are almost certainly late interpolations by
1 Cf- TiulMttby, PrpKTp&t pp. jci £
1 Whilnry, Ski fir., |
'1 it WWi 1 vi -J tiilpi' * md dlfyt firth.'- phvir %ti$r im V&dttt&f m%
1 c\k TP i. «7; Sttrt‘H7vm**urte*Siksri, 40; l.'pvltitho. 1. ft,
1 L 6*
1 i. Etflifi*™ Mdhu-mSfwti tml-atttkam apt jMmatfpt- onAdm.-
mfrm.
1 4*1, L'J+iii tu f-xJait rnittrtim jti+mdfrmp tmtm t
J t.tAi f ritiiff r«d mikidai fv mdh^nuUraktiwL
PROSODIES Sj
pin timing kail} incompetent scholiasts: (hey appear in almost
identical form in many of the Lite $ik§as,! and it is significant that
a similar verse which appears in out text of the RP! bears no com¬
ment by Uvata.
. .
3 23 Quantity
Beside the terms ’short’ and 'Jong' [hrm-a, tUrgha) Wr End listed
in the introduction to the RP and in the concluding stanzas of the
TP the terms ’light’ {i<u>hu) and ’heavy1 (guru).1 These latter terms
refer primarily to the quantity of the syllable for metrical purposes;
but Since the term uiqtpti, ’syllable’, is also used to mean 'vowel’,
the vowel rather titan the syllable is regularly stated to he ’light’
or ’heavy’, Whitney's statement that 'for metrical purposes syl¬
lables (not vowels) are distinguished by the grammarians as heavy
or light’4 is unfortunately not true: but we cannot but agree when
he remarks,
The distinction xti terms between the difference of Iona jimJ short bi
voweUoonil -md that ijt heavy und lighi in ij-lLihlf-rotisttucdon is vulu-
sblc and should be observcrij.'
The failure of our authors to observ e this distinction m fact leads
to some confusion. The actual rules as to svlltbic quantity, liovi •
ever, are not in any doubt;*
’Heavy’ arc:
{<r} (A syllable containing) a long vowel (including a nasalized
vowel)
(fr) (A syllable containing) 3 short vowel followed by a consonant-
group ur by a final consonant in pausa,
'Light' is;
(A syllable containing) a short vowel not followed by a consonant -
group.
1 f"f tihrmb, iPS Xdlti, [). 77, * m. $<>.
T HP iHtrnd. 5: TP wav. 4 Skt Gt„ 4 76.
1 Ibxd cf. t-mh, TPS, m+a.p. 1 jo
* Cl, rPxafi. u-,5.
y&tf ymt a cApi dhgkam
yoga -^rirr i^.'ri mthdimmSuluim
*hIm IvlH-t tfdj fUrujs; Cufydt
ifftinv nflp 1 If 1’jIpjj' jtfftt tafMuf.
ux:y&nv.w;{iwi yati eu y<il

ttVjftritinfrfl ■ uufjpuktam etui lughu wbwfhatiL
CL API 5t-j4; HPL to-SI; p&f. t. hr. ro-!*; PinRab. Ch, S. tnfrwi 4,
u PHONETICS IN ANCIENT INDIA
A final short vowel in pci us$ may be lengthened ami so form a

heavy syllable.1
It will he noted that sylkbk-division and length of consonants
arc not employed in liune statement* of *jllabk quantity. One
exception ifl provided by the RPt which static,
(A ivLliibEc eon limine) ft long vuwd U heavy ? and heavier if iKCffln-
pantt-d by A censcmam; (n sylbhk containing) a uliort vowel with a (pre¬
ceding) corttcuumt ts li^hr; 4iut tighter without a rannotiafii.1
This suggests a scries of four quantitative values, via,
‘Heavier' «p)
'Heavy1 (2 in.)
Tight* (11 m+)
1 Lighter" {1 m,)
But this distinction has no metrical consequences, and is not other¬
wise mentioned by the phoneticians* who regularly apply the
mâ-ctmcejit to vowel-length and not to syliftbk-quanUty. Bui
a dangerous step bss been taken in the statement of the I P lhat,
A (short) vowel before n conwyrmnt-ynnup La equivalent to 2 mdirui *
We may assume the logical background of such a statement to

have been somewhat a* follows:
(a) Short vowel before consonant-group = heavy syllable
(4) Long vowel = heavy syllulik,
.%(<■) Short vowd before consonant-group — long vowel.
(d) Long vowel — 2 rnStr&t,
.\(e) Short vowel before consonant-group — 2 matrm.
The phonological conaequni£» of such reasoning are &criou&
enough* but the greater danger lie* in the temptation to lake the
mxt step, viz*
by (a) and (>), heavy syllable — 2 matrix.
And this step has in fact been taken by the metrical authorities,
who employ mutra m a quite different sense From that intended by
the phoneticians, with the convention that a light syllable is equiva-
1 LT r/* iv. log, AP be, cit ; Pinsiln, be. cit. Set llLjo Gamble*, Ftm jJr
chuf?. ■*-
- Kfiii, 4t*4+- .r; unu \tirgfoiM; gariym fu yack unyanpiHoip hkorfi; taghu let-
jtijfpMp krantam: taghiyt ^naS^mtdd f it,
* Except m an titferpietnliolial quibble by P*l40jali frJuCulKd by Koejow*
op at-. p. jra),
* h\ ley . kaniyiîL-pJun.'JM^yaândntdva^rmyâldh MTurd dt'i-rtnsll¥4fy-
PROSODIES ft
lent Id I rrtutra and a heavy syllable to 2 mat raid That the twn
uses of the term have nothing in comm 1 m is recognb&cd by the
author of the Ifttauiuhtaviditarala, a manuscript work quoted hy
Varnio, whose tnm station E Dike die liberty of reproducing:1
Ry a cortvcrttiatial tradition the qimnlity of a I<mg syllable in att-uûred
*» ivro mytaaH whivh Ate attributed even m the tfuta vowel (that ha* ihiw
rnoruaV. In thx Matrrr w*yr iiMiough the quantity of a amsoiumi ih a h*If-
mem. 4 syllable coding in ■ consonant is measured two moraa, That a
ccm*«njint aboukl nnt increase the quantity of the syflahk is due to con-
vmrion.
The U&i sentence is reminiscent of the remarks of the RP quoted
above.
h b most unfortunate that the term mfltra ^hmild have been
employed at two different hut related levels of analysis. The con¬
fusion is iikinntdy traceable to the equation of syllable with vowdfc
and the extension of the single term abora to cover both*
3.24, Tone (warn)

The Endian authorities recognize the existence of three tone-
classes in Vcdic:s
’There are three tones* uddita* unudattnt and starit a * uddita
means literally raised'. tmudfitta ’imraised', ind svmta 'imottcd'-
Tfaafcc are described by the AP as follows:
In st given register a syllable with I ugh tone ts iufdtta, with low latic
nrrscdtfrfo and w ith falling {dkn/Ha) tone svaritn\ the first half of die ttttrifd
it tiiUtlaJ
The literal meaning of akftpta is ‘cast down*. The description

given in the VP read*:
Tin- uxi&tU7 i« high* the armhUtu low, mid cJtc reared « ojmbi rating of
*hx: two; , the frret half of thr it an (a is tiddite and the bluer part is nude
id fall (pmÂanyiitd)*
1 a. c.h. S- 1 ijfF.. with Hiliyudbif* cotmtarm

1 P *□-
1 The TP appears further Oo have ebervd the neutralczdbon of tonal dia-
bnclic iui. in tvhtapcfrd *peedi iscriii 4- cr« cl' pjkc T.w Uin.tt-w--rrr p 14
* PS 11. ctddttrtf idmtdititni ro nmriM ca nwirdi Trayah,
* P 14.-1 7 CiJTWifPTr^ VHTrfli’ ' A/«l r^m u; L flir u44ttOm: Ttidrir iT7TJt±itUimI dkfiptam
tUtinSum; it vjri'Jn^tdfilO imttfdndhiUi uddttam,
* k. joS-id* uremt udtUtak: irtm amutdituh: ubhas-errant mwflakr
l lb, taiyddila uiMlltim Wa*4n&** mdtram,
iv- 140 n witwyti t&ltara dddh prtnrifNaityiiSr,
MiOKETfCS IN ANCIENT INDIA
and in the TP\

The ttdfiltsi is high, the amufutfn [aw, «nd the iF4irtfts a compound
tone -, . it begin* at die level qf the uddtta und the rest ia m die level of i he
itntitUiMtn to tuy ihe teacher*; same say it ia a cniitinutma full (ffrotpiu*),1
The term pravtma means literally * downhill slope*, Lr,
An exception is provided by the RP* which shtte^
TUn drat | or ] wdfrd of the Jtaorita t* higher (hm the udlUtt: die rest
is
ThU statement suggests a relation of udatta : narita : onudftlta
somewhat as follows:- ; J ^:und this is supported by the peculiar
sysiem of toiic-murkinE employed in Hie texts of the J?K where
the svaritit7 and not the ud3ttat ls indicated by a vertical stroke
abovc tlie syllable:1 thus a word such aa agnirmn, where the order
of tone?. Ia arwddtta—uddtia—warUaI is marked ggninah, the
medial tidatta -syllable being unmarked* Thft peculiarity of mark¬
ing hi ako followed by the A V and by certain tests of the IT;
elsewhere in the YV the marking is in accordance with the pre-
scribed realisation, c,g. agninaa-
For the 'svarita wp find various other terms occasionally em¬
ployed, for example, in the TP dii-yama* i.e< 'of two pitches'.1
and in the RT 'uri-nira\ s.e, * high-low* ;* the Ndradt &%9-$ays of
the ttarila that it is pronounced ^between the udaita and the
OHuddtta {tiddUamuidtlu-uutifaye)* a description that is reminiscent
of the marking found in the texts of the SSmmttdti* where the
tidatta i* marked with a figure I, anudatUi with v, and tttmita with
i a a
2 (e.g. agninun).5
The wariim are phono logically divided into two main sub-
classes, ihr Ao-callcd H independent' and Hdependent’ (or "enclitic'),
'The former is generally explained as the result of a coalescence of
an udiViti frith an amuiiitta syllable (». - -h-* = S\ -*~ 4.
dlvl Ifa Ltl’.Lji* T||-
1 1 4<> 47- tut'tti* u Jut ink. w^t* taWnihiirah mnituh (cf,

Pdn, 1. 11 10 . ddi* tiiyoddita-iAmati r/m 'nutldsta-Mma ily dMryM -nr*. jJi
pro&Epi 1taV *ki, Cf- ftPiii, In
1 til 4~$. fcaij -xhlihi fairvriut11U _ j j JJ r- t -mJtrdfdbtiH n>j nj'.‘
3 CE iW.* V-jfuu Thr Volk Accent ami sh* Interprets (it P&nini\
Bwrthdy 36- Ik iflja. pp f ff
1 3“ 1 N # £5, JiH 41Sl
’ Cf. Vmrtii- ’StudlM tn tht Acc^ncuntion ct the S^mavcJij /Ver FI JUt*
JrrJtn Orfotttil C'rri/,, Ptnn^ i<J3e*pp
PROSODIES
); the latter ss a variant of the umtclntia immediately

-m* i
following an —as Panini puts it;
Afft_r itn udtttitj n mania aubitimted tor ihe
The dose ph on ctic relationali p of t he ind eptndcm and depen deni
maritas* however, in- nor denied,* and in some texts we find inde¬
pendent rcflTtfflJ indicated, like ihe enclitics, by an uddtta on the
preceding syllable.* In both cases it seems certain that its descrip^*
tkm as a Tailing* tone was justified, whether it represented the
glide from one 'register* tone [udiiito} to another (amdfiti<j)t or an
independent ’contour" tone (to use Pikes terminology).* The state¬
ments which refer to the first J mom as high and the remainder
as low ara probably speaking not in phonetic but in a type of phono¬
logical terminology ihat js prophetically redolent of Fragile.*
it wilt have been noted Ihat the APt in describing the uddtta as
high and die amidutta as low, made the important specification,
'within a given register ; the pilches, dial to say, are relative and
not absolute—a point which is specifically made by Patsnjali:
The terms ’With' nnd kIow’ have no Absolute %s mi heat ion*
To quote a modern parallel,
it i* the relative height of the tonemes, not ttuir scrnal pitch, which i*
pertinent to therr linguistic analyst!.1
Certain of our treatises attempt to give some account of the
physiological processes w hereby the tonal distinctions are elfcctcd.
The RPs statement is as follows:
Tlir three Kmrt art uddtta, tmndAua, and mmte: they eh effected by
lertsmcw* lames* ami 'dkit'pa respectively."
T VI11. (T- 6* widlt&l uTtiulJttmya wiariUth Cf. TP iiv_ uiLlttdi pur#
1 nuiidfUih tumiiam. Upri*kk& viU. 7. uddttdd tnntddUfnji ftdr,nri, . .
1 Cf, WickeiTWêJ, At- Ur. ir * 140 But for vinuui mmncjtA, which we can
flcarcelv hope to [nl^Tpitrt, ef. TPxk. 9-ta; VFL
* Cf. Mnaknwa V*d GV.h « Sy.
* Tam* Tart£W£*t, pp f ff
1 Cf. Trubctzli-rry, ffmrtpift, pp. »4 ft i Maitmei, FWdttfy aJ
Pftottdfici, pp. trj f:. ("If. ..we make use cf ihe concept of mora, we mn cannidcf
nil melodic tones as iucccseiUcip of two punaua) toivr. and thus reduce the
number cf distinct pn^wiiol uaiti . JukcbHcn, 'Die Brt4mmg und Ihtt
RoUu eh d*r Wort- mid Symajptiaphoiiologie*, TCLFiv. thj If. On the Inniiii-
bc.ni uf ihii Tcchniqut. cf. uu, I. 1. 51.
* l. S I. on K. ii. *30 < Kidhofft, i. -07]. sum*warn flajjwflhYiK
jSjjîifâ/ca/rr, 1 Pike, Tvn* JUxPttftWtf**. p 4 J cf pp. 17 ff.
* i£L 1.
fj.f-UJ.it tdTsttfdtuti * •* tr.izriliii ca frdVuji n
to uayu^¥
w PHONETICS IN ANCIENT INDIA
The terra ‘akfrpa\ 'casting down', lias already been encountered
and can hardly be related to a physiological basis: the uther terms*
however, are extremely suggestive of a reference untie state nf the
vocal cords, and this is in fact borne out by the explicit statement
of the l*&ri-ftiitfS,' examined by Vanna,- as alto by die correspond¬
ing passage in the TP;
The high tone d effected by ictitiun, iwdnr*> ami constriction of the
gloltts; !he taw tone is effected liy Itotnesi, softness and widening of the
glottis.5
*nic*c statements are not so very remote from a modem descrip¬

tion such a? that of Forchhammcr;
Lits MiltclreipStcf ktihneti wir die Toobfibc aoivohl dutch SpflfUicti vyjc
wefa durch AI»sctiliinLcn tmw Veftlfckcn tier Stimmtipfititi verandem,*
and certainly do not merit the criticism made of them by Whitney—

There is evidently much mart guwwoftt than true uh*cmition ui thin
nile. , . P There £■ nothing m hU io commend in such u drtdnptinn of the
WWf in which kfw tunc ik produced.
Even today the precise mechanism of pitch variation is a difficult

tuid debated topic,1
h should here be mentioned that in the musical treatise* each
oetas l- i:' referred to as the 'drrigmit ’or 'double' of its predecessor.*
□nd it would be tempting to eon dude From this that the am hors
were further acquainted with the theory' of vibrational frequency-
rat jo*: this interpretation ts mvali dated * however, by die cum-
mcrtliiLor'seLatemcnt dot'dunhie1 means simply 'double in effortV
Ulc weal is Lhu> here in advance of the east with Euclid's
observation:
Of tlic laund-ntcvvemeus* tome are of higher md «otnc of lower frt>
ijum-ty: the high Frequencies produce sharp lotiodfi and the Jam fit-
qwenrir* %kep ttrund-i*
1 . inutfiff k*mth*~hitfitya , rfrfrhi kimfia-bilmpa.
1 Op, cir., p. sfij,
* kin:. •t-lQ, rAtfirnyrnu tin:*!;* kfowywiy mc<&h-kardrf tebdatw: atsea-
eiwyo m&rdmmm umud klituxiti mtemb-kardm
* i>p eirM p. 3oi
4 tf R\u-«]l. Sjvn/i ami Vuu*. pp. 1*1 IT. Ser fiuthtf IK. Cxujf, The
hleÛCUim O! pitch Oanjgc in ihr Voice', ytnmt. iV/ Jttprfrigp. mi. |Wr
PP *54 ^
* f, iih 7.
Kjtlfi^rha on to*:, cit. (I imkhipl 10 Dr A. jlib for ihn icfmnet)
1 ijfltw Ctttwiai 2 1.1 M 1 Tf* & -;41 rt£T* iifp at *ii« tia W4 j*t m U Upatvnp** ,
™ ■* mKwAttj* ~<?tv ^ 3# iW^repsi fiajvri^v-

PROSODIES
If the ancient linguists employed any graphic method of indicat¬

ing tone, it lm {apart from the orthographic systems of our tests)
been long aiiice lost to us; but we have something closely parallel
in the manual gestures prescribed to accompany reciuEion of die
Vedie hymns- We in fact find the terminology of ihcfcc prescrip¬
tions to lie related to dial of the phonetic accounts. The root
Ajtp-P *iq throw', which was used in the phonetic description of the
svi&iia appears also in the gestural terms urdhta-htpn and tidkih-
kitpa* 'throwing up' and 'throwing down*, referring to the manual
gestures accompanying the udatta and onudMta tones respectively'
the gesture for the toarUa is said in dm case io oonfiiai of n com¬
bination of the two,1 The verb pmnikunyatt, "is made to fair, like¬
wise used in the phonetic description of the svtmuI, again appears
in the description of the gesture accompanying it.- Uvata, com¬
menting on the RF* phonetic u*? of *akstp<i\ had equated it with
ttbyag-gimuinn\ "a moving tcn»*V and this term alio appears in
the gestural section of die VP, where Kama is quoted as teaching
that for one type of nuriio the lond is to S>e moved "across and
downward 5 7 he. diagonally downwards, hxhm appears probable
that some of the tonal terminology is really based nn gestural move¬
ments. which are of course related in turn to the kinacsthetic and
acoustic phenomena.s
The Vcdic tonal system lias left tittle if any trice in the modem
Indo-Arvan languages-* Hang, in a study of contemporary Yodic
chanting, claimed to have found the ancient tonal system preserved
(a svimta, for example, l*dng sung on two notes of which die first
wasthc higher);7 these ckims cannot be accepted however, without
further investigation.
Greek, like Yedicp had preserved the tndo-Eurropeart tonal sys¬
tem with considerable fidelity', and it will be of some interest to
1 5S, Rlmftaaifuan nn I*rati}rh£ S., ftl Wtbef. p. -> . C f. 2 OH
Vi3 L tZl- tatrodditt ^rdhraîimifnnfji AurfUiyrt mm&Ul*
h&tuixyit. ,, -
L VP 1, 134, ftvn* mhtilyit priinihimyiilr tid&ilt,
1 On HP in t. riMf* *&ma
* i- izy til pUttWfl trryuf* nihaiyu Runtwn; 3
cf. I. I Z. Hie term
a u«d in ■ Kiiaur.U ten» hy the Ahtrirniyadatpa*-^ {wo}. in e phonetic tem*
*t flppnan u% the < B (XL iv. 2- 7, Mfldhy. alM hutim lityak. yti yOuntm**
<IalU tsm&tti mrfadhyUU , .)
f O al*> 5. K. QwtEeijh P***- ztsJ f"*- Cons- Phan. St., p 176
* Cf Itkich. Ufmh-Aryen. p 4&; ^ ■!*> Ft. L. Turner, JHASt ipi6+
pp. Soj if
" Chi' dai Wtun u*iJ dm Wtrth dn nrtJjitJwrt Acttnft, pp. *3-;4
Q* PHONETICS IN ANCIENT INDIA
see id whut extent the parallel system* have given rise to parallel
description* by the native authorities.1 As regards the high and
low tones, the Greek writers are agreed in according to them she
titles ofi>, sharp, acute", and J&qnf,'heavy, grave’, terms which are
parallel to the Sanskrit udatta and amdatta. For the compound
tunc we find the terms fn'mins1, 'of two tunes* (Comparable with the
i/rt-ydfiio of the TP) and "acute-grave" (comparable with
the of the fiT); the common traditional term is mqxoirw-
jjwitir, "bent round, circumflex*, but according to a Byzantine source
this term referred simply to the graphic prosodic svmbol and was
substituted for Afvfiiip*w by the great Alexandrian grammarian,
Aristophanes of Byzantium, when he changed that symbol from
to " ; the same source aho provides a phonetic deftcriptiun w hich
parallels that of the Indian rvarita, via,.that for the ‘circumflex’ the
voice stans at the acute pitch and falls to the grave, thus, constitut¬
ing a combination of the tno* Dionysius Thrax distinguishes
various types of "circumflex", which are further subdivided by
Glaucus of Samos, but, as in the case of the Indian descriptions of
particular sub-categories of ftarita, their precise identification is
impossible.
Regarding the physiological processes of tonal differentiation,
tJlauous provides a parallel to the Indian sources when he de¬
scribes the acute as imrertifiiuij, ’tense', and the grave as
"lax". UV iD-iy also note tiut for tone in general the Greeks use not
only the term u^JOtripSm (‘tunc’ — l,at. accoiftii, equivalent to Ski.
trttm) but also t*W (‘tension"}.*
The Greek sources also mention a further type of tone, the pi&ov
or ‘middle tore'; the identification of this mth the circumflex,
as assumed by many scholars, can hardly be supported m view of
the fact that wx find them mentioned as distinct categories within
individual statements. The ‘circumflex’ tones referred to bv nur
authorities sre those which we find marked as such in our texts,
and these are all in the nature of ‘independent’ tone*(cf. the ‘indt-
T«t» 'll the Cir«k HntoutEtta which tollttw will be ftmutl cotLit«l -n p„,t.
Ifn'c, Gwd* r.- GtrA' .-L-smtiuttjfm; Stuitmsl, ftp, cit.P ppP 94 if
1 , , , crsiK'jpaiiq tuU BKpunrwpfrtM. Ufithi Tl
tMnjAftV T= -uJ * -1 k!■ '!'■ H.'.x *U T9 finpv, nv&rv iUfl jj -ai; av-W .r£ tqE „
6f*c f *:=ti . .
* Th* nymohi^caliy equivalent Stt. b used &dy ui the un*e of ‘mano-
t^nr' {el Uvafa <m VP t. i;o; RfcrmUniim 1m Prxtijnd SHtta&; Web<tt Imj
ac. 411 m.
PROSODIES
pendent jwtrita* mentioned above); but on the evidence of Greek

musical fragments Professor Turner has long since made the
observation that,
When 4 long unaccented vowe! wrs sung on two note*, there Wn$ a
tpndenry , . . in che 4 those fnlkm'in^ bit Accented ay liable to in nice
the first the higher This is in agreement with what we may mnnim*
to have been the nature of the vowels folio wing on accented syllabic from
a comparison with the Yediu accent iJ
the unmarked syllable following an acute was thus equivalent to

the 1 dependent1 or 'enclitic' svarita of VediC. It seems reasonable
to proceed, with Grammont, to identify’ this tone with the jxkjof of
the Greek authorities
Cr Km figure toujeun* k ctitc dc I’niKUr e’eat qn'cm sic pent pas mom her
dc Paigu Jiu gnre par unr chute diponrvuc dc timet, maii par Lin «Ui
mttitmihijaircf qm partanl dc la httureur dc 1 Wu arcive progress! v’emmt
ii rrllc du srravc. . . a Le pitm esl cxactement Nquivitirtu du ion que ics
Hindoo ippc&km mmita (descendant) ct qui eat imcmiedmire entre
Vuilfittii (haul, sigiM d iVmtdfittii fnon himr, baa, grave).*
Sturtcvam’s objection that 4 **ime mediate*" would be a pecu¬

liarly inept name for an accent which omened within itself both
the extreme*' loses much of itfi force in view of the description of
the stariin by the Soradl Siha (viz. udattanudaiui - madhye) and
the marking ^system of the Samavrda (see above).1
3.3. Tempo
Ii will be appropriate to conclude die discussion of synthesis
with a feature which is proven it nentiy related tu the major unit*
of utterance. The following statemcni on the subject of tempo
appears in the RP:
There are prescribed three apsed> of utterance—reduced* medium and
rapid; for repetition one should employ the rapid speed, for recitation the
medium speed* nnd for thr untruetiem of pupils the reduced speed.*
Thla recognition of the pedagogical function of the ‘reduced’ tempo

(xdlamithi) finds an echo in Paul Pasty's \ . * ** pronunciation farm-
* Ooxr. mi, 1015, p. 156,
■ Phsmttiijur Ju Grrc Anrim. J>. 38S-
1 Cl, ahn Vctidfy.hs Trail* d* AtstutUnliim Grrtqut, { +5. ‘L* fflM jJ»S
■j^jnsruji hEqri id non jmji interni^diairs . . rutii combine dea tirm.1
* sin. 46, 49. Auto ryrrir UpadUfHOi * viffl 14?^^ mfuftjftinrfp 01 idfrui'r^ rd r
tihhyti-rdttfie drutJî ijnim tm madhgmmBm
Hjydryim upudciiSrthf ktirjitd Ifttitft vMmdittim*
hire rslciuic1*. que jc Liens pour fipcL-iakjr.ent con Venable a Ten*
BCis{êTntllL,.1 But of greater interest would be a recognition of lbe
significance uf tempo as a linguistic category; in addition to impor¬
tant semantic functions its consequences are evident at all levels uf
analysis- anil modem linguistics must sooner or later find appro¬
priate measures for its description. The [',£ does in fact go so far
Ja to observe that the medium speed is the basis for the establish¬
ment of phonological length-distinction?.1 In this, as in so many
of the matters discussed, the Indian pioneers have pointed the way
for future research; the details of their work arc such as often to
evoke die comment,
nihil rat lam dinuin quod nim dictum »ie prius
—their principles arc such as may inspire us to disprove it,
T Cfflffvfuj/iojji brimful, p vi,
J a™ e g. geriwn. Muter i’tarfm, pp. *7 fl1.. tea, 124 f., and (with Kudgmi
«d i'alas^juiu dmuio- with me* ttf iiilitulatinir, AS PE *v, 51 f|.
prirt end Piito, Lttnguo#*- »v, 4ft t Cf. also TPS. 19JO, ftp, JW f.
* (Ludert, P. 971.
INDEX OF ANCIENT TECHNICAL TERMS
Tut folio winy list referi in lemm JjKuued in the mum Mf, md doe* M!
indude iiutfinccfl occurring only in fwHaenti quouiiunn.
(i> SANSKRIT
8or 8f* 85* 87- dOÂur, 85,

4t^:aP Si, *5.
nmt^ftlttp Hj hqj, gB. 92.
£0. It. 19, +£, 43- filftfgtu, 73.
mJjgVLJifc:.T.v. S|K l}h nati, 86.
unuivdru. ifr, 20. 40-4*. 51, 1+*
aflJif, $6. ndiiAw-wifAj, 75 f
unfoMH !*■ «^yd, 5^. 77-
70-75. wma-iprrta, 15-
ujrâiÂa-p 16-17, 5D-
jumtirtQMti 16.
flWfM 2Q* 71.
Jhipto. 87. ptv.kftiw id,
akstpa, K<> giT pmtnrrtflta, j*.
rfJpCHfcm (-pFAjutart), Jj, 41, *"*««*». 41, JJ. 41. «*.
dilk4pit(f r 72. 88.
&??*t f-ikflqyfllflU), 2 = pralff^. j6+
inn*tprm> *?■ pnfnniffkh 23-u-
35. /v/urrt, i4P fij+
udtfHo. 87-92. Aarm, 15-
VffHifeff* Hyp 92, bdkya {-pvaystwi), *x<
u^d^m4ui>«h i*p so, ^7t bhutfmi, 69.
urfMya. 48, makdpF&mt it-
<P?t =6, 51- 8^38.
«ko-£*Tttmi-fihthv}. u. fflierdAon, 1^-53,
"Tthyti* 61, mfiTdhiTnya, 52,
kiiHihn* 33. 5&V*, 69.
knr.fhytj' 48, 52, ;g, 6i, ™, 16, 7J^&
^4-wr 17-18, 13-=4, S7- iôtd4ri (»ct aiâvd^d).
karfiiTUi, 7+ nrittu \
Mkatahai it. *%«* I 40.
■ifeW* 14,
&&*. 58. ^t. 14, 54.
Afafrtflp 79. /qflbu %-
AruJ7H7*j>rjrnil< 11, Âu/irnydlKjr.rrcTp 63 .
Ijp 6a< WWi 46-47, 53.
fw*t 85, f^na, 13-16.
tfegmf, u- fjl*!ja-kuUu f 4.
JttefeTdj. l$L *.;ii*KiJ-’-11 rtrujfrftiiij,I, 8* 14. 15k 20.
/j^d-mrtdfcytfp 18. S7i^h
/]'/jtu-pweiVu, 18, 5J. ««W 54-
jrAttfnaifiyri, 18. $0-5=, 55- vikdrat to.
idJuvya, 61. W£jUddp 7U.
t&U, 18. $L tqp*tfjb&?d, 75.
danfd, 18. rijtitmhiirt, 7Htt
mflfa, 18, S4, LIT'JrrJ. 58.
INDEX OF ANCIENT TECHN LL' Al. TERMS
ftijnJ/ji'ttA^+J-J'jTp 6j.
ntfiiJ, ^4-35^ frjr
romun^Lina,,
rwjUi, lam^KO^Fa^], 13.
vitaxeQt 50,
lWiPjimV, 5*r ^K>-
jrttwrrtf,. 45<
»iMfflL 45-
t'jcicWdp *lp 6i*
I.îlTJJiTTW. *4, ■ rtfeiwz. 1J-1L *4t4*i Si. 57. SO- t6'
ijJnrfn, *4, J4. 4*. 7*. 75-
riwfdtfl*,. 4*-
j^jlAii^frWJteN ?l. ijifffii, 14-15. as* is.
tphoftma, 74 '
ibatev 3^
rtuu^fl I'VflWfl'h *4. 49. *«■
MTO^p 71
n'liro C'innt*). h.~ 91,
rJsrnrfUir Hi 58* 6jt &*-
*4
ivtfraMujrJt, “ Si
73 7
IciraAn fri P to-11.

Amnr-fnitfrJ, IS, J1h
u*aitfNi*aT fcy-
(aflutiW («WP^jp 6j. .'.i>iirt j . Sj,
fd) GREEK
93*
iV' 4^.
paâ yi- Wj^^Lrtp $2.
4#“*°* 3flp Ji.,
+«Wr 3&-
J+, -or ttunO* 9 a* Tdiâ (jlT-
M^ 4e*n**i $Q°
At*$apn*f 92, ^s«(**"* a^
cpjpqra^ 5J» jo.
(iM) LATIN
wdiMp 03 . Tflpt/jfl, ;i6«
otpifdlij, 36-37-
*4, i6.
?k*™. i4i *6- jwf/tJio, J-h |6.
/orfEl. 36. J7-
ffroittp 2, j6.
rfmu, j6r J7>
iiquiiuj,
j^yrm k euiL.iT tiPCTAK

AT TErfi UillMIll f OUFMt'
IIT |UJ |_; •• Iklll'Y, FfllNtEU SO THE VMVIPiim

*
IW
^ 1^4 &«w • l ^ ■(*-**+- *
— J" ‘‘'c■•::'■■■f ;
|J ^ «., • ., ,w '.j ’ - I'
'• ' * ! ■
* 1 " ,2 > * *
\
*11- - ■;£--Ti-ii-m 1
£
Cl ^m\L ARQ-IAEOLOGICAL LIBRARY,
XEW DELHI
litut t
Catalogue No. -l1 * 1 iC*; * ^ * - *15
rti 1 . , »< . j ,
Author
—
* :ic .-’4Gi. In i.;i c Lo • t L Jiv
Title-
£
hook that it fihut u bulG block'1
Ê°LoG/r^
GOVT Of INDIA
Depiirtmejnt of Andiftfolcgr
.< NEW DELHI. y
O **•
Fteav bl ip na to keep Lhe hook

clean ami ruuvuig.
IU fw V UlLlilt

Allen - Phonetics in Ancient India - OCR PDF

Uploaded by

Copyright:

Available Formats

Allen - Phonetics in Ancient India - OCR PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Allen - Phonetics in Ancient India - OCR PDF

Uploaded by

Copyright:

Available Formats

—-

Call noJ U2i4fl3S4=fiU

Lomtim Oriental Series

Groffrty Cumhttitge, PubSiihcr to tkt UmvtTtily

OENTit 1 ' i*'ilUAL

Duty £ff 7 ff'-3

OINTTUL aH<:im »*»<'*! lUAIa

MiTSTO W Cxut uBritaes

EDITIONS OF TECHNICAL WORKS REFEAJtHD TO IX

0*0. The Grammatical Achievement l

0.40. Word and Sentence 9

Ed. Sadtui Pam, JFS ii. ». 1935- PP rog

^ ETT^^t»‘4hury,jra ti. -=. t9J5, PP- tqfjS.

o.o. The Grammatical Achievement

* U Bloomfield, r.onri«tfr, v. *7* ff , ■ _

tUnu«UidW» PrakriyjtuiumuK- Wwttojt SMh

Httf commendable as the cause may be, the non-Sanskritist can

1 Sytitmt of Sanikrit Gummarr p.

0,1, The Indian Influence on Western Phonetics

The influence nf the Indian works on the phonetic views of

The "seemingly obvious’ distinction nf voiced and voiceless here

1 TPS 1^461 pPr *>4 ff~

! ** vr sttmttwd Alphabetfir rfdmirv Umnittm Fi*He« Ompkit

referred to was subsequently rcCOgnuwjd by Lepsius as 'derived

o.z. The Sources

The former are phonetic treatises relating to the pronunciation of

Mtikabh&iy&i and it h uvidvtit that in India, unlike Europe, gram¬

-a felicity which the commentator interprets as applying nut only

1 Fatttfijajri-tM.i It I- OH hin i. iL^j, Kteliwni, i, loSiiuilue, iLcunportjint

refer to each other’s opinions in a coinmendnbly objective manner.

0.3. The Sanskrit Alphabet

* ft. S, Prasad, A Pbmttif tituJ Pk'it<'lospf<d Study f t flfcnj'frtirfp '■*!»-

employed: for purpose* of textual quotation h the atandnrd Roman

0.4. The Principles of Description

' t’i M. n. Erne:.!*!*, "Hie N*«»l iTvifinnt, oi S/in^nt'. Lix^lgr, ttii

0,40. Word md Sentence

in which the commentator adds,

Some statements of the rcbtiotiship of Pada to Samhitfi, however,

1 Cf. Mbh. 1.1 i [Kidhom, L i4h

iht juncti{jfi“farrm. A convincing solution to the difficulty i& pro¬

of outside classification, in which someone has itmused himself by

.. just at a Howcr-picker may alto pick fruit, and b wood-gadwrermay

0 41, Phonetics and Phonology

This* however, b in Lhe nature uf a pious aspiration which regret¬

The working of this ahemauon is seen in vetbd forms mth as?

be moat conveniently discussed by reference to the terminology of

But * drawing unyoked1 seems hardly 10 l>c a natural metaphor for

Uan and 8Articulator; 1 In a large majority nf the articulator

0.43. Order 0/ Analysis

THE SANSKRIT ALPHABET

ComftMJLtfta rGUUaS' J'litr f-llltaf

* Abo oram ft}

The meaning of wmc of these terms is made dearer by Uvaja, who

But it will be noted that in the TP's statement of 'distinctive

* lrVj.f to-mc hw Uvn^u Oft I^P iJL iIO^gF. IS* It li ffjtcwd

of which the buccal, phafyngul and nasal may be justified the

i. 11 o. Vmcth and CmunuinLs

Maximal closure, on the other hand, provides the criterion for

1 Cf. f00i, .l-OMi/j* Fhcinrtit'i. pp_ 58 f . Forcftharnrnti, Thmri> imd

OINTTUL aH<:im »»<'! lUAIa

* U Bloomfield, r.onri«tfr, v. 7 ff , ■ _

' t’i M. n. Erne:.!!, "Hie N*«»l iTvifinnt, oi S/in^nt'. Lix^lgr, ttii

™T Jt^iKlS 5£ l. i. W iKtellwm,.. 44). Tir,tm iptern i,te ir m

* 3£TS£3i„ to mm, wS1 ^ "• -• *

I he Open ipiram* refirrJfUttiK ft in paima.P

prubkms connected with h This voiceless breathing primarily

* Wwd <Mhfcrit* uiuEtmcd in I he pfroii eitc ineruturr. ti ionic KiuStftlCi of the