Skip to main content

Full text of "A Comparative Grammar of the Modern Aryan Languages of India: The noun and ..."

See other formats

This is a digital copy of a book that was preserved for generations on library shelves before it was carefully scanned by Google as part of a project 
to make the world's books discoverable online. 

It has survived long enough for the copyright to expire and the book to enter the public domain. A public domain book is one that was never subject 
to copyright or whose legal copyright term has expired. Whether a book is in the public domain may vary country to country. Public domain books 
are our gateways to the past, representing a wealth of history, culture and knowledge that's often difficult to discover. 

Marks, notations and other marginalia present in the original volume will appear in this file - a reminder of this book's long journey from the 
publisher to a library and finally to you. 

Usage guidelines 

Google is proud to partner with libraries to digitize public domain materials and make them widely accessible. Public domain books belong to the 
public and we are merely their custodians. Nevertheless, this work is expensive, so in order to keep providing this resource, we have taken steps to 
prevent abuse by commercial parties, including placing technical restrictions on automated querying. 

We also ask that you: 

+ Make non-commercial use of the files We designed Google Book Search for use by individuals, and we request that you use these files for 
personal, non-commercial purposes. 

+ Refrain from automated querying Do not send automated queries of any sort to Google's system: If you are conducting research on machine 
translation, optical character recognition or other areas where access to a large amount of text is helpful, please contact us. We encourage the 
use of public domain materials for these purposes and may be able to help. 

+ Maintain attribution The Google "watermark" you see on each file is essential for informing people about this project and helping them find 
additional materials through Google Book Search. Please do not remove it. 

+ Keep it legal Whatever your use, remember that you are responsible for ensuring that what you are doing is legal. Do not assume that just 
because we believe a book is in the public domain for users in the United States, that the work is also in the public domain for users in other 
countries. Whether a book is still in copyright varies from country to country, and we can't offer guidance on whether any specific use of 
any specific book is allowed. Please do not assume that a book's appearance in Google Book Search means it can be used in any manner 
anywhere in the world. Copyright infringement liability can be quite severe. 

About Google Book Search 

Google's mission is to organize the world's information and to make it universally accessible and useful. Google Book Search helps readers 
discover the world's books while helping authors and publishers reach new audiences. You can search through the full text of this book on the web 

at http : //books . google . com/| 

Digitized by LjOOQ IC 

Digitized by LjOOQ IC 

Digitized by LjOOQ IC 

Digitized by VjOOQ iC 

Digitized by LjOOQ IC 













TRUBNER & CO., 57 and 59, LUDGATE HILL. 

{All Rights reserved.) 

Digitized by LjOOQ IC 




• • 

• • • « 

• • • 

• •• • 



•• •• 




• • 

• • 

• • 

• •• • 

•••• .,« 

• * • •• 
•••• ,•• 

• •• • 

••• • • 


• • 

I • • • • 

*^a ••• • 

• « 

• • 

•* r ••! 2 

Digitized by LjOOQ IC 


Two years and a half have elapsed since I sent home 
my First Volume. This Second Volume would have 
been ready sooner but for the constant interruptions to 
which I have been subjected. The pressure of official 
work has largely increased, and, at one time, when I 
was called on to take charge of the administration of 
the Province of Orissa, was so heavy as to compel me 
to lay aside all literary work for six months. 

To this cause must also be ascribed much of the dis- 
jointed and unfinished appearance of some parts of the 
work, for which I must ask the reader's indulgenca 

John Beahes. 

CuTTACK, Oeissa, Aprilf 1874. 

Digitized by LjOOQ IC 

• • • • • I 

• • • • • 
•• • •• • 

• ••• • 

• • • • 

• ••• • 

*•• • • 

••• ••• « 

• • ••••••• 

Digitized by LjOOQ iC 



Chapter I. — ^Formation op the Stem. 

§ 1. Structure of the noun 

§ 2. Suflixes .... 

§ 3. Stems in A . 

§ 4. Adjectives formed from A-stems 

§ 5. Exceptions . . . 

§ 6. Stems in -KA and -ANA 

§ 7. Stems in -A preceded by a semivowel 

§ 8. Stems in -MA 

§ 9. The KA- sufix and its lamiflcations 
§ 10. Stems in -TEA . . ' . 

§ 11. Treatment of feminine stems in -A 
§ 12. Group of stems in the palatal and labial short vowels 
§ 13. The long vowels of those organs . 
§ 14. Stems in /? . 

§ 15. Dissyllabic and consonantal suffixes 
§ 16. Stems of uncertain origin or partial application 
§ 17. Secondary stems — ^formation of abstract nouns . 
§§ 18. 19. The same — possessives and attributives 
§§ 20-22. The same, continued 

§ 2S, Stems with double suffixes . . . . 

§ 24. Formation of diminutives 






















Digitized by LjOOQ IC 



§ 25. Compound nouns ..... 123 
§ 26. Numerals — cardinals ..... 130 
§ 27. iN'umerals— ordinals ..... 142 
§ 28. Other numerals . . . . . .144 

Chapter II. — Gender. 

§ 29. l^atural and grammatical gender . . . 146 

§ 30. Use of gender in the seven languages . . 147 

§ 31. Typical terminations of the adjective . . . 150 

§32. Terminations.of the masculine . . . 151 
§ 33. Terminations of the feminine .... 159 

§ 34. Terminations of the neuter . . . . 161 

§ 35. Formation of feminines from masculines . .163 

§36. Gender of words ending in consonoDts . . 171 

§ 37. Decay of Gender — ^its cause in Bengali and Oyiya . 177 

Chapter III. — ^Declension. 

§ 38. Inflection ...... 181 

§ 39. Preparation of the stem in Ofiya and Bengali . .183 

§ 40. The same in Hindi and Panjabi. . . • 186 

§ 41. The same in Gujarati, Marathi and Sindhi . . . 187 

§ 42. Table showing terminations of the stem . . 195 

§ 43. Pormation of the plural in the uniform languages . 198 

§ 44. Formation of the plural in the multiform languages . 202 

§ 45. Origin of the plural forms . . . . 203 

§ 46. Origin and analysis of the singular oblique forms . 209 

§ 47. Oblique forms of the plural . . . . 218 

§ 48. Eemnaats of the synthetical system in other cases . 222 
§ 49. Absence of oblique and plural forms from certain languages 227 

§ 50. Internal modiflcations of the stem in Marathi . 231 

§§ 51. 52. Quasi-synthetical forms of some cases . . 233 

Digitized by VjOOQ IC 



§ 63. Adjectives 
§ 64. Numerals . 
§ 55. Case-affixes 
§ 66. The Objective 
§ 67. The Instminental 
§ 68. The Ablative 
§ 69. The Genitive . 
§ 60. The Locative 
§§ 61. 62. Postpositions 



Chapter IV.— The Pronoun. 

§ 63. Pronoun of the first person singular . 


§ 64. Plural of the same ..... 


§ 66. The second person singular and plural 


§ 66. Genitive of the two first persons 


§ 67. Pronoun of the third person 


§ 68. The Demonstrative ..... 


§ 69. The Relative ..... 


§ 70. The Correlative. ..... 


§ 71. The Interrogative ..... 


§ 72. The Indefinite ...... 


§ 73. The Reciprocal . . ... 


§ 74. Adjectival pronouns . . . . , 


§ 76. Pronominal suffixes in Sindhi 


§ 76. General scheme of the pronouns and pronominal adverbs 


§ 77. Miscellaneous pronouns .... 


§ 78. Gipsy pronouns ..... 


§§ 79. 80. Concluding remarks .... 


Digitized by LjOOQ IC 

Digitized by LjOOQIC 


Page 11, line 7> for nadd read nadd, i.e. the mark on the final a is the accent, 

not the long vowel ; and so in the six words which follow. 
Page 25, line 21, for ?]f)^|4| read gij^ TR?- 
Page 26, line 14, for f^ read flj^. 
Page 30, line 5, for M\^ read ^ff^. 
Page 39, line 6, for iTW read TH?- 
Page 53, line 10, for ^^ifT read ^NgfiU- 
Page 53, line 11, for xH^i^l read ^>^^|. 
Page 58, line 27, for jdjarU re&djdgard. 
Page 65, Ime 23, for ^^TRT read ^^ifl- 
Page 70, line 8, for f^lfRTT re&d f^T^Tirr. 
Page 70, line 14, dele "a" after "with/* 
Page 70, line 19, for ^T^HK read m^H^- 
Page 78, line 9, for B. read Pr. 

Page 81, line 3 from bottom, for derivatives read desideratives. 
Page 83, line 1, for gaurdsd read ganfdsd. 
Page 93, line 8, for ^HfT^ read VTT^* 
Page 95, line 20, for ^q% read ^S^, 
Page 100, line 19, for quarrel read quarrelsome. 
Page 113, line 18, for it read them. 

Page 122, line 7, transfer the comma from after 'mute' to after 'consonant.* 
Page 128, Hne 5, for ^TiRi^ read WiT^^. 
Page 128, line 5, for T^ifT read ij^. 
Page 128, line 20, for WN^Tfl^ read ^T ^dil^ T- 
Page 128, line 20, for <frN read Tflf. 
Page 132, line 17, in " so-much-used" dele the first hyphen. 

Digitized by LjOOQ IC 


I, line 3, for "fifty-three" read "fifty-five." 

), line 1, for iH^ ^^^^ ^1^* 

[, line 1, after 'except' insert M. 

I, line 17, for ^ra* read ^ra*. 

), line 22, (or future read past, 

I, line 7, for X^i\^ read ^flf^l^ . 

I, Une 7, for ftrfjlft read ftrfWt. 

[, line 4 from bottom, for these read those. 

5, line 23, for 1[^f^ read if^f^- 

5, line 11, for 1^1^ read iflfw I 

), line 8, for ^«l||4f read ^e|||«|. 

i, line 3 from bottom, for Wf^ i"^*! ^TEf . 

r, line 13, for ^^ read ^^. 

), line 6, for Ifft read 9|^ . 

>, line 13, for "^^ read '^^af . 

\, line 20, for ^IT^ read ^ftO- 

\, line 18, for termination read comdination, 

{, line 3 from bottom, for ^fj|4, read 4if||<. 

>, line 23, for 1|^ read Iffftj^. 

f, line 22, for 7?^ read lf^. 

>, line 4, for Tfff^ read 4|f|{S). 

3, line 11, for you read yore. 

Digitized by LjOOQ IC 






CONTENTS.—} 1. Stbucturb op the Noun.— § 2. Suppixbs.— § 3. Stbms in 
-A. — § 4. Adjectives FoBMED pbom -I-stems.— § 6. Exceptions. — § 6. Stems 
IN -NAt AND 'ANA. — § 7. Stems in -a pbeceded by a Semivovel. — § 8. 
Stems in -JTi.— § 9. The Z-d-suppix and its Ramifications.—} 10. Stems 
IN 'TRA, — § 11. Tbeatment op Feminine Stems in -^— § 12. Gboup op 
Stems in the Palatal and Labial Shobt Vowels. — § 13. The Long Vowels 
OP THOSE Gbgans. — § 14. Stbms in -8, — § 16. Dissyllabic and Consonantal 
SuppixES. — § 16. Stems op Unceetain Origin ob Partial Application. — 
§ 17. Secondaby Stems. — Fobmation op Abstract Nouns.— §§ 18, 19. — The 
Same — Possessives and Attbibutives. — §§ 20-22. The Same. — § 23. Stems 
with Double Suppixes. — § 24. Fobmation op DiMmuTivES. — § 26. Com- 
pound Nouns.—} 26. Numebals— Cabdinals. — § 27. Numebals — Obdinals. 
— § 28. Othbb Numebals. 

§ 1. Nouns are divided into two classes: those which name 
objects, and are called appellatives, or in the older phraseology 
substantives ; and those which describe the qualities of objects, 
and are called attributives, or in the older phraseology adjectives. 
Both classes will be considered in this Chapter. 

The noun as employed in Sanskrit contains three divisions — 
the root, the suffix, and the termination; the root and suffix 
taken together constitute the stem, the whole three combined 
form the perfect noun fitted for use in speaking or writing. 

Digitized. by VjOOQIC 


Thus naras, "a man," consists of three parts, nar+a+s, where 
nar is the root, a the su&x, 8 the termination ; nar + a = nara 
is the stem. 

The terminations vary with every relation or connexion of 
the noun, and their variations constitute what are called cases. 
In the modem languages the terminations of the Sanskrit noun 
no longer exist in their original form; in some of the seven 
languages i^o trace of them is perceptible in the singular, and 
only weakened and half-obliterated traces remain in the plural ; 
in none of the languages are more than slight and doubtful 
indications of their presence still observable. The variations 
of case are expressed in the modern languages by particles 
placed after the stem, just as in modern European languages 
they are expressed by particles placed before the stem. The 
English declension, fisk, offish^ to fish, stands in the same re- 
lation to the Gothic fisks, fiskis, fiska ; and French champ, de 
champ, cL champ, in the same relation to Latin campus, campu 
campo; as Hindi nar, nar kd, nar ko, does to Sanskrit naras, 
narasya, nar&ya. 

The first point then to be considered is the formation of the 
stem, and in this inquiry our attention is entirely concentrated 
on the final syllable or suffix. When treating, in the First Book, 
of Phonetic Changes, I exhibited, as far as the materials at my 
command permitted, those alterations and corruptions which, 
have operated to change the form of the root; and the next 
step is to discuss the processes by which the stem, consisting 
of the combination of root and suffix, has been built up in its 
manifold developments. 

Sanskrit grammarians divide nominal stems into two clasfies. 
They do not admit the possibility of a noun being a primitive 
word, derived from nothing else, but hold that all the words 
in their language are derived from verbal roots (called dh&tu)^ 
by additions and changes of various kinds. The first class, 
therefore, consists of those nouns derived from t\iQdh&tu directly : 

Digitized by VjOOQ IC 


these they call kridanta ; the second, of those derived immedi- 
ately from other nouns : these they name taddhita. I shall use 
for kridanta the term " primary stem " ; for taddhita^ *' secondary 

§ 2. Derivation is effected by the addition to the dhdtu, or 
root, of certain syllables called pratyaya, or suffix* These suffixes 
in many cases not merely add a vowel or syllable to the root, 
but also effect some change in the root itself, either by lengthen- 
ing the vowel, eliding a final consonant, or in some other way. 
In order to indicate the effect the suffix has on the root, the 
Indian grammarians have prefixed or added to each suffix 
certain letters, which serve as a sort of memoria technica. Thus 
V[XM " cooking," would be described ds xj^H- isn^, that is, derived 
from the root Tl^ " to cook" by the suffix ^fT; the effect of this 
suffix is principally to add a final H to the root, but it also 
lengthens the root-vowel, and this effect is indicated by the 
letter "sj^; it moreover changes a final palatal into a guttural, 
and this is denoted by the ^. The number of pratyayas is 
very greats the grammarians having multiplied them without 
stint, in order to give expression to. the large mass of minute 
distinctions with which they have encumbered their writings. 
As, however, we are not discussing Sanskrit grammar, but only 
the simpler grammar of its modem descendants, it will not be 
necessary to consider all these suffixes in detail. They may be 
conveniently thrown into groups according to the vowel or 
syllables which they actually add to the root, the technical 
letters being referred to only when the distinction which they 
mark is of importance in its bearing on the forms of the spoken 
languages. Among the kridantas or primary stems the Indian 
grammarians include participles and verbal forms of all kinds, 
as well as abstract nouns. It will, however, be more in accord^ 
ance with the principles of linguistic science to reserve the 
consideration of participles ^d all verbal formations till the 

Digitized by VjOOQ IC 


chapters on the Verb are reached, and to discuss in this place 
merely those sufiixes which form substantives and adjectives. 

Suffixes are monosyllabic or dissyllabic. Out of the vast mass 
of suffixes mentioned by the Sanskrit grammarians, many may 
be dismissed as unconnected with the present subject, inasmuch 
as the stems formed by them were so formed before the rise 
of the modem languages, and the question of their construction 
is a matter not of modern, but of ancient, grammar. Our seven 
languages took over these words ready made, and, as far as we 
are concerned therefore, they may be regarded as primitive 
words. Only such suffixes will here be introduced as have left 
traces in the speech of the present times ; and if it be necessary 
to speak of certain of them which are of purely Sanskrit use, 
it is merely because of their bearing on phonetic or structural 
processes of recent introduction. 

§ 3. Stems in -a. The nouns of this class are of all three 
genders, and make in the nominative case of Skr., masc. -as, 
fem. -d, neut. -am. The « of the masc. nom., however, is not 
permanent; before a word beginning with a sonant letter, it, 
together with the a of the stem, changes into o. Inasmuch as 
the sonant letters in Sanskrit are more numerous than the 
surds, the form of the nominative case in -o was much more 
frequently heard than any other ; and the vulgar, who are not 
careful of minute grammatical distinctions, appear even in the 
earliest times to have used the termination in -o to the ex- 
clusion of all others. Thus Vararuchi (v. 1) gives it as a 
general rule that o is substituted for sw (= as), the technical 
name of the nom. sing, ending, in all words whose base ends 
in a, as vachchho, tasako^ purtso, for vrikshaSy rrishabhas, puru- 
shas. He considers that the a oi. the suffix has been elided, 
though it is more probable that the whole termination ^as 
changed into -o through the intermediate form -ah, the labial 
vowel owing its origin to the involuntary contraction of the 


Digitized ^y Google 


lips during the passing of the soft breathing indicated by the 
visarga. In Magadhi Prakrit the nom. is sometimes formed 
in 'By and in Apabhransa in -w. 

In more recent times two separate sets of stems developed 
themselyes out of this a-stem. The first probably ended in u 
in all the languages down to about the fourteenth century; 
since then it has ended in -a in all but Sindhi, which still re- 
tains the termination in -u. In old Hindi and Marathi the 
nominative in w is distinctly traceable, and in the former is 
even lengthened to u at the end of a verse where the metre 
requires a long syllable. This is especially the case in Tulsi 
Das's Eamayan, a work which exhibits a wide range of popular 
forms and phrases. Thus we have ^0%^> ^"ft^? 'WT^j M-cti^^ 
at the ends of rhyming lines for i(0^> wV^ in modem H. 
;mO^ ^^V^, iarir^ Mr. The second ended in o, which in the 
broad pronunciation of old Hindi sounds au, though the form 
in is often used by the earlier poets ; and there is no dis- 
tinction between the two : it is merely a matter of manuscripts, 
some using one, some the other form. This o softened sub- 
sequently still further into d, in all but G. and S., which still 
retain o. 

Concerning the cause of the parallel existence of the two 
forms there has hitherto been much doubt. Dr. Trumpp con- 
siders it a mere matter of accident, and thinks the presence of 
one form or the other is due to habit and daily usage.^ Dr. 
Hoemle, however, holds, and with justice, that this reasoning 
is insufficient, and would ascribe all words which exhibit the 
d=io termination to the forms of nouns with an added ^, which 
are extremely common in Prakrit. To this opinion I in the 
main subscribe. I think myself fortunate in this second volume 

^ " Eine Begel scheint bei dieser Separation nicht vorgewaltet zu haben, wenigstens 
habe icli bis jetzt noch keine entdeoken konnen, sondem der tagiaglicbe Gebrauch. 
scheint dch fiir die eine oder andere Endnng entsohieden za haben." — Zeitsoh. d. 
D.M.G. vol. xvi. p. 131. 

Digitized by LjOOQ IC 


to have hints from other and better scholars than myself, which 
Were wanting when the First Volume was written. 

There is, however, one other possible way of accounting for 
the peculiarity. This is the influence of the accent.* It is 
vigorously denied by some that the Sanskrit accent has per- 
sisted irntil modem times ; but it may be urged that the idea 
is a new one, not altogether without support in facts, nor in 
any way d priori impossible, and it is not quite fair to dismiss 
it contemptuously without due consideration.^ The influence of 
the accent is naturally confined to words of the early Tadbhava 
class, namely to words which have -come down uninterruptedly 
fi:om those times when Sanskrit was spoken, and whose form 
depends on the ear, not on the eye. This is all that is tdaimed 
for it: it is not pretended that the Sanskrit accent has been 
preserved in words resuscitated from the written language by 
learned men, centuries after it had ceased to be spoken. 

With these reservations, the reader is invited to examine the 
lists hereinafter given, given not in a spirit of dogmatism, but 
merely because I cannot satisfy myself in rejecting accentual 
influence in them, till I am met by something more convincing 
than sneers ; though in the desire to arrive at the truth, which 
is the only object which these researches can possibly have for 
one in my position, I shall be ready to throw down my weapons 
and acknowledge myself beaten directly I feel myself fairly 

In any case, whether the influence of the accent be admitted 
or rejected, the collection of words now given will be useful as 

» See Vol. I. Ch. I. § 6, pp. 17 ff. 

' Unfortunately little is known of the spoken accent in Skr. In the following 
pages, the rules laid down in Bopp's Vergleichendes Accentuationssystem hare been my 
guide ; but perhaps since that was written advances haye been made by European 
writers. These, in my isolated position in the depths of an Indian proyince, I liave no 
means of obtaining. I am, howeyer, quite conscious of the unfavourable effect isolation 
has on this, and all other parts of my work, and only wish my critics would bear it 
in mind sometimes. 

Digitized by VjOOQ IC 


afiPording instances of the development of the -a suffix^ whatever 
be thought of the cause of it. Those who hold that the termi- 
nation d=o is due to the Prakrit (and at times also Sanskrit) 
suffix ^ must explain to us why this '% producing d = o, has 
been so capriciously employed, why, in fact, Skr. lit^ "lip," 
should not have taken a ^ and become Pr. 'iTi j'^* H. ^HZl* 
just as much as, according to them, "V^ " ^S?/' ^^<>^^ ^ ^^^ 
became ^U:^4> ^«i2sVi> ^TOT* If, as is asserted, all nouns 
might and did take this ^, why do not all nouns of this stem 
end in the modems in d = o, why do some end iaa = u? 

(1). Barytones, — ^The Sanskrit words in the following list are 
all accented on the penultimate; they are all primary stems 
formed by the addition to the verbal root of one or other of 
the simple suffixes, such as ^l^> VC> V[\i and the like : they 
are all therefore fair cases in point. 

Skr. ^ft^ "lip," Pr. ^. H. ^, P. fte, G. ftZ, ^PtJ, M. ^, 

Skr. ;«li| "ear," Pr, ^|^, H. ^Hlf, P. ^m» G. M. B. O. ^^nir, S. ^IJ. 

Skr. 1|T^ "wood," H. 1|T7, and so in all, but S. WT?* 

Skr. IK^ "armpit," Pr. 11^, H. nf^, P. H^, G. M. B. O. HT^, 

Skr. -9^ «ko8," H. ;ajprtr, p. 'itf . the rest IJt^, S. ijftir. 
Skr. iiru" trouble," H.^rtT, Rl^OT, M.ft(%^,ftWrtr, S.^Sl^. 
Skr. ^nif " moment," H. ^Qlf , ^if , ft^, the rest ?^, S. fl^llf (fem.). 
Skr. inl "womb," H. irW» TW» P. w/. and 1(9S{, M. G. ^rW» S. 

Skr. ?rn? " viUage," H. ^ft^, G. IfRT, M. ift^, B. O. m, S. iTT^, 

Skr. ?rnr "net," H. ?rnr, so in all, but G. HT^, WTt* »• 'H^- 
Skr, IfTf "tone," H. ^IPT* ^ in a". G. ^fffUr, S. TfT^- 
Skr. ?Nl "oil," H. ?Nr, so in aU, S. %^. 
^ Skr. -^ "tooth," H. ^, so in all, P. ^, 8. nJJ, 

Digitized by LjOOQ IC 


Skr. f^ "day,'' H. f^, so in aU, S. t%m. 

Skr.^lJ "country," H. ?9, in the rest ^^, 8. ^^. 

Skr. vnif "noose," H. IJTO, IRt^, G. and the rest XTftj, P. Vn%» S. 

Skr. T^^ "ploughshare," H. XliPif , and so in all, but P. XC35> 3. lil^. 
Skr. J(\it "road," H. iHT* TPTf in tl»e res* ^RW> S« T^- 
Skr. ^it^ " world," H. ^ft^ (" peopje"), in the rest ^^1, S. ^I^.^t^ . 
Skr. T^ "bamboo," H. '^jm, ^tH <uid so in all, S. '^Vt^* 
Skr. ^7 "banian-tree," H. WW* so in all, M. O. ^7, S. ^. 
Skr. f^ "poison," H. f^, B. fB^, M. G. id. and f^TO, P. ftl^, 
t^> f^» 8. fi|^ and f^W- 

Instances might be multiplied to a large extent, but the 
above will suffice to show that the rule holds good in a vast 
majority of instances, Sindhi having uniformly u, the other 
languages a, though Gujarati now and then retains u. This 
final short a is not pronounced at all in Hindi, Panjabi, Ma- 
rathi, or Gujarati, and very slightly in Bengali and Oriya. 
We hear jdl, tdn^ tel^ d&nt, and so on, not jAla ; words oi this 
class may, for practical purposes, he regarded as ending in con- 
sonants, except in M. and G., where it is necessary, as will be 
shown hereafter, to bear in mind the existence of the final mute 
a, in order to efiect the necessary changes of the base in the 
oblique cases of the singular. 

(2). Oxytonea. — The list which follows comprises, like that 
which precedes, none but primary stems formed by simple 
suffixes, as 'W^i ^Bf^ and the like ; but the words which it con- 
tains are all accented on the last syllable, and accordingly it 
will be seen that they all end in the modem languages in 

Skr. l^igr "egg," H. ^i^, P. tU, B. ^rtljfT* O. lU, S. ''VX^y M. and 
G. are exceptions, M. ^%, G. ^#. 

Skr. list^ " worm," H. i£^^|, end so in all, G. and S. ^i^^, , 

Digitized by VjOOQ IC 


Skr. ^ '< knife," H. ^TT> and so in aU, G. and S. ^1^. (In aU the 
fern. S^ is more common.) 

Skr. %S "slave," H. %n5T. %Wr» so in all. G. and S. ^^ft. 

Skr. ^ "dust," H. Wf ("powder"), so in aD, G. and S, ^jft, M. 
and B. also ^T. 

Skr. ^ " dust," H. ''l^ ("lime"), P. id., M. ^prr» "^J^* B. O. ^IfT, 
B. also W9\, G. and S. ^^t ^J^' 

Skr. 1^ "candle," H. f^, O. B. id., P. ^V^, f^[WT, M. f^, G. 

^Wr, s. fwitt. 

Skr. ^^eni "flag," H. V^, P. id., Q. ^[^. 

Skr. «rrW " stream," H. ^irWT> and so in all, G. 9||cbt> and WC^» S. 


Skr. liirr "snake's hood," H. Jfl^, M. B. O. id., Q. Ifl^, S. Iff^ (fem.). 
Skr. ^fcW "young," H. ipH, l^T, and so in all, B. ^T^, G, l^t, 

Skr. H9 "spear," H. ^TT^T. P. M. O. id., Q. and S. ^T^. 
Skr. IJSrnfr "brother-in-law," H. mmXy P. M. ^Idbf* B. O. JfWU, G. 
and S. corral') G. also Hldot- 

Skr. l^S^ " shoulder," H. i|^|T, P. 70^^, M. ^t^J, O. B. nf^, G. 

Skr. Wn "post," H. ^dn, P. B. O. id. 
Skr. Ijjsr "mouse," H. ^n» P« B. irf., O. ^ffT. 
. Skr. ^ "rain," H. ^^7^, W^^, O. ^J^, B. M. id., G. ^. 

The Hindi in this, as in so many other instances, vindicates 
its right to be considered the leading language of the group, by 
the fidelity with which it adheres to rule. The other languages 
are less^ faithful to the long vowel representing the oxytone, than 
they are to the short vowel derived from the barytone stems. 
Marathi especially diverges in this respect, but the divergence 
is probably due, as will be seen hereafter, to its practice in 
modifying the final vowel of the stem in the oblique cases. 

Digitized by LjOOQ IC 


Bopp's rule^ for stems formed by the addition of simple a 
is that they carry the accent on the root-syllable in most cases. 
He excepts, however, a class of abstract stems fornaed from roots 
in t and i, which allow the accent to sink d6wn to the final 
syllable. They are as follows : 

Skr. f^ ^^ << destructioD," H. igpj, P. %, S. ifNt* O. %, M. ^ . 
Not used ia B. and O. except as Tatsama. 

Skr. "^ ^^ "heap.'* Ia compounds H. ^^, M. ^pS[, ^^^, etc. 

Skr. t% IRI " victory." In all ^IRI or % , 

Skr. 1^ ^9^ " smiling.'* Not in use. 

Skr. f^ ^pf "horse.'' yf^ and ^ in the old poets; often with an 
additional syllable^^^f^ in Chand. 

Skr. "ift ^ini " buying," In compounds H. f^^ (f^niW)> B. O. id,, 

Skr. Ht ^^ "fear," H. ^^, B. O. lU, P. 5|, S. Hi; and ^^, M. 

Skr. ^ ^nr ''abode." In the compounds with ^^R|, as below. 

Skr. ^^rr^RT "temple," H. ^lETW, P. ^^T, S.*^f%^, G. ^^35. 
M. ^^S35, O. lU, B, ^^^^r. 

Skr. ^^XT^RI "father-in-law's house," H. ^X;!^, M. ^ihc. 

Skr. f^ ^51^ " refuge." In the compound ^i^M, H. ^|^<^|, P. 
tU, M. "m^f B. O. id., G. ^Jiroift. S. ^ir9T» ^Wft^t' 

In all these words the inherent weakness of the '^ has led to 
the corruption of the two syllables of the Skr. into one in such 
words as dird, devdldy while in others it has been practically 
softened into a diphthong in combination with the preceding 
consonant. It may also be conjectured that though the learned 
accentuated the last syllable of stems of this small class, the 
masses did not at any time observe this distinction, but treated 

^ Vergl. AccentoationssyBtem, § 115. 

Digitized by VjOOQ IC 



them as barytones, and the final vowel has therefore not been 

There is also a rule of a somewhat intricate character, which 
declares that stems of this form, when simple appellatives, which, 
according to their original meaning, are nouns of agency, bear 
the accent on the last syllable. Under this head are such words as 

nadd, " river," or the sounder (root If^ " to sound "). 
plavil, " sliip," or the stoimmer (root ^ " to float"). 
chord, « thief," or the stealer (root ^ " to steal"). 
hard, « liand," or the maker (root ^ " to do"). 
maghdy " cloud," or the wetter (root fiTf " mingere"). 
devdj "** god," or the shiner (root fe'f " to shine"). 

It is obvious, however, that nouns of this class may be formed 
to any extent by a little exercise of the imagination. Every 
noun in fact formed by the suffixes (ich, ghan, or op and the 
like, which merely add a vowel to the root, might be classed as 
a Tiomen affentis, and expected to be an oxytone. There is no 
reason why gdrhha^ "womb," should not be considered as a 
noun of the agent, meaning the container (root lfi% "to hold"), 
or kdrnaf "the ear," as the piercer (root W^"to pierce"), or 
visha^ "poison," as the pervader, from its action in stealthily 
creeping into the blood (root f^ "to penetrate"); and as a 
matter of fact the grammarians do so regard them all. The 
rule appears to be too subtle for general practical use, and the 
following list of oxytones of this class, mostly formed with ^I^t 
which is said specially to form agents, will show that in a 
majority of instances the modem languages have not regarded 
these words as oxytones. 

Skr. ^[^ '* hand," H. ^, and so in aU, S. iP[. 

Skr. IT^ "river," H. if^, M. trf., perhaps 8. If^ "tube," but also ira. 

Skr. ^ "cloud," H. ^, and so in all, S. ^. 

Skr. ^ "thief," H. ift?:, and so in aU, S. '^. 

Digitized by LjOOQ IC 


Skr. ^^ **god,'' H.^^, ^^, so in aU, S. %^ and ^^ . (In Muham- 
madan Sindh it means <' a devil.") 

Skr. ^ ''gadfly" (the biter), H. ^, ^, and so in all, S. ^ '<the 
bite of an insect or reptile." 

Skr. Tf^ ** bridegroom," (he who chooses the wife), H. "if^, P. B. O. id., 

§ 4. Under this rule is included a large number of adjectives 
which are oxytone in Sanskrit, and end with long d or o in the 
modems. It may, however, be observed, that the majority of 
adjoctives from a-stems end in 4 or o in the masculines, except 
in B. and 0., where the habitual neglect of quantity h^s led to 
the final vowel being shortened. The rule therefore hardly 
covers all the examples, and is perhaps not meant to do so, as 
Bopp merely includes adjectives with the meaning of the present 
participle. We may suppose that the rule, once established for 
this class of adjectives, gradually extended itself to the others 
also, on account of the facility which the variety of terminations 
so obtained afforded for making the distinctions of gender. In 
those languages which pay little or no attention to gender, as 
the Bengali and Oriya, the distinction was not needed, and we 
hear in 0. such phrases as bord bhdi, " eldest brother," and bord 
bhauni^ ** eldest sister.'* 

Examples of adjectives — (p) indicates oxytone, (V) barytone 
adjectives : 

Skr. ^ir^ (*) "clear, good," H. ^^f^^T, P. «</., G. 11^, S. ^"iraft 
("wfiite"), B. O.^IHBf. 

Skr. lap^ (o) "bUnd," H. ^\3rT, P. HffT, G. ^«hsia5t, M. ^*V35T. 

B. o. ^iro, s. ^\fr. 

Skr. ^ {b) "half," H. ^SIHTT. P. ^VIT, G. ^W^, ^T^, S. ^/ 

M. n^, B. yisa^gn, o. ii^, ^} 

1 Used as a substantiye. 

Digitized by LjOOQ IC 


Skn ^Tff W "high," H. ^1^, P. y^^, 8. ^nfr. G. ^T^, M. ^, 
B. 0. id. 
8kr. ^rnir (b) "Wack," H. mWT, P. M. 1|T35T, G. 1T3§t, B. ^|TW, 

Skr. urnr (o) ** one-eyed," H. W^f P. M. IHUTT* <>• 8- 'ITOt^ B. 
inrT> O. WWT- 

Skr. iM^ (o) "deep," H. irff^, ^f<|, and bo in all, O. J|ff<. 
The Tatsama JJ^Hl. is also in use in all. 

Skr. KJJg (o) "intense,"* H. ITreX. P- Wm O. IfTOt. M. B. O. IHM- 

Skr. 1^ (*) "pale," H. ^f|TT, P. icf., S. G. lifV^t. M. (dimin.) ^ZT, 

O.lftTTjB. iftT- 

Skr. ^pf (o) "dense,* H. ^iprr* P- ^nilT* ®' G* ^^ (many, much), 
M. ysr^, O. B. IR . 

Skr. ^rfOT (o) "hard," H. ^rf^, O. B. tU, P. l^S^, •Hf, G. HRPHr. 
M. ^iThlT' G- B. 4n&^> S. WZ^' 

Skr. ^f^ W "nglit'' (hand), H. ^ff^, O. ^fTHf* B. TTITT. 

Skr. ^\rC ((^) "deaf," H. ^f^, ^f^, M. lU, G. i|^, 8. ^tit, 
B. ^f^, O. ^rffTT. 

Skr. ^TER (o) "dry," H. ^J^T, P. ^|WT, G. ^, m'. ^JUT, 8. ^,^WT,B. ^i|T. 

There are very many other instances. Hindi preserves the 
long vowel which naturally arises from oxytones, and lengthens 
the vowel in many cases in barytones ; in a few Tatsamas like 
*lfi*l^ the long vowel is not taken. The accent is thus virtually 
neglected, and in this respect Hindi is followed generally by 
P. S. and G. Marathi halts between two opinions, sometimes 
taking the short vowel, sometimes the long ; the latter, however, 
appears to be the more common of the two, as might be expected 
in a gender-ridden language like this. More remarks about the 
adjectives will be found scattered amongst the various stems. 

1 Deep (colour), strong (iniiision of a drag), thick (cloth), etc. 

Digitized by LjOOQ IC 


§ 5. The influence of the accent is only claimed for the early 
Tadbhava class, and it is natural therefore that there should be 
many instances in which it does not apply. Seeing that there 
are among the late Tadbhavas many which approach closely in 
point of antiquity to the latest of the early Tadbhavas, the line 
is hard to draw. We merely give examples therefore of these 
exceptions, and wish for explanation as to why, if the stems in 
d'O are due to the addition of 'H, that suffix has not been used 
here also. Many of the words now given appear, from their 
phonetic structure, to be early Tadbhavas, as for instance ^^, 
which exhibits the Prakrit process of assimilation from ^^rq" 
into ^^» and the modem further elimination of the first 
element of the nexus and compensatory lengthening of the pre- 
ceding vowel. 

(1). Of Sanskrit oxytones which take the short or barytone 
ending in the modems, the following are some of the more 
prominent examples : 

Skr. f^^ <<lion," H. '^^, P. id., the rest spell f^ but pronounce singh, 
S. both f^^and ^fty. 

Skr. ^B^ "bear," H. xtW> ^' «^» M. '^, P. f?:^, S. f^, O. (rare) 

Skr. ITf "house," H. ^TT, and ro in all, S. ^EPI, O.-H. 5}f , M. trf. 
Skr. Zf^ "year," H. ^?;^, and so in all, but S. ^rf. 
Skr. -qif "leaf," H. Vil^ (betel), and so in all, 8. JU^. 
iSkr. m^ "nature," H. ^m, T^J^t, and so in aU (^T^), 
Skr. ^^m " slave," H. ^TO, and so in all, S. ^fTO. 
Skr. 1^ "milk," H. ^^, and so in all, S. ^ft^. 

These words are all very common words in constant daily use, 
and as such should, according to the ^ theory, have taken 
that suffix in Prakrit, and consequently end in 4 -o in the 
modems. Some of them have an alternative form in 4, as "mj, 
which has ViffJ "a leaf;" ^, which has 4KH\ and q<m in 

Digitized by LjOOQ IC 


the sense of " rain ; " the meanings of mf and ^T^ are secondary 
or restricted, as compared with VlftT and WWT, and they are 
perhaps late Tadbhayas in the former, but early Tadbhavas in 
the latter sense. 

(2). Barytones in Sanskrit, but taking the long vowel in the 
modems, are the following : 

Skr. ^ftf " iron." H. ^ft^T . P. B. id., but M. G. 0. ^tf (O. also ^^) , 

Skr. ^^ "gold," H, ^ftfT, P. B. O. id., but M. ^pf^, G. 8. ^pflf. 

I cannot find many undoubted examples of this class, and 
even those that do occur do not run through the whole seven 
languages, which seem as a general habit to be more faithful to 
the barytone accent, especially in substantives, than they are 
to the oxytone. Or, if we accept the ^ theory, to be derived from 
words which did not take the ^ termination, more frequently 
than from those which did. 

§ 6. Stems in -na and ^ana. The former of these is in use 
only in a very small class of words, all of which, with one 
exception, are oxytone in Sanskrit. The words are — 

f/ajnd from f/aj\ " to sacrifice,*' 
prasnd „ prachh, " to ask." 
yatnd „ yat, "to strive." 
viSnd „ vtchh, " to shine." 

and fem. y&chna ,y ydch, " to ask." 
truhna „ trish, " to thirst." 
The exception is swdpna from shwap, "to sleep," which, 
however, the grammarians derive by the suffix f^. while the 
others are formed by T^- 

In the modem languages these words, many of which are 
much corrupted, take the under-mentioned forms : 

Skr. ^T^ "sacrifice," Old-H. WaPf, ^RT, ^W, H. ^10^, P. ^pf, M. 

Digitized by LjOOQ IC 


Skr. 1(1^ "effort," H. ^snnr » P. M. G. tU, 8. ^TTf^. 

Skr. IHT, ''question." Not in use, except as a rare Tatsama. 

Skr. f^ "shining, light," H. f^R (« dawn "), ftfiRn; or fifif- 
^TT» ^' f*lf^O^» S. fiprn^' f^fn^T • (The M. and S. are compounds, 
of which the first part represents "pf^.) 

Skr. ;gTT *'dream," Old-H. ^gXTif , ^ift^^, P. HUT, ^HW^, S. Hirft, 

In these words the accent is entirely neglectiod ; strangely the 
only word in which Panjahi and Sindhi exhibit oxytone termi- 
nations is precisely that one which is barytone in Sanskrit. It 
will be more convenient to take the feminines of this and all 
the other -a stems together later on. 

The suffix -ana is one of the most commonly used of all. It 
is of all three genders, and in the large majority of instances is 
barytone, carrying the accent on the root-syllable. A few cases 
exist in which the first syllable of the suffix bears the accent, 
and some in which it is oxytone, but the rule is that they 
should be barytone. 

The masculine contains simple appellatives whose original 
meaning was that of the agent. They are not much used in 
the modenjL speech, and when they are, occur as Tatsamas, or 
as very slightly changed Tadbhavas. Being mostly simple 
words with strong consonants, they offer no opportunity for * 
phonetic changes, and may therefore, in spite of their identity 
with the Skr. form, be ancient words. I give a short list, as 
there is not much to be learnt from them, beyond the fact, im- 
portant to the present portion of our inquiry, that they take in 
every case the barytone form, 

Skr. ifv^if "son" (the delighter), H. if^Ty and so in all. Chiefly used 
in poetry. 

Skr. ^TRPI "singer,*' H. J||i(«|, and so in all. Chiefly used in poetry. 
Skr. ?iil|| "mirror" (the flatterer), H. ^^, and so in all, except S. 

Digitized by VjOOQ IC 


Skr. ?^ "fire'' (the burner), H. ^ipf > and so in all.- 
Skr. ^^if << tooth " (the biter), H. ^^pT, and so in all, except S. Poetic 
Skr. Tf^qf "raft" (the crosser), H. TR^f » and so in all, except 8. Poetic. 
Skr. f%^l!I "ray" (the shiner), H» fi|^^, and so in all, S. ftff^ftl. 
Skr. ^n^'^ "boiled rice" (the moistener), H. i^^'^* <uid so in all, 
except S. Poetic. 

Three adjectives, oxytone : 
Skr. ^Bt^if "angry," H. ^fflOT (rare). 

Skr. '^MUf "angry," H. ^(|l| (rare), S. ^f^CQV comes from another 

Skr. ^V|i| "obstructive," H. "'ft^if (rare). 

Far more widely used, and in every sense more important, is 
the neuter form. Two classes must be hfere distinguished : the 
first, simple appellatives, or names of objects or actions; the 
second, abstract nouns which have the meaning of the perform- 
ance of an action, or the being in a state. The latter are in 
fact the infinitives of verbs. 

The first class is always barytone, both in Sanskrit and the 

Skr. ^>J*«f» "courtyard," H. '^A'Pf > ^iTf> P- B. 0. id,, G. ^it^Rir. 

Skr. ^rr^rif "gold," H. ^(x|^, and so in all but S. 

Skr. ^^if " sandal- wood," H. ^Tf » in all, B. ^?^« 

Skr. T^t^R "life," H. ^fNif , so in all, P. ^, S. ^^. 

Skr. IfPf "song," H. in^» so ^a all. 

Skr. ^^TT^ "bath," H. ^^TT^, ^TO«fT«r, P. B. 0. id., G. i^fm and iffTHT. 

Skr. "iRTf "eye," H. if^if, %f , so in all, G.^^if (masc. pi,), S. Sflff. 

Skr. m^i^^ "memory," H. ?Tf9T^^, so in all. 

8kr. ^qr*f "rolling pin," H. ^W[^$ so in nil, 0. %W11|T« 

VOL. n. 2 

Digitized by LjOOQ IC 


The second class consists of words which, though simple nouns, 
and as such regularly declined, perform neyertheless in four out 
of the seven languages the functions of an infinitive. 

Thus from Skr. ^R^ "the act of doing," H. W^^ " to do," P. ilRTfT • 

It is unnecessary to multiply instances, as every verb in each 
of these languages forms its infinitive in this way, absolutely 
without exception. Iji Bengali and Oriya the form exists, but 
simply as an abstract noun, almost 'entirely restricted to gramr 
matical and other works. The real every-day infinitives of 
those languages are formed in a diflferent way. Gujarati also 
wants this form. 

In Sanskrit the formations of this class are regularly barytone, 
and accentuate the root. In M. and S. they are also barytone, 
but in.H. and P. they take the long vowel. It does not appear, 
however, that this is a regular exception. In old Hindi the 
infinitive of this class — I say "of this class" because there is 
another infinitive in ^ — ends always in the short vowel. For 
example, Chand (a.i). 1200) uses the phrases: W^ fT'T ^ " (in 
order) to seize the victory," ^i>jR t^^^fC "having plotted to 
bind," ^ ^^ ^iRi«i ^^nr " a warrior terrible in the tug of 
war." In the modem idiom we should have ^% ^» «ii^«li» 
and ^>««i*^ respectively. This consonantally terminated form,* 
or rather (as it is pronounced in poetry) this form ending in 
short a, is used by the poets as far down as the middle of the 
seventeenth century. Gambhir Rai {circa 1650) has ^it^^ 
IJfT Vi^ '* (that) no one might be able to touch." Tulsi Das 
(1600) employs this form regularly in his Eamayan, e.g. T^Wf^ 
Jn 'TC'R Wr5t "the keepers then began to forbid (them)," 
Sundara Kdnd. 60, 15. From BhaktamsLla (1630) : iTt'R ifif 
ISTRft" "he ran to beg." I have pnly picked out an instance or 
two at random, as the practice is universal. 

It will be more appropriate to discuss this matter at full 

Digitized by VjOOQ IC 


length under the Verb, ^nd I therefore content myself here 
with saying that the long vowel appears not to be in any way 
whatevpr due to the accent, but to have arisen from the influence 
of the anusw&ra (see Vol. I. § 65), by which karanam became 
first qit,^, then «IIWtj *<*ll, and lastly kamA. In M. the 
infinitive is still neuter, and it is masculine in the H. P. and S. 
merely because those languages have lost the neuter, and only 
retain the other two genders.^ 

This infinitive is also in certain phrases used as a simple 
noun, as in Hindi %'fT\in or %f ^T, literally "taking and 
giving,'* but idiomatically expressing the state of one's loan 
account with a banker or mcmey-lender ; ^TT hW*1T "to give 
and to receive,^ idiomatically the outstanding assets and lia- 
bilities of a mercantile concern. In Sindhi, words of this class 
sometimes take the form in o, as 

r^i|(qt "a debt.'^f^J^fl^^ra^ or f%llljt ^rpjt "debts and credits" 
(but f^nnir " to give **). 

^IXuH or *nC^ "embroidery'* (lit. "filling up,'*^ Hf^ "to fill"). 
4iri»Un ' "betrothal" (Tfl^ "to ask for"). 

In these cases the T inserted before the termination is 
inorganic, and merely due to the preference of Sindhi for that 
sound (Vol. I. § 32). In Hindi and the other languages, nouns 
of this kind are more generally feminine, and so also in Sindhi ; 
in Marathi and Panjabi also the fern, form is more usual. Thus 

Sindhi ^^Rlft "saying," verb ^^^ "to say." 
>» f^5Wt "going," verb ^^" to go.** 
And with short i — 

Sindhi fi(^^ "swallowing," verb f^Ff^ "swallow." 

1 Hoemle in a recent article in Jonm. As. Soc. Bengal, 1873, vol. xlii. p. 66, 
contends for the derivation of these infiuitiyes from Skr. karaniyaniy etc., oyerlooking 
the intermediate forms of the medisBTal poets, which militate strongly against his 

Digitized by LjOOQ IC 


When added to causal verbs, this suffix, which then takes 
also the fonn ^Rrfl' ani, is used to express the idea of the cost of 
doing, or wages for doing, an act ; as 

Sindbi <^U||l|f1' "porterage," verb ^Uill[^U|' "to cause to carry.** 
>» TrTPTt "grazing fees," verb '^TCW "to cause to graze.** 
»9 ^^ipLUfl " cost of wasbing," verb ^^|<^ur " to cause to wash.** 

Hindi has. words of this class, as 4|J|^ "betrothal,** H^^ 
"woof,** m^^ "lading,** but more frequently as consonantally 
terminated masculines, as *IT11I mImUI " feeding and protecting," 
"maintenance.** Marathi is perhaps the richest of all the 
languages in words of this formation. It is, however, as will 
be seen from the following examples, very capricious in its use 
of such words, sometimes using a masc. in TITT^ at others fem. in 
IffV or W, or neut. in W or^. 

Verb ^^^ " to stop,*' ^vfUJT (^0* a slip of bamboo to fasten a door witb. 
^K^l[ft (/.) I a slip of bamboo to strengthen the 
^vgill (ft.) } edge of a winnowing-basket. 
Verb lliflii " to bind," l|S^r^ (/.), ligature. 
Verb ^KTZ^ " to cut," ^KTZ^ (/•)> a cutting or reaping, Le, the quantit)r 

cut at one time. 
Verb WT^^ " to take out,** «iiT«^ (/.)i removal of crops from a field. 
„ ,« qii^ (n.), a pitcher to draw water from a 

well with. 
Verb ^1^ "to dig," ^Pr^ (/.)» a digging. 

Verb yaltdoii " ^ stir,*' ^BS^l (n.), a sifting, the quantity sifted at one 

„ „ MtcbUII (fn.) ) a stick for stirring grain while it 

m.) ) a stick i 
».) ) is bell 

„ „ M^cbltf (^0 ) is being parched. 

^ 7. The next class of -a stems is that in which the final 
letter is preceded by a semivowel, as ya, ra, la, va. Of these 

Digitized by VjOOQ IC 


ya 18 chiefly used in Taddhitas or secondary stems ; there are 
only a few primaries, which, being feminine, will be treated 

(1). Stems in ra are of two kinds, those in which the suffix 
is joined immediately to the root, and those where a joining 
vowel intervenes. Both kinds are fcnr the most part oxytone 
in Sanskrit, and the accent is lost by the rejection of the 
last letter of the nexus ; the whole suffix thus ordinarily dis- 

(a). Examples of words where the suffix is joined immediately 
to the root — 

Skr. IRf W « doad,'' M. Hpf, ^^WT, a Hpf, S. Hg. In the rest 

Skr. 18ITO W "mango/' H. ^im, ^, P. «., B. O. ^RTW, ^RW, 
G. ^iNt (the tree), M. ^R^T, it^, S. ij^ and ^invt- 

Skr. ^ (p) " otter," H. ^, P. ^J7f , M. G. ^, B. O. ^. 

Skr. ^^ (o) " moon," H. ^ft^, and so in all ; bnt P. ^, 8. ^. 

Skr. ^ {h) "vulture," H. fiRT, P. ^-^ M. G, fir?[, VU[[, fiHil, 
S. f^VI}, B. 0. not used. 

Skr. Tfnsr (o) "copper," H. 1ffti|T> P- ^d. and ?|^, 0.ti/.,B. WRT. 

Skr. ;E3qT9 (<») " tiger," H. ITT^, and so in all, S. ^^. 

Skr. irgsf («) *^ocean," H. ^^4<> iRJ^, ll*}^^, P. W., the rest 

Skr. Tp[ (o) " Sndra," H. ^?T* In the rest TIfttsama. 

Only two of these words take the long -d, and of these TThO 
may be derived from the Sanskrit form TTT^RI, which would lose 
the ^, and the two vowels would coalesce into a long vowel, as 
will be shown hereafter. 9^ was probably regarded as an 
adjective. ^ifX in M» is fem.» the change of gender accounts for 
the long voweL 

Digitized by LjOOQ IC 


()9). Examples of words where the suffix is joined to the root 
by a joining vowel— 

8kr. ^^HK W " chowry,'' H. ff^, 4fi^, P.^,Q. ^sftxt, M. id., 
0. ^^, B. ^, S. ^. 

8kr. ^^ (o) " fr<^," H. irriT* P- G- ^^^ 8- 1^- 

Skr. ^^ (o) •* husband's younger brother," H, ^^^ P, U. Id., Q.'^, 
B. 0. ^^, 8. %^. 

8kr. \pgTC (•) ""dhatura tree," H, V^TT. P. B. M. tU, G. V^^t* S. 

Skr. ^^ (o) -jujube," H. ^, %^, P. id., M. Q. ifr^, B, ^1^^^, O. 
Skr. ^TOT (o) -bee," H, ^WK^, ^X^ ^^ ^'> ^ ^t^TT' M, i^, B. 

^fm^f 8- ^• 

Skr. MV^i, (o) « temple," H. 4tfi | (< , and in alL 
Skr. ^rt^ (o) "goaV H. iRn[J, B. P, id., 0. iRRT, CL ^^, S. 
;^f«|^, but M. If^^isl. 
Skr. ^Hfft W "father-in-law," H, ^rgT> P- ^ITT^ 8. ^|ffr. M. 

vrerr, g. ^rarr, b. o. ^b^. 

There is not much uniformity in this list. Sanskrit oxytones 
end in the short vowel in Hindi and the rest; and the one 
barjrtone is in P. S. M. and (J. terminated by the long vowel, 
though the others keep the short vowel. Under the head of 
nouns in ^ will be found an attempt to explain a good deal of 
this discrepancy. My method of reasoning does not admit of 
the usual slipshod way of accounting for the difficulty by setting 
it down to ** caprice'* or "lawless licence/* There is a reason 
for everything in this world, if we can only find it out ; and 
if we cannot find it out, it is only honest to say so, and not 
to try and cover our ignorance by saying there is no law. 
Some words of wide daily use have all sorts of forms ; if we 
knew more about the subject, we should be able to give a reason 

Digitized by LjOOQ IC 


for all of them': perhaps some day we or our descendants may 
be able to do so. For instance, the word ^H (oxytone) "a 
wheel/' has the following long list of forms in modern times. 

H. ^im^ "cart-wheel;" "^TSET' '^T^' ^WT^V, ^^it (a miU-stone); 

p- ^raR> ^^^f 'tsrtj ^^wtt, ^Bpwr^* ^W3^; ^rst, ^fgrarr. 
o. ^raR^» ^^» "vn ; '^raft. ^^, 'qr^. ^i*cb1. 

M.^iH; ^iraT» ^Bpspft; M*fl<; '^SRT. ^^SB^; ^Wl; "Wf^f 

o. ^^, ^T^viT, ^^^A\ ; ^wwT> ^'iTf ^wit* ^^; ^TW- 
B. ^in, ^qi^r» ^^. ^'iT» '^Fii ^wr» ^riit* ^n^- 

All these may be imdoubtedly accounted for by special rules. 
Their significations are yery various, all resting on and derived 
from the primary idea of roundness. To enter into a detailed 
examination of aU of them would take too much time and space. 
I therefore pass on to the next form. 

(2). Stems in la. Many nouns ending ia la do not come 
under this head, because the I is part of the root, ^TTW^ ^^> 
ifTWy WRT and others, which are to be considered as formed by 
pratyayas leaving a only. Of those which are really formed by 
the suffix la, the following are some of the commonest. 

(a). Substantives. 

skr. ^mrn w "lotus,'' H. p. ^nm, *to. m. g. ibtob, o. w., b. 

Skr. ^VSm W "blanket,'' H. VfH* ^b|RW> P. mmSSr 'NSB, G. 
m\MdQ > ^\Hdo\ , V(dot> M. Ili^ldol, O. lKf(dS. B. JOm^, nf^, S. 

Digitized by LjOOQ IC 


Skr. ^SIT^ra (o) " mouthful," H. IS^, ^, P. ^, f^, G. ijlldo*!), 

M. wim, B. 0. ^, grfir.^ 

Skr. 7^1^ (o) "plantain," H.^WT, P- 0. id., S.^pflV, G.%9B,%T» 
M.^i35, B. ^mji ^rf^WT- 

Skr. ^l^^r (o) "caning," H. ^FW. P- '*535, S. ^^, G. ^4cb> 
M. B. 0. ^J^m- 

Skr. ^Ifift^ (o) "koil," H. ^iitl^, P. S. B. id., G. ^Mdo, M. 

^i?r#w, 0. ifwfao. 

Skr. fvm^ (b) "pipal-tree," H. ift^W, P. ti. ftnsRr, O. fqiqdb.' 
M. id., Q. fxTIWi S. fqfq^. 

Skr. ^Tl^r^ (d) " circle," H. #9^, and so in all, M. G. O. also 71^398 • 

Skr. ^^ira (o) "pair," H. ^^TW, P. G. B. id., 0. ^pnaS, M. ga5T, 
^qcloi ^«<cbl (twin, adj.), ^J (twins, «.n.), m^, TO5. 

Skr. ^^ (o) "chain," H. ^tNT, ^rf^^. ^^, P. t^K* S. Irf., 
G. ^rNOo, M. lU and B. fjf^m, 0. ftp«35. 

()8). Adjectives. 

Skr. xf^^(o)." tremulous," H. ^^^, 't^WT, *^N^. P- "^.^W* S. 
^^, G. ^^35, M. B. O. ^^^. 

Skr. fiir«|5| (o) "loose," H. ftm, P. t%«T. »- ftft. t%^, G. 
it^o|;o^, M. 1^, O. f^^, B. id. , 

Skr. ^ft?ra («) "cold," H. #cr5r. ^Nn^t' p- ^^' ft^^* »• 

^d^> M. TJ^craB, 0. id., B. ^f^TTW- 

In the case of adjectives, the Tatsama form does not take the 
long vowel, while the Tadbhava forms do to a great extent 
Tms confirms the general theory. Tatsamas resuscitated after 
the Sanskrit language had ceased to be spoken would naturally 
not follow the accent. 

. ^ Chiefly used In the sense of rinsing the mouth with water after eating. 

' The Onyas prohahly horrowed this word from the Marathas, as it is only fonnd 
in the names of a few places, prohahly founded during the MarathaTule. The Oriyas 
generally use ^TO3 or mn^, from ^t|<««. 

Digitized by VjOOQ iC 


(3). The suffix va is employed to form both adjectiyes and 
substantiyes. It is not of common use. Examples : 

Skr. ^ni (*) " horse," only used rarely as a Tatsama. 
Skr. xrr^ (o) « side," 8. T|TOt, M. IfRRf. 
Skr. f^ (*) « bel-tree," H. %^, B. M. O. id. 

Skr. X(^ (b) « ripe," H. IRffT* P- M. 0. id., alsofw, 8. Jpit, G. Il^ft , 
Skr. 1T$ << eastern, former," H. "JT^. in the rest TJ^, S. n^. 
Skr. ^ «aU,» H. iT^, ^[TTT, B. id., O. ;irf , P. ^RTT, iW, S. ^. 

in^9 as a substantive, has descendants only in M. and S. ; 
but in the locative case tnW it forms an adverb of place, IJTO 
" near,'* in all the languages. As, however, we are discussing 
the formation of the stem with especial reference to its ter- 
mination, this adverb does not concern us here. In the case of 
so familiar a word as ^ "all," many irregularities might be 
expected to have crept in ; thus we have the barytone ^if, as 
well as the oxytone mi^l, the latter by the rejection of the ^. 
Chand uses an oblique singular ^, which would point to a nom. 
^Qprr> if we could place any reliance on so rude an author. The 
final vowel is, however, often merely inserted to eke out the 
metre ; as in the hymn to Ganesh (i. 27, 26) : 

" Before all affairs, thy name is prefixed." 
Here the metre is Chh4nda-vir&ja : 

u- I -u I — I u- I -u I — V 

§ 8. The suffix ma forms adjectives, and masc. and neut. 
substantives. It is generally oxytone, though there is also a 

^ This constant use of tabbai in Chand may be nothing more than a Prakritism for 
the nom. p). of Skr. iarvve. 

Digitized by LjOOQ IC 


class of appellatives which carry the tone on the root-syllable. 
Examples : 

Skr.m^ (b) "mud," H. HTf^, TfT^, G. 1|T^, 0. ^ITfH, 

Skr. ?rm {b) " viUage/' H. irf^, 8. ^n^, IfX^, G. ITRT, M, ;qt^, 
B. 0. ^. 
Skr. ^ (o) « heat," H. ^TTiT, M. 0. B. W. 
Skr. ^ (•) « tree," H. ^^if, and so in aU. 
Skr. ^ (o) « smoke," H. ^irf, P. irf., 8. ^, M. G. ^, B. ^^, 

Skr. \f^ (6) '< justice," H. \ar^, and so in all. 

Skr. ^THT W "left-hand," H. ^TT^, ^nrf, P. TnC^, G. M. ^m, 
B. irf» 0. id. and ifR. 
Skr. ff If (o) " snow," H. ^H, M. fflf. In the rest Tatsama. 
Skr. |if {o) « gold," Old-H.l, |if. 

There are not many examples of this suffix. In most of them, 
where not still in the Tatsama form, the weakness of the if, 
which passes into ^ preceded by anuswara, has caused the loss 
of the final syllable to be very common. In many cases the "if 
has itself disappeared, leaving only the anuswara; and in 
i||^, 41^1, etc., even the anuswara is lost. 

§ ^. The suffix ka is of very frequent use in Sanskrit, and in 
the modem languages its use is extremely common. It branches 
out into many different classes, and its discussion is embarrassed 
by the fact that it is employed both as a primary and a 
secondary suffix. We are not directly concerned, however, 
with the minute distinctions which Sanskrit grammarians find 
necessary ; for our purpose it suffices to take the Sanskrit noun 
as it stands, without troubling ourselves to inquire whether it 
be formed by adding the suffix to a verb, or to another noun. 
As regards the modem languages, some of the classes of this 

Digitized by LjOOQ iC 


suffix which are regarded in Sanskrit as secondary are perfectly 
primary, and may be so treated without any loss of clearness. 

Ka is generally joined to the root by a vowel, and the termi- 
nations most in use in forming nouns which have lived on to 
the present day are aka^ ika, uka, dka, and &ka. 

Whatever may be the strict philological theory as to the 
origin of this ka, in practice there can be no doubt that its 
meaning is primarily that of " the doer/' It may perhaps be 
supposed to be a shortened form from the root kar, " to do ;" but 
this is a matter of Sanskrit philology, into which it is needless 
here to inquire. Starting from this point, however, the mean- 
ing, like all meanings in all languages, widens out and loses 
in distinctness as it goes on. A fourfold division may be de- 
tected, which has this advantage, that it subsists in the lan- 
guages of the present day as cleariy as in Sanskrit and the 
Prsirits. If it be found in Sanskrit, it will of course be also 
found in the Prakrits, as the latter are the mere apes of the 
former, having no independent ideas of their own ; or perhaps 
it would be fSairer to say of them, that they are identical with 
Sanskrit in so far as they are the popular side of that ancient 
Aryan speech of which Sanskrit is the learned and literary side 
only. It is not till we come to the Indian renaissance in the 
twelfth century that we find the popular dialects possessing any 
originality, and striking out for themselves forms which are 
something more than mere colloquial and phonetic corruptions 
of Sanskrit. When they do begin to do so, they often leave 
the ancient path cmd go into ground where it is difficult to 
follow them, or account for the origins of their forms. When 
therefore they do carry on an anci^it system into modem times, 
it is a fact to be laid hold of and brought to the fore. 
.The fourfold division which I make is, then, as follows : 

(1). Words which mean purely and simply "the doer:" as 
mr^ ^ doer,'' xrnTH " cooker," ^psni " giver," %^i|| " writer." 

(2). Wbrds in which the sense of "the doer" is only to be 

Digitized by VjOOQ IC 


detected metaphorically: as ^^ "a road," i,e, that which 
goes on and on; ^x|^ "pupil of the eye/' i.e. that which 
shines or glances ; VJ^^ " fire," i.e. that which purifies ; IfTRI 
"asthma/* i.e. that which chokes. This class includes words 
which are derived from intransitive verbs, and those in which a 
passive sense must be substituted for that of the agent: as 
R^^ " leopard," i.e. the spotted animal ; ^THRS " a pit," i.e. 
that which is dug ; mit " lotus," i.e. that which blooms ; ^n 
" birth," i.e. the act of being bom. 

(3). Words which have entirely lost all idea of agency, if in- 
deed they ever had it, and have become pure appellatives : as 
^ICTI " an army," which, if it ever had the idea of " the encom- 
passer," from the root iR^ "to surround," had lost it long 
before the times to which we can look back ; J^l " a wolf,'* for 
which we have only a dubious verbal root W^ " to sdze," which 
looks as if invented for the occasion ; ^^ " hell/' ^fS 
"world." . • 

(4). Words in which the suffix has no meaning at all, but is 
merely added for metrical purposes, or to avoid the intricacies 
of declension. This is a very numerous class, and we have 
side by side two forms, one with the suffix, the other without 
it: as^rtzand q^f^^ l "waist," ;E|nci^ and ^STl^^ "box," ^HW 
and '^^RR "black," ift^ and '^tW^ "ball," ^rWT and ITRRR 
" net," 7T^ and 7[Wj[^ " thread," and hundreds of others. It 
may almost be said that this suffix may be added at will to any 
Sanskrit noim, and in fact it would be allowable for any one 
who was composing or translating into that language to add the 
suffix to any noun he pleased, and authority would no doubt be 
found for any individual word in the vast stores of classical 
Sanskrit writings. 

In the modem languages, although the same shades of mean- 
ing can easily be traced, yet as our business is rather with form 
than meaning, it will be more convenient to exhibit the various 
details of this large and important class according to its forms, 

Digitized by VjOOQ IC 


allading to the meaning of the words only where they are in 
any way essential to the clear perception of the subject. 

(1). "ire has two fotms. — (a). In Tatsamas it retains the k 
with the masculine gender, and with the sense in general of the 
agent, as in ^RTT^ " a doer," ^TT^ " a taker," UU^ " a taker." 
In Tadbhavas there is a numerous group of interesting words, 
which have come to us through the Prakrit, by virtue of the 
rule laid down in Book I. § 53 (3) and § 54. A single conso- 
nant in the middle of a word would be elided by that rule, and 
the termination akah becomes thus ao. In the moderns the a 
drops out and the o is retained in G. and S., but changed to d 
in the others. Thus we get the following ; 

Skr. ^^rnr^ra "cmWic myrobalan,* Pr. ^H^^, H. '^rt^WT^ P. 

Skr. ehU^c^i « thorn," Pr. iSZ:^, H. ^itZT. P. ^ilT. M. ^fUzi, WZJ* 
B. «., O. ^P5T, G- 'RtZt* S. *it. 

Skr.^^tn"horse,''Pr.l^t^^, IL^t¥T,P.M.B.0.t(/.,8. G.^t^. 

Skr. f^^^ "leopard," H. ^^TfT, P. f^W, M. f^^TTT, B. O. id., G. 
f^Tit. S. f^^ and ^IHt- 

Skr. trW^ "description," Pr. 1?Z^, H. HTfT* P- ^f M. irTTT. B. 0. 
id., Q. S. Tl^. 

Skr. ^^Ri "book," Pr. iJtc^'^^ H. ijt^, and so in all. The fern, 
ify^ is however more common. 

Skr. VJZ^ " liire," H. HTIT, B. O. id., P. HWT. S. ^ifff. 

Skr. ira^ "head," Pr. ^f^nSt, H. J^m, B. M. m/., P. ^(^, S. 

Skr. -^^iZ^ "tumour." Pr. ^Pt^J^, H. lit^, B. 0. P. id. 

Sindhi uses this affix, which in that . language naturally 
results in o, as descriptive of trades or occupations. Trumpp 
gives as examples the following, formed by Sindhi out of its 
own modern materials : 

Digitized by LjOOQ IC 


^^TP^ "carpenter," verb q^QI " to cut." 
^^ "sawyer," „ ^^^^ " to spUt." 
^V^ " diver," „ ^^ ** to dive." 

I^t?^ "seeker," „ V^\TS " to seek." 

In ViTZ^ M. is irregular, having HT^ (fern.) and ^JT3 (w-)' ^ 
also G. ^>4 and iT^ instead of ^T^ from THOTI- M. has also 
contrary to rule i^t^ from ^^^, but also Mit>jl, which, as 
will be shown hereafter, is from a Skr. fem. form ^(d^l- 
The feminines of the form in aka always end in Skr, in ikd, 
e.g. bdlaka, " boy," bdlikd, " girl." The comparative neglect by 
M. of the long d is possibly due to the fact that masc. nouns 
ending in short a, i.e. a mute, in that language, change the a tod 
in the oblique cases ; thus from ^^ " a house," '^XT ^ " of a 
house," tnj ^ "to a house," so that the distinction between 
this class of nouns and that which ends in long d exists only in 
the nominative, and is thus of comparatively rare occurrence. 

Here also it may be admitted that, as the suffix ^ may be 
added at will to all nouns in Prakrit, it is, probable that many 
of the nouns ending in long d or o, which I have held to be 
derived from Skr. oxytones, do in reality owe their final long 
vowel to the fact that the word from which they are derived 
had in popular, though not in classical usage, a ^ tacked on 
to it. This would account for Sanskrit barytones like ^Y^, 
OTi$, becoming ^^1^, H\A\} with long d. The difficulty, as 
already mentioned, is the existence of ant/ nouns in a-u ; if ka 
is added to all nouns of the a-Qtem, why do not all end in 4-^ ?^ 

()8). The suffix ^Rl appears also in a great many words of 
apparently modem origin, as well as in a few which can be 
traced back to Sanskrit, most of which are feminine. 

^ This view is taken, as I have stated above, by Prof. Hoernle of Benares, in 
Jonm. As. Soc. Bengal, vol. xli. p. 154. The learned writer is not disposed to 
admit with me the influei\ce of the accent at all. There is, however, not only much 
that is difficult to understand in that essay, but much t^iat requires farther proof. 

Digitized by VjOOQ IC 


Some of these are pure appellatives, and If derived from 
Sanskrit feminines, all of which end in T^, they must have 
lost the final long d, and changed the i into «> neither of which 
processes indeed are without a parallel, though it seems un- 
necessary to suppose that they have taken place here. Examples 
of such words are — 

H. ^rg^ " a road" (/.), P. M. B. O. id., G. ^TCTJ (m.). 

H. Mil^cfi '< a gate " (m.), and so m all except S. 

H. %Z^ " a seat" (/.), and so in all but S. G. writes %7^. 

We find in Sanskrit ^9^: and ^^ "a road," as well as 
^fffif , from a form of which, ^rfl$^> or from ^ET^ by change of 
the semivowel into the media, this word may have been derived. 
Mil^oS, like all the neuters of its class (see Vol. I. § 80), is a 
formation from the root '^^ " to split,'' " to open," though we 
cannot point to any actually used Sanskrit word from which to 
derive it. ^tZ^ is quite a modem word from the verb ^JifT, 
concerning the origin of which s^ Vol. I. p. 179. 

Now comes a long string of words, all feminine, which 
express noises, pains, violent actions, and are like our wonls 
buzz, thump f crack, bang, jingle, tingle, and the like. It is almost 
impossible to do more in any of these cases than refer the word 
in a general way back to some Sanskrit root. The majority of 
them are in all probability onomatopoetic. As the same words do 
not occur in all the languages, I give each language separately, 
and add that every one of these words has its verb with the 
same meaning. I give the verb in the first few examples to 
show how it is formed, the reader can supply the rest for 

Hindi. "WZ^ "stoppage" (^R^RT "*<> be stopped," ^JZ^THT " to 
restrain), ^R^Ri "pain" (i|P5^CTT" to ache"), ;E|r7^ "crash" (^TW^TT 
"to crash," "to fall with a crash"), ^B^^ "sprain," li^OTI "clang," 
^IZ^ "pit-a-pat," ^f^ "drunkenness," "reeling," ^t^ "starting," 

Digitized by LjOOQ IC 


fq^qij "^^PSfif ipra> «|«1^> ^^Rl, ^^ra, all mean "glitter," "flash- 
ing," ^Z^ "crack,'' "snap," ftjipi "start," IJZ^ "jerk," f^g^W 
"wrangling," "quarrelling," "Z^^ "twinge," ^if^ "harsh sound," 
"crunch," Z^^ "throb," "drip," ^^ "ache," T66^ "amazement," 
^if^ "sob," ^^OTI, \^i|S "trembling," Y^TiR "roUing," "wallowing," 
^fH " glare," 15i(:H| " flutter," « twitter," 9)^9^ " starting," " flash," ifCTI 
"ogle," ^r^^ "bound," "spring," ^(Z^ "hanging," " dangling," ^fljil 
"flash," "bound," ^T^T^ "gush," "bursting," ^f^ "flash," "glare," 
a^H " tapering." 

Panjabi, Many of the same words as Hindi, as H||^c|{, ^EKIT^, ^^^, 
f^^^nV, ^7!1|S, ^Z^y IgZm, ^H^f also Zlir^, ^Vm ; and some words 
of its own, as f^Z^ "filth," "scum," f^"^ "pus," "mucus," ^Jl^ 
"wink," ZtCTI "jugglery," " hocus-pocus," ^"^^ " crouching," " skulk- 
ing," V^RI "fear," "dread," VT^ "creaking" (of shoes), ^^f^ "fond- 
ness," " petting," and others. 

Sindhi does not exhibit many words of this form, owing to its invariable 
rule of making all its words end in a vowel. Examples are ^J^1|S (/I) 
" lustre," ^-tf^' (m.) " caprice," " whim," and a few others agreeing with 
Hindi, as ^?^, ^91^, ^J^^>etc. ; but Sindhi, as will be shown hereafter, 
has another form for words of this class. 

Marathi has ^35^ " glitter," cnrH "bang !" " crash !" ^^iR " shock," 
If^liS " copiousness," ^ZIR " liking," " taste for a thing," Yf^^ " blow," 
flpipil "twinge," ZH^ "jingle," ^^[^ and ^^iR "handful," ^R^ 
"a dab," or "mass of mud," tl^Gb'cli "a splinter," together with some 
of those that occur in Hindi. 

Crujarathi is, like Sindhi, not very partial to this form. Instances are 
ZiHR "strut," "swagger," ZW^ "sob," TJ^eR "shock," ^q^ "slur," 
" blot," ^[ITiR " glitter," qab^ and l^BR "glitter." 

Ofiya and Bengali have mostly the same words as Hindi, which need 
not be repeated. 

Digitized by VjOOQ IC 



From the above lists have been excluded words of similaer 
form, in wbich the ^ is organic, or a part of the root, as ;s^ 
"staring," ^f^l "shove," "puff," ^ "piece," tf* "hind- 
rance." But under this head come miscellaneous words like 
^ft^ (w.) "square" (in all the languages), which is probably 
from Skr. "^T^^^iR. The H. ^t^ " starting," as in ^\M ^Z^ 
"to start up" (as a man suddenly aroused from sleep), is, it 
would seem, contracted from ^7fc7, a shortened form of Sanskrit 
'4^^K " surprise." 

All these words beiug iu form identical with the root of the 
verb, are most usually employed to form compound verbs with 
the addition of ^nfT? itTTj^TTj TWTT or other semi-auxiliaries, 
especially iu Hindi, as WZ^ XW^ " it remained hanging," ^sltWt 
H^^ 'RTT "the horse started (from fear)," ^cTENh* ^TCTI t^^ 
" he dashed it against (the ground)." ♦ It is doubtful, therefore, 
whether in many cases they should not be rather considered as 
a part of the verb, than as noims. 

Sindhi, Gujarati, and Marathi, and to a less extent the other 
languages too, use also a closely allied form of this suffix, in 
which the final vowel is long, kd or ko. In many cases both 
forms exist, as iu H. V^^ and ^SRRiT? G. ^Z^ a.nd ffZ^. In 
Sindhi this form is the favourite, and is used almost to the 
exclusion of the other in eR. Examples ia that langiuige are 
the following : 

^^4i \ " fear," " dread," ^f5<v<^ " retchiDg," ^gp^yt " c»re," " anxiety,^ 
^>^cji^ " crack," " crash," " thunderpeal," ^>^iJY " rattling," " pattering," 
^^^fit w/'j ^^«Rt " rumbling," Tf^iit " rub," ^4j4V " fondness," " taste," 
If^Rfr " burst," « blast,'? " gust," ^JZ^ id., ^J^Nst "puff," Zf^ " boil- 
ing," "bubbling," V^^ "trembling," \annst "rumbling," "gurgling," 
ll^^ " quivering," " shivering," " rustling," XRXIRJrt " bubbling." 

The other languages have also this form. Thus Hindi has 
^nRiT as well as Wi^, ^ST^RRT and ^^^n^ V^I^ and Vit^- It 
is not necessary to give detailed examples, as in popular and 
VOL. n. 3 

Digitized by LjOOQ IC 


Tulgar words like these no very strict canon is observed ; some- 
times one form is used, sometimes tlie other. In 0. and B. the 
forming is most common. 

(2). The suffix l[;i| follows the analogy of ^nC- Iii Tatsamas 
it retains the 1%, in Tadbhavas the ^ goes out. Here, however, 
it is not left in the same condition as 'H^, because the dis- 
appearance of the ^ brings into contact t and o, and later i+ d. 
In the case of aka, the result of the elision of k was a + o and 
a + d,m both of which cases the short a was easily absorbed by 
the long vowel ; «, however, resists absorption, not being homo- 
geneous with the vowel following. What we really get is a 
double set of forms, of which one ends in long i %, the other in 
X^, T^, or JiJ. 

Taking first the form in |^, we are met by the difficulty that 
more than one Sanskrit •termination results in |^, at least in 
H. P. B. and 0. For instance, there is the form ^= |^, as in 
Skr. xrrfhsf, which is in all VfP^ or liTtlH'; and again, Skr. 
1[H = |; in ^grf'T'l, modem ^rrit or ^fff^, \jrfini^^\jnft; as well 
as Skr. |^ itself. Further, it must be observed that the suffix 
1[J|| is in most cases a secondary suffix, so that a notice of it 
would hardly come in here. The cases I shall now give are 
chiefly from the Skr. fem. J^, which, as mentioned above, 
belongs really to the "H^ series. 

The best example of a bond-fide primary word of this form is 
the following : 

Skr. ^Pftfilrtl " pearl," Pr. iftPrT^. ifWt (»».), M. lft?ff (n.). In all 
the rest 4ft cfl (»>•)• 

In M, the anuswara is a retention of the neuter form in 

The following are from f eminines in 1[HIT - 

Skr. wif^m "%>" H. p. 1^^, m^ (A o. B. imft, M. irnft, 
0. ^frreft* s. ?!f%. 

Digitized by LjOOQ IC 


Skr. 4jPi^l "earth," H. f^ft, if|^, P. G. id., M. ?Tnft, O. B. JH- 

Skr. ^4l(d«ll "cucumber," H, ^l^lf^, and so with slight phonetic 
variations, but with final t, in all, 

Skr. grf^^iT "key," H. B. ^, P. S. M. yft, G. ^, O. 

Skr. I^Qr^f " chalk," H. ^rVt> ^^^ so in all, bnt O. ^rf%. 
Skr. ^Tfwrr "beard," ft. G, ^iT^, P. ^Tf€t, ^. TRfft, M. 1^, 
O. ^Tf^, B. ^Tft- 

Skr. iftfz^ " rapeseed," H. <fr€^, M. fft^gJT- 

Here we may introduce a group of Maratlii agents and 
adjectives, which, as derived from Tadbhava verbs, are primary, 
and appear to have originated from Sanskrit nouns in ika, by 
rejecting the k and hardening i into y. Such are — 

^|^<e|| " a borrower," verb ^ST7^ " to extract." 
^rorr " laborious," „ ^nji)f " to work." 

^^r^^TTT " scraper," „ ^^vSl^ " to scrape." 

^9F^^ "jeweller," „ ^f^^f " to set jewels." 

t^3B^" prying," „ tf35^f " to pry." 

In many of these the alternative form in long i is found, as 
^1^, ^^nft, etc., and in some cases the latter is the only form 
in use, as ^|cft "owner of a field" (Skr. ^f^), frojn Jfj^ "a 
field'' (Skr. %w). 

We may now dismiss ika till we come to the secondary 
formations to which it more properly belongs. 

(3). ^<1 is exactly similar to ^R| and 1[H|, but its various 
forms can be traced with greater accuracy, as the vowel is not 
so easily confounded with other suffixes. 

(fit). The full form uka is retained only in Tatsamas. 

(/8), The k is rejected, together with the vowel which follows 
it, and the u lengthened in compensation. Words of this class 

Digitized by LjOOQ IC 


haye almost always the sense at the agent, either direct or only 
very slightly metaphoricaL Instances are the following : 

Hindi ^Wl^" destroyer," verb ^S^VThRT ** to destroy/* 

u ^TTRf " spiller,** „ ^nTTTTT " to cause to descend,* P, 

id. but in sense of ^deg^ra^d," 
M.UL**tL passenger." 
„ liT^« cutter," „ ^ffJWT "to cut" 

„ ^n« "eater," „ ^gTin«toeaV'P.S.M.«.,G.^Pr5 

(adjective) "destructive." 
„ ftl^" player," ^ %fRT"toplay." 

„ Hf^" watcher," „ — [Skr.Hf^, H.l|f^" watch"], 

„ ^nPlf" fighter," "beater,",, ?nTirT"tostrike,"P.S.irf.,G.^Tn| 

„ ^T^' " caUer," « robber," „ ^J^m " to shout," P. id., O.B.^T^ • 
„ V^^^" pusher," " shover," ^ipilW^ " to shove," P. id. 
„ in^ " ascetic," " one who 1 4JI|4«|| " to shave," P. id. " a child 
shaves his head," „ ) whose head has been shaved for 

the first tin\e." 
,, ^"wrestler," „ ^W^ " to puU," P. i^f. 

>» Wl^ " waller," „ ^TPRT " to walie," P. id, 

„ i»M " metal caster," „ Tf^PfT " to cast metals," P. id. 
" t^niT|^"8p<Hler," „ f^1ITain"tospoil," 

„ lg\M " sweeper," "broom," f||44|| " to sweep," and so in all. 

In addition to the words noted above, Panjabi has also ifpg. 
" tearer/' UTiraf" reader," HT^"prover," "trier," ^5^"dipper," 

^ f>Ak(t is a rural form of Bakait (dacoit). The word is derived from the verb "to 
shout," because the dacoits always shout and call out at the entrance of the village they 
are about to plunder, whereby all the inhabitants, being terrified, hide in their houses, 
and the dacoits, who are generally quite as afraid of the villagers as the viUagers of 
them, can plunder the house they select without opposition. 

Digitized by VjOOQ IC 




^^n^ " adulterer/V ii^TIf "lasting," from imWT "to tear/' 
il^VflQT " to read,'' ^l^ip " to test," ^WQT " to be immersed," . 
^3\jITW^ "to abduct," "seduce" (Skr. ^1?:^), f^nfTflTT "to 
endure" (Skr. fvR^ in the sense of not + destruction). In some 
of tbese, as irm^ » ^^^' ^^^ ^^ radical vowd is lengthened, 
or even gunatized, a practice of which more exam{des will 
be shown imder the Verb. 

From the habitual omission of the Ghijaratis to distinguish 
between short and long u, it comes to pass that it is almost 
impossible to tell which of the words ending in u should cor- 
rectly be written with long 6. As each word must be oonridered 
on its own merits, the dictionary is the fit place for the dis- 
cussion, ^rr^, ^IPPI? ^WT^, should probably be written with 
^, but in many others the point is doubtfuL The difficulty 
is increased by the fact of the existence of the pratyaya u in 
8kr., which also has the sense of the agent, so that in a language 
which does not distinguish between the long and the short 
vowel, it becomes impossible to say whether we are dealing with 
u, or uka, or &ka. The following list exhibits the most common 
words of the class : * 


„ ^" to sing," 

„ ifZ^V^ " to wander,'* 

„ ^Q^^ " to qpend,'' 

99 ^Q'^l^lj " ^ trnderstand,' 

» %1^" to till," 

„ 1]T74 ** to sweep," 

„ ^S(Xyi " to wander," 

^BT^ "eater." 
^n^ " singer." 
1)7^ "wanderer." 
^i|^" intelligent.'' 
%n| " cultivator." 
^]T¥ " broom." 
^^ "wanderer," 

1 For ibis list I have to tbank tbe Bev. I. Y. Taylor, of Abmadabad, an acoom- 
plished Onjarati scholar, and author of an excellent grammar of that language, pub- 
Hsfaed at the Irish Histion Press, Surai, 1868. It is unfortonately written in 
Qnjarati, so that one is eiq^ected to know the language in <Hrder toJeam it ! 

Digitized by VjOOQ IC 


Verb HH^ "to do," ^iRl^ "doer." 

Of these H^, H^, ^n^^ IT^? are probably formed with w, and 
the rest with uka. The word ^pS is a hybrid from Persian 
^j^ "expense." 

Sindhi has many examples of this form; the following are 
the principal : • 

inr^ " swimmer," verb ?r^^ " to swim." 

HT^" fugitive," „ Iflfig " to flee." 

IHra « breaker," „ ^^ " to break." 

Tff^ "one who obeys,^' „ 9TSra "to obey." 

XU " dweller," „ "^f^ " to dwell." 

"q^^ " goat " (i.e. " browser^') „ tn(X^ " to put out (cattle) to graze." 

^"beggar," „ fiRIJ « to beg." 

i^Jxr "vagfabond," „ ^fira '' to wander.*" 

^Jf " patient," „ ^f^HT " to endure." 

%NRf "fighter," „ fSr^i^" to fight." 

Ift^ " robber," „ ^^K^ " to rob." 

fT^ " destroyer," „ [Skr. fi^. " to kiU "] . 

In the majority «f <;ases it will be observed that the vowel of 
the root is lengthened or gmiatized. Marathi does not employ 
this termination very largely, and I have not found many 
instances worth quoting* Nor do Bengali and Oriya much 
favour it : a final long vowel is in no case agreeable to the spirit 
of these two languages ; and the forms in use belong to the next 

Hindi has a few words of this form which are simple appella- 
tives as far as their present use is concerned, though, perhaps, by 
some far-fetched and fanciful chain of metaphor, they may be 
capable of being resolved into agents* Such are ; 

Digitized by VjOOQ IC 


Skr. "ilT^^ "sand," H. WW* G. has also W[^» but O. WrfW» S. 
^1^, B. mwt and ^fw> probably from a Skr. form l|l(lil|| (see 
remark on ^[f^^ under JjBH). 

Skr. Hf^. also mn and H^ "a bear," H. mm, M. id., O. ^J^. 
Bhojpnri Hindi per metathenn Vit^SMf ^^tWi* 

Skr. W^n (P ^TT^) " maternal uncle,'' H. Wm> M. id., O. i|T^, 
but B. and P. 4||4|h ^- and S. TTRTt* 

(7). Especially in B. and 0. the form ending in ^5^ (often 
shortened to ^5^ in 0.) prevails, and in Hindi also this form 
is frequently found, particularly in words denoting occupation 
or trade. In Bengali it is ordinarily written ^mi, but in this 
ease, as in many others, the ^ is merely a fulcrum for the 
following vowel, and is not pronounced. * In this class are in- 
cluded many simple appellatives, and numerous words for which 
it would be difficult, if not impossible, to find a Sanskrit origin. 
Examples are : 

Bengali ifTf^ "fisherman,*' O. TfT^^ and •^j H. ?!^1|T, also 
written K^^, P. ^fTS^y S. Vf^^ "a fishing-boat," G. 9!^^ id. 
[Probably from an unrecorded Skr. form TTT^RlO 

Bengali i)^i|| *<pimp," O. H^pUT and o^, H. H^T^ (chiefly as a 
term of abuse), M. Hl^- ['^^^ classical Skr. is ^^TZ, but we must 
suppose a form V[\^ or ^Zm$ from ¥f^ " to hire," t.^. one who hires out 
women ; cf. our English whore, German Hure, Ang.-Sax. hure, with kyre, 
heuer, "reward," "hire.''] 

Bengali 2^€H "skewer," verb %^ "to stick." 

Bengali ^^i|| " stake," verb id. 

Bengali ^VT^RTT '' postman," verb ^Pi " to call." 

Bengali ^TZ^ and %Z^ " palm of hand," verb ^XZ " to stroke," " lick." 

Bengal %^p!TT '' widower," probably connected with '^t^^ '' a stick," 
g.d. "a withered branch," "fruitless," O. ^i^pTT* 

Digitized by LjOOQ IC 


Bengali Xt^ "glutton," from XtZ "bdly.** 

^°Sr<^ f||tA€|| "pungent sauce," verb ^f[W( "to burn." 

Bengali ^|<dm " cast in a mould." 

Bengali l^^ITT " snarling." 

Bengali WTW^ "watery." 

Oriya has, besides those mentioned, WT^^ "full of holes," 
from m^ "a net ;" ^T^W "bachelor,'* probably from ^rrf% "a 
stick'* (see %y^ above) ; H^^ "a fool;" V(inRT "honeyed," 
from iT^ "honey;" 7T^^ "a drunkard," from H^ "wine;" 
^n^^ "a bearer," from Skr. ^ "to bear," HIT "a load ;" m^^ 
"an oil-bottle," from ^ra "a reed" (oil being generally carried 
in a joint of bamboo) ; iTHirW "a big nose," from irnn "nose ;" 
%^[^ " a clod," >il^^l " a kind of rice," etc. 

Many of these words are secondary forms, if we have regard 
to the rule which holds that only nouns derived directly from 
verbs are to be considered as primitives; but in wrords whose 
derivation is admittedly obscure, it seems mmecessary to keep 
up this distinction. 

In Hindi there are often two forms of the same word, just as 
we saw in aka. Thus we have T|f^ and ^f^nSRy inff and ^IJ^ ; 
and in many cases there is only the form in ud, as in ^TTW 
" one who' files," from ^Jj^fV " a file," or^jfifT " to file ;" Wf^, 
ifS^, mentioned above, and many others. In Bhojpuri Hindi 
the termination ud is added capriciously to all nouns whatsoever, 
and as we cannot suppose an origin from a form in uka for all of 
them, we must be content to see in this practice merely another 
instance of the common rule that a form, when once introduced 
into a language, is extended to all sorts of words with which 
it has no legitimate connexion. Thus we hear the peasant of 
Tirhut and Ghamp&ran use the following forms : 

%I(19T " 8l»ve," Hindi %1T. 
f^^ "plough," „ fur. 

Digitized by LjOOQ IC 


t5in"8on,» UindiifeT. 
^r^^ " house," „ IR;, etc. 

(4). Aka. — Connected with this suffix is a form AkUy which 
appears principally in Sindhi, and to which, with elision of ^, 
Trumpp refers a common form in Sindhi 46, in which the final 
u is lengthened. It is clear, however, from the structure of 
other parts of words in Hindi which end in this form 46, that 
they are not derived from Aku^ but from causal verbs, in which, 
as I have explained in Vol. I. § 60, the aya, which is the 
characteristic of the causal in Skr., is changed through Au into 
a, to which the long (l representing the suffix uka is appended ; 
thus we get — 

IjiUiJlMI "to coax,"* H. ^HWRT "a coaxer," "wheedler." 

fa^l^l "to feed" [caus. of liTTT "to eat"], H. f^WTST "feeder." 

^3^TTr "to cause to fly" [caus. of ^vj^l "to fly"], H. ^>^|^ 
"spend-thrift" ("one who makes the money fly"), P. W., S. id. 

^l[Pn^ " to earn" [a quasi-causal or nominal verb from ^iTT "work"], 
H. ^fiiTT^ " a worker," " bread-winner," P. id,, M. id, 

^^TPff "to make clear" [causi of ^T^PfT "to see," Skr. ^ftpf], H. 
^^IXSI " an indicator^" also as adj. " significant," "perceptible." 

fa^ f ^i r "to detain" [caus. of tZVIT " to remain"], H. fZlRT^ {adj,) 
"durable," "stationary," P. id. 

ITSTfT "to cause to be set" (jewels) [caus. of ^!|^^ "to set jewels"], 
H- ^rer^ "jewelled," "studded with gems,** P. W., S. id. 

^^i!S[^ "to travel" [unused irreg. caus. from '^JZ "road"], H. IfZT^ 
"traveller," P. Mf. 

^[^TRT "to melt" [caus. of ^JW^ "to be melted"], H. ^^RT "a 
liqucfier," " solvent." 

M. ilTCIxA "one who is inveigled away" (wife or servant), from IVT^^ 
"io take away." 

1 Causal of an unused l|i^<tj|«l| " to slip," " to wayer," i.^. " to cause to waver." 

Digitized by LjOOQ IC 


This class is exceedingly numerous^ especially in H. and P. ; 
in fact; a word of this form might be made from every clausal 
in the language, and would probably be understood by all classes. 
The dictionaries do not give every word, perhaps because the 
compilers do not happen to have met them in writings. 

Instances of nouns in <itl in Sindhi which Trumpp refers to 
dku are the following : 

4,f |*A " dweller," from ^^^^ " to remain." 

$HIT^«puUer," „ fipif^ « to pull." 

t^rtrar" wearisome," „ t^r^TT^ « to tire.» 
t^^nrar " a destroyer," „ f^^THC^ " to destroy." 
%iT^ *' saleable," „ f^m '< to be sold." 

^^X^ " an ill-wisher," „ f^^Hf " to curse." 

Of these words, however, f^rtlBF may be regarded as formed 
by uka from the causal, as the word t?T^ "to be wearied,^' 
" to be tired by any one,^' is in use, and f^^ll^J^ is its causal ; 
and the same may be suggested of most of the others. Sindhi 
does in many cases retain the k ; as, for instance — 

ri|4^|M "a receiver," verb fipC^^ **to take." 

fl|l||S "a drinker," „ ftnra "to drink." 

"^fT^ "d^eUer," „ ^^flj « to dweU." 

fi|^Tl| "sleeper," „ fit>J "sleep" (Skr. fifyr). 

%^ra " tamed," „ ^^TJ " to tame." 

But there are many instances in Sindhi, in addition to those 
given above, in which the form dii is undoubtedly from the 
oausal, as 

^[;|rr^ "a shirker," verb ^pnt]^ "to miss," "evade," causal of ^p^ 
" to be missed." 

^ Although this verb now means "to devoor," yet its origin (from Skr. ^Hpll) 
shows that it had at one time the meaning " to take," which has now attached itself 
to another deriyatiye from Vi^^t namely, Nl'^IU* 

Digitized by VjOOQ IC 


^nn^ "extravagaot," verb ^Mlf^fff "to spend," causal of ^R^ "to 
be spent." 

ie|ST^ " deKverer," ^TTC^ " to deUver," causal of ^^ « to escape." 

In Hindi this suffix takes usually the form dk; as £xr as my 
inquiries have gone, I am not aware of any forms in d& in 
Hindi which can be referred to dku; they appear all to be 
like ^9WI^> ftwrar, and the rest, derived from causal verbs, 
and the d belongs to the verb, and not to the suffix. Instances 
of dka = dk are the following : 

Hindi ^ST^HV "a flier,'' '<a bird that has just begun to fly." 
» ^^Tl| " a swimmer." 
•n Wnm " hissing sound," 
»> IJ^n "speed," "hurry" {literally "being swept along," from 

|||>il|( "to sweep"). 
« ^inTRI "quarrelsome." 

Most of these have also a form with the long final d, as 
II^niT, ^nnT, and it is diifl^cult to distinguish them from 
formations in which the long d represents the causal; thus 
Ttin and JRjm "crash,'' IflT^ and TTWim "thud,'' ^TW 
"clink," seem to be from the <5ausals l|^fi|T, <!!¥RT> and 
OT^TRT, where the final k represents aka, not dka; and it 
will be seen that the sense of agency is as much obscured in 
nouns of this form as it is in the cognate forms i|m, ^B(ZM, 
IPRI, and the like, given under aka, 

Qnjarati, like Sindhi, has this form in frequent use : 

^%^ "to fly," ^41^ "one who makes the money fly,^ 

Z^ " to endure," ^l||<^ " enduring." 

M^€| "to shape," ^bWT^ "that may be shaped or moulded, 

^^j "to be mounted," ^^THS "rideable" (a horse). 

» ((, 

Digitized by LjOOQ IC 


W^ "to fight," ^r«T^ "fighter." 

I|44 " to quarrel," l|4|^ " quarrelsome " (this word is rare). 

IR^ " to be hot," 7tm^ and THTTS " hot-tempered," "Irritable." 

The only one of the above which can be referred to uka with 
a causal is ^>if^; the others have no causal sense, but are 
either agents, or adjectives with a secondary meaning, and in 
one or two cases even, as, for instance, in %|>i|^, ^>91^, the 
meaning is passive, like that of the Latin part, in -m?w«, as 
faciendmy etc. 

(5). Ma is a rare termination in Sanskrit, and is not traceable 
in any of the modem languages except Marathi and Gujarati, 
and in the former its presence is to be ascribed more to the 
habit which this language possesses, of lengthening the final 
syllable, concerning which see § 50. 

Instances are the following : 

^1^ "sample," "taste," Skr. W^ "to select" 
^^"frog," „ 7n^,l.«. ifl^+^rV' 

^rn^m " teasiDg," from ITRltf '' to tease," Skr. ITHIl* 
^rS^fira "remembering," „ 1K3^|^" to remember "(perhaps Skr. ^Vl| 

with ^IT in sense of "abiding "). 
^rpn^ " wakeful," Skr. id., from mnv + ^TV * 
1^>fipra "stopping," from ^4^^ ^to stop." 

As this suffix is especially used in forming diminutives, it 
will be more appropriately considered imder the head of 
secondary formations. 

A Gujarati instance, written with the short tf, is ^nS^pi 
"carriage,'' "deportment,'' "behaviour,? from 'l^^ "to be- 
have/* but this is evidently a modem word, probably borrowed 
from Marathi, and bringing with it the l|f of the Marathi in- 
finitive, which has no place in G. itself. If we derive ^nS^pi 
from Tif^> we must treat the if as part of the sufi^, as Taylor 

Digitized by VjOOQ IC 

FOBHATtON 6lt fH£ StEM. 45 

does, which is erroneous. It is really ^f^^ +* BWI. Moreover, 
the form of the word with the comlniiation ^ is that of a very 
modem Tadbhava, and the meaning is one of a sconewhat 
secondary kind, so that, aU things considered, the word can 
halrdly be admitted as genuine Oujarati ; and as I find no other 
example of the kind, I am induced to conclude that this suffix 
is, except in Tatsamas, confined to Marathi. 

§ 10. The last of the -a stems is tra. Words formed with 
this suffix indicate the instrument. In the majority of instances 
the T goes out, leaying only the t (see examples in Vol. I. p. 
337). Some words, however, preserve both letters by splitting 
the nexus, and Sindhi, as a rule, substitutes ?, which is pro- 
nounced tr, aiid as such is to be regarded merely as a peculiar 
method of writing. This class contains a large nimiber of 
common words, some of which are extant in two forms, the 
earlier Tadbhava and the later Tatsama ; thus, while ihf > Ttm, 
^ are ia common use, Chand iavariably writes ^, Thf, IfJTT, 
and from the Sapta9atakam we know that the r was dropped as 
early as Prakrit. Sindhi has a class derived from the stems in 
this suffix when preceded by i and forming ^, in Sindhi 
^ or i;5 ; thus : 

^TT^? ''a musical instrument," verb "inra ''to sound." 
^if^J " a beast of burden," „ ^T^^ " to carry." 

But there are not, I believe, any parallel instances in other 
languages, except those already given in the First Volume. 

§ 11. Much interest attaches to the stem which comes next 
in our list, both on account of the widespread and deepseated 
ramifications which it exhibits, and because one of its develop- 
ments is of the highest importance in the elucidation of the 
mystery of genders in some of the languages. The suffix in 
question is technically known as ZT^; but its efEect is to add 

Digitized by VjOOQ IC 


^ d to the root of a verb, so as to produce abstract nouns, or 
nouns denoting the idea involved in the verb; as ^ "to 
endeavour/* ^^ "effort/* Closely connected with it is ''if^^ 
also leaving in, which only differs from TPl, in the class of 
verbs to which it is added, a distinction not at all important to 
our present inquiry, and not of very much moment even in 
Sanskrit. The Pandits, being rather oppressed with the amoimt 
of idle time on their hands, have employed themselves in mul- 
tiplying useless distinctions, which in this busy age we are 
forced to disregard. The suffix d, whether technically classed 
as ZT^or ^Rp, is practically the feminine sufix of the majority 
of nouns whose masculines in a have been discussed in the pre- 
ceding sections. It is important, however, to note that Sanskrit 
masculines in a form their feminines both in & and in i ; and as 
the rules for the adoption of the one or the other termination 
are somewhat intricate, dictionary-makers in most cases add 
the pratyaya in brackets : to wit, ZT^ or 1|n^ when the f em. is 
in 4, ^t^when it is in L 

In Tatsamas this suffix, of course, remains, as ^WT "worship,'* 
f^«^| " thought,** and the like. In Tadbhavas it is invariably 
shortened to a, which is, as before noted, mute for practical 
purposes in all the languages except Sindhi. This suppression 
of the long vowel is, in cases where a double consonant pre- 
cedes, compensated for in the usual way by the lengthening of 
the preceding vowel. A few instances have been given in 
Vol. I. p. 182 ; but as the question is really one of ^ the forma- 
tion of the stem, it will be better to give a full list in this place ; 
and as the words now quoted are of very frequent occurrence, 
the exhibition of a consiiierable niraiber wiU be useful, for the 
sake of the individual words as much as for the rule itself. For 
our grammar- writers, being more of the rule-of-thumb sort of 
people than philologists, have, especially in Marathi, been much 
exercised on. this point, in their endeavours to account for the 
fact that the majority of these words are feminine. They seem* 

Digitized by VjOOQ IC 


to consider that the final consonant has been the deciding 
element in the matter, and lay down, or attempt to lay down, 
rules for each letter ; as, for instance, that final ^ is feminine and 
final "3 masculine, evidently not being aware that neither ^ nor 
^ have anything to do with the matter, but that the words in 
which those letters now appear as finals are derived from Skr. 
words ending in d, which has been absorbed, leaving the pre- 
ceding consonant, no matter what it be, as a final; and the 
cause of these words being feminine is not any peculiarity in- 
herent in the consonant, which has now, .as it were by accident, 
become final, but results from the words having been feminine 
in Sanskrit and Prakrit. For the rule holds good for the 
modem Aryan group, as well as for their Romance and Teutonic 
cousins, that the gender of the ancient mother speech is faith- 
fully preserved, in spite of all changes. In Oerman much of 
the diflSculty which foreigners experience in determining the 
gender of nouns would be removed were they better acquainted 
with the forms of the Old High German. " Gender was, in the 
older language, easily recognizable from the form and method 
of declension of the word itself. When once we know the full 
Old German inflexion of a substantive, we can have no further 
doubt as to its gender. In our modem speech, however, these 
marks of gender have to a great extent been worn away and 
obliterated. Compare, for instance, *der Dom' (masc.) and 
*da8 Horn' (neuter), 'der Wind' and Mas Land,' Mer YogeP 
and *die Nadel,' etc., with the Gothic equivalents thaurn-us and 
haum, mnd-s and land, fugl-s and nethla, Mer Same' (or 
Samen), *die Staude' and *das Ende,' with the Old-High- 
German samo, studtty entV ^ In the same way in the modem 
Aryan languages, our doubt as to why ddnt should be mascidine 
and bdt feminine is removed when we look back to the Sanskrit 
daninis and vdrtd respectively. It would be well if those who 

^ Heyse, Lehrhueh ef. Deutsehen Sprache, yol. i. p. 443. Compare also the forms 
of tlie Gothic declensions at p. 96 of the same yolume. 

Digitized by LjOOQ IC 


Write our Indian Dictionaries for ns wo]ild put the original 
word in a bracket for our guidance. Unfortunately they flo not 
usually know the original themselves. 

My list, which is only a specimen, and by no means ex- 
haustive, is as follows ; it consists of Skr. feminines in d, 
irrespective of the pratyayas by which they are considered by 
native grammarians to be formed : 

Skr. ^nJT "wool," H. ^T^T, P. ^ff , S. ^^y but G. ^R is n. 

Skr. ^ "bedstead,'' H. 15Z, ^TTZ, P. ^, S. ^Z, G. M. ^TTZ- 

Skr. ^ftWr "iguana," H. iftf , P- S. tU, G. 1^. 

Skr. ^fT^ "shadow," H. ^, ^f(ir> Wtf > IfT^* P- ^> ^BfN, S. 

Skr. ^Nn "leg," H. ^5^1^, G. M. tU, P. ^N, S. ^. 
Skr. ftfST "tongue," H. ^ft^, P. G. M. icf., S. f^. 
Skr, -^ "ddb- grass," H. ^. 
Skr. ^p^T "vine," H. J\^, P. id, S. TRf. 

Skr. fif^ "sleep," H. ^'l^, P. id., S. fif^, G. l^, M. ^, ift^. 
Skr. '^tirt" pain/' H- "^S^* Old-H. I^T' P- "^5^' «1^ *^<* *■ ^^^ 
rest iftlT- 

Old Hindi confounds ^ and ''J^ ; thus Chand writes : 

^It! ^ ^rr^ ^% ^Oa i 

" "Who knows, mother ! the pain of a barren woman? 
The dart of a rival wife pierces the body ! *' — Pr, R, i. 178, 

Skr. Ill^l " nlother," Old-H. ^FTPT, as in the line above quoted, ordi- 
narily ^, HT^ and in^. 

Skr. THWr "garland," M. and G. TfTBo- 

Skr. t:^ "ashes," H. rj^, P. G. M. id., S. I^IT- 

Skr. T^IOT "widow," H. -^InJ, G. M. O. B. id., P. 1^, S. "^ (see 
Vol. I. p. 299). 

Skr. ^5gx "streak," ^^g, P. G. id., M.^^, ^. 

Digitized by VjOOQ IC 


The Sindlii ^^ means the first streak of down on the cheeks 
of a young man, and may be derived from ^^, in the same 
sense that the Persian and TTrdu poets use la^ "line" in the 
sense of whiskers or moustache. Thus, to take an illustration 
from k popular Indian poet, Wall says : 

^ L5^ ^j^ ^r" v^ l/ \s^j^ 

*' The moustache fears the mirror, 
As the thief fears the watchman." ' 

Skr. ^r^n "shame," H. ^TTH, G. M. O. and Old-B. id,, P. ^TWT, S. 
^^, The form H<4j|| is also in common use. 

Skr. ^tTT "kick," H. WTW, B. O. M. G. id., P. 1^^, S. ^Rf , B. 


Skr. ^f^n <'rein," H. ^PT, and so in all. 

" When Prithir^j the King turned rein, 
The heavens stood still, the earth trembled, and the earth-serpent." 

— Chand, iV. J2. xx. 33. 

(g^ is for ^ "heaven,^^ fipr^^^I^Tf "sky,*^ \gr^=^5|^ 
" earth,'^ VHcf = "^IJc^ " falling to pieces,'* and •fl^ is the ser- 
pent Sheshnaga, who supports the earth on his head; or we 
may take \r^ and ^iTR to be separated parts of a compound 
^r^ifXT " the earth-serpent.'*) 

Skr. "J^nfr " speech," H. '5^T<f » P- G. id., M, also, but rare, S. ^Tfif • 
Skr, f^f^n "«gle marmelos," H. ^, P. M. B. O. id. 
Skr. ^l^ « bed," Old-H. H^fTT, H. %g|, P. id., G. M. ^. 

^ Page 8, line 13, of M. Garcm de Tassy's beautiful edition of Wali (Paris, 1834). 

YOL. II. 4 

Digitized by LjOOQ IC 


Skr. "ffim "hall," H. ^TW, P. WKf »• M. ip3S» B. O. (vulgarly) 

Skr. fiprr "stone," H. ^RfT, P. ftr35, S. ftT, M. l(te> B. f|pf , 

o. ftias. 

This word in the modem languages is generally restricted to 
the meaning of a peculiarly shaped flat stone on which spices 
are ground for the native dish "curry." In Sindhi, however, 
it means a "brick." 

Skr. ^[1^ << elephant's trunk," H. ^, P. ^, S. ?|f%, M. ^ITT* 
G. B. O. ^, G. also ;|^ and ^9. 

Skr. ^TJn "evening," H. HtlJ, P. #ff, S. ;et?ft, ^NjT, but also 

^4^f G. ^^fr, M. id., B. o. ^rnj- 

Nearly all of the words in the above list retain the feminine . 
gender throughout all the languages; but this point will be 
more fully dwelt on in the next chapter (see § 36). 

Besides words of the class given above, there is an extremely 
numerous class consisting of abstract nouns, which may be 
formed at will from infinitives of all verbs by dropping the 
final syllable, and they then convey the sense inherent in 
the verb. Thus in Hindi ^TTTTr " to beat,** and UTT " a beat- 
ing.** Thus they say ;j^ ^ ^^ ?nT TTTT "he beat me a 
great beating." It would not be correct to say that these 
abstract nouns were derived from the infinitive ; on the contrary, 
in respect of formation, they stand on the same level with it. 
The original Sanskrit root i|^, for instance, forms two nouns ; 
by adding U^ (^) it forms the abstract (l\^\ " pain,** and by 
adding 'g^ or ^Z (^1) it forms iftlTif " the act of paining^' ; 
from the former we get ift^ "pain,** from the latter, liH^i^f 
" to paio,** infinitive of the verb.* 

1 See § 9, (1), (/9}, feminines in ak. 

Digitized by LjOOQ IC 


Some few of the commoner pairs of words may here be set 

Hind} iriT " a beating," WX^ " ^ beat." 

„ ^^ " a running,** ^^^1 " to run.'* 

„ HW " an error,** ^'S^ " ^ «"^-" 

» ^n " ^ rising,** ^7^ " to ascend.** 

» ^TCTI " a stopping," ^^41 [ " to be stopped.*' 

It is these abstract nouns which are used with a long list 
of auxiliary verbs to make the compound verbs so common in 
all the seven languages, as iTTT ^TWIT "to kill/* ^n ^TTT 
" to mount upon a seat/' and the like. In Sindhi all the verbs 
are capable of being used as abstract nouns by the rejection 
of the final syllable of the infinitive : as in these instances from 

^rnj *' wakefulness,** ^TPTS " ^ be awake.** 

. ?r^ " pardon,** TT^IIF " to pardqp.** 

^re " strength.** ^I^W " ^ be strong.** 

In Marathi also there are numerous abstract nouns of this 
sort, with which may be joined the corresponding words in 

M- Wnm " adherence,** Wnf^ " ^ adhere.** G. ^TPT and WHT^ • 

,, -^'^fuUness,'' Hlf^^tofiU.** „ If^ „ HTf- 

„ ^TW" motion,*' ^ra^ "to move.** „ '^TW w ^TRRf- 

» ^rr^" running,'* \8|T^ " to run.** „ VT^ „ ^sfT^ (poet.). 

Instances may also be found in quantities in the other lan- 
guages, but it is unnecessary here to adduce them. The 
formation of these abstract nouns in some cases necessitates 
the lengthening of the radical vowel of the root, and in cases 
where that vowel is i or u, it is changed into the guna vowel. 

* Sindhi Orammar^ p. 46. 

Digitized by LjOOQ IC 


This process is more clearly seen in Sindhi than in the other 
languages ; thus we have — 

f)|^ ** snatchiog," 1|EV ** ^ snatch." 
Htir " error," ^If^ « to err.?' 

Thus also in Hindi, where the verb WRT corresponds to an 
abstract noun WPT, and ^TOfT to ^TO; tH^IT to ^<^; and 
^^l^T to ^ft^. As usual in these languages, there are very 
many of these abstract nouns which it is difficult to trace back 
to any Sanskrit root ; the principle, however, is the same in all : 
when once established in the popular mind, it was by degrees 
extended to words in which it had ijo business to appear. 

§ 12. The group of stems ending in i consists principally 
of e, ni, and ti. The former added to verbs composes abstracts 
or appellatives ; but the final short e is in most cases rejected in 
the modem languages, except Sindhi. Thus, taking all three 
suffixes together : 

Skr. ^ifq " fire," H. ^r^T, M. G. id., P. ^VRT, B. ^nftiT aod ^Rfipif , 

Skr. flfif "hurt," H. fj^, S. flftj, G. id. and fT^, M. fnj. 
Skr. ^fif "earth," H. ^, ^, ^, P. id. and h1^, S. ^, Wf, G. 
Ij, ^fr^, wr, M. ^, B. ^. O. id. and ^. 

Skr. Xjf^ " night/' H. -^Jcf , P. G. M. id., S, -^tfif , B. O. -^tfif and -^Tfcf . 

See also examples in Yol. I. p. 315, as gfs, ^rffe, "fffe, and 
^he like. 

Skr. ifVfif " song," H. i(t^, and so in all, but S. iftlj (m.). 
Skr. ^nf?f "caste,*' H. ^cf, and so in all, but S. B. O. ^TlfTT- 
Skr. ^|f?! "memory," H. P. WK^t- 

Skr. 7Tf?| "opinion," H. J(^, P. id., and so in all, though Hfffl is used 
in literature. 

Digitized by VjOOQ IC 


In this class also the gender depends upon tliat of the 
Sanskrit, and has nothing to do with the consonant which may 
happen to be left final by rejection of the vowel. 

In Sindhi there is a class of words, not very extensive, 
formed with the suffix ti, which expresses abstract nouns, 
haying also a verb of the same meaning. Trumpp's instances 
(Gr. p. 49) are: 

l9TTf?T "expense,'' verb l§l(l|f "to spend/' 
^n^f^f " pleasure," „ ^miB " to please." 
^q^^f?| "remission," „ 5^^W "to remit" (Old-H. ^^fTT* 

Mod.-H» Stn^f^O* 
^l|rf?f "behaviour," „ ^5^f^"togo." 

So also the double noun ^9mf^ ^rWfiT (literally, "coming and 
going") "income and expenditure." 

In the other languages the short final i is generally rejected ; 
when retained, it is mostly lengthened to L An example of the 
former is the curious H. word ^9innT or 'W^, which now means 
"brokerage," " commission." This occurs in Chand. I. 3, in the 
form "^I^Ht, and with the meaning probably of "increase." 
There is no modem verb from which it can be derived ; but it 
is perhaps to be referred to the Skr. root ig^J ridh^ " to increase,** 
through a form "VjffiT- In the other languages the corresponding 
word is S. 'Wfn (/.), G. '^sa^Jt (/), M. '^riTf , 0. B. Wlcf . 

With long i we have — 

H. P. ^rot "profit," "rise of prices," from ^OTT ".to rise," M. id., 
G. ^^ "rise," "advance," S. ^^f^, O. '^ifTf, B. id. 

H. Wnft "assessment,'^ from WPTT "to be attached," M. WRi, G. 
^^n (/.) "intimacy," 8. IT^, B. O. WRf?T, Wrft?T. 

H. ^Wift " movement,* from ^^flPfT " to move," M. id. " influence," 
P. ^W^y 8. ^Rjft " custom," "habit," O. ^fWtfn, B. id. 

H. P. inRft " settlement," from 1|^RT " to dwell," M. ^(irf?! , G. '^^H\* 

Digitized by LjOOQ IC 


3* ^lj^'> with totally uncalled fbr anuswara and softening of ?( to 1^, 
probably induced by a belief that the word was the feminine of the present 
participle active, as in English ''a dwelling," 6. 'QtHd, B. id. 

H. ^ITTft' "filling," "completion,"^ from HTTT "to ^^'' ^- w^-» ^^- 
H?7ft, G. id., S. H^, O. B. IS^. 

The words of tliis form are not, however, universally com- 
posed by the addition of the suffix fif ; for instance, H. VT^ 
"the earth,^^ is from Skr. "^ft^ "*^® supporter/^ Thi^ 
suffix is commonly, but erroneously, added to Persian words, as 
^^^ "deficiency," from S "less.** Here may also be men- 
tioned those two excessively common vulgarisms ^'jdsti" and 
'^pancasti" so perpetually in the mouths of the lower classes. 
Panvasti, meaning "protection," "favour," is used instead of 
the correct Persian (Jiijjjjparwarish, and is perhaps derived from 
th^e participle ^^^jy^ parwasta, "protected." But '^jdsti/^ which 
is used instead of ii^\ij ziy&da, " more," defies analysis ; and I 
have never heard any attempt to account for it : it is perhaps 
in some way corrupted from (^^^ ziy&datL 

The labial vowel is found throughout every branch of these 
languages in strict parallelism to the palatal, here also it is so ; 
there are stems in «, nu, tu, and ru. They present, however, no 
particularly noteworthy peculiarities. Common examples are 
as follows : 

Skr. ft^ "a drop," H. ^, ^, Of^, P. ^, S. ^^, ^, M. 
Skr. Trg "wind,'' H. ^TR, ^, B. ^r€t, O. m^, M. ^m, G.^, 
Skr. Wrar "arm,'' H. irff » P. S. id., G. iffff , M. id. and ^HT. 

B. O. m^. 

^ Yulgaxly used for "enlisting in a regiment," "engaging in a seryice," with 
Mi<,4^1 when used of the person who hires or engages the soldier, and with ^TfT 
when speaking of the soldier himself. 

Digitized by VjOOQ IC 


The general rule for these groups is that the final short 
vowel is in a majority of cases rejected by all the languages 
except Sindhi; and when retained, is generally lengthened. 
Cases occur in which i is substituted for u, as in bindi for vindu. 
Of the. common suffix ^ng or "W^j I treat in § 16, because there 
are some peculiarities in its use which render it both uncertain 
ill origin and partial in application. None of these suffixes are 
used in the modem languages as additions to verbal roots, so 
that they do not come under the head of g^uine modem 

§ 13. As in the case of words ending with the short vowels 
of the labial and palatal organs, there is a tendency to lengthen 
in order to preserve them, it is natural that the nouns ending 
in the corresponding long vowels should, as a rule, retsdn them. 
Long f^ is the termination of a nimiber of difEerent classes, 
which will be detailed in § 18 (1), and in the next chapter 
(see Chap. II. § 33), where the subject comes more fully 
under discussion with reference to gender. Long ^ ii is repre- 
sented in 

Skr. ^"wife," H. ^^, P. ifflf, M. S. l^T. »• O. ^Iir, W^, G. 
7f^ ; these three, .heing careless of quantity, shorten the vowel. 

The monosyllable ^j" "eyebrow,** undergoes considerable 
changes, as H. ^, P. ^jfff. S. introduces its favourite i in 
fli^, while O., on the other hand, inserts u in ^Jlg:. G. H^> 
M. ^t^ and ^St^ are probably derived from some formative, 
rather than from H^ itself. ^ " earth,** has been illustrated 
above, under ^fif (§ 12). 

§ 14. The termination V of a large class of nouns in Sanskrit, 
where it is preceded by TI, represents an older form ^ITT^, the l[ 
of which, though rejected in the Sanskrit nominative, still sur- 
vives in such Old-Hindi forms as ^i^TnT " ^ doer,** and is found 

Digitized by VjOOQ IC 


in all the oblique cases of the Sanskrit noun. At p. 165 of 
Vol. I. these words were somewhat cursorily noticed. It may 
therefore be as well to examine the whole subject more ftdly 

As the noun in all the modem languages takes its form &om 
the nominative case of Sanskrit, and omits entirely the gram- 
marian's fiction of a separate base-form, it would be expected 
that in this class of stems the groundwork would be the 
nominative in a, as ^mftl; but this is the case, strictly speak- 
ing, only in Tatsamas. The large and important class of 
words denoting relationship and professions exhibits numerous 
difierent forms. 

In Prakrit there are several systems ; the simplest and com- 
monest is the substitution of dro for n, as ^nrnct" husband,'' 
Skr. ^^1^. In this case the Prakrit merely perpetuates the true 
old Aryan nominative, rejecting the corruption which has 
taken place in classical Sanskrit into n. From the analogy 
of the cognate forms in allied languages we see this, as Latin 
dator^ Greek Sa}Ti]p, which postulate a Skr. ddtdr; the final o 
in Pr. arises from its custom of requiring a vowel-ending, which 
leads it to attach a vowel to Skr. nouns ending in a consonant 
(Var. iv. 6, 8), or to reject the final consonant itself. From 
this form arises the Old-Hindi form ^RTTft! quoted above, still 
in use in modem Panjabi W^TfTT ^-nd S. ^^|^. 

Prakrit follows the Sanskrit in shortening the vowel in the 
familiar and much used words denoting relationship, finf 
"father," ^J^ "brother,'' WRTT^ "son-in-law," which stand 
for fi|?n[, ^^nr^, WWTcr^, respectively, as is shown by their 
making in the other cases ftlW^J^, not frnTTTI* ®*^* Thus also, 
while Lat. has dator, datorem^ it has patir^ patrem ; and Greek 
honrip^ Samjpa^ but Trarfip, iraripa, and irarpo^ ; Prakrit has 
ft'Vft, HTinct, ^l* IT ^O (Var. V. 45). In these words, how- 
ever, there is also the contracted form fxf^, Vil^» and this 
is apparently the only form permissible in the corresponding 

Digitized by VjOOQ IC 


feminine ^TW (HFTJ) "mother/* whicli makes TTHIT. This 
latter is the form in use in Pali, as fxnn> mV\, WnTT.* In the 
oblique cases comes out a form in u, thus 

Nom. Plural ^^^ for Skr. V^StK** 

Instr. Sing. H^pTT » ^^' 

Loc. Plural H^ » 1?^ (Var. v. 33). 

This rule is not extended by Vararuchi to nouns of relation- 
ship, though in Pali the u form occurs in the genitive sing, and 
plur. as friTjW, ^n^JW,' pi. ft^> ^^T^^^ ^^' 

There are then in the mediseval or Prakrit stage three types 
of this class of nouns : first, that in dro, shortened in nouns 
of relationship to aro ; second, that in d ; third, that in u. 
^0 one of these forms nms through the whole series, or is 
found in every case of the noun, except perhaps the first. 
When discussing the phonetic changes of ^ (Yol. I. p. 159), 
it was shown that though this vowel migrates into u but rarely, 
and principally in words which already have a labial consonant 
adjoining or preceding the vowel, yet, that in the modem, and 
probably to a great extent in the mediaeval languages also, it 
was often pronounced as rw, so that we might expect to find 
this formation in u somewhat common in the modem languages. 
The Pali forms pitu, bhdtu, etc., may thus be taken to have 
arisen hoixipUru, the vulgar pronunciation oipitriy though it is 
also possible to derive them from pitaru, shortened trom pitaro. 
This latter derivation is, however, rendered less probable by the 
fact that Pali has this nominative in u for words which retain 
the older and fuller form dro, as kattu, for ^ " doer ;" satfhu, 
for ip^ "ruler,'* where the elision of the long vowel would 
seem to be too violent a supposition. 

^ Grammaire Palie de EaccHyana, par. M. E. Senart, Journal Asiatique, sixth ser. 
Tol. xyiL p. 220. It is Eaccdyaua's second chap, on nouns, rule 39. 
2 lb. ii. 40. 

Digitized by LjOOQ IC 


In the modem languages the termination u or 2^ is common. 
Examples are : 

Skr. ftm " father," P, 8. fini, and occasionaUy in Old-H. 
Skr. iri^ "mother," P. J(V^, ifpj, ;irt^, int^ ^> S. JH^. 
Skr. ^J^ "brother," P. ^sTR, ^n^, S. HT^, M. HT3f • 
Skr. ^ "grandson," M. HJ^. H. IfTift- 

The other languages, however, have f^ in some cases, as in 
H. m^ Wt€\) etc. The word for a "barber** may be intro- 
duced here. In classical Skr. its form is TTft7T> hut this is said 
to be from an older form inftTTT for ^rrf^?n (^WTHi^), agent of 
causal of ^T, in the sense of "to cleanse.*** It becomes inBT 
in M., but •f'rt^ in all the other languages, except B. and O., 
which retain the form «||fi|<f. Marathi, Sindhi, and Panjabi 
are, it will be seen, the languages which mostly affect this form 
in u. Hindi generally exhibits that in I or d. It is followed in 
most in the word below. Skr. i^f^m "worshipper,** H. M^lO^ 
^d so in all, M. also "gWTCT, and S. "WTXt- The latter is 
referred by Trumpp, erroneously as I think, to the suffix dru 
(see § 15). The Prakrit form would, we may suppose, be ^WO" 


This is again one of those cases where confusion arises from 
three or four different pratyayas, whose forms were quite 
distinct in Sanskrit, having by phonetic changes all come to 
have the. same form in the modem languages. Thus a word 
ending in e^ or il may either come from the pratyaya u, as 
kdru, "a doer,** or from uka, as kdt&, "cutter,'* or from 6A», 
as jdjarity "watchful,** or from n, as nd^6, "grandson.** It is 
not possible in each case to decide which of these terminations 
is the true one ; and in many cases it may be safely asserted, that 

^ In all the ceremonies of the Hindu religion in the present day a preliminary 
shaying hy the harber is a necessary part of the purification which must be under- 
gone by the celebrant. 

Digitized by VjOOQ IC 


uoT at having come to be considered the usual termination for a 
large class of words of agency, the vulgar tacked it on to aU 
sorts of words, as was seen a few pages back in the case of Q-. 
?a^^, where it is added on to the foreign word ^/^, without 
any regard to the hybrid nature of the word thus produced* 
This habit is common to all languages, and may be paralleled 
by instances in our own, as " starvation,** where a Latin tertni- 
nation has been unceremoniously tacked on to a Teutonic verb 
"starve'* (M. H. G. sterben). It will not be necessary therefore 
to pursue this question any farther. 

In the majority of instances the modem languages have 
formed words of this class from the Sanskrit nom. in d, and in 
these cases there is nothing remarkable to notice. Such words 
are for the most part Tatsamas, and do not therefore enter into 
the current speech of the people very largely. 

§ 16. The dissyllabic suffixes in Skr. are athu, dlu, and wAnw. 
The first does not seem to have left any traces in primary stems, 
though under various modifications it appears as the foundation 
of secondary steins in several languages. 

The second, dlu, is extremely common, both as a primary and 
secondary. An allied suffix is dru, and from the close connexion 
between the two, it comes to pass that a form with a cerebral / 
is in general use in the dialects which possess that letter. In- 
stances of primary words, according to the view of the Indian 
grammarians, are the following ; though they seem to make into 
primaries, by deriving them from almost imaginary verbs, many 
words which are strictly secondaries derived from nouns. As I 
said before, it is not worth while to stick very closely to this 

firgr^ " sleepy,*' H. Olj-I^, S. fi|4l0» G- ft?T^» M. Vn^i^^ , 
B. O. fM^I^ (rare). 

4^1^ "merciful," H. B. P. 0. ^«||^ and ^^THVy M. ^illcb) ^^l<^> 

G. id., S. Tm%^ 

Digitized by VjOOQIC 


This termination is of frequent occurrence, and is one of those 
which are attached to all sorts of words, without regard to 
origin* The common, and often noticed rule holds good here 
also, that when a people have once got to feel that a certain 
termination carries a certain meaning, they extend its use to all 
words in their language. Thus, from modem verbal roots come 
tlie following : 

Yerbia root |pr^ « quarrel," H. QiTlT^ " quarrelsome," M. G. <)J[>^ic^ , 


Verbal root ^ " fear," H. Hnj^ " fearful." 
Sindhi, as usual, changes / to r : 

^ > " pedlar," verb ^SYTW " ^ seek." 

ft^rnct "cotton-carder," „ ft^" to card." 
Marathi is particularly rich in. words of this type, such as 
^^^Idb "pitiful," verb IRVf^ "to moan." 
^I^ldb " itching," „ ^TPII^ " to itch." 

A long string of them will be found under secondary formations. 

The third suflix, ishnu, is of very rare occurrence even in 
Sanskrit, and I have not observed any words which can be 
referred to it in the modem languages. 

With regard to '^^^y ^p^, ?fi^, and l[i^, there is also very little 
to be said. The first three are similar in treatment. Masculines 
of this stem form their nominative in d, neuters in a, the modem 
languages accept the nom. as their type. Thus '^Tf^ " king/' 
nom. \m\y which is the form in use in all the modems. In- 
stances are : 

Skr. iTITn^ "name," nom. m^f H. 9|Ti(, and sain all. 

Skr. ^mm^ "birth," nom. ipij, H. ^K^^T^, P. ^PR^» WWt* 

Skr. irffn "festival," nom. ViHf H. 11^, P. B. 0. id., M. G. irf, 

Digitized by VjOOQ IC 


Skr. ^|4|^ "rope," nom. d|4i|, H. ^TR» ^Fl> <uid so iu all. 
Skr. ^<A«^ " skin/' nom. ^p|, H. t|J4|, ^^4(, and so in all. 
Skr. ^?f^"l6ve," nom. ^9f, H. Jfm ^Tf, and so in all. 

Noims in T^C form their nominatiye in f^, in which they are 
regularly followed by the modems. As this suffix will be 
more fully discussed in several other places, I omit instances 
from this section. 

§ 16. The stems, or themes, or bases, for all three terms 
are used by various authors, hitherto discussed, are all distinctly 
traceable to Sanskrit stems. But there are in the modem 
languages, with their rich and varied development, numerous 
classes of nouns whose terminations point to a common source, 
which yet cannot always be distinctly referred, in a manner 
admitting of no doubt, to either a Sanskrit or Prakrit original. 
Others again there are, which, though they can in some in- 
stances be brought back to Sanskrit, are only of partial applica- 
tion, being found in some languages, and not in others. It 
must be remembered that it is only in one language out of the 
group that any attempt has yet been made to classify or analyze 
these formations. In the rest the grammarians simply give 
rules for the declension of nouns, without troubling themselves 
to explain how the body of the word was formed. Only in 
Sindhi have the valuable labours of Dr. Trumpp put me in a 
position to tmderstand the formation of the noun in this least 
known of all the group. Often from this exhaustive work light 
has shone into all the languages, and I cannot too often or too 
fully acknowledge my obligations to it. It follows, however, 
from what I have just said, that it is impossible at present for 
any one writer to carry out to the full the somewhat minute 
system of classification that has been observed in the foregoing 
easily recognized classes. The Indian languages in this respect 
folly establish a right to be considered the equals of their 

Digitized by VjOOQ IC 


Eomance cousins, in the number, variety, and expressiveness 
of the derivatives which they have formed out of their own 
native resources. Frequent^, too, they have adopted a form 
of noun from Sanskrit, and finding it useful and convenient, 
have extended the principle to their own Tadbhava or Desaja 
verbs, so that we know them only to be primaries from the 
existence of such verbs. All the forms that I have as yet come 
across I now group together in this section. 

(1). f^ is affixed to causals to denote "the wages or cost of 
doing anything.'* 

Hindi ^jflT "to wash," causal ^Jt^TTTTi nom. VtWrf^ "art of wash- 
ing," M. ^gwrt^. 

Hindi i\^{\ "to carry," causal (j^m|l||, nom. iftwrtj "cost of 

Hindi ^IZTT "to twist ropes," cans. iraTTTt »©««• W2T^ "hire for 
making ropes." 

Hindi 4|^«ll "to beat out," cans. 9|^|l|r, nom. If^T^ "hire for making 
ornaments of gold and silver." 

Writers on Marathi grammar teU us that this form is only 
used in words of Hindi origin ; but it may fairly be doubted if 
there be in Marathi any such thing as a word of Hindi origin. 
Molesworth — although he sometimes incloses in brackets the 
corresponding Hindi word after a Marathi one — guards his 
readers against supposing that the Hindi word is the original, 
and tells us that he merely puts down the Hindi, because it is 
the same as the Marathi, leaving it an open question which 
is the original. It is more consonant with what we know 
of the relations between these languages to suppose that both 
formed these words independently from the Prakrit. This view 
is further strengthened by the fact that a similar form exists 
in more or less frequency in all the languages of the family^ 
except Sindhi, which expresses the idea of '^ wages'' or '*ex- 

Digitized by VjOOQ IC 


penses" by a form in ^nit or ^Rnjt, as mentioned in § 6. 
Bengali and Oriya have numerous instances of this form, as 
litwrt^ " cost of carriage," but in O., from some forgetfulness 
of the original meaning, we often hear the phonetic expression 
?ft^ll[ W^ or ^mi^ i^. Panjabi sometimes inserts a ^, 
owing to the existence of this semivowel in its causal, and the 
same practice prevails in rustic Hindi, as ^?Ji!T or '^TTfTTT "to 
graze cattle;" P. H. ^4^||; or ^^Tft^ "wages of a herdsman 
or shepherd;" P. ^TTT^ "cost of pulling down a building," 
from ISTPIT " to demolish ; " twrt^ " wages for grinding," from 
fTOT^TOT " to cause to grind," and many others. G. also ex- 
hibits numerous words of this type, but alsa expresses the idea 
by a form in inj, the if of which arises from ^f , as in VflW^ "to 
find," H. qi^Ml (Skr. TTTTlf)- In ^<IHU| " wages of a herds- 
man," the older form would be ^^I^Hf, which is analogous to 
the Sindhi form in/"^nit, mentioned above. 

(2). "^ is employed after causals to denote an act of any kind, 
and is frequently written ^. In many cases the sense would 
point to a derivation from the simple verb, and in these cases 
we must treat the termination as ^KT^, and derive it by means 
of the Skr. affix ITJ or ^g, the 7f of which is preserved in 
Sindhi, but changed into ^ according to the genius of that 
language, and takes a feminine in i, probably from the u having 
been regarded as the ordinary Sindhi masc. nom. of a-stems, 
and not, as it really is, an organic portion of the suffix. Sindhi, 
however, in a few instances, rejects the 7f, thus coming more 
into conformity with the other languages. 

Verbal root inz^ "stop," H. ^ITZ^iT^, P. ^UZ^RT^, S. id., G. ^IZHT^, 
M. id. and ^HZ^ITRT. 

Verbal root ^^ "puM," H. fa^N > P. f^T^T^- 
Verbal root IfWH "c^use to melt," H. ITWPfy P* 195T^. 
Verbal root f^Wt " rub," H. f^^|<|. 

Digitized by LjOOQ IC 


Verbal root ^ifT "whirl" (active), H. ^iTPV, P. "WH^.* 

Verbal root ^pg "ascend," H. ^TTR "act of ascending," "rise," P- 
^>ff l"^, S. id., G. ^^97r?» M. id., but rare. 

Verbal root ^^ "cause to graze," H. ^^^f* ^' ^TTT* 

Verbal root fJSpn "to conceal," H. f^^tTHT. 

Verbal root V[^ "fall,* "alight," H. T^H^ "encamping-ground," P. G- 
u/., M. id. and XfTPTi O. TTIT- 

Verbal root f^ni " be sold," H. f^^J^ " sale," 0. f)^^. 

Formations of this class are so common in Hindi, that it may 
be said that every verb in the language may give rise to such a 
noim. They are less common in the other languages. In 
Sindhi the examples, besides those given ahove, are 

ySfJ^ " surrounding," verb ^Bf^HT " to surround." 
^IT^HT^ « jingling,* „ f^^T^ « to jingle." 

^ ^ [ " humming,* „ ^filT^ " to hum." 

Owing to the fact that Sindhi, when it omits the ^, as 
^IZ^rra> ^fiPRT^j writes the final vowel as ^, it is at times 
difficult to distinguish words of this class from those derived by 
the suffix ^Sirj, as noted in § 9 (4). The only distinction is that 
the latter class makes the u long. In Gujarati, where the dis- 
tinction between long and short u is very seldom observed, the 
difficulty of distinguishing is still greater. Fortunately, how- 
ever, G, frequently writes words of the present class, like H., 
with ^. They are not very common in G. Besides those given 
above, I find also — 

^<^ie| "determination," "certainty," verb "y^TT^ " to determine." 
Tf^^f "stirring up (a quarrel)," „ 4|x||e|^ " to stir up." 

> In special sense of a measure of land, originally as much as could be ploughed in 
a day, from the bullocks turning at each end of the furrow, much as we use the word 
"turn" in such expressions as "a day's tum of work;" a ghum&u would be "a 
torn of the plough.*' 

Digitized by VjOOQ IC 


This termination is rare in Marathi, though the syllable 
^ITSf is used in the formation of secondary verbs ; and it seems 
also foreign to the genius of B. and 0. The few words that are 
to be found appear to have been borrowed in comparatively 
modem times from Hindi, such as B. ^H^, 0. ^7r9> "an 
attack," H. ^(^J^ ; and 0., as in the instances "WKJ, VrWl, f^^, 
given above, generally drops the final if or ^. 

(3). Probably connected with the last-mentioned is the very 
common Hindi form in "^IWZ, with allied forms ^vn^Zy ^WZ- 
The fact that many of these words imply sounds of various 
kinds has led to the supposition that they arise from compound- 
ing the verb with the noun '^S[pfZ "noise," " sound," but this is 
not altogether probable. It might rather be conjectured that 
the process is just the reverse, and that '^sn^ is a mere ono- 
matopoetic word derived from the termination, on account of its 
having some resemblance to the sound. The word is used to 
express principally light and repeated noises, as ** tap-tap," 
*' pit-a-pat," and the like. The two forms, that with ^ and that 
with ^, often exist in the same word. Instances are — 

Hindi f^^^KJd " slipperiness," f^WWRT "*<> cause to slip;" 
fafii^liSd "fretfulness," fafil^l ^ "to vex;" ^WT^Z "itching," 
^^HMl "to itch ;" VT<t< | ^d "confusion," V< ^ <I^T "to be confused ;" 
M^n^ "melting," ^pTT^ "to melt" (active); ^ ^^[^ ^ d , ^4<^|fd 
"splendour," -cf^j^HI "to cause to glitter;" ^"RZ "plaiting," 'J^ 
" to plait ; " ^^i^^ " touch," ^fifT " to touch ; " fj^ff^ l lgd " tinkling," 
^JiRJifT«n "to tinkle;" ^?JWT^ "whining," ^IJWRT "to whine," 
" fret ; " ^ilTR? " stooping," ^ifTRT " to stoop." ^>^MiNi|||d " flutter- 
ing," dv^d^ifd " trickling," dd^md " tingling," mMiN<lijd " throb- 
bing," are all formed from verbs of the same sound. 

There is a very large class of these words in Hindi, both in 
dhat and dwat, as well as contracted into l|t^, as TflTT? 
" breakage," from g^^TPTI " to cause to break." These forms 

VOL. II. 6 

Digitized by VjOOQ IC 


all run into one another to so great an extent that it is very 
difficult to draw the Hne between them. Thus from ^^JTTT 
"to explain/' we have ^^iq^ "the act of explaining," but also 
4tn^, i.e. g^JT^nfV, with the same meaning. The origin 
of these forms being unknown to the masses, it was natural that 
they should confuse any two of them which had at all a similar 
sound. Moreover there is some obscurity about the exact deri- 
vation of this and the preceding form in ^SIHT. The presence 
of long d does not necessarily prove that these words are taken 
from the causal verb in all cases. In some instances the sense 
leads to a derivation from the simple verb, as in '^TR, M>^|^, 
which come from ^OTT, tf^TTT, respectively, and not from 
^WRT> M^^HI. The long vowel must therefore be assigned to 
the suffix, and in cases where the word has, on account of its 
meaning, to be referred to the causal, it appears that the two 
long vowels, that of the verbal root and that of the suffix, have 
coalesced into one, without any further lengthening. 

Panjabi possesses a great number of words of this triple form, 
mostly identical with Hindi, as ^nr^T^Z, M ^Ml{^ > and ^«|<T^- 
From the peculiar softness of Panjabi articulation, the ^ in the 
second form would be very indistinctly heard, so that, writing 
according to sounds, the third form would represent more cor- 
rectly than the others the spoken word. Thus the nimierals 
JTEfil^ 71, ^TfZ 72, ^^ 73, ^tW^ 74, and the rest, sound 
generally ikdt, bdt, tet, chauL ^WTJ "father-in-law** is in 
most districts saurd. The Hindi- speaking people, on the con- 
trary, pronounce the 1| generally very clearly and distinctly, 
perhaps rather overdoing it ; so that they would naturally 
retain the forms in ^1^^, and in the eastern Hindi area WTO, 
where Panjabi prefers the shortened form WZ. 

Sindhi does not appear to have any words of this form, but 
instances of a similar and possibly connected form will be found 
in § 20. Gujarati has W^, WTO> and WZ, I do not find W^. 
Of the former, instances are ^apTRZ and iRl^ " fiction," ^TTR^ 

Digitized by VjOOQ IC 


"to make;'* ^^^{^<^^ "afEection/' ^Wt "*^ love;'' of the 
latter, xldb^l^ "brilliancy," 'RKT^ an incorrect spelling of 
Hm^l£ "confusion," 4N^v^ld "muddle," HiU|^U!ld "shudder- 
ing," "tingling," etc., from verbs of similar form. 

Marathi generally follows ways of its own in the numerous 
forms of derivatives in which it abounds ; in the present instance, 
however, it is found to be to a limited extent in accordance 
with Hindi. The class of words which I am at present dis- 
cussing appears in M. with the terminations ^TZ,, ^ITJ; 'fZ, 
^nz, and occasionally WZT. Nouns of this class are both pri- 
mary and secondary ; the termination, once established, having 
been extended by the vulgar to all sorts of words. This free- 
dom, or rather licence of formation, is very noticeable in Marathi, 
which in my opinion is far richer in nominal formations than 
any other language of the group, and it is therefore noteworthy 
as a specimen of the blindness and imscientific method of the 
ordinary grammarians, that they either dismiss the question of 
stem-building without notice at all, or dispose of it in a few 
cursory remarks. It is really one of the most intricate and im- 
portant questions of the whole subject, and if fully worked out, 
would demand a volume to itself. 

Undoubted primary formations in M. are the following : 

^ Cm^d "the state of beings much used or worked," "practice," 

" routine," verb ^Hiriq " to rub," Skr. ^Rllf . 
yj Tte'fZ " residing," " state of being settled in a place," verb ft?^ » 

"to dwell," Skr. if^if. 
» ^rrSRTZ " act of burning the weeds before ploughing," verb HT^Sf^ » 

"to parch," Skr. ¥|^. 
_._ ( ^fflpCZ I " conduct," " management," verb €|l^((j " to carry," Skr, 

^ ^^Md * * burnt in cooking," verb ^E|r^xf^ " to be burnt," Skr. iR^H} .^ 
^ From ^n^in its original sense of being spoiled. 

Digitized by LjOOQ IC 


WZ ^9^rC^ "troublesome,* "vexatious," "vile," verb l^^^^nf or 

^9q3E»!)f "to sink," "fail," "miscarry," Skr. ^Q^.^ 
» ^Tf^ " tough," " clammy," lit, " much chewed," verb ^TW^ " *o 

chew," Skr. ^ntOl* 
» %T^ "throng," "crowd," "crowded state of a street," verb 

^^nt^ " to squeeze,* Skr. .* 

y> %flZ "compressed," "flat," verb %q5f "to press," Skr. (Vol. I- 
p. 212). 
^WZ 'WTZ " dilemma," " scrape,* " impediment,* verb ^V^f^ " to yield," 
"faU" (orig. "to be impeded"), Skr. ^|^ "binding," "re- 
straining,* "impeding.* 
>, ^>€^><U "rattling," "grating," verb 4d4^>fi)| "to rattle" 

„ T^^Crra "confusion," "bustle,* "muddle," verb ^U^n^iK "to 
be in a muddle" (VoL I. p. 336), 

» MOIMUII^ " a Fal»" " Mging*" verb ^niT^rar^ " *<> "ng loM^y " 
linrT ^'WZT " bawling," "outcry,* verb IfRJ^ " to call"' (ori^n un- 

From the above examples it will be seen that M. agrees 
pretty closely with H., but that in several instances the form is 
adjectival; this peculiarity probably arises from the looseness 
with which these words are employed; they were originally 
substantives, but have passed over into adjectives by degrees. 

There are words in Oriya which end in ^^, but I am not 

^ In the primary sense of limping, jostling, and more generally going badly or 

' I take this word to be a formaliYe of a familiar character deriving its origin ulti- 
mately from the Skr. f^ "to heap," substantive ^TO, perhaps with itl "body" 
added, so that we should get a word ^^^TT for the more regular Skr. Iff^T^II' 
<* crowd of bodies/' Inversions of this sort are common in the modems, and not Un- 
known even in Skr. 

> The ordinary meaning "to drive" arises from the Indian practice of driving 
herd by frantic shouting and calling. 

Digitized by VjOOQ IC 


prepared to affiliate them to tliis stem. Of words in dhat, dmt^ 
ox At, I do not remember to have heard any^ nor are there any 
instances in the dictionaries. The formation seems foreign to 
the genius of the language^ as also to Bengali, though neither 
of these languages have had their depths searched out sufficiently 

yet, to enable a decided opinion to be given. 

(4). Hindi possesses a range of words ending in ^ and ^irr> 

which, though for the most part secondary, are also in some 

instances primary, though comparatively rarely so. The other 

languages have occasional analogous forms : 

xf^ifl "to mount," H. ^^^, ^I^WT "<>»« ^^'^ mount*," "a rider," 
P- ^nt^> "^^^f ^itlfT "a mounted groom," "a trooper," M. ^nTTlTf > 
5ffT^^ "horseman," "climber.* 

Vniin "to shont," H. ^Im « robber,* B. O. ^fnTTTT- 

The origin of this form will be discussed, together with the 
allied forms ^9^ and ^ft^, under the secondary formations to 
which it seems more q)ecially to belong. 

(6). A widely-spread group of stems is that in ^m^V, with 
variants ^ifN, ^NT, and "^unf , which may possibly be connected 
with stems in a«a, and differ from th^n only in the long vowel ; 
while, on the other hand, the full form ^m^V seems to run into the 
pure secondary form "qif, and has often a long vowel prefixed. In 
Bengali this form occurs as ^irfif , which points to the fem. ^RJT 
of other languages. I place this form among those of uncertain 
origin, because of the long vowel and the labial ; also because, 
like many of the preceding stems, it occurs both as a primary 
and secondary in some languages. All these peculiarities throw 
a haze of doubt over its origin and development. Examples 
are as follows : 

!H. Mlf 4^14^ ) '* a garment bestowed on guests at weddings or 
Trff^T^nV) feasts," verb Trff?JlfT " to clotbe." 
,y ftf^Fnft "cultivating land by stealth," verb fSpTRT "to 


Digitized by VjOOQ IC 


^UPR WFR^ " thatchiDgf," "an encampment," verb^ftTT " to thatch." 
despatch of goods," verb ^^^T^ "to cause to go." 


>» 'f^TW " a cake of cowdung used as a charm," verb «|«|«|| 

" to increase." * 
^^tt^ f^re^T " toy" (ue. f^RTRPfT), v«rb f^mV^ " to cause to 
f^n^TT " bedding," verb f^Q^TTIT ** to spread." 
» W^^ " a load of wood," Verb HTTTT " to cause to load." 

"^Unr ^ZTT " leisure," verb WZTT " to be released." 

It would seem proper here also to insert the word ^Flt^ or 
^^TST^ "a wooden slipper or patten," which should probably 
be written ^[^«f or i^>^|e|«|, from the participle ^^TT " stand- 
ing." It possibly owes its present shape to some fancied con- 
nexion or jingHng with a irf^ " foot." 

Panjabi has fSrajTOTrT "bedding," f^^ftnn "toy," IBfl^lj^ 
"encampment,^^ Mff^l'dun "present of clothes,^' and in general 
the same words as Hindi. In Sindhi this form is secondary 
only. Gujarati uses a form Wnnir> as ^f^TWlft' "dress given 
at a wedding ;" ^BTVITTIQ^ "congratulation,^^ and "congratulatory 
§ifts," verb ^nTRPt "to welcome ;^^ and a contracted form in 
t^TOT^ or •^ "bedding." Instances of the Bengali form above 
quoted are f^^TifT "carpet," verb fi^ri,1^ "to spread ;'* 
f«(JWl«n "spoiling," t^PRmit "to spoil ;^^ ^TTTf^ " interlac- 
ing,^' " wrapping,^' WITI^ "to wrap,'' "wind ;" ^iTSlfif "pack- 
^g ^P>'^ ^^r<^ " to pack." Oriya being in the habit of giving 
a sound of o to short a, uses forms t^WIT, f^RPTT? HTTT? for 
the H. f^re^TT, etc. The few words of this kind in Marathi 

1 From its supposed efficacy in increasing the amount of grain in the heap on 
which it is placed (see EUiof s Baces of North-West Provinces, vol. i. p. 235, my 

Digitized by VjOOQ IC 


do not seem to be the genuine property of the language, and 
are in many instances undistinguishable from stems in -ana ; 
they need not delay us. 

§ 17. Leaving now this somewhat obscure and imsatisfactory 
class of words, let us go on to the secondary stems. There is a 
vast mass of these, and several of the purely secondary suffixes 
have been capriciously extended to verbs, thus bringing the 
words formed by them, according to our classification, imder 
the head of primary stems. It will, however, be more con- 
venient to treat these all as secondary, merely pointing out as 
they occur the instances in which they have passed over into 
primaries. Being all derived from nouns by the class of suffixes 
called Taddhita, these secondary formations fall into two great 
divisions : the first, those which add to a substantive or adjective 
a suffix which converts it into an abstract noun, descriptive of 
some character, occupation or quality; and the second, those 
which by the same process create appellatives or attributives. 
In this section I include only the abstract nouns. 

(1). The first I shall take is the very common suffix T|«^, 
with its numerous variations. This arises from the Skr. suffix 
tt, which I have conjecturally connected with >Brra^^^ "self^^ 
(Vol. I. p. 330), and which passes into ^^ (also tHIF, Var. iv. 22), 
and, by the process described in the passage above cited of Vol. I., 
becomes J^x^ and xgiif . It has many different forms in all the 
languages, as — 

H. ijif, v[^, in. 

p. ^, j^rTj ^» ^^' 

s. TT^, ijifi*, lit, II, int, ift. 

M. mi, 1WIT- 
B. ijif, iniT. 

Digitized by LjOOQ IC 


Examples : 

Hindi. ^gjT "acid,* I^ZTT^ and •Vm "acidity ;" ^^STT "Bharpcr," 
^^SIM^I "sharper's tricks,* "fraud ;" ^^tZT "small," ^ZTTT "small, 
ness;* %9 "fop," ^W^^T "foppishness;" IfTW^ "child," ^TW^IRif 

"chudhood" (Chand imninr, *.«. ^ninnr); ww\ "great," ^rph 

"greatness;" ^ "old," ^ffTR and •in "old age;" WflRT "boy," 
1ir>nnir "boyhood;" ^fTf "dissolute,* l^l^miV "dissolute living;" 
Xtl "iHdow," 4,I>HMI "widowhood;" ^T^^ "ignorant,* ^1:^^11 
" clownishness." 

PanjaH. fiTH, ^J^ "old age ;* WfigiHT "boyhood ;* VJH^Siyim 
"fraud ;* IJ^I^I^yqfT "clownishness ;" ^[^^UfT "debauchery;" 111^4 U4I||> 
•'Q^» •"3^» •]prT "childhood;* ^^^\ "widowhood;" and, peculiar 
to itself, '^WK "boy," ^Jf^JJUIT and $%f<JjQT "boyishness;" 7nr( 
"stupid" (Skr. IjJJ), IJlfJUnr "stupidity ;" ^tH?J "boy,* If^^lTyirT 

Sindhu ^VT^^> •^Hlt* ^JWJt, ^Wt> l^'nt. ^[%4t, ^^P^» 
"old age;* "^If "widow," "^^TR^ "widowhood;" WTIT "woman," 
WTW^^ "womanhood;" VlfWfl "P«"<^^*»" ^f^^TOUft "panditship;" 
■^ra "seizing,* ^TT? "assistance;" '^fz "near," l^fz^ "partiality;" 
l|it "small," lUwrt^, iUi:mj, ^<%^t "youth;" ^rnjt "watchman," 
^lffMt» Trfipi "watchman's work;" nfvV " coflin-bearer," ^itf%ra» 
•xnj "duties of a kandhi;" ifT^ "man," <RTf^ipJt» ^I^HMUlt 
" humanity ; " flf^ " wild beast," Ri^MUH " bestiality." These are all, 
except one, from Trumpp, Sindhi Chr. p 61. 

Gujarati, Taylor's Grammar (p. 144) gives several examples which 
are not found in the dictionaries ; but as the author is a resident of the 
province, he has probably heard the words from the people round him. 
Edulji's Dictionary does not apparently contain more than a third of the 
words in the language ; and Narmada Sankar's, though much fuUer, does 
not g^ve all the formatives, except incidentally as explanations of some other 

Digitized by VjOOQ IC 


word. The words I have found are: U<>f1 "aged," Mi^MIQ; •^ 
"old age;" ihSt "blind," ^VT^lf > •^ "blindness;" IvN^Bt "dark," 
^JVIcblM^ " darkness ;" jjf^ "widow,* (^>f 11^ "widowhood ;" IRfnftt 
•1|m "old age ;" ^Ifdbt "inverted," ^4db>TM^ "invertedness ;" ^^llfri 
" dissoluteness.'' The word I|dbl41 " grief," " anxiety," " torment," seems 
to be derived rather from the verb Hdo4 "^o bum" (^^^^ than from 
1|9B "strength," and may be set down as an isolated instance of the use of 
this suffix as a primary. 

Marathi. ^W^ "good," ^fltqiT' •^TOT "honesty;" ^Ift^m "good," 
^rt'r^lWr* ^t^I^RHr "goodness;" %^ "chUd," ^TfT^WT^ •^TOT 
"chUdishness;" 'Smnf ''» woman in childbirth," ^lM< 1 MU| ' "attendance 
on such a woman." In M. also is the softened form ^1|f, ^Tnify ^^^ a 
form ^, ^,'^, corresponding to the in ^nd ift of the other languages ; 
as from ^tr "tbief," ^ftT'nil* •TTO. ^* and ^^tU^ " thievishness," 
or the "conduct of thieves;" ITfT"! "smaU," WfPF^ "smallness;" 
^ "great," ^t?:^ " greatness ;*• ^ntz: "bad," ^rTt;:3qirr "badness," 
says the grammar, but ^ll^^^llQl, ^nfts s^ys the dictionary ; ^f |IQ[ 
"wise," llfniFnrr "wisdom;" 4^MK1 "old,'' »;M1<MU1 "old age." 
The words formed with this suffix are not given in large numbers in the , 
dictionary, as it would appear that they can be formed at will from any 
adjective in the language. 

Bengali. ^TPIT* ^J^TTO "old age;" iptHnfT* •^WT "talent," 
«* virtue;" WCrnHIT "debauchery;" but the form, though occasionally 
heard in conversation, is not very common, as Bengali has another and 
commoner form in IfffiT for words of this class. 

Ort^a. Words of this form, as irT^I9^^> MHI^M* I» are .sometimes 
heard, but the form does not seem to be indigenous in the language, being 
very rarely met with. A genuine instance is Hfl9 "^ rogue," Vrfl9* 
"qfJipilT "roguery;" also ^Iff^Mf^^l "the duties or profession of a 
Brahman," such as studying the Sh^tras, performing religious ceremonies. 

Digitized by VjOOQ IC 


and the like ; f«ir^^(Mll| '< laziness ; " finHHIT ^or fq^nqm << dissolute- 
ness," from fq^M " dissolute." 

The two points to be observed in the treatment of this stem 
are the method of joining it to the root, and the terminations 
which it takes. As to the firsts if regarded as a mere pratyaya 
or sufl^, it should in the case of Tatsamas be affixed to the bare 
root, so that from ^ we should get 4^H^ = ^ct^ and this 
is what does actually happen, and so far the usage is regular. 
But when affixed to Tadbhaya adjectives which have taken the 
d or oxytone ending, this d would be retained, as in ^^FTt, 
in which case we must not consider the word as having been 
derived from the Skr. ^?t> but rather thus, that the termina- 
tion was regarded as a thing apart, as a sort of qualifying 
particle which could be appended to all adjectives at will. 
From the detached character acquired by this particle arises the 
peculiarity that it is in H. sometimes, in M. G. and S. very 
frequently, attached to the oblique form of the noun, because 
the nominative form is regarded as appropriate to that state 
of the noun only in which it is not subjected to any influence 
from without, but is either an agent or ^ mere indication. 
Directly the noun becomes subject to influences of any sort, it- 
passes into the oblique, and the addition of the syllable TJi^, or 
any of its variants, was regarded as subjecting the noun to an 
influence somewhat similar to that exercised by the case particles, 
and therefore demanding the oblique form. When we see the 
suffix added to nouns in the direct form, it would show that in 
these cases the form came into existence when the suffix had not 
yet won its detached character ; and the unsettled nature of the 
terminations of the root in all the languages points, in confirm- 
ation of this supposition, to a time when the languages them- 
selves were in a transitional state, and had not attained to fixity 
of practice. 

Equally undetermined is the termination of the suffix itself. 

Digitized by VjOOQ iC 


There are first the forms H. P. M. TTf (W> ^Hj) and JPTI, cor- 
responding to wliich respectively are the S. Vi^ and ini^. These 
two forms reproduce the barytone and oxytone forms of noiuis, . 
and if, with some writers, we derive the latter form from an 
addition of ^, we should have to suppose a Skr. form ?rti 
giving Pr. tHU^ or xcTHni > as however in words of the form 
^T^ in Skr. the accent is on the last syllable " sattwd,** it may 
be fairly reasoned that it would remain on that syllable in 
Prakrit also ; and as the suffix itseK has been conjectured to be* 
shortened from "^If^n^, the form in ipTT would be the original 
one, shortened as time went on into Jl9{ from f orgetfulness of 
the accent. The uncertainty in the method of affixing this 
ending, which is sometimes added to the direct form of nouns, 
and sometimes to the oblique, and at times even to a shortened 
form which is neither oblique nor direct, as in Sindhi, proves 
that words of this type came into existence at very difierent 
periods; and in those which were created in later times the 
accent would naturally have been forgotten, and they would 
take the form ^f. 

Secondly, the forms in and xft point to oxytones, and seem to 
be derived from some such form as Vf, omitting the HJ ; such a 
form may well have existed in spoken Prakrit, although no 
traces of it survive in the written works. This form would 
come direct from the Sanskrit ?t, whereas that in Tquf must 
come from an older Skr. ?p(^. 

Lastly, Sindhi has forms in^ and ^, the final vowel of which 
seems to be quite inorganic, as we cannot trace it back to any 
corresponding peculiarity in the older forms, and is probably 
due to an unconscious imitation of the analogous forms TfT^ and 
ift, which have been or will be discussed in their proper place. 
The Skr. word ^nw^ " self,** retains the n in aU cases of the sing, 
but the nom., and in this, as in so many instances in Skr. and the 
cognate languages, the nominative stands apart, and the deriva- 
tive forms are taken from the oblique. Thus Greek and Latin, 

Digitized by VjOOQ IC 


wliile faithftilly retaining the original type of ^ in 7roSo9> pedis, 
reject the d in the nominatives irovf; and pes ; yet it is from 
^o&>9 and pedis, not from irov^ and pes, that the deriyatives are 
formed, as well as the nomi itself, in their modem descendants. 
So also oBopT<y;, dentis, reproduce ^[^ more correctly than oBov^ 
or dens, and in deriyatives and descendants the former, and not 
the latter, are the base. The reason of this is clear ; the obliq^ue 
form being used seven or eight times to once of the nominative, 
sticks in the popular mind, and is used in grammatical forma- 
tions, while the classically modified nominative is forgotten. 
This was probably the case with the termination now under 
consideration ; the vulgar would know little of an euphemistic 
high-flown nominative ^H^, while they would be familiar 
with some ten or twelve forms having for their base ^RT^P^ ; 
the Old-Hindi poets use '^^TfiT or XJ^f while the modem 
language does not begin to use the classical nominative '?[T^ 
till after Pandits had begun to resuscitate the accurate form. 
So also, while the Brahmin wrote ^Prt, the people may have 
said ^Tt^PO or even ^?1T9l*t9 and thus the form ini or J^^ 
reveab itself as the older and more accurate. The ITOT type 
is still in existence ; in Old-Hindi we have ^^iTPni " virility,*' 
from y^miHI', Skr. ^ipf?!, and Sindhi has from ;r^ " a head- 
man,**^ f|(i9^^, Pr. <R^QRnif> where the S., rejecting one of the 
two consonants, does not, as H. would, lengthen the preceding 
vowel, but inserts instead its favourite short t. There may be other 
instances in the other languages, but I have not yet met with any. 
Sindhi has a few rare examples of a form in ^^T^ which 

' Dr. Tmmpp will forgiye me for pointing out a slight error in liis English here ; 
he t ranslatee this word "headsman/' not perhaps rememhering the difference 
between that and ** headman." The former means ** an executioner/' i.e, " one who 
beheads criminals." Dr. Trumpp's English is so excellent, correct, and graceM 
throughout as to command admiration ; it is in a friendly spirit that this little slip is 
noticed, as it might mislead an English student. The difference between <<heads-« 
man" and " headman" is slight indeed in form, but yerj wide ii^ meaning. 

Digitized by VjOOQ IC 


Trumpp refers to this suffix, as Q^HKI^' " boyllood/^ " youth/' 
from if^Hl^ " a boy ;" but it is open to question whether this 
is not the primary affix mentioned in § 16 (2), which has passed 
oyer into a secondary form in a few cases. 

As a corollary to the connexion which I see between ^l* and 
^IT^PC' ^ would here introduce the Bengali stems in mfiT, 
which I derive from ^IRiSr in such Skr. words as 4||f |fU|. 
Oriya uses TT^Tcni as a religious word, in the sense of " cele- 
brating the greatness or merits of** a holy place or festival, 
thus we have ipiT ^ffTTfif the "glories of Ganga,** ^ %!^ 
IVTTf^ "a description of the religious merits acquired by a 
pilgrimage to the shrine of Jagannath,** and others. Bengali 
has elided the ^ and softened the if to 1[. Instances are ^PTH 
" mad,** Ml< ! < gi lfil " madness ;** ^^ "debauched,** ^^^rf^ " loose 
living ;** ITtl " buffoon,** Ht^nf^ " buffoonery ;** J||^| "an ass,** 
lll^lfi r " stupidity ;** Jtzi " obstinate,** ^^^Zlf^ " obstinacy ;** 
^t^ " greedy,** Q^M|Tf4| "greediness ;** II^TT " base,***|f^irrf^ 
"baseness;** ire "profligate** (lit. "destroyed,** compare our 
use of " dissipated,** also the original meamng of " profligatus**), 
irerfif "profligacy.** In familiar or colloquial words which 
end in a vowel other than a or 4, the initial vowel of the suffix 
is elided ; thus %% "a child,** %%f7T "childishness ;** lli^*^ "an 
impudent boy** (French gamin), Xi^%f^ " impudence ;** and in 
vulgar speech the favourite u of Bengali exercises an influence on 
the following vowel ; thus from ^ " wicked,** ^^fH " wicked- 
ness.*' By a curious caprice, also, a duplicate form isoccasionally 
employed, as from ire we get ire?fT " profligacy,** and ifg^ini. 
Oriya also uses this form, as UTT " old,** ^7rf^ " precocity in 
children ;** r&Hl " loose,'* " lazy,** (inilfil " laziness," " inatten- 
tion,** B. id. It will be observed that this suffix is almost 
universally employed in a bad sense, as expressive of stupidity, 
loose living, or the like. One instance, ^[ITrf^y occurs occasion- 
ally in Hindi; but, with this exception, this form appears to 
be peculiar to the two easternmost members of the group. 

Digitized by VjOOQ IC 


(2). Of very wide use in Sanskrit is the suffix Jf, generally 
neuter when forming abstract nouns, though of all genders in 
appellatives. This suffix generally changes to "Ij in the modem 
languages, and in most cases the words become feminine, owing 
to the general use of ^ to express that gender. The typical 
word of this class is — 

Skr. '^ " theft," Pr. ^tfT^. H. ^t^, and so in aU, but O, ^tf^- 
So also 

Skr. ^rn^ "cheating" (^pTt "a cheat"), B. ^plij , H. Z^ (vulgo 
" thuggee"), and so in all. 

This termination is extended, as usual, to words of uncertain 
origin, and notably to Persian words, inasmuch as that language 
also adopts the suffix i to denote abstracts ; thus we have ^^^ 
"goodness," i^«A; "badness," ^^ "deficiency," ^^^^ "excess;" 
and even colloquially Arabic words take this ending, as ^^j^^^^ 
"sanction," fromjjlai^ " seen," "approved." S. has also if^ 
"badness," from it;^ "bad" (Skr. ^R^ "slow") ; H^ "good- 
ness," from H^ "good" (Skr. ^^). In many cases, however, 
the final ^ of the Sanskrit is elided alogether, as in 

Skr. ^Tn4 " fortune,* H. V[^ , and so in all. G. retains Tatsama ^fH^ 
and S. V^. 

Skr. TaTRT "rice," H. VTT* and so in all, except S. ^sTW dhdnyu or 

This latter word seems to have been originally an abstract 
meaning "possessing," "wealth," root ^^^ as stores of rice 
were, and still are in many parts of India, the principal source 
of wealth to all classes. 

Where, in modem formations, the adjective had acquired, 
as adjectives almost universally did, the termination in long 4, 
the i of this suffix does not supersede this vowel, but is attached 
to it, forming &L Thus we have a very large range of words in 
all the languages. A few are — 

Digitized by VjOOQ IC 


Hm "good/' H. H^ "goodness," P. H^T^, S. ^i^, G. M. 

O. id., B. ^TTwrt- 

f^^m "pure," H. fiflSwrt* »»^ so in all. 

-^^ "great," H. ^^^ "greatness," P. ^ff^Uft;, S. ^it, O. B. 
O. id, ; not in M. 

^[^ "high," H. ^JNrrt; "height/* S. id. ("goodness"), P. ^^tI;, 

This form is not very common in M., but is of almost xmi- 
versal application in tlie rest. P. anomalously inserts short i 
before the termination, but this is a dialectic peculiarity which 
does not prevail in aU parts of the province. It has already 
been noticed that so many Skr. terminations fiise together into 
I in the modem languages as to render it difficult to distinguish 
them. The form now under consideration is restricted to ab- 
stracts, and has no connexion with any other. 

(3). The Sanskrit suffix J[\ exists in most of the languages 
in a large number of words, but occurs chiefly in Tatsamas and 
modem Tadbhavas ; so that it would be more correct to suppose 
that it has been resuscitated, together with the word to which 
it is attached, than to speak of it as having come down unin- 
terruptedly from early times. 

Skr. f^TH "firmness," H. f^cH, P. ^PM^dh 8. fj^fdl . In 
the rest ^^cfTy generally pronounced drifhatd in H., and so written also 
in P., but in O. it sounds drufhatd, 

Sindhi has a few words not strictly Tatsamas, though only 
slightly altered, and it occasionally adds ^, making TfT^, as 

irfe "deficient," Mfddl' ''E|fe?n^, "deficiency." 
^ftj " fit/' Wtform " fitness." 

In B. and M. words of this class are extremely numerous, but 
are all pure Tatsamas, and as such do not strictly come within 
the range of development of the modern forms of speech. 

(4). Another conmion suffix is 1[in^, which is used in Bengali 

Digitized by VjOOQ iC 


in its Tatsama form^ but rather among scholars than in the 
mouths of the people, as also in M. 

'^^ "red," B. M. "^f^tSTTT ''redness," and so in all the others, but as a 
rare and pedantic word. 

^Hr ''white," B. M. '^f^RTT "whiteness," and so in all the others, but 
as a rare and pedantic word. 

Trumpp affiliates to this suffix a large class of words in S. in 
^mf^ and ^niCftr> supporting his theory by allusion to Vara- 
ruchi, V. 47. This passage merely directs that certain words, 
such as brahman, t/uvan, adhwan, are to be declined like Atman ; 
thus, vamhd makes mmhdno ; Juvd, Juvdno, etc. It is not, how- 
ever, stated that this rule applies to formations in iman; and 
even granting that it does, we get imdno, the change of which 
into dni or dini is not supported by any proof, and seems some- 
what harsh, and opposed to the genius of the language. More- 
over, in the closely allied and interchangeable form di, see (2), 
the d belongs to the adjective, and is a solitary instance of the 
retention of the older form of the oblique, which is still long in 
Ghijarati, as '«5T^ " good,*' obi. ^TTT> ^^^ ^^s been shortened in 
S., as "^Srs^ "white,** obi. IRf . In the case, therefore, of the words 
about to be quoted, I should say that the suffix was merely ftf 
or Xp^f 3^^ regard it as the same as the primary suffix ^i^^ 
mentioned in § 6. With this agrees the fact that the cognate 
languages have a form 'W^ or ^TTT, which is made only from 
adjectives in long d, so that the long vowel belongs to the stem, 
and not to the suffix. In Sindhi iU^TfUI "whiteness,** from ^raft 
"whitfe,** I should divide achhd-nL 

From ^R, then, used as a modem suffix to adjectives which 
had already arrived at the stage in which they ended in long 4, 
the following examples may be taken to have arisen : 

H. ^P^ "high," 5*qTT "height," P. ^^, ^3^1^, S. ^^ 
"superior," ^r^Tftr, Q. ^T^, ^T^TW- 

H. ^^ " broad,'' ^TR " breadth," P. id; 

Digitized by LjOOQ IC 


H. ^#iT "deep" (Old-H. ^^), ^#iT^, ^I^Tif , "depth," G. ^fc^, 

S.'^ftjft "broad," qfii<|fu| "breadth." 
S. ^if^ "great," ^jftT "greatness." 

(5). ^ is suffixed to a large class of adjectives in Hindi, to 
denote condition^ and agrees with our termination "-ness." 
This form is connected with that of the desiderative verb in 
Sanskrit, and some of the words of this class cannot be regarded 
as formed from 'nouns by the addition of a suffix, but are to be 
derived whole as they stand from Skr. The substantive in ^ is 
accompanied by an adjective in Wt, as ftRTra or T^ETHBT " thirst,^^ 
ftl^rreT " thirsty,^' which are not to be regarded as recent for- 
mations from xft, the root of TJVifT "to drink,** but as from Skr. 
PmNI^I and fi|Ml^ respectively, these being the substantive and 
adjective from fqill^ " to desire to drink,** desiderative of root 
TU "to drink.** Skr. fllMI^I loses the long final d of the fem. 
and becomes H. fiT^rW, while ftniTO apparently takes H, and 
becomes ftrrr^^y whence would come a Pr. fxTOT^^, and the 
u Tanishes, leaving H. ftHTP^T. The modem words would be 
more accurately written fi|i|l^ and firVraT> as the medial IJ 
has disappeared (Vol. I. p. 199). Accordingly these words in 
most cases bear the meaning of "desire to do a thing," even 
where it is impossible or difficult to trace the word to any 
definite Skr. desiderative like ftm^- And in those words which 
we cannot refer to a Skr. desiderative, it must still be held that 
the form is in itself by nature desiderative, and having appro- 
priated to itself this meaning, has been tacked on to modem 
nouns to form abstractions, so that in a number of instances it- 
legitimately comes under the head we are now considering, 
namely, that of secondary abstract nouns. Examples are found 
of three classes : first, words derived direct from Sanskrit deriva- 
tives ; second, those derived from modem verbs (primary stems) ; 
third, those derived from modem nouns (secondary stems). 
VOL. n. 6 

Digitized by VjOOQ IC 


Skr. finrWT "thirst," H. frrvm (tWRT, WW), and 80 in aU the 

H« 0*11 "to weep,* "^^^Sire "desire to weep," "vexation,* P. '^^(W 

H. *J^«IT " to urine,*' Mf[m "desire to make water,* P. id, 
H. f^ifT "cacare,* ^irFET "desire to stool," P. id. 

Skr. fif^ "sleep* (H. iff^), H. ^^TO, p | ^ | 4j , "drowsiness." 
H. ^m " sleepiness,* ^V||^ " nodding," " incipient sleep." 
H. ^J1| "violence," H. |Jirn9 " a violent burst of rain.* 
H. 4J^'dT " sweet," H. flTTra " sweetness.* 

Each of these words has its corresponding adjective, as 

fnnrer, ft^w, ^^, fiiror, f«r^»?[TOT, ^wmi , but not 

^JifTOT ; hut the stem itseK is almost entirely confined to Hindi 
and Panjahi, only rarely occurring in the other languages, as 
M. >1ZTO "soumess,^^ from ^^fZT "sour ;*' fif^lfl " sweetness,*' 
from fif^ "sweet ;** and with the short vowel '^f^^ "plump,** 
from 4\i^[ "a swelling;** G. VtoETO "whiteness,** from \ft3Bt 
"white;** ^^tsm "sweetness,** from l^tzt "sweet;** ^Eirsiirnr 
" difficulty,** from 'WtS^ " difficult,** and a few more ; and we 
may fairly assume that all the other words were modelled upoa 
finiraT, where the long d belongs to the root "qr. In O'^NI 
we cannot look back to the desiderative of Skr., which is ^;^t^> 
which could in no way produce rodsd, but would result in some 
such form as ruisd. Skakespeare, in his Dictionary, absurdly 
derives these words by adding the Skr. ^in|T " hope,** in which 
he is blindly followed by his faithful plagiarist Forbes. There 
are nouns with this termination which do not belong to this 
stem, as J]4i(JT "a pole-axe,** the former part of which is 
probably from HftS (Skr. Vl^) " a knot,** with some derivative 

Digitized by VjOOQ IC 


of the root ^ "to cut/' just as JfJJS "an axe/' The gaurdsA 
is a formidable weapon of bamboo with iron knots and clamps 
on it, and a small axe at the head. We might assume a form 
l|iW||^ as its origin. There is also a word 4J\||^| "a small 
turban/' the former part of which is ;rf^ " head," and the latter 
probably a derivative from ^f^ "to clothe," so that a form 
^[^irnfii may be assumed for its origin, unless, indeed, which 
is highly probable, the word itself grew up in modem times, 
and has no Sanskrit original at all. 

The stems given above are the most common and widespread 
abstract forms. There may be in the various languages others ; 
but they are rare and confined to special dialects^ and do not 
require detailed notice in a Work of this kind. 

§ 18. The next class of secondary formations is that of adjec- 
tives denoting the possession of any article, or of any quality 
or tendency : the former are called possessives, the latter at- 
tributives, xmder which are included words denoting trades or 

(1). The most important, and from its variety perhaps the 
most difficult to treat, of the stems of this class, is a group which 
has the following terminations : 

Hindi f;, J^, J^. 
Panjabi the same. 

Sindhi f;, ^, fTJff. 

Gujarati ^, t;^, T^- 

Murathi ^, ^, l[i||. 

Oriya }^'^-'^- 

Five Sanskrit stems lie beneath the threefold division whicl^ 
i 18 shown by the above list : 

Digitized by VjOOQ IC 


' 1- jfitf leaving ^|[H^, nom. ^, as 9flt%^ "gardener," nom. '4||^. 

2. ^ and ^fj^, leaving \jl, as t|^?| ''mountain/' t|^?Tt^ "moun- 

3. TS^ and 'Ji^, leaving f[i|| with guna, as %^ "Veda," ^f?lC 
"teacher of the Veda;" l[iR without gunia, as VIZ "jar»" Iffd^ "con- 
tained in a jar." 

4. TST, leaving \J^, as XTff "kingdom," <^|f^'i( "belonging to a 

5. Several suffixes leaving "^, as Ipf "man," ^|#^ "popular." 

The difficulty consists, first, in the mutual interchange and 
fusion of these several stems, and secondly, in the fact that we 
cannot in all instances trace the modem word up to any distinct 
Sanskrit word. Of course when we can so trace it, all difficulty 
ceases; but there are many modem words which we can only 
vaguely identify with some Skr. root, but cannot teU through 
which of the above formations it has descended to us. A partial 
clue is affi)rded by the meaning in some cases, as the Skr. forms 
denote sometimes possessives, at others attributives or appel- 
latives. But the lapse of time, the changes of meaning, and 
the facility with which one word lends itself to reasoning which 
would lead it either to a possessive or attributive origin, obscure 
the matter very much. ^ may come from 1, 2, 4 and 5 of the 
Skr. stems; J^ and its fellows from 2, 3 and 4; i[cR can only 
come from 3. 

It will tend to clearness to give first those words which can 
be traced without doubt to one definite Skr. stem, and then to 
discuss, as far as our present lights permit, the question of the 
doubtful forms. 

(a). Following the order of the Sanskrit forms, we first 
treat of ^;i^^, having for its nominative ^. The commonest 
words of this class will be found in Ch. 11. § 32. Others are 
the following : 

Digitized by VjOOQ iC 


H; >3V>j) and ^f^ "rower,** from 8kr. ^11^, one who uses a ^i|^. 
in the sense of "oar," (H. ?t^ and Tt^)» P- t^-* M. d^ "one who 
carries a staff," B.- O. ^^^ (ddhrt) " rower." 

Skr. Ifl'^ " reader," " learned," H. id., O. Mf^, the name of a sept 
of Brahmans. 

Skr. Hfin j^ " one who has read the three Vedas," H. B^TT^, O. f ?tf |0^ 
the name of a sept of Brahmans. 

Skr. %llft "lion," H. ^fft, P. %^, S.%ft, G. i»ft. 

Skr. ^4D' "prosperous," H. ^H\, and so in all. 

Skr. ^ "leprous," H. cjt^, P. ^^, S. ^^sH (^l>>ch points 
rather to a form ^if%7|), G. ^fsf^, M. c^t^, O. id, and IRV, B* 

' Besides numerous Tatsamas in very common use in all the. 
languages, as "qpft "sinful ^^ (mf^)> Vi^ "virtuous^' (^f^'fr'l)* 
O^n " diseased'* (O'f'r^)* ^.nd the like, S. lias also ^(^ 
*' complainant,*' from ^t^ " complaint,'* which does not seem 
to be an Aryan word, but is probably corrupted from Arabic 
^>\j "claimant," J^y^J "claim." In some cases of undoubtedly 
modem origin S. retains the final vowel of the original word, as 
lTf?f "opinion," '^^ "tenacious of one's opinion," "opinionated," 
Tvhere the final short i has been changed to a, 

(13). The next form |^ is used to form principally patro- 
nymics, or adjectives denoting nationality or caste, and results 
in f^ and JJ^. 

Skr. f^hlV^ " a Sindhian," H. flh|^, and so in aU, In the 
same way are formed in S. ^fT^ "a man of WT^ or Lower 
Sindh," ft^J^ "a man of RjlO U^ppor Sindh;" Hj*^!^ "that 
which belongs to a f%^ Hindft." But 0. uses the form J^, 
as ^Hfid " an Oriya," from a Skr. form ^Jt^^, from ^St^, 
the old name of Orissa; also iftf^^ "a Bengali," Skr. 
if)>j)i(, from ^ftlf, the old name of Bengal. Common to all 
is m^\S\ or li^at " a Mar&tha," from Skr. IflglAliU^ ' 0. ha» 

Digitized by LjOOQ IC 


also ^Mlllir or ^Ji\ "worthy to be ofEered," Skr. ^f^lffhl, 
which, however, rather means "worthy to be honoured with 
offerings." Further instances are : 

Skr. Vlilft^ "mountaineer,*' H. XT^cf^, Trififirr (chiefly used of the 
Nepalese), P. id. and IT^^hNTT, O. VH^nt^t. M. B. O. ViUft^, O. 


Skr. ^^^ "belonging to a country," H. ^l^ "native,* as distinguished 
from "foreign," P. id,, S. ^^||^ (from %3f) and %^ (from %^), G. ^^* 
M. ^^^ and ^^^|, O. ^Ij^ and ^^M, B. id. 

Skr. TJ^Ul "belonging to the kingdom," B. "^T^, O. H. id. This 
word is not known in the other provinces, and denotes a class or caste 
coming from the '^^T7> i.e. "^T^, or kingdom of Bengal, that is, the settled 
and central parts of the country, as contrasted with the outlying and thinly 
peopled regions. Skr. has both forms 4^|)^t| and "^itj^j hut the former 
is commoner as a Tatsama in the modems. 

(y). We now come to the most important of these forms, that 
in l[c||. There are two classes of this stem in Skr., one which 
augments the vowel of tbe root, the other which does not. The 
distinction is maintained in Sindhi with tolerable accuracy, but 
not generally in the others. Words of this stem are generally 
used as names of trades or castes. 

Skr. ^ri^ "spirituous liquor," ^fl^lii "a distiller," H. ^j|^ {s(inr1)y 
B. ^^, O. ^t^. 

Skr. ^^ "oil," 9f%fS|j " oilman," H. %^, and so in all. 

Skr. '^(};^ "betel-nut," ?y |l^fi | q [ "a seUer thereof," H. ^vHt^, 
P. ?Wt^, G. like H., M. TrtiPtaft, B. O. irf^. 

1 Used chiefly in composition : when standing alone, it means an inhabitant of the 
J)e8h, an expression which implies the high table land of Maharashtra above the ghats 
or mountain range of the coast, and is opposed to Konkan, the low narrow strip of 

Digitized by VjOOQ IC 


Skr. ^ <<yedm" %^ ^'learned in the Vedas," H. ^. 
Skr. 11^ "plant,^ ^if^PC '* grower of plants/' H. m^ ^'a caste of 
gardeners,*' M. id. 

Formed upon the same model are the names of tliree great 
septs of the Kanaujia Brahmans ; thus ^^ is shortened from 
^1%^, and that is from Skr. f(%f^9 " one who knows two 
Vedas;*' ^^^ is ^ift^^, from V||^f^4 "one who knows the 
whole four Vedas.'* The fuller form is retained in t?f^^ "one 
who knows the three Vedas," ^^(^il. It is, however, equally 
correct to derive these forms from number 6 of the Skr. stems 
above, as in Skr. there are forms ^^, i^^y %^^, 41 1<^^. In 
the case of ^1T^9 the origin of the nieaning is doubtful. The 
K&chhis are a widespread and very industrious caste of culti- 
vators and market-gardeners, and their name may either be due 
to their growing vegetables, as one of the meanings of Sanskrit 
H^ is " a plant," or perhaps, and on the whole more probably, 
from the kdchh or tightly-girt and tucked-up cloth round the 
loins, which is the only garment that they wear. Here may, 
perhaps, also be classed the common word H. ^^ and ^^, 
M. ^Clinft> the name for ordinary peasants in M., and in H. that 
of an extensive caste. I do not know to what Skr. word this 
name is to be affiliated. It is also spelt ^pRt and ^^, and 
the name probably originated in comparatively modem times, 
and may have no connexion with the older language. 

Sindhi examples of this stem are : 

Skr. ^ff " camel,* iftfipi " camel-driver ;" S. ^^ " camel,* ^ftS\ 
** camel-driver.* 

8. ^f^ " earthenware," %ftrO " **^®'* ®^ pots." 

S. ^H^ " vegetables," WTftl^ " 8cU«r of vegetables.* 

This last word is, I think, not Aryan, but comes from the 
Arabic <{i^ "potherbs," whence the word so common in the 
other languages, Jl»; Ja^^d/, "a vegetable-seller." In Hindu- 

Digitized by VjOOQ IC 


stani the phrase baniyd bakkdl is colloquially used to indicate 
petty traders and shopkeepers of all sorts. 

(S). The suffix ^ is hardly to be distinguished from |^, 
except in words which can be traced up to the Sanskrit. The 
typical word is 

Skr. ^rt^n^ '* a Kshattriya or man of the warrior caste,* H. q^tH^ ^^» 
%^, P. id,, S. t^Nft, M. ^^. 

(e). Stems formed with ^ are very numerous^ and, owing to 
the variations in meaning which modem words have undergone, 
are sometimes confounded with abstract nouns in ^, like ^t^ 
" theft," from ^i|, and, like them, frequently drop the ^ alto- 
gether. Of stems which retain ^, the following may be cited: 

Skr. M\Mi "belonging to a village," H. ITTlfty and so in all, but H. fre- 
quently mfijcl and 4||I4, also ^^t^ and H^, from the Tadbhava ifm. 

Skr. ^ "principal," H. qj#, P. ^, ^'^, S. ^, G. ^, 
gt^>, M. ^, B. id,, O. id, and igf^T^. 

A very large class of words exists in all the languages which, 
from the absence of any special Sanskrit form to which to refer 
them, or from the existence of more than one Sanskrit form, 
cannot be definitely referred to any of the above heads. Such, 
for instance, is \jn?t " master,'* " lord,'' also " wealthy," a word 
in use throughout. Skr. has ^slf'nt^, ^iHnii and ^jRf. In 
Old B. we find the word igrfif "lady," which may either be 
shortened from Vf^nft, or from ^^RTT* 

M. is peculiarly rich in words ending in lU, which are per- 
haps generally to be affiliated to the stem in l^^^l, as mentioned 
in § 9 (2), but may also be attributed to (e) by supposing the 
usual addition of 1, making ^5WI. Some of the more striking of 
these words are given with the Marathi primary stems to which 
they are allied : 

'^^^ITf "commission," ''ITS^ "commission agent," "broker." 

^n^ "smartness," "a shock," 4x|^| ** smart," "sharp," "active." 

Digitized by VjOOQ IC 


^i||^ "scheme," I||t|;^J "scheming,** "treacherous." 

^ZIBS "bullock-pack," 4^|dbi|l "fit for carrying burdens," "a 
draught-horse or bullock." 

cftt^dbT "a coarse blanket," ^(^db^l "one who wears a kdmbafd" 
" a labouring man." 

^ITRf^ " pole for carrying loads," 4i|^4i|| " porter." 

^it« " leprosy," 4\iM\ " leprous." 

^qdb "scab," 4^«|db^l "scabby." 

^nZ "ghat," Ml^ili "Brahman who attends at a ghdV* 

in^l "horse," ^t^^TT "groom." 

Here we have a curious preservation of an older form. ^tTT 
is Skr. vh^^, Pr. ^t^ljt. The M. word points back to a form 
^ftzf^CT, MlifSftH, ^^ M^d^My in which the ^ has been preserved 
as a fulcrum for the termination. 

Gujarati has also words of dubious origin in J^ or t^. 
The double form arises from the unsettled nature of Gt. spelling, 
in which no distinction is made between long and short i. The 
two examples most frequently given, ^f^ilt "grieved,*^ ^^^ 
"happy," do not appear to come from any of the above stems, but 
from Skr. "^tf^R and ^f^ respectively. Other instances are : 

ViUldo "school," fi | ^|fdd41 "scholar'' (perhaps Skr. *|:^nf%ii). 
^fif "earth," ^jf^pft "earthy," O. ^t^ (perhaps Skr. ^TTHT, or 

^it^ "leprosy," ^Stf^pft "leprous" (Skr. is ^f^), see (a). 

Without spending more time in considering the subject, it 
may suffice to suggest, that the fusion of several cognate forms 
into one, which is so common a feature in these languages, has 
been at work here also, and that the Skr. terminations given 
above may have been by careless or ignorant speakers added 
often to nouns which did not take such terminations in Sanskrit 
itself, so that words of this group may be referred at will to any 

Digitized by VjOOQ IC 


of the five SanskriC steins given, and in isolated instances' to 
others also, as, for instance, H. firf^fin "uncle,'* which is from 
none of the five steins, but from Skr. filH^. 

(2). Principally indicative of qualities or dispositions of the 
mind is the suffix ^(TWy which, although occasionally found as 
a primary, is far more extensively used as a secondary suffix. 
Vararuchi (iv. 25) treats this suffix in Pr. as a substitute for 
mat (and vat), suffixes denoting possession, overlooking the fact 
that ^(Rf exists in Skr. already. In Pr., as also sometimes in 
Skr., the final u is dropped. This stem is found in different 
words in the seven languages ; each language may therefore be 
illustrated separately. 

Marathi,. in which, whether from the greater perfectness of 
the dictionary, or because the language is really richer in these 
forms, a larger number of instances has been noted than in its 
sisters, uses the form ^Cfas with the cerebral 35 and long 6; it 
has also ^Idb^ and pccasionally even/^TO5T, which latter points 
to the use of the universal ^ suffix, so that we should postulate 
a Skr. ^ftlRC. There is also foimd *4||gI^€||, which would indi- 
cate a former l^rf^ra, with reference to the remarks in the 
concluding portion of the last section. Thus we have : 

Skr. W^ "pity," M. WK[^. H. ^RT^. 

Skr. ire "trouble," M. *ftgldb "laborious," "painstaking." 

Skr. %JJ «hair,«» M. ^^TT^ and %HT35 "hairy." 

Skr. IJir^ "ball," M. JnTT "knob," "tuft," ijl^ildb "tufted grass." 

Skr. [uncertain], M. Vf^ " stone," ^jl^ldb " stony." 

Skr. 7J^ "snout," "beak," M. fjf^ "mouth," j JI^ T db "foul-mouthed," 
" scurrilous." 

These three are combined in a proverb descriptive of the 
peasant's three greatest troubles — 

^^•^mraB ^ witob Tr^wrr ^mras 

'^ Pasturage coarse and knotty, fields full of stones, and a scolding wife." 

Digitized by VjOOQ IC 


Skr. t^ "hand,'* M. fPf, id., f t^T ldb "thievish," "with itching palm." 
Another proverb runs — 

"The itching palm may he satiated, hnt never the scolding tongue." 
Slir. ^fz^ ^'& measure of time, ahout half an hour," M. ^T^ id., 

^^Idb "&i^ instrument for measuring time," '<agong," H. Hf^^l^* 

P. id,, and so in all. 

Skr. TBTTf! "wound," M. TTRI id., HWildo "wounded." 

Skr. t^gPT "greasy," M. ^^ "sap,* "juice," f^<Hdb > f^<«ldol and 

f^^lcb^r "viscid," "gummy." 

Slir. [uncertain], M. ^jt^ ^^^ ^flt^ "sleep," fj^YVldb' "sleepy." 

As to this last word, the root ^ in Skr. means "to move,'* 
among other things, and the root '^^^, which is in effect an old 
causal of ^ in its meanings of "sowing (grain),'* "weaving'' 
and "sewing," shows that the movement implied by the parent 
root is an oscillating movement backwards and forwards, like 
that of a weaver's shuttle, or a sower's hand ; so that we may 
fairly suppose it to have obtained the secondary meaning of 
"nodding," as one does when drowsy. In confirmation of this, 
M. has also ^tiTT or jfm "a swing ;" so that we may assume a 
form ^rf^ + ^, which would give l^n, the anusw&ra, or rather 
anun&sika, being inorganic. (See Vol. I. pp. 177, 327, for the 
U, and p. 143 for the ^.) 

Skr. ftrofl "covering," M. fjf^ "a mat,''|jtfn35 "matted or bushy" 
(a tree); see Vol. I. p. 177. Hence is formed a verb f]iMldbi)i "to be 
bushy and close of foliage." 

Skr. MHTil " mmeness,* " selfishness," M. id., HMJi\ch " selfish," " self- 

Skr. ^51^ « milk," M. Ijp^ id.,^ "^^(135 " fuU of mUk," " a milch cow." 

There are also in M. instaiices of this suffix taking the form 
dr&y as in Sindhi, thus — 

M. ^1^ "a tile," if^i^rT^" tiled" (as a bulding). 

Digitized by LjOOQ IC 


We now pass on to Gujarati, wliicli, like M., principally uses 
the Bo, though words of this stem are also found with ^, but 
perhaps only through a mistake of the writers. Authorities 
disagree very widely as to the spelling of Gujarati, which 
is not to be wondered at in the case of a language with a 
scanty literature, no £xed standard of pronunciation, and 
numerous dialects. 

G. ^rnf " shame," i^^ldb " shamefast," " bashful." 

G. ^r^r " sleep," ^V||<^ " drowsy." 

G. ^I^cjl " word,** ^l^Td^ " loquacious," also in a good sense "eloquent." 

G. ^Hl" " sand," ^TTT^, T7fJ3S , and by some ^^TRT " sandy." 

O. M^O " stone," M\!|4,r<^ " stony." 

G. ^ftf^ "blood," ^^jl^ l <^ " bloody." 

G. ^T^ " beard," ^ij^-^ l c^ " bearded." 

G. ^r^ " a ghari," TT^^fig^t " gong." 

With words ending in f^ this language affixes the termination, 
without modifying the stem word. In the case of nouns in "^j it 
adds the suffix to the oblique form in d ; the fem. in I is the same 
in the oblique as in the nominative ; so that we may say that in 
all cases the suffix is added to the oblique form. An exception 
is\?!T<jp, but here we derive from an earlier form\?f in use in 
the other languages. The language of course uses, like all its 
sisters, Tatsama words, at the discretion of 'the writer, and words 
like A^l^ are almost Tatsamas, the cerebral 3^ being merely 
a vulgar pronunciation of ^. This suffix is found appended 
even to Persian words by the indiscriminating vulgar ; thus from 
A^ "shame,'' is formed ^4^4||<^ or ^<4{ldb "bashful.'' 

Sindhi more frequently changes ^ to ^ (Vol. I. p. 247). Thus 
S.Uff " buffalo," %fT^ " buffalo-herd." 
S. V^ " herd of cattle,* ^SROTC " herdsman." 
S. ^^. " camel," ^t3T^ " camel-herd." 
S. ir^ " barley," ^Wt^ " (wheat) mixed with barley." 

Digitized by LjOOQ IC 


S. ^rit " hour," ^TgJT^, o^t '* gong." 

S. W| "clouds" (collective), ^ITWt "cloudiness." 

The final vowel, whether it be the u of the masc., or the i of 
the feminine, disappears before the suffix ; but long i is retained, 
merely hardened into ^, as in M* Quasi-Tatsamas also occur, 
which are simply due to the fact of the words being honestly 
written as the people pronounce them, and not as Pandits would 
have them; such are f^rSl^, 4M\^y for ir?!T^, 4«||^. 

The word f^W^ ^^ candlestick,^' from {^^ "a light,'* is 
quoted by Trumpp in this place; but it would appear to be 
questionable whether we have not here a compound from 
^t^ + ^l^^. I shall show presently that compounds of this 
sort are not imfrequent, and often render it difficult to determine 
whether any word belongs to this class or not. In this and other 
languages compounds whose last member is i^ire "time,'' or 
^jTWr "hall," or '^W^ "abode," as will be exhibited hereafter, 
come under this category, and the confusion is increased in M. 
and G. by the substitution of the cerebral 95 for W in these 
words,' as well as in the suffix 'm^. 

Panjabi has not any special fondness for the ^ sound, nor does 
it particularly afiect stems of this formation. Instances are, 
"^d^ldbl "woollen," from ^?r or ^3ST " woqI ;" 4iJiHI "distressed," 
for c||iJdHi > from m^7{ "trouble" (Skr. ^^). In this case the 
possessive signification is sometimes lost, and the word used 
merely as an intensive substantive. c^r^'^lcbT "thorny," from 
%fT "thorn" (Skr. i|Sl^^), where the short i represents the e of 
the oblique form ?ji%. ^U^SIdbT "a large earthen pot," from 
^siri^ "a pot," here again intensive. isteT^ (ironical) "a vain, 
pompous man," that is, one who has a gong beaten before him as 
lie walks, from ^ZT "a bell," "a gong." There is also, as in the 
at)Ove-quoted languages, the common word Vlf^^l^ "gong," 
from VFS^l but the language prefers the cognate forms of this 
stem in «7, ul, and ail, as we shall see presently. 

Bengali and Oriya, being languages originally poor, and 

Digitized by VjOOQ iC 


having, therefore, a smaller stock of Tadbhavas than the others, 
are not fertile fields for secondary formations. In them such 
stems are mostly Tatsamas, like ^ERT^; and it will not be 
productive of any large results to adduce instances from them. 
The words which the student may meet with will be readily 
recognizable as belonging to this stem; such as, ipiTT^ 
" quarrelsome,'' already cited ; 0. f%\sn95 " burglarious,'' from 
fth»l "a burglar's hole," sometimes written QjMilldb for f^^TT^; 
Mmdo "middling," from ?Tfij (Skr. J{^) "middle;" f^s^Tdb 
"piercing," from f^ "a hole." 

With regard to Hindi, usually so rich in primary and 
secondary forms, it is somewhat strange to find so few illus- 
trations of this widespread stem. Apart from Tatsamas, it has, 
however, forms in dlu, and, owing to its habitual neglect of 
final short vowels, also dl, but displays, like Panjabi, a marked 
preference for the forms in «7, ail, el, and uL It also in many 
instances changes / to r (see Vol. I. p. 247), thus producing dr, 
with variants drd and dru^ where the final u is preserved by 
being lengthened. The following list will suffice to illustrate 
all these forms. 

H. Igfff " shade," ^(^14,1 " umbrag^eous" (not from ^f^f + TTTT, as 

Forbes puts it ; the |[ belongs to the root). 

^ rfMMdlV \ " one who finishes," " destroys," " makes 

H.finra"end," L ^ ... 

(f*RZTTr) an end of." 

H. ffte " beUy," } ^|*^^^ | « pot-bellied." 

H. ^€r^ « hour," Mfigill^ " gong^." 

H. ^ " milk," ^'mr " milch," and ^aTT^, B. id., O. ^f^ . 

H. liW " baU," iitwrTT " globular." 

H. f%^ " mark," f^fTT " recognition." 

Skr. Tt^^^ "love," fqi|K| " beloved" (as though fimi^ for 5*). 

H. ,^7T "sand,*^?fnrr "sandy* (more generally ^^, see further on). 

Digitized by VjOOQ IC 


Many of these words, as in the other languages, have lost 
the possessive or qualitative sense, and have become simple 

(3). Closely allied to the last, and like it used to express 
possession of a quality, is the Skr. X^, which in Pr. often 
doubles the ^, making 1[V« This suffix in the moderns takes 
a lengthened form fj^ or ^J[^, and combines with a preceding 
a OT d into ipir, tf^TT? and ^^. As remarked above, it is a 
favourite in H. and P., though less frequent in the other lan- 
guages. These two languages have forms in elu, where the u 
probably arises from confusion with dlu, unless we here admit 
a change of d into e, noted as not unfrequent in Vol. I. p. 136, 

H. irt^ "knot," iritm "knotty," P. id. 

H. ir^ " pit," irlWT " large pit" (intensive). 

H. ^ii^ " gravel," *inf^ "gravdly." 

H. ^T^ " house," V|^^| " domesticated," " tame." 

H. ^rra " grass," MiD^I " grassy." 

H. ^^\Z " wound," -je^^i « bruised." 

H. ^^ " beauty," ^4)^1 " beautiful." As in Bihari LaPs fanciful verses : 

ftnc ftnc ^ftc ^ ^ ft^ ^wRcHt SNr • 

Satsaif 234. 

H. Ifr^' " shade," ^H^WT " shady." 

H. \t^ "sting," nI^S^WT "armed with a sting" (said of insects, as 
^Tl[ ^it^ "j^ll^l % " this insect has a sting"). 

. H. ^SRT " shove," '^%^ "in the habit of butting" (said of homed ani- 
mals), P. ^%^» fem. \|^i^. 

H. IS^ " form," ^4[m " weU-shap^," " comely." 

H. 5t^ "belly," #ff^ "pot-bellied." 

H. IPT " forest," ipNiT " wild," "jungly." 

H. ^ " sand," ^^VWT " sandy," " gritty." 

Digitized by LjOOQ IC 


H. "^tlj " load,* «|%^ " load-bearing^,* subst, " a beast of burden." 
H. T^ttn " rival wife * (Skr. ^Qc^), ^^tHST " belonging to a rival wife,'* 

as in the phrase 4^^^! <HTI^ that is, *^ a brother, sou of the same father, 

but by a different mother," " a half-brother.* 

H. ^^ "start* (Skr. ^IH^irnC), ^f%W "a horse that is given to 

shying or starting." 

There seems to be no rule for the use of one of these forms 
iijj:^ead of amHwir, though in practice we find that one word 
always uses eld, anl another ild, while a third has only the 
form aild. With respect to the use of the final d, it may per- 
haps be laid down that such^ords as have a distinctly adjective 
sense always take the final d, but sich nouns as are used sub- 
stantively reject it. Thus, if we shoald say, "he is my half- 
brother,^^ we must use eld, as ^^iNtT W^l ^^^ ^ ^^ would say, 
" this man and woman are my brother and lister by the father's 
side, but by different mothers," we might S^-y, " they are my 
sautek," f^"^ ^^^ %*. So also derisively, "Fo, pot-belly ! " 
would be, ^ ^Iff ^ . 

Those words in which Panjabi agrees with ff. bave been 
noted above; other instances peculiar to that language are these 
which follow. ^^4 in the sense of halting, gives ^*Z%^ (/•) 
"a woman that walks mincingly, or affectedly," sometimes, 
but incorrectly, written ifd^^, possibly from some idea of 
its being in some way connected with %^ " play." From ^Hf 
" stoppage," ^TO^ " a horse that jibs," G. ^f^^^ . 

P. ili^ "strength," "drawing-power," ^^^ "astringent" Jsaid of 
medicines). \ 

At. <l1c "suffocation," (in India) "rage," ^^WT " angry/' ^^ "* 
bad-tempered person." 

P. ^RTT "petticoat," ^9f^^ "a woman who wears a petticoaV* *-^' 
a virtuous woman ; prostitutes do not wear the petticoat in some parts> nor 
the drawers, but only the loin-cloth, sdrht \ 

Digitized by VjOOQ IC 


P. ^ " Storm/' 1}^ mb9t. « quarrel," « affray," " uproar." 

P. IRBT " weary," ^'WT " easily fatigued." 

P. ^ " milk," ^?>^ « a milch cow." H. has also this word. 

P. XRifJi " stone," XW^hrT* ^Wt^ " stony." 

P. ^r^ " shame," ^nftWT " shamefast." 

P- ^i^l " the middle" (the name applied to the central part of the 
Panjah hetween the R&vi and Bi&s rivers), 7i%^ ''a man from the 

P. makes tlie feminines of words of the form elii in o, see Ch. 
II. § 33 (3). As between ail and el, it may be decided that the 
latter is a softening of the former, which is again shortened from 
dil W^y where the d belongs to the root. This applies to H. 
also ; thus 9 ^ [iim " shady,'' is for l^fr^NTT, which is ^FRH + T^ > 
while ^W\wr is '^f'T + IJ'Sn. The difficulty lies in this, that 
often where we have eld, we cannot find the termination in long 
d in the root; thus there is no reason to believe that ^f "forest," 
was ever ^tfT, yet we require iHT to give us il'Nn', i*e. ^^T+X^. 
The rule will not therefore apply in all instances. 

Sindhi has fj^ft and "H^, but also, in conformity with its usual 
custom, ^^ and H^. The long vowel, as in H. and P., seems 
to have arisen from rejection of one of the two h of Pr. i^jf- 

t?^ W^ " ohstinacy," fT^llt " ohstinate," G. id. 

t?^ ^rtfVf " patience," ^tNV^ " patient." 

If^iY ^ " the desert," ^^^ " a man from the Thar." 

^"leather," ^^^ " leathern." 

5^ " shade," ^t%^ " shady." 

iPCt «<^- Tf%^ id. 
„ >5^ " a kind of g^ass," ^ITOXt " made of dt^hu grass." 

With Sindhi on the one hand, and Hindi on the other, is 
connected in respect of this suffix Gujarati, having in some 
instances the dame words as the former, as in fdl^l*, and in 
VOL. n. 7 

Digitized by VjOOQ IC 


others the same as the latter, as ^^sn35= H. ^V||<, also ^^|^= 
H. and P. '^v)W- Its usual forms are ^^ and H/^. As 

HJH "relationship," ^"nftwt "related, by blood or marriage." 
^m "much," %|Ij|<^ "abundant" (only used in poetry), 

irr^ "bad, defective," nf^O "scanty." 
Vf|4f " sweat," ^4)db\ " sweaty." 

The language does not seem to be rich in this class of forms. 

Marathi has numerous formations in %^, ^^, and if^, also 
in IPC and Hi;^. Care must be taken to distinguish from words 
of this class those words which present long i from the tendency 
of this language to lengthen vowels in a final syllable (Vol. I. 
p. 155), such as ^dldb " perverse,'* "malicious,*' which is merely 
a vulgar pronimciation of Skr. ^RffZ^ "crooked.'* Further are 
to be eliminated a small class of compoimd words in which 
^^f^ forms the last member, as'^^if^^ "temple," =^e|H^; 
also familiar words such as ^f^Vw or ^£\^ "ancestor," from 
Skr. TO through a now disused ^RT> and ^^T?^ "a wife," from 
^f^. These seem to be a sort of diminutives. 

As genuine possessives may be cited — 

;E|^ "side," "direction," ^fSxf^^ "relating to a quarter or direction." 
^[^^ " rock," ^^<iV9 " rocky." 

^rr^ (adv.) "below," 4^|^^ " belonging to the lower part." 
iJt^ "sweetness," jft ^cfe " sweetish." 

^itZT " thorn," ^RtZ^T " thorny," also iRli^ " a thorny shrub." 

id. mtf\m "a thorny creeper." 

(Skr. ^r^) ^T!f "a } 

> ^HI\ "divided into squares" (a cloth), 
division," ) 

> ^Wt^ ''digging up plants" (subst.). 
excavation," ) 

SffTT " tuft," 'ft%TT " tufted." 

"^[[^ "stench," ^n%U "stinking," also ^TT^VT- 

Digitized by LjOOQ IC 


^ETTif " sweat,'* ^H^VT " sweaty,'* 

^ "old,'* ^ji|^ "old clothes," "rags'* (subst). 

1]^ "encnmbrance," ^^^^"tronblesome/' 

zrre " smell,'* q|ii<| , ^n%Wr " fet»d,** " rancid.** 

Here^ as in the other languages, the meaning of the possessive 
has in several cases passed over into that of a substantive. 

In Oriya this suffix is not common : an instance is ^(f^YWY 
"handsome,^' from '^g^, I have not found any others, except 
Tatsamas. The dictionary (Sutton's), however, is a very meagre 
one, and I have not heard the form in speaking. The same may 
be said of Bengali, in which language secondary formations of all 
kinds are comparatively rare, or, if they exist in rural districts, 
are not fully recognized as other than mere local corruptions. 
There is a vast field for research in this direction. 

(4). The combination of the suffix in ^ with a preceding u 
is rare, though not altogether non-existent in Prakrit. Vara- 
ruchi gives only one instance, vidrulio, for Skr. f^HITT^ 
" changed." The suffix ^J^ with the long vowel is found in a 
few cases in Skr., as 'mTfWf (also TT^JW) "mad,'' "gouty'' (i.e. 
affected by wind), from ifnT " wind.^* In the modem languages 
it is also rare. For Sindhi Trumpp gives no examples> and 
does not even introduce the suffix in his very full and well- 
arranged list of secondary stems. There are, however, a few 
words which may be referred to this suffix in all the languages. 

Skr. TT^ "a stroke of wind,*' S. ^TT^W "blight from wind,** H|^<^ 
" winnowing g^ain." H . 4||^<^| " mad " (from the meaning of Slir. ^ETRra 
given above), also written WT^TWr* ^TlfWr ^"^ ^fFTTT' P- writes the 
word ^|>^<| , ^I'j^l , and IR^Wr* G. ^(T^ " foolish." M. ^^V^ "flatu- 
lent," also a corrupt form ^ I *| ido " whirlwind. " The word IVIVftdb» mean* 
ing "a kind of soil on the banks of rivers," seems as if it belonged to this 
stem; but the connexion is difficult to trace, unless it be "driven hj the 

Digitized by VjOOQ IC 


wind," such as are the sand-drifts which the violent winds in the dry season 
heap up in the beds of rivers all over India. B. «| | y5>4 1 and WTSW 
" mad," «||yj>i) " whirlwind," ^njt^ " winnowing." O. ^ I findb t " mad." 

From M. I have picked out the following, wliicli should 
probably come under this head. 

Wl^ "curly" (Skr. ip^W), ^TRTtTT "gluttonous," from ^TT^ "to 
eat," a primary formation, iflMclb' "bloated," "pulpy" (like a pregnant 
female), from ^PTHT foetus. 

Of more extended use are— 

Skr. ^ETTif "heat," "sweat," M. V|14r>db1 > ^Wliife" prickly heat," "an 
eruption caused by excessive heat of the weather," H. B. M^0> ^"* ^• 

Skr. ?rur "middle," H. mflWT "middle-sized," "intermediate," P. 
^^cbl and 44^Gbl> S. however 4{fJlO> B- ^^^ ^- 'i^^- It « said 
in O. of the second of three brothers, thus ^>f HT^ "eldest," ^R^¥IT 
"second," ^ftZ "youngest." 

From Skr. ^TRT in the sense of "defect,'' "failure,'* comes 
P. ^TT^raB, Hl^dbl " lazy,'' " careless," not f oimd in the other 
languages ; also ife^db " quarrel," from yJR^ " snare," " deceit ; " 
IfflSBT "dear," from WTf "fondling," and some few more. 
H. ^6\^ "jocose," from "3^ "jest." The formation is not, 
however, a very common one in any of the modem languages, 
any more than it is in Skr. or Pr. 

(5). All the various forms of this group of ^ stems still 
farther soften down to an obscure al, generally but not always 
preceded by y, thus giving '^QTW. The typical word is 

Skr. TBTTf! "wound," H. IBTR, M. G. id., S. P. ^TT^, B. O. ^, whence 
comes Old-H. ^^[^5 H. lETT^I^' P- G. id., P. also V1IC^» S. ^T^T^* 
M. m«(|db» B. irrf^' inverted from TSTTI^W, O. id. 

Here only M. has the long d, and is probably from dlu, while 
the rest are from ila, so that they might be referred to a Skr. 

Digitized by VjOOQ IC 


form ^srrfif^, where the ^ has gone out as usual, and the inter- 
mediate form was probably ^nt^Tir. Panjabi has a great pre- 
ference for this shortened form, owing perhaps to its fondness 
for short vowels and habitual neglect of the principle of com- 
pensatory lengthening. Even where a word is written with a 
long vowel, the Panjabi peasant will often slur it over so that it 
sounds short. Thus 

Itf^ "knotty,'' from Iff "knot." 

llV3CT "muddy,'' "turbid," „ l}^ "filth." 

^i^^ " cavity," " hoUow," „ %R: " house." 

'JiW for ^^<^^' " belonging to the other side," TWC " *^ other side." 
They say 9f^^ ^^^n* f^T3R8B ^ff^ "the bullet went right through, and 
came out at the other side.^ 

^cHSIfT "sandy," „ *^"sand." 

xitlWfT "insipid," „ Xgt^ "dregs." 

iiT5a5T"dear," „ WR " fondling," H. ^T>i^. 

As an instance of this form may be cited 

H. ^RlWr "foremost," "in the front," P. id., S. ^ir^jPCt "superior," G. 
^'IT'W (i^^O "before," M. H9|cbl "superior," B. O. ^i(nrf^» to which 
corresponds H. fim^X "hindmost," with a similar series in the rest, as P. 


Sindhi has a set of stems in lu preceded by a short vowel, 
which, however, do not seem to be connected with this suffix, 
but rather with the Skr. diminutives in ^ and T> and may 
therefore be relegated to the section which treats of those 

This form is usual in Oriya, as 4liUdb "fleshy," from ^t9 
"flesh;" Wi^Bo "deceitful,'' from ^H^ "a cheat." 

One form fuses into another, or into several others, so con- 
stantly in these languages, that the inquirer is at every step 
bewildered by their similarity, and as the meanings, which are 
after all th6 safest guides, have also commonly, been lost or 

Digitized by VjOOQ IC 


shaded off by metaphors, or by local circumstances, into other 
classes, there is often little help to be found from them. It is 
probable, however, that in all except Sindhi, the group of forms 
in I preceded by short a has really been derived from dlu, ila, 
or one of the cognate forms in Sanskrit ; and it must be per- 
petually remembered that the modem languages often add a 
termination, which has for them acquired some special meaning, 
to roots and primary stems in which it does not exist in Sanskrit, 
or, if it existed, has not been handed down to us, so that the 
fact of the form itself not being found in Sanskrit is no argu- 
ment against its being a genuine one for the modem vernaculars. 
When also the modems tack on these expressive endings to 
Tadbhavas whose origin is uncertain, the inference is that they 
fully recognized the effect produced by such endings, and used 
them at will, thus constructing numerous quite recent words 
of their own. 

§ 19. The Sanskrit sufiSx 1[?f , though in form identical with 
the past participle, is also largely used to form secondary nouns, 
originally possessive in meaning, but shading off like the other 
similar suffixes into attributives and mere appellatives. It is 
the parent of a very large range of forms in the modem lan- 
guages, the most common of which is that which amalgamates 
the i of the suffix with the long d of the stem, producing in H. 
the common form in ait, though the others frequently keep the 
two vowels distinct, and M. more suo lengthens l[ to |^. Hindi 
having got hold of ait as a termination expressive of habitual 
occupation, possession or profession, adds it, regardless of ety- 
mology, to words which have no final 4, shortening the long 
vowel of the root. Thus from 

H. 7TW '"shield," H. ^^ <« shield-bearer," P. 1^, M. ^THrtn* B. 

H. ^IRn '' strut," H. V4i^?T "a strutting, swaggering fellow," P. 


Digitized by VjOOQ IC 


More regularly, however, from words ending in d, as — • 

H. cn^fQi "war-song," "paean," 4>C^7T "a warrior-bard," P. id. 

H. "^4(1 "report," "news," ^^ "a gossiping person." 

H. "^Wt "wrangling," "affray," ^3|?f "a brawler," P. id. 

H. if^ "spear," ^i|l^ " spearman," P. id. 

H. TRTT "relationship," ^^ih^ "a relation." 

In Sindhi this suffix is found in two forms, etc and aito. The 
e of the former arises from the fact that this suffix is added to 
the oblique form of the noun, thus TO " son," oblique ^putra, 
whence we get putra+ito^=putreto, ^^ "having a son/' In 
f eminines in a the oblique also ends in a ; thus f%ni " daughter,'^ 
f^nptt "having a daughter,^' from dhia+ito. Some words in 
u masc. do not change in the oblique, and in this case the u 
is elided ; but a consciousness of its having been there prevents 
the i of the suffix from amalgamating with the final vowel of 
the root, thus HfT^ " brother,'' HTT^ " having a brother," for 
bhd{u)+ito. Irregular is ^BjtipTt "having a wife," from ift^ 
"wife,*' where we should expect ^itfjuSt for joi+ito; it has 
probably been constructed on the model of putreto, without 
reflecting on the origin of the form. Another irregular word is 
^Idll^d) " having a road,'* from ^TZ " road," which should be 
tdteto like dhieto. 

With adjectives the suffix ito is added to long d, as 

^[T^T1[nt "timely," from ^[T<t "time." 
HlJT1[?lt "opportune," „ iT^ "opportunity." 
^^T^??!" "sultry," „ ^"sultriness." 

Oif I|,?f1 "sleepy," „ fn^ "sleep." 

iNni:nt "regular," „ 5|g"rule." 

This last word is also used as an adverb " regularly," " me- 
thodically," and this is the case with several other words of this 
stem, in Sindhi ; as 

Digitized by LjOOQ IC 


M4lf,d> " madly," from ^^ « mad." 

IlcSmr'ft "respectably," „ trfTf "honoqr." 
fSr^mnfr" safely," „ f5|^ "charge," "care" (f^ 

"in charge of"). 

Working round in our usual circle, we come next to Gujarati, 
which has forms in iyo produced by the rejection of the ?T 
of JJi. These were mentioned in § 18 (1), €. Tlie form in 
ait is not given in any of the grammars or dictionaries, and 
does not appear to exist, though it is strange that words like 
4^Qlc1, which breathe the spirit of the old Rajput heroes, 
should not have survived among the descendants of the war- 
like Chalukyas. Of ^^]l!ft there are a few examples, as 

IHrgit^ " recognition," smdbtf^ g JY " an acquaintance." 

^fXT " honour," ^if^dl' " respected," " a man of rank." 

^TRT "knowledge," WTftwt " skilled," " a connoisseur." 

These are also written with long f^, as WT^llWtj an approach 
to Marathi pronunciation. In the latter language the forma- 
tion in ^ is common as an attributive principally : thus • 

%'^^4 "a crackling sound,"* ^>^^^d "that which crackles,* also 
metaphorically "prompt," " smart." 

4^iJ^4J " sound of rustliug of dry grass," l^4J^in ?f "rough," "hlunt;" 

^T5^"bangl bang!" ^UI^oDd "noisy "(said of a festivity). 

^^^ " squash !" ^^^4^^ " soft," " pulpy," " squashy." 

So fond of expressive formations is Marathi, that a very long 
list of words of this class might be adduced ; they are mostly 
reduplicative. Of the form ^?!lrt7T/ one or two instances have 
been given already, as ^^ll^^, ^f^lf^ff ; another is J^JtIjTC 
"one that takes the lead," from JjTT "front.*' M. and 0. 
have from ^|^ " sword," ^^ImiT 0. " swordsman," and l^Ull;^ 
M. It would not be unreasonable to suppose that the Marathas 

Digitized by VjOOQ IC 


borrowed this word from the Oriyas ; the Khai^daits of Origsa 
formed once a very powerful and numerous body of soldiers. 
Under the Hindu Rajas they were the militia or landwehr 
of the province, and are still found as village watchmen in the 
South, though in northern Orissa they have beaten their swords 
into ploughshares, and are industrious cultivators. The khandd, 
or short double-edged sword, was the sign manual of the kings 
of Orissa, and is found rudely engraved on all their copper- 
plate grants, thus. The Marathas may have 
borrowed the word during their long period 
of sway in that country, as it is strange that it is only found 
in these two languages, and has a political significance only in 
Orissa. 0. has also 

^n^lf^cl "one who collects firewood, straw, etc., for sale," from ift^ 
"a bundle.* 

%7fn[cT "a worshipper," ''one who attends regularly on an idol," from . 
%'qT "worship." 

And with a further suiBix aka — 

Mdbl^^l "a fugitive," from Mdbl^^ "flight," for qdblf<f!^ . 
Ql^lf,^ "heir apparent to a throne," from f^miT *' the tilak or mark of 
sovereignty," H. zt^^* 

<id||^d "chopper," from ^KZPf "cutting." 

^I^ll^<1 "watchman," "guard for travellers," from JtWH "traveller." 

The gant&it was a man who escorted the pilgrims to the shrine 
of Jagannath over the dangerous and difficult roads of former 
days. His occupation is gone in these peaceful times, but the 
title remains to his descendants, who are now ordinary peasants. 

Bengali inverts the suffix into dti^ as in %^^ " worshipper," 
for %Tn[TT> but does not use many words of this stem. 

Tatsamas of this form are common in all the languages, and 
inasmuch as Ha would in Prakrit drop the ?^ and become iya^ 
and thence ly the early Tadbhavas formed by this isuffix add 

Digitized by VjOOQ IC 


another element of uncertainty to the numerous words in long 
^ in all the seven languages. This has been mentioned above 
under that stem. 

§ 20. Most frequent in Sanskrit as a suffix indicating pos- 
session is ^HH^. This forms adjectives having for nominative 
TP^ {m.), inft (/)> Wl(n.), also ^fP^, 'hR', ^. In Prakrit the 
form becomes ^n?f^, ^^ , ^pi?t. The Prakrit form is preserved 
with some modification in early Tadbhavas; but all the lan- 
guages have a certain number of instances of late Tadbhavas 
and Tatsamas with the Skr. formation, such as ^ifTfC 
"wealthy," which is in use in all. Sindhi, .however, must 
haye a vowel-ending, come what may, so it writes ^<|je||«j 
"compassionate,^^ and with very slight corruption ^ft^^RTJ 
"virtuous,'' for Skr. ^ft^RTPl.; "fif^np^ "leamed,^^ for Skr. 
fwrrr^. Of early Tadbhavas with the Pr. form ^^, the 
following are examples : G. ^ilH^ " pitiful," ^SPT^Tf "wealthy," 
'^^RfHcf "prosperous" (Skr.'^f^o)^ ^rWR^cT ''bashful'^ 
(Skr. ^r^nTf^)> but these are not very common* 

Far more general is the contracted form ^fitTf , with oxytone 
forms ^^cfT and ^cft'. With the change of forin has come 
a very extensive change of meaning, so that it is not easy in 
all cases to see by what mental process the modem signification 
can be traced back to the original idea of possession. 

In Hindi the form in '^UtTt is almost identical in meaning 
with that in ^, and indicates only rarely possession, as ^41^9 
the name of a troublesome class of vagrant beggars, who in the 
Panjab lay claim to miracidous powers of warding off evil. 
They are so called from the ^^, a hideous little instrument 
like a drum^ with a bullet at the end of a string attached to it, 
with which they keep up a ceaseless rattle ; so that the word 
woidd mean "possessing a dak,^' P. >f^d and ><^d, the 
latter for ^^^(f. Less directly possessive is H. m\ft "son 
of a I|3T'' (Skr. ^), or "husband's elder brother,'' P. t«lf?r, 

Digitized byVjOOQlC 


wliich, however, may be a compound from ^H + tRf. Hindi 
has a string of feminines in ^^SlTft, indicating generally action, 
in some of which we can get the idea of possession, by treating 
them as adjectives compounded with a fem. substantive. 

Thus ^ifiMt "patrimony,*^ "paternal inheritance,^' is ex- 
plainable as 4MH;f^ ^qf^ " wealth possessed by, or belonging 
to, a father," where, however, the idea of possession is rendered 
passive, as we could not translate it "possessing a father," 
though this would be more in accordance with the original 
meaning. We must not be surprised at inversions of this sort 
in a popular and unreasoning language. Thus, for instance, 
the Latin suffix 'bilis, which originally meant "able to do," 
is now generally used in the passive sense of " able to be done," 
amabilis, aimabk, amiabk, do not now mean " able to love," but 
"able to be loved," "that which may be loved;" and in the 
numerous hybrid words which we have formed by adding -abk 
to Teutonic roots, the same rule prevails: thus we say, for 
instance, eatabk, drinkable, meaning " that which may be eaten 
or drunk," not "that which can eat and drink." The mon- 
strous modem word reliabk, which is creeping into our language 
in spite of protest, can have no meaning at all. A reliable state- 
ment means literally " a statement that can be relied," which is 
nonsense ; we say " a statement that can be relied on" so that 
the word, if allowed to exist at all, shoidd be " relionable" ! 

The majority of these feminines cannot well, by any inversion 
or supplying additional meanings, be invested with a possessive 
sense. They refer mostly to accounts and business, and we may 
supply a substantive ^nn or mft* Thus 

ii4<f|1rf\ " act of explaining," from ^rVRIT^ " to explain," \ generally 
n^rft " settlement of accounts," „ ^||T^ '' to explain," ) used to- 
gether, P. irf, 
ififtlfV " giving security for one," „ 4(^|^| "to cause to trust," P. id,. 

Digitized by LjOOQ IC 


'J^ ^ ** closing an account,^ from ^^HlfT "to settle,* "wind-up," 

M ' ^ m " payment in full," „ «T^ « full," " comjrfiete." 

Others, again^ are general in meaning, and have no reference 
to the primitive idea of possession, as 

^f^^lft "the beginning of the {toughing season," from ^^ "plough" 
(also ^<4JhA). 

^pftlft " act of encouraging troops," from ^ipiT "to array." 

n^vH^) "supplication" (probably only a vulgar corruption of Skr. 
'PPRRnft ITNNt "modest prayer"). 

^j^Hift " flesh," " meat" (derivation uncertain), P. M. W., G. ^BTT^* 

%^41 " ransom," from a»\j|«|| " to ransom." 

Panjabi has also a fair number of words of this kind, mostly 
identical with Hindi. Gt. uses principally ^t^ aiid ^<¥t> as in 
4!«|^d), quoted above, Ij^^cf) " explanation.'^ M. writes ^nft, 
as ^^irapft, corresponding to H. ^ift?tt, or BRI, as ^<T^, H. 
41^^^. In B. and 0. the practice of pronouncing a as o 
has led to the confusion of words of this form with those formed 
by the suffix ti. In Hindi also it is a fair presumption that 
many of the words just quoted may be ascribed to the suffix 
f?f, the l[ of which is lengthened, as in the cases quoted in 
§ 12. Thus words which have ^ft added to the root direct, are 
from the simple root, while those in which the ^ is preceded 
by ^, are from the causal root, whose old form ended in du ; 
thus ^ff^ftY woidd be for a|^|HJ + fif , from ^IJl^ifT, the older 
form of the causal. 

Here comes in more confusion. In H. and P., and occasion- 
ally in the other languages also, are found words which we are 
tempted to affiliate to this suffix, but which are written with 
?. It is probable that these words should have 7t, and the use 
of Z is due partly to ignorance, and partly to their having 

Digitized by 



been confoimded with words of the form "^UtTJ, already dis- 
cussed in § 16 (3). Some are found in P. with both <f and Z, 
as in the following list : 

H. H^ " true," ^-^{^^1 " truthfulness.* 
H. fT^ " hand," ^^£\ " dexterity," P. id. 
H. fT^ "to grasp,'' f^JI^^ " steff." 

H. ^5tT " horn," f^jH^ " a burnisher made of horn," P. tU, " a small 
P. ftRRTT " teaching " (Skr. ftf^) faljl^ and flre^Zt " instruction." 
P- fW^ " upright," fll^^S " uprightness" (also ft^cft)- 

Under this head, and to be distinctly referred to the suffix 
^ by elision of the ^, and occasional lengthening of the pre- 
ceding vowel, are to be classed the following Sindhi words : 

ir^ « load," nr^ « a porter." 

Mi^ " labour," (S£r. xft^pt) iHsI^ " labourer." 

Fig^uH " debt," r^-HIHIId " debtor." 

With characteristic change of ?f to ^, occur several words 
which may with great probabiKty be ascribed to this suffix, 
though Trumpp would refer them, judging apparently chiefly 
by their sense, to the stem in ilTlf through the Pr. form tHJ. 
Such are 

Wt^ " W>" ^r^<T5 "boyhood" (quasi Skr. ^TTIRR^WI^BIIT^). 

This is the only instance he gives among secondary forms. 
When treating of primary forms, he deduces this stem from the 
Skr. affix ^IRT, as noted in § 16 (2). From my own notes I take 
the following : 

4^ "shoulder," ^^STT^ "a bullock's collar" (quasi Skr. ^IT^il^^in^ 
''machine for the shoulder"). 

§ 21. Closely allied to the preceding is the Sanskrit suffix 
IHi which is divided into two, ?E|9 being added to nouns of 

Digitized by LjOOQ IC 


place to indicate attributives, as ^f^lQIAI *' belonging to the 
South/^ and ?Qi(^ added to indeclinables to form words implying 
production, as ?nn9 " produced there/' 

The use of these two forms has been yery widely extended in 
the modem languages under the form ^, which arises naturally 
from Hf (see Yol. I. p. 327). I shall show reasons in a future 
chapter for believing this sufl&x to be the origin of the Marathi 
genitive in ^, ^, *^. At present 1 confine myself to the 
stems which it forms, merely observing that these stems are 
more frequent in, if not entirely confined to, Sindhi and Marathi, 
— a circumstance which adds confirmation to the theory of the 
genitive, by showing that the use of this sufl^ was familiar to 
the natives of those two provinces. 

In Sindhi the suffix is preceded hj i or e; the former is used 
where the stem is a feminine in short i, in which the long i 
of Sanskrit has been shortened in the primary stem, because it 
stands at the end of the word, but preserves its full length 
when the suffix is added. Thus 

Skr. ij^ "inclosure," S. %lff "thicket," "jungle," %^lfHt "be- 
longing to the jungle." 

Added to masculines in o or t^ it takes e, the old oblique 
form, as 

S. '^tS " village," 'Jlft^ "belonging to the same village." 
S. ^J^ " quarter," m^xft " of the same quarter (of a town)." 
S. tnt\ "opposite,^' in^'tft " from the other side." 

Marathi does not insert any junction vowel, as 

M. ^f^ " house," ^r^^ " domestic." 

M. ^T^ " above," q4^^| " superior." 

M. ifftn " within," ^Hf^T^ " inner." 

M. ^ " front," ^1^ " anterior." 

M. iTjJf " behind," 4| HI V| | " posterior.?' 

Digitized by LjOOQ IC 


Substantives are also formed in this way,, as ^t^lf^ " salted 
food/* " pickle/* from ^RHtf " salt.** I can hardly venture so 
far as to say that this stem is not found in the other languages 
at all ; for I think I have seen isolated instances in P. and Qt. : 
but I may safely say that it does not in any, except Marathi 
and Sindhi, attain to anything like general use. In those two 
languages it vindicates its claim to be considered as a de- 
scendant of stems in W, both from the phonetic consideration, 
and from the fidelity with which it retains the meaning of pos- 
session, combined most frequently with that of place. I cannot 
accept Dr. Trumpp's theory, which would connect this stem 
with the Skr. ika through a change of k to ch, 

Ika is one of the great k group, of which so much was said 
under the primary stems, and which must be again introduced 
here, because it is extensively worked to form secondary stems. 
"We have already seen what ika becomes as a primary among 
the rest, and shall not be led to suspect it of changing to ch. 
In Yol. I. p. 269, it was shown that this organic change, though 
there are traces of it in Skr. and Pr., is not by any means 
a characteristic feature of the modem languages, and the few 
instances in which it does occur are those of initials. 

To come to the group in ka, which need not, however, detain 
us long, as in the discussion of its use in primary stems, the 
method of its application was explained. Of aka as a distinctly 
secondary form little trace, if any, exists, — that is to say, we 
cannot point to a class of words being either abstract nouns, 
possessives, or appellatives, which are evidently formed from 
other nouns. Of ika^ in Marathi, forming nouns in ^, and of 
its connected forms in the other languages, notice was taken in 
§ 18 (1), 7, to which I have only to add some cases in which the 
^ is retained, as in Sindhi 

^imlX^ ' "mercantile," from ^TRT^ " trade." 
^tftplit " commercial," „ ^THlot " shopkeeper." 
^Tf??Rt" rustic," „ fX^ "peasant." 

'Digitized by LjOOQ IC 


In the other languages this form is rare, and generally found 
in Tatsamas or yery recent Tadbhavas, such as iH4ilf^<i in 
use in all, " relating to an assembly/* from WTT^ " an assembly/* 

^9^ is used in Sindhi to make adjectives implying habitual 
actions or states, thus also inhabitants of any country. 

Ifra " injurious," from ^xfsf " injury." 

^^" revengeful," 9, %H "revenge." 

C^lfXf^ " mountaineer" „ ^PT^ " hill." 

7^Tf^n^"amanofBakhar," „ ^^pf^ " Bakhar." 

Such secoAdary words of this type as exist in the other 
languages, as, for instance, H. "^te " a glutton,** from ^, have 
been sufficiently exhibited under primaries, and need not be 
again referred to. 

There is only one other member of the ^-group which 
remains. Sindhi adds oko to nouns and adverbs to signify 
adjectives of time, as 

^r<i8^4l " yearly," from ^rfT¥ " Y^^" 
-^JTft^" nightly," „ '^jfTT" night." 

In all these cases the final vowel of the stem is rejected, and 
the suffix joined to the bare root. 

§ 22. As we draw near to the end of this long series of stems, 
the illustrations become more and more confined to one or two 
languages, and we seem to have exhausted the powers of those 
members of the group which are least prolific. Marathi and 
Sindhi keep up the game to the last, and in this, as in all other 
respects, show themselves more fertile in varied developments 
than their sister tongues. But this fertility is not for them a 
legitimate cause for boasting ; they are, on the contrary, suffer- 
ing from an objectionable plethora of forms. Among all the vast 
range of secondary forms there is very little variety of meaning ; 
the English -y, -ness, ^ship, and -hood, do between them as much 

Digitized by VjOOQ IC 


work as thirty or forty different suffixes in Sindhi, and it is not 
pretended that each of these suffixes has in itself some subtly 
restricted shade of meaning which separates it from the oth^s, 
and renders it in a special manner appropriate to the individual 
word with which it is bound up. Bengali gets on very well 
with one or two, and Hindi with eight or ten suffixes; but 
Marathi and Sindhi cannot be contented without exhibiting 
some suffix for each of the innumerable Sanskrit pratyayas, 
both kridanta and taddhita. Under these circumstances it seems 
needless to wear out the reader's patience by going minutely 
into each one of them, more especially as some of them are 
of such rare occurrence as only to have a word or two apiece.. 
I shall therefore wind up this portion of the subject by briefly 
enumerating, with one or two examples to each, such remaining 
forms as I have met with ; and if the reader should be acquainted 
with any other forms than those I have set down — as readers 
acquainted with the spoken vernaculars probably will be — he 
will have no difficulty in determining where to place it in the 
general system. 

The Skr. suffix l{^ produces a stem in ^9JV^ in Sindhi, as 

^^ " turmeric," ^"it^l " coloured with turmeric." 

I am unable to find any traces of this form in the other 
languages, except in Tatsamas. The ^ is dropped, giving a Pr. 
41^^', 3^d the if softens into anundsika, with lengthening of its 
vowel, which is then brought forward in order that the anund- 
sika may be utilized in filling the hiatus. The process as a 
whole seems imique, though resting on phonetic changes for 
which there is abundant analogy. 

•Sindhi has also stems in irUy from the Skr. suffix T, in 
which it stands alone, as in Skr. "^ forms diminutives, in 
which practice it is followed by the other languages. Sindhi, 
however, makes appellatives and attributives, as 

^fv^ l ^ '' housebreaker," from ^fy '* a hole made by burglars." 
VOL. n. S 

Digitized by LjOOQ IC 


The same language has also a series, peculiar to itself, of 
stems in ^!infft> T^* ^Wft> which are in the main possessives. 

'q^ " thief," ^4^<inn " of a thief." 

^ " moon," ^rttl'^ " moonHght." 

%ft[ " merchant," %^l[lt " of a Seth." 

Connected with which are stems in 4n^, as patronymics, thus 

And in ino, as 

^"^luH '' affianced," from ^i^ '* connexion by marriag^e." 

To these may be added dtho, perhaps from Skr. WZ : 

qT^ldlr " damp," from xrn^ " water." 
And dso, from Skr. II, as 

^ I I^IMV " sandy," from ZfT^ " sand." 

As to the origin of all these forms there is much doubt. 
Standing so much alone as they do, and unsupported by corre- 
sponding forms in the other languages, it is difficult to know 
whence to trace them. Trumpp deduces the first from the Skr. 
suffix fyi, which, however, does not seem sufficient for the 
vowel-changes in dno and Ano. The second is probably from 
Skr. ■^Rraf'Tj and the third from f^. 

There are several other forms given by Trumpp, but most of 
those omitted from this list are to be accounted for under the 
verb, and may therefore be passed over at present. The same 
holds good of several stems in Marathi, such as those in Hl^, 
ift and the like, and in Gujarati. • 

§ 23. Some of the languages are not content with one suffix, 
but will take two in the same word ; in this case it is generally 
ika in one or other of its forms which does duty as a second 

Digitized by VjOOQ iC ' 


suffix. Thus in Sindhi stems dno^ uno, we find subsidiary forms 

dnikoy ULTttko, as 

^^ttrftNrt "of a thief," from 5EftT + ^'TO + T«t- 
^^pB^iEt " preceding," ^, ^ + ^mr + IC'lt- 

These double forms do not differ in meaning from the single 
forms, and it is probably to this tendency to use ika as a mean- 
ingless expletive, which recalls to us the similar habit in 
Prakrit of attaching a ^ to aU and every Sanskrit noun at will, 
that we should refer a numerous class of obscure words in all 
the languages, in which, by the side of the regular stem, we 
have an augmented stem in id or ud. Thus 0. uses 4lf|l||Mn(f^l 
for " the business of a Brahman,^' where we have the suffix ilTlf 
augmented by i[^= ^. Rustic Hindi has ^f if^^l "milch 
(cow),'^ which is '^^ + W^ + l^^i. In the Bhojpuri dialect of 
Hindi the terminations id and ud are added by the vulgar to all 
nouns, whether primary or secondary, without altering the 
meaning. Other cases, where there are apparently two suffixes, 
would be more accurately described as compoimds, where the 
latter member, being a verbal root, has become so common as to 
look almost like a suffix. 

§ 24. Diminutives, as well as a nondescript class of words 
either contemptuous or jocose or familiar, besides jingling and 
fanciful formations, are very common in all these vernacular 
languages, and in m^ny cases the terminations in themselves 
are absolutely meaningless, and incapable of being solemnly 
and scientifically traced back to the ancient languages. In 
such a maze of playful or vulgar developments it would be but 
waste of time to attempt an elaborate arrangement ; the human 
mind makes a tool of the tongue, and strikes out for itself 
sounds which satisfy its needs without regard to grammar or 
etymology. It wiU be more useful and more interesting there- 
fore to state wl\at forms the languages employ, than to try and 
find out why they employ them. 

Digitized by^VjOOQlC 


The relative size and quality of material objects are elegantly 
shown in aU the languages l)y change of gender. In tliis 
respect they agree with the old Teutonic family, and other 
members of the great Indo-Germanic ckss. Masculine nouns 
express objects which are large, strong, or coarse; feminine 
those which are small, weak, or fine; and where the neuter 
exists, it expresses things dull, inanimate, or sexless. This will 
be more fully drawn out in the next Chapter. 

The commonest type of the diminutive is that which has for 
its characteristic the letter ^ ; the words of this class occasion- 
ally end in this letter simply, but more frequently take the 
long vowel-ending, as ^, xi^,*^, ^. This ^ appears to have 
arisen from the Sanskrit suffix ^, which in that language also 
forms diminutives, and is retained under the forms "^T, "'(t? etc., 
in many of the seven languages. It also occasionally modulates 
into ^, as might be expected, and in a few rather exceptional 
cases, chiefly in Marathi, appears as Y. In point of meaning, 
this termination has a wide range, from words which are pure 
diminutives to those in which the sense of smallness is only to 
be made out by researches into the original meaning ; then to 
those in which the idea is that of contempt or familiarity, fond- 
ness or trifling ; till at last we come to words which have lost 
all sense of smallness, and are simple appellatives. In this last 
class it is often impossible, or, in the present state of the subject, 
very hazardous, to suggest a primitive word, from which that 
in actual use may have been formed. The words of this class 
are the pure offspring of the popular mind, and consequently, 
though the principle involved in their formation is the same 
in the whole seven, yet the speakers of each language have 
formed their own words separately, so that, except in closely 
allied languages, as Hindi and Panjabi, it is rare to find an 
instance running through two or more. It will, therefore, be 
advisable here, as it was in many of the primary and secondary 
stems, to take each language separately. 

Digitized by 




* Sindhi is clearer in respect of tliis form than its sisters, 
Laving a greater number of the words of this type in nse as 
pure diminutives, and not so many as contemptuousi familiar, 
or appellative terms. Trumpp gives the following : 

li^ "journey" (Skr. tf^ " path") t^^Tft " short journey." 

^f^ " shop," ^tTVt " small shop." 

^ " life/* Hi^ " short life." 

f^ " lightnmg " (Skr. f%^), f^Hfi^ " brief flash of Hghtning." 

In the case of words ending in u masc, the suffix is either 
added to the oblique form in a, or the final u is changed to 
i, — ^in the latter case probably from the influence of the i of the 
fem. termination; but where the u is the feminine ending, it 
remains unchanged. 

Similarly, nouns in o change that letter to a or t ; of the former 
an instance is fif^T^ " sweetheart," from fif^ " heart ; " of the 
latter Htf^¥t " a small monkey," from Ht^ " monkey.^' 

Feminines in a and i retain those vowels unaltered ; but i and 
6 are changed to ia and ua respectively, or in other words the 
suffix is added to the oblique form, as 

?hSt " churning staff" (Skr. 7T«8|); ^ttvfl^'ft " & small churning staff." 
'¥[Z" scorpion," ^TZ^IRA ** a small scorpion." 

There is also a suffix ro, joined only to adjectives in the ob- 
lique form (nouns in o make their oblique in e), to signify some- 
what of the quality indicated by the primary, and corresponding 
to the English termination " -i»A," as 

f%^ " long," r^VlO " longish." 
ift^ « smaU," ^XO " smalUsh." 

This latter often takes a double suffix, as ^"XR^J "very 
Marathi, with its customary fuUness of forms, has a wide 

Digitized by LjOOQ IC 


range in diminutiveB, and sometimes uses two to one word. 

IttZ "knoV* i llaf^ "a bundle," gtlAl>H "a smaU bundle," also 

^rf¥ "piece," ^rtnTl^ " a smaU bit," also IQtl^. ^i><f%;. 

fH^ " rag" (f^tlft), f^V|^' " a poor rag" (contemptuously). 

^Sfm "leather," ^TT^ "skin," "hide," ^Hfit "the human skin." 
Here the sense of diminutiveness is lost. 

IT^r " village," IftBf^ " a small village." 

Skr. JIJ^ " young of an animal," ^|q>91 " darUng," " little pet" (said 
to a child). 

Marathi has for adjectives also the termination Wt, cor- 
responding to Sindhi "^^y, and, like it, with the union* yowel 
e after oxytone nouns, as 

irtn " ^^^y" " light-coloured," g [ Y^y r " fairish." 

But it uses the form in TT also for this sense, as wNtlTT^ 
Wt^ttdSl "longish," from ^lt^ "long.'' 

The termination >ft, xft, if, is extremely frequent in (Juja- 
rati, so much so indeed that words of this type have in many 
cases lost the sense of smallness, and stand alone for the primary 
idea, having pushed the old primary word quite out of the lan- 
guage ; such are : 

^TfTlt» ^Tft "day," from f^[^9, through f^, ^fT- 
lftft^«face," » ^ ». ^• 

That this latter word is strictly a diminutive is shown by 
the existence of the intermediate form ^^^9^; the original 
word ^i§ meaning " the whole face,'' ^J9^ is restricted to the 
sense of " mouth,'' and this brings it under the definition of a 
diminutive, because that class of words not only describes a 
smaller article of the same kind as the primary, but also an 
analogous or allied article of a pettier, narrower, or partial kind. 

Digitized by VjOOQ IC 


With the pure sense of diminutioii are found the following 
words in ordinary use : 

J||4|^ "hamlet," from TPTTf " vUlagc." 

TrtriY " smaU beU," „ trfe«beU» 

ftR^"KttledarHDg,* „ ^ " life," P. fsRlt. 

^^n^fN^t " somewhat hungry," „ ^^i^t ** hungry." 

ll^taf^ '* a small bedstead,'' „ X^^lf <<bedstead" (Skr. Viki^). 

^i^\4['4\ '* a rascally sham hermit," „ %^T^ '' an ascetic" 

^tf^q/t " ft trumpery old idp],*" „ TTfT^ '' an idol of Siva." 

These last two are contemptuous, a very common use of the 
diminutive in all languages. Familiar and slightly contempt- 
uouSy but at times with somewhat of a kindly meaning, are : 

Ifpgf^t " a wife," from IR^ " lady." 

^Tl[Hft " husband," „ ^rrf^ " brother* (used as a respectful 

term to males). 

These two words correspond to the vulgar English "the 
master/' and "the missis'' (mistress), or to the rustic terms 
" the good man," " the good wife," as in the Marathi nrljU. 

<llj1il "ashes," « rubbish," from ^^ " ashes." 
^t?:^" a string," „ ^ "a rope." 

il^nft" a footpath," „ ^7^ "street." 

As in Marathi, so also here, the termination ift (Sindhi '^) 
is used as a diminutive, thus : 

^1^^ "a small crescent used as an ornament," from ^ft? "moon" 
(Greek firivlaKO^^ Latin lunula). 

MI^H) " a small writing-board," from V[X^ "plank." 

Hindi and Panjabi go together; and (Jujarati, in virtue of 
its position as an isolated dialect of Old-Hindi, exhibits often 
the same words. In the first-named language the tendency, 
already mentioned, to express smallness by changing a mascu- 
line word into a feminine^ operates to reduce the number of 

Digitized by VjOOQ IC 


diminutiyes of the type ^, and a large proportion of the 
words of this type are now simple appellatives with no special 
sense of smaUness. Common to nearly all the languages are 
the following words of this class, whose meaning is in most 
cases somewhat obscured. 

H. tnrft " hut," « hovel,'' P. ^Jfq^, G. ^, M. ^t^rft , ^gttf^, 

O. tiM>|t> B. ^ff^Hlt- The primary word ^(tf is not, as far as I know, 
in use, but it is probably connected with ^jflf " covering." 

H. ^(WW\ "leather," literally, "a small piece of skin" (see Vol. I. 
p. 345), found in all. 

H. ^ft^V>3T " boy," Skr. Iff^Ri " young of an animal," perhaps already 
a diminutive from "Jf^f "body" (see Vol. I. p. 261, for the numerous 
modem forms of this word). 

H. fZ^IT^ and fz^iwt " a wafer," "a small round cake," probably a 
diminutive of '£\^\ '* the round sectarial mark which Hindus paint on 
their foreheads ;" this word is derived from Skr. fTI^QPli (see Vol. I. p. 226), 
which ag^ain seems to be a diminutive from f?f^ '< a grain of tU seed." In 
P. and G. Zf^R¥t> %f^fi>ft> means ''a small round hillock;" and in the 
various meaning^ given to the cognate words in the other languages we 
may detect as a general thread of meaning running through them all that 
of any small round object, as " a mound," " lump," " cake," " wafer," 
" piece of broken pottery," and the like. 

H. ^tT^ " a vagabond Domy* from NffV " ^n impure caste." 

H.tWf"lazy," „ *twr" slack," 

H.;TO^" bundle," „ irra"knot." 

H. ^4^ "fishhook "(smaU hook), „ ^«hook." 

H. ^A^TVt " glance of the eye," „ ^|t^ " eye." 

H. "WZP^ " small upper room,*' „ ^fZT " upper room.* 

H. ^>i^T " small or bad egg,"* „ ^H^ " egg^*" 

H. ITOTT " skin," " hide " (contemptuous) „ ^TTW " skin." 

H. ^UTT , •^ " smaU piece,* ,» ^ " p5ece* (Skr. ^f^). 

In some of these, in Hindi fashion, ^ and T are interchange- 

Digitized by VjOOQ IC 


able, and additional emphasis is given to the idea, of smallness 
by the use of the feminine termination ^. 

Hindi has also a few words in Wl, with joining vowels e and 
0, as ij^HVI " a young peacock/^ from ^R^T " ^ peacock ; " iliJ^HI 
"young snake," from ^^rf^ "snake;" or without the joining 
vowel, as ^(^^ " a bamboo flute or pipe," from ^tNT " bamboo" 
(Skr. ?tij), also written '4(^0 ^^^ ^NO* There is also a 
form which is interesting as exhibiting the retention of the 
Sanskrit affix i^ in a diminutive sense ; 7V%^ " a small pond," 
from TfT^ "a lake ; " ?|%^ " a small hut," from W^^ " a shed." 
The first of these words is derived from Skr. Tf^PT, Pr. 7raT^> 
through a fem. ^TSTf'niT, of which the Pr. form would be 
^^If^l ; the last is from if^ftf^. The masculine form of 
this suffix is seen in ^t^RR "a small drum," from ^t^ "a 
drum," and perhaps in the rare form ^, Gt. Tft; aka in Pr. 
becomes ao, and the ^ is inserted to soften the hiatus, as in H. 
^9^, from infT " calf" (Skr. '^TO) ; G. ^t^ " small sprout," 
from ^5t^ " sprout ;" ^T^^ " small bush,^^ from ^JUT " bush." 
In some f eminines from originals with barytone or mute a ter- 
minations, the diminutive is J^, as 

^f^nrr "small mango,*' from ^)^, ^|lf ''mango.-" 
^f^nrr^boddice," „ ljir"body." 

In B. and 0. the forms of the diminutive ^ are generally 
the same as in Hindi, and, as in that language, have to a great 
extent lost their special meaning. In 0. especially, and to a 
great extent in the others, the addition of ^ may be made to 
almost any noun at the will of the speaker, and would in most 
places be perfectly understood as indicating contempt. 

There is to be found more or less frequently throughout the 
group another diminutive form in Z, generally with joining 
vowel ^ or 0. If we consider this Z as allied to the ^ of the above- 
named form, we should have to put it first as an older form, 
which has been subsequently softened into ^; but there are 

Digitized by VjOOQ IC 


difficulties in the way of this process. The junction vowel 
of the forms Z and ^ is either o or e, which two vowels repre- 
sent, in mjr opinion, the nom. masc. and oblique endings 
respectively of the noun, whereas the junction vowel of the ^ 
form is generally a, only occasionally in M. we find o. From 
this it would result that ^ is used with words whose ending is a 
mute, or a consonant Z and ^ with oxytones in o or 4, oblique 
e\ and inasmuch as these vowels in a large number of cases 
indicate the presence of the suffix ha in Prakrit, words formed 
with them would be early Tadbhavas, as H. 'J^WT and the like, 
while the words in .^ would be to a great extent late Tadbhavas. 
Instances of the Z type are — 

8. ^ijlj^ " smaU fire," from ^TTft " fire" (Skr. ^f^). 
S.^STTiftZ" village headman," „ ^TT^ " a chief." 

Trumpp, from whom I take these instances, points out that 
the diminutive in Pashtu is regularly formed by utai. Bellew 
(Pukhto Gr. p. 108) restricts it to nouns ending in u/, and 
gives as instances : 

^j^ sarai, "man," j<5§*«9 sarofai, "a small man" (which may be 

written witZf)- 
^J largai, "stick,** ^^ largofai, " small stick," (JK^ftd^* 
^^^ jina-t, "g^rl," ^^ jinotat, " Uttle gW," tMYzt. 

The existence of this form in Pashtu is a confirmation of its 
antiquity, and justifies my attributing it — as I did above from 
independent reasons — to the early Tadbhava period. It is not 
so common in the other languages, though instances are not 
entirely wanting. 

In M. t^\JIZ and f^vft'^' " a wretched rag," from f^^ " a rag ; " 
but more frequently without the junction vowel merely, as ^, ^^(tWZ ''a 
small post," from ^^rf^ "post;" ^^ "high," ^^Z "bighish;" ^giT 
"salt," ^rrr^ "saltlsh," "brackish;" ^ db^T "scorched," m<J6mi. 

Digitized by VjOOQ iC 


"slightly scorched;" %q "pressing," %tIZ "flattish;" f^ "wet," 
^^Z "moist." 

H. ^r^tft **a small strip of leather," P. G. id.y M. ^qTRtZT» •^Z 
from ^HT "leather;" (f ^uH^I "* fawn," from ff^^ "a deer;" G. 
?r^ " a writing," W^tft " a note," " a smaU letter." 

§ 25. In the formation of compound words the modem lan- 
guages, while lacking the exuberance of the Sanskrit, are not 
altogether wanting in strength. It will be useless to discuss 
those words which have been formed in Sanskrit, and borrowed 
whole from that language in modem times. The modem 
grammarians needlessly bewilder their pupils by leading them 
through all the mazes of Tatpurusha, Bahuvrthi, and all the 
other methods of making compound words, which are in use 
in Sanskrit. In the present section it is proposed merely to 
note the instances that have been found of compounds made in 
modem times principally from Tadbhava or Desaja materials, 
and to endeavour to ascertain the law that imderlies their 

That simplest of all methods of forming compound words, 
which the Sanskrit grammarians call Tatpurusha, is still in 
force in our seven languages, as it is in most modem languages 
of the Indo-Germanic family, being a special characteristic of 
that family, and surviving through all the changes brought about 
by time. The seven languages therefore have formed com- 
poimds of this sort from their own stores, from Tadbhava and 
Desaja words. Under this class are included those compounds 
formed from two nouns one of which governs the other; 
familiar instances in our own language, which is rich in these 
words, are shipowner, horseman, homekeeper, and the like. In 
these words we have an inversion, the governed or dependent 
word being placed first, whereas in the separate construction 
they would stand last, as owner of a ship, man on (or with) a 
Jiorse, keeper of a house. Further, it is not only in the genitive 

Digitized by VjOOQ IC 


relation that the dependent word stands ; according to the gram- 
matical rules of each language, it might stand in many of the 
. other cases or relations. In some of the languages, consequently, 
the dependent word, when put first, takes the oblique form, 
which is common to all cases of the singular, and by itself 
denotes simply the state of dependence, the particular kind of 
dependence being indicated by case-particles, which are omitted 
in the compound. Thus : 

8. M^^UJl " housemaster," from ^^ ^ V^ " master of the house ; " 
S. ^ ;f <q|4 " north wind," „ ^TPC Wt ^^ " wind of the north ; " 

— where ^^ ^, ^TPC ^> are the genitives respectively of TBT^ 
" house,^^ and ^Rf^ " the north,^^ which ending in the nom. in 
Uy form their oblique in a. Similarly, in 

8. 9?i)i^T^' " head-eater," from jf^ ^ l^[^ "eater of the head" 

{fig. tormentor) ; 

8. <i|i<f*1j>dt " hand-broken," „ ^B^fif ^ ^^ " broken of hands " 

(Jig. lazy) ; 

— ^we have in the first the oblique singular in ^ of the noun 
^T^, and in the last the oblique plural in ani of the noim ip|. 
Those languages which use the Persian character obscure the 
real nature of such compounds as these by writing the two 
words separate, or rather — inasmuch as some Persian letters 
cannot be joined to the following letters, and native scribes 
seldom leave any space between their words — it is impossible to 
say whether the word is written as one or two. Thus in the 
Persian character ^J»^<^j^ may be meant for one word or two, 
for all that we can tell from the writing. It is only in the case 
of those letters which have a separate final form that we can 
tell : thus jL^ \Jf^ must be regarded as two words on account 
of the final long 4/; but if written jl^Cf^, it is one word. 
This is, however, merely a detail of writing ; in speech, these 
words would be regarded as one. 

Digitized by LjOOQ IC 


In Hindi the rule of placing the dependent word in the 
oblique form is disregarded : only one class of nouns, that in d, 
has an obKque form, and this is not cared for; thus we have 
from Ml^l " horse,^^ oblique ^^¥%> the following compoimds : 

^'S'qcr " horseman," from vh^ ^ ^CT " rider on a horse." 
^"^^^ " horse-race," „ 1^^% ^ ^^ " runnings of a horse." 
^l^TT^ " horse-stable," „ vn>?I ^ ^arW (mWD "house for horses." 

In all these cases no sign of the oblique form remains ; but 
the Hindi, being sensitive about quantity, shortens the vowel 
of the first member : thus d becomes a, i becomes «Vtt becomes w, 
and the diphthongal vowels o and e turn respectively into 
u and u 

Words ending with the long vowels d, i, {l, also generally 
shorten those vowels, in which case the so shortened vowels 
drop out altogether : thus, as we have seen, ift^ becomes in 
composition ^^; so also ^T^ becomes Jm, and ^^ be- 
comes ^ . 

The following examples will illustrate the above remarks : 

ITTOTWT .^ ^^j^^„ from TTpR'" water," and ^ITO[T«haU." 

tJifllZ " a quay," „ ViT^ " water," and ^ETTZ " a ghdt or 

M*ien>f\ « a betel-garden," „ TfR " betel," and ^T^ " garden," 

also written TJ^TT^- 
^^TQ^ " a kind of firework," „ IJJ^T " flower," and ^^ " cascade." 
iRfir^W a prolix talker," »> ^"RT "word," and ^^J^ "pro- 
l^tlTTT " * highwayman," „ mz " road," and iHT " striking." 
^fi^U^I (^trr) "iron pot," „ ^iftfT " iron," and ffsST "pot.** 
^f^ " iron filings," „ ^tfT " iron," and ^ « dust." 

There is also a class of compounds of the Tatpurusha kind in 

Digitized by VjOOQ IC 



which the first letter of the second or governing member is 
eKded, as — 

H. '^%W " an oil scented with flowers," from TM " flowerj^and B^ " oil." 
H.^Qf^rh^ "a smell of rotting substances," „ ^^^ '' rotten," and 9^9 

« smeU." 
This class is very numerous in Marathi and in Panjabi, less 
so in the other languages. Instances in Panjabi are — 

^ilfnS^i^ " smell of burnt cloth," from ^ll|4| " clothi" and 1^ " smeU." 
HrfX:^|hsr" smell of a camel," „ l|R;fT" camel," „ id. 
^f^T^lt^ " smell of raw meat," „ W^ " raw," „ id. 
^rfCTlN"8onr smell," „ ^r|T"80ur," „ id. 

— ^in all of which the final long d of the first elemtot in the 
compound is shortened to i. 

One familiar set of words, however, runs through all the 
sevai, namely that formed by "JlT^ " maker.^^ These have been 
partially given in Vol. I., but may be again detailed here — 

Skr. fBpj^irTT "goldsmith," Pr. ^imO > H. T^^nUi, P- ^Pt^K > 
Skr. WOTTTT "potter," Pr. ^pf^HTTt* H. W^fTT^ P- ^^y M. G. 
Skr. ^^^ITT "leather-worker," Pr. ^M|^|<t, H. ^RFiTT* P- C^. id., 

The Chingana retains the older form in ^CTTj ^^^ ^ *^® 
plural with softening of k to g, as — 

machengorOf " fishmonger," from macho, *' fish." 
maseskaro, " butcher," „ mas, " flesh." 

The former word is constructed with the plural because it 
implies " one who sells many fish ; '* macheskoro would mean " a 
man who only sold one fish.''^ 

1 Paspati, Tching. p. 63. 

Digitized by LjOOQ iC 


Similar is the treatment of ^sTTT "holder," in 

Skr. ^^ g W TT " carpenter," Pr. ^^g f O (?)» H. ^jmT* G. id., B. O. 

^n: (i7!K). M. ^j^nr* ^'^^^ dim. ^^K><i > s. ;|zf^ ; 

Skr.l^lnTOTT "Palanq«een-^^«5r," Pr. iJ^SCfnct (?) or ^iif?T<t» H. 
^IffH!, M. *«/. ; 

— ^where the aspirated letter has been changed to 1 and 
ultimately elided. Further illustrations are — 

H. ^1%^ " scented oil," from ^fff ** sediment of an oil mill," and 

H. '^(VJ'fl " name of a caste," „ 4^m\ " king," and Hlf " messenger," or 

i^m\ "king," and yf for ^Hf" son." 
M. \)|Xf^ " a resinous oil," „ Wf " smoke," and ?f^ " oil." 

ic o siYiell aF I 
„ „ lSimZ"80ur,"and^mir"8i»eU." 


a smell of 

• Af Ani »> ^riz" burning," and iBnnr"s™«ll-' 
singed food, ) 

Marathi has numerous compounds indicating various kinds 
of prepared, scented, or medicinal waters, the last element of 
which is xrnit '^water,^^ changed into ^pjt : thus— 

i|^e|([f|' " mango-water," from ^«(f '* mango," m^ " water." 

fldb^uf) " water heated in the sun," „ ^jgg " sunlight," id. 

Of compounds which would be classed as Karmadh&raya by 
Sanskrit grammarians, many have been preserved from Sanskrit, 
but the modem languages have created some out of their own 
stock. Trumpp gives the following from Sindhi : 

^nrraX^ " a glutton," l^rU " great," ^^fT^ « eater." 

Vm^O " weU-wisher," ^Rlft " much," ^ («./.) " wish." 

But others which he adduces, as 4{fT^^ " a merchant,^' from 
7f|X-h^i|if=''a great man," and TRT^it^ "the other world,^' 
are Sanskrit, and are in use in all the languages. 

Digitized by VjOOQ iC 


The following instances of modern Xarmadhd.rayas are from 

^T^^^'big-bellied," from UTT " Wg," and^«bcUy." 

if^Tirr^'' polite,"! f, HWT "good,*' „ ifpj^" man." 

^Rfcfirr " long-legged," „ ^m^ « long," „ tm " leg." 

^|\;Rim^ « hemiplegia," „ ^ffW " half," „ HHTTW "skuU," 

^ra^r^ "hdf-ripe," „ ^ITOT "half," „ W^ "raw." 

?5Rra^"ahare"(^#.long.eared) „ WWr^long," „ ^if "ear." 

The adjective in Hindi typically ending in d, an adjectival 
termination has been added to the nouns which form the last 
member of these compounds, in all cases in which the com- 
pound itself is used adjectivally ; and the first member of the 
compound, if it contains a long vowel, has that vowel euphoni* 
cally shortened. 

Marathi is very rich in compounds, idmost more so than any 
other language of the group. A few instances of Karmadh4- 
rayas are here given : 

c|ildb^i^< " a pole-cat," from llldbl " black," Mlm^ " cat." 
TTNlfimft" red earth,*' „ Tft^TJT " red," iTpft "earth." 
^rf^nfTOT"loDg-legged," „ ^ft^"long," cTN "leg." 
^Idbcf(v&|| " lackless," „ ifiTdbT " black," ffflf « face." 
«lO^W " prosperity," „ ^TTT " good," 1J?T" condition " (^tRt). 

In Gujarati, as far as I can judge from the works available 
for consultation, the practice of making compounds out of 
modem elements is not carried to so great an extent as in 
some of the other languages. Instances are : 

>1^4i^^l|^ " politeness," from ^J^ "good," M\^H " man." 
^ft^in^pr" a grandee," „ ift^t " great," HTipr"man." 

Dvigu compounds, namely those in which the first element 
is a nimxeral, are exceedingly common in Hindi, and ^almost 

Digitized by VjOOQ IC 


equally go in the other languages ; thus with ^ " two/' as the 
initial element, sometimes shortened to ^ : 

H. ^WTPRTT " an interpreter," from '^mj " speech." 

H. ^q^T " a cloth of two breadths," „ IJgT " strip of cloth." 
H. ^^>ij " a necklace of two threads," „ ^r¥ "thread." 
H. <(\41iH "pregnant" (with two lives), „ ^ « life." 

H. ^^ffffV " a two-anna piece," „ ^VPTT " •" anna." 

With cfVir " three," shortened to ft : 

H. fTT^rrr " & necklace of three threads." 

H. f?ri^rftRrr "» building with three doors," from iftw "a door." 

With ^rq " f our,^' which becomes in composition ^, i.e. ^^ 
from Tfg<^, with dropping of ?^, according to rule, and elision 
of final T : 


H. xn^ifl "a four-anna piece." 

H. ^4^ " a framework for a door," from ^JTZ "wood." 

H. ^^Hqft "earring," „ Hfit "ring." 

H. ^Vfi^T "a box with four partitions," „ ^|^ "house." 

H. ^'^ 44111 " the four months of the rainy season," „ Wm "month." 

All the compound numerals ar6 Dvigus, as will be noticed 
presently. It is not necessary to give illustrations from the 
other languages, as they are formed in precisely the same way 
as in Hindi. 

Dwandwa compounds, strictly speaking, do not exist. The 
Dwandwas of Sanskrit consisted of two nouns or more com- 
pounded together into one word, the latter of which took the 
terminations of the dual noun of the a-stem, or that of the 
plural, according to the number of things expressed in the 
ToL. n. 9 

Digitized by VjOOQ IC 


compound. But as the modem languages no longer possess 
the inflexional terminations of the old synthetical languages, 
it is clear that no Dwandwas in the Sanskrit sense can exist. 
There are abundant alliterations where two or more words are 
put close together, and only the last word takes the case-aflBxes : 
thus one may say, ^T^ vh^Y '^ "on elephants and horses ;" 
but these are hardly compounds. The same remark applies to 
the very numerous cases where a second word is added to the 
first so as to jingle with it, such second word being for the 
most part quite meaningless, as in S. ^RS ^pR " talk,^^ where 
If^ has no meaning ; such are, ^?cre Vim H. " near,^^ where 
^JJire is meaningless ; ^3?rZT J^^fZT " upside down,^' and many 

Bahuvrlhis are easily formed in all the languages by giving 
an adjectival termination to any of the foregoing classes of com- 
pounds, and in fact several of the examples given above are 
Bahuvrthis. Further illustrations are unnecessary. 

§ 26. Numerals, — The cardinal numbers up to ten inclusive 
are simple adjectives, some of which are declinable, others not ; 
from eleven up to ninety-nine they are, as was stated in the 
last section, compounds to a certain extent of the Dvigu cha- 
racter. Each numeral, though a strong family likeness runs 
through them all, stands on its own basis, and is derived 
directly from Prakrit, through the operation of the phonetic 
laws of its own language, arid it will therefore be necessary to 
examine each one separately and in detail. 

One.— Skr. ^iR, Pr. ige, H. tpp, P. ^, S. tff ,^, HJ, 
fff^lrit^ \f^irit (Sindhi is never contented with one form) ; 
G, Jf^y usually written %^ ; M. TJ^, ^qf, 0. B. n;^ ; Gipsy yek. 
It is indeclinable in H. and P., but may take case-affixes when 
standing alone ; the two Sindhi forms in \fft are diminutives, 
and that language has put on an |[ in front, the reason for 
which is not apparent. B. pronounces it very short and harsh. 

Digitized by VjOOQ IC 


something like yaek^ rhyming with the English words pack^ 
staek. The H is long in all the other languages except P. and 
S., and the double ^ of Prakrit has therefore been reduced to a 
single. Old-H. has constantly ^ and igj. 

Two.— Skr. base % and % Pr. ^, H. ^, P. ^, M. ^sft^, 
0. B. "5|[, Gipsy dui; but S. ^, G.%. Here S. and G. stand 
apart from the others, they having, as pointed out in Vol. I. 
p. 331, assimilated the ^ to the ^, and thus made g", out of 
which S. gets its peculiar sound ^ ; S. appears to have taken 
the base fl[, and G. the base f^. H. and P. have probably also 
taken the base fl[, and resolved the semivowel into its vowel, 
thus getting ^^, whence ^; the forms ^t^, ^tt^, T1[, and 
^ are, however, f oimd in the earlier poets. M. ^g^ff contains 
the type of the neuter plural; this is unknown in Skr., and 
has probably been extended to this numeral from M. ?f^ 

Three. — Skr. base f^, nom. pi. neut. '^|Vftr> whence Pr. 
ftf^, and Gipsy trin, H. Ht^, P. fj^, G. 71^, M. l^, 0. 
f?lfif, B. "fifif. All these are neuter plurals by origin, and 
from this, as observed above, M. has apparently imitated its 
• i^ff "two." P. has also a commonly used form ^, which 
agrees with S. %, as far as the subjoined ^ and vowel are con- 
cerned ; but S. has changed 7f into Z, under the influence of the 
following '^ : both these forms lead back to the Skr. masc. ''SRi:. 
If it be asked why two languages out of the group should adopt 
the masculine, while the others take the neuter, it might be 
answered, that S. and P. (especially S.) are often more archaic 
in their forms than the other languages, and that H. has pro- 
bably not deliberately adopted the neuter plural ; but as three is 
the first numeral that has a distinctly plural meaning- — one and 
two being respectively singular and dual in the parent speech — 
H. has indicated the idea of plurality by adding the sign which 
indicates plurality in the noun, where, as it will be seen here- 

Digitized by VjOOQ IC 


^fter, the terminatioa of the neuter plural has usurped the pW^ 
of all other plural forms. Kashmiri, a very archaic direct, 
uses f^, and we may therefore place it, together with the 
Gipsy, P. and S., on a higher level of antiquity than the others ; 
Gipsy by its additional 1 (unless this is to be considered merely 
as anundsika) standing halfway between the three western 
languages and H., and its followers 0. B. and G. It is charac- 
teristic of G. to be careless about short vowels ; THJ for f?(^ is a 
parallel to im "ye," for g^. 

Four. — Skr. ba^e ^Tf^, neut» pi. ^^^fTfT> whence Pr. xiTflfi;, 
B. and 0. ^TfT> S. ^qif^Q, H. P. G. M. ^TT, Gipsy star, 
shtar, ishtavy probably by inversion from tsar, ^ having in 
M., and possibly formerly in S. also, sometimes the sound of 
ts, also in Kashmiri ^or = ^^tT, a soiind due to non- Aryan 

The entire loss of the nexus ^, which had become ^ in 
Prakrit, is unusual, but possible in the case of so-much-used 
a word as this, and it need not surprise us when we think of 
1T?^rPl=T!l1, colloquially Trf. 

Five. — Skr. base ^^(% but the final n drops in the nomi- 
native, and in fact throughout the declension. Pr. Tj^, H. 
v^y and so in M. G. B. and 0., but S. and P. have ij^, with 
the tenuis softened to its media. Gipsy and Kashmiri have 
pdntsh, which is written in the latter trf^, and M., though it 
writes trf^, pronounces pdnts. 

Six* — Skr. base tf^, but nom. ^Z, which is also the declen- 
fflonal base. Pr. ^, whence H. ^, and dialectically*^, P.*^, 
S. ^ and 15^, G. ^, 0. ^p^ (pronounced chhdh), B. ipij; but 
M. ^ifT, where the ^ is regular Marathi for ^, see Vol. I. 
p. 218, and the ^ akin to S. |pf . How If came into ^ it does 
not concern us to inquire, as the process took place in earlier 
times than we are dealing with ; Kashmiri has still ^, as fjf^,, 
with. which may be compared Gipsy (ift, iflw) shQ,. sho% 

Digitized by VjOOQ IC 


Seven. — Skr. ^, the H^ being dropped as in ti^t, Pr. iRf , 
whence H. ^THT (see Vol. 1. p. 287), where this numeral is 
treated of as iUustrating the' strong nexus). The Gipsies use 
eftay which shows traces of Greek influence, eirrh becoming 
e<fnd; as Paspati (p. 77) points out, there is a frequently oc- 
curring change of tt to ^ in the transition of words from Greek 
to Chingana. 

Eight. — Skr. ^i|^ or ^[re, Prakrit ^|^, H. ^RTJ, P. ^Uf , 
S. ire, M. G. 0. lira, B. "mz. All these are regular phonetic 
changes. Kashmiri has also m?, but Gipsy again approaches 
the Greek with ochtOy which Paspati says is modem Greek o;frci 
for o/icrco. 

Nine. — Skr. if^, Pr. H^, H. "^, which is also written inr> 
P. Tfff or ^, S. ^, G. M. 'i^r, O. ^nr {n6h)y B. ^TO, Kashmiri 
^ and ^, Gipsy again Greek enea^=:ewed. Had the Gipsies 
only learnt to count as far as six when they left India? With 
the exception of ten and twenty, they have few distinctly Indian 
numbers above six. 

Ten.— rSkr. ^, Prakrit ^, whence H. ^, B. and 0. ;^ 
in literature, thanks to the Pandits ; but people when they talk 
or write to one another use ^ ; G. ^, P. "^ and ^, M. ^^, 
which is strange, as M. does not usually change ^ to ^. S., as 
might be expected, has ^,' Kashmiri ^, Gipsy desk. This 
numeral presents no special features. 

There is a regular ascending scale in the numerals. The 
cardinals up to ten are isimple derivatives from Sanskrit ; from 
ten to nineteen they are compound derivatives from the same, 
but inherit the system of composition handed down from the 
earlier language. From nineteen onwards they form their 
numerals in a way of their own, which agrees at the same time 
in principle with that of Sanskrit, but differs from it so far 
that the ihaterials of which the compounds are composed are of 
modern origin in many respects. 

Digitized by VjOOQ iC 



It will be more convenient to exhibit each group in a tabular 
form. Here follows the group of " tens" : — 


I f 




F F 


w w w w w 



f III ^ I I 

rill rif ^ 

^ ifc a^ (K a^ Sc & So ^ 

I I % I ^ % %% 

1 1 If 11 




These all follow Vararuchi's well-known 
which the ^ of ^^ is changed to '^j and ii. 

rule ii. 14, by 
44, by which ^ 

Digitized by VjOOQ IC 


is changed to f . This type "^ is throughout retained by H.> 
which even in "sixteen** has dialectically ^BftT^ as well as the 
more classical iH^f ; the final ^, however, not being a favourite 
or common ending in any of the languages, is mostly dropped, 
and its place supplied by long d, produced by combination of 
the two short vowels ; thus ^TR? becomes WTT^ ^7 olision 
of f , and then M. HTM- The final long o, which is exhibited 
above in B., is quite inorganic, and arises from the habit Ben- 
galis have of pronouncing ^ as d ; this short o is lengthened in 
counting, and we thus find Wf^, but it would be quite as 
correct, or rather more so, to write 'WIX* pronouncing bdr6, 
P. and S. add anim^ika throughout, and S. retains the Y, but it 
can hardly be said that the nasalization of the final syllable has 
any organic foundation : it is, I believe, merely another instance 
of the nasal way these people speak. One often hears a distinctly 
nasal twang given by natives to words where no nasal is written ; 
and until some bettier and more rational standard of spelling 
shall be introduced, it is merely fighting with the air to reason 
about the origin of these anun&sikas, many of which are nothing 
more than local accent, and have no foundation in etymology. 

As to ^'eleven," it is noticeable that the H, which was 
dropped in scenic Prakrit, still holds its place, though some- 
times softened to ^, or even ^, in the modems; the ^ or ^ 
which appears in H. ITHCf > P. and G., is probably by inver* 
sion from Jll^, and the prosthetic ^ of G. is inorganic. In 
"twelve*' is seen the rejection of the initial ^ already noticed. 
"Thirteen" seems to have been wrongly worked out by Cowell ; 
following Vararuchi (p. 121, note), he would make trayodasa 
first into treyodasa, which is not, I think, what Vararuchi 
means, as in i. 5 he couples teraha with sunderam = saundaryam^ 
and similar words. He probably supposes a transposed form 
taryodaia, which by his rule becomes terqdaka^ by the palatal 
vowel leaping over into the preceding syllable (Vol. I. p. 136) : 
the steps are trayOf taryo, tario, tdiro, tero. Then the syn- 

Digitized by VjOOQ iC 


chronoiis change of da&a to raha produces teroraha^ easily con- 
tracted to teraha in Pr.^ because o is habitually short, and 
terdraha, terWaha^ is a natural contraction. 

In the case of "fourteen/* Prakrit does not change ^ to "^j 
probably because the "^ of '^J^ had been assimilated to the 
following ^, making 1^ ; this is followed by all the languages 
except S., which sticks to its favourite ^. 

More irregular than the foregoing is "fifteen." Here the 
anusw&ra operates, and by Var. iii. 44, the nexus ^ is changed 
to IJ (it would in Pali have been '^). This double n has become 
anuswftra, and ^ inserted, or in some cases ^, owing to the 
.influence perhaps of the following T. Bengali has preserved 
the Prakrit type, merely lengthening the vowels by its broad 
rustic pronunciation. Here also the'V is an eccentricity for 
d=^, and one might accurately write niN^. 

Skr. has in " sixteen*' changed the ^ of 1^ to ^, under the 
influence of the final 7 of ^, and this cerebral soimd is ex- 
pressed in the modems by gg in those languages which possess 
that sound, and by '^ or if in those which do not. I am not 
aware of any authority for a Pr. form ^Qft^> and give it as 
mere conjecture, following Trumpp; though perhaps ^ftTf 
would be more correct for ^= ip^ahd '^ss ^. 

Of the remaiiung two there is only this to remark, that P. 
lengthens the second syllable of "seventeen" in imitation of 
" eighteen," while M. reverses the process, shortening " eighteen " 
in imitation of " seventeen." 

The form for "nineteen" means "twenty less one," and was 
probably originally ipJit^rf^hTf^> from which we get the G. 
^^Un^» ^y inversion for ^Rtljt^, and Old-H. mi{«f)ffl and 
ipft^; M. preserves an accurate form in l^^lj^fl. The same 
system prevails in " twenty-nine,** " thirty-nine,** and the rest. 
Prakrit has l[i(1ir^ and VStjj^^ (Weber, Bhag. p. 426). 

The series "twenty-two," "thirty-two," "forty-two," etc., has 
been exhibited at Vol. I. p. 331, and the series "twenty-seven/* 

Digitized by VjOOQ IC 



"thirty-seven," and the rest, at p. 289, where also is shown the 
series "seventy-one," "seventy-two," etc, so that it is unnecessary 
to give these in detail again ; the forms of "seventy" are included 
in the following table for completeness' sake, and because there 
are some inaccuracies in the forms given in Yol. I. p. 288 : 


a. ai ai ^ '^ M ^ % 


t f f t ^ ^ % I 






€ t g I? f I E 


I I 

%. I 




I 'I 




$ i js s ^ ^ I 



^ Weber, Bhagatali, p. 426. 

Digitized by LjOOQ IC 


Oriya does not use any derivative of Ot^Hf) tut instead 
^rf^, i.e. "score/* H. also uses tftlft for the same. The 
languages are quite regular in ''twenty" and "thirty/' in the 
latter occasionally retaining the nexus tr, but more usually 
dropping the r. In "forty'' the Prakrit of the Bhagavatl has 
changed T to W, and the nexus tt is dropped throughout. 
More irregular is "fifty," in which, as in "fifteen," Vararuchi's 
P,rakrit substitutes ^ for 5; in this it is followed only by 
Marathi ; the other languages refer back to the Sanskrit form, 
softening the ^ into ^ in S. and P. only. " Sixty" is also quite 
regular, as are also "eighty" and "ninety," except that M. 
and Gt. make the initial into e and ai respectively, under the 
influence of the fin the next syllable. 

It is when these numerals come to be compounded that they 
exhibit such wide variations of form, as to render it necessary 
for the student to learn each number up to "one hundred" 
separately, and even when the phonetic and structural mechan- 
ism is explained, there remains a large residuum of eccentricities 
and peculiarities. Following the example of Skr., the modern 
languages do not merely, as do most other families of speech, 
add one number to another to express the numbers intermediate 
to each decade, but compound the two elements : thus H. does 
not say do aur bis, for " two-and-twenty," but makes a single 
word bdia; so do all the other languages. I shall first show 
the changes which the units undergo, and then those of the 
tens, as in every case the lesser numeral precedes, and they say 
" twotwenty," " threetwenty," and the like. 

" Two." In by far the greater number of instances the form 
is ^, as in ^|<^f quoted above, and in the examples given in 
I. 331. In "thirty-two," however, the long d is shortened to a 
before IFf, making iPfftiT. In " forty- two,*' the ^ of the 
T|i4J^9 is elided in Pr., when it becomes the second member 
of a compound, so that we get '<eir^|^4 in Bhag., the ^ being 
inserted to fill the hiatus. This H in the modems either 

Digitized by VjOOQ iC 


softens to H, as in H. %W^9 "forty-two/' or is added to the 
letter preceding, as 0. ^QTSV^. P. here has Wfi^t ^ imitation 
of "thirty-two ;^^ G. is irregular in ^f^7)<^ "seventy-two/* 
and B. and 0. insert an inorganic "^ in f^TW^ " eighty-two/* 
Pi^H^ " ninety-two.'^ With the exception of these few irre- 
gularities, the general type iff* is regularly kept throughout. 

" Three.'' The regular type is ft; shortened in some places 
to f?f, and occasionally with an euphonic "^ added, making 
t^: thus— 

H. %^ " twenty-three," 7^^ or ^^fV^ " thirty-three," %'4||^^, 
also%c!« and^o "forty-three," Of^lfJ I "fifty-three," f7|?;i|^, ?!• 
" sixty-three," finpRT " seventy-three," O f i^liH " eighty-three," fTRpi^ 
** ninety-three." 

P. %i: "twenty-three," ^7^ " tlurty-threc," ?t m^ > ftT* and jf^^ 
« forty-three," f?f^^, fift* " fi%-three," ?}f3 " sixty-three," finpRT 
" seventy-three," ffli^liH "eighty-three," f?f^nr% " ninety-three." 

S. ^ftf "twenty-three," "^ftf "thirty-three," ^TTT^iVf "forty- 
three," t^NTf "fifty-three," ^ff^ "sixty-three," ^f ^ f< "seventy- 
three," fj^Rrtft "eighty-three," f^pRil^ "ninety-three." 

G. B^ " twenty-three," 7>aO^ " thirty-three," Thrrtf^T, ^•'^^ 
"forty-three," ^[^im "fifty-three," ^^07 "sixty-three," g^g^Tii. 
" seventy-three," B^Rf^, wrtfV "eighty-three," ^W " ninety-three." 

M. ^^t^ " twenty-three," Jt^t^ " thuly-three," ^^IdS^fi "forty- 
three," %qs "fifty-three," ^t^re "sixty-three," %nnC "seventy-three," 
-aft|ii|4^ " eighty-three," WTfl^ " nmety-three." 

O. ?H3I "twenty-three," ?H^ " thh-ty-three," B^Tnfhl "forty- 
three," 7h|if " fifty-three," ^Imfz " sixty-three," "il^tX " seventy-three," 
?|irnft " eighty-three," f^PTT^ "ninety-three." 

B. ?rt3J "twenty-three," IHtll "thirty-three," <^fl|tf\n "forty- 
three," ntlMli| "fifty-three," iNfj- "sixty-three," ^fT^T^ "seventy- 
three," fH<|^ " eighty-three," ffl<|^g|^ " nbety-three." 

Digitized by LjOOQ IC 


"Four/^ takes the compound from ^ throughout, with 
variant spellings ^, wn, and ^^ ; the numbers may be made 
up after the example of the threes, an euphonic "^ bdng inserted 
before a vowel, like f^fX* " Five'' is in composition xl^, inffj^j 
and P. and S. l|^, M. has ire in i;i^t9 " thirty-five," and G. 
ftir in filK T IdblH " forty-five/' " Six'' is always ^, ^f^, orlf , 
except in M., which is imlike the rest, changing ^ into ^ and 
9 in its usual fashion : thus while " thirty-six" is ?f^ffrir, and 
"fi%-six" igpjcn, it uses ^ff^ "twenty-six," ^I'Tnifhr "forty- 
six," mW^ "sixty-six," IHf%|< , ^Jf^fTT "seventy-six," IfNl^ 
" eighty-six," and "Tfmp^ " ninety-six." ^' Seven" and " eight" 
exhibit no irregularities, and "nine" is not used as a prefix, 
the periphrasis with ^m " less than" the next higher number 
being used in its stead. H. however has iJRrrtft " eighty-nine,'^ 
and OpTR^ "ninety-nine," in which last it agrees with P. 
1I^W% and the other dialects. S. has two singular forms, 
il\irni% and ^|VTf%> apparently from l|^ 4- ^'Sff^ + 1% "nine 
upon ninety," or some such expression ; in one case the ^, in 
the other the ^ of '9(^ has suffered elision. Chand uses this 
method of compounding by means of I9f%| in the lines — 

^ ^hni w'l ^w ^nr • 

" That (was) the year nine hundred (and) ovei^ 
Years thirty and six before." — ^i. 221. 

meaning to indicate Samvat 936, a.d. 870. The reduplication of 
the ^ is metri gratid. 

With regard to the latter part of the compound, ift^ be- 
comes ^^ in H., "I^J in 0. and B., |^ in P., but ?ft^ in Q-. and 
M., '^t% in S. 

'^t^ remains unchanged. x||^^ loses its initial, and some- 
times takes TT) as in H. %?fT^t9* 

"Fifty" undergoes very wide changes. In H. it changes 
from VI^T^ into ^^ in fpjr + tpj "fifty-one," HT^ "fifty-two," 

Digitized by VjOOQ iC 


ift^ "fifty.four/^ ^m^^ "fi%.8eveV' HOTm '^fifty-eight ;'' 
and into Xfm in fd^M^ " fifty- three/^ iPflif " fifty-six/* il^m 
"fifty-three;*' both these are probably the first half of an 
older form i|^n9> the anusw&ra of which has been hardened to 
^, the initial tj is softened to '^T after vowels, unless it be 
thought preferable to see in these forms a relic of the Pr. 
il^STOT- In P. 4^VTf changes to ^^, except in ^ft'fn " fifty- 
four '^ and fipi)^ "fifty-six.'' The same ocburs in Sindhi, 
whose liHT^ is regularly changed to #9|Tf , as ipphlTf " fifty- 
one." G. follows H., as do also 0. and B., with very trifling 
changes; for "fifty-five," however, they use Q-. Xl^nnf, 0. 
M^'1^^9 B. tfrrS, in opposition to H. TT^nnV* 

With "sixty" S. and P. have recourse to their characteristic 
change of ^ to ^ throughout ; thus 

P- T'lTfZf S. li;^Sff3 "sixty-one/' 

„ mfZf » ^Tff^ " sixty-two," etc. etc. 

This course is followed by all the languages except 0. in the 
analogous case of i|Tf<, but only by P. and S. in the case of 
"sixty." (For the series "seventy-one," "seventy-two," etc., 
see Vol. I. p. 288.) 0. combines the ^ of ^M^ with the fol- 
lowing IFf , one of which it rejects, and inserts a labial vowel, 
tjius producing ^^f^, as ii^ljR, 'CT^fTj ®^- 

"Eighty" presents no noteworthy features; but in M. W^ 
"ninety" rejects the final ^, whicl^ is perhaps a relic of the 
Jl of ^^Ul in composition, and doubles the ^y at the.same time 
restoring the cerebral type of Prakrit, which was inadmissible 
m the single word, because the ^ was there initial ; thus it gets 
^pf, as in Tlj^rnq^ "ninety-one." G. follows an analogous 
process, changing ^ or %^ into m preceded by anun&sika, as in 
^ftrtW " ninety-four," Mxljiy^ " ninety-five," ^fte " ninety-six." 
The other languages are regular. 

In the junction of the two elements of these compounds it ia 
to be observed that a long & is often inserted, but I am unable to 

Digitized by LjOOQ IC 


lay down any rule for its insertion or omission, unless it be that 
where the initial letter of the second member is weak, d is in- 
serted ; where it is strong, omitted. Compare H. «|||^i| " twenty- 
two" for ^ + ^f^ with 4iX\H " thirty-three'^ for ^ + ift^ ; but 
even this rule hardly carries us through all the series. The 
Prakrit had originated these compound forms long before the 
modem languages came into existence, and the secret of the 
composition must therefore be looked for in that stage of speech ; 
and the data for Prakrit numerals are imfortunately so defec- 
tive, that as yet I have been imable to formulate any rules on 
the subject. . Marathi introduces a joining vowel e in the series 
with "forty," and ^ in the "seventy," "eighty," and "ninety" 
series ; but the reason of this is not easy to find. 

Above "one hundred" composition ceases, and the words are 
written separately, except in M. G. and S., which, from " one- 
hundred-and-one" to "one-hundred-and-nine" can combine the 
lesser numeral with the Skr. VJtI?; " above," thus : 

" One-huDdred-and-one," M. li^ifc^ tI <% S. |{cjJY^<^ ;^, G. I^^tK, ^ . 

§ 27. The ordinals of all the languages are formed each on 
its own basis for the first four numerals, but after that follow a 
regular system of genuine Sanskrit origin. 

" First " has the following forms : H. trff^, P. w?., M. id,y 
S. Vi^y ftrff ^, ftffff , G. ^f^. 0. and B. generaUy use 
the Skr. Tf^m, but the H. xrff^ is gaining ground, and is 
heard among the lower orders constantly. If we derive these 
words from TRPR, we are met by the difficulty of admitting the 
change of i? into W or x;, tod I therefore suppose that they 
come from an imrecorded comparative THT^j which would stand 
in the same relation to the superlative TRPR as Latin prior 
does to primus, Bopp (§ 293) has already established the fact 
that ira^ is the superlative of H, and similarly (§ 321) he shows 
that all the cognate languages use the superlative form, as 
primus, irpSyro^ for Trporaro^, and eristo =^ erst, from er^=eher. 

Digitized by VjOOQ IC 


So also our own "first," in Old-English ^rs^, is the superlative 
of " fore." There must also, one would imagine, have been a 
comparative, and this could be nothing else than h^\, 

"Second'' is H. ^^^TTT, P. J^, jyn, S. f^^, fNt, G. 
«n4t, M.. ^^<|, 0. ^[^, but 0. and B. often use f|[<f^. It 
is probable that the rude uncivilized Oriyas and Bengalis did 
not possess the idea of an ordinal till late times, when they 
naturally had recourse to Sanskrit to supply the want. Old-H. 
has also <f[3rr, which, together with the Panjabi form, are from 
Skr. f^<n*l, Pr. ^f;i(, also ^[t^, but the modem ^^ refers 
to the form ^f^^, formed by hardening the ^ into Wl(^ and 
then dropping one ^ (Vol. I. p. 249). The S. and G. come from 
another Pr. form f^ff?nf, reduced by dropping the ^; this is 
further shortened into ^ft^. For the H. and M. ^« and ^'^T^, 
see under the next paragraph. 

"Third." H. 1!^^, P. cft^, cfNT, S. f|^, 5^, G. 
HWt, M. firorr, O. and B. "RtoT is in use, but clefts is com- 
mon also. This ordinal is parallel to the last, and is derived 
from the Skr. Hcft^ by the same process as the words for 
" second.'' The form Tft^T with its analogous ^4j4,r seems to 
be a recent compound of the numeral themes ^, eft, with some 
word meaning progression, and probably connected with the 
root H "to go," but there are no certain facts on which to 
found an opinion. 

"Fourth" is regularly derived from the Skr. x(fj\$, see 
Vol. I. p. 144. 

From "fifth" onwards the ordinals are regularly formed 
by adding H. ^, Old-H. ^, P, ^, S. ^, G. ^, M. ^. 
B. and 0.^ having no ordinals of their own, use the Sanskrit 
terms when necessary. All these terminations come regularly 
from the Skr. Tfif, the termination of the ordinals.^ Hindi uses 
for "sixth" W^9 the regular Prakrit form of Skr. W, and 
with this agrees G. lafztj hut all the rest are regular. 
^ See for farther discuiBsion of this question Chap. III. § 64. 

Digitized by LjOOQ IC 


§ 28. Besides the cardinal and ordinal series, the languages 
have other sets of numerals formed in ways peculiar to them- 
selves. Such are the fractional- numbers, in which all the seven 
languages are rich. 

"A quarter'" is H. XTRT or ITRT, P. id., S. ITR, G.iTnr, M. 
^TRTj 0. ttTHj ^> B- M^^l ("^nn) ; the origin of aU these is Skr. 
irrf^, p. "^TT^ "a quarter/' the compound ^1(41^^= "with a 
quarter," becomes in Pr. ^^rn^V^ and in H. ^B^, P. ^WT? ^TT1C> 
^RmC^, S. ^OTTt» ^' ^Bnn> M. ^RTT, 0. ^^3^m[> B. ^r^^ («a«M). 
Thus H. ^?n ift^ " three and a quarter," and so in all the rest. 
On the other hand, the deduction of a quarter is expressed by 
derivatives of Skr. Ml^t^, i*e. IIT^ + ^3JT "a quarter less." 
H. iftif, P. T|^^, VP^, S. xftljl', G. iftlirr, M. in^nr (nearer 
to the Skr. than any), 0. xftif, B. id. Thus xftif H^T " two and 
three quarters," or " a quarter less than three." In H. P. O. 
and B. we appear to have the locative form, in the others a 

Two other numbers also are peculiar to this group : " one and 
a half," H. %^ from Skr. Jf^, concerning which see.VoL I. 
p. 238, and "two and a half," H. "^[^y P. ^, S. "^^^ G. 
^rot, M. ^viH, 0. ^7T^> B. ^in^Ti;^ the origin of which 
seems to be Skr. "ll^ + ipf, to which M. adds the conjunction ^. 

For " three and a half," " four and a half/* and the rest, the 
languages add to the complete numeral the word H. P. ^fl^» 
S. ^J, G. m^y M. ^btI, 0. ^, B. ^1, from Skr. ^ + ^ 
= "with a half;" thus "three and a half" is H. ^ri% TTN, 
"four and a half" ^51% 'qTT, and so on. 

• The other species of numerals, such as "once," "twice," and 
the rest, will more appropriately be described under the head of 

'Digitized by LjOOQ IC 



CONTENTS.—} 29. Natueal and Grammatical Gender.—} 30. Use op Gender 
IN THE Seven Languages. — } 31. Typical Terminations of the Adjec- 
tive. — } 32. Terminations of the Masctjlinb Gender. — } 33. Termina- 
tions OF THE Feminine. — } 34. Terminations of the Neuter. — } 36. 
Formation of Feminines from Masculines. — } 36. Gender of Words 
ending in Consonants. — } 37. Decay of Gender— its Cause in Bengali 
and Oriya. 

§ 29. Q-ENDER is of two kinds, natural and grammatical. 
Natural gender is that which refers to living beings, and is 
threefold : there being one form for males, a second for females, 
and a third for mankind or animals regarded merely as such 
without reference to sex. The human mind has, however, not 
rested content with this simple and natural use of gender, but 
has, by an effort of imagination, extended the distinctions of sex 
to inanimate objects, abstract ideas, and, in short, to all nouns 
of every kind. All languages are not alike in this respect: 
some retain all three genders, others only two, and some have 
had the good fortime to emancipate themselves entirely in the 
noun, and to a great extent in the pronoim, from these awkward 
alid cumbersome swaddling-clothes of speech. This is happily 
the case with our own beautiful and practical language, and is 
an advantage for which we ought to be deeply thankful to our 
Norman ancestors, whose keen common sense led them to reject 
much that was useless and unwieldy in the speech of our English 

The older languages of the Indo-European family have all 
vol. n. 10 

Digitized by VjOOQ iC 

146 GENDER. 

three genders; those of the Semitic family got on very well 
from the beginning with only two, having never thought of 
developing the neuter. In the middle and modern Indo- 
Germanic languages the German still retains all three genders, 
while most of the others have only two, and some none at alL 

ITatural gender exists in all languages, grammatical gender 
only in a certain number. In those which do not observe gram- 
matical gender, natural gender is frequently distinguished by 
having two separate words for the two sexes; in those which 
acknowledge both kinds of gender, the distinction is afiPected by 
a variation in the form of the word, most frequently in the termi- 
nating syllable or vowel. In consequence of this habit, it has 
>come to pass that the form of the word has created the gender ; 
thus, in deciding what gendet should be ascribed to a word 
which indicates an object from its nature incapable of classifica- 
tion by sex, the form of the word is the only guide ; and if it 
happen to be of a form similar to that which is appropriated to 
the male sex in living beings, it will be classed as masculine ; 
if to the female sex, as feminine ; and if to neither, as neuter. 
Both classes of gender exist in the languages we are con- 
sidering, but in a very different degree. Gender is in all of 
them indicated to a great extent by variations of form, especially 
of termination; but it must be observed that whereas substan- 
tives have, as has been shown in the last chapter, a very great 
range of terminations, adjectives and .the participial forms of 
verbs have very few. It is principally by observing which form 
of an adjective or participle is used with any given substantive, 
that we can tell what its gender is. For instance, H. sarak, ''a 
road,'* is the name of a thing in itself incapable of natural 
gender ; it is only by noting such phrases as bari sarak, " a big 
road," sarak bantiy " the road is being made," that we discover 
it to be feminine. Hence it follows that in those languages 
which use very few participial forms, or whose adjectives have 
no distinctive forms for gender, we are unable to trace the 

Digitized by VjOOQ IC 

GENDER. 147 

gender of substantives very clearly ; such languages have only 
natural, not grammatical gender, and even the natural gender 
only in a limited number of instances finds expression in the 
form of the word used. 

§ 30. Sanskrit has all three genders, so have the Prakrits/ 
I^ the modem languages, only Gujarati and Marathi have all 
three. Sindhi, Panjabi, and Hindi have only masculine and 
feminine ; Bengali and Oriya have no gender at all, except in 
the pure Sanskrit Tatsamas, now so largely introduced, which 
retain the form of the Sanskrit gender, but even this only in 
the higher style. In B. they say iftz ^mnn " a little boy,'' 
WtZ MTf^RTT " a little girl,'* and iftz ^ ^^ a little thing.'' In 
the highflown literary style they would write 'I4«^< flC ^^ a hand- 
some man," ^«^'<^ '^ "a beautiful woman ;" but in the common 
style, which is, after all, the true language, one would hear 
^'^'K 5^ for ^* a beautiful woman," without regard to gender. 
The same holds good in Oriya, and in both, even in the high 
style, there is no distinction between masculine and neuter. In 
this Chapter, therefore, we may dispense with these two lan- 
guages altogether. 

In Hindi, Panjabi, and Sindhi^ the reduction of the three 
genders to two has been efiected by turning both the masculines 
and neuters of Sanskrit into masculines; and as the common 
people in the two first-named countries are very careless about 
the use of the feminine, it may almost be said that grammatical 
gender scarcely exists out of books. The use of gender is still 
further reduced, even in literature, by the fact that a great 
nxmiber of the adjectives in those two languages end in con- 
sonants, and do not possess separate forms for masculine and 
feminine, so that they afibrd no clue to the gender of the sub- 
stantive, which is only perceptible from the numerous participial 
forms of the verb. 

In Sindhi some of the neuter nouns of Sanskrit have become 

Digitized by LjOOQ IC 

r '^ 


feminine^ but the most part are masculine. As beftwre mentioned, 
all nouns in Sindii end in a vowel, and the distinction of glider 
is therefore more clearly recognizable than in the other lan- 
guages wherein consonantal terminations aboimd. There are, 
however, some adjectives, mostly of Arabic or Persian origin 
(though some are pure Sindhi), which, although ending in a 
vowel, do not change either for gender, number, or case ; as^p^ 
^^ plentiful,^' Skr. ;|R^ ; WW " mixed," Skr. ITW ; ^rff^ ^^ easy,'' 
Skr.^lT; which are Sanskritic; and aUJ "whole," My^ "good,'^ 
(^)jts>^ " distressed," which are Arabic and Persian. 

Marathi still retains in full and every-day use the whole three 
genders, and the same is true of Gujarati. Not only have the 
substantives three genders, but the adjectives also, and the 
greater part of the tenses of the verb, being participial in form, 
have also three genders, so that to a foreigner the difference in 
this respect between theses two stilted languages and the simple, 
easy, but in no whit less expressive Bengali, is at every turn 

In all the five languages which have gender expressed, the 
masculine is used to denote large, strong, heavy and coarse 
objects ; the feminine small, weak, light and fine ones ; and the 
neuter, where it exists, represents dull, inert, and often con- 
temptible things. So far is this carried, that in cases where the 
original word was only masculine or neuter, a feminine form 
has been invented to express a smaller or finer article of the 
same kind ; and, conversely, where only a feminine form existed, 
a masculine to express a larger or coarser object has been struck 
out. Instances are : 

Skr. ^rn^ "pot" (n.), H. j|(4| "a large cooking pot" (m.), ^flfV 
'< a small cauldron," or "earthen pot" (/.). 

Skr. ipi and iff^ "haU" (m.), H. ^ft^ "cannon-ball" (i».), ^ft^ ' 
"bullet," "pill" (/). 

Skr. •^:f^ "rope" («i.), H. ^^^^ "cable" (m.), ^tsft "string," 

D\g\t\ze6 by VjOOQ IC 

OENDER. 149 

Here lihe final t of ra^mi having been lengthened^ and I being 
.universally regarded as a feminine termination, a masculine in 
d has been created. This distinction runs through all the lan- 
guages, and is even foimd in B. and 0., which, although careless 
in the matter of gender, still recognize the difTerence between 
the termination in d, as denoting larger objects, and that in i, as 
denoting smaller. This will be more fully treated imder the 
heads of the various terminations. 

In M., with its three gei;ders, the different shades of a 
common meaning, as indicated by gender, are illustrated by 
the following examples : 

f^f^TT (m.), "a large piece tom< from a plantain leaf^ used to 

wrap up goods in." 
f^^T^ (/.), " a small piece of the above.** 
fxl^^lQI (tw.), the same as f^TZT- 
ri(4,Un (/•)» t^e same as fij^/^ , also "a small chisel.'* 
f^^lQ (».)> " a small chisel," " the act of tearing." 
J|14I (^O' *' ^ large cart, or waggon, for conveying loads." 
•^rnft if')* " a small cart or carriage for conveying persons." 
irr% ('S')' "^ clumsy or ricketty old car or dray.'i 
;g^ (»!.),« a cable." 
^t<t(/)," a thread." 

^trr ("»•)» "a string," " thread," "streamlet," "track." 
\jfin (»»•)> " a stone." 
Vf^ (y*)» "a large mass of rock." (Here the fem« eipresses the 

larger object, which is unusual.) 

In the case of animals, the masculine and feminine indicate 
the respective sexes, and the neuter either the* young or the 
whole species collectively; as 

ysff^ (ot.), "horse." 

Digitized by LjOOQ IC 

150 GENDER. 

IJt^ (n.), '' tiie horse/' i.e. the genus Equui genenUj. 

Iflf^ (wi.), « a hc-goat.*' 


^I|^ and iPirJI^ (**-)> '' ^ kidy** also the goat species in general. 

The same practice exists in Ghijarati ; as 
1^t^(m.), "horse.'* 
^Efr* (/.)," mare." 
M14 (n»), '' a wretched screw of a horse," also horses in general^ 

or collectively. 
imft (»».), " a buffalo-buU." 
UlNft (/), "a bnffalo-cow," 
VIT^ («.), " a buflfalo-calf." 

These peculiarities of gender and its terminations will come 
out clearly as we work through this Chapter, and need not be 
further illustrated here. 

§ 31. The adjective in all the languages exhibits the general 
type of the terminations for gender, and is so fixed and regular 
as to afford a standard for each language. The general type of 
the adjective is as follows : 













In and Mn 


Panjabi .... 




' idn. 







Gnjarati . . . 





Marathi. .. . 






It is not of course meant that these are the only terminations 
of the adjective or participle, but that among the numerous 

> Narmadds'ankar's example is: <<When the Qaekwur came to Bombay, he 
brought a great number of honses dgihai^un ghorun) with him." 

Digitized by VjOOQ IC 

GENDER. 151 

terminations that are found these are the chief, and, so to speak, 
radical ones; the others jnay be regarded as exceptions and 
deviations, to be explained in their place. 

In the participial forms which so largely enter into the con- 
struction of all verbs, these teniiinations are unvaryingly used 
for the respective genders, except in some instances in Marathi, 
in which the incorporation of the substantive verb into the 
participle has, by the operation of euphonic laws, wrought a 
change in the voweL 

Taking then the terminations of the adjective as the central 
type, it will now be necessary to go through the range of sub- 
stantive terminations in order, so as to get at a clear comparison 
of them all, and to ascertain which are regularly derived from 
the central type, and which are exceptions, to be accounted for 
in other ways. 

§ 32 (I). The masculine termination 4 in H. P. M., and oc- 
casionally B. and 0., corresponds to the termination o in S. and 
G. in a large majority of instances : as 

H. irerr " chUa," and so in aU ; bat G. i(^, S. ^^. 

This termination is in S. masculine, without any exceptions ; 
in G., however, some words in o are feminine, in which the o is 
not the pratyaya, or formative syllable derived from the Sanskrit 
-a^, Pr. -0, as in the above-mentioned class, but arises froni 
some phonetic corruption of the root-syllable. Such are the 
feminine nouns. 

G. TOt " leec V Skr. ^T^fNTT (/•)• 

G. lEfl" "mortar," „ ^t^ (m.). 

G.^" lizard," „ ^fr^(/.)(^foriftf). 

The number of these words is not large, nor are they in them- 
selves important. 

The termination in d exists throughout the whole seven 
languages, including G. and S., in certain words : these are — 

Digitized by VjOOQ IC 

152 GENDER. 

(a). Words derived from Skr. nouns in an, like m^i,, the 
nominative of which ends in Skr. in d. Such are TTTT " king" 
(m.), in all; ^IcHi "soul" (w.), in alL Nouns neuter in Skr. 
of this form end in the short vowel, as.TRP^ "name," H. TTf? 
m. in all, except M. and Gr., where it is neuter. «if^*fi " great- 
ness," however, is used as a fem. in H. P. and S., and as both 
m. and/, in M. G. has preserved the real gender of the Skr., 
in which language 41 Hl^^^ is masc. 

(^). Words derived from Skr. noims in TJ, making their 
nominative in ?n. Of this form are a great number of Tatsamas 
denoting the agent, as W^J, ^T^TT? etc. 

(7). A few words of foreign origin, running through all the 
languages, and chiefly relating to males, so that they are mas- 
culine by natural gender. These are mostly Arabic, Persian, 
or Turkish, but have been in use in the Indian languages from 
very early times, and have even undergone some corruptions, as 
will be seen in the following list : 

H. ^TWr "master," « sir," IIS, P. WT^, WMT, S. ^TWT, O. id, 
H. mm "child," bb . P. id., S. ifrft, O. id., M. ^^. 
H. ^prr "sage," Ub, P. id., S.^pn^, 41^1^ , ^fVmW, O. ^TTR. 
<i«lV. M. ^PIT- 

H. f^r^m "river," b^J, P. ^H,^!^, S. < r< ^ |^ , ig[Pc^Riy, G. 

H. "^m^l "nobles" (pL, also singular), *l^'\ and ^4|<i^ , P. ^4^4,1^ 

singular, S. ^3iRJ^j *¥ sing., G. ^iPCR, •ft sing., M.^?nj^ sing. 

H. ^?[T "God/' 1 j^, P. id., S. id. and ^^T^, G: ^^, ?iY^, 

H. ^TfrtT "overseer," iXi^^b, P. ^O^U , S. ^ff^, G. id., M. 

The word ^MJit, though really a plural, is constantly used as 
a singular in modern speech. It will have been noticed that 

Digitized by VjOOQ IC 

OENDEB. 153 

there ia a tendency to add a t« or t? on to many of these words, 
and so to account, as it were, for their being masculine, by 
giving them the old masculine termination. G-. also effects the 
same by boldly changing the final d into o. 

(2). The termination i is the typical termination of the femi- 
nine in Tadbhayas and Desajas ; that is to say, in noims of the 
mediaeyal and modem kinds, in which, as we have frequently 
before noticed, a principle of development prevails of a character 
essentially different from that which distinguishes the classical 
Sanskrit or Prakrits. In opposition, however, to this generally 
feminine character of i, a number of nouns are found in all 
the languages, which, though ending in i, are masculine. This 
class seems to have given some trouble to our grammar-makers, 
and their difficulty has been increased by the laxity of practice 
in some of the languages. Especially is this the case in Hindi, 
where, owing to the leading authors having been men of strong 
poetic instinct, but of little learning, a tendency has grown up 
to give to each word the gender indicated by its typical termina- 
tion ; and words which, from their origin, ought to be masculine, 
are sometiines found coupled with feminine adjectives or par- 
ticiples. This source of error is, however, confined to words 
which have only grammatical gender ; those which are mascu- 
line by natural gender, as describing male beings or their occu- 
pations, are exempted by their nature from the possibility of 
being mistaken for feminines. 

Masculines in I may be divided into the following classes : 

(a). Those derived from Sanskrit agents in l^^^, which in 
that language form their nominative in f^ I ; as the modem 
languages ignore the grammatical fiction of a base, as distinct 
from the actual nominative, it is from the latter only that they 
take their forms. Instances of this class are the following (see 
also the list in § 18) : 

Skr. f^ ''elephant," base ft^(m.), H. G. S. fT^, P. J^, M. 
f^, B. O. fl^. 

Digitized by LjOOQ IC 

154 OBNDBR. 

8kr. ^rnr^ "lord," base |j| 1(4(1^ (m.). In all ^J^, except 
though in the commoner word jj^^l^f), which becomes in the modem 
languages ift^llj, M. has ^h^TPf^- 

Skr. TUWlt "gardener," base inftP^ (»»•)• ^^ ^^ frtft* but G. and 
M. M\<£\. 

Skr. inft "bird," base Vtf%^, H. ^, M. G. P, id., S. H^, B. 
l|TQt» An<i ^ £ast-B. TH^* 

Skr. ^iX^ " witness," base irrf^^> H. iTRi^* ^n^ so in all^ 

There are, moreover, in tliis class many words of uncertain' 
origin, as well as others whose origin is not at all uncertain, 
though, owing to phonetic changes, it is not so apparent at first 
sight. In some of these words we may attribute the form to a 
feeling that long I as a masculine was appropriate to names of 
trades or professions, so that it was added to words for which 
there existed no prototype in Sanskrit. Such are — 

H. Vt^ "washerman," and so in all. B. and O. have Vt^ <^d 

H. M^^lO ''village accountant," so in all, but rare in M. 
H. inrNft "neighbour," in all, but more correctly with ^. 
H. M\^ " shopkeeper," in all, but S. ift^« 
H. Xdft "a non-resident cultivator," so in P. B. O. 

The exact form from which ^IJt^ is derived is imcertain. It 
should be Skr. Vlf^PO ^<>^ VF^ "to wash,^' and the vulgar 
language probably used the word in this sense, though in 
classical Sanskrit the word seems to be used, exclusively in the 
other sense of the root, namely^ as ^^a runner.*' xn^ is from 
irNft, and that again from ^f^'^, from tf^ in the sense of a 
" side ;" a pdhi cultivator being one who lives, not on the spot, 
but in another village, and is thus, as we should put it in collo- 
quial English, an ^^outsider.'* q^^lO ^ a somewhat abnormal 
compound of the mediseval period ; the first part JfZ is f ^rom iff 

Digitized by 


GENDER. 155 

"a letter*' or "writing," and the second seems to postulate a 
form ^rf^ ^^doer/* perhaps from the root h.* XT^t^? more 
accircately iJ^fWftj is TrfTRTt^P^. H\^ has been by some de- 
rived from the Skr. ift^^ ^* sweetmeat ;** but I believe it really 
comes from the Arabic J^J^ "provisions/* "stores;" the ordi- 
nary Indian modi is not a sweetmeat-seller, but deals in grain, 
and eatables of all sorts. 

In a subsequent section (§ 35) it will be shown that the 
majority of these words testify to the existence of a Sanskrit 
original in X^ ^J forming their feminines in Jj^^ or in forms 
derived there&om. 

(^). Noims derived from Sanskrit substantives in ^, with 
the sense of a male agent. Here the gender is natural, not 
merely grammatical Common instances are : 

Skr. ^J^ « brother," H. V^^j so in all, but P. also Ifp^, S. HT^> 

Skr. IW "grandson," H. irnft> «> ^^ B. and C, but M. ITHf . 

The number of words of this class is not large, nor do they 
run through aU the languages; in fact, as has already been 
mentioned (Ch. I. § 14), the tendency of ^ is rather towards 6 
in this class. 

(y). Words derived from Sanskrit masculines and neuters in 
1[, by lengthening the final vowel ; as 

Skr. ^^ "curds" («.), H. ^ (m.), P. ^ff (m.), M. and G. id. 

Skr. xrfTf "lord" (m.), H. I^ft («».), but oftener xm- 

Skr. ^Ol^Mffl " sister's husband" (m.), H. ^fflft^ (m.), P. H<«lll^> 


Skr. If^Sfii " carpenter" (iw.), H. If^, O- B. M. id., but G. S. ^xfts 
whieh is an independent formation, from the modern verb " to cut." 

1 The origin of the termination vdrt will be discussed under the Verb, to which it 
properly belongs. 

Digitized by VjOOQ IC 

156 GENDER. 

(£). Deriyatiyes from Sanskrit masculines and neuters in 1[9, 
X^ and ^. Concerning the former of these, see examples in 
Ch. I. § 18, (^), (7). Of the two latter examples are : 

Skr. tfTlft^ ''^ water" (n.)» H.irnft (''i-X and so in all,^ bat generally 
inift* Ii^ ^* and M. it is n. # 

Skr. flTEJ "beloved" (»».)> H. ift («».), "lover," "husband." 
Skr. ^^f^ "a class of Brahmans," H. ^^FWt.* 
Skr. ^f^ " a Kshattriya," H. ^nft, W^, IKt, P. id^ S- ftT^V, 

Under this head probably comes the list of obscurely deriyed 
words in Sindhi given by Trumpp at p. &6, though, with some 
three or four exceptions, the origin of these words is not at 
present traceable. 

(e). Words deriyed from Sanskrit masculine and neuter nouns 
containing 1[ or ^ in the penultimate syllable, but in which, 
the last syllable haying fallen 'away through phonetic changes, 
the l[ or f^ has become final, the former being lengthened 
to t;. Thus : 

Skr. ^ "clarified butter" («.), H. 1^ («*.), M. 1^ («.) but rare, 
P. G. id. («.). 

Skr. TJt^ " life" (m. and n.), H. ^ (m.), and so in all. 

So completely, except in the case of professions and trades, 
has the idea of the feminine character of f^ taken possession of 
the popular mind among the modern Aryans, that, many words, 
which ought to be regarded as masculine, are treated as feminine, 
simply because they now terminate in f*. Thus in H. '9(^wf\ 
"butter** is found with feminine adjectiyes, though it is pro- 
perly masculine, being derived from Skr. ipnftTf (n.), and 
the corresponding word in M., ^t^> is neuter. So, also, H. 

^ The Sotis are one of the highest clans or ffotrat among the Maithil Brahmans. 
The Bajas of Darbbanga belong to this gotra. 

Digitized by VjOOQ IC 

OENDEB. 157 

f^ni^ ^^ sale'' is treated as feminine^ though derived from Skr. 
fi|%€r (w.). In M. several words which are neuter, of this 
termination, are written with anusw&ra as ^, because the 
anusw&ra, being derived from the final ?1[ of Sanskrit, seems to 
the people to be more appropriate to that gender. Thus they 
write Miuff *^ water,'' ^HtSt ^^ pearl," and others. It is by the 
operation of a similar tendency that we in England now treat 
such words as riches, aims, as plurals, though they are really 
singulars corrupted from richesse, almesse ;^ so also our old sin- 
gular pease, from pimm, has been changed into a plural peas, 
and a singular pea formed from it, though the s has nothing to 
do with the plural sign, but is a genuine radical part of the 

(3). The termination ^ ^ is masculine, as in the cases of the 
other long vowels, in words derived from Skr. masculines. This 
rule is a good guide, in spite of a certain number of exceptions, 
in all terminations, namely, that the gender of the Sanskrit 
original is fairly kept in all modem words, and affords a clue 
to the many apparent irregularities; as, for instance, in the 
case of words like ^, tlT^^ etc. Inasmuch, therefore, as the 
termination ^ in Sanskrit is of all three genders, we cannot say 
that it is either regular or irregular for a word in any of the 
modem languages to be masculine, feminine, or neuter. We 
must in each case trace the word back to its origin; and we 
shall, in a large majority of cases, find the modem word retain- 
ing the gender of its parent. In Ghijarati no distinction is 
practically made between long and short u ; and even in Marathi 
some confusion exists. In Marathi, however, and Sindhi, the 
long 6 is generally masculine ; in H. and P. words of this ter- 
mination are about equally divided between the two genders. 
No examples of the masculine nouns of this termination need 
be given. The rules for the formation of the stem contained in 

^ The translators of our Englisli Bible knew better when they wrote '*an alms" 
(Acts iii. 3), though they use ^< riches" as a plural, <* Your riches are corrupted." 

Digitized by LjOOQ IC 

158 GENDER. 

the preceding Chapter will serve to gnide towards the Sanskrit 
word from which the modem word is derived, and the gender 
can thus be easily ascertained. 

(4). The short vowels are generally elided at the end of a word 
in Hindi, as well as in P. and G., and to a very great extent 
also in common Marathi. It will therefore be more convenient 
to reserve the consideration of such words till we come to the 
section on consonantal endings. When the short vowel is not 
elided, it is frequently lengthened, to enable it to retain its 
position, and the words in which it still- remains as a short 
vowel are all Tatsamas, like Tffif , ftt^, ^^IHJ, If^j the gender of 
which is identical with that of the Sanskrit. 

It is only in Sindhi that short vowel terminations are of any 
account, and our discussion of them is confined to this language. 

Short t« ^ is the typical ending of masculines. 

Short a ^ is the typical ending of feminines. 

Short i X is chiefly feminine. 

Masculines in ^ are derived from the Prakrit termination o, 
and represent the class I have called barytones in the last 
Chapter, thus corresponding to the masculine consonantal end- 
ings of other languages. 

Under this head come also nouns derived from Sanskrit 
masculines and neuters in ^, thus — 

Skr. ^a|T^ " wind " (n.), S. ^TR («».). 

„ ^ "liquor" (n.), S. ^.^ 

There are also a few words in which the final u is derived 
from the H of Sanskrit, as quoted in Ch. I. § 14, fi|g= fiT^, etc. 

There appear to be no instances of masculines in a ^, and 
very few in 1[. Of the latter, instances are vn^, from Skr. i|f?f 
" lord f ftr^Rft , from Skr. ^f^[ft[ « the planet Jupiter f ff^ 

1 This is Trnmpp's derivation (Sindhi 6r. p. 33). I am disposed to derire the 
word regularly from Skr. T^, Pr. M^» 

Digitized by VjOOQ IC 

GENDER. 159 

" Vishnu ;'* and as shortenings of f^ from l^^^, as noted above 
in (2, a), lifft from lilltt'l ^^lion;'' ITfTT ^^^ ^Tpt;, from 
fi^m " helper," are familiar examples. 

§ 33. As in the masculine, so also in the feminine termina- 
tions, there exists a double system. The termination 4, which is 
typical of the feminine in Sanskrit, is in the modern languages 
a resultant from o, and typical of the masculine ; yet it is found 
as the termination of many fem. words, just as I, the modem 
fem. type, occurs in many masculine words. 

(1). Final d is feminine in the seven languages in the follow- 
ing classes of nouns. 

(a). In Tatsamas which are feminine in Sanskrit, such as 
"^^ ^^ worship/^ fcjcll ^^thought/' J^^ ^Hongue/' ^|5^ " story," 
and many more very common words. 

(^). In a considerable number of modem Tadbhavas, which, 
though changed from the pure Sanskrit form, still retain the 
distinctive termination, and with it the gender : such as 

Skr. Yim "patience" (/.), H. ftpRT, W^f P. s! f^m* O, ^^T. 

Skr. ^TRT " pilgrimage" (/.), H. WT^, P. ^HTPCJ, S. ^HTZTT, O. id., 
also irnrCT» M. W^, all/. 

Skr. ^>gnr "hunger" (/.), H. ^piJH, P. ^fW^- 

Skr. ^mBT "order" (/), H. ^mWi, P. ^lITfirW, 8. ISf^l, G. 

Skr. fiqr " mnrder" (/), S. f^. 

(7). In some words of imcertain origin, but purely local in 
type, as— • 

H. r^ftei T (/•)> " a small box," but in all the rest m. in d, with a fem. 

H. f^f^^ l (/)» " a bird," in P. and M. without the final syllable and/. 
H. ^ft ^l T (/)» " an old woman." 

Digitized by VjOOQ IC 

160 GENDER. 

In these three very common words the final ^ is probably a 
recent addition, and the words were originally, as they are still 
in the cognate languages, f^pftj f^nft, and gftj the i having 
been shortened, as is usual in Hindi, in consequence of the 
additional syllable. Or we may suppose them to come under 
§ 9 (2), and to have been feminines of the form ^Tf^WTT, where 
the sufi^ ikd has been softened into ii/d,' though it would be 
useless to look for a Sanskrit original for the two first words. 
The third may, however, well come from a Skr. form ^f^RIT. 

This remark leads to a consideration of somewhat extended 
application. It may be asked why d, which is in Sanskrit a 
fem. ending, should in the modems be so typically and univer- 
sally a masc. ending, and similarly why I, which in Skr. is quite 
as much a masc. as a fem. ending, should have in the moderns 
so almost exclusively attached itself to the fem. P The answer 
would appear to be found in the extreme prevalence in Pr. 
of the practice of adding the suffix aka to nouns of all classes. 
We have to begin with the oxjrtone nouns in Skr., which 
become noims in o and d in the modems, and to them we must 
also add the Pr. formatives in aka^iao, of the type ghotalca^=- 
ghorao=ghord ; the union of these two sets of words results in 
a preponderance of mascs. in d-o. Now, as the feminine of 
aka is ikd, and ikd becomes ipd, and more frequently still I, 
if we suppose that to every one of the words to which Pr. tacked 
on aka for the masc., it also tacked on ikd for the fem., the pre- 
ponderance of f a? a fem. ending in the moderns is eixplained, 
the more so that we can add to the i from ikd a large class in 
which already in Sanskrit the masc. in -as forms its fem. in i ; 
such as ^^T» (»*.)> ^•^O (/)• Moreover, the typical ending 
of the adjectives being d-o for the masc, and i for the fem., it 
was natural to use an adjective ending in i, with a substantive of 
a similar termination ; and so it comes to pass that aU grammar- 
writers treat masculines in i and feminines in d as exceptions, 
though few of them attempt to account for their existence. 

Digitized by VjOOQ IC 

GBin)EB. 161 

(2). Final i is feminine then in all oases except those noted in 
the last section, and this holds good for aU the seven languages. 
Final ^ is feminine in words derived from Sanskrit feminines, as 
^^^earth/^ ^^^^ daughter-in-law/' H. ^^, P. T^fj^y S. ^y, etc. 

In Sindhi, as noted above, the feminine terminations are 
a and i, and the majority of these words are identical with the 
feminines in mute a of the other languages, which have been 
short^ed from Skr. feminines in d. 

(3). In some classes of adjectives, mostly very common and 

vulgar, a feminine in o is found. This is very common in 

Panjabi, and frequently with the suffixes ^, IJ^, or if^, where 

the may be regarded as a lengthening of the u of the mascu- 

, line. Such are — 

\^T%^ ''(a cow) that thrusts or butts," masc. >i|^W. 

y^ "whore," from ^^^ "cunnus," also written W^> fl^, and 

^^ "adulteress," from ^RTTin' "to copulate," masc. qra*. 

'rf|rf^' " a woman with projecting teeth," from ^^ " tooth," masc* 

1 '^^ " ft coquette," origin uncertain. 

•rwt " ft** immodest woman " (probably from 4if |^, for ^1^ = ^^f). 

Most of the words of this form are words of abuse used by 
women to one another, the fair sex in India being possessed of 
a remarkably fertile invention in the matter of vituperation. 
In Marathi also ^ is often a feminine ending, as in ^TRrtlt 
" woman," where the o is probably a lengthening of the mascu- 
line ^. In Oriya, when addressing women, they always 

§ 34. The neuter has less variety of termination than the 
other genders, and exists only in Marathi and Gujarati. The 
type of the neuter is M. l(, G. ^^ but M. also has a neuter in ^, 

VOL. u. 11 

Digitized by VjOOQ IC 


which, however, is not frequent, and occurs principally in 
words denoting the young of animals, which have also masc. 
and fem. forms for the two sexes of the animal ; as q^iM (m.) 
"he-goat,'* ^Wn(^ (/) "she-goat," ▼l^ (w.) "a kid of both 

Both the typical terminations point back to the ^ of the 
Skr. neuter, that of G. in ^, by virtue of the process so often 
observable in the case of final ?1[, by which that letter resolves 
itself into its two elements, the labial and nasal, the former 
of which passes through ^ into ^, and the latter into anusw&ra, 
and then into anun&sika or a simple nasal breathing. The 
ij of Marathi would seem to be derived from a weakening of 
the ^ of ^ into X[. 

Though the other languages have no neuter for the noun, 
yet the infinitive, which is a verbal noim, is derived from a 
Skr. neuter, and in most of the languages retains a neuter form. 
Hoemle (Joum. As. Soc. Bengal, vol. xlii. p. 66) combats the 
theory that the neuter 'form observable in Old-Hindi, and in 
several dialects of Modem-Hindi, as well as in Gt. and M., and 
he might have added in Sindhi also, is derived from the Skr, 
neuter in ^. Thus he wiU not take Old-H. W?I^> dialectic 
^^RTlt^ «li"i,*l ti ^^j from Skr. WK^y ^s he says the process 
is opposed to certain glottic laws which he appears to have 
formulated for himself. He would apparently derive the Ma- 
rathi infinitive ^%' from a Prakrit form i|S?;f^^; and the 
other infinitives with a labial type, as iR'^ij^, from a further 
Prakrit form ^<,(!|^. The phonetic changes thus indicated 
are indeed possible, and quite regular; if we once concede 
Prakrit forms ^^Hu^ and <4)<,lir% there is no difEiculty in 
deducing from them modem forms ^?;*^ and ^^iff respectively ; 
but we want more evidence as to the said Prakrit forms, and 
their meaning hardly corresponds with sufficient accuracy. 
But leaving this question of the infinitive for its proper place, 
we may follow Hoernle in his process of applying the principle 

Digitized by VjOOQ IC 

GBMDBR. 163 

to nouns in general. No sucli general ending in |^ or ^^ is 
of course to be found in Prakrit for neuter nouns of all kinds ; 
but the author would have us belieye that such an ending did exist 
in a great many cases, and that its origin is to be found in the 
favourite suffix ^, so that we are to postulate in the case of 
every Marathi neuter in "if a Prakrit form in X^y derived from. 
XjtHy and in the case of every neuter in ^ or ^^ a similar ^9^ 
from ^ : why Marathi should always select 1[<, and Ghijarati 
^3^, is not explained, and seems in fact unexplainable. It is, 
however, highly probable that we have in this theory an 
indication of the direction in which we should search for the 
explanation of such forms if dissatisfied with their derivation 
from the simple Sanskrit neuter; and it must be admitted 
that the author's illustrations from the forms of* the oblique 
cases of stems in ^ in Marathi are strongly confirmatory of 
his theory. 

§ 35. A large majority of feminines in all the seven languages 
are formed from the corresponding masculine noims, though 
there are of course many which have an independent origin ; 
as, for instance, words which possess natural gender, in which 
the female animal has a distinct name from the male one. But 
in names of trades, castes, and the like, the female is generally 
derived from the male. 

Masculines in <) G. S., or d H. and the rest, form their femi- 
nines in J ; as 

H. ^ft^inn " boy," ^t^Fit " girl/' 

S. also shortens final i, as Q^l|(\ „ 

This is an extremely common process, based on the typical 
endings of the two genders, and need not be further illustrated. 
It is extended also to the case of mascuL'nes ending in short u 

Digitized by LjOOQ IC 

164 GENDER. 

in S., and with mnte a, or in other words, with a consonantal 
ending, in the other languages. Instances are : 

S. trf " washermao," 1|f^ " washerwoman." 

8. W^ " he-ass," ^T^rflf " she-ass." 

S. ^fN " firmus," \i^f^ " firma." 

P. ^TR[ " sugar-cane," ^TITT^ id. 

P. f^RT " vetch," ^RT^ id. 

P. il^H " a cowry," cf^vj^ " a small cowry." 

H. ^«^4^ " monkey," 4«^0 " she-monkey." 

H. ^R " forest," ^ifV « copse." 

H. ^t€ " bamboo," ijMt " flute," « reed." 

Although, "however, instances are to be found, such as those 
given above, of a feminine in t^ being derived from conso- 
nantally ending masculines, yet the practice is not very common, 
the feminines in i being more generally derived from masculines 
in 6 and d ; especially is this the case in masculines derived 
from Skr. by means of the common suffix ^«5, which, as has 
been shown, produces H^, and then ^ and ^, while its 
feminine is regularly XjSi[[, yielding X^9 ^^^ *^®^ ^» ^^ ^ 
^tZ^, ^t^^, ^t^ (m.) ^^ horse/' H\Vl^[, ^f%^, ^Tit (/) 
" mare." 

Extremely common, and spread throughout the seven lan- 
guages, is the feminine ending whose varied forms are given 
below, and which is confined to the expression of nationality, 
caste, occupation, bodily and mental qualities, and other attri- 
butes of living, and chiefly human, beings. 

H. has, in the first place, femiaines in ^, ^^ifV or ^, 
derived from Sanskrit jyfty the feminine of masculines in 

Skr. f^ "elephant," H. fT^, Skr fern. ff%^, H. Iflf^if, 
frtMV, or f^ift* 

Digitized by LjOOQ IC 

oendbr; 165 

This form it also uses in feminines derived from a masculine 
in ^ which ^ is thereby shown to represent the Skr. •l['^; as 

ift'fY " washerman," Vtf^^ " washerwoman ; " * 

4||^ ** gfardener," iTTf^lPf and ifrf^I^ " gfardener's wife ; " 
but the termination is by no means confined to those words 
which have, in Skr. the form ^^ (w.), 1[^ (/.), but, like aU 
terminations which have once become typical, is added to words 
derived from all sorts of formations ; thus 

H\^\i, " goldsmith " (Skr. ^e | lifgftK )> ^TTfTf " goldsmith's wife." 
HcTH T^ " leather worker " (Skr. ^^^n^)* ^^lR*l Ws wife. 
^Ffn: " potter" (Skr. f1|Wn;)> ^5PfTfT^ Ws wife, 

irra " tiger" (Skr. c^ra), ^f^, W^^ " tigress." 

4Rt '' a caste of cultivators," ^f^ ^^® female of that 

Through the habitual carelessness of this language in respect 

of unaccented short vowels, this termination is often sounded 

merely aslR. 

lilf^ " a seller of vegetables," ^Hf^if and ^Slrf^if ''a female seller." 

ierTWr"cowherd«(Skr.lfrtTW^). Wm^ and ^frfW^ "cowherd's 

ISH << cheat" (Skr. ^R(), T^Hf and ^fj|l| '' female cheat," or 

lynft " sinner," ^PW* TTftf > MlfM^Y " peccatrix." 

^^^ " washerman," W^^f l^is wife. 

^^^ " carpenter," ^^ his wife. 

In this latter word there has been contraction from «i«i,«i. 
Regular derivatives from the Sanskrit are the words indicative 
of conditions peculiar to females : 

Skr. irfih^ "pregnant," H. infiPT and IITH^* 

So also a feminine in ^^, IHT., or ift^ may be made out of all 
sorts of masculines, to indicate the wife of a man of any class or 

■Digitized by Google 

166 OBNDEB. 

trade, rather than a woman who practises the trade herself. Thus 
one often hears familiarly ^^tT^ " a thief s wife/' from '^t^ " a 
thief.y This does not necessarily imply that the woman is her- 
self a thief, but that her husband is. The same remark applies 
to the following : 

l|f^ « a Pandit," Qr^dllfl Ws wife. 

IJ^ " a teacher," ^Mii* l his wife. 

3T5T " a chief," d l jiMf* ! and ^|f<ll,* l " a chief's wife." 

^^V^'' a headman," ^"^M^f^ and ^MVTI^I "a headman's 


These forms 'Wf'f and ^RTTf ^^e from Skr. "^RPlVj the lattw 
by a common inversion. They are even added to words of 
Persian origin ; as 

jx^ " a sweeper," ^f^i^lOl " a female sweeper." 

JxJ^ '<a Mughal," ^^ifWrf^ ''a she-Mughal." 

^U- " a servant," '^I^Klf* ! "a female servant." 

Even in words ending in ^ we have the feminine ending 
^in[f; as 

Ifffpirt " trader," ^H l ^H,^ " a trader's wife." 

The Muhammadan government conferred the title of 'Khkn 
f^^ on Hindus in some parts of Bengal, and from this the 
common people have formed a scarcely pronounceable feminine. 
IDidn throughout India is pronounced with the final n nasalized, 
as though written ^, and the feminine is therefore ^tNtHCT? 
vnlgo ^9t^rnt or ^rfm^j which the reader may pronounce at 
his leisure. 

Panjabi has also this method of forming the feminine, and 
chiefly neglects the l[ and uses the termination Ut or Hflj, the 
former used after a cerebral, the latter after other letters, a 
practice in which Panjabi reverses the Skr. usage. The appli- 
cation of this feminine ending has grown to be irrespective 

Digitized by VjOOQ IC 

GENDER. 167 

of the terminatioii of the masculine, as in H. This will be 

seen from the following group of examples : — 

VY^ or Vt^ " washerman/' VY^Ilf his wife, 
^flf^orVtfT«acheat/' VtfW <rf. (/•) 

^nr^ " leprous man " (Skr. ^[^), Uf <^ " leprous woman." 
"^S^vft " bad companion " (*/i.), ^^TRUI id. (/.) 
^|<t^ "angry" (Skr. Tlftvt), ^BRKt^Smr id. (/.) 
^i^'d^^r " trickster," ^aTf^db'^ " trickstress." 

The intelligent and progressive caste of Kayasths, which 

is so leading a class in Bengal, is very scantily represented in 

the Panjab, and the name is somewhat corrupted. 

^n:^ or nti[^ (m.), ^inj^, nfTi:wn^, ^i:ftr^iinft, and 

IqV is used also alternatively with ^RfllT? as 

'^yi^ « headman," '^VT^* ^^SnOT'T his wife. 

4<^M " peasant," ^ifflHUH W- (/•) 

((Skr.irrOTTfV,) e. 

ip?ITT"viDager"< , > ^piT^lfT " village woman." 

^^ « menial," W^t^ id. (/.) 

71f " cheat," T^TIqt his wife. 

From the feminine again, by some forgetfulness of its origin, 
has been formed a masculine ^^W^ ; so deeply seated is the feel- 
ing that a feminine in ^ corresponds to and postulates a masculine 
in d. Precisely in the same way in Latin the feminine marita, 
literally ''manned," from maSy tnari^, has given rise to a mascu- 
line mariius. 

Persian and Arabic words also undergo the prevailing Indian 
conversion into feminines. Hence spring the common but curious 
hybrids — 

Jl^t^ " angry," i.e. Arabic ,*ac with Aryan ^^ , fem. ^%^R| . 

9|frr? " executioner," Arabic Sip^ fem. ^RTT^^ his wife. 

41 4^ " artificer," Persian Jj^l^ fem. ^ilO'K^ ^^* ^*^®* 

Digitized by VjOOQ IC 

168 GENDER. 

iP[5r^Tl[" debtor," Ar.andP.jb^y fem. IK<9<|f,U|. 

^PQirri^ id.f corrupt Arabic v^L^i fem. qi\^ii,<ii- 

Masculines in ^irrt^ shorten the ^ and add HT for the fem., as 
\|^<!|T|; " weighman," ^4^11,^ ^^^ wife. 

When a word of this form ends in anun^ika, that is rejected, as 
IJ^^ " devotee," ^pTTi;:^ ^^^ ^^^®- 

So also with words ending in ^, as ^fi|[«(| '^merchant," 
fem. ^fU T iUf^UI , ^ft^nij^ "a merchant's wife." In P. as well 
as in H. this feminine termination implies rather the wife of a 
man who is described by the masc. word, and not a woman who 
herself does the thing implied, as in the H. instance of ^tT?^, 
which does not contain any imputation on the honesty of the 
lady so styled, but on that of her husband. 

Sindhi afl&xes this group of terminations to masculines of all 
sorts ; thus it is added in the forms ftj and Ijt to masculines in 
short M ; as 

WJ " a Jat," Hlzf^ and ^f^uf), also ^^^illllH) " * J®*** w'^«-" 

ip^t^ " a Belucb," «|^T|lfj)j his wife. 

iifV " lion," ^EfT^fnr " lioness." 

To masculines in o, but less frequently than the fem. in X • ^ 

it^ '' a dioimmer," ^f%rf^, ;^f%ni^, ^^ETRQI') And ^^ZniT^ 

Trumpp teaches that in the case of masculines in short u, as 
^IZ, the final u is changed to i, before adding the fem. termi- 
nation ; but if this were the case, we should have a double 
feminine. In the cognate languages, where the masculine ends 
in a consonant, we have feminines in jyi^ and this leads to the 
conclusion that the 1[ is p^rt of the termination, and ^^rfzflr 
should be divided thus: ^?Z + T^> ^^^ ^Wft + ftf. In the 
original Skr. of this form we have the masc. base ^aififn., and 
though the final ^ has been dropped in the nom., yet all the 
other cases retain it, and it must therefore be regarded as the 

Digitized by VjOOQ IC 

GENDER. 169 

true ancient form, and the feminine sign would be only 1[ ; but 
the modem languages having got hold of jyfty and regarded 
the whole of it as the feminine termination, have tacked it on 
to all sorts of masculines. The only real instance of a double 
feminine is in such words as H^^CTPOt? which is doubtless 
^ifZ+'^Rntft. A similar confusion of forms gives ^Wni[ftl> 
where there is ^f^, the original or regular fem. of it^, with 
the ^ changed to a semivowel before the fem. ending ^JHIhVj 
which again has been regarded as in some way connected with 
Xjfty and has thus come to be pronounced as Wl^ftl^ or perhaps 
^SCn^t having become by inversion ^QmC^y ^ ^ ^^^ other lan- 
guages, Sindhi has clapped on final ^ from a feeling that the 
word is feminine. 

In the case of masculines ending in A, this vowel is shortened 
to t* or elided altogether, and the phrase, "a Hindu female," 
may thus be expressed in six different ways ; thus 

tff^, fff^; fff^j tffftr; trf^imft, fff^ifrftr. 

Gujarati has the terminations in considerable variety, as ^^RJf, 
IOTj ^W^5 ^ and ^, of which ipil is peculiar to this lan- 
guage. Examples are : 

ifWY " servant," ^ft^RlTf 'ft^W " fen>aJe servant." 

ift'fY " washerman," Vt^TOT' Vt^^ " washerwoman." 

li^clo) *' a certain caste," l|^dbU| a woman of that caste. 

Taylor (p. 28) gives also feminine forms \Jt^P9, ^itdo^; 
but these are not noticed by Narmadft Shankar. They might 
arise from the final ^ of l[ift being shortened to the semivowel ; 
and the form tpm is froni 'W'fti Pr. "^Rnjt^ through the H. in- 
version ^1[f . Gujarati retains the Prakrit form of the nasal Hf. 

mfw^ " trader," ^rTftrwir. 'TTftr^Jinir. ^anftmrftr. 

mXH " a Maratha," 4J>>'^ia . 

IfR " tiger," Wl^^nt , TT^HT- 

^SP^ « master," Wft^lrtfV. 

Digitized by LjOOQ IC 

170 GENDER. 

It appears that in cases where there are two forms of the 
feminine for the same word, one form is in use in one part 
of the country, and the other in a different part ; but it is not 
stated which form is used in any specified part. 

Marathi in particular affects a long vowel in the final syllable 
of words ending with a consonant ; and in the class of words 
now under consideration it accordingly takes regularly ^[ij in 
the feminine nominative, but shortens this ^ to l[ or "^ in the 
oblique cases, because this syllable is no longer final in those 
cases. Thus we have — 

g^^ " skilful roan," ^4|0ui " sk'ilful woman,*' " a good housewife." 

IfT^ " devotee," ^TRfhlT " fenaale devotee." 

^TTO " tiger," TPI^ " tigress." 

npfy « peasant," ^^iPltW kis wife. 

Sanskrit feminines in \^, whether they have any corre- 
sponding masculine in Marathi or not, are treated by it in the 
same manner as the above : 

^1^^ " procuress," Skr. sf|pfV. 

^ifj^lir" sister," » Hf^rf^- 

"With regard to 0. and B., as they have no means of indicating 
gender, so also they do not possess any regular method of form- 
ing feminines. In modem times a large number of Sanskrit 
feminine words have been dug up out of dictionaries, and are 
now used in their Tatsama form ; and there are old Tadbhavas, 
like the word for " sister," which have not been formed from mas- 
culines in any modem language, but have come down ready made 
from ancient times. Of these it is unnecessary to say anything. 

Nor need we here introduce those words which are names 
of female animals, the male of which has a different name ; like 
H. ^ "ox," ^rtl "bull," 3^ "cow." Eaoh of these words 
stands on its own basis, and descends from some Skr. word 
which has no connexion with the masculine. 

Digitized by VjOOQ IC 

GENDER. 171 

§ 36. "With regard to the gender of that large class of words 
which, in all but Sindhi, end in mute a, or in other words in 
a consonant, I am convinced that it is quite impossible to lay- 
down any law. Each word must be traced back to its Sanskrit 
or other origin, and the reason for the gender will then be 
apparent. A large number of these words have come to their 
present shape from rejecting a final d or other feminine ending 
in Skr. (see the remarks in § 11), and it is therefore mere wtiste 
of time to attempt to find out any rules founded on the con- 
sonant which has thus accidentally become final. Stevenson, 
in his Marathi Grammar, occupies several pages with lists of 
feminines ending in a consonant, which it would be impossible 
for the student really to carry in his memory ; practice alone in 
speaking the language can teach him the gender of these words. 
It is of little consequence to know the gender in H. or P. In 
the forms of Hindustani spoken all over India, gender is habitu- 
ally neglected by all classes, and it is only in the area in which 
Hindi is the mother-speech that much attention is paid to it, 
and even there only by purists or accurate speakers. In a great 
portion of Bihar one hears such phrases as Tp^T TT W^TT % 
"your mother is coming;** and in the Parbatia or Nepali 
dialect, gender is not preserved at all. It exists, however, in 
literature, and its existence cannot therefore be ignored in 
Hindi even; and correspondents in the Marathi country and 
Gujarat inform me that the threefold gender of those languages 
is always correctly used, even by the lowest and most ignorant 
peasantry — an assertion which I should hesitate to receive in its 
entirety until confirmed by actual observation. 

In Marathi gender is distinguished by native grammarians 
in three ways, and the rules which they lay down are useful as 
fer as they go, though necessarily incomplete, and leaving much 
that is imclassified. They distinguish gender — (1) by significa- 
tion, (2) by form, (3) by both combined. Of these three, the 
second, or that by form, applies only to the termination, and 

Digitized by VjOOQ IC 

173 GENDER. 

bveaks down as soon as it gets from words ending in a Yowel 
to those ending in consonants. It is a good guide for vowel- 
endings, and consequently embraces the whole of Sindhi. In 
Marathi it helps ^s to some few rules for consonantal endings. 
Thus nouns in ^Jff are of course feminine, as shown in the last 
section; so also are those in Hf^ and V^. It is also asserted 
that nouns in nt, that is ?! preceded by anun&sika, are masculine ; 
but there are so many exceptions to this rule as to destroy its 
value for practical purposes. Similarly untrustworthy is the 
rule that nouns in ';^, ^, and ^, when forming the last member 
of a nexus, are neuter; for thereupon follows a long string 
of such words all masculine. 

In all the languages the large class of stems in ^, mentioned 
in § 9 (/8), is chiefly feminine, but contains enough masculines 
to render it unsafe for a foreigner to treat all such words as 
feminines in speaking the language. 

Reduplicated nouns are. nearly everywhere feminine ; but as 
many of these have vowel-endings, they pome imder head (3), 
which is a very perplexing method indeed, and leads to more 
confusion than either of the others. The only really trust- 
worthy rule under this head is that which prescribes the gender 
of abstract noims; those in Xfif or tpff being neuter in those 
languages which have the neuter gender, and masculine in 
those which have it not. "qip is masculine in all, and TfT 
feminine, as in Sanskrit. 

Under head (1) we are led into a maze of conflicting con- 
siderations. Leaving out, as sufficiently accoimting for them- 
selves, words which have natural gender, the artificial gender 
reminds us of the incoherent old " Propria quae maribus" of our 
school-days. Winds, mountains, rivers, periods of time, and the 
like, appear to range themselves under various genders with a 
most inextricable caprice, and tlve usual tagrag of exceptions 
hangs on to the skirts of their .army like a mob of unruly camp 
followers. No one perhaps really believes that the speakers 

Digitized by VjOOQ IC 

GENDER. 173 


who unconsciously developed their respective languages ever 
stopped to think what the gender of a word was, or intention- 
ally made it masculine because it was a mountain, or feminine 
because it was a tree or a river. The utmost we can admit is 
that instinctively large coarse strong things were treated as 
masculine, and their opposites as feminine, while things really 
or constructively inert were usually spoken of as neuter. 

It will perhaps be safer to conclude that all words whose 
gender cannot be ascertained from their present form must be 
traced back to the older form, where they will generally be 
found to be in possession of e termination which at once ex- 
plains the reason of their modem gender. 

Even this rule, however, is not free from exceptions, for 
there are not wanting words which, though clearly derived 
from Skr., have changed their gender in the course of ages. 
Especially is this the case with words denoting the body and its 
parts, where a sense of tenderness or familiarity seems to have 
operated to cause them to become feminine, that gender being 
used to denote small delicate objects. From the same sentiment 
the Bomans turned the names for parts of the body into diminu- 
tives, as auricula, ocellus, and the like. Instanqes are — 

Skr. ^f "body'' (m.), H. ^f (/.), P. and G. also ^ (/.), but in M. 
it is maac.y and S. ^f^ (/.), where the fern. 1[ has been added. 

Skr. ^fg^ "oath" (w.), H. ^f (/.), so also is the late Tadbhava 
tkM^ ; M. l|XI^ is m. and f, when meaning " an ordeal," but m. when 
meaning " a curse." 

Skr. ^1? "eye" (n,), H. -^Jt^ (/.). P. '11?^ (/.). G. ^rf^ (/.), 
S. ^rf% (/.), where the retention of the final ^ has probably led to the 
word being considered as fern. Chand uses %f^ ( = "^f^). 

Skr. UTt^ "sickness" (»i.), H. "^Ilf^ (m.) << mucous excrement," M. 

Skr. ^y " arm" (m.), H. Wff (/•)» P- ^d. (/.), S. ^ (/.). In G. 
it \%f., but takes a fem/ending iftft; so &lso in M. ^TlfV (/.). 

Digitized by LjOOQ IC 

174 GENDER. 


Skr. ^ "thingr" («.), H. IRT (/). P- ^WRT (/)» S. ^HJ (/.). G. ^RT 
(/.), M. -^ (/. and «.)» ^(/.)- 

Skr. f^^ "poison" (n.), H. f^RT, fSwr (»»0, S. f^, fty (/.). 

Skr. ^ "entrails" (mostly in pi. irt^nftT) (»»•)» H. ^iTf^r'^lt^ (/). 
also with fern, ending liTTO" *°^ ll?f^- 8« "^Rt^ (»».), M. ^ftTf («•)» 
G. lf?f^(n.). 

Skr. VTTJ "root," "metal," "element" (m. and n.), H. vra (/) 
"semen virile," S. ^[{^ (/.), P. \mT, ^?f (/), G. ^JTTf, VJ^ (/), 
M. mI. (/. and m.). 

The following are from Smdhi ;^ they are all feminine, though 
the Sanskrit original, and in some cases the derivatives from it 
in the cognate languages, which I have added to Trumpp's Ust, 
are either masc. or neuter : 

S. if^ " offspring," Skr. '^, H. ^^ (m.), and so in aU. 

S. f^ "semen," Skr. t^ (m.), H. fif^, ^ (/.), f^^ (/) 
" drop ; " in which sense S. has also f5t^ (/.), M. f^ (/.), G. f^ (n.), 
probably because ^ in G. is the regular neuter ending, and the u of this 
word has been confounded with it. 

Trumpp here inserts S. irT5 (/.) and "^nil^, which he would 
derive from Skr. TJ^i^^ (w.), but this is apparently an error ; 
the word should be derived from Skr. '^TTT or ^fiqifT " the day 
of the full moon,*' which is feminine in Skr., and carries that 
gender into S. Stack writes it ^^^, and gives the forms ^IJ- 
^rrtrt', M(\ui4||in> etc., which exist in all the other languages. 

S. ^ "thread," "wire," Skr. cfH (w.), H. ?Tf<t (m. and/., but I 
think more usually/.). G. M, id. (/), P. cftcf (/)» cf ^ (/). and l^ (/) 
" catgut." In all it has frequently the sense of the string of a musical 
instrument, as a lute, etc. 

S. ^ "woman's milk" (/.), from Skr. ^^if (m.) "breast." S. has 

^ Trumpp, Sindhi Gram. p. 89 et seqq. Only such words haye been taken from 
the list as show a change of gender from Sanskrit to Sindhi. 

Digitized by VjOOQ IC 

GENDER. 175 

also ^fW (i».) in this sense, and H. ^TTf i 'Pf (^O* ^^ Modern-H. the 
form ^(9{ is usual, but Chand has — 

f3S ^(T^ jm m^ H 

"From the same breast drank milk," literally (there was) a drinking 
ofmilk.— i. 170, 7. 

S. T^ " sealing-wax," Skr. ^ " lac" (m.). 

S. sre "ringworm," Skr. ^^ (m,), H. ^T^ (m.), G. ^1^<^ (m.), M. 
^TT (/•)> <»149 (>»•)• The fern, gender of ?T^ in M. is probably due to 
the existence of the Persian jl J "justice," which, like all Persian words in 
^, in M. is treated as a feminine. P. ^|^ is however/, also. 

S. WT^ "potash," Skr. ^^ (m.), H. ^TTT ("*•)» M« *^- » '^ s^""® 
senses it is m., in others/. P. ?§T^ (/.), G. ^TPC (''*•)• 

S. ini and S>^ "du-t,'' Skr. -^ (n.), H. S>^ (m. and/.), M. ^^ 
(m. and /.), G. if35 and ^^ (w.), P. W[3o and if^ (m. and /.), ^35 
however is always/. 

S. ff^ " assafcetida," Skr. ffj^p (m.), H. ff^ (m.), P. f^ (/.), M. 

Sindhi does not always stand alone in its change of gender, 
and it will have been noticed that the various languages are 
capricious in their use of gender as regards these words. Most 
of the instances given are monosyllables, and there is a ten- 
dency in all the languages to regard monosyllables, or nouns 
whose fin^l a has become mute, tis feminines. It would seem 
also that there is a faint and not easily definable tendency to 
attribute a feminine gender to certain consonants, as ^ and 7{, 
This may have arisen from the fact that a majority of words 
ending, in those letters are really by origin feminine. Although 
the stuff and backbone of these seven languages is pure Indo- 
Germanic, yet we must not ignore or imderrate the influence 
which Arabic vocables have exercised. This influence began 
in Sindh eo early as Muhammad Easim's conquest of that 

Digitized by VjOOQ IC 

176 GENDER. 

country in the early part of the eighth century. Mahmud of 
Ghaznin's numerous expeditions extended the knowledge of the 
sacred language of the Musulmans to all Western India in the 
eleventh century ; and by the middle of the thirteenth, Arabic 
words were heard in almost every city and town. Our seven 
languages were then only just growing into their present shape, 
and Arabic words were thus woven in with their structure as it 
grew. The idea of Hindi or Marathi ever having been without 
these words^ is a mere dream of Sanskritizing purists, In the 
most obscure corners of rural India these words are heard, and 
it not unfrequently happens that the old Arabic term is more 
familiar to the masses than the grand new-fangled Sanskrit 
word invented by the Pandits. Thus the somewhat hybrid 
word, <JJU^ jurimdna, is used in the sense of "a fine," and is 
understood everywhere, while the newly-coined "^^5^1^ is not. 
Now in Arabic the termination CL^^ is distinctly feminine ; in 
fact, it is the regular method of making feminines from mascu- 
lines: thus, C^jfc3, l:^j), ^^:-^^Jy '^^y f <^^*^, and many 
others of the same kind being feminine, there would doubtless 
grow up an impression that whenever a word ended in t it was 
feminine; and as the masses know nothing and care nothing 
about derivation, the use of that gender would extend to all 
words in t, no matter whether of Aryan or Semitic orgin. This 
final ci> is, in Arabic often written as h, though still pro- 
nounced ti thus we may write lt^jJ or ^j^, both pronounced 
by Arabs daulat. In borrowing these words, the Persians fre- 
quently pronounced the final as h : thus we get ^j>^ darjahy 
<u7^ martahahy and the like. From the frequency of these 
words an impression would, as in the case of t, arise that there 
was something inherently feminine in final h, and we thus ac- 
count for such words as ^f , ^"t^, being feminine. In some of 
the languages these words having become, by usage, thoroughly 
feminine, have had one or other of the vowel-endings peculiar 
to that gender added to them, as in S. ^ff and others. 

Digitized by VjOOQ IC 

GENDER. 177 

With regard to the other changes, which are less easily 
explainable, we must, I think, have recourse to the argument 
that the mediaeval Aryans, before entirely doing away with 
any peculiarity in the structure of the ancient language, 
began to be careless about its use, just as in the Merovin- 
gian period in France, the Latin case-endings were used care- 
lessly one for the other, till confusion resulted in their eventual 
abolition. So in the decay of gender, which has evidently 
taken place in the Indian group; the first step was a care- 
less and irregular use of the genders of individual words, 
by which, if any one word of very common usage were femi- 
nine, a whole group of other words of similar sound would 
be made feminine too, and the same with any familiar mascu- 
line word. 

§ 37. That the use of gender has shown signs of becoming 
less habitual, and gradually dying out, is undeniable. .While it 
has died out entirely from B. and 0., it is not much regarded in 
H. and P., and only two genders remain in S. The full range 
of three genders remains only in M. and G. When we cross 
the frontier into the territory of the Iranian languages, we find 
no gender at all. To ask why this is so, would be to ask a 
question which is virtually unanswerable. The neuter is cer- 
tainly a very useless abstraction, and it is not surprising to find 
it the first to be thrown aside. In the modem Romance lan- 
guages this has happened as much as in all the modem Aryan 
languages except G. and M., which have no parallel except in 
Modem High German. In the Indian group, the Prakrits retain 
all three genders ; but the earliest mediaeval Hindi has only two, 
the masculine and feminine, and even these two are much con- 
fused. It can hardly be said that Chand deliberately means to 
use a neuter, when he claps on an anusw&ra to a formless nominal 
stem to eke out his metre, or uses a word in its original Sanskrit 
form, as in the lines — 

VOL. II. 12 

Digitized by LjOOQ IC 

178 GENDEB. 

"First (there is) a taking of the well-adorned Bhujangi, 
Whose name (though) one is taken in many ways.'* — i. 5. 

Here the numerous anusw&ras are introduced merely to make 

the line scan, the metre being U I U"""" I U"""" I U K 

and a nasal is commonly doubled by prefixing anusw&ra : thus 
H^ is to be read as though it were written ^CTfWC^ ^^'^ IT'W 
as MV|M(4{^. Frequently, too, we find a masculine participle or 
adjective with a feminine noun. Thus, while in one line we 
have ^ ft?! ^'W (—where ^ "she became" agrees with 
Mf'ff*! "queen'' in a previous line — there occurs, a few lines 
below, W^it ftRsft" ^^R^t? "the bride made lamentation," where 
the verb is masculine, and in the next line \sn^ "took," which 
is also masculine, refers to the same noun URTT.* The same 
indifference to gender, even with living beings, exists occasionally 
throughout the poem, and it may therefore be concluded that at 
that age, or before it, the strict observance of the three genders 
of Skr. had ceased to be usual. The masculine being the most 
common of the two genders that remained after the disuse of 
the neuter, gradually absorbed the feminine in ordinary writing, 
imless there was any special necessity for the employment of the 
latter, as, for instance, in the case of living beings. While, 
however, the poets retain tolerably accurately the two principal 
genders, the people must have grown careless about therii at a 
comparatively early date ; for Nepali, whose origin as an inde- 
pendent language dates from a.d. 1322, has little or no cogni- 
zance of them ; and the earliest Bengali and Oriya poets, who 
wrote in the first part of the fifteenth century, show no traces 

^ Though we may here argne that we haye an instance of the ohjectiye constrnetion, 
though the subject is not, as it should be with that construction, in the instrumental. 
It will, however, be seen from Chapter III. } 67| that the early and mediseyal poets 
regularly omit the instrumental in the objective construction. 

•Digitized by Google 

GENDER. 179 

of the feminine as attached to anything but living beings* 
Grammatical gender had perished in the eastern area of the 
seven languages, then, by the fourteenth century, leaving only 
natural gender, and even that but indifferently attended to. 

It is not surprising that Bengali and Oriya should have lost 
the distinction of gender earlier than the other languages, see-* 
ing that they had so little occasion to use it. In these two 
languages the participial forms of the verb, which have in the 
other languages usurped the place of nearly all the old syn^ 
thetical tenses, do not exist ; and by their absence a great and 
constantly occurring necessity for the use of gender was taken 
away. Thus in Hindi the verb has only one synthetic or 
Prakritic tense remaining, namely, the indefinite present, the 
third person singular of which ends in "^ ( = %f), as in iR^ 
"he does/^ and the third plural in 1^, as in ij^ "they do." 
All the other tenses are formed by participles : as 

Present ^If^TfT " <Ioes," ^^RTT " sees." 

Past fiimi " did," ^?5T " saw." 

Future i|S^3]f "wQl do," ^%^ "will see.** 

Although the future is not a participle, but formed by adding 
itj to the indefinite present, yet this IJT, like the terminations 
of the present and future, changes its vowel for gender and 
number, and makes a feminine ift, pi. m. 3f,/. iff. So that in 
all three tenses there exists a necessity for remembering the 
gender, so as to make the verb agree with its subject or object, 
according to the nature of the construction. 

But in Bengali there is no such custom, thus : 

B. Aorist ii^, O. H^. 

„ Present ^arf^Bt^. „ 'RT^- 

„ Past Ifrf^, „ qiP^^. 

„ Future ^^^ or o^, „ ^irfT:^. 

None of these tenses change their form in any way for gender, 

Digitized by LjOOQ IC 

180 GENDEB. 

and there is therefore no need to remember the gender of the 
subject ; the objectiYe construction also is fortunately unknown 
in those two languages. 

If to this potent reason for disuse of gender we add the uni- 
versal shortening of filial long syllables, which led ultimately to 
their suppression, it will be readily understood that languages, 
which had no means left of marking gender, should soon cease 
to be aware of its existence at all, and in this respect should 
go even further than English. While our language retains 
distinct words for natural gender in the pronoun of the third 
person, these two do not ; % means " he," or " she," or " it," 
and all the cases of this pronoun are the same throughout, as 
will be seen more in detail in the Chapter on the Pronouns. 

Seeing how much the existence of distinctions of gender 
tends to make a language difficult to foreigners, it is not per- 
haps a mistake to regret that all the seven languages have not 
followed the example of these two, and got rid of gender before 
literature stepped in to arrest their development, and stereotype 
the forms they at present possess; and we may certainly set 
our faces against the obnoxious pedantry of some modem Ben- 
gali writers, who, in resuscitating a Sanskrit adjective, bring 
back with it the gender which the spoken language has long 
^go got rid of. 

Digitized by LjOOQ IC 



CONTENTS.— ^ 38. iNrLEcriON.— § 39. Pkepabation op the Stem in Orita 
AND Benqali. — J 40. The same in Hindi and Panjabi. — J 41. The same 


OF THE Stem. — § 43. Formation of the Plural in the Uniform Languages. 
— § 44. Formation of the Plural in the Multiform Languages. — § 46. 
Origin of the Plural Forms. — § 46. Origin and Analysis of the Singu- 
lar Oblique Forms. — § 47. Oblique Forms of the Plural. — § 48. Rem- 
nants OP the Synthetical System in other Cases.— § 49. Absence op 
Oblique and Plural Forms from Certain Languages. — § 60. Internal 
Modifications of the Stem in Ma^la.thi. — §§ 61, 62. Quasi-synthetical 
• Forms of some Cases.—} 63. Adjectives. — § 64. Numerals. — } 55. Case- 
affixes. — § 66. The Objectivb. — § 67. Instrumental. — § 68. Ablative. — 
} 69. Genitive. — § 60. Locative. — {§ 61, 62. Postpositions. 

§ 38. The modem noun in all the seven languages has the 
same number of cases as in Sanskrit, nominative, accusative, in- 
stnunental, dative, ablative, genitive, locative, and vocative. In 
Sanskrit these cases are distinguished by changes in termi- 
nation, as naras, naram, narena, nardi/a, nardt, narasya, nare, 
nara. This is the fashion with the old inflectional languages— 
a cumbrous and somewhat clumsy system, which the human 
race, in its onward march, has now in many instances discarded 
for the simpler and more spiritual method of detached particles. 
In the Indian group, Hindi stands, as usual, prominently for- 
ward in this respect ; while the opposite pole is represented by 
Sindhi, the rude and complicated speech of backward and im- 

Digitized by LjOOQ IC 


civilized desert- tribes. It is false philology to say that to Sindhi 
must be assigned the first place among its sister-tongues^ be- 
cause it ''has preserved an exuberance of grammatical forms, 
for which its sisters may well envy it.'* As well might the 
active rifleman of to-day, in his tight-fitting easy dress, and 
with his handy but deadly weapon, envy the warrior of old, 
staggering along imder half a ton of steel armour, and with no 
better tools than sword and spear ! As well might the modem 
traveller, carried at the rate of thirty miles an hour in a com- 
fortable railway carriage, envy the ancient German plunging 
through the muddy forest-roads in his vast and unwieldy 
bullock- waggon I Nature never works backwards, but ever 
onwards. The granite peaks of the Himalaya are worn by 
rain into a thousand wrinkles, and their substance is carried 
by countless rivers down to the plains of India ; should we call 
the fertile soil of the Gangetic delta the " degenerate descend- 
ant'* of those ice-bound peaks P Had the languages of India, 
and its soil, remained to the present day frozen hard in the 
bonds of a rigid synthetical system, or imbedded in the granite 
of its hills, they would not now suffice for the daily needs of 
its active and versatile millions. There is no language on earth 
so widely spoken as English, nor is there any tongue that has 
so freely and fully shaken off all inflections, genders, cases, 
tenses, and the rest : yet who shall dare to say that the lan- 
guage of Shakespeare and Milton is wanting in poetry ; that of 
Bacon, Locke, and Hamilton in precision and clearness ; that of 
Burke and Macaulay in power or eloquence P If the words of 
Sanskrit have in the present day lost many of their consonants 
and vowels, it is because they had too many to start with ; &th 
is a handier word than ashtau^ and no one would care to waste 
his time in saying ahhyantare who could express his meaning 
just as well by bhitar. Let us not be misled by imphilosophical 
talk. The modem languages are not corruptions of the San- 
skrit; they are improvements on it: and those that retain the 

Digitized by VjOOQ IC 


greatest amount of its antique and elaborate forms are the least 
advanced of the group. 

In none of the seven languages are the case-endings of 
Sanskrit preserved. The only trace of an inflectional or syn- 
thetical system is to be found in certain changes efiEected in the 
last syllable of some nouns, to prepare them for taking the de^- 
tached particles which express the relations of case. Strictly 
speaking, a noun has in none of the seven languages more than 
four forms : the nominatives singular and plural, and the modi- 
fied stem or oblique, or crude, or formative, for both numbers. 
Of the various names suggested for this latter case, I shaQ use the 
commonest, or oblique. The crude form of the noun is a* term 
Vfhich I shall use when speaking of the oblique employed without 
particles, to signify vaguely all cases except the nominative. Of 
course the languages differ very much in this respect, as in aU 
others. Oriya and Bengali have only one form for both nomi^ 
Dative and oblique in both numbers. Hindi has an oblique 
form only in one class of nouns, while the rest indulge in a 
great many variations, whereby they gain nothing in clearness, 
while they lose much in simplicity and practical usefulness. 

§ 39. To begin with the simplest of the group, Oriya and 
Bengali. Oriya has one form for all possible conditions of its 
noun. The case-particles, though not detached, do not affect 
the form of the stem. Thus, in a noun ending in mute short a, 
which is as much as to say, ending in a consonant, we have the 
declension — 

Sing. N. ghar, "house," 
Ac. ghar-A^fi, 
G. ghar-ar, 
L. ghar-e, for 

In nouns ending with a vowel the result is the same. Thus, 

PL ghar-m^ft, '' houses.'* 
for kar^re. 

Digitized by VjOOQ IC 


Tkjky "king/* gen. r&ji-r, "of a king," where the initial vowel 
of the particle has been sacrificed, but the stem remains un- 
touched. So in the loc. lAjk-re. The genitive, however, is the 
only case whose particle begins with a vowel, and consequently 
the only case in which any hitch can arise. Further instances 

pati, " lord,'* gen. pati-r. 

bahu, "wife," „ baha-r. 

swftmt, " master," „ swftmi-r. 

nat), " actress," „ natl-r. 

bhft, "earth," „ bhft-r. 

Some pedants profess to teach that words of the type ^TH!^ 
(^rf'nO shorten the i in the oblique cases, but this is a mere 
attempt at aping Sanskrit. The popular speech takes no heed 
of such refinements. 

Bengali is similar to Oriya in its treatment of the stem, which 
it subjects to no preparatory change when used with case-aflixes, 
though, as these affixes are different from those in use in Oriya, 
it is necessary to give examples. Thus, in nouns ending in a 
consonant, we have the following cases in which collision might 
occur, owing to the particles beginning with a vowel : 

Sing. N. kukkur, "dog." 
G. kukkur-^, 
L. kukkur-e, 

PL kukkur-^^. 
kukkur-^ diger. 

In the plural, however, it is more usual to add a word denot- 
ing plurality, as will be shown hereafter, to which the case- 
particles are affixed. The form of the plural in -erd is generally 
restricted to words descriptive of human beings, as santdnerd, 

Words which end in vowels retain the form of the nominative 
stem throughout, and avoid collision by eliding the initial vowel 

Digitized by VjOOQ IC 


6f the particle, just as in Oriya, though — owing to the very un- 
settled character of the language, which has not yet made up its 
mind which of its varying forms it will regard as the classical, 
or generally to be adopted— some difference of practice is ob- 
servable among good speakers and writers. I follow the guidance 
of Shamacharan Sirkar, who, however, like most learned Ben- 
galis, leans rather too decidedly to the Sanskrit and high-flown 
side of his language, though he is practical enough to notice 
most of the peculiarities of the spoken language. Thus — 

Sing. Nom. ghof&, "horse," PL N. ghort-gaft, etc, 

G. ghof&-r. 
L. gliof&-e. 
= ghorft-y. 

Further examples are — 

nftri, " woman," G. n&ri-r, N. PL nftri-r^. 

pasu, "beast," G. pasu-r, L. pasa-fe. 

jau, " lac," G. jau-r. 

Very frequently the plural signs are entirely omitted, and 
the fact of plurality left to be inferred from the context. In 
Old Bengali, instead of eliding the initial vowel of the case- 
particle, collision is avoided by inserting ^. This letter is not 
pronounced, but acts as a fulcrum merely. Thus the G. of 
^ "lac," would be written ^ft%^, or in its own characters 
WC??, the dot imder the ^ indicating that it is to be softly 
pronounced, and not, as usual, like / For instance, in Eabi 
Eankan's Chandi, where the merchant's wife Khullanft is sub- 
jected to the ordeal of being burnt in a house built of lac and 
other inflammable substances, Vishwakarman builds for the 
purpose — 

jau'er dfd, jau'er p&p, jau'er kapdt. 
Beams of lac, rafters of lac, doors of lac 

Digitized by LjOOQ IC 


§ 40. Hindi, with which in all important respects agrees 
Panjabi, comes next in respect of simplicity. All nouns ending 
in a consonant remain imchanged throughout. The only class 
in which any change takes place is that of noims in d, when 
derived from Skr. nouns in -aSf through the intermediate forms 
au and o. Nouns of this class form their oblique singular and 
nom. pi. in e. As Sing. N. ^Zl " son/^ Obi. ^, N. PI. ^. 
The oblique pi. is in Old-Hindi am or an, in Modem-Hindi -on. 

Sing. N. 11^41 " horse." 

N. PL ^=t%- 

Obi. i^t^-. 

To this oblique form are added the case-particles, as iftf% ^ 
"to a horse," Mt^f ^ " to horses." Exempted from this rule 
are those noims in d derived from Skr. noims whose nom» is 
already d; as 4,1^1 "king," ^TWT "giver:" these do not 
change in the singiilar oblique, or nom. pi. ; thus they say 
Tnrr ^ " to a king," ^inr " givers." The rule is carelessly 
kept in old writers, and even in the present day among the 
peasantry one may often hear iH^l ^ ; moreover, the neglect 
of the plural is very common, and it is colloquially more usual 
to employ the singular, as ^t^ ^^tTT " twenty horses." Instances 
of neglect of the rule in Old-Hindi poets are these— 

ftf^ ^TT W^ ^"1 II %TT Tiff ^TTT II 

" At that time came somehow Into the tent a snake." — Chand, i. 246. 

Where we should expect ^^'it. A similar passage is 
Tiw ^nn %TT Tf^ B 

" The king came into his tent." — Chand, i. 194. 

And in the Bhaktam&l& occur ^i\W if " in the cup," ^^rfT'lT % 
"of the boy" (Namdev.). 

The feminine noun in I undergoes no change in the oblique 
singular ; in the plural the inflection of plurality is appended to 

Digitized by VjOOQ IC 

D£CLEI(SI0)7. 187 

the simple uncliangecl stem, as %^ ^ " to a daughter," %^^^ 
^ " to daughters." No other preparation of the stem occurs in 
Hindi, which is thus, except in the one instance of nouns in A, 
as simple as Bengali or Oriya. ' Panjabi retains unchanged all 
nouns ending in a consonant ; those which end in a vowel are 
treated much as in Hindi. Thus — 

Obi. «jr«^i* 

So also in words which have anuswara over the final 4 ; as 

N. Sing. ^irX" boy.' 
ObL iT%. 

N. Sing. 4fU|^i' " trader.' 
ObU ^3Tfip{. 

N. PI. ^ftrt. 

Obi. ^rf^^. 

Panjabi has no fear of the hiatus, any more than Hindi has, 
and even in nouns where the final & is preceded by a vowel, it 
makes no effort to prevent collision ; one instance in point is 
the word last quoted, another is TH^^raf^ "a trier," obi. 
^T^i^R parakhaue, pi. n. the same, and pi. oblique M<4^xAI^^I* 
pardkhaiii&n ! where no less than four vowels follow one another. 
It would not be inaccurate, however, though imusual, to write 
M<^*)^i> thus avoiding the hiatus altogether. Still, a few 
words, ending in ^ and ^, change that letter to ^ before the 
termination of the oblique plural ; but even this is optional, and 
in a language so split up into dialects as Panjabi, no hard and 
fast rule can be laid down, fq^ "father," writes the oblique 
pi. ^^ "to fathers;" m^ "crow," IRT^ ^ "to crows;" 
?fT^ "mother," ifT^ ^ "to mothers ;" butft^^, ^RT^'^rt, are 
also heard in some districts. 

There is nothing more to be said about these two languages 
at present. 

§ 41. Gujarati is older in form than Hindi, and is in fact little 
more than an archaic dialect of that language, brought by the 
Ch&lukya Rajputs into the peninsula of K&thi&wad, and there iso- 

Digitized by VjOOQIC 


lated and out off from the other Hindi dialects, and thus gradually 
developed into a separate language, retaining antique forms whio]|r 
have dropped out of use in the parent speech. Its grammatical 
formations are consequently more complicated than those of 
Hindi, and in respect of the preparation of the stem it exhibits 
special peculiarities. Nouns ending in a consonant (t.^. mute a) 
in this language have more than one form, but the oblique form 
is not universally applied. In some cases the case-particles are 
affixed to the direct or nominative form, in others to the oblique ; 
the accusative and dative (T^hich are really only two different 
branches of one case, the objective) and the genitive affix their 
case-particles to the direct form ; the instrumental, locative, and 
ablative use both the direct and oblique forms. Thus from ^ 
" a god," there come 

Ace. and D ^if if. 

G •^^. 

Inst ^ . %^ ^. 

Abl id, 

Loc '^ Kf. 

The oblique form is the same as the nom. ; but there is also 
an oblique in e, as ^, of which more will be said in a subse- 
quent section ; this form alone is used as an instrumental and a 
locative ; but both these cases ordinarily take case-particles to 
define their meaning more sharply ; in which case we sometimes 
find the direct, and sometimes the oblique stem used ; thus, abL 
^y( ^ and ^% ^, imtr. ^^ SK^i), id. It will be pointed out 
in its proper place that Gujarati is fond of heaping up pleoneustic 
case-particles, this is one instance : ^^, already bearing the 
meaning "by a god," the form ^ ^ is pleonastic, but is 
utilized to express a slightly different shade of meaning. The 
result in Gujarati is a striking proof of the essential unity of all 
the languages in the group, the termination e running through 
them all in a singularly homogeneous way. It would be strictly 

Digitized by VjOOQ IC 


correct to say^ that in nouns \Fith a consonantal ending, 
Gnjarati, like its parent Hindi, has no separate form fot 
the oblique, the form in e, which has now crept into use, 
being, as will be shown presently, an old case-ending, or 
rather two separate case-endings .confused together. The plural 
of nouns of this class is formed by o, as ^^ "gods;'' to this 
the ending in e is added, as ^eHu, but only in the instrumental 
and locative. 

Masculine nouns in o, corresponding to H. in 4, from Skr. 
a-stems, have three forms for the singular ; the nominative in o, 
the purely oblique in d, and the instrumental-locative in e, 
thus — 

1. Sing. Nom. ^ ' (^I-J) "day." 

2. Objective ^TfTTT'^f. 

3. Instr. and L. ^TfrC%* 

But the instrumental and ablative append their case-endings 
to both 2 and 3, as ^TfTTT ^ and l^TfT% ^. The genitive 
uses only 2, as ^TIT^ •ft' ; and the locative, when it takes the 
case-ending, uses 2 ; when not, it uses 3 : thus we have both 
^i^|>j| irf and ^1^1% for "in a day." Strangely, too, the 
instr. sometimes adds its e to form 2, and appears as ^TJimi' 
The plural of this form is regularly d, as ^TfTTF " days ; '' but 
this appears to have been felt too vague, and in modem times 
an has been added, which brings the plural of this form into 
homogeneity with the plural of consonantal nouns ; thus it is 
now spoken ^^T^TTT^? to which latter forms case-endings are 
affixed. Here, again, the instrumental adds its e to the fuller 
form, giving a string of vowels, as ^TfT^T^t? ddhaddoe, " by 
days." Precisely similar in all but the nomiuative is the 
corresponding neuter noim in ^; thus, iftlt^ "a face," pi. 
ift^ft^, where the anuswara alone differentiates it from the 
masc. ; the modem form with o, however, drops the anusw&ra, 
and is thus identical with the masc., as in 4J)Cl'^l^* Noims 

Digitized by VjOOQ IC 


ending in all other vowels, including o, when it is not derived 
from Skr. bases in -as, add the ^s and os to the final vowel, 
without making any phonetic combinations. Examples are : 

N. Sing, pati, " lord," Instr. patie, N. PL patio, lostr. patioe, 
„ fiadf, "river," „ hadie, „ nadio, „ nadioe. 

„ vastu, '* thing," „ vastue, „ vastuoy „ vastuoe. 

„ vahU, " wife," „ vahde, „ vaMo, „ I , ', 

( vanuoe. 

„ chAo, "mortar," „ chhoe, „ chhod, „ chhode. 

The spelling of Gujarati is still very unfixed, and the fulcrum 
^ is often used, as in Bengali, where vowels come together, and 
has no effect on the pronimciation, Trf?nj and qf?f^ are written 

Marathi, which comes next, is in every respect a complicated 
language, having been unable to work itself free from that 
maze of forms and terminations which an ancient synthetical 
language always leaves behind it. In the matter we are now 
discussing, its usual ill-luck follows it, and the student is irri- 
tated by the variety of the changes he encoimters. In addition 
to special forms for the locative and instrumental, it has the 
regular four forms, the two nominatives and two obliques, and 
is, moreover, encumbered with three genders. The variations 
in the oblique, which is also the crude form, are divided by 
grammarians into six classes, a division which will be followed 
here, though it is not quite free from exception. There 
is some want of fixity in Marathi in this point, and authors 
are not quite at one as regards ^the forms to be used in some 

(1). Masculine and neuter nouns ending in mute a, lengthen 
that vowel in the oblique form of the singular; masculines 

^ No distinction is ordinarily made in G. between long and short t or u. In fact, 
the ordinary current hand has not distinct characters for the two sounds. 

Digitized by VjOOQ IC 


have the same form for pi. nom. as for sing., while neuters add 
ij; the oblique plural in both cases 'is formed by ^» Thus 
we have — 



N. Sing, mq "father," 


^" house," 


Obi. ^vn 

Obi. 5^mt- 



Similar to this class is that of nouns in ij^ and ^, which 
lengthen their respective vowels ; all three genders are treated 
in the same way. 

Obi. ^R^, Obi. ^. 

N. Sing. ^" honey/* N.Pl.IT^. 
Obi. inj. Obi. ^. 

The three short vowels are so far perfectly homogeneous in 
their treatment. 

(2). Feminines in short a, such as the words given in § 11, 
where the short a has arisen from shortening the long (i of a 
Skr. fem., form their oblique sing, in e, their pi. nom. in 4, 
and pi. oblique in dn ; and words of the same class, which have 
retained their original long d, are formed in the same way. 
Thus ^^t^ "tongue," and UTffT "mother,'' differ only in the 
nominative singular. 

N. Sing, ^i^ " tongue," N. PL ^ftHT- 
Obi. ^Ptil, Obi. iftHt. 

N.sing.?rnn» N.pi.incn. 
Obi. in%, obi. imrt. 

(3). These same feminines in short a, however, exercise the 
feminine privilege of not knowing their own minds ; for while 
some take e in the oblique, others take i, and a large number 
vacillate between the two. 

N. Sing, ^inr "fire," N. PL ^^Ipft. 

Obi. wifti Obi. ^inft. 

TSfail "a sKce," has HiHtWT and iffiNr "to a sUce;" 'q'^i ^ 
" offence," ^^1%^ and ^ ^ m . 

Digitized by LjOOQ IC 


(4). Long vowels follow the same general principle as the 
short ones ; their vowel being already long, no further change 
takes place in the oblique singular. The formation of the 
plural is multiform^ and will be discussed in a separate section ; 
the oblique plural, like that of class 1, only differs from the 
singular by the addition of anusw&ra in nouns ending in (i : in 
other cases anusw&ra is added to the form of the nom. plural, 
which is a recrudescence of the vowel of the singular into its 
semivowel, effected by the addition of d, 

N. Sing. ^T^" father," Obi. ^T^T, N.Pl.^-RT, Obl.^^. 

{"mother-) __^^ . 

in-law,") " ^^' " ^^^^' " m^^' 

yy- ^rsrat " woman," „ ^TSrat, „ ^^^, ,, ^T^Tlf- 

This may be considered the typical method of treating nouns 
ending in a long vowel, but there are yet two other processes. 

(5). A few words ending in ^ and ^ masc. and neut., and 
all diminutives in ^ and ^, have a way of their own. They 
reject the labial vowel, and form the oblique sing, in d, pi. dn. 
Masculines have the same form for both nominatives, neuters 
having the invariable neuter pi. im|. 

N.smg.jJJ^^„j obL-^rre^rcT. N.pi.^frz^^, obL^rz^- 

„ If^ "pony,'' „ TfgT* » c!^> >> Tfft. 

(6). Lastly, as if purposely to complicate their language, the 
Marathas have a totally different method of treating nouns of 
all three genders ending in long vowels, as well as those neuters 
which end in an anuswalra. The process here adopted is the 
hardening of the final vowel into its semivowel — W in this 
case having no semivowel of its own, takes ^ — and affixing to 

Digitized by VjOOQ iC 


the secondary stem so formed the masculine oblique termination 
d for masc. and neuter nouns, and the fem. e for fem. nouns. 
This class, however, is not well fixed, for many inasculine words 
in it may optionally form their oblique as in class 4, and the 
feminines as in classes 2 and 3. Thus f^^ " a scorpion," m., 
may in the dative form be t^raWT or f^^TTW. The fem. 
^^TO may either make as above ^|^|, or under this rule 
^rra%Wr. Ther^ are even some words which are spoken in 
three ways : thus ifTTf " grandson," m., IT^^TT, TTrITWT> and 
TRPTTWr. ^^ n. '' tear," -^m^}, ^9TWr, ^iRWT. 


of this class are the foil 



N. PL. 

' OBL. 











^^ "leech*' (/.), 





^m^ "brother" (m.), 




Exceptional is ^ " woman," making Obi. f^^ ; 'H, PL f^^ll, 
Obi. f^il'j. The above are all the forms in use in Marathi, 
and in this last class it must be noted that polysyllabic nouns 
in ^ do not join the v arising therefrom to the preceding con- 
sonant, but keep a short a between ; thus, cfT^ "ship," rTTT^Wr, 
not TfT^TWr ; also that monosyllables in u, not only change that 
vowel to ^, but still retain the vowel, though shortened, as ^ 
"needle," ^TT^T- 

We now pass on to Sindhi, in which language we have the 
good fortune of being able to avail ourselves of the inestimable 
labours of that sound scholar, Dr. Trumpp, whose grammar 
of Sindhi is the only grammar of any of our seven languages 
which has as yet been written on correct philological principles. 
In the following abstract I work entirely from Dr. Trumpp's 
materials, though I have altered his arrangement slightly in 
TOL. n. 13 

Digitized by VjOOQ IC 

194 DECLENSJ6N. ^ 

order to bring it into harmony with that followed in respect 
of the other languages. Sindhi has, like Gujarati, the locative 
in e, as %f "in a country" (Skr. ^^). It has, besides, the 
usual four forms, to wit, the two nominatives and two obliques, 
as in Marathi. 

(1). The consonantal ending or mute a of the other lan- 
guages is* represented in Sindhi by u for the masc, and a or i 
for the fem. Nouns in u have the following scheme of forms : 

N. Siog. ^re " slave.' 
Obi. ^TO. 


Obi. ^rot, ^f^,^[refir. 

Where the final u is preceded by a, the semivowel is inserted 
to prevent hiatus ; it is also inserted after ^, and optionally 
after other vowels, except i or i. 

^[[^"weed," Obl.-^. 



irt^^dish,- „ wt^. 

t^" mound," 

», f*^. 

in^^wind,'* „ ^rr^or^T^. 

^^" meadow," 

>, ^1^. 

Nouns which end in short u^ derived from other sources than, 
the Skr. a-stem, remain unaltered in the oblique singular, and 
in the oblique plural either follow the nom. plur. or sing. The 
words of this class are chiefly, if not entirely, the old words 
of relationship, which in the formation of the plural follow the 
Prakrit system. 


t^rar, ft^3f*T or fiRTf^r. 

HT^, HT^f'f or 9rR1Cf%- 

^or^ljp^, ^^f% or ^linif^- 

In the feminines in u the vowel is shortened in the oblique 
plural, as — 

N. Sing. ;TO "mother-in-law," N. PI. ^, Obi. ^Rrf%* 

'» ft^ "Hghtning," „ t^, „ t^r^. 

Digitized by VjOOQ IC 

t?r3" father," fll^. 
ITR "brother," ITR. 
^ " sister," 5^. 

N. 8INO. 




^dU "wish," 



(2). Nouns in short a, wMcli is always fern., do not change 
for the oblique sing. ; their oblique pi. is given below. 

N. PL. OBL. 

IT^orilT^, ^iflf or^nr^f^- 
And the same holds good of nouns in short i, both masc. and 
tern., the only peculiarity being that masc. nouns use the older 
and longer form in l[fif exclusively, as %^f?C " lion," obi. pi. 

(3). The termination o, corresponding to the d of Hindi and 
the other languages, modulates into e for the oblique singular, 
and dn, in, or ni for the plural ; thus— 

N.S.irBft"head," Obi. il^, N.PI.ITOT, Obl.^T^, If^, iT^. 

When thid final o is preceded by a or d, it inserts «?, to 
prevent hiatus, but not when preceded by other vowels. 

(4). The other long vowels may be grouped together, as 
below : 

isa N.Sing.WJ^, ObLWJ^T, N.Pl. ^ITOT*, ObL^TWrnrfif. 

"gardener," .•ICf*!, •l^^rf'f, •1[^f«f. 


"When the anusw&ra precedes the final vowel, it is retained in 
the oblique forms also. 

g 42. The following table exhibits the whole of the typical 
terminations of the nominal stem in all the languages except 
Bengali and Oriya ; these two languages having no change of 
the stem need not be included : 

Digitized by LjOOQ iC 



I V> W W ii 



I.' 1^ "Tf 




' ^1 1 4,( 4it ^ 

"i| .^ Ir ? 

1 1 1? 1 i 


4i( W Hnf 

IP IP IP <|p 


1' I ^iF ^If 



1^ I" 'S f 






^ ^ Zf 

p IP p 5 

I— 1 

i 1 I *^^ 
1 1 1 1 '? 


^ ^% 


1 1 53 § 
a ;r a ;§" 



1 s s 


.a .2 Ri ^ 

OQ CO Pk A4 

;^ d ;25 6 






Digitized by VjOOQ IC 




►y |r **» 

1^ s 


^ I ^ 














9 w V 

4^ U^ 

1P IF il-- 
^% ^ 






■ I 

*^ *^ *^ i 






^ 4>f 4if ^ 






i : : 

•J K-' S 

, d z d 

• • • • 

.a r j j 

OS <3Q S », 

ii 6 ;i 6 

• • • 

cS OQ S 

15 © S5 






Digitized by LjOOQ iC 


From this table, wliich is designed to show merely the lead- 
ing and typical terminations of each language, are purposely 
left out all rare or exceptional forms, such as the small handful 
of words ending in o, not deriyed from a-stems, a few in e and 
at, and the like. 

§ 43. Before entering upon the attempt to explain and ac- 
count for the manifold variations of the oblique form of the 
noun, it is necessary first to exhibit the system of forming the 
plural. The terminations of the nominative of the plural have 
been given along with those of the oblique in the foregoing 
sections, but nothing has been said in explanation of them. 

In respect of the plural this group of languages may be di- 
vided into two classes: first, those which form their plural 
always in one unchanging way; and, secondly, those which 
have more than one way of forming it. In the latter class 
stand Marathi and Sindhi, in the former all the rest. 

Of the former, or imiform languages, the simplest are, as 
usual, Oriya and Bengali. Oriya formerly made its plural by 
adding e, as kumdr, "boy," pi. kumdre. This plural is still 
found in poetry. Thus Dlnkrishna (a.d. 1520) wi'ites of Xrishna 
and Balar&ma — 

'* When the children weep with wailing cry, 
They easily surpass the song of the Kokila." 

—Masakallolaf iii. 110. 
And again — 

^^ tT ^ ^ ^ II 

gi*f it,qi ^^fil% ^ f^ II 

" Giving ear, listen, virtuous men ! 
Some days after the children were bom." — id. iv. 1. 

-^where kumdre^ jane^ and dine are the nominatives plural of 
kumdr, jan^ and din respectively. This form, however, was 

Digitized by VjOOQ IC 


probably found too indistinct, and was liable to be confounded 
with the instrumental and locative singular. A noun of multi- 
tude was therefore adopted; and the word selected was ;RTir 
^* measure/' a word already in use in isolated expressions, such 
as t^iRTR " daily," in other languages. A distinction is made 
in the use of this word between animate and inanimate objects ; 
thus they say ^)5r^ " ox," pi <^^44IM ; ^V4^\ " neighbour," 
pl TRfWrnX^; but ^^ "net," pi ^^RHf; WT "a load," 
P^' WTTTf . The e indicative of the plural is here retained, 
but tacked on to the plural sign mdny only in the case of living 
beings; it is omitted in the case of inanimate objects. The 
case-particles are aflixed to a form ^RT'f ; thus the genitive sign 
is ^rrt^, now-a-days often, but erroneously, written ITR^f^* 
The Oriya language is absolutely and undeviatingly regular in 
its way of forming the plural, the method above described is 
the only one in use, and the language does not contain a single 

Bengali has more than one way of expressing the plural, but 
I reckon it among the uniform languages because of the methods 
in use only one is a true plural ; the others are periphrases or 
compounds of two words, and not, strictly speaking, plurals at 
all. The regular form of the Bengali plural is H]'^ er& ; the 
initial letter of which is elided when it follows a noun ending 
in a vowel. It is not unfrequently elided, even when attached 
to nouns ending in mute a. This form of the plural is now 
in practice restricted to rational beings, and even in their case 
the periphrastic plural has gradually come to be used. Ex- 
amples of the true plural are — 

5^ « man," PI. Xf^j^ or yWTT- 

^T^ra « boy," „ ^T5rl^ or "Jn^RIITT. 

ifKt " woman," „ illO^T. 

Sometimes a form simply in ^, as in Of iya, is used for the 

Digitized by LjOOQ IC 


plnraly as lrt% "people." This is the genuine old plural, of 
which erd is the modem representatiye. The periphrastic plur. 
is formed by adding to the singular some word indicative of 
multitude. The words so employed are the following: Hlff 
"multitude," ^ "class," ^Emnr, ^RTO, W^, " all," K^ "whole," 
^> ^JWT, ^jfW or ^{f^, "number." Of these ipDI, ^, and 
perhaps '^[^Wy are more commonly used with rational beings, 
the others with irrational or inanimate things, and ^jftf is 
famOiar. Thus they woidd say, Mt4l ^CRKW " horses," %§ ijfir 
"children, brats." When these words are used to form. the 
plural, they take the case-particles and endings, leaving the 
noun quite uninflected, thus — 
• N. Sing. «5|[^ "dog," N. PL ^^nC^r^ "dogs." 

^^j- • fine ^nji^ "dogs," "to dogs." 

Instr. gf|p^ ^n55 '9JK^ "by dogs." 
AW. • ^p^TC ^T^ itlTl "from dogs," etc. 
And even in Words which use the true, and not the peri- 
phrastic plural, all cases but the nominative preserve the noun 
itself from change by inserting f^ " side ;" thus — 

N. Sing. I^cn "god," N.Pl. \^^\\\ "gods." 

G. ^^dlf<^< » or shortened ©^ "of gods." 
Obi. I^nnf^^ "to gods," etc. 

Lastly, Bengali being in the verb careless of plural forms, it 
is idiomatic to use simply the singular for the plural, leaving 
the hearer to understand what is meant by the context. This 
is especially the case where a numeral is included in the sen- 
tence, as ^tf^ ^ "^^ ^ "^ ^rtf% <2riir '^TO ^^f^^ «?ir®<i ^^^ ^*l? 
H&ri Muchi ei duij&tiprdy any a sakaljdtir anna khdy, "H&ris 
and Muchis, these two castes eat the food of almost all the 
other castes." Here, though the sense is clearly plural, the 
whole sentence is actually singular, and would literally be ren- 

Digitized by VjOOQ IC 


dered, "Hari and Muchi, this two caste almost of all other 
caste the food eats." An interesting example of this peculiarity 
may be adduced from Bharat Chandra's well-known poem, the 
"BidyA Sundar/* where he is describing the strange nations 
collected in the fort of Bardwan (a.d. 1740), 

^V5ll« ^^SPRT^ f^fs^ ^?t^n 

" In the first fort is the dwelling of the black-coat 
Ingrfij, Olandaj, Firingi, Far^, 
Dinemar, Elem&n, practises artillery, 
Wanderer various goods brings in ship." * 

The whole of these nouns are, of course, in meaning plural, 
but in form singular, agreeing with the singular verbs kare and 
Anaye. The passage, like all genuine Bengali, has a large 
sprinkling of Persian words, as ci;;'*^^ Jj^, ^J^^ S^V^* 

Hindi forms its plural very simply, with its usual practical 
common sense. Nouns with consonantal ending have the nom. 
pi. and sing, alike, as ^T!9fi|| "boy,'* nom. pi. "^T^Ri "boys.'' 
The oblique plural is in Old-Hindi formed by ^Rpf, which 
modulates into U^ and Vf • In the modem language this 
becomes ^*. Feminines ending in a consonant form the nom. 
pi. in % as ^XTT "night," nom. pi. .'?Q?f. Nouns ending in 
d masc, from the Skr. a-stem, make their plural in X[, and the 
corresponding fem. in i becomes ^^. Thus — 

^TCTIT "boy," N. PI. W^, Obi. WFif . 

^re^"girV' „ mwiNt, „ inrtWt. 

^ These nations are ourselves and onr European neighbours, who were known then, 
as now, to the Bengalis by their French names, the French heing in these days the 
most iinportant of the foreign settlers. Thus, Ingrdj is ** Anglais," or English; 
OlandSj, **Hollandais," Dutch; Firingi are the Portuguese, andFards, **Fran9ais;" 
Binemir is a corruption of Denmark, and Elem^ are ** Allemands,'' or Germans. 

Digitized by VjOOQ IC 


In Old-Hindi the nom. pi. fern, was ^, as ^StV^ "book/' 
nom. pi. ifWt ; tliis form is still preserved in the participle. 
These few forms are all that Hindi finds necessary to ex- 
press ideas, while Marathi and Sindhi have not had the skill to 
reduce their working apparatus into one or two plain homo- 
geneous classes. 

Panjabi is also quite uniform and simple in its plural. Nouns 
ending in d form the nom. pi. in e, all others have the two 
nominatives identical. An exception is made by a small class 
typified by l^vi "mother/' which forms its plural iTRtj though 
^rr^ would not be wrong for the plural. 

Oujarati is extremely uniform, simply adding o to the singular 
of all nouns of every description. In nouns ending in o, from 
the Sanskr. a-stem, this o becomes in the plural nom. 4, and 
this was originally the only plural termination for nouns of this 
stem. But o having become the general type of the plural for 
all other stems, the popular feeling could not rest content with 
an anomaly : the plurals in d were not felt to be true plurals so 
long as they lacked the universal plural sign o, and this was 
accordingly tacked on ; so that we now hear l^^WCt " ^j/' pJ- 
^ft^K^T^ ; and neuters of this class, which formerly made their 
plural in ^, now also take an additional "^j as ^ "ogg/' 
nom. pi. l^^Nl* 

§ 44. Marathi plurals are multiform, but need not be given 
here, as they have already been shown in the last section. The 
plurals of Sindhi, however, which are also multiform, require 
to be drawn out more in detail. Trumpp exhibits seven classes, 
which are here given. 

( Nouns in ^ masc. formN. PI. in H as ^7 << a well," N. PI. 1||{ (Skr. ^). 
( „ ^fem. „ „ ^ „ tf^ " thing," „ ^(Skr.^). 

a. „ ittCmt) » „ ^ „ Tret "carpenter," N. PL ^in 


Digitized by VjOOQ IC 



3. Nouns in ^ (*) { ^"'p^^?; ) ^ „ f^^ " scoipion - N.Pl.f^(Skr. 

^- » ^ »> tf ^SJRfe „ f^ " murder,*' N. PL f^T^ (Skr. 

^ » ^ or ^ „ 71^ « nostril," N. PL Tfjf or THTt- 

l^masc. „ „ ^ „ %t7" merchant," N. PI. %f7 (Skr. 

There are a few exceptions^ or rather words which preserve 
a Prakritio method of forming their plural, and these will be 
noticed in the next section. 

§ 45. Having now given the fiicts as regards the plural and 
oblique forms, we may attempt to trace their origin, and the 
steps by which they arrived at their preswrf condition. In 
doing this we naturally look to the Prakrits, rather than to 
Sanskrit, because these modem languages are in reality de- 
velopments of the Prakrits, or common colloquial speech of 
ancient India ; and although in the matter of phonesis they 
have been subjected to influences alien to those which prevailed 
over the Prakrits, yet in the present instances they are almost 
strictly Prakritic. In fact, these modifications of the stem, to 
fit it to assume case-particles, are nothing more than relics of 
the old case-endings of the early Aryan speech, which have 
been so reduced by use, that what remains of them no longer 
suffices to indicate the case-relations clearly, and particles have 
to be called in for the purpose. 

A remarkable family likeness runs through aU the plurals 

Digitized by VjOOQ iC 


of the uniform languages, and extends to a certain extent into 
the multiform languages. The difference between the two 
classes is this, that while the former have selected for use one 
or two Prakrit forms only, and have applied them to all kinds 
of stems indifferently, the latter have taken the separate Prakrit 
form for each stem. In this respect these latter are decidedly 
behind their comrades, and it may be suspected, and indeed 
partially proved, that Hindi and Gujarati have centuries ago 
passed through the stage in which Marathi and Sindhi still 
remain ; though, from the oft-lamented absence of literary monu- 
ments prior to Chand (a.d. 1200), we are unable to trace the 
steps more than by few and faint indications. 

The most widely uised form of Sanskrit noim is the a-stem ; 
this forms in the nom. plural masc. and fem. dA, neut. dni. 
The dh of the masc. and fem. is in many of the Prakrit dialects 
changed into e ; this change rests upon the tendency to break 
down d into e, already discussed in Yol. I. p. 137. In those 
Prakrit dialects which received most literary cultivation, the 
Maharashtri and Magadhi, the pi. masc. ends in d ; this ending 
has been preserved by G. and S. as the plural of masculine 
nouns in ; as G. miff, nom. pi. HTWT (modem >||ll||^), S. id. 
The ending in e has been adopted by H. P., and, in some cases, 
by M. also; as H. P. WZJy nom. pi. ^. In the case of 
masculines in d, Marathi has both forms, that in d and that 
in e. Of these two, however, the latter is the regular form, 
and the former i? only found in a few peculiar terms of respect 
and the like, as mm "father," nom. pi. irRT, where we can- 
not ascribe the word to a Sanskrit o-stem, but must regard it 
rather as an onomatopoetic or fanciful formation. The reten- 
tion of the d in the nom. pL of B. and 0. is not attributable 
to the Skr., but is simply due to the general tendency of those 
languages to reject all sign of plurality in favour of the peri- 
phrastic construction with i^, or iTTf > already described. 

The special Bengali plu^'al in erd divides itself into two parts : 

Digitized by VjOOQ IC 


the first of which, e^ is the old plural form of nouns ending in 
mute a ; wd the second, rd, by the fact of its being confined 
to living beings which possess a distinct individuality, as con- 
trasted with inanimate objects, connects itself with the redupli- 
cated plural of personal pronouns, as dmard **we," tomard 
"ye/* The origin of this form will be searched out in the 
Chapter on Pronouns. It is sufficient in this place to note that 
this ending, by origin a pronominal one, has passed over to 
a certain class of nouns, and that it is in this latter usage of 
modem date, not being found in the oldest literary documents 
of the language. So modem is it, that in the form erd we 
have really a double plural, the old ending in e having lost its 
sense of plurality in the popidar mind before the rd was added. 

The general preference of the lower kinds of Prakrit, and 
after them of several of the modem languages, for the 
form of the masc pi. in e, rests, in my opinion, on the fact 
that even in Skr. this is the regular form for the class of 
pronominal adjectives, as ^: "all/' nom. pi. igj^. It is ad- 
mitted that this method of declension represents an older and 
more genuine form of the a-stem ; and it is in keeping with the 
general pecidiarities of these languages to suppose that they 
derived their forms from the ancient Aryan language, and left 
on one side the modifications introduced into classical Sanskrit. 

"Next comes the neuter plural, which is of special importance, 
because it has usurped in many cases the functions of all other 
plurals. In Skr. it ends in dni, as \irt "wealth," nom. pi. 
^Jirrf'f. In Prakrit this form becomes di or din, as \9tirT1[ or 
VHn^ > *^ nsus addit nasalem," says Lassen, p. 307, but it seems 
more likely that the nasal is a relic of the if of the Skr. ^fif , and 
it would be more correct to regard the form ^^ as the older of 
the two, and to say of the other form " usus omittit nasalem." 
As early as the scenic Prakrit this form has passed over to the 
masculine, and we find ^frnt ^ pL <>f the masc. ^ilft. In 
Old-Hindi, the imiversal form of the pi. for all genders pre- 

Digitized by VjOOQ IC 


serves the Sanskrit form better than Prakrit does, by writing 
^ffif; as ^IXfif "men/' ^^rff^f "words/' At a kter stage the 
final i is rejected, and the form becomes Ifif , as 41^i|i«| "boys." 
The language, however, does not feel the need of any sign to 
distinguish the nom. pi. from the sing. ; and this termination, 
which, as will be pointed out later on, has become fused into 
one with that of the genitive in ^RTft, has remained in use only 
in the ojblique cases, to which the Middle-Hindi adds an ^, 
making the termination V!f . Thus in Tulsi Das we have (a.d. 
1650) regularly such plural forms as ^i^TT*i{ "wrestling-places," 
T^spf "some," •c|4,«|4^ "feet." The Middle-Hindi poets use aU 
three forms side by side ; thus in the same page one sees ^i^jRff^ 
"devotees," ^^TTTf "youths," and ^^ " houses." Only in the 
f em. of words with a consonantal ending is the Prakrit form in use. 
Thus from ^inf with anusw^ra we get the Old-Hindi form in 
ain, as "^^ "nights," from which comes the modem form ^^fH- 
The masculines of this form do not change in the plural, and 
the retention of this ending for the fem. is probably due to a 
desire to mark the difference of gender. Although the ending 
^Wt is, as we have seen, neuter, yet Hindi has lost the neuter ; 
and this ending had even in Prakrit got loose from the neuter, 
and was used for all genders, so that its application to the fem. 
in Hindi is not surprising. Marathi is more exact in using 
% which is contracted from ^inf, as a neuter, thus preserving 
the original gender. This i| is used for the nom. pi. of neuters 
of the consonantal noun, as well as those which end in ^ and ^, 
which harden the final vowel of the stem into the corresponding 
semivowel before this temmiation, as M\^ "pearl," pi. iftrif ; 
?rrt5^ " ship," pi. ?ff?f or wnC^. In the case of neuters of the 
d-stem, which already in the singular end in ij, the plural does 
not shorten the Pr. ^^^^ into J^, but, rejecting the first vowel, 
makes the ^^ long, and writes ^, as"^^ "plantain," pi. ^Bof* 
The same process takes place in Panjabi after feminines in long 
d, as irar "calamity," pi. ^Rrrtt or 'TWlt:^. 

Digitized by VjOOQ IC 


Gujarati has adopted one, and that the most common, of the 
Prakfit plural forms for other than a-stems, and simply adds 
"^ to all stems without exception, even to the class ending in 
0, which already had made its plural in d, so that it now 
becomes do, as noted above. 

In the same way as it retains if as the termination of femi- 
nines which end in a consonant, so Hindi uses the Pr. form 
^Vnf in another way after feminines in long i and A. Thus 
^f\ "daughter," pi. witlXT; ^PT^ "wife," pi. ift^^, where 
the final «' has been dropped, and the anuswara carried back to 
the d. The same takes place in Panjabi, as \it "daughter," 
nom. pi. \iHrt, and is extended also to feminines of the con- 
sonantal ending, as «irer "a word," nom. pi. irert- ChandMises 
this form also with masculines and neuters, with ^ or "? inserted 
to prevent hiatus, and, with his customary disregard of quantity, 
often makes the d short : thus we see M^M for ^T^iWf " spiritual 
guides," J|^'<|| " pots." To be connected with this is, I suspect, 
his not unconmion practice of adding UTT, as B^m "breasts," 
ippf "gurus," ^TfTI "observances;" but the passages in 
which these forms occur are so obscure, that it is difficult to 
pronounce a decided opinion. It is clearly a plural, however, 
in the following : 

"Addicted to great sin, blinded by riches." — i. 137. 

There remains to be deciphered the mystery of the midtiform 
plurals in M. and S. In the former the consonantally ending 
noun, if masc., undergoes no change ; when fem., however, it 
has two methods, it either makes d or I. This arises £rom the 
fact, that fem. nouns of this type are shortened either from fern* 
nouns in li or t in Skr. Thus ^ft^ " tongue," from Skr. f^iyr, 
pL ftiyr^ makes its pi. accordingly ^^fhn; but ^ITRT "fire," 
from Skr. ^rf%r, Prakrit lrf?»r, pi. ^RTifWt, simply rejects the 
final of Pr., and takes for its plural ^WHJV* In declining 

Digitized by VjOOQ IC 


words of this type in M., it is therefore always necessary to go 
back to the original, either in Skr. or Pr., in order to ascertain 
which vowel has been rejected, and thus to know how to form 
the plural, — a difficult process/ which the more advanced 
languages avoid by simply having one termination for all 
classes. Even to the natives of the province who have spoken 
this crabbed language all their lives, this practice causes diffi- 
culties; for the grammarians give a long list of words which 
may be formed in both ways, though, from not knowing the 
real reason of the co-existence of the two forms, they are unable 
to pronounce which of the two is correct. Nouns in short 1[ 
and ^ of all three genders make no change in the plural, being 
rare in the vulgar speech, and consequently not the subject of 
any regular development from Prakrit. Masculines, except 
those in 4, do not change in the plural. This is a practice which 
seems to run through all the languages to a greater or less 
degree, as has been remarked under Hindi; and all neuters 
of whatever class have a tendency towards if or its prolonged 
form ^. Finally, feminines in ^ and ^ form their plurals in 
"m and ^ respectively, in which we may see the Prakrit plural 
ending "^ changed into d, as in Hindi singulars. Prakrit, 
however, does not dislike the hiatus ; it makes plurals ^41^> 
^c^^lVt? from singulars ^^, ^^, where Marathi hardens the 
stem- vowel into its semivowel, and uses ^^, ^ig^|. 

Sindhi is quite as multiform, though less irregular than 
Marathi. The masc. in u, corresponding to the consonantally 
terminated noun of the other languages, forms its nom. pi. in a, 
shortened from the Prakrit pi. of nouns in o whose plural ends 
in d, m ^^3^, pi. ^Wl' Trumpp (p. 105) points out very 
justly, that as the singular ending in u is shortened from Pr. o, 
it is consistent that, the plural form should be shortened also; 
though I cannot agree with him in thinking that the Hindi and 
P. have done the same, as Old-H. makes the nom. pi. of this 
class in ani,. which has been subsequently rejected, and which^ 

Digitized by VjOOQ IC 


as we have already shown, is derived from the Skr. neut. ^L ia 
^rrf^* The feminiiies of this class ending in a and i have 
separate forms ; those in a, being shortened from Skr, feminines 
in d, like the corresponding class in Marathi, take either dn 
or ^n. The former of these is to be accounted for, in mj 
opinion, in the same way as Panjabi feminines of the type 
iniT, pi. ^irt, by the passing over of the Prakrit neuter ending 
^^ into the fem., and not, as Trumpp suggests, by rejecting 
the of the Pr. pi. form ^RT^t, because this process would not 
account for the anusw&ra. It is true that M. forms like whn> 
^rnft, have rejected the o of Pr. ^4)1^', ilJifl^' ; but then 
they have not taken the anusw&ra, which S. has. I therefore 
think it more correct to connect Sindhi in this particular with 
its neighbour Panjabi, than with distant Marathi. 

The other form of the pi. in ^ is common to all fem. nouns 
in the language. No satisfactory reason for this form has been 
shown. Even if we admit that the o of Pr. plurals has been 
changed to il, this does not account for the anuswara, which 
is too important a feature to have crept in by accident. It may 
have been extended to the nominative from the oblique cases 
of the Apabhransa pi. (see Lassen, p. 464), as is often the case 
in other languages. The remaining masculine nouns do not 
change in the plural. 

§ 46. Our next business is with the oblique forms, and this 
is perhaps the most intricate and difficult part of the inquiry. 
The oblique forms are, like the nominatives plural, remains of 
the synthetical declension of the Prakrits, and the mystery is 
not so much what they are, as how they came to assume their 
present shapes. 

"We must start from a fact, patent enough in the Romance 

languages, but not quite so patent in their Indian congeners, 

though even in them it can be established by illustrations as 

well as assumed from analogy. It is this, that at an earl^ 

VOL. n. 14 

Digitized by VjOOQ IC 


period the distinctive case-endings of the synthetic system be- 
came confiised, and one was often used for another. Some 
examples of this practice in mediaeval Latin were given at 
p. 113 of Vol. I. Now in the Prakrits the first step was the 
total loss of the dative, and the substitution for it of the geni- 
tive. The forms of the other cases also began to approach 
nearer to one another by degrees as they lost the distinctness of 
Sanskrit, so that in time much confusion crept in, and the 
terminations, which had been originally very different, all 
merged eventually into one form, which constitutes the oblique 
of the modem noun. The languages show their consciousness 
of the fact that the oblique form is a relic of the various cases 
(other than the nominative) of the old inflectional noun, by 
adding the particles which they now employ to indicate cases 
to this oblique form, and not to the nominative or direct form. 

It is not, in my opinion, correct to derive the oblique from 
any special case of the Prakrit ; it rather results from a general 
fusion of all the cases. I am here speaking of the singular 
only; the plural oblique, though analogous to the singular, 
must be traced separately. Taking first the ordinary Sanskrit 
fl-stem, which comprises a very large majority of the nouns in 
the language, it appears that, after the rejection of the dative, 
the Maharashtri, or principal Prakrit dialect, retained only the 
following scheme of case-endings. N. o, Ace. «m, Instr. ena, 
Ab. ddo, ddu, Gt, assa, L. e. But the distinction between the 
N. and Ace. was early lost, and there remained, even in this the 
most Sanskritic of the Prakrits, only four forms, ena, ddo, assa, 
and e, for the oblique cases of the noun. 

Maharashtri is to a great extent confined to poetry. "When, 
it is used in prose, it loses some of its distinctive features, and 
assimilates to the Sauraseni, the principal prose dialect. In. 
this latter the ablative is found to end in d, and dhi, and in. 
the Magadhi, the genitive has also begun to draw towards this 
type, exhibiting the termination dh. In the dialect of the 

Digitized by VjOOQ IC 


Saptasataka the ablative terminates in ddo, but has also the 
forms d, do, du, and dhi. The Bhagavati has a very similar 
range of case-endings. 

It is, however, especiallj to the Apabhransa dialect that we 
must look with regard to the modem languages, as this dialect 
seems in most respects to be the truest representative of an 
actuaUy spoken dialect. Here we find a still greater fusion of 
case-endings, abl. ddu, dhe; gen. dhe, dho ; loc. e, i, dhin. 

Here we must stop as regards Prakrit, for at this point we 
reach the gap of many centuries which has not yet been bridged 
over. When we come to the beginning of the modem period, 
all case-endings have been lost in most of the languages. The 
work of fusion went on during that obscure period, and we have 
no means of discovering how it proceeded. 

It is next specially to be noticed that Old-Hindi possesses 
a singular crude form ending in f^, which is applied without 
distinction to all cases of the singular. This form takes us back 
to at least three cases of the Apabhransa, namely, the ablative, 
genitive, and locative, and virtually includes four, as the dative 
was already fused into the genitive. The universality of appli- 
cation of this termination in f^ will be seen more clearly when 
we come to the PronouD, which in all languages retains archaic 
forms with peculiar tenacity. But. also in the noun its use is 
very frequent; thus Chand employs it as a dative, or sign of 
" direction towards," in 

ftrff ^iW f?t*l 'ITRft' M<fi| I 

"For what cause, Sishi, hast thou come to the house f" 

—Pr. R. L 45. 

And as an ablative in 

''Who (am I) from what race sprung.** — ih. i. 167* 

Tulsi Das has numerous examples, as ^fhrrfV ^f^ ^Vl^ 
" Having seen Sita speak (to her) ;" l^vnt 'WT ^fti^^rflF 'IS^ 

Digitized by VjOOQ IC 

£12 PEOiBNSiON. 

"tells the story of them in brief" (^%^) ; ^Wf^ff ^ W( 'WT 
^ifri^ " having told all that tale from the beginning." 

I^ow it will be observed that in the general fusion of forms 
the only two types that remain are the instrumental in ena, 
which is not changed as far as the data for classical Prakrit go, 
and the ablative-genitive-locative-dative in dhi; but by the 
simple process of the elision of the ^^ we get di, which naturally 
slides into e. The transition is well shown in the scraps of the 
later Apabhransa dialect preserved in old Chijarati, thus — 

H^^^ 'J ^^?^ ^^ '^^iw ^rwt 

"A hundred and eight (kinds of) knowledge were in the head of Eavan, 
Kot one knowledge profited him at the time of the destractioa of 
Lanka." » 

■ Here i^l^lQ^llll^ is the genitive formed by adding to the noun 
the termination Skr. TR (Pr. "?nj)> ^y which the noun is changed 
into an adjective, and agrees with the governing noim in 
number, gender, and case. <^Mlf^ (for ^tn%) being in the 
locative, the termination of the adjectival genitive must be in 
the oblique form to agree with it, and HUlf, is therefore used, 
in which at shows as the shortened form from dhi. At a later 
stage of the language this is written <T%. 

The instrumental retains its form of ena in Chand only in 
Gd.thst passages ; but even in them, for the sake of his metre, 
he often rejects the na, leaving only the e, and this again he 
sometimes modulates into at/a. 

From the above considerations it results that the distinctive 
features of the case-endings in Sanskrit all, in course of time, 
melted down into one form distinguished by the ending dhi, di, 
or e. We thus account for the fact that in Hindi and Panjabi 

^ '^Gujar^tlbh^bdno itihas," p. 44, quoted from a poem called Munjar&sa, the 
dato of whick » not glTen. 

Digitized by VjOOQ IC 


the oxytone noun in d forms its base in ^^ as ghofd, ghare. The 
same takes place with Sindhi nouns in o, which correspond to 
Hindi oxytones in (i, as wft " head," ^|%. Hindi and Panjabi 
have only this one method of forming the oblique, and, as I have 
stated before, only employ it in oxytone nouns ending in 4* 
All other nouns yirtually end in a consonant, and are not there- 
fore open to any change ; they have rejected all inflection ab* 
solutely. In mediaeval Hindi, words of this class use the form 
tf , with no connecting vowel, as i^T^ff " to EAm," and when, 
in process of time, this affix was dropped, there remained nothing 
but the bare stem, incapable of inflection. Even in those nouns 
which, strictly speaking, end in other vowels than d, the same 
rule is followed, because these languages take no heed of final 
vowels, and in speaking, at least, reject them always ; and ev^i 
in writing they are of little value. 

It is in the languages of the western group— as might be 
expected — that the greatest diversity exists, and to them we 
must now turn. Sindhi takes for the general type of its 
singular oblique the vowel a, for which Trumpp hints at a 
derivation from the genitive; but we have pointed out that, 
before the period of the rise of the modem languages as such, 
the Apabhransa Prakrit had already nearly obliterated all dis- 
tinction between the genitive and other oblique cases, bringing 
them all down to the common form dhi. If this be the case in 
the written Apabhransa, — which, though wandering far from 
the central type of Prakrit, must still, as a written language, 
be supposed to have retained greater regularity than the spoken 
language, — we are justified in supposing that, in the spoken, a 
still more complete fusion of all the case-endings must have 
taken place ; and it is not likely that a rude pastoral race would 
carefully observe such minute distinctions as that between dhi, 
dhe, dho, and dhin. Moreover, we notice that even in the 
written language, in one case at least, the final short vowel had 
been rejected, so that the ablative ends in dh or d. Sindhi is 

Digitized by VjOOQ IC 


very prone to the shortening of vowels. Thus it would be quite 
in harmony with the gen^*al practice of the language, to pre- 
serve out of all these endings nothing more than short a. In 
masculine nouns ending in u, we have really the consonantal 
Ending of barytones of the a-stem, and it is therefore only what 
we should expect to find that the oblique of this class should 
reject the w, which is hardly audible in pronunciation, and con- 
clude with a. Thus ^TH "slave," obi. ;^TO, may be traced 
back to a Prakrit form nom. ^\H\ ^T^j obi. ^I^lf^, ^TOT^, 
^rer^, ^lifllffi, all of which in the spoken language would 
fuse into ^TOf^, as in Hindi, and thence into ^TOf, and finally 
^W. The reason why nouns in o form their oblique in e after 
the Hindi fashion, appears to confirm the view I have taken 
of the origin of these nouns. When they owe their long final 
vowel to the fact of their being derived from Skr. oxytones, the 
presence of the accent on the final syllable prevents the termi- 
nations dhi, dhCj etc., from shortening their d into a ; all that 
takes place therefore is the rejection of ¥, and the termination 
thus becomes d'i, which by a natural process becomes e. Sindhi 
nouns in short a and t do not differ in their oblique from the 
nom. This is a further confirmation of the view expressed above. 
The Prakrit oblique of such nouns would end in ihi, ahi; but the 
i and a belong to the stem, not to the termination ; and when 
the hi is rejected, there remains nothing, so that the oblique 
cannot imdergo any change. Kouns in long i and A add an a 
to the stem, which is again a relic of the common form ahi 
deprived of its final hi. Long before the epoch of the formation 
of these modem cases, the Prakrit had disencumbered itself of 
the habit of making an euphonic combination between the final 
vowel of the stem and the initial vowel of the termination; thus 
the long i and A hold their place, imchanged by any commotions 
which might vex the termination. 

In Gujarati the only change that occurs is in the oxytone 
nouns in o, which make an obi. ia d. I have often before 

Digitized by VjOOQ IC 


mentioned that Ghijarati strikes the student as an archaic 
dialect of Hindi, whose development has been arrested by its 
isolation ; and it would be consonant with this view to regard 
the oblique form in d as derived from the full Prakrit dhi, by- 
simple rejection of the final hi — a process which we have shown 
to be followed by the other languages. Beyond this there is 
no further change for the oblique in Oujarati ; the termination 
in e belongs to the instrumental, which will be explained here- 

Marathi remains to be discussed ; lumbering along as usual 
with its old-world Prakrit baggage of terminations, it ofiers 
many troublesome problems to the inquirer. In nouns with a 
consonantal ending it lengthens the mute a into d : as bdp^ obi. 
bdpd; ghar, obi. ghard. In Old-Marathi the nominative of 
nouns of this class, like the corresponding class in Sindhi still, 
and Hindi and the rest in former times, ended in w, the Prakrit 
barytone o. Thus we have ^J]^, ^^, if«^4J, and the like. The 
Marathi consequently did not consider words of this class as 
ending in a consonant, nor does it now ; technically these words 
still end in short a. Not having, like the Hindi, rejected this 
final vowel, and with it all power of modifying the termination, 
it has been able to retain the oblique form in d from dhi, merely 
rejecting the hi, as in the others. This it has been able to do 
in barytones, whereas the others only preserve this ending, 
softened to ^, in oxytones. Similarly in a certain class ending 
in long d, it exhibits the oblique in d ; but in this case it may be 
equally correct to regard the d as merely a retention of the form 
of the nom. ; and this is rendered more probable by the parallel 
case of nouns in I and ^, which also, as shown in the table, have no 
separate form for the oblique. The noims in long vowels would 
be unable to form a separate oblique, because the obKque of the 
Prakrit woidd merely differ from the nom. by the addition of 
hi, as dhi, Ihi, Mi; so that when the hi came to be rejected, there 
would remain nothing. Marathi differs from Sindhi and all 

Digitized by VjOOQ IC 


the other languages in being still bothered with notions of 
aandhi, and still hardens the final vowel of a stem before a 
termination beginning with a vowel. Thus in a class of nouns 
ending in d, i and H, it hardens the vowel before the imiversal dhi 
of the oblique, making from bhdu, bhdvd and the like. In mascu- 
lines and neuters of this class, the Prakrit masc. form dhi has been 
used, leaving d: thus they say, hhdvdld "to a brother,'* which is 
hhdv'd{hi)'ld\ but in the fern, the Prakrit fem. is used. In 
these words so early as the Mahar&shtri Prakrit, the fusion of 
cases had taken place. Thus in the feminines mdld^ devi, bahu, 
we find only the following narrow range of endings : 

ifTWr'' garland," ace. TTRfy abl. 4i|<j||ff , Instr. Gen. Loc. ?nwni- 
^^ "goddess," „ ^pl „ ^^VfV f» » » 4^ll» 
^If "wife," „ Hjf „ inrff » " »» ^^^8fII- 

The abl. differs very slightly in sound from the other oblique 
case, because e is short in Prakrit, and consequently to the 
vulgar ear the general type for the fem. oblique would be e. 
Thence it would result that in the words ^rW» ^jft* after the 
final vowel had recrudesced into its semivowel, the form of the 
oblique to be added would not be d, as in the masc, but e, and 
we therefore find adsaveld, stripeld, which are sdsav-e-ld, striy- 
-e-ld. In this case the Marathi is more s^isitive than the 
Prakrit, for it does not permit the hiatus where the other does. 
The principle of changing the final vowel into its semivowel 
having been once introduced, has been iguorantly extended — 
through the influence of that blind groping after analogies 
which has been so fertile a cause of change in many languages — 
to nouns ending in d ; and as these have no semivowel of their 
own, the most frequently used of the two semivowels, % has 
been applied to them, so that we get an oblique sdsaryd from 
a nom. adsard. It is precisely on the same principle that the 
weak declension of nouns and the weak conjugation of verbs 
have gained so largely, and are still gaining, both in English and 

Digitized by VjOOQ IC 

DECLENfilOIf. 217 

Germany on the old strong forms. The sixth declension in 
Marathi, namely that which we are now considering, is one of 
the weak declensions in fact, and as such has gained ground on 
the strong declensions. Another weak declension is that which 
comprises masculines and neuters in ^ and ^, which lose their 
final vowel and form their oblique as though their nom. had 
ended in mute a, or in former times in u. Thus rd/sarii makes 
its oblique neither vdtsaru nor vdtmravd, but vdtsard, where the 
final vowel of the nom. has been ignored, and the masc. oblique 
sign d has been added to a stem tdtsar. On analyzing the 
words which fall imder this class, it becomes apparent that in 
most cases the final A or An is a modem invention, and not 
organic. They are, first, words compounded with the Skr. ^S^ 
and 9^, in which the final u is short, and might thus easily be 
confounded in Old-Marathi with barytones of the a-stem like 
^j;^, so that they formed their oblique in d, and the 
lengthening of the final vowel of the nom. is only another 
instance of the fondness of Marathi for final long syllables : 
secondly, they are neuters ending in the diminutive syUables 
vi and ^ which, as I have shown in § 24, are in the other 
languages TT or ^, and "^ or '^, respectively, and thus come 
imder the head of oxytones of the (Z-stem, and the oblique 
would regularly be d. There is, however, very great irregularity 
and confusion on this subject, the language not having made up 
its mind as to which of the three forms available it will use. 

Nouns ending in short a, corrupted from feminines in Skr. in 
d, of which the type is ftQET, M. ^sft^, form their oblique by 
adding e, as ^i^. This is the same rule as that followed in 
feminines in long i and A, and the e is the regular Prakrit 
oblique. Thus the Pr. obi. would be TTPirnC; but in this case, as 
the Marathi has lost the long d, it merely adds the e to the final 
consonant. This it does also in Tatsamas which retain the long 
d ; thus Yrr?n makes ^nif • Here I smell the Pandits. I suspect 
that the nom. had become* ifm? ^ in Hindi, in which case the 

Digitized by VjOOQ IC 


form J(ift for TrnfTT would be regular, just as ^^ for ^in[; 
but the Pandits have subsequently restored the long d. It is 
no answer to this to say that 4i|dl is found in the earliest 
writings in the language, because, in the first place, the writings 
are no guide to what the speech of those days really was ; no 
Indian writer could ever resist the temptation to use grander and 
more Sanskritic words than occurred in the spoken language ; 
the attractions of the so-called sddhu bhdshd have always been 
irresistible; and, secondly, the formation of the oblique form 
took place long before the earliest writings that we have; and 
it is therefore quite possible that when the oblique in e was 
formed, the nominative in current use was ^TRT. 

There is, as before stated, also another method of forming the 
oblique in use among the nouns of this class, namely that in I, 
which arises from the fact of their being derived from noims 
which in Skr. ended in t, or I. In masculines of this class the 
oblique Prakrit is in ihi, in feminines it is in ie, both of which 
have left in Marathi only L Examples of this class are the 


Skr. lrf%r "fire," "MU^, HnHtf. ^fPTt ^IPft- 

Skr.^ "belly," ^Bf^, f^frtf, ^, f^. 

Skr. ^ffe "fist," Tfff, 4J^lff, Jj^, ^^. 

Skr. ^Jt^ "assembly," ^f^» 'ftf^* ^^t^» 'ft^- 

This last word is almost a Tatsama; it would be completely so 
had it not lost the final i in the nominative ; it is used in the sense 
of " talk, gossip, conversation,*' also of "an affiiir, case, business." 

§ 47. We now come to the oblique forms of the plural, which 
are in all respects simpler and more imiform than those of the 
singular. Hindi has but one form for all classes of nouns, 
namely ijf , which must, I think, be distinctly referred to the 
genitive of the older languages. Sanskrit forms the genitive 
of the a-stem in ^IPlt for all three genders; the nom. and acq. 

Digitized by VjOOQ IC 


plural of the neuter end also in dni, which I have shown to be 
the origin of the nom. pi. in Hindi. The similarity of the 
genitive ending to that of the nom. has perhaps led to its 
preservation for the oblique cases, as in the period when the old 
inflectional case-endings had died out, and the use of the modem 
particles had hardly become fixed, there would be no distinction 
between the different cases of the plural; and Chand accordingly, 
as pointed out previously, uses such forms as ^T^l, ^«*IR, both 
for nom. and obi. plural. In his Gslth& passages, where he 
employs archaic constructions, we find a genitive in ^Ipt, as in 

"The sweet sound (made) by the anklets ofwamen.^^ — ^Pr. R. i. 17. 

(^=W^> ^=Ip^, ^5T^ = ^3T^ instrumental of a 
fem. form ^TJ>) 

Prakrit, in the principal dialect, makes its genitive in inJ^, 
and extends this form to all classes of nouns, totally rejecting 
the Skr. genitive in dm used in so many bases. In fact the 
terminations of the a-stem have, as a rule, completely over- 
ridden and supplanted all the others. Hindi has rejected the 
final anuswslra of the Pr. and turned the n into anusw&ra, and 
this rejection and softening are the probable causes of the 
present form in ^, the long vowel o having its origin in an 
effort to compensate for the loss of the n. Panjabi, which is not 
so sensitive, retains simply ^ for the oblique plural. There 
seems to be no room for doubt that the Pr. genitive is the origin 
of these forms, because the other cases have a different type 
altogether. Thus the Maharashtri has instr. in ehi or ehin, abl. 
in aunto or hinto, loc. in esu, emn ; and though the Aprabhansa 
has a different range of endings, yet they do not, on the one 
hand, approach the Skr. genitive, nor afford, on the other hand, 
materials for the construction of the Hindi oblique pL, the long 
of which is in my opinion to be accoimted for by a still 
further lengthening of long d, a letter which occurs only in the 

Digitized by VjOOQ IC 


genitive. That tlie Hindi form is comparatively modem is 
shown by the fact of its not occurring in any of the middle-age 
poets, in whose writings the form in ^rfif , ^1, or ipf, is used 
for the oblique as well as for the nom. This is why I referred 
at the beginning of this section to the similarity between the 
nom. neuter in dni and the gen. of all three genders in dndm. 
I believe that this similarity is the cause why no separate form 
for the oblique was struck out for so many centuries. It is a 
further confirmation of this view, that Gujarati, with its arrested 
development, has no oblique form for the plural, nor have 
BengaU and Oriya, both of which languages must have sepa- 
•ated themselves from the central Hindi type certainly earlier 
than A.D. 1400, as we find Bidy&pati in a.d. 1433 in full 
possession of a distinct set of forms. The Bhojpuri dialect of 
Hindi also does not possess the form in ^, but makes its 
oblique plural still in ^if , as ?ftTT ^> etc. ; so also the Marwari 
dialect, which uses only irt, which, like the Panjabi form, is the 
legitimate descendant of Chand's plural in ^if . 

By the aid of this view the terminations in use in Sindhi are 
also explainable. The Apabhransa dialect, which is more 
especially connected with Sindhi, has fused all its plural endings 
into a small range of forms, as instr. Mm, dhin, ihin, (Lhin, abL 
ahuUf gen. ahan, nom. and ace. du. Only the loc. retains a 
distinct shape dsu; and even in this, when we remember the 
facility with which Sindhi changes ^ into 1, it becomes 
probable that a form dhu would not be long in making its 
appearance. Later Apabhransa genitive forms in iheUy uhen, are 
also found ; so that we really get as materials for the Sindhi 
oblique little more than one form with trifling variations. 
The oldest and fullest form of the oblique in this language ends 
in fif preceded in each class of nouns by the final vowel of the 
stem. Here we have the if or ^ of the Skr. forms 'SRTf'f and 
^Hliftt Pr. mftr and W^, fused together. The other forms nt> 
% are readily deduced from the Apabhransa forms ^ and*f> 

Digitized by LjOOQ IC 


the vowel preceding whicli is merely a yariant derived from the 
final vowel of the stem. 

Precisely similar are the Marathi forms, which are strictly 
analogous to the singular oblique forms of each class, only 
differing by the insertion of anusw&ra, which evidently points 
back to the ^ of the Fr. gen., and preceded in each instance by 
the phonetic peculiarities which mark the singular. 

On the whole, then, we conclude that in both singular and 
plural the terminations of the oblique descend from a general 
form produced by the fusion of all the oblique cases of Sanskrit ; 
but there is this difference between the two numbers, that 
whereas in the singular no one case has retained its individual 
existence, or impressed its own special type upon the modern 
nouns, in the plural the genitive, by virtue of a special strength 
of type, and by its similarity to the neuter nom., which had 
usurped the place of the other nominatives, has preserved its 
individuality, and in a great majority of classes absorbed into 
itself the other cases. It might also, however, be said that to a 
certain extent, even in the singular, the genitive has had the 
preponderance, as the form in f^, although its earliest appear- 
ance in Prakrit is in the capacity of an ablative, is yet more 
easily derivable from the Skr. gen. in '^TS than from any other 
classical form. Thus, although the imiversal written Prakrit 
termination is ^TW, yet it is phonetically more natural that a 
form ^irfH should have arisen, which — by the operation of the 
tendency to change ^ into ^, a tendency which certainly exists 
in all the languages, though more extensively in the western 
members of the group — would become "^rff . It must be noted, 
also, that the change took place at a time when these western mem- 
bers were most powerful — ^Eastern Hindi, Bengali, and Oriya, 
not having then arisen. The period of the origin of these 
forms cannot be put later than the seventh century, when the 
decay of Buddhism brought about those great linguistic changes 
which laid the foundation of the modern languages; and at that 

Digitized by LjOOQ IC 


epoch the eastern parts of India were, as far as we know, com- 
paratively sparsely peopled by men of Aryan race. A phonetic 
change, therefore, of the character we are now discussing, would 
naturally be in accordance with the tendencies and peculiarities 
of the western tribes who then constituted the immense majority 
of the Aryans. 

§ 48. The forms of the oblique are not, however, the only 
traces which still survive of the old Sanskrit inflections. 
Simplest of all of these, the locative, which ended in H, has held 
its own down to the present day in many languages. In the 
Oriya poems this locative exists, as ij^ " in the village," iflf^ 
" in Gop," though it has now been superseded by the analytical 
locative formed by ^, and modem Oriya uses lf(^ \ or iffq yf^ ; 
in the latter of these '3T= WPf • 

In Bengali it still survives, as in Ift^ " in anger," ^if^ " in 
fear," "^ " in a chariot." Here also, as in Oriya, the tendency 
to an analytical construction led at an early date to the addition 
of the particle %, so that in Kasi Das's Mahabharat forms 
^i% It, T^ ^9 occur, although pleonastic, and often more with 
the sense of an ablative. After nouns ending in long d, this 
ending takes phonetically the form of J(, as ^tTRI "in a 
horse;" but after nouns in other vowels, the modem termination 
% is more usual. 

Hindi does not know this locative form : having adopted e as 
the oblique ending for the only class of nouns in which it 
admits a separate oblique form, there was no room in its system 
for the special locative. Ghijarati regularly retains it in all 
cases, with complete disregard of phonetic combinations, so that 
it is added to nouns ending in a vowel quite as freely as to those 
which end in a consonant. Thus we have ^itl( " in a custom," 
?f1^ " in a tent ;" but in nouns of the masc. o-class, in which 
the oblique differs from the nom., the change of termination is 
possible, and they consequently write ^|f |^ " in a day." So 

Digitized by VjOOQ IC 


also in the plural, which universally terminates in o, the e of 
the locative is added, giving oe^ no account being taken of the 
fact that the locative of the plural in Sanskrit ends in Tjg. 
This is the way with the modem languages. Having got into 
their heads the idea that a certain termination is typiced of a 
certain case, they stick it on to their nouns all round, over- 
riding the more intricate distinctions of the older languages, 
and thus gaining in simplicity and regularity. 

Marathi has a locative in ^, universally employed in the 
older poets, but now going by degrees out of use. The form 
is the same for both singular and plural, and appears to have 
arisen from the Pr. locative in ^lrf^[, which in Bhagavati 
appears as %f^y or as Weber reads it ^ftf. The later form 
was probably ^f^, which, by rejection of the a, is, strictly 
speaking, a portion of the stem, and throwing forward the 
anusw&ra, becomes t^. Marathi has by degrees got rid of the 
^, as in the similarly constructed forms of the oblique, and the 
lengthening of the final vowel is the usual Marathi custom. 
It agrees in practice with Gujarati, in using the singular form 
for the plural also. This may be pointed out as another instance 
of the preference of the Prakrits and modem languages for the 
older or pronominal declension, as this termination comes ulti- 
mately from forms like fl^f^^-^ 

Panjabi resembles Marathi in having a locative in \, which, 
however, is not restricted to the plural, and is not of very 
general use ; thus, ^T^ " in houses,*' If^ " in hands." In the 
singular, a locative in ^ is occasionally found, as lEf^ ; but 
this is more strictly an ablative, and I suspect we have here, 
not a relic of a synthetical case, but an abraded particle, as will 
be explained in another place. 

* The corrupt Konkani of Goa uses a locative in X> as ?rf% " on the bank," from 
Tf^, Skr. IfJ, where classical M. would have ?f^T. (Bumell*s Specimens of 
S. Indian Dialects, Mangalore, 1872.) This is probably only a shortening of the 
Skr. locative in 1(. 

Digitized by VjOOQ IC 


In SindU there is a locative, but only in nouns of tlie u-class 
(=mute a). It ends in t, shortened probably from the Skr. e, 
as in ^^, and not, as Trumpp writes, identical with the locative 
termination t, because this latter is not used in the declension of 
nouns of the a-stem, from which the Sindhi u-stems descend. 
Moreover, the declensional forms of the a-stem have to so great 
an extent swallowed up those of all the other stems, that we are 
hardly justified in looking to any forms but those of the a-stem, 
unless it be the old pronominal forms of words like ^. 

Besides the locative, several of the languages have also a relic 
of the old synthetical instrumental. This case in Skr. ended in 
the a-stem in ena ; and Marathi retains this form shortened into 
Tt> as ^*^ " by a weapon,'' ^^ " by a house/' Inasmuch, how- 
ever, as this termination is identical with that of the nom. pi. 
of neuter nouns, it has become customary to use a pleonastic 
construction by the addition of one of the modern particles 
41^^ " by means of," so that they would now write ^Ji§f ^d^^ 
" by means of a weapon." Here, in consequence of the back- 
wardness of Marathi, we are enabled to see in force a process 
which has occurred in the other languages also at a former 
time, namely, the gradual wearing away of the synthetictd 
case-ending, and the consequent necessity for employing a 
particle to bring out the meaning more clearly. 

Old-Bengali possessed also an instrumental ending in e, pro- 
bably arising from the rejection of the na of ena. This ending 
being identical with that of the locative, was abandoned by 
degrees in favour of an analytical construction with particles, 
though it is even now occasionally used in colloquial language. 
Instances from Bidy&pati, the oldest Bengal poet (born a.d. 1433, 
died 1485), are as follows : ^^^ ^%^f?f ji^dl ftT " that a 
virtuous woman becomes unchaste through love." — Padakalpata- 
ru, 980. t(^ ftl^ wfT Tf^ ^ T% "HtW "1^ ^7 nitnd I nothing 
counted, being foolish through that love." — Pad. 982. ^CT^ 
iW^ if^'l >^ " love has adorned his bow with, lamp-black." 

Digitized by VjOOQ IC 


•—Pad. 80. fif^ fT^ ^5^1^ ^TWF ^'T " having made itself a seat 
by means of its own new leaves." — Pad, 1450. Kabi Xankan 
(a.d. 1544) also uses this case frequently, as «nR(V5l*^'3 TRfT 
^ ^ WrSj " from without all hung down their heads through 
shame." — Chandiy 149.^ In the same way it is used in many 
later poets, so that the existence of the form is well established, 
though it has been banished from the modem literary style. 

Gujarati retains this form of the instrumental in ordinary use, 
though it, like M. afid B., has felt it necessary to have recourse 
to separate particles to define the meaning more fully. In this 
latter case, it is, as usual, pleonastic, having, besides the simple 
form ^ "by a god," also the forms ^ iffO^ ^^<1 ^% ^> ^s 
well as ^ ^. With its usual disregard of the hiatus, Gujarati 
adds this e to the final vowel of nouns ending in i and t/, also 
to the plural in o, as shown by the examples given in § 41. 
No traces of a separate instrumental remain in H. or P., though 
Chand in Gkihk passages uses the Skr. forms, as stated above. 

Sindhi is the only language which possesses a synthetic 
ablative, as %y " country," abl. %^, with variant forms ^iff, 
^if > %?T^, %1^. All these clearly proceed from the Sanskrit 
abl. in ^5(R^, as ^^|cl^, which in Pr. becomes first ^RT^ or ^51!^, 
then ^^, and in Apabhransa also ^HTST. The variant forms 
merely testify to the unsettled state of this rude language, in 
which, from lack of literary cultivation, dialectic forms abound. 
It would seem that while the most correct form is %|[T^, the 
most used is %^, and the anim&sika appears to be nothing 
more than a modern inorganic addition, such as Sindhi is fond 
of. The first of the two nasals in %f^ is also anunasika, and 
is merely the Sindhi method of softening a hiatus. Inasmuch 
as this form is purely synthetic, and not a mere case-particle, it 
naturally takes the place of the final vowel of the nom. ; if it 
were a case-particle, it would not do so, but would simply be 

1 This is the page of the Calcutta ed. by Gopal Chandra Chakrayarti, 1278, B.^. 

VOL. n. 15 

Digitized by LjOOQ IC 


appended to the oblique form. The form of the nom. represents 
the nom. of Prakrit and Sanskrit. Thus if^ is the modem 
form of Skr. T^^; the Sindhi abl. ifTt or mj^ represents on 
the other hand Skr. if^T^, and has never had anything to do 
with the nom. iT^. It is, therefore, not a correct way of 
putting it to say that the final vowel of the nominative is 
changed or dropped before this ending; on the contrary, it 
should be said that this case is derived direct from the cor- 
responding case of the older language. In nouns which end 
in ^, ^, ^, ^, the ablative case results from the custom already 
adverted to, of using the case-endings of the Skr. a-stem for 
nouns of all classes. In Sanskrit, while T?^ formed its abl. 
H4^|c^, iff^ made, not ^^i(1c^, but ^RTTt, ^^made its abl. 
^fHTTJ ; but all this was too complicated for the rustic folk. By 
far the larger number of the nouns in their language were of the 
type 1^^, and the minority were soon made to follow that type 
too. So it came to pass that in Apabhransa the common abl. 
ended in hu or he, with the final vowel of the stem preceding it, 
as il^W, 'f^ty, or ^iT^"^, ^^f , or oy . Sindhi goes a step 
further than this. It knows only one form, ^S|17; and this it 
simply sticks on to the stem, merely shortening the final vowel 
by the weight of the termination; thus, ^TJn^ "rope," abl. 
. 'ftt^nrt; f^f(% "wild beast," filf^^f. Thus that which was 
a bona-fide synthetical case in nouns of the a-stem, becomes 
almost a separate case-affix or particle in other nouns. This is, 
in my opinion, not an isolated instance of this process. If my 
method of interpretation be correct, there are, as we shall see 
when we come to the case-particles, several other instances of 
bonS,-fide synthetical case-endings having been broken off from 
the stem and used as particles. In the plural, this ending has 
come to be regarded quite as a particle, and is appended to the 
obKque form of the noim, as M<f«l^f "from houses," which is the 
obi. ^GT^fif +'^. It is easily seen that this ending has no business 
at all in the pluiral, as it represents distinctly the Skr. singular 

Digitized by LjOOQ IC 


fonn ^W?^; but so strong is the tendency to extend the use of 
one good simple form to all possible cases and classes, that no 
hesitation was felt in tacking the case-sign of an ablative 
singular on to a form derived from a genitive plural, in order 
to compose an ablative plural ! 

§ 49. In striking contrast to the wide range of synthetical 
forms observable in Marathi and the western languages, stands 
out the naked simplicity of the Bengali and Oriya, which have 
scarcely any variations of the stem. With the exception of the 
now almost obsolete form in ^, which did duty at once for 
instrumental, locative in the singular, and nominative in the 
plural, Oriya has no modifications of any kind; and Bengali 
has but one, the nom. plural in erd or r&. The question then 
arises, what is the reason of this difiference P Why should two 
members of the group have so thoroughly emancipated them- 
selves from the trammels of the old system, while the others 
are still so tightly bound up in themP The question seems 
to be parallel to that of the difference between English and 
German, the former of which has altogether rejected, wlule the 
latter has very largely retained, a synthetical type. In the case 
of English, whose development lies open before us, we can see 
the influence of the Norman race, — ^a race kindred in blood, and 
originally kindred in speech, to the subjects of Harold whom 
they conquered, but who had been put through a preliminary 
training by a long sojourn in France, as though purposely to 
fit them for the task of raising our rough English fathers to 
their present high position in the world. To the manly vigour of 
the old Norse pirate, the descendants of Eolf had added the grace 
and polish of the vivacious Frenchman. With that teachableness 
which was so pre-eminently their characteristic, they had sucked 
in all the sweetness and light which Europe then had to give. 
They came amongst us as a leaven of cultivation, and they 
made us what we are. On our language they worked a mighty 

Digitized by VjOOQ IC 


change ; and it is to this that I would especially draw attention. 
They taught us by degrees to throw away all terminations as 
useless, retaining only a very few which were absolutely necessary. 
Under their guidance, the language softened and simplified itself 
amazingly. Gender was the first thing to go, artificial gender 
especially ; even natural gender remained only in a few objects, 
and those indicated by uniform and regular methods. The 
numerous systems of forming the plural all fused into the 
addition of -es or -8 to the singular, and the case-endings dis- 
appeared, till at last our language stood forth clear and active 
like a trained athlete with his loins girt for the running.^ 

It is son^thing of this sort of influence that we should be dis- 
posed to seek for in Bengali and Oriya, and the difficulty of the 
inquiry is that we cannot find it. "We may, however, guess at 
it, and there are scintillations afforded us out of the gloom of 
Indian history which confirm our guesses, till at some points 
they almost touch on certainty. The first of these is the fact, 
now almost beyond a doubt, of th^ very modern character of 
Bengali. The earliest writers in that language, the Vaishnava 
poets, use a language so much akin to Bhojpuri and the dialects 
spoken in the eastern parts of the area occupied by the Hindi 
dialects, as to force on us the conclusion that the Bengali itself 
is nothing more than a dialect of Eastern Hindi. It is not till 
the beginning of the sixteenth century that we come to any- 
thing sufficiently marked to deserve the name of a separate 
language. Now long before that time, we know that Hindi had 
cast aside the greater portion of its synthetic machinery. The 
only relic of the modifications of the stem consists of the e of the 
obKque of d-stems, as in ^Zl, obi. ^^. But it has before been 
noticed that down to a late period this form was not fixed, and 
the oblique ended vaguely in ahu After d-stems this would 
naturally take the form dhi; and the rejection of the hi, which 

1 See on this subject Grimm, Deutsche Grammatik, toI. i. pp. 694, 709 (first ed.}, 
and Bapp, Comp. Gram. Verb&l-Orgaiiismus, yoI. iii. p. 163. 

Digitized by VjOOQ IC 


we have seen was a common process in all the languages^ would 
leave an oblique base in d for the d-stems, and in a (mute) for 
the a-atema; or, in other words, there would be no difference 
between the nominative and oblique. Further, it must be 
remembered that Bengali descends from the peasant language 
of Bihar, in which, as I know from long residence in those 
parts, it is not customary to form the oblique of the d-stem in e^ 
as it is in classical Hindi, The rustic of those regions will say 
ghord ko instead of ghore ho. The classical Hindi is not based 
on the speech of the eastern area, but on that of the western, 
and especially of the regions round about Delhi and Agra. It 
is not surprising therefore that Bengali, an ofi&hoot of the rustic 
dialect of the eastern area, should be ignorant of this peculiar 
custom. The crude form in ff is very common in the 
Vaishnava poets, as dnahi "to another'* (ant/a), premahi "in 
love," karahi "in hand,'* and the like. Perhaps the rejection, 
so universally occurring, of the My arose from its being con- 
founded with the common particle ff "indeed," and so being 
regarded as a mere emphatic addition easily rejected without 
altering the sense. If Bengali had attained an independent 
existence as a separate language at the early period when the 
other languages were passing through the stage of transition 
from synthesis to analysis, it would probably have struck out a 
course of action for itself. During all that period, however, 
Bengal was but thinly peopled, and its language was identical 
with Hindi, and it therefore partook of the changes which went 
on in that language. Its independent existence dates from a 
time when the sentiment of the necessity for indicating various 
relations of the noun by modifications in the terminal syllable 
had entirely passed away, and it does not therefore partake of 
any such changes. This modemness of Bengali must always 
be kept in mind in considering its present structure, because in 
recent times the language has been so overlaid with words 
borrowed from Sanskrit, in their Tatsama shape, that scholars 

Digitized by VjOOQ IC 


unacquainted with historical facts have been led to regard it as 
that member of the Aryan group which most closely approaches 
to the old classical speech, and to give it the position which is 
held by Italian in the Romance group. It is time that this 
misunderstanding should be removed. If we strip Bengali of 
all the Sanskrit words which have been brought into it during 
the last fifty years, and examine minutely its grammatical forms, 
and the true peeisant vocabulary, we shall find that it is more 
removed from Sanskrit than any of its sisters, and it wiU stand 
out in its true light as a coarse rustic dialect destitute of refine- 
ment and precision. 

With regard to Oriya, the same remarks hold good. "We 
know, from history that the Oriya race did not enter Orissa from 
the north, through Bengal, but from the west, across the moun- 
tains which separate it from the southern limits of Bihar. Many 
of the words of the language have the Bihar type of Hindi, and 
resemble Bengali only in those respects in which Bengali itself 
resembles Hindi. If we place the immigration of the greater 
part of the present Aryan element into Orissa at the beginning 
of the tenth century A.D., on the decay of Buddhism, it will 
result that the language which they brought with them from 
the valley of the Ganges must have been already to a great 
extent analytical; and their subsequent long isolation will account 
for the retention of forms which the onward march of the parent 
Hindi has long ago discarded. 

In both these languages there is also great reason to suspect 
non- Aryan influence. Eecent inquiries into the component ele- 
ments of the Hindu population in both provinces lead to the 
conclusion that a large portion is still, c^nd has always been, 
non- Aryan. In fact, it would not be going too far to describe 
the inhabitants of Bengal and Orissa as aboriginal non- Aryans 
converted to Hinduism by, and mixed up with, an immigrant 
element of Gangetic Aryans, whose language, religion, and 
physical type they have, notwithstanding their political in- 

Digitized by VjOOQ IC 


feriority, largely and deeply influenced. When the scholar 
whose non-Aryan dictionary has led us to expect a non- Aryan 
grammar from his pen, shall have given to the world the result 
of his labours, we shall be in a position to measure the extent 
to which the wild hill tribes and rude fishermen of the coast 
have reacted upon their invaders. . At present we are not able 
to do more than hint at the existence of such influence ; we can 
point out neither its direction nor extent. 

§ 50. Marathi possesses a process peculiar to itself as far as 
the seven languages which we are working at are concerned, 
but which is also apparent in some of the cognate dialects which 
it has been necessary to exclude from the present inquiry, as 
Kashmiri and Pashto. In our seven languages the changes 
which take place in the stem are confined to the termination, 
but in Marathi a class of words exists«in which internal modifica- 
tion is found. These are principally feminines in ^, formed 
from masculines in ^, examples of which have been given in 
§ 35. These words either reject the long I altogether in the 
obKque, or retain it shortened to t, or change it to its semi- 
vowel ^. Thus, Skr. ^1p|4), M. HT^tHf "a female devotee,'* 
obi. Hlf'Rin', where the i is shortened according to Molesworth ; 
or ^HT!nift> where it is changed to a according to Stevenson ; 
and in either case the oblique termination in Ms added, as in 
nouns of the form ^(PF, obi. "^Jinft. In cases like this it would 
probably be more logical to say that the noun retains its correct 
form in the objiique, while in the nom. it is lengthened in accord- 
ance with Marathi custom. 

There is a large class of these words, and many others not 
derived from ^if^ are treated in the same way without regard 
to their gender ; thus, 

^^V^ (m.) " rat," obi. ^f^J^T > ^^^h and even ^jfT- 

H15^(/.)" flour," „ ^RT^, sometimes ilRipJit. 

ipft^(n.)"earofcom,'' „ ^fjiprT. 

Digitized by LjOOQ IC 


The Sanskrit originals of these words are ^1^^ or '4<^4,, ^WT^ 
or ^QKf^nS, and ivfipf respectively, so that the oblique represents 
the older form more accurately than the nominative does. The 
practice has even been extended to words of Persian origin, so 
deeply seated in the popular mind is the principle of analogy. 

j^^ " a pack-saddle." M. Ijh^Hi , obi. iJl^K T • 
A-li^ "reverence." M. TfT^ft^ yi WHT^ and «^. 

1^ " a chain-bow." M. %9fVif „ %W^- 

Similarly, in a large number of nouns of all three genders, a 

long it. in the final syllable is changed to a ; as, 

Skr. ^%K " sprout." M. ^^^' , obi. ^NTT • 

^f^ " arrangement." M. ^9ra?f " trick," obi. ^Rfcft- 
^rf^^ « fly." M. iTTf^, obi. 411^^1 . 

In all these words, and they are tolerably numerous, there is no 
authority in the derivation for the long u ; and the form of the 
obKque, in consequence, is due to the slightness of the difference 
in the vulgar pronunciation between the three short vowels. 
The above process, it must be explained, is only applicable to 
words where the vowel is preceded by a consonant ; when it is 
preceded by another vowel, phonetic considerations induce a 
different process. ^ becomes hardened to ^, and BT to ^, in 
such cases, at least in the Konkani dialect. In the Dakhini it 
is more usual to shorten the vowel, as c|||<^ "woman," obi. 
Konkani ^T^%, Dakhini ^Tl[5f . Thus Tuka says, 

g[^ ^=1% ^^ «<ii^ "^i^ I ^^tf*Rrf T% ft^ ^Rtft II 

" Quoth Tuka, thus the good wife breaks out, sobbing she weeps and 
at times laughs." — ^Abhanga, 567, 3. 

And of words in BT, ^xftdb "temple,'' obi. Konk. ^efdbl, Dakh. 
^^351; as again from Tukaram — 

"Sings in the temple before the god.'* — Abh. 569, 3* 

Digitized by LjOOQ IC 


Of the two forms, the Dakhini, which merely shortens the 
vowel, is the older and more natural; the Konkani agrees with 
what I have called above the weak declension of Marathi, which 
bears evident signs of being comparatively modem. It is, 
perhaps, a sign of the increased tendency to Sanskritize, that 
the old hiatus, which neither Prakrit nor any of the modems 
objected to, should have begun to be replaced by the semivowel. 

§ 61. In noticing the peculiar ablative of Sindhi, I said that 
it was not an isolated instance of an inflectional case-ending 
having been detached and used as a case-particle. The other 
instances of the same process must now be exhibited, as holding 
a middle place between purely synthetical terminations, such 
as those of the locative and instrumental of Gujarati, Sindhi, 
and other languages, and purely analytical methods of indicating 
cases, such as the particles ^j*^, ^, and the like. They owe 
their existence to the tendency, arising out of the general con- 
fusion and abrasion of case-endings, towards adopting for all 
cases one good strong form of the older language. It is the 
Darwinian principle of the "survival of the fittest,'' noticed in 
regard to the phonesis of the group at p. 27 of Vol. I., and by 
virtue of which the strong forms of the neuter nom. pi., and the 
common gen, pi. ^Hf^ and ^irf, have usurped all the cases of 
all. three genders of the pluiral in Hindi and Panjabi. In the 
Sindhi ablative the form ^^ or irt is purely synthetical for the 
thstera, but it has, strictly speaking, no business with any other 
stem ; yet it has been applied to all stems indiflferently. Conse- 
quently, in all but the u- and o-stems, which descend from the 
Skr. a-atem, it is no longer a relic of the purely synthetical 
system, but has half migrated into a case-particle. On the other 
hand, it cannot be classed with case-particles, as kd, ke, ki, because 
these latter are modem formations, not derivable from any case- 
ending of Skr. or Prakrit, but independent words fused down into 
particles. I would therefore put these forms into a separate class, 

Digitized by VjOOQ IC 


and call them quasi-synthetical forms ; synthetical they are by 
origin, being derived from Skr. or Pr. case-endings; but they are 
not so in the manner of their employment, being used frequently 
in places where neither Skr. nor Pr. would use them, and some- 
times even quite detached from the noun, and used with the 
oblique stem, which itself contains all that is left of the syn- 
thetical case-ending. 

The first of these forms to be noticed is the Marathi ablative 
sign tf^, which is now used as a case-particle, and appended to 
the oblique stem, as sing. M^l^^ly pi. M<i$^; it is sometimes 
contracted still further into ^3R , and combined with the nom. 
form, as M^^* Lassen has long ago identified this form with 
the Prakrit ablative plural, which has two forms, ffiii^ and 
ifiHI"; from a fusion of both arises 3^ljt. This form is found in 
Early-Hindi Thus Chand,%ftn ^ ini%^ ifcT I "How far 
is it from Ajmer?"— Pf. E. i. 178, and ^IfTf finite JT ^F^ « 
"Quoth the Siddh, 'from what city?'"— 16. i. 184. Here we 
have two forms, 1^ and y<!ft, in the former of which the 
softening of the nasal into anusw&ra is compensated for by 
lengthening the vowel, and in the latter the anusw&ra has been 
shifted forward on to the last syllable. Lassen points out that 
these two forms are both pleonastic or composite,* that in (^^ 
being composed of the termination of the plur. instrumental f^, 
and that in ^<sjlt of the plural locative ending ^, with the 
particle ift? from Skr. 7|^, an adverbial particle with an ablative 
meaning, generally indicating " from a place.'* The form ff^ 
thus means "from by," and is a causal ablative; that in ^<>fft is 
"from in,*' and is a local ablative. Of course, with the fusion of 
the two forms into one, this distinction was lost, and, what is 
more important for our purpose, the distinction of number was 
by degrees lost too, so that in Old-H. and M. we have this 
ending used for singular as well as plural. M. has rejected the 
final lit, and lengthened the vowel, and the anusw&ra not being 
merely the anundsika or nasal breathing, but a method of writing 

Digitized by VjOOQ IC 


^ as the first member of a nexus, liaving been restored to its full 
form, the result is the form Ipf . 

In Nepali, a Hindi dialect of the eastern area, carried into 
the mountains by the last king of Simr&nw and his followers, 
when they fled there on the capture of their country by Tughlak 
Shah in 1322 a.d., occurs the ablative form flr^ "from," which 
may be, perhaps, connected with this form by a reverse process. 
If Am^o and aunto could be fused into hunto, they could just as 
easily become sintOy and this form would result in sit Nepali 
has not yet been studied scientifically, and its phonesis is as yet 
imperfectly known. In a brief memorandum on the language 
which I have,^ a tendency to favour the palatal sounds is observ- 
able, as in tTflft " ye," H. TO, and the like ; it is not therefore 
improbable that the form ainto would be preferred to hunto; 
and by a people so far east as the Nepalis, the preference for 
1 over ^ would not be felt. There is, however, another theory, 
which would derive sit from Skr. <Qf^. This is open to the 
objection that Ijf^ means "with," whereas sit means "from." 
As matters at present stand, I do not venture to decide for either 
theory, but leave the matter sub judice, but with a preference 
for the former. 

To return to Marathi. There is an older form of this case in 
yf^Rltj IffM, the existence of which would seem to militate 
against Lassen's theory, as it 'is scarcely possible that hunto 
should have become huniydn. To this it may be answered, first, 
that these two forms occur in poetry, and the lengthening may 
be due to metrical causes ; and, secondly, a confusion has pro- 
bably arisen in the popular mind between this form and that of 
the Prakrit gerund in ^Rir= Skr. in ?rr. In this form also the 
old M. adds;^;^, as in the word ^^tPRTt "sobbing" (Skr. 
perhaps ^rfS^^, Pr. tt^lir)> ^ ^^^ quotation from Tukaram 

^ And for which I beg to express my obligations to Dr. Wright, Residency Surgeon 
at Kathm^ndo, who courteously complied with my request for informatioui by sending 
this note drawn up by a Nepali Pandit. 

Digitized by VjOOQ IC 


in the last section, though modem M. now rejects this termina- 
tion, and would write i^^«|. This jyff does not seem to be 
organic, but a mere fashion of a certain period. There is always 
an element of confusion in words beginning with hu or ho, on 
account of their resemblance to the substantive verb ^, which 
has from an early period sounded ho. It is then possible that 
the form hunto was regarded as a tense of ho " to be," and the 
shortened form hiJtn would be confounded with the gerund 
^t^if; and as this latter was subjected to lengthening into 
ift^rfiRrf, so the ablative form may easily have been at the 
same period lengthened to :Ff'n5rt« The principle of false 
analogies here at work is to be found constantly in all depart- 
ments of language, and to it may be ascribed numberless eccen- 
tricities in the vulgar speech. Parts of the verb ho are used 
in the other languages as case-signs. Thus Bengali uses its 
infinitive (originally a locative of the present participle) ip|^ 
for lft^> in the sense of "from," a usage only to be explained 
by supposing the idea to be that of having previously been at a 
place but not being there now, which involves the idea of 
having come away from it; thus ^r^ ^f^ ^lRfMiH> "I came 
from the house," would be literally, "in being at the house I 
came," or, as we might say in colloquial English, "I have been 
at the house and have come away." Similar is the use of hoke 
or hokar "having been," in the sense of "through" in Hindi; 
thus, to express "I came through Benares," one would say ^^^- 
T^ f1% W^ ^, literally, "having been in Benares I am come." 
It is not surprising, therefore, that the Marathas should have 
thought that ^if "from" was contracted from ^V3I«| "having 
been;" and were it not for the Old-Hindi forms used by Chand, 
we might have been prepared to acquiesce in this view. As it 
is, however, Lassen's theory seems decidedly correct, and has 
been therefore adopted in this work. To this formation may 
also be ascribed the ^ of Panjabi locatives, as in iPCt, 
mentioned a few pages back, which is thus to be regarded as in 

Digitized by VjOOQ IC 


reality an ablative, and contracted from ^r; tff. In fact, the 
locative may itself be called an ablative of place, the meaning 
"in** being expressed by the ablative in many Aryan languages. 

§ 52. Another quasi-synthetical form is the dative in Marathi 
which ends in ^, and is derived from the Prakrit genitive in W, 
Skr. ^, the dative having been eariy absorbed into the genitive 
in most Prakrit dialects. This form in Marathi cannot be 
classed with the locative and instrumental as a purely synthetic 
relic, because it has ceased to be restricted to those positions 
where it would occur in Sanskrit. In the parent speech, the 
genitive in aaya belongs exclusively to the declensioti of the 
a-stem ; in other stems, the genitive is formed by the addition 
of ahy and in some cases &h\ the vowel is also rejected or 
amalgamated in some nouns of the i- and e^-stems, and in some 
few classes the ending is uh ; so that the singular genitive type 
may be generalized as simply visarga preceded by certain 
vowels, whose variations are determined by the form of the 
stem. But visarga is too weak a thing to last, it is almost 
entirely swept away by the Prakrits and their modern descen- 
dants, and recourse is had to the stronger form asya, or rather 
sya of the «-stem, with which also agrees the older pronominal 
declension. This is by the Maharashtri Prakrit applied to all 
masc. and neut. stems indiflferently, thus 

Pr.^^ "tree," ZR^^ Skr. JHf^. 
„ ^rfl^«fire," ^pJIW „ ^: 
„ ^ "friend," ^TOW » ^Wtt 

For feminine stems, however, Prakrit does not use this form, 
and even for those given above it has an alternative form in 
lift for the i- and t^-stems ; «o that we may trace the ending sya 
thus : in Skr. it is used only for the masc. and neuter genitive 
singular of the «-stem, in Maharashtri for all singular genitives, 
masc. and neuter, in modem Marathi for all datives of all three 

Digitized by VjOOQ IC 


genders, singular and plural. It has therefore,, in the latter, 
come to be almost a case-particle; but, probably owing to its 
shortness, it is, as we learn from the grammarians, used in a 
somewhat restricted manner, especially in the Dakhin or central 
part of the Marathi area, in which a more correct form of the 
language prevails than in the Konkan or coast-line. The identi- 
fication of this ^ with the ^ of the Skr. genitive would seem 
to be fatal to the theory which would derive the M. genitive in 
^, etc., from ^. The same form can hardly be the origin of two 
separate cases. This, however, will be more fully discussed in its 
proper place. 

It is in Marathi more especially that traces of these quasi- 
synthetical forms are found, though even in the other languages 
faint indications may be seen. They are valuable as supplying 
a link in the chain of development, and as showing how, as the 
spirit of inflectional construction and expression died out of the 
popular mind, the old case-endings fell into inextricable con- 
fusion, one swallowed up another, four or five fused into one, an 
ending peculiar to one case was appropriated to another, the 
distinctions between diflferent declensions were obliterated, and 
the languages, like new wine left to settle in the vats, deposited 
all their sediment, and were racked off clear and sparkling. It 
was all very well for a dreamy old Brahmin, who had nothing 
better to do with his time than to sit in the shade of a tree and 
doze over philosophical abstractions, to have a dozen different 
ways of declining his noun or conjugating his verb ; and it was 
no difficult task for him to recollect each one of a vast growth of 
terminations and inflections : but life is too short now-a-days for 
such minutiae, the business of existence is too varied, and time is 
too valuable. The modem languages are not objects of pity, as 
having degenerated from a higher level ; they are rather to be 
congratulated on having known how to bring order and simplicity 
out of a rank chaotic overgrowth of forms and types, and having 
thus become fitted for use in these bustling modem. days. 

Digitized by VjOOQ IC 


Among the particles that are now used as signs of case, there 
are several which may possibly be classed as quasi-synthetical ; 
but they are not given here, because some doubt still hangs 
over their origin and real nature. They will be duly noted 
in subsequent sections. 

§ 53. Adjectives partake of the nature of substantives in so 
far as their form and structure are identical with them, both 
being nouns, though the former are attributive, while the other 
are appellative. But as adjectives are generally coupled with 
substantives, and as these latter are the principal words in the 
sentence, indicating by themselves the object referred to, it is 
natural that they should do the bulk of the grammatical work, 
the adjective being merely appended to qualify the substantive, 
and not therefore requiring to be so accurately inflected or 
declined. In several languages the adjective, consequently, 
undergoes less change than the substantive with which it 
agrees. Here, again, we have the old common sense system of 
simplifying as much as may be. In Sanskrit, Latin, Gqthic, 
and the older languages of the family, the adjective was made to 
agree with the substantive in* gender and case, so that each 
adjective presents a triple declension, masculine, feminine, and 
neuter. Sanskrit sometimes shirked all this elaborate concord. 
Its array of declensional and conjugational forms was so for- 
midable that Sanskrit writers themselves seem to have felt the 
burden of so vast an amount of wealth, and to have endeavoured 
by various tricks of composition to shake off the load. Thus 
the necessity of inflecting the adjective to follow all the varying 
phases of the substantive is to a great extent evaded by com- 
pounding the two together by the method known as Karmadh&- 
raya, by which, whether, as is more common, the adjective precede, 
or, as also occurs, the substantive precede, only one inflection is 
used. Thus, instead of saying ^V^^rM^ "a blue lotus,*' they 
combine the two words into one, producing ifl^cM^) and thus, 

Digitized by VjOOQ IC 


when they require to use the genitive for instance, ^M||rM<^^ ex- 
presses the meaning quite as clearly as the longer 4^^^^ ^(^T^R9- 
Latin has not this power, and every one is familiar with the 
clumsiness that results when one has to string together a 
number of adjectives and substantives in the lengthy genitive 
plural forms 'Orum and -arum. 

All that is really necessary in the inflection of adjectives has 
been retained in the modern Aryan languages, and much that is 
not necessary in some. So long as there exists some clear 
means of knowing which substantive in a sentence any given 
adjective qualifies, it is only needful to decline the substantive. 
Confusion can only arise in a few instances. Thus in English, 
when we say, "I have found an old man's cloak," we may mean 
the cloak of an old man, or an old cloak such as men wear, as 
distinguished from such a cloak as women wear^ but even in 
this extreme case confusion can be obviated in our flexible 
language by a different arrangement of the sentence, and in 
nine cases out of ten the adjective would by its meaning indi- 
cate the substantive which it qualified. If we speak of a "blue 
sailor's jacket," the word blue can only refer to jacket, as such 
a thing as a blue sailor would be absurd, and we know that 
sailors habitually wear blue jackets. Such a language as ours is 
worthy of a civilized and enlightened race, because by its very 
absence of forms it assumes that those who use it are people of 
intelligence and do not require to have their minds guided to 
the meaning by the leading strings of synthetical forms. A 
Roman required this aid. To him "caerulea nautae tunica" was a 
diflferent thing from '^caerulei nautae tunica;" and had he been 
unprovided with the help supplied by the variation of the final 
letter of the adjective, he would have been at a loss what to 
understand. Thus we may say that synthetical languages are 
fitted for the childhood of the human race, analytical ones for 
their manhood. 

Bengali and Oriya do not change the form of the adjective at 

Digitized by VjOOQ iC 


ally whether for gender or case; the adjective is placed just 
before the substantive^ and one case-ending does for both. 
Hindi gives to those adjectives which end in d, a feminine in I, 
and an oblique ^gular in e^ but does not make lise of the 
oblique form of the plural. Thus one would say ^51% ^t% ^ 
"of a black horse," and not ^jfNI vH^f ^ "of black horses," but 
'RTW Ml><t IT. The reason of this is obvious. The adjective 
having been put into the oblique form, common sense shows that 
it must refer to the substantive in the oblique form, and there is 
no need for a closer method of indication. The speaker is 
supposed to be able to use his wits to this small extent. Panjabi> 
however, is conscious that its speakers' wits are not sharp enough 
to be trusted, and the adjective is therefore put through aU four 
forms in each gender; as fffifNit WRff ^ "of low castes" (/.), 
t^rfZ^ ^n^li ^ "of white horses" (w.). This gives a clumsy- 
appearance to the language, and ought not to be necessary for 

The Qiijarati adjective has aU three genders with the typical 
terminations, lit masc., fj fem., and ^ neut. The feminine 
remains imchanged for number and case, except that it option- 
ally adds to the plural the universal o. The masculine forms its 
oblique in Ay like substantives of the o-stem, and like them has 
the locative and instrumental in e, which ending is here also 
allowed to pass over into the ablative; the plural oblique is the 
same as the singular, giving in consequence a type quite 
analogous to its parent Hindi. The neuter differs from the 
masc. only in adding anusw&ra to the nom. plural. The range 
of forms may be thus drawn out: 

Nom. Sing. Masc. ^tO "good," Fem. ^TPC^ Neut. ^9T^. 

Nom. PL „ ^TTT » » ^BT^ »» ^TTT- 

®^"8^- 1 nw -frrr^ f ^TTft and ) 


"Obi. „ ^TTTT M „ |„.jrvA | »» ^TTT- 

The adjective is thus precisely similar to the substantive of 
VOL. n. 16 

Digitized by LjOOQ IC 


the o-stem thronghoat; and from tlie position whicli Gnjarati 
holds in respect of Hindi, as well as from the analogy of 
Panjabi, it may legitimately be inferred that Hindi itself 
formerly had all these forms, though it has now got rid of some 
of them. Adjectives ending in a consonant do not imdergo any 

Sindhi declines ite adjectives in the same manner as the 
substantives of corresponding terminations. It follows that 
adjectives agree in gender, number and case with their substan- 
tives, with the exception of that somewhat numerous class of 
words of Arabic and Persian origin which are not liable to 
change. Although the genius of Sindhi requires that every 
word should end in a vowel, either pure or nasaUzed, and in 
consequence words derived from Persian ending in a consonant 
have the short vowel u added to them, as lOT frran c-jy^, tj^- 
?rnj from ^U-juj , yet it seems to have be^i thorougUy imder- 
stood and clearly felt that this final u was something different 
from the final u in pur6 Sindhi words, and it has therefore not 
been subjected to the changes which the rules of the language 
exact from indigenous words. The same reason has apparently 
protected Arabic words, whose un of the nominative, the tamoin 
of grammarians, had probably died out of the popular speech 
of the Arabs before the period of their conquest of Siadh, so 
that the words at that time were as perfectly consonantally- 
ending as Persian words, and had the u given them by the 
people of Sindh. 

In Marathi the adjective takes, as in the other languages, the 
typical endings, "^masc, "|jfem.,*|[ neuter, for the singular, and 
H masc., JICl fem., ^ neuter, for the plural.. This type of adjec- 
tive is the only one that changes for gender: all others, whether 
ending in vowels or consonants, remain unchanged. As usual 
with this language, there is some indecision in the forms used for 
the oblique. In the majority of instances the adjective merely takes 
i^e form yd^ in analogy to those substantives in d which insert 

Digitized by VjOOQIC 


the semiyowel in the oblique, and whicli grammarians call the 
sixth declension. The adjective may, howeyer, take the form e, 
which is also used by substantives in 4 as the ejiding of the nom. 
plural. Thus they say H^IT ^fTHreTOT or WT ^RPPTWr "to a 
good man/* and this pair of endings is used also with fern, and 
neuter substantives. Adjectives ending in any other vowel or 
in a consonant do not change at all. 

It will thus be seen that there is in all the languages a 
tendency to deprive the adjective of its full range of terminations 
— a tendency based on the rational principle that such an elaborate 
variety of endings as the adjective is capable of affording, if ex- 
panded to its full limits, is quite unnecessary for clearness, and 
may therefore be dispensed with. In this respect even the 
Marathi, usually so prodigal of forms, has allowed itself to be 
influenced by practical considerations. Only the imcultivated 
Sindhi still retains all this useless apparatus, for which, perhaps, 
among other similar perfections. Dr. Trumpp would call upon its 
sister-dialects to envy it ! 

§ 54. The numerals, whose elegfiint and symmetrical develop- 
ment from the ponderous Sanskrit compounds has been exhibited 
in § 26, are simple in their declension, taking generally the 
usual signs of plurality and case when required, but inasmuch 
as they are strictly adjectives, no^ oftai requiring any such signs. 
In B. and O. they possess no peculiarity; in H. there are only one 
or two points requiring notice. "When special emphasis is re- 
quired, or a numeral is used with reference to scone object pre- 
viously discussed, the plural form is used ; but in this case it is 
the obUque form in ^ that is used, none of the nonunative 
forms taken by the noun being customary. Thus we say 
?fV^ *n^ ^t% " the three brothers said,'* alluding to some trio 
of brothers well known to the speaker. The use of the oblique 
form is only another instance of the tendency of common forms 
to extend beyond their proper and original sphere into all and 

Digitized by LjOOQ IC 


every other as occasion may serve. The forms of the nom. 
plural of the substantive vary, but the oblique plural ends 
invariably in ^, and this form therefore has acquired an 
ascendancy, and come to be regarded as the common and un- 
mistakable note of plurality. As no numeral in Hindi ends in 
long d, there is consequently no opening for any change for the 
oblique singular, and we hear i;;^ ^ " of one," i(t ^ " to two," 
Vr^J^ % " with fifty," and the like. The habit of not using 
any grammatical forms which are not absolutely necessary has 
led to the .almost xmiversal rejection of aU signs of plurality in 
a noun constructed with a numeral. If we say "fifty hauses," 
it is evident that "houses" more than one are meant; what 
need is there to put the noun in the pluralP X|^T9 ^TT*^ "in 
fifty houses," is quite intelligible without any plural sign. It 
is thus that language gradually simplifies itself by the aid of a 
few natural reflections. 

In Fanjabi it is usual to add ike plural sign to a numeral, as 
^^ if^pgrt ^ " of ten men," but this only in the oblique cases, 
and it would not be incorrect to omit it. Gujarati follows the 
Hindi customs in this respect. 

All three languages, H. P. and G., treat their ordinals as 
common adjectives of the oxytone type. Hindi adds the 
terminations — 

Masc. Nom. ^, Fcm. Nom. ) ^^ 

„ Obi. •^, „ Obi. ) ^' 

Panjabi the same ; but it, as well as H. in many local dialects, 
still preserves the older ending in ift, ^j"^, which is derived 
from the Sanskrit ordinals in if: Whence comes the anusw&ra 
I am imable to say. Gujarati knows only the simple ending in 
fn. ^,/. ift, n. ^ ; pi. m. ^TT, ^, / ^, n. J(f or ^. The anu- 
8wd,ra in H. and P. is probably quite a modem addition, and 
begins only at the "fifth ;" the first four, and the sixth, are 
regular (see § 27). 

Digitized by VjOOQ IC 


Marathi declines its cardinal numerals regularly after the 
manner of the substantives of similar form; thus ipi "one," 
forms its oblique by adding d, as inTT, after the model of masc. 
and neut. nouns in mute a (§ 41, cl. 1). In the feminine the 
oblique is formed by the addition of ^, as ipft, like ^int, obi. 
^Unft (§ 41, cl. 3). But the numerals ^ "two," iftif "three," 
and ^TT "four," take a new form in declension, with crude stems 
4^|f(, f?f^> xH^, which resemble the plural forms such as 
^It(> ^StUfy in nse in Hindi, as mentioned above. When they 
refer to persons, another set of forms is used, produced by the 
addition of % to the shortened bases ^, fif and ^, thus making 

M. PL Norn, ^t^, Fern. Norn. ^vH* M. Obi. ^M(> F. Obi. ^vflf. 

This process does not extend beyond the number " four," but 
is parallel to the practice, in the Bhojpuri dialect of Hindi, of 
adding ift to a number when it stands alone ; thus they say ^jH 
"two," Hlvl^n "three." And in Oriya the longer form 4\\^[ is in 
use; thus ^iftZTH "one (thing)," ^ iftTT "two (things)." In 
all Indian languages the practice of adding some word meaning 
"piece, portion," etc., to numerals standing alone, is common. 
The Marathi here differs from the rest by restricting the particle 
to persons. In Oriya jftZl means a "piece," and so also does 
the curtailed Bhojpuri form jft; the derivation is perhaps from 
Skr. ^ft'^ra "family," in the general sense of a class or collection 
of persons or things. 

Sindhi treats aU its cardinals as plurals, and, except in one or 
two cases, inflects them as nouns with a regular oblique. Thus 

^ " two," PI. Obi. f^rf*r. 

^"tbree," „ f^. 

^Tf?^ "four," „ t^i^fii. 

where the ''^ of the nominative is dropped, as it is in the form 

Digitized by LjOOQ IC 

^46 DfiCLlSKSlOK. 

^ used in compounds. In these three numerals the form of 
the oblique in ^[fi| is used ; in the rest that in l|^, as— 

^ "fivtf," PL Obi. ^«rf8r. 

As the words "two" and "four" have no neuter plural in ^ 
in Skr., it is probable that the Sindhi forms have been extended 
to these words from fsjfi|, Skr. 4)lU|> gen. ^i|lU|(, Prakrit finf. 
The Pr. oblique of " four" is ^^ or ^^3*^, from which MtJPi 
may have arisen, though we can hardly derive f^rf'l from Pr. 
^ft- It is more probable that the ordinary termination of the 
oblique plural of nouns has here been extended to the numerals 
than that these words have been separately formed from Prakrit 

The series of tens from 11 to 18, which end in ^, drop the 
anusw&ra as well as the a which precedes the ^ ; thus — 

Similar to this is the Hindi dialectic pronunciation ^TTlf 
"twelves," from ^TTTf "twelve." The numerals in 1[ and Tj, as 
'^(fi "sixty," and ^ "ninety," are not inflected; ^RfV "eighty," 
however, takes an oblique ^VflfV- When we get among the 
higher numerals, much of the sense of plurality is lost, and 
singular nouns are often in all the languages constructed with 

The word W "one hundred," is in all the languages treated 
as if it Were a substantive, and in Sindhi has a regular declen- 
sion, as follows: 

Nom. Sing. ^ETS* Nom. PI. ^pB( or ^. 
„ Obi. ^^, „ Obi. W^y ^rrf'Tj ^y Wl* 

The numerals, however, often remain umnflected before a 
substantive, which saves a good deal of trouble. 

Just as in Hindi the numerals, when treated as collectives, 
take the oblique plural, as ^Iff ^^t^TI^ "scores and scores 

Digitized by VjOOQ IC 


of cartB," Wntf Wret ftnrf^ "soldiers by laJdhcs,** so also in 
Sindlii they take the regular plural ending in ^ of fern, stems 
in ^ (see § 42); thus ^ "ten," makes T$, as 5'in^ "three 
tens," and the oblique in ^pf, as ^||\ ^3Vf^ "four tens," 
and the like. This attribution of the feminina is due to the 
abstract character of the collectives, just as in Greek Tptd<: 
"a triad," efa? "a hexad," Be/cas "a decad," are feminine; but, 
as far as I am aware, no such change takes place in the other 
languages. Gujarati expresses collectives by adding ^, as in 
Sanskrit. Thus G. Z^ "a triad," Skr. ^, G. ^rtl, Skr. 
^W^, and so on; but it also adds the neuter sign ^ in super- 
session of ^, as in ?JT^ "a heptad," Skr. ^Tff^? GI-. ^in^ "an 
octad,*' Skr. ^^. In this respect it appears to stand alone. 

Sindhi often adds the emphatic I or M, as do also the other 
languages, to all cardinal numbers, to denote completeness; but 
Sindhi sometimes incorporates this emphatic syllable into the 
numeral, and adds the terminations to the word thus formed, as 

^ « aU three," Obi. f^pft, fff W*^ • 

^^rnd:" all four," „ ^^ijyit. 

The ordinal numbers in all the languages are regularly 
inflecfed as adjectives in d: thus H. m. ^HK\y /• T^^' ^^^ 
sin^. ^^Kf obLpL ^XS- But Sindhi has added an anusw&ra 
to all its ordinals. The usual inflection of the adjective in lit> 
fern- "Ij, must therefore be supplemented in all its forms by an 
anusw&ra superadded. Trumpp would derive this anusw&ra 
from the ^ of the Skr. termination K^\ but this seems opposed 
analogy of the other languages, in which the m of ?R is 
preserved, or softened into ^, with the anuswSra still existing, 
as in H. "^sfK ^. It is, however, in accordance with the general 
analogy, that the H should be elided in Prakrit words, and ^ 
softened to anusw&ra, and we should probably concede to Sindhi 
greater regularity than to the^ others, and suppose that it is H. 

Digitized by VjOOQ IC 


and P. whicli are irregular. In tlie following examples Sindhi 
certainly answers to Trumpp's derivation : — 

8KB. PB. 8. H. 

Fifth 1^^ M-^^n ^^ but MMqt. 

Seventh ^TTW ¥Wfft ^Rlt » ^TTTPrt. 
On the other hand. 

Eleventh ipRT^ TBETT^Rft ^HT^f „ T^^TTTFTt- 

Twentieth fihX ^fV^^ T^t » ^ft^Rt- 

In these two latter there is in Skr. at least no ?|IT to start 
from, and though Prakrit has IQRTT^R^ which would give 
^ITff , yet it has preferentially ^fWft, which will not account for 
the anusw&ra of ^ftft- It is true Pr. has also 4^^10 1 *1, which 
would lead to 4)t{l^; hut the Sindhi form of this would 
prohably be ^ff^^, not ^t^, unless we suppose an elision of H 
and the throwing forward of the anuswftra. In Panjabi, how- 
ever, we have both the If and the anuswS-ra standing together, 
as in ^rRrt "seventh," so that the latter certainly cannot liere 
be a relic of the former. Chand's ordinals are as follows: 
1st inER. 3rd^TfV. 6th li^?T- 7th ^ct- 

2nd^^. 4th ^. 6th ^. 8th '^. 

Pr. R. i. 5. 
in which, while the final ^ is sometimes written as a consonant, 
and at others as an anuswd.ra, there is nevertheless nothing to 
indicate the source whence the modem terminations arose. The 
only way that they can be accoimted for is by supposing redupli- 
cation of the ^. Further research is required on this point. 

§ 55. We now arrive at the most interesting and important 
section of our whole inquiry into the noun, namely, the origin 
of the case-affixes. Having rejected the synthetical method as 
a whole, and retaining merely certain half-efiPaced traces of 
declension, our languages have had to betake themselves, like 
their European relatives, to added particles, in order to draw out 

Digitized by VjOOQ IC 


and express fully the yarious relations of the noun. These 
particles are placed after the noun, in contrast to the European 
method, which places them before it; but this practice is in con- 
sonance with the order followed by the mind of an Indian speaker, 
who constructs his sentences always in a sequence directly the 
reverse of that used in the languages of Europe, so that, in trans- 
lating from an Indian yemacular, one has always to begin at the 
end of a sentence, and work backwards.^ A great deal has been 
written on this particular branch of my subject, but for the most 
in a desultory and inconclusive manner ; and I hope, therefore, 
to be able to put together, in something like order and arrange- 
ment, the results of the inquiries of others, as well as my own 
discoveries and beliefs, so that, if the question cannot at once be 
settled, it may at least assume a more concrete and manageable 

It m^y be assumed as a starting-point, that the case-affixes are 
remnants of noims or perhaps pronouns, which have been cut 
down and worn away by use. I think it will be admitted by 
all philologists that any other assumption would be irreconcile- 
able, not only with the fundamental principles of modern Aryan 
glossology, but with the universal laws of language. In the wide 
field of Indo-European comparative philology, the great master 
Bopphas conclusively proved that this principle everywhere pre- 

^ The difficulty of following the sense of a long passage in a yemacular document 
which this practice induces, will have struck many of my brother magistrates in 
India. Take the following, from a Police report : " To day at 1 watch of the day 
A. B. watchman of Tillage G. brought to the police-station a brass drinking yessel 
and platter which he found during his rounds last night on the edge of the public 
road leading from D. to E. at the south-western comer of the pond lately repaired by 
the Manager of the estate of Baja F" This would run in Bengali thus: <* To-day of 
the day one watch of village G. watchman A. B. of Baja F. of the estate by the 
manager lately repaired of the pond at the south-western comer from D. to E. leading 
of the public road at the edge by him last night his rounds during found of brass a 
drinking yessel and platter to the police station haying brought, etc." All the yerbs 
throughout the report are in the conjunctiye participle ^^ haying done," and at the end 
of perhaps the fourth page one comes at last to the only finite yerb in the whole, ** I 
haye reported it" ! 

Digitized by VjOOQ IC 


vails, and that eyen the synthetical case-endings of the early- 
classical lang^agej9 are relics of independent words. It is there- 
fore safe and rational to assume that in the languages of whidi we 
are treating, allied as they are closely and indissolubly with the 
old mother-speech Sanskrit, the same sentim^it exists, and the 
same method of word-building still survives. Throughout the 
material world we see that the process of reproduction is one of 
such a nature that it can be repeated time after time f<^ ever. 
Man begets man throughout the ages, and tree produces tree; 
the mountains are washed down into the sea, and the forces at 
work in the bowels of our planet upheave fresh moimtains, which 
are in their turn washed away. So also in language, words 
originally independent are seized and bound into slavery to 
other words, become case-affixes, are incorporated into case- 
endings, and are finally abraded altogether. Then the mind 
seizes fresh words, and binds them into slavery again, till they 
also wear out by use; and, if the world lasts long enough, will in 
their turn pass into case-endings and disappear, and a third set 
will have to be captured and miade use of. The process repeats 
itself, and the modem Indians, when they had recourse to the 
words which have become the case-affixes of to-day, only did 
what their remote ancestors had done before them, when they 
took pronouns and nouns and made them into the terminaticma 
which Sanskrit literature has preserved to us, such as -e«a, Aya^ 
a%yay and At. 

Literature, however, has a tendency to arrest the process of 
change; and the modem languages of Aryan India are so rapidly 
becoming cultivated literary tongues, that we may suppose that 
they will not in future develope so quickly as they did in former 
times. The literature which they possessed before the advent of 
the English schoolmaster was not of a kind to influence greatly 
the spoken language, but rather held itself proudly apart, and 
looked down on the folk-speech. Even in the present day this 
silly feeling is strong. A generally sensible writer like BankimL / 

Digitized by VjOOQ IC 


Ohandi^ the editor of that excellent Bengali magazine the 
''Bangadarsana/' for instance, in writing a serial novel, puts 
into the mouth of one of his characters the familiar word "diyft- 
siMi" (meaning a match for lighting a candle) ; but in the very 
next line, when writing in his own person, uses the highflown 
Sanskrit equivalent "dlpasil&ka,*^ though he knows perfectly 
well that for a thousand Bengalis who understand the former, 
not ten would know the latter word. It is to be hoped that this 
sort of nonsense has had its day, and that in all the seven 
languages literature will by degrees become more natural, and 
that men will begin to see that there is no disgrace in writing 
as they talk. 

But this, is a digression. To return to our subject. It follows 
from what has been said above that we must look for the origin 
of the case-affixes in nouns of the older language. It follows 
also that the nouns in question must have been in use at the 
period when the modem languages began to be formed, — ^in other 
words, they must have been words of the lower and more popular 
dialects of Prakrit. We should hardly be justified in looking 
for them in scenic Prakrit, but rather in Apabhransa. The 
argument used by scholars in Europe, that the dialect of the 
plays and of HSla's songs must have been a spoken dialect, 
because players and dancing-girls could not have used a 
language which their audience did not understand, has in 
reality very little weight to the mind of one who has lived long 
in India. It is a curious but quite undeniable fact, that dancing- 
girls do in the present day sing many songs which only the 
educated portion of their hearers can imderstand, or, if the 
humbler and more illiterate part of the audience do understand 
them at all, they do so, not because the language is that which 
they themselves speak, but because it is fine talk, such as they 
hear their betters use. In an Indian language there are always 
three or four shades or strata of talk existing side by side at the 
same epoch. Thus there is in the Bengali of to-day the highly 

Digitized by VjOOQ iC 


Sanskritized style of the pandit^ the somewliat artificial, but less 
Sanskritized style of gentlemen of education and refinement, the 
practical every-day speech of the middle classes, which contains 
only the simpler Sanskrit words, the strange jargon of the 
women, and the rough homely patois of the peasantry. It is 
quite possible for a foreigner to know one of these languages, 
or strata of language, without knowing the other. It often 
happens that the English indigo- or tea-planter, mixing only 
with the lower classes, speaks with fluency the peasant speech, 
while the high official speaks equally well the dialect of the 
educated ; and the planter cannot talk to a native gentleman in 
the habitual dialect of the class, nor can the official understand 
the peasant without an interpreter. These things are so now, 
and they probably were so a thousand years ago, and, for aught 
we know, will be so a thousand years hence ; and we are therefore 
justified by experience and analogy in looking to the lower or 
Apabhransa dialects for the origin of modem forms — ^all the 
jargon of Hala and the plays notwithstanding, I shall now 
proceed to exhibit the results of such investigations as have up 
to the present time been made by myself and others, taking each 
case-affix separately. 

§ 56. The Objective. Under this head are classed the 
accusative and dative of the Sanskrit. Prakrit had already lost 
its dative, and the modem languages make, strictly speaking, 
no distinction between the nominative and accusative. There 
remains therefore no way of designating that form which the 
noun takes when it is the object of an action but that of 
"objective.^* Marathi alone has a distinct quasi-synthetical 
dative, which, as we have shown above (§ 52), is a descendant of 
the Sanskrit genitive in asya} In Marathi, however, as well as 

^ The Gipsy language has also an objediye in dsy wluch is nsed only with mascu- 
line nouns and in the singular; thus Eom '^a gipsy/' obj. rom48\ rat "a lord," obj. 
ra»V#; raklo (H. lafkd inverted) "a boy," obj. raklds. — ^Paspati, p. 50. The same 
form occurs in Eashmiri, as mdul "father," obj. mdlis; nichu "child," obj. niehavis. 

Digitized by VjOOQ IC 


in the other languages, there is no sign to mark the object. In 
H. ^(^fir^ IRTT "the house fell," we have the same form as in 
^IT '<^«fl^l "he bmlt the house/' When it is thought necessary 
to emphasize or express the state of objectivity — a question the 
discussion of which pertains rather to syntax than to formlore, — 
the following particles are added to the oblique form of the 
noun: — 

Hindi ko. Marathi Id, 

Panjabi nus. Bengali ..... ke. 

Sindhi khe, Oriya ku. 

Gujarat! neu- Nepali W. 

These affixes are the same both in the singular and plural. Of 
the seven languages H. S. B. and 0. fall into one group, P. M. 
and 0-. into another. 

The H. affix ko is softened from an older form ^, which is 
stni in use in the Braj Bh&shS, and many other rustic dialects. 
An older form still is ^ ; and the oldest form of all is that 
found in Chand, iRf . The form ^, which Trumpp^ hastily 
affiliates to those given above, I hesitate to connect with them, 
believing it to belong to a different root altogether. What, 
then, is the origin of this affix P It is derived clearly from some 
noun, as I have already shown reason for believing ; and the 
problem is — ^What is that noim P 

Of the use of ^ it is unnecessary to give examples, as the 
form is in common daily employ. Similarly, ^ is to be met 
with on every page of a Braj Bh&sh& poem, m^ is also very 
common. , Thus, in the Eamayan of Tulsi Das — 

^f^ tIrfM ^hrftr 'f^ ^fT^ II 

Ayodhya KSnda, 7. 
"Prosperity, success, wealth, (like) fair rivers 
Overflowing, came to the sea of Avadh." 

* Sindhi Gr. p. 116. The author also errs, in my opinion, in considering the anu-. 
swira in <^ a modem addition. It would rather seem to be the older form of the two. 

Digitized by LjOOQ IC 


"King, give tlie heir-apparentship to Ram." \ 

Sundara K^dft, 8. 
" Thus having said, bowing his head to all, 

He went rejoicing, holding in his heart Eaghunath.'^ 

It occurs on an average ten times on a page throughout the 
poem, and in some of the recent lithographed editions is regularly 
written «nf — a point to which I shall refer again presently. 
Sur Das uses ^, #f , and ^ ; but not, as far as I know, ^ or 
^U^. It is impossible to speak with certainty, as it would take 
half a life-time to read through the vast ocean of the Siir 8&gar. 

Chand uses several forms, but the anusw&ra is retained in all, 
thus shovring that it is not a mere modem addition. Hjs forms 
are ^, ^, and «|jf. We cannot tell how far Chand's forms 
have been modernized by copyists ; but that ^|jf is a true f Opn, 
and not a copyist's error, is evident from the fact that in t^ 
places where it occurs, it is demanded by the metre, and occa-\ 
sionally appears vdth the final vowel lengthened in cases where 
a long vowel is wanted. Instances of both ^|jf and uni^ are — 

" He seeks one of you." — ^i. 88, 9. 

irni ir^ ^ ^^f ^ II 

" At monung-time a gift to the Brahmins 
Dividing with his own hands gave." — ^vii. 5, 3-4. 

This latter would run thus in Modem-Hindi — ^UTTT ^ 'B^RI ^ 

ift ^n^^ 'WT ^ II 

"Having made obeisance to all." — vi. 38, 2. 

Digitized by LjOOQ IC 



"Having thas said, you find (».*. cause) fear to me." — ^i. 160, 1. 

''For the war with Prithiraj at Mahoba, 
Parimal has summoned us." — xxi. 84, 6. 

Of earlier forms than ^|^ we have no examples, and the gap 
must be filled by inference. Where actual evidence fails, we do 
not, as some carping critics say, fall back upon mere guesswork, 
but, applying the known laws of phonetic development, endea- 
vour to reconstruct an older form, and are not guessing any 
more than Owen guessed when he reconstructed an extinct 
animal from one bone. "We cannot do this ; but it may be 
possible to get to a step or two further back by analogy. It is 
admitted that initial letters seldom change (see Vol. I. p. 190) ; 
we therefore assume on good grounds that the ^ of this word 
has come down imchanged. Further, ^ generally results from 
the dropping of the organic portion of an aspirated letter; arid 
^ as a termination represents the Sanskrit termination of the 
neuter in 'am. 

The following theories have been advanced, based on the above, 
and similar well-known phonetic processes. Trumpp (Sindhi Gr. 
p. 115) derives from Skr. ^, which he supposes took the form 
fi|?|, and he thus elides ^, which, on going out, aspirates the pre- 
ceding consonant, thus producing the Sindhi % ; the Hindi ijt 
he derives from Br?f , through Prakrit forms f^Rft> ftwt* and 
ijit. This theory fails, as it- has been shown above that the 
older forms contain an ^ and anusw&ra, neither of which could 
be got from Trumpp's process, and there is no reason to suppose 
that ijit has a different origin from ^|^. To accoimt for the f , 
Trumpp supposes that the ^ of ^ on going out aspirated the 
following consonant, thus producing msf. That m?t would 
naturally produce ifilf , ii^, and perhaps even the allied form 

Digitized by VjOOQ IC 


itif, is imdeniable ; the only difficulty is in getting to ^ff from 
IRt. There is another way by which, still retaining fRt as the 
origin, we may account for the if, namely, by supposing that 
when irt had become, as we know it does in Prakrit, W^ and 
1^, the next step was to elide the ^ (Vol. I. p. 202), thus 
getting m^ ; but the place of a single consonant which has 
been lost by elision is often supplied by f in Prakrit,^ and by 
this process a form m^ is obtainable. 

The weak side of all this argument is that ig^ does not com- 
mend itself as a probable origin for an affix meaning primarily 
and generally " towards. '* Trumpp says that in Skr. igU is used 
as an equivalent of ^ in the sense of " for the sake of," " on 
accoimt of," " as regards." But even if this statement be true 
of classical Sanskrit, it is doubtful whether the use of Ig^ in 
this way was ever sufficiently common among the lower classes 
to have given rise to so very common an affix as the ijit of the 
modern objective. It is difficult to see how a word primarily 
meaning *' that which has been done " could come to mean " on 
account of." With ^ and ^^ there is no such difficulty, 
because these words mean respectively " in that which has been 
done," and "by that which has been done;" and the transition 
from these senses to that of "on accoimt of" is easy, resting as 
it does more on the acknowledged meaning of the locative and 
instrumental cases than on the root. 

The Sindhi objective affix % khe, can, without doing violence 
to probability, be admitted as an offspring of ^ ; but the mind 
is not easily satisfied with the parallel affiliation of ijit to ^cf. 

As far as concerns the meaning, a more probable origin for ^ 
is that partially suggested by Hoemle (J. A.S.B. 1872, pt. i. p. 
174, et seqq.), who, however, has not traced the steps of the trans- 

1 Weber, Saptasatakam, p. 29. " Es scheint vielmehr das A ill diesen Fallen 
nur als eine Art spiritus lenis zui Yermeidung des Hiatus gebraucht zu sein, ahnlich 
wie im Mdgadhi der Jaina das i/, in einigen Fallen aber aucb h selbst oder v 
verwendet wird (Bbagayati, i. 399, 409, 411, 426)." 

Digitized by VjOOQ IC 


ition correctly. The meaning of ^ being evidently, in the first 
instance, " towards,'' " near," and the like, we are led to look at 
the analogy of Bengali, in which language 4]^ means " near," 
and is used with a genitive as an independent postposition, as 
TTT^TT ^$ " i^^r him," or " in his possession," or " to him." 
Now this word ^$ is from the Skr. 7^, locative of ^i^, and 
means literally " in the armpit," or, as we should say, " at the 
side.*" <g[^ becomes in Old-Hindi WRT^ and the accusative ^rtf 
would become first W^, then ^"Nl. As ^, like the rest of the 
aspirates, migrates into if, a form m[^ is legitimately presum- 
able; whence, by shortening the vowel, we get the already 
established form i|^, with its variant ^|#. I confess that this 
derivation approves itself to my mind in preference to any other. 
m^, as a substantive, meaning ^' armpit," is in common daily 
use in the present day. It is a Tadbhava of somewhat later 
origin than those very early Tadbhavas which have given us 
the adverbs and case-affixes, and it is therefore no objection to 
this derivation that the case-affix should have tmdergone more 
change than the noun. As a parallel instance may be cited the 
adverbs like ^5ft, Wft, where ^ is all that remains of the ^ of 
'4|||«f, which word, when used as a noun, has kept the fuller con- 
sonantal form ^TR. When used as a noun, the word ^i^ loses 
its final inherent a, and becomes a barytone monosyllable; but 
the affix comes from the accusative, which is used adverbially, 
and consequently retains its anusw&ra. ^IR^ = ^im^> ^, is 
equivalent to ^I^ = ^ and ^^Tf! = TF^. 

HT^ actually occurs in a place where the metre requires a 
long vowel, in Tulsi Das's Ramayan, Ayodhy& K&nda 330: — 

W^ tI?! ^^w ^r^ ^R ^rr^ I 

**In writing the moon Rahu has been written, 
Fate is always crooked to all," 

VOL. II. 17 

Digitized by LjOOQ IC 


unless we here take nrrj without anusw&ra to mean "why;" 
but this is strained, and does not accoxmt for '^^. 

In confirmation of this derivation as regards the meaning, a 
matter in which Trumpp's derivation fails to satisfy, it may be 
further added that in modem Urdu the same method is still 
adopted, though with a diflferent word. The Arabic word Jx) 
"armpit,'' is always used in the sense of "near," "dose by," 
"to;" thus Jm ^^"near the house." 

Hoemle is, I think, wrong in saying that ijit is derived from 
Bengali ^KtI^, as it is impossible to derive a twelfth-century 
Hindi word from a fifteenth-century Bengali one, and Hindi is 
the parent, or at least the elder sister, not the daughter, of 
Bengali; but Hoemle has certainly, in my opinion, indicated the 
direction in which we should look for the origin of the word, 
and I believe the steps were as I have shown above; and further 
research will probably establish the intermediate forms for 
which we have at present no actual proof. A striking analogy 
to the assumed genesis of iRjf from i|n^, is afforded by the old 
H. and B. affix X|3f "beside," "to," which arises from Skr. xn, 
from tl^ "a side," through tn^ and VX^, which last form is 
actually used by the early poets. 

The Oriya if may be considered as a mere variant of the 
Hindi ijit, which itself, in the corrupt dialect of Hindustani 
spoken in Southern India, sounds w . The Oriyas do not pro- 
nounce the "^ at all f uUy, but give it a soft short soimd, 
which it is very difficult to distinguish from ^. The Telugus, 
the near neighbours of the Oriyas, with whom they have for 
centuries had close intercourse, also mark the accusative by a 
form ^. Telugu, however, being a Dravidian language, is not 
in any way connected with the languages of our group, though, 
as it has received a large number of Sanskrit and Prakrit words, 
there are often great similarities between it and Oriya, and it is 
singular that the structure of the verb also is very similar. Dr. 
Caldwell (Dravidian Comp. Gr., p. 225) asserts the identity of 

Digitized by VjOOQ IC 


the two forms ; but we have already traced the Hindi ko to an 
earKer form ?|^, which, whatever be its origin, is distinctly an 
Aryan word, and has no connexion with the Dravidian afi&x. 
The Telugu afi&x is as often sounded hi as kUy the vowel varying 
according to the final vowel of the noun to which it is afi&xed; 
thus Hari makes Hari-ki "to Hari," but dora "a lord,'* dora-ku. 
Bengali ordinarily in its earliest recorded forms, in the 
Vaishnava hymns, does not mark the objective any more than 
early Hindi does. When, however, it is necessary to indicate 
this case, we generally find it noted by the affix*^, which still 
survives in the so-called dative of the modem pronoun. In 
this early stage we also occasionally find the Hindi crude affix hi. 
Instances of "^ are common in the Bengali Mah&bh&rat of 
K&sir&m Dks — a book much beloved and bethumbed by the 
ordinary Bengali shopkeeper. Thus ^^ng% ^ftT <|^| ^BRffW 
H^T M "sitting at the root of a tree, the king said to Bhima." 
—414, a, 5} ^^Mi^ <*R«ilX Wl^ hO^K II "to drink (Kt. to 
the making of a drink) water he goes to the lake." — 415, b. 25. 
MlPimi ^r^ WRT ^if^^f^ ^sntr ll "taking a water- vessel goes 
to fetch (lit. to the bringing) water." — ib. 40. H^ in^ TS(t 
^r$^ ^rra ^lUi^dl II "Feeling fear the virtuous woman invokes 
Sri Krishna." — 416, a, 2. These examples have been taken 
from the first page that I happened to open ; many hundreds of 
others may be culled from Bengali poems. It will be seen that 
in affixing this X[ to nouns ending in mute a, an e is inserted; thus 
^rS^, instead of IPSI!!^. This practice results probably from 
the addition of the emphatic e to the singular nom., instances of 
which are common ; thus Bidy&pati ^apf g^ ^I^TT'^ II "<>f 
such a kind is thy love." — Fad. 984. U ^^^ ififf 'Rfft ^I^- 
^tit tl «irn5 ^ ''rfH ^IRI^ ^?nfl% ll "Ah! dearest, why dost thou 
question? (lit. make the question). Even now thou shalt make the 

^ Calcutta edition, published by Mabeslichandra Ghose, 317, Chitpore Road, 18721 
The first number is the page, the letters a and b denote the column (there are two in 
each page), the last number is the line. 

Digitized by LjOOQ IC 


feast of love with Kdnli." — lb. This fonn is really the nom. 
plural; and when used in the singular, thus naturally carries the 
idea of abundance or emphasis — an idea, however, which soon 
faded, leaving the poets free to use the e pleonastically when- 
ever it suited their metre. 

The explanation of the origin of this form is, according to 
Trumpp, with whom I agree, to be sought for in connexion 
with that of Marathi, Nepali and the other languages using an 
aflix whose characteristic is ^, to which we shall come presently. 

The modem Bengali uses instead of "^ the form %. Now 
Bengali is very prone to softening A into e ; in ordinary con- 
versation one constantly hears this pronunciation (Vol. I. p. 142). 
Thus IS^rf^RTT f^Wn^ "I have abandoned," will most usually be 
so uttered as to sound chhMy6 diUm. It may therefore be 
safely assumed that the afiBx ^ is changed from an earlier i|rr> 
shortened from iRTir, which we have above taken as the origin 
of i|#, ^, and ^. 

Passing from the group whose characteristic is ^, we come to 
that whose characteristic is ^, of which the most important 
member is Marathi. In that language the dative affix is Wly 
and in the earlier language ^TRpf and WT^OT, allied to which is 
the Nepali ^TTI^, obtained by elision of the l{, whereas Marathi 
has rejected the whole of the last syllable. 

From the root ^^ there is a large host of derivative forms, 
both verbs and particles, in all the seven languages; and as the 
root means primarily "to stick," it would naturally be adopted to 
express the idea of nearness or adhesion, which is very suitable 
to the objective. The simplest form in which it occurs is 
perhaps Hindi ^5R, meaning "up to," "as far as;" but owing 
to the facility with which ^ and ^ interchange, Hindi has in 
some rustic dialects an affix if or 5}, originally*^, used with 
the dative. This stiU survives in the neighbourhood of Delhi, 
as, for instance, in the proverb quoted by Elliot (Races of 
N.W.P., vol. i. p. 6):— 

Digitized by VjOOQ IC 


^R^ ^ti^ ^Rni ^t! ftf^*^ ^^ ^TfT 

**What they bow they eat themselves, and give not a grain to the 


It also survives in G., which marks the objective by*if . The 
form % is found in Chand: — 

f'rf^ ^M1«n^ f^ II — iii. 3. 
**In his youth to Prithiraj, 

In a dream at night (came) a sign." 

^^n gimt ff?r 8— v. 13. 

** Hearing it, Prithiraj himself 
Invited the prince kindly." 

This is, I now think, the correct translation, though I formerly 
translated the passage otherwise (J. A. S. B., vol. xlii. p. 165), 
which translation led me into difiGlculties about the construction, 
which are obviated by taking ^^IT^^ as the objective. 

Marathi has lengthened the vowel of the root, but Hindi 
has not; and it is to a participial form WpT? the meaning of 
which would be "adhering to," "close to," and finally "to," 
that we must attribute the present aflix. ^ffir becomes ^n[, and 
then %, %, and %. The anuswslra would thus be an inorganic 
addition, common in Hindi. From ^J|^, a regular Prakrit form 
of the relative participle, by a similar elision of l( and con- 
version of ^ into ^, comes the Panjabi objective it for if^, 
where the final if has been weakened into anusw&ra. Analogoris 
to this is the form ^, contracted from W^, in Old-Hindi used 
in the sense of " up to," "until." Thus Kabir : 

^ ^ ^n^* ^I^ 4|i|^ n — Ramaini, s. 46. 

* * How far shall I teU. (of them)? they have gone into unconsciousness," 
literally "till where?" And again 

Digitized by LjOOQ IC 


—Sakhi 201. 
** So long (only) sliines the star as tlie sun does not rise, 
So long (only) do worldly works suffice as knowledge is not full ; " 

literally '^ until that (time) shines the star until which (time) the 
sun does not rise/' 

In a precisely similar way Chand uses ^tf^, as 

**H!ow long shall I describe my inferiority?" — ^i. 22. 

The other change to which W is subject, namely that into *?[, 
yields the Bengali objective*^. 

In Ghijarati works the affix of the objective is sometimes 
written without the anusw&ra, and thus resembles one form of 
the genitive affix; but it will be shown hereafter that the 
resemblance is accidental only. 

§ 57. The Instrumental. This case possesses a special im- 
portance in the seven languages, from the fact that in most of 
them, owing to the peculiar system of prayoga or construction 
which prevails, it takes the place of the nominative before verbs 
in the past tenses — a practice which makes these languages in 
this respect difficult to foreigners. In the cumbrous Marathi 
especially, the correct use of theprat/oga is a sort oipom asinarum 
to beginners, and even in some instances to natives themselves. 
The forms of the instrumental are these : 

Hindi . . 

. . . «ff. 

Panjabl . 

. . nai. 

Marathi . 

. . iMfff, eih »fff, pi. ntn, htn, ih- 

Gujarati . 

. . . e. 

Nepali . . 

. . . /e. 

\ Digitized by Google 


Sindhi has no fonn for this case, but uses the simple oblique 
or crude form without any affix; so also do the early and middle 
age Hindi poets, as will be shown in the latter part of this section. 
Gujarati, in addition to the e, which may also be regarded as 
originally merely the crude form, has also certain pleonastic 
affixes; thus it writes"^ ^IP^^ ^^ 1^ "by God," "by means 
of God," also ^ ift, or "^ jft, which latter is strictly an 
ablative, and will be considered under that case. 

Bengali uses the direct prayoga or construction in the past 
tenses of verbs, and has no need of an instrumental to take 
the place of the nominative ; but when it requires to indicate 
instrumentality, the literary language employs such words as 
^ll^j T^r^lj^, ^ETTTj ^nf^; while the common speech uses the 
participles of the verbs "to do" and "to give," ^f\i|| ".having 
done," r^ill "having given." Both high and low alike also 
borrow the affix of the locative %, as in Gujarati the ablative. 

Oriya, like Bengali, has only the direct prayogay and, like it, 
supplies the place of an instrumental by '9JKJ and similar 
words, and colloquially by*^ "having given," and such like 
words ; but the instrumental is very rarely used in Oriya, the 
locative affix*^ usually supplying its place. 

The Gipsy uses sa in the singular and /a in the plural, but as 
a pure instrumental only, and not as a substitute for the 

In those languages in which the past tense of the transitive 
verb requires the instrumental construction, the verb does not 
under those circumstances agree with the nominative case. There 
seems, however, to be an exception to this rule, if my informant, 
a Nepalese pandit, be correct, in the Nepali language. The 
pandit gives the past tense thus : 

% % ^grot "I ate," fTift % ^ "we ate." 

t%^rn[^"thoaate8t," f7I?ft%^§T^"ye ate." 

^OT % ^J^ "he ate," ^cTfUr^aTOT "they ate." 

Digitized by LjOOQ IC 


In this paradigm we see that the verb changes its termination 
in each person, whereas in Hindi it remains fixed, thus: ^^ 
^TRIT, cf 5f ^TRlTj etc. The peculiarity of the Nepali usage 
is undoubtedly to be explained by the fact that this form of 
the instrumental has become so completely identified with 
the nominative as to be mistaken for it by the vulgar; 
and in all probability, as I shall endeavour to show, the use 
of the % ( = Hindi ^) is a mere modem custom, and origi- 
nally the direct construction was used, and they said If l^|^(, 

Although the question more strictly belongs to syntax, 
yet, in order to establish the correctness of the theory as 
to the origin of the instrumental affix which will be brought 
forward in this section, it will be advisable to give a brief 
description of the question of prayogas as they exist in the 
modern languages. 

T\iQ prayogaa are three in number: kartari, karmani, bhdve, 
which may be Englished respectively, subjective, objective, and 
impersonal; and what they are will be imderstood from the 
following Latin phrases : 

Karta. Rex urbem condidit. 

Karma. A rege urbs condita. 

Bhava. A rege urbi conditum est. 

In the first the verb agrees with the nominative case ; in the 
second it agrees with the object, and the subject is in the instru- 
mental; in the third the verb is impersonal and neuter, the 
subject in the instrumental, and the object is generally in the 
oblique form. 

The Kartd prayoga is generally employed in the present and 
future tenses ; the other two in the past tenses. In Marathi the 
potential, however, takes the Bh&va as well as the Kartd ; and 
in all the languages except Marathi the Karma and Bhdva con- 
structions are restricted to transitive verbs. 

Having thus briefly stated the general system of construc- 

Digitized by VjOOQ IC 


tions, we return to the instrumental case of the noun, which, it 
will be seen, occupies the place of the subject in two of the con- 
structions. It is, in the first place, necessary to observe, that in 
Gujarati there is an additional form of construction, in which the 
subject is in the dative case ; and this is, strictly speaking, the 
Karmani construction: for in native grammars the dative, as 
well as the accusative, goes by the name of Karman or Karma, 
just as we have in these pages called them both the objective. 
The conirtruction with the instrumental would more accurately 
be called the Ehrane, Karana being the name for that case. 
Secondly, not only in so archaic a language as Gujarati is the 
dative used indifferently with the instrumental in the frequently 
occurring constructions noted above, but in Nepali the forms of 
the case-affixes are very similar, the dative having "^TT^, the 
instrumental %; and the same similarity exists between the if of 
the Old-Hindi objective and the % of the Modem-Hindi instru- 
mental; and so also, while *^ is the sign of the dative-objective 
in Gujarati, it is the sign of the instrumental in Marathi. 
From these considerations it would seem to result that the two 
forms are identical in origin, and have been confounded with 
one another by the vulgar. For, as regards Hindi and Panjabi, 
certainly my own experience is, that the objective and im- 
personal constructions are never properly understood by the 
unlearned, and in the rustic dialects of the Eastern-Hindi area 
are more usually omitted altogether, and the direct or subjective 
construction employed. 

It would be out of place here to go into more detail on a 
question of syntax; but it may be noticed that the participial 
form ^^rf^F, with its variant ^xfif, which is almost if not quite 
proved to be the origin of the case-affix*^, is often used with a 
very wide range of meanings, and with great laxity of applica- 
tion, — as is natural from its meaning, which may, without 
violence, be diverted to many uses. Thus in Old-Bengali it is 
used in the sense of "on account of," "for," as— 

Digitized by VjOOQ IC 


fti wrf'T ^^tr ^^^ UT^^iftr » 

\TW %7nf ^RIT »— Bidyftpati in Pad. 

"For what, fair one, dost cover thy face? 
It lias ravished my senses." 

It is, however, necessary to infonn the reader, that the theory 
above stated is not the generally received one; or, as these 
languages have never yet been studied scientifically, it would be 
more accurate to say that it is not the theory held by the few 
scholars who have looked into the question. Trumpp probably 
means his remark upon p. 113 to be taken as a mere obiter 
dictum^ and it seems not to have occurred to him that there were 
serious objections to his derivation. He and others state that 
the affix 5}, with its variants, is derived from the case-ending 
of the Sanskrit, The instrumental in Sanskrit ends in ena in 
the case of o-stems; and down to a late stage of classical Prakrit, 
this ending is preserved, and occurs even in Chand in G&tha 
passages, and, as stated in § 48, survives iDL Marathi as if, 
where the final ^ has been weakened to anusw&ra. In Gujarati, 
also, the e of the instrumental has been mentioned in the above- 
quoted section as a quasi-synthetical termination ; but it will be 
observed that this ena loses its w, and that we cannot in any way 
get i} out of it, imless we suppose a termination ijif to start 
with, which does not exist. In general, the modern languages, 
throwing aside the complications of the various Sanskrit inflec- 
tions, adopt those of the a-stem only, and the few traces of the 
synthetic system that still survive are, without exception, to be 
referred to that stem. "We should hardly be justified in looking 
to the terminations of any other stem for the origin of modem 
forms; nor in this case would it much help us if we did : for, in 
all but the a-stem, the instrumental termination is ^; and 
although in masc. or neuter stems ending with a vowel an >H is 
inserted, making the whole termination "^y yet this >H is 
regarded by Bopp, i. § 158, as euphonic only, and not an 

Digitized by VjOOQ IC 


organic portion of the word. Even from ^, however, we 
could not get the annsw&ra of ^; and if this a.ffix were really, 
as the theory assumes, a relic of an old synthetical case- 
ending, we should certainly find it in full force from the earliest 
times ; whereas, on the contrary, the modem origin of 5f as an 
instrumental with the objective construction, has struck many 
scholars, and is, I believe, now generally admitted. Some 
iastances may here be adduced of this construction in the older 
poets to show that they did not use an aflix % or* ^ as an 

Chand uses the direct or subjective construction even with the 
preterite of transitive verbs ; thus — 

f^ 1?p8ft ^^<!T liW «— Pr. R. i. 49. 

"Ten times the King asked, 
The Brahmia gave no answer in the matter." 

—where, in modem classical Hindi, we should have 4^|^| % IJ^ 
and ftpif ^ f^^. 

Where the subject is a pronoun, it is often put in the oblique 
crude form, and the verb agrees with the object, as — 

fiR T^ ^ftift ^ f^ «— i. 136. 

"He protected the Brahmins." 

Here f?fif i^ plural oblique, and ^fWV the old form of the 
preterite fern, of ^4,«f|, agreeing with "^^ ; again — 

"Who made heaven, earth, the seven hells." 

The various nouns agree with '^, which is neuter plural ; 
while ftrf|[, the subject, is oblique singular. In the passage 
above quoted, i. 49, occur two lines close to each other, each 
with a different construction — 

Digitized by VjOOQ IC 


Wf f^ ^W ^ T!T?r TC II 
'*B!e who killed the snake on my father's neck." 

and in the next line but one ^ 4g^^U4. In the first we have 
the agent in the oblique ftrf^; in the second it is in the nomina- 
tive ^ — 

• ^ ^ifTTW ftnaft ^ifr ^f^ «— xxi. 124. 

** Good speed the Chandel made, 
(Saying) 'Parimal hath written this,' he gave it into his hand." 

Here the agent is in the nom. singular, although the verbs 
iftift* and ^t^^, being both preterites of transitive verbs, would 
in Modem-Hindi require the objective construction. 

Coming down to later times, we find Kabir employing the 
oblique construction without %. 

«h4<d Titf^ f?r% ffi5 ^rnrr ll — Ramaini, 38. 

**B!e who taught the Kdhma in the Kali Yug, 
Having searched, did not find the power (of Allah)." 

f^p^ and f?r^ are both plural obHque ; but the construction 
is irregular, as the Arabic words Jui^ kalama and c:^jj kudrat 
are both feminine, and we should expect TT^T^ and XTT^. 

Perhaps it would be unfair to expect such a refinement 
from the weaver-sage, who wrote his language as he found it, 
without troubling himself much about the words he used. The 
fact that Kabir was a man of the people, and not a pandit, 
gives us great confidence in his writings as evidence for the 
way the ordinary folk of those days talked ; and this confidence 
would be greater could we be sure that his writings had not 
been tampered with by meddling scribes. Again, he uses 
the direct construction in 

Digitized by VjOOQ IC 


Tftr qiMi^ ^^t T|^ ^si^R T|^ 'ffif fi^ n 

— Sakhi 183. 
**!Kabir touclied not ink or paper, he took not pen in hand ; 
He made known the lord to whom is glory in the four ages.'* 

Here '^ifWK ^^ nominative to the three transitive preterites wift 
"touched," iTI^ "took/' and IRT^ "made known/' and with 
regard to the first verb, its objects are irft, which is fern., and 
mWf^ {s£^), which is masc. ; the second verb has for its object 
^li^RT (/J^)> which for some reason is always /em. in Hindi, and 
the verb is fern, also, and thus agrees with its object. It must 
be observed, however, that the word Kabir would, if treated as 
a Hindi word, have the same form in the oblique as in the 
nominative, and we may thus fairly regard it as oblique in this 
place. The verbs would then all regularly agree with their 
objects, as in the Xarmani construction; in which case, if Kabir 
had known of the modem use of % as a sign of the instrumental, 
he would doubtless have used it. Another — 

'ft ^ Tt9 ^ ^^ ^ ^i^ f^# lYi; II 

—Sakhi 176. 
"I have wept for aU the world, no one weeps for me, 
That man weeps for me who contemplates the word." 

Here Sir, though used in Modem-Hindi as a nominative, has 
not yet lost its true force as instrumental of the first personal 
pronotm, of which ^f ("^TFR;) is the true nominative. 

The use of % as an instrumental is quite unknown to Behari 
Lai, in whose Sats&t it does not occur once. Tulsi Das is 
equally ignorant of it, as — 

^5R TIJTRT ^OTT tT3 ^^ II 

"When Raghunath conquered his enemies in battle." — Jranya-k. 265. 

Without prolonging this inquiry by adducing any more 

Digitized by LjOOQ IC 


examples^ it may be said^ as a general dednction from the 
practice of the Old-Hindi poets, that they are ignorant of the 
use of % as an instrumental case-affix, and use the objective 
construction, as a rule, with the oblique form of the noun, indi- 
cated, where there is any oblique form, by the affix e; arid that 
in this respect Sindhi, Gujarati, and,'as far as we know anything 
about it, Old-Panjabi, agree with Hindi, while one at least of the 
Marathi forms of this case is a relic of the Sanskrit instrumental. 
It would thus appear that, on the decay of the synthetical system, 
and the fusion of all the case-endings thereof into the one oblique 
form of the analytical system, no trace of the instrumental as 
a separate case remained, and its place was supplied by the 
objective for many centuries. A partial revival of this case 
took place at a later period, probably about the reign of Shah 
Jahan, when the form ^*, hitherto used for the dative, began 
gradually to be extended to the noun when used as the subject 
of a transitive verb in the past tense, and thus ^ came in High- 
Hindi to be used as an instrumental. 

The reason for fixing the sixteenth and seventeenth centuries 
as the period when this innovation. occurred is, that, as my 
learned friend Professor Blochmann^ has shown, it was about 
that time that the general admission of Hindus to offices under 
the Empire began to bear fruit in the Persianized form of 
Hindi which we call Urdu ; and this new phase of the language 
was based, to a great extent, on the dialect of Hindi spoken in 
the provinces adjacent to the Court, in which the form^ was in 
use as a dative. 

Marathi supplies yet another argument in favour of the 
theory now imder discussion. In that language the instru- 

1 I wish to take this opportunity of paying my tribute of respect to this profound 
scholar, whose knowledge of all that pertains to the Court and Goyemment of the 
brilliant Mughal dynasty of India is singularly minute and accurate, and whose 
splendid edition and translation of the " Aln Akbari " is the most yaluable contribution 
to our information on this subject that has yet appeared, displaying an industry, 
depth of research, and range of knowledge, worthy of the highest admiration. 

Digitized by VjOOQ IC 


mental in ^, wliicli has been shown to be areKc of the Sanskrit 
IpT, is felt to be too weak by itself to express the instrumental 
relation proper, as distinguished from the merely constructive 
instrumental; and where direct instrumentality is required 
to be expressed, we find a case-affix added: thus, "by means of 
a weapon," is ^f^ <!fi'^«|, where ^f^ alone is not felt to be 
explicit enough. On the other hand,*if, so far from being felt 
to be an old half-abraded synthetical case-ending, is recognized 
as a distinct word, and has even yet hardly taken its place as a 
case-affix, but is often used as a postposition attached to the 
genitive case: thus we have such forms as ^|i£||'if= "by him," 
where WT^TT is genitive masculine. * It must be borne in mind 
that there are three grades of formative additions to a noun in 
these languages. First, and oldest, those affixes which are 
remains of the Sanskrit case-endings and are fully and com- 
pletely incorporated with the stem and inseparable from it. 
Second, and next in point of time, those affixes which have been 
adopted to supply the place of case-endings when these latter 
had become so much abraded as no longer to mark case-relations 
vnth. sufficient accuracy, ^d which are npt integral portions of 
the stem-word, but are attached to its oblique form. Third, and 
latest, those adverbs, particles, and postpositions which are 
recognized as independent words, and are attached to the stem- 
word after it has already received its case-affix. It will be seen 
that the degree of coherence of these three grades is exactly in 
proportion to their antiquity. Consequently, when we find 
*^ in the third of these categories, we cannot admit it to be a 
relic of the synthetical system, because, if it were, it would be 
in. the first category. 

Marathi has plural forms for aU its methods of forming the 
instrumental: thus — 

SiDgular t( has pi. ^ and ^. 

|> •f 99 it IT* 

Digitized by LjOOQ IC 


In this particular, popular usage seems to have extended to all 
three affixes a plural which was originally appropriate only to 
one ; and the use of "^ft in the singular is parallel to that of the 
ablative form yif, which, as shown in § 51, has been extended 
to the singular from being originally plural only : for ipf must 
be taken to be originally a plural also. The form ^ is a synthe- 
tical relic, being merely a lengthening of the Prakrit f^, 
Sanskr. fil^ (Lassen, p. 310) ; ^*is more correctly written ^, 
and is, I think, another instance of the connexion between the 
instnmiental and the dative. Such a word as^^^nHf, or^^j^, 
"by a god," would correctly be divided ^^TW + ^ (for ^), and 
not ^^T + lif. Hoemle has shown (J. A. S. B. vol. xlii. p. 61) 
that the form "^^TRTT^, tised as a dative in Marathi poetry, has 
similarly been treated as though it were ^^ + fl I Xf , which is 
erroneous, as there is no such word as ^raY» The word should 
be divided ^"^fTO + W3¥ ; the latter word being a Prakrit form 
of ^Bf^ "on account of." In these two cases we have a con- 
struction exactly parallel to that of TUT^TT*^ above, where the 
affixes are only in the third degree of cohesion, and are attached 
to the genitive of the noun: for, as explained above, § 52,*^4[ij, 
though now used as a. dative, is really "^^^ = Skr. *^c|4Sf. 
Lastly, for the confusion still existing between the two cases 
may- be cited the curious construction still common in Panjabi 
in such phrases as ^^^ ^ ^^3^1 ^ = "he was to have come," 
literally *'by or to him to come it wa8"="illi veniendum erat,'' 
or "ab illo veniendum erat," where we may call ^^ % a dative 
or an instrumental, as we please. Native grammarians call it 
the latter. 

The above considerations leave no doubt in my own mind of 
the truth of the theory that the forms of the instrumental case- 
affix now in use are originally datives which have been trans- 
ferred to the instrumental. Other similar cases of the affixes of 
one case having passed over to another will be met with as we 
go through the remaining affixes. 

Digitized by VjOOQ IC 


§ 58. The Ablative. — ^The terminations of this case in 
Sindhi and MaratM have already been shown to be remains of 
old synthetic case-endings. Marathi has no other method of 
expressing the sense of "from;*' but it, like aU the other 
languages, uses the oblique form of the noun, together with a 
long string of affixes, or rather postpositions, to express mean- 
ings which it has been customary with grammarians roughly to 
class together under the head of ablative. 

Strictly speaking, however, the ablative is that case which 
expresses procession from, or, as it is called by Sanskrit gram- 
marians, "^MI^H; and I shall here therefore only notice those 
affixes which convey the meaning "from." These are — 

HiDdi . . . . "^, ?|, ^. 

Sindhi . 
Oriya . 

^» #t> ^. 

Bengali has no form for the ablative, but uses a postposition 
^%^, which has been explained above, § 51. The origin of the 
form ^ appears to be the Skr. adverbial ablative cf^, as in 
ITWff^ "from the village," in Pr. ^, where the o has been 
softened through dto e. In Chand it is often written^r}, as — 

"Prom his race sprung." — i. 164. 

^ ^^Sft ^ ^ft^*c| ^\J II 

"Say ye (and) I make him slain from Hfe." — i; 178. 
{t,e. give the word and I kill him). 

I think the anuswS-ra here is merely an inorganic addition, as 

it so frequently is. Concerning the origin of ^, there is much 

obscurity. No scholar, as far as I know, has as yet thrown any 

light on it. The most probable supposition is, that it is of the 

VOL. n. 18 

Digitized by VjOOQ IC 


same origin as % but with the particle t% added, so that ifV 
would stand for 7(^, the vowel being lengthened. This appears 
to be Vans Taylor's idea (Gujarati Grammar, p. 64) ; but he 
seems also to think the form may have arisen from a fusion of 
the two forms of the Prakrit ablative in W^, ^IT^, and f^, 
which is possible, but not probable, as ^[ra is too old a form, 
having passed into ^y and "^fft^ before the period of the origin 
of Gujarati. 

For the Sindhi ^, etc., Tnmipp ofEers no satisfactory 
account, merely remarking that it is allied to the objective sign 
% (^), but with the termination of the ablative '^, ^, etc., 
added. More will be said on this point when considering the 

The Oriya ^ appears in this form in the earliest documents 
we possess, and a fuller form "31^ is common in modem times. 
It will not, I think, be contested that we have here the Prakrit 
ablative in ^(TJ; so that the forms would be Skr. |^1«|X(^, Pr. 

The Hindi affix %, usually applied to the ablative in modem 
writings, does not really mean "from," but "with," and comes 
under the head of postpositions; but as it is now used* as a case- 
affix, — that is, with the oblique form of the noun, and not, as 
pure postpositions are, attached to the genitive or other case, — 
it will be better to consider it here. % is softened from an older 
form #t, still used constantly in the rustic dialects of Hindi, 
and this leads us back to the full form ^RT? which is the Sanskrit 
adverb ^Rf "with." Chand uses it after verbs of speaking, just 
as % is used in Modem-Hindi — 

^t f7Tflrf^\T«l BIT n 

"Quoth the messenger to Prithiraj." — xii. 16. 

^t ^ifif B^ *rr W 

"Quoth the wife to the husband." — i. 7. 

in which latter instance it precedes the noun to which it belongs. 

Digitized by VjOOQ IC 


Sindhi has fomls ?5t, aiid%; Gujarati H ot ^, used in poetry 
in the general sense of "with," though in some grammars 
erroneously called a locative, probably also belongs to this group ; 
and in Old-Hindi we meet with an intermediate form ^1, to 
which correspond the Sindhi forms ^RW and %i^. In Old- 
Bengali we meet with ?a^, which, however, is probably for ^, 
the form used in the modem language. Thus Chandi Das 
(a.d. 1460) has— 

^TTT f % ^4£>< 1^ firf%^ ^rWT ^ I— Pad. 1307, 26. 

^^On the banks of Yamima, beneath the Kadamba tree, she met witb 


and Kasi D^s (Mahabh&rat 415, S. 6) : 

**With whom in the forest fight the three men?" 
Tulsi D&s uses ^if in his Ramayan frequently — 

** With him Yajnavalkya afterwards found (it)/' — ^BMa-kand. 14. 

"I afterwards having heard it with my own guru." — ih, 

**She spake with Trija^a, clasping her hands together.'* 

— Sundar-kdnd. 300w 

§ 59. The Genitive. — In all the seven languages, and in 
most of the subordinate dialects, the system that prevails for 
this case is to add to the oblique form particles which vary their 
terminations so as to agree with the governing noun. In other 
words, the genitive case-affix is adjectival, and Agrees with the 
governing noun just as an adjective would, so that, as has been 
frequently remarked, the construction is not that of the Latin 
patris equus, but that of paternus equus. Bengali and Oriya 
having lost all gender and all means of marking the oblique 
form, have also rejected as useless the adjectival form of the 

Digitized by LjOOQ IC 


genitive; but it will be shown that the aflixeft which they use 
are closely aUied to those used in some of the other languages. 
The forms are as follows : 

Hindi ki. 

Panjabi ..... d^. 

Sindhi je, 

Gujarat! no. 

Marathi cM, 

Bengali er, 

Ojiya «r. 

The first five languages inflect this form. Thus Hindi has 
1T> %> ^; Panjabi ^, ^, ^, tl[^; and so with the others. 
Bengali and Oriya remain unchanged ; and omitting them for 
the present, it will be seen that the principle of making the 
genitive case of the noun into an adjective runs through aU the 
other five. It also prevails in the allied dialects. Thus Mar- 
wan has '5(t> T> ®*c.; Mewari 35t; the Konkani dialect of 
Marathi wt, ^, %, etc., as well as ^, ^, ^;^ Kashmiri ^, 
^rt^, ^, etc. 

The genitive is the most difficult of all the cases to account 
for; and, as there has lately been considerable discussion about it 
between high authorities, one cannot but approach the knotty 
subject with some trepidation, taking as our starting-point the 
modem Hindi forms ^, %, liSt. When the governing noun 
is masc. singular nominative, the genitive takes the affix ^, 
as ^TTSRIT ^41 "the horse of the father." The affix ^ is used 
when the governing noun is feminine, no matter what be its 
number or case, as ifR ^ ^t^ "the mare of the father;" 
^TR ^ ij^vj^^li "the father's mares." When the governing 
noun is masculiiie, but not in the nominative singular, % is 
used, as "^T^ ^ ^i1% ^ TTTT " he beat the father's horse." 
Hindi has only these three forms, and the reason of this is, 
that its adjective is not inflected any more than this (see § 56). 

It may next be noticed that there is evidence to show that a 
letter T has dropped out of all these forms. 

1 Cimha-Blvar^, Grammatica da lingua Concani (Goa, 1859), p. 38. 

Digitized by VjOOQ IC 


The Bhojpuri, a widely-spoken rustic dialect of Bihar in the 
eastern Hindi area, has in its pronoun of the third person the 
genitive affixes W(^, ^, though I have never heard nr(^. 
With this agrees the language of the Chinganas or Gipsies 
of Turkey, whose genitive is formed by koro, which does not 
change for gender or number, though the closely allied adjective 
in koro, which Paspati (Tchingian^s, p. 53) distinguishes from 
the genitive, does. Thus, BalamesAoro manush, " a Greek man," 
BalamesAm gaji, " a Greek woman." 

The Marwari dialect, as q)oken in the present day, uses, as 
mentioned above, the forms "^^j '^, '5^: thus ^HT^Jt ^ TTTTS 
Tt Trot TT^TTCt mSI^I^IT " Here come I the merchant of Ram- 
garh, the Raja's treasurer ;" iii\dl<lO ^«^ H«rO ^3fW ViJ^ 
TTH^ " I hold the treasurership of the kingdom, from which I 
have acquired dignity."^ 

Nepali has ijt = H. ^; ^ = H.%; and^=H. ^. Nepali 
having set up as a distinct language on the formation of the 
Hindu kingdom of Kathm&ndo in a.i>. 1322, we should expect 
to find it perpetuating the Hindi of the date when it separated 
from the parent stock,, and we are thus led to conclude that, as 
early as the fourteenth century, the T had dropped out of many 
rustic dialects as regards the noun ; while as regards the pro- 
noun, it is retained in all of them to the present day ; and, with 
the exception of M. and S.,. the genitive of aU the personal 
pronouns in all the languages of this group is formed by an 
affix whose typical letter is T- 

One step further back from modem times takes us to our only 
authorities for mediaeval Hindi — ^the poets. These eccentric 
gentlemen are very fond of omitting case-affixes altogether, and 
stringing in one line a number of nouns in their crude form, 
leaving the reader to make sense of them as best he may. I am 
not here alluding to Chand, but to writers far more modern 
than he, as Tulsi DS,8 and Bihari Lai. When, however, they do 
1 Marwari Khydls, p. 3, ed. by Eev. J. Eobson, Beawr, : 

Digitized by VjOOQ IC 


condescend to use case-afi^es^ they use several sieparate sets 
of forms. 

First, and commonest, are the forms now in modem use : 
ITT, %, ^; but of these ^ is generally and in the best editions 
written ^ orift. From this we gather that this affix, what- 
ever its origin, is undoubtedly an adjective of old standing, the 
form ^ being the older form of the masculine nominative of 
a-stems, which always end in o in Prakrit, and still retain this o 
in G. ^ and S. ^, as weU as in Konkani ^. Examples are : 

IPI f^^ftro ^ ^ TT^ II— Sundar-k&nd. 305. 
"He burnt the city in a moment, 
Only not the house q/'Yibhlshan." 

# ftpf ^ 4rt*^fil ^irftTc! «<>ill, II— i*. 

"Of whom he has made unlimited boasting." 

% ^¥ twhur TT^ iT^ ll-i*. 298. 

"Hear, Yibhlshan, the lord's custom." 

In this last instance "5^^, though feminine, has the masculine 
oblique form of the genitive ; and this would seem to show that 
*^ was used with the oblique of both genders, for ^^ is here 
the objective, though without the case-sign. 

A few instances may be taken from the Bhaktam&la of 
N^bhaju (circa a.d. 1600, the tik& is a little later) : — 

^nt^ II— Bh. 133, 1, tika. 

"Came Yamdeb afterwards, he asked Kamdevjii, 'Tell me in fuU the 
very pleasant account of the milk.'" 

"A new zest in amorous poetry, an ocean of impassioned leve." 

—ib. 44, mill. 

"In imitation of l^arsingha, he smote Hiranakus (Hiranyaka^ipu)." 

— td. 49, mCU. 

Digitized by VjOOQ IC 


Extremely common in Tnlsi Bks is the fonn ^; this is 
noticeable because affording a connexion with the Bhojpuri 
•fCT and the Gipsy koro. It is not declinable, and does not 
therefore make the genitive into an adjective. Instances abound, 
the following are a few : — 

^rfq ^ ^^iT ^vt^ ^fif n 

" Hearing the friendly speech of the monkey." — Sundar-kand. 300. 
'* To-day thefruit of the good deeds ©/"all has passed away." — ^Ay.-k. 343. 

"Quickly take away (this) fool's life."— Sundar-k. 304. 

In some copies a feminine form ^flRf^ occurs; but this is 
written IR^ in others. 

"The pride of a monkey is in his tail." — Sundar-k. 304. 

'nnr ^g'TTf uidm ^rtr ^n# n 

"Hearing with his ears the word of treachery." — ib. 309. 
Kabir uses this form — 

"There was there no day nor night, 
Of him {i.e. the Creator) tell what is the race and caste?" 

— Ram. Ti. 6. 

It is not, however, frequent in his poems; and, when it occurs, 
is chiefly used with the pronouns, as in the lines just quoted. 

At this point comes in Bengali with its genitive in TJT, 
which, like Tulsi D&s's genitive in IR^, does not change for 
gender or case. In Old-Bengali this case takes the afi^ ^, 
which still survives in certain adjectivally used forms ; but, in 
accordance with modem Bengali pronunciation, now takes long 
dy and becomes ^FTT- Thus they say, '^llOiaftK "of to-day," 
et|faiH<^ "of to-morrow," or "yesterday," in such phrases as 

Digitized by VjOOQ IC 


"your letter of to-day has arrived/' "the aflfair of yesterday;" 
also 7ffmrc> TWT'in: "of that place/' "of this place." The 
Oriya genitive in ^1^ also belongs to this category. In the 
singular it has lost its ^, because in Oriya the final a of stems 
ending in that letter is preserved, so that we must read IT^ 
ghara, not, as in H. or P., ghar. From this it results that when 
the affix H^ was added, the ij stood alone between two vowels, 
and was, according to the general rule, elided. It ought to have 
thus produced ghara+ar=ghardr ; but the tendency of Oriya is 
to shortening unaccented vowels, and, whatever may have been 
the place of incidence of the accent .in ancient times, in the 
present day it is certainly on the stem-syllable in ghdrar. Con- 
sequently, instead of ghardr, we get gharar. In Bengali the 
same thing occurred, and the present form of the genitive 
in er is, in all probability, another exemplification of the 
often-remarked tendency of this language to corrupt d into e. 
This view is supported by the fact that, in many cases where 
custom had previously led to the rejection of the final a of the 
stem, the genitive even in Bengali is formed by ar, as in Oriya. 
If further confirmation were needed of this origin of the T geni- 
tive, it would be found in the Oriya plural, where the ^ is still 
preserved. Oriya, as above noted, § 43, adds to the. stem the 
word iTR, to form the plural. This word takes in the nom. the 
form Jr(w( for animate, and ^TR for inanimate objects ; and in 
the oblique ^fTPT^, which is probably for iTPrf^f, like the Old- 
Hindi plural. Adding to iTPnC the genitive form ^P[, we get 
the modem Oriya genitive ^TPI^, wherein the ^ has been 
preserved, quite according to rule, by being incorporated with 
the preceding nasal. Bidyapati sings — 

^fBTf^nft flpj ^WfZT 5^f5l5 ni««\ ^^'T WtHTT I— Pad. 984, 5. 

**I was a virtuous woman, I have become unchaste through deKghtmg 

in his words." 

In his poetry, however, as in Kabir's, this form is more 

Digitized by VjOOQ IC 


frequently met with in the pronouns ; in the genitive of nouns 
he generally drops the T, and uses only ^ : as — 

'wnrar ^tfif m^mir ^R TfT i-Pad. 985, 2. 

"The love o/a good man is like a line (engraved on) stone." 

wm ^ ^iif5 'CT ^frtc «— Pad. 

** Every one will say (it is) the quality of love." 

?>^ W(Z^ ^JprrW^ ^ «— Pad. 109, 6. 

**It increases like the fibres of the lotus." 

To be classed together with these forms is, I apprehend, the 
Marathi afl^ ^, denoting "a resident of," as in fcm^^< 
"a man of Chiplun." Here also I would place the Marwari 
Tt> where the initial ij has been rejected, and the still further 
corrupted Mewari 35t, and Konkani ^. 

Passing on from IR^ and its variations, we must come to a 
fuller form ^"^j %'^« This was first pointed out a*s a genitive 
by Hoemle in his articles in the Joum. As. Soc. B., and I have 
since found additional confirmation of his view from other in- 
stances. The two passages which were first noticed are in Chand. 

Hif^ f^ffe ^ f^ '^nr^TTf ^(^ II— xxi. 29, 9. 

**The kettle-drum made a noise, the armies, wheeled, 
The sight of the Ch&huvan was lost from view." 

That is, the two armies lost sight of each other fronpi the dust 
they raised. The second passage is very obscure, and the render- 
ing is tentative only; there is, however, no doubt about the 

^r^ nj<<«! f^^fr ^a %^ 1— xx. i4i, 7. 

"Blindly ran the elephant of the Ch^uv&n ; 
Making a circuit, he surrounded (the enemy) on all four sides." 

There are other passages also where this form occurs. In 
fact, it is used whenever a rhyme is wanted for the preterite of 

Digitized by VjOOQ IC 


T)i4,«|| "to turn." Kabir, however, uses it in the middle of a 
verse : — 

^ ^ ^^1 Tra Tg[rCT ^ilTT »-Sakbi 163. 

"The society of humaii beings, into that let no one hastily plunge; 
In one field feed the tiger, the ass, and the cow." 

TTnre ^ Tftif ^ ^ftr %i; <fff Tw I 

1^ irrCt HWif XW irtZT ^R Wm H-Sakhi 142. 

** Cy ignorance are three qualities, the bee {t,e. life) has there taken up 

his abode; 

On one branch three fruits, the brinjal, sugar-cane, and cotton."* 

W^ iV^Ml %War Ht ^pPr W(T^ WT 8— Sakhi 77. 

" (As from) the clouds of Sravan drops fall from the sky. 
All the world has become Yaishnava, giving ear to the gurus. ^^ 

(Here ^RRTPf = ^^UJT, ^(^^1= Iji). Tulsi Bka also knows 
this afl^, thus — 

^ ^nnr f^ ^rt?f "vsi^ft h 

^^ inftftl "ife tTRT %<t y-Ayodhyft-k. 63. 

"Por many days there was great expectation. 
Omens and certainty of a meeting with the .friend." 

^ffif ^RT ^ftir xwr ^ ^BRtT II 
t^^ i^Mfd ^f^ ?R "l^t «-Lankft-k. 6. 
** Hearing this speech of the son of the wind, 
Snules Raghupati looking at the monkey's body." 

^^ ^f^ ^T f;^TO %TT 11— Aranya-k. 265. 

"Seeing the smoke of (the bodies of) Khara and Dushana." 

The form ife''^, with its changes for gender and case, is also 

^ This is one of Eabir's obscure didactic utterances. He means that ignorance, 
or rather the condition of unassisted human nature, has the three qualities of sattva, 
rajah, and tatnah, in which life, compared to a bee among flowers, is sunk. It is, he 
says, as if one branch should bear such different products as those mentioned. 

Digitized by VjOOQ IC 


in use in Old-Gujarati, and the range of forms is fuller than in 
Hindi ; thus we have — 

M. F. K. 

With a n. sing. |p(^ %^ %t|. 

„ n. pi. %^ id. llR^t- 

„ oblique %^ and ^ id. ^|i^ and^. 

Illustrations are — 

Samul Das, Padmayati, 64.^ 
**I have kept at home the son q/" Champak." 

"Let there be an order o/" Your Majesty, I will suck the sea dry." 

—'id, AngadavUhtM in Kdvya-d, i. 23. 

'* Having left the waves of Ganges, he goes on to the bank and drinks 

from a well." 

— Narnngh Meheta in KdvfO'd. iL 4. 

It is a well-known and frequently used affix in Gujarati. 
To complete the range of illustrations, I may add here a 
remark which was accidentally omitted from its right place a 
page or two back, that Bidyapati's genitive in is, formed by 
omitting the -^ of m^, is also found in Old-Hindi. Thus, Tulsi 
Bis has — 

ftg ^rro^ HW ^SR?f ^ ^V^ »-Ay..k. 334. 
"(To obey) a father's conmiand is the crown of virtue." 

And it occurs in Chand : — 

"Hari can save in an instaat." — ^i. 60. 

In this place-irf|[ is = 1^, and is used in the third grade of 
cohesion with the genitive of the governed noun, so that fl^^ra 
is a genitive (flpf = ^^^), and ^ is shortened from IR^ or ^RJ. 

1 From Lecky'8 Gujarati Grammar, p. 250. 

Digitized by LjOOQIC 


The varions facts that have been adduced have brought us up 
to the edge of the gap of the dark ages ; and, before leaving 
terra firma for the region of analogy and the cloudland of 
inference, it may b6 as well to sum up what we have gained. 

Hindi forms its genitive by affixes which produce an adjectival 
construction ; these are ^, %, ift. Of these ^ is written ^ 
in Nepali, and in many rustic dialects, also in mediaeval poetry. 

Bhojpuri has a pronominal genitive in W^j 5^1^. Marwari 
has "^j"^, '^. Tulsi Dds and other writers have IR^, with a 
feminine ^flRf^, which is doubtful. Oriya has H^ in the genitive 
plural, andiir^ in that of the singular, which latter is due, as is also 
the Marwari "^j to loss of the ij. Bengali had IR^ in pronoims, 
and has ^jpf^ still in certain adjectival formations ; and its pre- 
sent genitive Tj;^ is formed from "^ by loss of il and softening 
of the A (resulting from the coalition of the final a of the stem 
with the initial a of the termination) into ^, more ma. The 
gipsies have preserved the in and the "^ both in their genitive 
koro, which, when used as an adjective, has a feminine keri. 

Old-Hindi and Gnjarati have a fuller form of this genitive 

affix— *i|RCT,*%^, ^. 

Beyond these facts, there is, as far as we know, nothing further 
to bd adduced ; and the question now is, what is the origin of 
these forms ? The form%^ is found in one of those scraps of 
Apabhransa saws which Sastri Vrajlal has quoted in his admir- 
able little work. He unfortunately does not give the author's 
name, and gives the date as Samvat 1551 (a.d. 1495), which is 
very modem for Apabhransa ; but the language of the lines is 
older than this date : — 

"Confdipation at eighty-six, falling in love with one's female slave, 
Wrapping oneself in a blanket, show that death is near." 

This might pass for Old-Hindi or Old-Gujarati ; and it is note- 
worthy that the language is precisely identical with that which 

Digitized by VjOOQ IC 


Hemachandra writes about as a Prakrit. It may, therefore, be 
permitted to us to regard the form%^ as one which ascends 
into the Prakrit stage, or that of decaying but still existent 
synthetical forms, and it is consequently a form which has not 
yet become a mere affix, but still retains something of the nature 
of an independent word, and, as such, has a meaning of its own. 
We shall see, as we go further back in point of time, that in a 
slightly earlier age this word is capable of being used alone as 
an integral constituent portion of a sentence. 

There has lately been a controversy between Hoernle and 
Pischel on this very subject ; and in a work like this, which pro- 
fesses to lay before the reader such deductions as the author has 
been able to make from his own observation, and such facts as are 
generally admitted by scholars, it would be out of place to enter 
into a lengthened argument upon questions which are still sub 
judice. After reading and testing as far as I am able the argu- 
ments of both sides, the conclusion that I have come to in my 
own mind is that Hoernle is right, and that the objections of 
Pischel refer rather to details than to the general argument. 
.The reader may judge for himself, either from the original 
articles (J. A.S.B., pt. i. p. 124, 1872 ; Indian Antiquary, vol. ii. 
pp. 121, 210, 368), or from the following brief, but I trust 
accurate, abstract of the arguments. 

The Prakrits have a word kera, which is used in the plays in 
the sense of "done by," " connected with," "belonging to," and 
the like. It is to this word that the genitival form %'(t is 
referred by. the writers above mentioned. It also occurs as 
keraka. Hoernle derives it from the Skr. past participle 'gnr; 
Pischel, following Lassen and "Weber, derives it from kdrya. 
Hoemle's process is as follows : WfT becomes in Prakrit ^lf^cft> 
just as W{ becomes ^(\^ and Vtf changes to VlPidt ; then the 
71 drops out according to rule, leaving ^fl'^t, which is the 
same word under a slightly diflferent method of spelling as 
Chand's ^R^, a form in use all through mediaeval Hindi. 

Digitized by VjOOQ IC 


Hrf^^ or )i^ makes %^ just as naturally as "^ICT^^ makes 
ir%^ through IT^ff^lit (soe Vol. I. p. 135). Pischel seems 
to mistake Hoemle's meaning, and to labour under the dis- 
advantage of knowing nothing of the modem languages. A 
question like this is not to be settled, like a matter of Greek pr 
Latin scholarship, merely by collating texts of written works. 
He urges that keraka is far too modem to have undergone " so 
vast and rapid" a change as that into Bengali er. But, unless 
the whole of this work which I am engaged on is wrong, the 
most striking feature of the seven languages is precisely the 
"vast and rapid" change which they have undergone; and, in 
this particular instance, a change that took fifteen centuries to 
effect is hardly too rapid. He relies on Lassen (p. 118), who, 
however, admits that his reason for doubting the derivation of 
%^ from ^TTT is merely the absence of an ^ which could form H 
in the first syllable by epenthesis. Hoernle has shown how this 
might be, and has proved the existence of intermediate forms 
mf^jcft and itilXiH'. The derivation from wd^ is a sort of pis 
aller, a casual thought which appears to have occurred to Lassen 
when he was hunting for an analogy to dkhari/a=achchhero. . 
It does not seem to have struck him or Pischel that the meaning 
of kdryam will not in any way produce a genitive, while that of 
krita wiU. Krita means " done ; " and to take as an instance 
the line of Tulsi Das quoted above, cRfrf ^^ ^^if = liifqgkTt ^^ 
"the speech made by the monkey," is a phrase which would 
easily glide into " the speech of the monkey," while the phrase 
9rU<4iT^ ^'^•f would require to have its meaning violently 
wrenched before it could be got to mean anything of the kind. 
I cannot perceive the drift of Pischel's objection founded on the 
fact that keraka is always inflected. Of course it is, and so is its 
descendant Gujarati %'^, %Ct, etc. ; and it is this very quality 
of being inflected that makes it suitable for an adjectival genitive 
affix. It is farther urged that in the later Prakrit dialects keraka 
becomes kelaka. This does not concern us, because the later 

Digitized by VjOOQ IC 


written Prakrits are as artificial as Sanskrit itself, and probably 
were utterly unlike the spoken language. 

It may then, as a conclusion, be safely stated that Skr. ^TTI^ 
becomes in Pr. ^if^^, then%?^ and'ifc"^;iit, from which we 
get"^"^^ and %0> and that from this, shortened into WfJ, 
comes the Mod.-Hindi HT ; while from nr^, rejecting the final 
Towel as useless, in the two languages which have dispensed 
with gender and case, the Bengali forms ^nC aiid TJX and the 
Oriya iffx; and ^ are derived. It is possible that the Mewari 
3ot and Konkani wt niay be similarly derived from iJi^niS; but 
I prefer to regard them as modifications of an original '?^, such 
as we have in Marwari. 

Gujarati has for its genitive affix a series — ijt, ^, ^, etc. 
The early form of this affix, which is found also in Apabhransa, 
is TfTlft, fl^, ^. Instances are — 

^Tf ^sr; fftTi^ ^^roft ^iwrR^n 

*'Eix thy thoughts on Hari, thou of little wit, indolent." 

— Narsingh Meheta, Kclvya-d. ii. 1. 

(Literally, " grasp the meditation of Hari,") 

"The man who has cultivated the excellent virtues ofRsn., 
The world celebrates Ms glory." — ih. 

*'0n account of the shame of Drupadl, he rushed from Dwarika in a 
breath." — ih, ii. 5. 

In this line both forms of the genitive are used, that with %^ 
and that with 7f^. Another instance from the Apabhransa 
was quoted in § 46. Another is — 

"The King of Lanka thus speaks."— iJ. i. 16. 
Side by side with this affix, the modem affix in ^, etc., occurs; 

Digitized by LjOOQ IC 


but this does not prove that it is not a later form of 71^, 
because the poets habitually employ archaic forms side by side 
with those of their own day, as we have seen in the quotations 
from the Hindi poets in this Chapter. Nor does the change of 
Iff to 1 present any difficulty, as Gujarati is very lax in this 
matter, and in common parlance, like Hindi, uses ^ in places 
where Iff is etymologically correct. 

Bearing in mind that the principle which runs through the 
genitives of all these languages is to apply to the noun some 
affix which shall give it an adjectival sense, it will not seem 
strange that each language should have selected its own affix 
from the extensive range of adjectival affixes which Sanskrit 
offers. In the present instance it is natural to look to the affix 
^fif , in such words as ^TRfT " i)erpetual," from ^pfT " always ; " 
H^TTPT "ancient," from gij "formerly;" ^ipr "new," from 
^ "now;" f^^l "old," from f^ "long ago." if is gene- 
rally written IJ in the western Prakrits, and thus ^fif becomes 
in Apabhransa, as we have seen (§ 46), Tfuft. The meaning of 
(ffif accords well with that of an adjectival genitive ; and it is 
not strange to see an affix which in the classical language is of 
restricted use, extended by the vulgar to every noun in their 
language. Although in Sanskrit this affix forms adjectives out 
of adverbs of time only, yet in the Gipsy language we find the 
corresponding affix ino extended to adverbs of place ; as, 

anglalutno "former," from angldl "before" (Skr. "^RT)- 

arattutno) ^ ^ \aratti) * ^ ^"^'' 

avgutno "firstborn," „ avgo "former." 

duritno "distant," „ dur "far" (Skr. ^). 

And even to nouns, as — 

gatmdno "villager," „ gav "village" (^> H. T^f^), 

dakarutno *'kmg\y/' „ dakar ** king." 

Digitized by LjOOQ IC 


"With elision of t, just a& in modem Gnjarati, 

angarun/9 " made of coal," „ angdr "coal** (Skr. ^(^fT^). 
^ariMio ** stony,** „ 3ar "stone.*** 

The aflix ^JPTF is also in use in some rustic dialects of Hindi, 
and will probably be found in Chand, though I do not remember 
to have seen it there. 

The Marathi genitive aflix ^ has been asserted to be derived 
from the H. ^ by the change of « into ^; but after much 
searching I am unable to find a single other instance of this 
change in the whole of this language, and hesitate to believe 
that so common a word as the genitive affix should be the result 
of a phonetic change, which is absolutely isolated and unsup- 
ported by a single analogous process (see Vol. I. p. 209). It 
seems rather to have arisen from the Skr. affix ^, meaning 
"produced in," "sprung from,*' in such words as ?f^i||T(Q 
"southern," trr^THQ "western." Here, again, it is assumed 
that an affix of restricted application in Sanskrit has obtained 
a wider use in the vulgar language. In old Marathi poetry this 
affix has a lengthened form f^eil, t^f^, and %fiT ; but this 
additional syllable is merely added to eke out the metre, 
and is commonly found after other case-affixes also, as in 
1^ and ift^Rlt for ^i|. 

Sindhi takes as its affix ^, which Trumpp wishes to derive 
from the Skr. adjectiavl affix ^ through ^ to gj, a process for 
which there is no authority. It will be in analogy to the 
derivations suggested for the G. ift" and M. ^J, to derive 
this affix also from a Sanskrit adjectival termination, namely 
^ (^), in such words as IRT, t^, %X!J, \3f^, and the like. 
The only obstacle to this theory is that ^ is added to 
the oblique form, whereas, if every substantive had been made 
into an adjective by the addition of ij, only the stem would 
remain; and the same objection applies to the derivation 

^ Miklosich, neber die mundarten der Zigeuner Europa's, ii. 23. 
VOL. n. 19 

Digitized by VjOOQ IC 


suggested for Marathi ^; thus V^^ilfti: equinus, miglit give ^- 
^H^, and thence Mi'iNU hut not ^^tim^, as it actually is. 
Perhaps among the unthinking vulgar the custom of adding all 
the other case-affixes to the oblique form was carelessly extended 
to the genitive also. As regards Sindhi, however, it is, as an al- 
ternative line of investigation to be pursued, worthy of considera- 
tion whether we have not here the relative prbnoun ij "who;" 
thus Vifil ^ ^sRfft "the master of the houses," would be origi- 
nally, "the master who (is) of the houses," — ^the oblique form here 
doing duty as a genitive, which is exceedingly probable when 
we remember how large a part the genitive occupies in the 
formation of the modem oblique. 

Sindhi and Kashmiri have a genitive affix, S. JJT^, K. ^^. 
The modifications of the Kashmiri genitival affix are very 
eccentric and peculiar, and are given as follows by Elmslie 
(J. A. S. B., vol. xxxix. p. 101). 

With singular C Siog. masc. ^^ fern, ij^ (sanzy^), 

noun ( PI. masc. ^tf^ fern. ^^||[ {sanzah aV;^), 

With plural i Sing. masc. 'ff? fem. f|^ (hinz yjt). 

noun ( PI. masc. f^f^ fem. fi^Wf^{hinzah ^jiA). 

Here not only does the affix vary in concord with the 
governed noun, but it varies with the governing noun also, thus 
introducing a double series of confusions. Examples are not 
given in the meagre article from which these forms are taken, 
and there are contradictions in it which detract from its trust- 
worthiness. Moreover, the author having laid down at the 
outset a system of transliteration, straightway departs from it ; 
and by giving us also the words in the Persian character instead 
of the Nagari, he stiU. further misleads the reader. All this is 
very tantalizing, because the few scraps of information which 
we have about Kashmiri are just enough to show that it is one 
of the most interesting and instructive languages of the group, 
retaining a singularly rich array of archaic words and forms. 

Digitized by VjOOQ IC 


There is one peculiarity, however, of which a firm grasp can 
be taken, namely, the change of initial ^ into ^. This is a 
well-known feature in Zend and in those Aryan languages which 
are situated close to the Zend area (see Vol. I. p. 258). This 
process seems to explain the origin of the Panjabi genitive in 
^, ^, ^y etc. In Panjabi, the change of ^ into f is extremely 
frequent, and it is in its general features a less archaic language 
than Kashmiri, so that it is not unreasonable to suppose, that 
while the latter only suffered ;^ to be weakened to ^ in the 
feminine afl^es, Panjabi suffered the change to take place in all 
cases, and we thus postulate an old Panjabi form l{^. This 
form is still used in many rustic forms of Panjabi, It and 
?i^ are simply the present participle of the verb " to be," and 
one often hears ^^ ^^= "being," or ^^ ^^, according to 
the dialect. ^^ is the Panjabi version of Pr. ^^, Skr. ^TO ♦,> 
from the root HJ^. One can readily imderstand how a word 
meaning " being," when added to the oblique form of a noun^ 
would gradually acquire the sense of a genitive ; thus, " the 
horse being to me," or "which is to me," becomes "my horse," 
just as in Latin we have the construction "est mihi." The 
anusw&ra of the present participle is dropped in Panjabi in 
many verbs, especially in the more cultivated portions of the 
country, where the typical or classical form of the language 
would naturally be developed ; thus, while the rude herdsmen, 
of the southern deserts say ^F^;^f mdrendd, "beating," the 
dwellers in towns say irR^.^<^^^^- ^^ tl^is way ^^ would 
become ^f^, and that again %^, whence by elision, of ^ we get 
the present genitive form i^. It is worthy of notice, that this 
form in ^ seems to be of comparatively late origin, and that in 
the Granth we more frequently find the genitive affix ^> as in 

§ 60. The Locative. — The traces of the synthetic system still 

^ Trumpp in J.E.A.S., n.s., toI. t. p. 198. 

Digitized by LjOOQ IC 


remaining in this case have been oommented upon in § 48 : liere 
we have to notice the modem case-aflixes nsed to express a loca- 
tive meaning. These are aa follows :— 

Hindi . 
8indhi . 

^. Bengali %. 

ft^. Oriya. . *^. 

"^> ^it^- NepaK . ifT- 

^. Kashmiri ntaiij? ( = ^Fil|?). 

^^rt?^. Gipsy . te} 

The most important group of these endings is that having j^ 
for its characteristic, and we are fortunately able to trace the 
origin of this affix fully back to the earliest times. It is derived 
from the Sanskrit adverbially used locative if^, meaning "in 
the middle." Chand uses a great variety of progressive forms 
of this affix, thus showing that even in his time the changes 
had been completed, and that he was consciously employing 
archaisms, as he so often does, when he used forms which were 
older than the latest^. Instances are, dropping the final e, 
and thus giving J(^. 

** Immortal dwelling among mortals." — ^L 3. 

*' Having said this speech, he came amongst the army." — xxi. 10. 

In this fuller form the word retains the fuller meaning 
"amongst." Next comes the solution of the semivowel into 
its vowel 7|f%f, and when a long vowel is required '^^. 

"Thousands three fell on the earth." — ^xxi. 7, 59. 

"The witch went amongst the queens." — i. 178. 
^ Paspati, p. 65. See also p. 500. 

Digitized by LjOOQ IC 


The natural transition from ^ -I- IJ to ^J (see Vol. I. p. 326) 
gives the form TTf^. 

"Fell headlong into the bottomless pit." — i. 79. 

7?^^ is here a form of the 3rd person singular preterite for 
TRFEft", which would be the Hindi rendering of Skr. iftt^ from 
ift^, and thus means "was set free;'* combined with Tn;ifT = 
ITOTT " to fall/' it means " was let loose falling/' i,e. " fell im- 
restrainedly or headlong." The order is here, as sometimes in 
Chand, and often in Sindhi, reversed, the postposition becoming 
a preposition. 

Connected with this last form are the forms l{t^ and 9r^> 
with inorganic anusw&ra, and the former with lengthened vowel. 
The final short ^ is rejected, a fate which frequently befalls final 
short vowels in Hindi. 

"They themselves went into the garden." — xxi. 5, 6. 

^Tl^rnr is probably a curious termination of ^31| with Persian 
i[) , after the fashion of "^JtWT- 

"What Mng, in what land." — xxvi. 18. 

In the next line we get — 

^<^A ^^ 'Rt II 

"In wealthy Tljain." 

xnt^= Skr. TI^ " abounding in (wealth)." A lengthened or 
secondary form, injTT> ^^T^ a quasi-diminutive termination, 
is also used, but with the more definite meaning, "in the middle 

"The modesty of men and women has gone, in the middle of the month 
of Phagun."— xxiii. 1, 4. 

Digitized by LjOOQ IC 


— ^alluding to the indecent songs song at the Holi festival in that 

** Having received the news, she arrived in the middle of the city." 

— i. 178. 

"The enemy fled into the middle of the hills and forests.'* — i. 206. 

A step further brings us to the rejection of the organic portion 
of the aspirate and retention only of the aspiration (Vol. I. 
p. 263, et seqq.), thus producing the extremely common form 

"Putting musk into lampblack, 
The queen streaks her eyes for ornament." — i. 20. 


"Grazing in Jh^khand." — i. 61. 
It is frequently lengthened to Mjf\ and irt^* 

"Seeing the king sitting in sleep." — ^i. 191. 

* "The hero Jalhan was smitten, he fell in the field." — xxi. 264. 
There is also a corrupt form ifflf. 


"If the husband dies in battle 
(and) the w^e does not become a Sati." — xxi. 175. • 

From this last form, which is properly ^, by rejection of the 
H[ comes the present ordinary form ^. 

"In one month he colonized a city." — i. 218. 

Digitized by LjOOQ IC 


The Sindlii and Gujarati forms are only slight dialectic varia- 
tions from the Hindi type ; as is also the Nepali, which has not 
the anundaika. 

Bengali and Oriya often express the locative by adding the 
full Sanskrit form irSt to the genitive, and Oriya oddly gives 
the locative ending to this word, so that we often hear TITITT 
4^U|^, where the last word is the locative of a noun W^. 
Marathi also uses irSt in the same way, and in poetry the 
archaic ^TWt, as in Tukaram, Abhanga 1887 : — 

t^^K f^'^ Tiff ^ 'irfft II ^^W g^ ^ ?prnrwt n 

*^Vithal is the universe, there is nothing besides, thou also art seen in 

And again — 

f%t^*^*^ ^njrt^ ?rai3B II TPTRT^ %9B TRT ^[fifT II 

*' Brahma has created the whole universe, in it are his various diver- 
sions and skill."— Abh. 1886. 

The most common method of expressing the locative, however, 
in Marathi, is by adding to the oblique form nt, apparently 
derived from the Skr. adverb IR!^ "within," which in Prakrit 
becomes ^tft* This particle is, even in classical Sanskrit, used 
as an affix, as ^^ifrRHf, between the teeth. In Marathi the 
initial vowel is lost, and the two letters nt alone remain. Thus, 
^TH-'cf=^5r^tcT "in a house," IjTpf "on an elephant," from 
flft; ^T^ "in honey," from i?^, with the vowel of the 
termination lengthened, as is customary in the oblique form. 

Of the Bengali % the most probable origin is from the 
adverbial ablative W^, of which notice was taken in a preceding 
paragraph; and Oriya *^, which, in its fuller form, is "af^, is 
probably of the same origin, — the two forms, TT^ and '^fit, 
having by degrees been restricted to special meanings of the 
same case, namely the ablative. 

§ 61. In addition to the case-affixes above noticed, which are 

Digitized by LjOOQ IC 


specially attached to definite cases, there is a very large number 
of particles which serve to modify the noun, but which cannot 
have any definite place assigned to them in the narrow range of 
cases to which the grammarians have bound down the Aryan 
declension. They fall into two classes, according as they are 
attached to the crude form of the noun, or to the genitive 
case. The classification cannot, however, be carried out very 
thoroughly, because some of them fall into different categories 
in different languages. Thus, H. XTW " near," (Skr. tn^, is 
attached to the oblique masculine genitive, as ^ % VIJM " near 
the house," literally "near of the house;" but M. iJWf, which 
is the same word as H. V[j^, is attached to the oblique stem, as 
frU^WT. It will be better, therefore, to go through the most 
commonly used and widely current of these postpositions, with- 
out attempting to class them under either of the two heads 
mentioned above. The distinction is important, as indicating 
the degree of cohesion to which each particle has attained ; and 
thus enabling the student, in the absence of literature, to form 
an idea of the comparative antiquity of each of them, and thus 
to measure, to a certain extent, the rate of progress of those 
phonetic changes which have given rise to the present modem 

§ 62. Of wide use is H. "^"St " before," used with a genitive : 
its older form is ^RW, still heard in rustic Hindi. Sindhi has 
^Ill5| and ^iPl^l, with oblique genitive in %, or with ablative 
in ^. The other languages have— 

Panjabi IH^i and Hf^n^. 
Gujarat! IRT^* 

Bengali ^n% (vulgo) 1{]^. 

In all these pairs of words we have derivatives from two cases 
of the Skr. '^Pff. The words in e are locatives from ^B%, 

Digitized by LjOOQ IC 


meaning "in the front;" those in u and o ablatives from 
^roT?t, meaning "from the front;" "IRirni. hecomes ^^* ^J^TT^j 
^Cnrra ; and the long d is in Hindi entirely omitted, and in S. 
and P. softened to «; while Oriya adds its own locative and 
ablative case-affixes*^ and ^. 

Similarly, for the corresponding word " behind," the Old-Skr. 
adjective T(^ (vedic) supplies its locative H% (not used in 
classical Skr.) and its ablative M^T<t. The words in use are — 

H. in%, ^ffil and J\VS. 

p. ft% 

» ft^. 

8. TJtH 

» ^• 

G. Xl% 

.. ''l^j'ITOt 

B. fi%. 

0. XJ% 

.. ^nf . 

The nexus ^ regularly becomes ^, and then ^, with a pre- 
ceding long vowel. H. tft% presents an anomalous long i, for 
which it is difficult to account. My theory is, that the same 
process has taken place as in f^f^il^ (Vol. I. p. 307), where the 
nexus is simdered, and the ^ passes into ^, producing ^{[^ ; 
thus Tj% would become t[^, or, owing to the tendency to insert 
i after h between two short vowels (Vol. I. p. 138), xrf^|%, and 
thus Tf^, which, from imitation of the kindred word trfit ^ l^^s 
been written with ^. The Sindhi has entirely dropped the nexus, 
and substituted a labial vowel, which is very anomalous. In 
every case it will be seen that the terminations are respectively 
locative, meaning " in the rear," and ablative, meaning " from 
the rear." * 

A third series, meaning " below," is afforded by the same two 
cases of the Skr. adj. ift^ "low." 

.H. ^ft% and ^t^* 

B. id. 

Digitized by LjOOQ IC 


This series is not so extensiyely used as the others, as there are 
two competitors for this meaning : 7f%, which is the favourite 
word in B. and 0. ; and^^, which has the preference in P. ^3 
" below," >^ " from below/' "^^ " along the under side of ; " 
G. 'I^ " below; '' S. "%% " below/' >Tf " from below." 71% re- 
quires no explanation, being a simple Skr. locative, one of those 
strong plain words which is not susceptible of any phonetic 
change. \z is a curious word, and has imdergone considerable 
change. Its origin is Skr. ^^SRT, which, in Pali and Prakrit, 
becomes \j, by rejection of the initial a, solution of ^sT into ^, 
and change of the second a to a short ^. Although ilM^K^ » 
common Skr. adverb, is generally quoted as the origin of this 
word, yet it would be more in consonance with the regular 
course of phonetic change to look to a form ^ff^^: the last 
element of which would be the root ^BT (OT), to stand. The 
P.\St and S.^^ are, however, regular ablatives; and G.^^ is 
the regular locative, just as the other above quoted words. 
(Weber, HaJa, pp. 42, 202.) 

Skr. ^qfff; " upon," gives H. ^fHTC, and so in all. With loss 
of initial w, it yields a long string of words, which have arrived 
at the position of case-affixes, being joined directly to the oblique 
form; these are, H. xf^, Vtfx,^ "on;" S. V^fx., G. xtTj 0. if^, 
B. xr;. With further softening of "q into Tf , it becomes in 
M. ^, and is closely attached to the noun, being written as 
one word with it ; and so also in Gujarati : not even taking the 
oblique form of the noun, but the simple stem, as ^^T^ " on 
the house," Tf^T^ " on a tree." 

Skr. "35n?f gives H. TH^ and 7{^, which, from having origi- 
nally meant " in the place," has grown gradually into an affix 
meaning "up to," and even simply a sign of the objective 
= ^; so you may say ^if ^ TTft ^^ ^^ <Tik ^TTCt "beat him." 
S. has also TfT^ and Tft^, P. TTT^. 

From Skr. ^Cl^, Pr. ^|f and ^^, are derived H. ^T^fil^ 
^fSl^f ^^f ^, and ft^; S. ITFft, ^TP^rwt, and ^VHf^ 

Digitized by VjOOQ IC 


used like the rest in the sense of coram, in front of, face to 
face, is rather a modem compound from modem materials, 
^ "with/* and 4{)^<^T "front," a recent diminutive from ^15. 
To this group may also be attached the Oriya ^fT^, meaning 
originally " in front of," then " in the presence of (a king)," 
and so, ultimately, as a substantive, meaning the court or audi- 
ence-hall of a king or great man, in which sense it takes the 
locative sign*^, as ^l^|^ ^|4iX t'R^ ^1^, " he supplicated 
the king," literally " in the audience of the king." The Hindi 
writers spell their word in a dozen different ways ; but from the 
terminations of H., as well as the other languages, it is clear, 
that here also we have the two Skr. cases locative and ablative, 
as in most of the other posl5)ositions. 

Skr. xn^ and ITT^ "near," H. "qro, P. Vfm and IIT^ "near," 
irrSt and mft "from near," S. ml, G. ^Tf%, M. JTTiRf, ^'^^• 
Here, again, the locative and ablative. 

There are innimierable other postpositions in use in all the 
languages : they will be found in the dictionary. The point to 
be observed is the prevalence of forms derived from the Skr. 
locative and ablative cases, necessitating the placing of the noun 
to which they are attached in the oblique genitive, or, to speak 
more clearly, in that form of the genitive affix which it takes 
when governed by a noun in the singular oblique. 

The structure of the modem noun is thus strikingly homo- 
geneous in all the seven languages, as well as in those cognate 
dialects which have not yet been thoroughly investigated ; and 
which, owing to the insufficiency of trustworthy data for them, 
it has been found necessary to exclude from the present 
inquiry. They have all a stem in four forms — the nominatives 
of the singular and plural, and two obliques for the two numbers. 
There are also here and there traces, faint and slight, but still 
quite unmistakeable, of the older synthetical system of the 
Aryan languages. These traces consist of abraded case-endings. 

Digitized by VjOOQ IC 


occasionally removed from the stem^ and used as postpositions. 
Some confusion has crept in — as was natural among generations 
which had lost the sentiment of synthetical construction — ^be- 
tween the yarions cases^ and a form proper to one case has 
passed over to another^ or been extended to alL To supply the 
distinctness of meaning which the mutilated case-endings no 
longer afforded^ case-affixes^ themselves for the most part 
shortened and simplified remnants of old adjectives^ nouns, and 
participles, have been called in ; but though this principle has 
been introduced into all the languages, the precise words so 
adopted vary in each case, every language having forms of its 
own not used by the others. The close relationship of the whole, 
however, is more fully established by this practice than it would 
have been had all the languages been in the habit of using 
precisely the same affixes. Uniformity of principle is a far 
deeper lying bond and token of esoteric unity than mere surface 
similarity of individual words : the latter might have been bor- 
rowed ; the former, being an inborn mental instinct, could not. 
The same remark holds good of the still further development 
of the noun's capabilities of expression, as shown in the post- 
positions which do not form cases, but are merely attached 
to the oblique noun. In these there runs throughout the same 
principle, though its exemplifications are different. The more 
we penetrate into the secrets of the structure of these languages, 
the more do they show themselves to be closely allied by the 
deepest and most fundamental ties, — ^the same blood runs in the 
veins of all, and the same fertile Aryan mind has found expres- 
sion in their rich and varied formations for its activity, wealth 
of resource, and 7ro\vfiifxavo<: ipifx/eui^ 

Digitized by LjOOQ IC 



CONTENTS. — § 63. Pronoun of the First Person Singular. — § 64. Plural 


Genititb op the two First Persons. — § 67. Pronoun of the Third 
Person. — § 68. The Demonstrative. — § 69. The Relative. — § 70. The 
Correlative. — § 71. The Interrogative.— J 72. The Indefinite. — § 73. 
The Reciprocal. — § 74. Adjectival Pronouns. — § 75. Pronominal Suffixes 
in Sindhi. — § 76. General Scheme of the Pronouns and Pronominal 
Adverbs. — § 77. Miscellaneous Pronouns.— § 78. Gipsy Pronouns. — §§ 79. 
80. Concluding Remarks. 

§ 63. The Personal Pronouns in all the seven languages are 
singularly homogeneous in type, and their analysis is rendered 
comparatively easy by the fidelity with which they have 
preserved the Prakrit forms. In this respect they stand in 
contrast to the nouns which have so widely departed from the 
ancient models. The first and second persons run parallel to 
one another, and have four fundamental forms, namely the 
nominatives and the obliques in both numbers. The genitive is 
a possessive pronoun, and, as in the noun, adjectival in form. 
It win be treated separately. 

Digitized by VjOOQ iC 



the first person the forms are as follows 





HiDdi ff, sr [%] 




P«DJ»bi f^[3|] 




sindhi ^^(i^,y^ 




Gqarati jf 



i^?T, ^«it« 

Marathi ^ 




Oriya ^ 



^Tif, ^N 

Bengali ^ 




Gipsy me 




NepaU ^ 




In the above scheme are given the forms which are etymo- 
logically correct, rather than those which the people really use, 
for much confusion has taken place in this respect in modern 
times in two ways. First, H. P. and M. use for the nomina- 
tive a form which is really the instrumental ; H. P. ?| has now 
quite superseded the old and proper nominative ^ in all but a 
few rustic dialects ; and in M., as far as I can learn, there is no 
trace of a form ft, or anything like it, either in ancient litera- 
ture or in rustic speech. Oriya and Bengali have for their 
nominative a form beginning with ^, which is characteristic of 
the Skr. oblique; so have the Nepali and the Gipsy languages. 
Secondly, Oriya and Bengali have adopted the habit of using 
their nom. pi. ^9|Tf^ and Hi^ as singulars, and have invented 
fresh plurals, 0. '^F^TTT^, and B. ^nWTTj ^^^ ^' ^^^ in addition 
struck out a plural for ^ in the shape of iftTT« Fashion and 
pandit-influence have succeeded in relegating poor ?T1[ and 
HKKS to the domain of "vulgar" speech, and, to a certain 
extent, 0|iya l^ also ; and so far has this habit gone that many 
natives refuse to admit that ^ and Wt^ are the true old 
singulars, but, in spite of the obviously plural nature of ^j^ 
and ^^nf^, persist in regarding them as the genuine singulars. 

Digitized by VjOOQ IC 


The nominative singular ft, ^f^, if or Jf — ^for these four 
forms are all one and the same — represents the Sanskrit "^j 
which loses its initial a in the Apabhransa or vulgar Prakrit, 
and becomes ^, which Lassen (Inst. Pr. p. 466) shows to be 
shortened from the Sauraseni ^T^, so that ^ stands for ^ 
(Yol. I. p. 254). Hence, by elision of if, comes f^, a form 
which Panjabi in many of its rustic dialects still retains. Sindhi 
goes a step further, and rejects the ^, while lengthening both 
vowels. The other form, '^, Trumpp considers as simply con- 
tracted from "^ ; but as "^ had been lost to ordinary speech 
long before the Sindhi forms began to be excogitated, it is more 
probable that "^ is to be regarded merely as a shortened form 
of '^rt^. Old-Hindi ^ is the same word as P. f^ with a 
slight difference of spelling ; while Middle-Hindi and Gujarati 
if are shortenings of the same by omission of the short internal 
a. Chand uses ^, sometimes written ^t> as — 

"Then I quit this body."— i. 157. 

^ ft ^ ^g'TTI ft THT H 

''All that I am hearing, mother."— i. 160. 

Modem-Hindi and P. ?f , which is now the only form in use for 
" I," is, strictly speaking, like Marathi ift, the instrumental ; 
in Skr. J(^, in Apabhransa i?^, and apparently also irtj, 
though Lassen is doubtful on this point (Inst. Pr. p. 480). I 
fail to see why Trumpp calls this an accusative (p. 189). The 
transition of the instrumental into a nominative is rendered 
natural by the use of the prayoga, in which the subject takes an 
instrumental form, and accordingly Chand uses % only before 
the preterite of transitive verbs, i.e. in the place where the 
subject is required to be in the instrumental ; in all other places 
he uses ff. Thus — 

^ ^^ ^rrff f^'T lift 'JIV'r N 

"I heard that the Shah had put out his eyes." — ^Ixv. 110. 

Digitized by LjOOQ IC 


— where in Modem-H. we should have 7i ^ ^^TT. The modem 
fashion of saying ^ % is founded upon ignorance of the true 
nature of the word^ and contains the instrumental twice over. 
In Panjabi this error is often avoided, % being used alone as 
an instrumental ; and so also in Gujarati, where ^ alone is the 
instr. of jf . Marathi ^ is probably from ^; but M. has also 
two forms of the instrumental, iff and iqf , the former of which 
refers back to ^, and the latter to some other corrupticm of 
Skr. iT^, the intermediate links of which are not extant.^ 

B. ^ and 0. ^ appear to have arisen from the Apabhransa 
form ^ by rejection of the initial ^, and have probably passed 
through a stage in which they were spoken ^9^ and i^^J : the 
first l[ in B. is inorganic. It is in singular accordance, as 
regards sound, with these forms, that Sindhi, at the other end 
of the Indian continent, uses also l| as a nominative, as also J(\; 
but, as regards origin, these forms have apparently, like H. ?f^ 
passed over from the instrumental. 

The commonest types of the oblique form in the singular are 
^ mo and ^^ mujh. IRt occurs throughout Old and Middle 
Hindi, Oriya, and Bengali. It appears in the slightly modified 
forms ^ and Iff ^ Sindhi; and, if we are to write Paspati's man 
as ^, in the Gipsy also. Although, in treating of the genesis 
of the oblique in nouns, it was stated that not the genitive alone, 
but a sort of conflation of all the cases in Prakrit fused down 
into one lay at the origin of the modem form, yet it was also 

^ Lassen probably reads %y^ for ^t^ in Vikramorvasi, Act iy. (p. 93, Calc. ed.), 
where the Calcutta edition reads ^f^ . The line is : H ^rf^f yigfq ^9?% ^f^ 
fvm Mp^f^ril. Here it is clearly a nom. '* If I, wandering about the world, 
shall find my love," as the Skr. gloss gives it: ^|J ^tMt WP^ ^^rf^ fW^i 
^f^% N Lassen uses Lenz'6 edition, which is not procurable here ; but if Hl^ 
is to stand, it is a noteworthy instance of a construction which has now become 
universal in Hindi, by which the instrumental is used for the nominative even before 
verbs in the present and future, and not only,^as in the Karmapi prayoga, before the 

Digitized by VjOOQ IC 


admitted that the genitive had exercised a potent and perhaps 
slightly preponderating influence. Here, in the pronoun, the 
genitive has clearly been the foundation of the modem type. 
The genitive in classical Prakrit and FaU is not here alluded to, 
so much as that of the Apabhransa. In the classical Prakrit 
the genitive stands as im, K%, it^, and %; H&la uses 9nf 
(once), 1^^, Jff, wi, ^, and in?; and Pali has im, Wl, iTff, 
and^V^ . Of all these forms, H&la's ^ seems the most popular, 
and stands one step higher than the Apabhransa inf> from which 
comes directly the S. genitive ^Jlf. Of the introduction of u 
into the first syllable in this word, and in H. ij^g", notice will be 
taken further on. ^nf would become ;fft and ^, just as the 
objective affix i|lf becomes, as we have seen, ijf and ^. Of ^ 
we have, indeed, no instances ; but then we have no writings of 
the period when it was in use. In Chand and his successors 
^, with its oblique affix ff, occurs frequently. First ^, with 
or without case-affixes — 

"How shall there be salvation /or me.^^ — i. 188. 
"Bhat by caste, King of poets, Lord! my name is Chand." — ^vi. 18. 

%^ ^rtf ^ ^ ^ Mi«i* II 

"Having thus said /or me you find fear." — ^i 160. 

^ ^ €t ^rr^ 'T ^^s^ II 

"If you do not speak truth with «w." — i. 157. 
Secondly, iftff and its shorter form '^f^: — 

^ ^ff fw f'nrfwl II 

"If Dhundha shall swallow me."— i. 170. 

^111^ ^Rffiff*r ^ ifrff II 

"Quoth the lord of Mohini to me."—-i. 192. 

Tfl' 'itfif ^UH finTT THTOTT II 

"There is no business /or »w in my father's palace." — bdv. 366. 
VOL. u. 20 

Digitized by LjOOQ IC 


In Old-Bengali also the closely-allied ift% is tlie regular 
form in use. Here the final e is short and=i. Bidyapati has : 

"What dost thou say ^ m^ after all?"— Pad. 

"Except me, in sleep thou shalt not see another." — Pad. 

The other form ^u is derived from J{^y one of the^Prakrit 
forms of the genitive, and owes its u probably to assimilation 
to the pronoun of the second person 7^. M. and G.^ while 
changing the ^ into ^, retain the vowel a, and use if^. Ohand 
uses ^1| not only as a genitive^ as in 

Tf ^»nc^ ^ fin! ^tirftw I 

"This land (was) my father's and forefathers'."— i. 279; 
but as an oblique generally, as in 


"TeU thou to me, mother!"- i. 147. 

Modem-Hindi uses a form ^%, which is made by adding e, 
the usual sign of the oblique in nouns, and indicates the objective 
case ; the affixes of the other cases are added to ^^ alone, as 
^^ "^j %> etc. The genitival origin of the form has been 
quite forgotten. 

Gujarati is very unfixed and irregular in the oblique cases. 
It takes if^ alone for the genitive, at least so say the gram- 
mars ; and, moreover, makes it the oblique stem, to which case- 
endings are attached, as 7nr% "to me," Tnr"^ "from me." 
The old poets use also IJ^. Thus — 

^pifi} a^iqi 5l^t^ <li^U| ^inc^ laft iff II 

"You form the cruel intention of leaving me alone." 

— TnLd in Kdvya-d. i. 4. 

In addition to this, however, it uses the form of the genitive 
(masc. nom.) m\J as a fulcrum for the ablative and locative, 
asifTCfft "from (of) me," ^1X^1^1"^^ (^f) nie;" and the 

Digitized by VjOOQ iC 

THE PKONOUlf. 307 

oblique form of the same ;Rf^ as an instrumental and dative. 
These are confusions arising from the consciousness that the 
oblique ttw was really an old genitive ; so that, when they got 
a new genitive, they used it also as an oblique. One hears in 
Hindi also, colloquially, such expressions as W^ % " from me," 
due to the same seiltiment. Marathi elides the final W of inr in 
some cases ; thus we have YHPirT aD.d HWt " to me,'* THPEft and 
^irtPf "against me/' In the locative, it, like G. and H., uses the 
oblique genitive in% shortened to inf^, aiid with the i further 
recrudesced into ^, as a fulcrum for the case-afi^, thus pro- 
ducing 4|||(«|M "in me,*' precisely parallel to Or. iTITT it- 
Oriya and Bengali use their oblique form IRt regularly through- 
out. 0. has, however, one curious exception, making the objec- 
tive ^rfjj, or shortened ^ (mote), instead of ift^^ which would 
be the regular form. As ff nowhere occurs as a case-afl^ in 
0., the only way that I see of accounting for this form is to 
suppose that we have here a shortening of the affix ?n^ or ?n(, 
which in H. and others has the sense of an objective, so that 

§ 64. In the plural, Oriya preserves the Prakrit form un- 
changed as regards spelling Hi^, but pronounces ambhe, the 
insertion of the b being due to the influence of the preceding 
labial ^: Bengali ^9|Tf^ appears to be merely a softening of 
^F^, which in Hindi hal3 imdergone transposition, the ^ 
having been thrown back to the beginning of the word, just as 
the verb^ "is," for |[l[, by transposition from ^rf%, shortened 
from 'll^t^. Nepali exhibits a form J[M^, as my informant 
writes it, which should probably be ^(if^; the long i being 
almost imiversaUy written* for short i in rural Hindi. This 
form is transitional to Bengali, and the short i must be regarded 
as a corruption of the final e of "^Q^. Gujarati writes ^^, but 
in a majority of instances the rural population use f^, which 
is not necessarily more correct than ^%, though the analogy of 

Digitized by VjOOQ IC 


Hindi would lead us to expect it. The Gipsy form amen, perhaps 
to be written ^Bjlf, agrees closely with this, and suggests the 
possibility of the ^ having been simply dropped from the 
Prakrit form. M., in its form ^!linF|ft, seems to have done for 
the plural the same as it does for the singular, namely, trans- 
ferred the form of the instnmiental to the Nominative ; for the 
instrumental in Apabhransa is 'Hi^flf , and the objective ^IFflt- 
From a confusion of these two woidd arise ^HTFlft : this, with 
the nasal ^RT^, is now used as the instrumental plural ; and 
the two forms ^9^7^ nom. and ^IIPF^ instr. stand to each other 
in exactly the same relation as the forms of the singular ^ nom. 
and ^ instr. It is difficult on any known phonetic principles 
to see in ^W*^ a derivative from ^|i^. 

P. and S. stand alone in having a nom. pi. "^RRf. Trumpp 
does not offer any satisfactory explanation of this. It is true 
that Pr. KV^ points back to Skr. '^m ; but we cannot leap over 
Prakrit and take our form from Skr. direct ; nor, if we did, 
would it help us with the long final i and anund^ika. A change of 
^ into ^ is a weU-known feature of these two languages ; but a 
reverse change of ij into ^ is quite opposed to their habits. 
Kashmiri has a similar form, which in one vocabulary is written 
^^ ais or "^SIX^, in the other ^rf%. 

Kashmiri and that group of ancient Aryan dialects still spoken 
in Dardistan differ from the cognate languages of the plains of 
India in having a fondness for ^, which they often retain in 
places where the latter would modify it to ^; and it may be con- 
jectured that P. and S. derive this form from some intervening 
dialect of hill Prakrit which has not come down to us : they 
also retain the ^ in the oblique cases of the plural. The other 
languages derive from the Apabhransa genitive ^F15, with 
which Oriya is identical. In H. the ij has, as in the nomina- 
tive, been thrown back to the beginning of the word. Although 
I have given the oblique form as ^^, yet- in practice this is 
hardly ever used, the case-affixes of the plural being added to 

Digitized by VjOOQ IC 


^, as ^ ^, ^ "q^. There is also a crude obKque fif used 
without aflixes, which comes from the Apabh. accusative ^97f^. 
G. aflBxes its case signs to a form ^nft or to ip^, the former 
being the same as H. ^wi, and, like it, seeming to postulate a 
genitive ITFl^TI^, formed on the same principle as the gen. pi. 
of nouns ; the other form ^RRf is clearly from "^f . M. '^n?^ 
is formed by lengthening the vowels from the same, and so is 
B. HRfT* In fact, all these forms are so closely similar to the 
Prakrit as to oflfer no difficulty in their analysis. 

§ 65. The second person is an exact parallel to the firsts and 
its forms are as follows : — 

sma. NOM. 





HiDdi ^[^] 




PaDJabi 1^ 




SiDdhi ^ 



TfSf, etc- 

nXt, etc. 

Gujarati 7f 



?W, ?!^>, 

Maratbi H 




Oriya g 



Q^^^fej fS^^' 

Bengali ^ 




NepaU ?r 




Gipsy tu 




There is a striking uniformity in the nom. sing., for even H. 
has in many dialects the form ^with anun&sika, though this 
is rejected in classical Hindi. All the early languages of the 
Indo-European family have as their base tu. The Skr. ^ is 
exceptional, and, as Bopp shows (Oomp. Gr. § 326), the m 
belongs to the case-ending, and the a is inserted between this 
ending and the base in all instances where the base does not 
already end in a, so that before this inserted a the u of an 

Digitized by VjOOQ IC 


original ?[ haa been hardened to ^. Thus we may assume a 
form IJ for the type of this pronoun. The Prakrits exhibit a 
considerable number of forms ; the commonest in scenic Prakrit 
is Tpl, side by side with which is ^flf ; and in Apabhransa ^, 
where the IJ is merely a stop-gap. By eliding ^ and lengthen- 
ing the labial vowel we get ^, the commonest form of the nom. 
in the modem languages. I take no count of the difference 
observable in the quantity of the vowel, though in some languages 
we find ?j, and in others i|, because these languages for the most 
part, as has been frequently shown,- ignore the difference between 
i and i, u and ^. Hindi ^ is like^ of the first person, the 
instrumental ?n( of Apabhr. brought over into the nom. It 
is a word of the rustic and vulgar side of the language, and is 
only used by the educated classes contemptuously when speaking 
to inferiors or domesticated animals. In P. it is stiQ the instru- 
mental, just as^. 

The oblique has two principal types, ^ and gu, correspond- 
ing to ^ and ^IJ of the first person. "Jft is used in Old and 
Middle-Hindi, and still in B. O. and S. and is derived from 
the genitive of Apabh. TJf , which S. preserves in its genitive 
imder the form ^^f alone, ipj, which becomes in M. and G. 
^, is from another Apabh. form of the gen. cflj. Nepali 
stands alone in having dropped the vowel altogether. 

The nominative plural Pr. g^^ is accurately preserved in 0. 
gi^ (pronoimced g^), slightly shortened in B. gfif ; and M. 
here, as in the first person, takes over the instrumental "g*^ in 
the form Ij^fft nom. and cJ4-fV instr., which some ignorantly 
write gijjt . H. here also, as in the first person, writes TJR; and ?f^ , 
which latter is the Gipsy form also, l^epali finft has singularly 
changed the characteristic labial vowel to a palatal, just as 
(§ 51, p. 235) it has ftr?T for ^. 

Panjabi makes its pi. nom. g'^, which is as great a puzzle 
as "^wt of the first person, and for which I can as yet assign no 
satisfactory derivation. In this case Sindhi parts company with 

Digitized by VjOOQ IC 


P., and exliibits a very great number of forms^ wliich, however, 
are arrangeable under two types. Thus — 

7(^9 7(^9 7(ff> ^^' 

^naPfj ^w!* ^Bftj "^5 ^* 

The first form in each of the above Knes is evidently the real 
form frolh which the succeeding ones have arisen by shortening, 
elision, and other simplifying processes. Tfjt is, I think, like 
M. "g^Fft, the Apabh. instrumental g^F^, where the u has been 
changed to ai>, and the m first weakened into anunllsika and 
then dropped altogether. It might also be accounted for by 
comparing it with Gujarati, which drops the labial yowel and 
uses 7f^. If Sindhi has done this, then the ^ is a softening of 
^, as in H. irN' from in^. The other series, of which ^Rjf is 
the fullest form, is quite imparalleled in the whole group. 
Trumpp's explanation is probably correct, that it comes from a 
form Pr. ^f , Skr. ^%, where the initial ^ has been elided 
and u changed to av, as in Tfjt. Even if this be the correct 
explanation, we have a most unexampled retention of a very 
archaic form which has never found its way into literature. 

In the oblique plural H, uses '^, as in the nom. The other 
languages mostly retain the form ^^, which is the base of the 
Prakrit pi. in all cases, and stands alone in the genitive. The 
modem languages generally add long d or dn, as M. ?piT, 
B. ?ft*ir, S. c!5tj the latter with its parallel series ^ISt* ^^• 
G. follows Hindi, but substitutes a for t* in jm and 7nft> which 
latter agrees with H. 'g'JFft* Panjabi 7j^ seems at first sight 
tiO agree with the other languages ; but the ^ is here in reality 
merely the ordinary Panjabi substitute for the ?5 of 7j^. 

§ 66. As in the noun, so also in the pronoun, the genitive is 
really an adjective agreeing with the governing noun in gender 

Digitized by LjOOQ IC 


and case. The formSy wliich depart very mucli in some cases 

from the type of the oblique, 

are here given. 

I. PBB8. 8INO. 

Hindi ^ 
Paojabi ^ 


n. PBRS. 8ING. 

n. PBBS. PL. 





ipmn. gfTii. 

Sindhi ^lit^etc 

. ^TOrt^H 

stc. i[titi, etc 


(Jujarati iTTTY 




Marathi VfTl^ 




Oriya iftT 




BengaU ift?: 




Nepali %7fr 




Bhojpnri f i^, ofr 




Marwari ifT^t 




Sindhi in this group merely uses the ordinary oblique with 
the genitive suffixes^ and Marathi does the satne in its plurals, 
simply dropping the IJ of m?^ and ^Ff , while in its singular it 
makes a compound form TRr9( + ^= irnjr, and gic + ^= g^. 

Leaving these two languages aside, the rest exhibit, under 
diflferent forms as respects quantity and attendant vowels, imi- 
formly a type in -5^ , which we have no difficulty in connecting 
with the older genitive of the noun formed by the affixes 
^'^ and ^flR^. It has been customary, however, to give a 
diflferent origin to these forms. Those who have done so have 
unfortimately taken two extremities of a long chain and com- 
pared them together, totally omitting the intermediate forms, 
with which they were probably unacquainted. Bopp, whose 
knowledge of Hindustani was necessarily very limited, derives 
H. %^, ^^, from Skr. 4|^i( "mens," ^^i| "tuus;" and in 
the same way ^^TnCT would be for ^FfTTT (as it is) from ^IM^^ 
a^d g^fTTT from ^^^i(. The process is said to be efEected 
by the change of ^ into ^ and then to '^, just as in the numerals 

Digitized by VjOOQ IC 


Skr. i;i|IT^ becomes Pr. ipWTf , H. IITTT^* ^^* "^® ^^7 ^^^ 

in vain in Prakrit for the intermediate fonns with ^, and the 
termination f^ of Skr. does not produce masc. d, fern, i, in the 

The pronouns use the same affixes for signs of all other casei^ ' 
as the nouns do — ^, ^j^lf , 7t, and the rest ; it is therefore fair 
to assume that they use the same signs in the genitive. Chand 
»uses IRt and l^t as genitives as well as other oblique cases.; but 
he knows already ^1^ and ipiTTTj ^isi^g> however, the former 
rather as a pure genitive, the two latter where possession requires 
to be clearly indicated. 

Thus iRt and the genitives in '?[T are contemporary forms, but 
perhaps with a slightly different meaning, the former being a pure 
genitive and oblique, the latter possessive adjectives. In 0. and 
B. ^ forms the genitive by adding '^; without this it is merely 
the crude form of the oblique cases. That this "^c is shortened 
from ic^ was shown under the noun ; and as in the Oriya noun, 
so also in the pronoun, this ;s^ is found in full in the pi. 
^l-i|4<) as well as shortened i|4^<. Old-Bengali has a geni- 
tive ipf^, in which, as in the genitive of the noun, we have the 
shortened form from iB|f^, with dropping of the '^. 

I see no reason, therefore, to assume any other origin for the 
genitives of the first two persons in '^ than that assigned to the 
similar form in the noun ; for even if we were to admit 4|^i( 
and its cognate forms to be the origin of the possessive pronoun^ 
we can find no parallel forms for the similar genitive in T of the 
noun. This X genitive is an obscure and rustic but undoubtedly 
ancient form, which has only recently been brought to light, 
first by myself and subsequently by Hoemle. It was not known 
to Bopp or Lassen; had they known it, they would probably have 
abandoned the Mi(\M theory. 

The only point in support of that theory is the curious Panjabi 
genitive pi. 'irBTTT asddd ; but the Panjabi plurals of the 1st 
and 2nd personal pronouns are formed upon a different system 

Digitized by VjOOQ IC 


to those in the other languages, and an argument derived £rom 
it would not be fairly applicable to them. 

Popular Prakrit has such forms as ITf ^'C^, iriym which ^^ 
would naturally flow ; and the rustic form of the same^ namdy 
411 0"* would equally be derived from ?rf 1^. So also in the 
plural, where Marwari VfTTt is IFf + 'RTt> and ^THCt for 
^f lO"* through a well-known Old-H. (flflO from ^Ff + 'IT^, 
where the '^ has been dropped. Documentary evidence is cer- 
tainly wanting for earlier forms, but this is because the modem 
forms were already in use at the dawn of mediaeval literature. 

§ 67. For the pronoun of the third person, a direct descendant 
of the third person of Sanskrit is not always to be found, its 
place being generally supplied by the near and far demonstrative 
pronouns, which are reduced to their simplest elements ^ wd '^ 

It is first necessary to pick out the few traces remaining of 
the genuine old third person, which in Skr. starts from a stem 7f , 
making, however, its nominative case W* ^., ^/, ^». Bopp 
shows (§ 341) that there is not in Sanskrit a pronoun of the 
third person with a purely substantive signification, but adduces 
the stem ^ as having originally occupied that place, and given 
rise to the Pr. %. It is not our province to go beyond what we 
find in Skr., and it is sufficient to remark on the traces of the 
stem 7{ which stiU exist. In Hindi we have ?^ " he," but often 
used as a correlative pronoun, answering to the relative ^. 
Another form ^ has now become an enclitic particle, but in 
Old-Hindi we find ?ft^ and the oblique form T^^ iRT, ^. This 
answers to ^if interrogative, and ift^ relative, and aU three 
forms arise from compounding with the pronominal stem the 
adverb IHI, Pr. ^9^, so that ^ftif is Tft ^^'T- Although ^: 
means strictly "again," yet in the modems it has changed its 
meaning; and when used as an enclitic, means merely " indeed," 

Digitized by VjOOQ IC 


or nothing at all beyond a little additional emphasis^ like the 
Greek [lev, ye, Se, etc. 

^ is used as a correlatiye in most of the languages, as in 

"Whoso drinks (it), he dies." 

"Whatsoever enemy attacks thee, he falls dead." 

—Elliot, EaeesN. "W. P., vol.,i. p. 66. 

In Panjabi it is used in precisely the same way. So also in 
Sindhi, where it occasionally stands alone; as — 

^ li^ ^ ^WTF ^ %f^ ¥t Tns^ II 

"He is this, he is that, he is death, he is i^Hah, 
He friend, he breath, he foe, he helper." 

Trompp (quoting Sh. Kal. i. 19), p. 205. 

In Marathi 7ft is sometimes used to mean " he," but its real 
meaning is " that," the far demonstrative ; it is not the third 
personal pronoun, but an adjective varying in gender according 
to the substantive with which it is connected — 7ft m., 7ft/., ^ ^• 
It may therefore be postponed till we come to the correlative. 

Gujarati, however, uses 7f , pi. TNJt, as a substantive pronoim 
= "he." Bengali and Oriya have% "he," not unfrequently, 
however, used as an adjective "that." These fQrms show a 
softening of the o of Pr. ^; Nepali 7f . 

The oblique singular is H. fTRT ; P. fTRT, fHf ; S. (Sf^f; G. 7^; 
M. IHT m., fTf /. ; O.-B. TTT^, Tff ; Nepali 7!^. All these forms 
come from the Sanskrit genitive 7r5ff, Pr. TIW- Chand uses 
7ra and THM, in the latter of which the long vowel is compensa- 
tory for rejection of one of the two consonants. He also uses 

Digitized by VjOOQ IC 


^rrff and shortened fM|f, forms resting on the Apabliransa 
Pr. ?n{> mi and ivff (Lassen, p. 482). Thus— 

" (He) of whom there is no body^ 
TTim how can one seize?" — ^i. 16K 

"(Him) whose arms are thunderbolts, who crashes the hosts of foes, 
Of Awi» the glory Chand speaks." — ^i. 46^ 

Here the u in tdm is a mere metrical tag. Even in Chand's 
time, however,^ this word was used as an adjective, and instances 
will be given tmder the section treating of the demonstrative 
pronouns. Bengali has in this third person, as in the other two 
persons, seen fit to have a finer term than the simple ordinary 
%, and for this purpose it takes t?rf'f> which is really the 
oblique plural corresponding to the oblique plurals of the other 
languages— H. finr, P. f^^iit and fn^, S. ?!fir, M. Wt 5 all of 
which come from the Pr. genitive 7{J^ with the long d shortened 
into a and still further into i. Bengali having made fiffif into 
a nominative singular, has struck out a new form f(f^ for the 
oblique, differing only from the ordinary oblique TH^ by the 
insertion of anun^ika in the first syllable. Oriya, on the con- 
trary, has for the nom. plural no organic form, but adds to the 
singular its modem plural sign ifT'tj making %in%; for the 
oblique it uses either an organic form ?rrf^> ITNT or ^H\A^i,, 
a genitive of the modem fashion. 

§ 68. The demonstrative pronoun falls into two divisions, — 
that which indicates a person or thing either present or near at 
hand, and that which indicates a person or thing absent or at a 

Digitized by VjOOQ IC 


distance. The former in all Aryan languages has the type i[, 
the latter ^ or ^. Thus in English, " this" and " these" have 
the palatal vowel, as contrasted with "that" and " those," which 
have the guttural and labial vowels. It would, be foreign to our 
present purpose to make any remarks on the evidences of the 
universal distribution of these two types throughout the range 
of the Indo-Germanic family ; but it may be observed that it is 
not confined to that family, but is foimd more or less in all the 
languages now classed as Turanian ; and if we are to admit the 
fundamental unity of these two groups, this uniformity in so 
elementary and radical a part of speech is highly significant. 
Thus in all the Dravidian languages i is the near, and a the far 
demonstrative, while u holds a middle place between the two,— 
as in Tamil idu, *'hoc," adu^ "id;" Telugu idi and adi; Cana- 
rese idu, adu ; Malayalam ita^ ata} So also — 

Tamil wa» "hie," avan "ille," ii;a/"h8BC," ara/" ilia." 
Telugu, indu « hie," vandu " Ule," idi *« haec," adi « ilia." 

To return to our own special subject. The seven languages 
exhibit the following range of forms : — 

I. Near Demonstrative — "this." 


Hindi ^,i:f ,H,TIS,^ ^[ir ^ ^T, T^. 

Panjabi 1[f , "Hf l!^* Tf Kf ' "W T^' T^- 

Sindhi t;, ft, % T'f* tff ft>\ ffft. 

Gujarati % (H) % (H) %^ %lft. 

Marathi fTtn.,lft/.,lfn. WT, ^ j^'^^T'^''} W^' '^• 

Bengali 1^ l[fT H i:it- 

* CaldweU, Drayidian Comp. Gr., p. 333 ; CampbeU, Telugu Gr., p. 77. 

Digitized by LjOOQ IC 


11. Far Demonstrative— "that." 


Hindi "9,^,^ ^W % ^f, ^^. 

Panjabi ^ ^W, ^ ^ ^iff , ^5^. 

Sindhi ^, |f, ft ^fT, IHT If , ft irfir. 

Gwjarati ^ ^ Hflit HHSt- 
Marathi deest. 
Ofiya deeit. 

Bengali ^ BfT ^ ^3ft- 

Marathi and Oriya use the correlatiye, which is also th^ pro- 
noun of the third person^ M. 7ft> 0. ^, instead of the far 

In Sanskrit, the primitive type i[ is overlaid by accretions, 
which render it somewhat diflBcult to identify (Bopp, § 360). 
The form which this pronoun has assumed in classical Sanskrit 
is ^R^ m., J^f., XJ^ n. ; and in Vedic Sanskrit there is a form 
^ neuter, which, however, is used merely as an emphatic 
particle. Scenic Prakrit has nom. ^ml m., fit/., 1[l^ w., and an 
oblique base XjR, also l[lj (Lassen, p. 326; Weber's Hala, p. 55). 
The ordinary Apabhransa of the plays has X<^, and in the 
songs in the fourth act of the Vikramorvasi are found loc. pi. 
1[^ (Skr. HJ) ; % "of her" (Skr. ^R^TT); %iT^ ^ % 'iHlN^ 
"and excessive is her affection;" ip«f> abl. sing, "than it" 
(Skr. "^ITOT^), and other similar forms. But it is perhaps 
useless to seek for the origin of the modem forms in any written 
works. They have their origin in all probability in a much 
lower stratimi of popular speech than ever found its way into 
writing before the time of Chand, whose forms may therefore 
in this, as in so many other instances, be taken as the furthest 
point to which researches can at present be pushed back. 

Digitized by VjOOQ IC 


The types of the demonstrative in the Prithirfij Rasau are the 
following : — 

N. Sing. l[^ ^^. 

Obi. njff mf%. 

Emphatic 1[^ ^^. 

N.PL %,ff ^• 

besides variations caused by the metrical necessities of the poem, 
and having no value as organic forms. 

*'To me this scripture is clear." — ^iii. 28, 2. 

Both the nom. pi. of the far and the emphatic singular of the 
near demonstrative are shown in the lines — 

** They ply their swords, he catching them in his mouth bites them." 

— i. 254, 5. 

**For what cause have you taken up this religion?" — ^i. 172, 6. 

"To complete this is a work firmly resolved on." — ^i. 87, 6. 
Shortened to nff and oblique in sense: 

'*In this fashion Anal uttered (his) speech." — ^i. 155. 

The ordinary modem form of the oblique, as in H. 1[?J, like 

the oblique of all these pronouns, appears to lead back clearly 

to the Skr. gen. "^rs, Pr. ^iw; and the forms l[f , l[fT, etc., 

' with the older forms in ff , as iftff , ifrff , TTTff , ^Tlff , ^ff > 

Digitized by VjOOQ IC 


show the ordinary oblique of the noun, produced by a general 
fusion of all the oblique cases of the old inflectional system ; 
while the plural oblique in if arises from the Skr. genitive <4||'i|f, 
Pr. 'mrf. Sindhi, however, has its singular oblique in il, and 
thus differs from all the other languages. This seems to derive 
its origin from the Apabhransa forms of the singular oblique 
. l[nr and l[nf mentioned above. 

In practice, no distinction is observed between these two forms 
of the demonstrative and that of the third person, the demon- 
stratives having to a great extent usurped the place of the latter, 
and being freely used to signify " he." 

To trace the steps by which the various languages have arrived 
at each of the slight modifications of the general type by which 
they are distinguished, is a task impossible in the present stage 
of the inquiry. Much might, doubtless, be accounted for by the 
phonetic tendencies of each language, but this would not be 
sufficient to explain the whole. In the absence of a continuous 
chain of literary monuments which could be trusted to reflect 
faithfully the actually current speech of their respective periods, 
it is beyond the power of any scholar, however laborious, to work 
out all the steps of the problem. Experience has taught us that 
in India literature is never a faithful reflector of popular speech ; 
and all that can now be done is to point out that one imiform 
type imderlies the whole range of forms in the whole seven 
languages, and that this type can be traced back to the earliest 
stage of Aryan speech ; while, at the same time, nothing more 
than the general type, the two leading ideas of i for the 
near, a and u for the far, demonstrative, can be with certainty 

§ 69. The relative pronoun meaning "he or she who," 
" that which," and requiring a correlative or answering word 
in the second clause of the sentence, is indicated in Sanskrit 

Digitized by VjOOQ IC 


by the type nj, whkh in Pxakrit, followed by the modem 
languages^ beoomes iif. The scheme in the seven languages 
is this: — 


OBL. smo. 


OBL. P^i. 

Hindi ^ 




Panjabi ^ 




Sindhi ^m.,^/. 



^rf^r. ftif*f- 

Oujarati i) 




Marathi ^m.,^f^n. Wlni.,f^f. ^i»^«rT/->Wt«- ^• 

Bengali % ^TUT % HffT- 

Hindi has also ^fi[^, Pr. ^ ^m, which is, like ^ft^, produced 
by the incorporation of tpii: . From this source, also, springs O. 
%^, in Old-Oriya sometimes written i|^, wh^e the guttural 
nasal ^ is the first step in the weakening of nr to animd^sika* 
The relative portion of this compound word takes in 0., as it 
does also when uncompounded in O. B. and G., the MagadM 
nom. sing, form 5}, which makes it identical in form with the 
nom. pi. The oblique forms in all the languages are from the 
genitive sing. JSkr. ^nB> Pr. ^n5^, with weakening of the 
vowel, and Skr. gen. pi. i|H( (instead of the classical Skr. ^ift , 
which has not been preserved), Pr. ^IQ. Old-Hindi has its 
oblique singular f^ff, which is more modem in type than the 
Modem-H. f^; this latter, however, was probably in use con- 
temporaneously with ^lf|[, as we have in Chand forms ^R and 
^TW. The plural in Ghand takes an emphatic addition, and is 
t^^; but this seems to be restricted to the instrumental. 
Bengali takes this plural in the shape of f%rfif , and makes it 
into an honorific singular nominative. 

VOL. II. 21 

Digitized by VjOOQ IC 


§ 70. The correlative is based on the Skr. ;g: and is the same 
as the pronoun of the third person mentioned in § 67. Many of 
the forms were given in that section, and all the others may be 
made on the model of ^ by substituting "if for ^. Nothing 
further need be said about it, as it is precisely homogeneous to 
the relative. 

Occasionally an emphatic form of these two pronouns is 
used formed in H. by adding ^ or f^, as ^ft^* Wtlft, but 
in Sindhi only ^ is added. It will be observed that, of all 
the languages, S. and M. alone treat these two pronoims as 
pure adjectives, and give them the distinctions of gender. 
S. ^ m.y ^ /., but in pi. only % for both genders. M., with, 
its customary redundance of forms, has all three genders for 
the nominatives of both numbers, but in the oblique singular 
only m. and /., and in the oblique plural only one form for all 
three genders. 

§ 71. The interrogative pronoim is just as uniform as all the 
others: the only difference is, that forms which, though they 
exist in the other pronouns, are in them kept rather in the back- 
ground, here come to the fore, and displace in common speech 
the forms which correspond to those more frequently used in 
the others. Thus, in the relative and correlative, ^ and ^ 
are in Hindi the commoner, ^a^^ and 7^ the rarer forms; but 
in the interrogative, ^^if is the ordinary form in modem use, 
while ^ is archaic, poetic and dialectic. The neuter, also, has 
a form of its own, whose origin ascends to a different Sanskrit 
word from ^. The type of the interrogative is everywhere 
^, just as iif is of the relative. The table of forms is given, 
because, although exactly corresponding to that of the relative, 
yet the exhibition of the whole set helps the eye to make the 
comparison, and brings out more clearly the symmetry of the 
pronominal forms, which is a striking and beautiful feature of 
this group of languages. 

Digitized by VjOOQ IC 


NOM. Snra, OBL. 8INO, NOM. PL. OBL. PL. 

Hindi 1|^ f^ js^ f^» f^. 

Panjabi ^1J f^* f^ ^fNf f*ift»ftlft. 

Sindhi ^JhlT n. i|5f|{ — ^rfif. 

Marathi ^jfrtl qTiuil 'frtF 'frtit- 

Ojiya ^, % lITfT %i?T% 1T^- 

Bengali % HTfT % iffT. 

Sindhi has iJtF only in the neuter; its plural nom. is not 
used ; and in commoner use both in Sindhi and Panjabi is the 
formS.%1 w.,^/., P.'^tif^ w., of;/., obl.%|%, which is 
from Skr. ^SVfl|^ Pr. %f^^ and later ^|f^ft> which in S. 
merely drops the ^; while in P. the T+W ^^^ ^®^ moulded 
into If, and then again split into ^4 1[. Old-Hindf preserves 
the symmetry by using nom. ^, obi. "ftrff , as — 

^ ftlff ^tirff xsm^ n 

"Who (am I), from what race sprang?" — ^i. 14T. 

Here, again, comes in the Skr. gen. m^, Pr. ^11W> as in the 
relative. Gujarati has singularly introduced a hiatus, writing 
i|^ instead of ^ ; this seems to have arisen froln a form w(ty 
which wiU be explained hereafter. The forms with Tjm added 
are here more widely used than in the other pronouns, perhaps 
because of the somewhat greater emphasis involved in asking a 
question. In all languages "who?" more often stands alone, 
almost like an interjection, than any other pronoun, and thus 
the Pr. '^Tj, which has sunk into an enclitic, would be more 
frequently used with the interrogative. 

The neuter stands alone in all but a few exceptional instances, 
and is as follows ; 

Digitized by LjOOQ iC 

824 THE PHONoxrir. 

Old-U. fintT, oblique ttf^. Modem-H. t^, obi. ^f^ and fipi; and 
th^ samiB in the plural. 

P. ^» ftnn, obi. iirtr» ^if • 

M« ^rr^9 obi. ^E|1(T» ^1^9 pi- the same, but obi. l|1|t> with the usual 
anunftsika of plikrals. 

B. ftl, t*Wl. 

The origin of all these forms is to be sought for in th^ Skt. 
fill, an old neuter. B. fii^, H. "ftw, P. UTO, refer back to the 
genitive, which in Pr. is ^IIW, sometuned also ift^. H. ^RfT is 
apparently a conflation of the oblique forms g. ^, abl. is^, loc. 
Hf\j, and ^IT the ordinary modem form, from supplying the 
hiatus of a form ?irW by Jl instead of ^, as in Gujarati. 
Prakrit has also an oblique ^rr^> whence Sindhi %ft by soften- 
ing of dtoe; from its retaining the d in the first syllable the 
Hindi oblique ^|f^ exhibits a form which postulates a Prakrit 

Gujarati has an unique interrogative ift w., ^/.^ ^ n., the 
only approach to which in the rest of the group is S.*^, used 
only as a neuter. Vans Taylor (p. 73) refers us back to the 
acknowledged alliance between ir and ^ in the leading Aryan 
languages, as in Skr. '^'C, Greek tcvwv. But it is important to 
observe that these greater phonetic laws work only in the sphere 
of the larger groups of the Indo-European family: within 
the limits of any one particular group, their working, if it exists 
at all, is very feeble and restricted. It is beyond a doubt that 
Sanskrit exhibits words containing ^ which are weaker forms 
of an older word with nj , the stronger form of which has been 
preserved by the cognate languages. But when once the parent 
language of the Indian group has preserved and stereotyped a 
form in 9, it is not found that its descendants modify this if 

Digitized by VjOOQ IC 


ii^to n; an initial n stands firm for the most part^ at least a 
cliange from it into j[ would be of too radical and wic^e a nature 
for the modem Indian languages, which do not deal in those 
vast organic changes which were at work in the infancy of the 
world. We must rather look for the origin of this form by the 
light of changes of such a nature as are prevalent within the 
restricted limits of our group, instead of hunting up aU possible 
modifications in every coimtry and age to which the Aryan 
speech has penetrated. If we can with difficulty find a single 
dubious instance of the change from ^ to ^, so frequent in 
Sanskrit, among its descendants^ we should hardly be justified 
in going ftirther back still to search for a change, which is in 
point of time prior to Sanskrit itself. 

The origin of ift is to be found in the earlier form m^^ a 
neuter signifying "what,** from which have been constructed 
a masculine and fepiinine l|^ and ^injfVy which are also i^sed 
as an indefinite pronoun, and, as will be seen presently, the only 
form in use for the oblique plural of the indefinite is ^1^. 
This form would lead us back to Pr. l||(\4\> Skr. 4V^, from 
which, as we have seen above, Sindhi and Panjabi draw their 
interrogative. ^UPwt becomes in H. %^, and in Old-Hindi 
the forms are ^Vl[^ and ^rerT^ the latter produced by elision of 
the short i. With this last-named form our Old-Gujarati K^ is 
identical, for 1( is not in this language necessarily the palatal 
sibilant, but rather the distinction between it and 9 having been 
obliterated, and only one sibilant sound remaining, the letter used 
to express that soimd is sometimes B, and sometimes If, according 
to the habit or caprice of the scribe ; so that we might here also 
compare the oblique in M. ^RHT or "^S^, instead of treating it as 
from a Skr. gen, li^, Pr. ^^W, which would not account for the 
final long d. Sindhi ^ is probably also of like origin, 1^ being 
often interchanged with il in all the modem languages; or ^qt 
might also have lost the a of its first syllable, and become '^, 
whence the transition to ^ is in accordance with the usual law, 

Digitized by VjOOQ IC 


It is more probable, however, that both .in S. and G. there has 
been loss of initial n, as it is not in harmony with the general 
phonetic laws of this group to suppose the creation of ^ from 
9+^ at so late a date as we must assume, to bring it posterior 
in point of time to the tenth-century iraT* 

It may here also be noticed in passing that the derivation of 
the forms ift^, ^ah^, etc., from ^ ^8R, ^ Ifif , suggested by 
Taylor, is erroneous. These forms are written in the earlier 
stages of Hindi ^m^ and ^i^if, where the labial vowel and 
semivowel are indicative of the ^ of yf:. The compound 
phrase ^ ^3H[ is not a conjecture, but is constantly found in 
Prakrit (Lassen, § 32). 

§ 72. The indefinite pronoun deviates from the homogeneous 
type of the other pronouns, and this deviation is explained by 
its origin. The forms may be given first, and analyzed after- 
wards. The typical letter is n, as in the interrogative ; and 
the neuter, as in that pronoun, stands apart from the masculine 
and feminine. The word now given means " any one.** 

NOM. 8IN0. 

Hindi ^ 


Panjabi ijitf; 

ftral.ftpft H 


Sindhi If."" ^ ^ 

m ^ 


Gujarati ijti; 

^ *T 


Marathi iitl|t, ^St^ 

litwr.^fniST 'itift 


Oriya j'«^'lt|^^'j 

«ffiR — 


BengaU %f 

WWi — 


These forms arise from the compoimd Skr. ^^fiv (^ ^ift) ; 
the enclitic particle ^fft in Prakrit slides into composition with 
the pronoun, and is written in one word ^f^, from which, by 

Digitized by VjOOQ IC 


elision of the ^, come the H. P. S. and.Q-. forms. S. often, 
however, rejects the final ^, which is lengthened in the other 
languages to preserve it, and because it is emphatic. M. has 
merely added the emphatic particle f^, with 1[ lengthened to 
^ and the f frequently dropped, to the modem form of the 
interrogative, so that we have a compoimd of three parts, 
^H-^ISr + fW' I^ Oriya, the final syllable ft appears to be 
shortened from the demonstrative % ; so that here, also, there is 
a triple compound % + ^IJT + %> literally "who forsooth is he ?'* 
0. %f^ and B. %^ are allied, and simply add f^, which may 
either be the emphatic particle of Skr., or more probably the ^ 
was inserted to prevent hiatus after dropping the ^ of "^fcft 

Old-Hindi has ^i\n3f and oblique Wlg> where the ^ or ^ of 
iitf% has softened to the labial vowel, and the final short i has 
been dropped, as is usual in Hindi. The oblique forms (c||(i| or 
fc|^ show the oblique of the interrogative with the ^, whose 
origin has been forgotten, so that it is regarded as a mere 
emphatic particle. Marathi ^lt^(3^ arises apparently from the 
fact that d is regarded as the general type of the oblique, and 
has been added without reflection. In ordinary current speech 
it is customary to add Tpi " one" to this pronoun, so that they 
say ^St^CfT Tpirnr "of some one ;" and the same practice pre- 
vails in the nom. pi. of H. ; thus Hll[ Tf^ or %l[i|| some " aliqui, 
aliquae.*' The Oriya ii|^|fX is a curious instance of how these 
forms arise : 'RTfTT is the genitive of the interrogative, and by 
adding the emphatic 1[ or t|[ to it we get ^T^TTft ^^ <(il^|fX', 
which should be used as a genitive only, so that the oblique would 
be ^innT y ^^^ *^® genitive form has been extended to all the 
cases, and they now say ^|||ir< ^ " to some one," ii|i|f |(\ "ZW 
" from some one," and so on. 

The plurals of this pronoun are seldom used in most of the 
languages, and in B. and 0. there are no plurals at all, the 
singular doing duty for them. 

Digitized by VjOOQ IC 


The neut^ form '^Qomething'' is expressed as follows : 
Old-H. 11^, Modero-H. ^; oblique, Old-H. WW* Mod.-H. fi|F|^ 
bnt the obUque form is rarely used colloquially. 
P. ^^» rarely if ever inflected. 
S. l|V, not inflected at alL 
G. <^ 

> do. 
o. fiiftf 

B. ftrf 

H. P. 0. and B. have a common origin from the Skr. fijif^^, 
the final ^ of which is dropped in Prakrit f^fcf^. The archaic 
and poetical H. Hfim seems to point back to a form in which, 
the enclitic t^^^^ had been afl&xed to a neuter m^, instead of 
fijiy thus making ^rf^c^; from 1V9, the u has leapt backward 
into the first syllable, making the modem ^f^. The change 
from ^ to H is hardly to be accounted for by absorption of the 
anuswftra, and must remain unexplained. The three western 
languages, S. G. and M., do not seem to have any connexion 
with ftsf^T^t^; but the last two use oblique forms of the nuis- 
cidine indefinite, and the first merely a lengthened form of f^ 
with loss of the anuswftra. 

§ 73. The reciprocal or reflexiye pronoun "self" is in most 
of the languages a derivative of the Skr. "limT "soul, self." 
As a substantive it means "self," and as an adjective "own." 
The former is 

H. ^w^, P. ^w^, 8. xfpn, G. ^im, M. ^\h^, o. ^^, ^inniT> 
B. "^iiMfn. 

The principle of phonetic change which lies at the root 
of these modem forms was indicated in Vol. I. p. 330. The 
process began in Prakrit, as we have 'W^, UTOT^ side by side 

Digitized by VjOOQ IC 


with ^J^, etc. This latter form has pot been adopted by the 

Hindi, followed by P. and G., uses the simple nominative, 
rejecting one of the two consonants, and lengthening the pre- 
ceding vowel in accordance with the regular practice in the 
treatment of the nexus. B., as in jfii, ^Pl, t%(il, etc.-, has 
made for itself a nom. singular out of a plural form. The other 
languages retain a final Hf or tf, which rests on the Pr. form 
in^rar (^ar. v. 45), which by his next following sdtra Yara- 
ruchi extends to all other Sanskrit nouns in n, as "^p^ == ^^|IQ, 
etq, Sindhi adheres most closely ta the Prakrit, merely reject- 
ing the first syllable; but the other languages, while they retain 
the initial ^, lengthen it to ^, and shorten the second syllable. 
I am disposed to think that this alteration of quantity is only 
apparent, and that tl^e forms ^9inilir> etc., are really derived 
from the oblique cases of the Pr, singular, as instrumental 
^rnnin, gen. "W^nft- It is observable in the modem Romance 
languages that where the type of the oblique differs from, that 
of the nom., the modem language adopts the former for all 
<»8es; thus we see in Italian monte, where the Latin nom. 
is mom, and the t occurs only in the oblique cases. The latter 
being used five times to the one of the nominative, naturally 
apquires the predominance in the vulgar usage. 

Uni is declined with the usual case-affixes ^, i|t> etc., in 
Hindi, when used as w honorific substitute for the pronoun of 
the second person, so also in P* and the other languages ; but it 
has a special genitive used, adjectivally and with a possessive 
sense, meaning " my, thy, his own,'* according to the person in 
which it is nsed. This therefore becomes almost a separate 
prcmoun, and haei the following foyms : 

H. ^H«ll fn., ^rnft/., 'W^ obi. sing., ^HXRT obi. pi. (raro ^IJifT, etc.). 
S- xrtf ^, ^, etc. 

Digitized by LjOOQ IC 


M. ^unr^; •^j •%, etc. 

O. imwrr^ (rare, f*!^ « «8«al)- 

B. mi\9(. 

These forms are derived from the Prakrit genitive, which is 
^(nmft, and the endings for gender and case seem to have been 
necessitated by the use of the word as an adjective. Hindi 
goes a step further^ and uses the oblique form, just as though it 
were the regular oblique of an adjective using ^^^^ ^, or 
more frequently '^jxnl ^ = seipsum ; ^Rqi} "if and "IpJ^ IHC 
may also be heard in conversation, as in the phrases ^^T^"^.^ 
5BWr "he thought thus in himself," nnnSf iTT ITf f^irm "he 
took the business on himself," ^W^ % ^THTi l^t "take 
counsel with your own (friends)." These forms, though common 
in the eastern Hindi area, might perhaps be set down as un- 
grammatical by authorities on the language, these gentry being 
apt to be capricious and fastidious, and prone to brand as wrong 
any phrases which they do not use themselves. 

There is also a curious word ^inre allied to this stem, and 
used in H. P. and 0. always with the affix of the locative, as 
H. ^inre^, 0. ^UTtri^ "among our-, your-, them-selves," as 
^Unro'il Hf£\ "divide it among yourselves." No origin for 
this form can be found in any of the Prakrits, nor can the 
form itself be considered as a locative apart from the case- 
affix. It is the case-affix which contains the locative idea, and 
when we remember that "if is from if^ " in the midst," we 
shall see that the complement of the idea involved in the whole 
phrase is a genitive " in the midst of selves." I therefore hazard 
the conjecture that this form is like l[ir, ^9, f^> and the rest, 
a genitive irregularly formed from a vulgar Prakrit ^HHiWy 
which would postulate a Skr. "^m^nQ* Now though no such 
form exists, or ever did exist, yet we have seen in the case 
of the noun, that the varied inflections of the numerous Sanskrit 

Digitized by VjOOQ IC 


nominal bases have all been rejected^ and the declension of the 
a-stem taken by the modems as the type of all nouns; it would 
not therefore be at all out of keeping with analogy, but rather 
much in keeping with it, for the modem languages to have 
in this instance also taken no heed of the peculiar forms of the 
declension of ^m^T^, but to have treated it like an ordinary 
noun of the astern, and made a genitive in ^, as in all the 
parallel cases. In my own mind there is not the shadow of a 
doubt that this is really what has happened, and this form 
may be set down as one more illustration of the admitted fact 
that a type in common use is often extended, to all classes, 
superseding entirely the minor types, and securing simplicity 
and uniformity in the place of a multitude of divergent 

§ 74. Although the pronouns, other than personal, above 
enumerated, are all, to a certain extent, adjectival, yet in several 
of the languages they have lost those variations for gender 
which mark the true adjective ; H. lft> for instance, is used 
both with masc. and fem. nouns, only in M. and S. are they 
treated as adjectives with separate gender forms. There are, 
however, certain pronouns which are adjectival in all the 
languages, and exhibit in their initial letters the types of all 
the above classes. They express quantity and quality, con- 
sidered demonstratively, interrogatively, relatively, and so on. 
As a type of them, the interrogative is here given, the whole 
series being exhibited in full in a future section. 

"How much?" qmntus? <*0f what kindP" quails? 
Hindi ftW^, ^^ %^j o^. 

Panjabi f^RTPn^ ^^ ft8fT,%?Tj •ift* 

Sindhi ^fiPCt, •'?^ ftlf^j •^• 

Gujarati %Z^, ^^(t, ©^ *Mtj •'ft, •^. 

Digitized by LjOOQ IC 

332 THE PKOI70UI7, 

Marathi ftnWJT, o|j^, o*^ mn, 9ig\^ %*. 

Marwarf f^Bn^, •'O ftT^j •^^ 

0, rad B, haying lost all geuder, do not inflect the adjeotiye ; 
they use the following; 0. iJRt "how much/^ B. H?|. The 
former series, that indicating quantity, is based upon Skr. f^RlTil^ 
(w,)> fSw^ (/.), ftW^ (».)> vhieh in Pr. becomes %f%^ 
(Var. iy. 25), also %^. The oldar Hindi form is %?n> which 
i$ aA inunediate descendant of the Pr. form. Thus Ohaud-^ 

'^How many men and Eajaxshis haye there been, (and) gods and 
demons of you?" — ^i. 162. 

The addition of the affix itT ^ Bengali indicates affirmation, 
and the affix seems to haye been at first distinct ; thus in Old- 
Bengali irr regularly follows i|7f {kdtd)^ as in Bidydpati : 

iW TT ^Rfit ^^ ^^V^ ^^^ ^^ ^^ ^ ' 

* * With how much lal^our, how many wonders has fate brought to thee I ' ' 

—Pad. 1352. 

** With how much labour, how much soever thou coyerest it, the snowy 
mountain will not be hidden." — ^Pad. ih. 

In fact, in the eastern area and in Orissa if and ifT are ^lot 
negatives only, but affirmatives also, the sense depending on 
the sentence or on the tone of the voice ; thus in Oriya : 

"Will you go to my house? Yes, I will go." 

ifT is probably in H. a diminutive (§ 24), and with this 
agrees the S. and the Marwari fSfccfO', where xt is also a 
diminutive ending, as is also the ^ of G. ^SZ^, and so 
perhaps is ^ of the Marathi. This ^ is a common addition 

Digitized by VjOOQ IC 


to the pronouns of this series ; thus we often hear in eastern 
H. fli^, and in M. f^Rft^V; it is, I think, erroneous to con-* 
nect this 1V> as some grammarians do, with IJJBii, supposing ytJBH 
to be used as unos, unas, in Spanish, in the sense of ^^some." 
The i( appears to be the sign of the nom. pi., and does not 
belong to the affix> which latter consists simply of ^, a longer 
form of which is %, and a longer still ^if^, from which % or 
m^ is made by omitting the '^. Both forms occur with the 
pronoim. %^ in one passage of Chand — 

"Of whom there is no body, him how can one seize?" — ^i. 161. 

** Where sight does not pierce, there how can one isee?" — ih. 

^^"^ and 1|i% i|if^ literally mean "what kind (of work) having 
done,*' and thus, too, f^%^ would mean fi^ Hjf^ " how many 
(works) having done." In the form%H<9| "some," however, 
the latter element may be admitted to be Jiss, and it is often 
written as two words, especially in modem Urdu uX>l ^. 

The group %^, etc., comes from Pr.'^Jsf^;^, Skr. ^^^^ as 
has been already stated. P. and S., more suo, change ^ into ^,. 
making ^R^ and shortened fi||^, to which S. adds the diminutive 
ending ^. G. ^^ appears to arise from the substitution of 
V for h, which is characteristic of the llajput dialects of Hindi, 
from which G. sprung.^ 

There is another series meaning "how great" in some of 
the languages, which arises from a composition of the Pr. 
wit (Skr. ^) or wit "great." This is S.'^fiit; P.%^, 

^ Dr. Biihler^s welcome announcement of his discovery at Jesalmer of the 
Vikramdrkacharitam, or Chronicle of the Chdlukyas, leads us to hope that we shall 
now hare some trustworthy data as to this interesting race, raluahle for the early 
history of Gujftrati. 

Digitized by VjOOQ IC 


%V; G.%^; M.Vl¥T; O.^ilt (indeclinable). Thus Dtna^ 
krislina — 

'R ^m Tft ^ 1^"^3CT I 

"Thou, drunk with the wine of love, forgetting 
Art, thy waist how slight (it is). 
What, knowest thou not how small it (is). 
Thy twin breasts how heavy?" — ^Basakallol. iii. 43, 

Hindi does not possess this form, saying instead fSBTprr WWl and 
the like. Sindhi here, also, uses a diminutive %^^ meaning 
" how small P*' a form which is not in use in the other languages. 
The whole of these series will be seen in their correspondence 
with the adverbs, the generic types running through the whole 
with admirable regularity. 

§ 75. Sindhi allies itself to Persian and Pushtu by a practice 
foreign to its sister tongues of suffixing pronominal signs to 
nouns, pronouns, and verbs, a complicated and difficult system^ 
from which the other languages are fortunately free. These 
suffixes are, according to Trumpp (p. 225) — 







Into the intricate changes rendered necessary by the addition 
of these suffixes in the terminating vowels of nouns, etc., it is 
not necessary here to enter, f^ is undoubtedly the Pr. gen. ^, 
and is still sounded me in Pushtu, but m in Persian ; l[ from 
Pr. B by elision of 7{, so also the third person ft from %. In 

Digitized by VjOOQ IC 


the plural, ^ or jf, which appears to be used only with pro- 
nouns, is by Trumpp referred to the Pr. form ^Fft (rather 
^'H^), which, howeyer, hardly accounts for the anun&sika: a 
more probable origin would be from ^VfTW^ reduced to ^rt^, and 
thus to ^ and jf; the second person ^ recalls Pr. 'eft* and the 
third person fif from ^[HTHSfj ^ I*^. gen. pi. of the base l['^^, 
which we have already seen is substituted for that of the near 
demonstrative used as a third person. 

The geperal rule for the method of adding these suffixes 
appears to be that the noun undergoes merely euphonic changes 
of its final vowel, but not constructive changes, and the discus- 
sion of them pertains rather to the study of the individual lan- 
guage than to that of comparative grammar. The Old-Hindi and 
Bengali, and to a certain extent also modem colloquial Oriya, 
exhibit the beginnings of a tendency to this system, which, 
happily for those who had to speak the languages, did not get 
beyond the first stage, namely, using the crude form of the 
personal pronouns, as in Chand's ift f^^ " niy father," 7ft ^H^ 
" thy feet,'* and in Mod.-Oriya ^ ^PC " ^7 house.'* It wanted 
but the shortening of the vowel and the change of position to 
load us with forms like ^T^, ft^, and the rest. 

§ 76. Not only do the substantive and adjective pronouns run 
in perfectly analogous series, each distinguished by the typical 
initial consonant or vowel, but a long string of adverbs also 
follows this analogy. At the risk of a little repetition, it will 
probably be as well to group together, so that the eye can take 
them all in at a glance, the whole of these concurrent forms, in 
a series of tables. 

Digitized by LjOOQ IC 


TItB PE6N0tl^. 





f I 

$ : : 

lee's I ^f 

I ' 



s fe 

S £ 










II ^f^ 


w d S 2 •" .4^ 
.2 ** 2 .2 «^ 8 'J 

r ? ? ? I -a " 


I ^ 



i 2 S S "S 

1 1 








^ s 

J - 

1 1 


5 8 

S 3 

I a- 

S S V 

p. r tfi 

Digitized by VjOOQ IC 






at ^ 

slfEieislll «,\<S:%tl%» 


«illSjl'-S «lllB3?i« 


« ^ 

m%mmm% » ?|^ltJc<&<tfci^^ 

^ I 

i?lllfl$'l iSt^lltliiS'l 


,9 > 




'2 bo 

I 3 

« I 

■a M 

S g g 

•g -I -s 






? ? s g I I ;a a 














Digitized by LjOOQ IC 






^ 1 li 1 1 1 1 1 




1 1 1 1 1 » 1 ^ 

II II 1 1 1 


« * c « _ 


1 1 1 1 IT 1 1 i 



« 1 1 * e f 1 f 

R •••••• 

.s *^ - 

^ -g ;l • • s 

"M 5 ft© .v s *!• 
? 5 f f .^ f > ^ 

1 •£ 

1 ^ 

Digitized by LjOOQ IC 


The reader will see by casting his eyes down any one of the 
columns how perfectly symmetrical all the languages are ; how 
H. has ^, irft, W^\ ^, ^, and 1[^, ^^S[^, fWT> 
fiWT^ ftWT ; Marathi Hjt, Sfjt, ^$T> %5t ; ^^^ *t® same in 
all the other languages. 

The adverbs will be explained in their proper place^ and all 
the pronouns have been commented on above. The forms in use 
are often numerous, especially in Sindhi ; thus the demonstratives 
ifV and If are pronounced ^^ and ^ in Lar, or the coast district 
of Sindh. In Gujarati the adverb of time appears as "^fcrft, 
^\\, ^irfr* successive stages of shortening; and many other 
trifling dialectic variations occur ; but in the above tables only 
one, and that the central or typical form, has been recorded, 
except in cases where two forms appear to be used with equal 
frequency, in which case both are given. Much of this re- 
dundance of form is doubtless due to the absence hitherto of 
any settled standard of spelling in all the languages. Native 
scholars have unfortunately set themselves to improve their 
mother- tongue by the resuscitation of Sanskrit words, instead of 
striving to fix the orthography of the words really in use among 
their contemporaries ; and from this misdirected energy of theirs 
it has too often resulted that the language presents a disjointed 
heterogeneous aspect, certain parts of it, as the nouns and the 
nominal part of compound verbs, being highly refined classical 
Sanskrit ; while other parts, as the inflectional and connecting 
particles, are rude in form and unsettled in orthography. This 
unfortimate practice, moreover, hsLS misled such European scholars 
as have taken a cursory glance at the subject into supposing that 
the modem languages are far more closely allied to Sanskrit 
than they really are ; and Bengali, which, from its phonesis and 
organic structure, is proved to be a very poor and rustic patois, 
has had so many "purpurei panni*' sewn on to it, that it has 
been regarded as the eldest daughter of Sanskrit^ which has 
preserved, with greater fidelity than its sisters, the family type. 

Digitized by VjOOQ IC 


It is in truth one of the youngest grand-daughters; and an 
examinaticHi of its essential features shows that it has wandered 
further from the original forms than any of the allied 

§ 77. Besides the pronouns which have been arranged in 
classes above^ there are certain miscellaneous pronouns which it 
is di£Blctdt to classify exactly under any of the heads which have 
preceded. Such is S. ^ "all," which is declined throughout, 
so is G. ^R^9 while in the other languages it is indeclinable, as 
H. ^^, P. ^^, 0. ^, B. 1S[^. Traces remain of this pronoun 
having been inflected in Old-H. ^/Jf, which seems to be a re- 
production of Skr. ^of^. S. has also an emphatic f orm ^^^t^ 
"every one," also compounds ^T^^ and ^l^^itl^. These two 
latter are treated as compounds in S., but their equivalents are 
written as two distinct words in the other languages. In S. it 
is necessary to regard them as compounds, because ^gm being 
capable of inflection, if it were written as a separate word it 
would have to be inflected also. Unique, as far as I have been 
able to ascertain, is the curious S. adjectival pronoim. fii^l^ or 
(4i>^|^ m., ftl^rrt^/j meaning "every," which seems to have 
come from Skr. in^, originally of course meaning "mortal," 
but subsequently shading off into the meaning of human beings 
in general, just as Latin homo has become French on, or as in 
German man has lost its distinctive meaning, and now implies 
merely "they," "people in general," and the like, in^ would 
become if^ift^ and by further softening, aided by the fondness 
of Sindhi for the ^sound, it' would successively be M>^ii\ and 
fi4>^€n* The ^ is merely the emphatic increment, as is shown 
by the fact of the inflection taking place in the syllable whicli 
precedes it as it does in ^Rftf^ also, thus : — 

Norn. BiDg. masc. \ ^ ^^ } Norn. sing. fem. I _ 

Digitized by LjOOQ IC 


Obl. sing, masc, <^ ^ ^> Obi. slag, fem. 

Nom. pL com. { 

fWf^ ^ or UiiiM ii* 

Obl. pL com. \ ^ ^ •*. ^ ^ ^ •» 

The abktive plural shows the form of the synthetic abl. in 
^, with the emphatic ^^ added^ thus ^fi|^j^. Only the obl. 
sing. fem. seems to show some divergence. The feminine of an 
adjective in o generally ends in £, and its formative or oblique 
in ^, as ^^ "dry"' (m.), ^gV (/)> obl. fem. ^cf^- Wkai 
appears to have happened is, that the type of the oblique ia 
has been added to the irregular feminine in d, and the emphatic 
i altogether omitted. 

In ^QVR^ a double inflection occurs ; thus fem. ?ERrVT9 where 
?ro is fem. of ^, and iH feni. of ifft ; but in the declension no 
further change takes place in the termination of the first 
number of the compound, which remains iqr) throughout. 

Although ^m is in the languages where it occurs indeclinable, 
yet H. P. and G. have a declinable adjective from ^, which 
takes the form m. mw, /. m(^, obl. ^jf^ and like S. ^, 
means " the whole.*^ P. declines Wl{ in the oblique, as ^QRTift; 
HH^' y the latter is an instrumental, but is used as an oblique 
with the objective affix ?fT^ in the first line of Bhai Mihr 
Singh's "Panjglb ik Keshan Ejss&.''i 

^'The true Gbd who has spread out his power for all." 

Similarly Panjabi makes a declinable word out of the in- 
declinable H. ^ifr^^ "other," also used as a conjunction "and." 

^ This work was written for me by the author, and is in the purest Panjabi of 
Gnjraty a town in the most fertile portion of the Panjab, near thje banks of the 
Chenab. The M3. is in my possession, and has never been edited or printed. 

Digitized by LjOOQ iC 



This is iQ Old-H. "^I^, and sometimes ^W, whicli betrays its 
origin from the Skr. preposition yf^. In P. ^f^ means 
''other/' and it is declined in the obKque pi. ft^l' "others," 
and iftr^* I^ ^^ a^ an indeclinable form ft^^^ which is 
analogous to ^inre, and has the genitive case-ending irregularly 
tacked on to it. In such forms as Wl(^i and ft^?rt, an 
irregular if has been inserted between the stem and the 
termination. This is probably due to the tenacity with which 
all pronouns retain archaic forms; the nom. pi. of the old 
declension would have been ^9^, f^^fif, and in adding the 
plural oblique {i,e. Prakrit gen. pi. W^) ending, which is ilrf, 
it has been forgotten that the if of ^Hf^ is an inflection, and 
a new plural base ^RHf has thus been formed, to which the 
obKque has been added. The same process has operated in 
H. ^ilift? where the if of ^RifT has similarly been incorporated 
into the stem. In all languages of the Indo-European family 
such eccentricities are found in the pronoun, and had their 
origin at that stage of the progress of language when the old 
synthetical forms were breaking down and becoming confused, 
and before altogether disappearing, were being used in a way 
which would have broken the heart of Cicero to hear. Precisely 
similar to such forms as ^QR^ is the French "leurs;" the 
Latin genitive masc. illorum becomes "leur*'=" their,*' and 
like the modem Indian genitives, has become an adjective, and, 
as such, has been supplied with the plural sign 8, which comes 
from 08, the termination of the Latin accusative, so that " leurs*' 
would be translated back into Ulorumo8! just as ^Rfift contains 
the elements of Skr. ?i^Tf^ + "Wit. So also Italians, forgetting 
that loro=ilhrum is already a genitive, prefix to it the genitive 
preposition, and say di loro^^'oi them,*' as also da loro, "from 
them,'' and con loro, " with them," as if a Eoman should have 
said de illorum, de ab illorum, cum illorum. Thus language 
plays sad tricks with ancient forms, whose meaning has ceased 
to be felt or understood, Not less eccentrically Hindi says 

Digitized by VjOOQ IC 


^TO'if " among themselves/^ as if there had existed a Sanskrit 
phrase ^IHR^ ^T^. Priscian in the one country, and Panini 
in the other, had, fortunately for their peace of mind, passed 
away before such processes were invented. Panjabi uses a 
genitive case-affix with ^SIHT^y and makes ^9IT^9 ^> and an 
objective ^SIPT^ ^, also an ^.blative 'W^ ?T- When a form — 
whether derived from a nominative or from an oblique form — 
in Prakrit had once established itsdf in the mind of the homy- 
handed Panjabi peasant, he, knowiiig nothing about direct or 
oblique forms, treated it as a stem, and added the usual case- 
endings to it. So true is it that man is not the master of 
language, but merely the instrument by which the processes 
of speech develope themselves according to natural laws. Man 
in fact makes nothing ; nature makes, and man merely places 
materials in such a position that the forces of nature can work 
on them. Man collects the wood, applies the fire, and sets the 
pot on it ; but the forces of nature thus brought together boil 
the water in the pot. So man utters sounds by means of his 
vocal organs, but nature controls the form which his utterances 
shall take ; and man unconsciously works out great and deep- 
reaching developments of speech far beyond his cognizance or 

Compound pronouns are in common use, but they present no 
remarkable features. The laws of their composition may be 
studied in the grammars of the several dialects, but do not come 
within the scope of a work which deals with structure only, 
because structurally they have nothing to be explained beyond 
what has already been stated when treating of them separately. 

Gujarati has two pronouns peculiar to itself. One is T^i^, 
pi. "self or selves,'* obi. '^cfttnt the origin of which is by Taylor 
(p. 73) correctly referred to the Skr. ^fTf • ; the i^ of ^ being 
hardened to Tf and the if dropped, the visarga becomes ^, and 
the ablative affix ?T^, having lost its special signification, has 
been applied to aU the cases of the pronoun. The word seems 

Digitized by VjOOQ iC 


to liave been formed in two halyes, as it were: ^: becoming 
separately ^ and ^, and the affix 7V^ becoming separately ?|. 
The word when first compotmded must have been used ad- 
verbially, "by themselves" or "of itself;" and then, throngh 
the common custom of forgetMness of its origin, the compound 
has been treated as a nom. pi., and regularly inflected. A 
regular form ^ff is still in use, but ais a distributive, and with 
the full form iJtTTT added to it ; thus 5nit ^ft^ ^ft?n% %T T9T 
" they went each to his own house," literally, '* they his own, 
their own, house went." 

The other pronoun is ^^, a remote demonstrative "that," 
which is also written ^fflft. The origin of this word is un- 
certain. ^^Wt means "first," but I think this is a diflEerent 
word from^lft* After e the Gujaratis often insert in speaking 
an inorganic ^, so that the two vrords come to be alike. My 
own idea is that %^ is shortened from in[^9 and that again 
from xrf^wtj a secondary formation from ir^ "distant," just as 
in Panjabi we have ^5^^ " on this side," q4,iil " on that side'* 
(of a river, road, etc.), so that ^ift would literally mean 
"yonder," just as in the colloquial English of rural counties 
we hear "yond* man," or "yon man," for "that man," German 

§ 78. The language of the Gipsies in various countries of 
Europe, though its vocabulary is a medley of words taken from 
the languages of all the lands in which this strange race has 
sojourned, is purely Aryan in its structure ; and Modem- Aryan 
too, being in many respects quite as far removed from tibie old 
synthetical system as any of the seven languages now under 
discussion. In respect of the pronouns, other than personal, it 
preserves the traces of its origin very clearly. Thus we have 
the interrogative kon "whoP" kdya "what?" with its oblique 
kas, also kaUa, pi. oblique kalin, and the indeclinable ka "which." 
In addition to this, they have a more definite interrogative kavd, 

Digitized by VjOOQ IC 


which in sound and meaning is ahnost identical with Gt. %^. 
It is singular that they haye also the unique (so far as the 
Indian languages go) G. ^, Gipsy so, from which comes an 
ablative 8ostar, which, when pronounced with the accent on the 
first syllable sdstar, means "why?" but with the accent on the 
last syllable sostdr, " because,*' literally "from what." From 
this again comes a further interrogative savo "qualis?" parallel to 
kavd. The Kashmiri, with which the language of the Gipsies 
is most generally in accord, here diverges from it, and the 
connexion is closer with Gujarati than with any other of the 
languages. Kashmiri has ^g^ and ^^ for "who?" or "what?" 
oblique has as in Gipsy, but here, as far as the meagre informa- 
tion accessible enlightens us, the resemblance stops, and Kashmiri 
runs into its usual similarity with Panjabi. 

The reflexive pronoun "self" is represented by forms derived 
from the stem "^n^j ^s masc. po, fem. pi, obi. pe. An older form, 
which Paspati (p. 71) says is rare in the present day, is pinro 
m., pinri f ., pinre obi. Here we see ^^ITtRir with the loss of its 
initial d and the old genitive affix ro, ri, re, shortened from koro. 
With this agrees the possessive of the two first personal pro- 
nouns minro, -ri, -re, "mine ; " tinro-, ri-, re, "thine ;" and plurals 
strikingly Indian amaro, -ri, -re, "our;" tumaro, -ri, -re, "your." 
But the language still possesses the simple genitives mo, to, as 
in H. B. 0. ^, ^, though it differs from H. in treating them as 
possessives, and consequently as adjectives, and inflects them for 
gender and case mo m., mi f., me obi. ; so also to, ti, te. In addition 
to the Te&exive po, pi, pe, it has also|?^«='^rnTO "each other." 

Demonstratives are aka and amka "this," oka, 09oka, "that," 
akavka and okovka "this here," "that there," which do not ally 
themselves very closely to the corresponding words in any of 
the Indian languages. Peculiar also are kadava m., kadayd f. ; 
kadald nom. pi. m., kadale f., "this," and odova "this." Perhaps 
we are here to suppose the operation of some influence other 
than Indian. 

Digitized by LjOOQ IC 



The demonstratives aha and oka form sets of parallel pro- 
nominal adverbs^ just as do the demonstrative types of the 
Indian group. Thus we have akati okoU "here and there/* 
akatar, okotar, "hence," "thence." The latter of these corre- 
sponds in the manner of its formation to the Panjabi, which 
expresses "hence" and "thence" by adding the ablative of the 
old synthetic system imder the form ^ to the words for "here" 
and "there" respectively, as ^^ "here," ^[c^ "from here," 
"hence," ^5^ "there," ^^t "thence." In like manner, tar is 
the sign of the Gipsy ablative. Thus too M. has l[ii% "hither," 
and the ablatival form T^ITl^ "hence," so also fd^^ and fTl^^if 
"thither" and "thence;" and the same method runs through 
all the languages, for which reason the words expressive of 
"motion from" have not been given in the lists above exhibited, 
as they are merely the ablatives of the words expressive of 
"rest at." 

The only two authorities for the Gipsy, namely Paspati and 
Miklosich, which are procurable here, do not give a full series 
of pronouns and pronominal adverbs ; and it may be conjectured 
that, in the rude speech of this people, they are not all to be 
foimd. Those noted, however, are in striking conformity with 
our Indian group. 

§ 79. In concluding, amidst constant interruptions, this volume 
on the Noun and Pronoun, I have a few further remarks to make 
of a general character. Attention has already been directed in 
several parts of this volume to the varying nature of the seven 
languages in point of simplicity ; but the geographical aspect of 
this question remains to be noticed. The most complicated of 
the seven languages are Marathi and Sindhi, and, as far as we 
know anything of it, Kashmiri. If Gujarati and Panjabi were 
as complicated as these three, then we might establish a regular 
gradation from east to west; for the Oriya and Bengali, the 
most eastern members of the group, are distinguished by 

Digitized by VjOOQ IC 


extreme simplicity; while the Hindi, the central language in 
position, is central also in this respect, being less simple than B. 
and 0., less complicated than M. and S, It will be remembered 
that in the First Volume Hindi was noted as central in respect 
to its phonesis, leaning neither to the peculiarities of the 
extreme eastern dialects, such as a fondness for long vowels and 
the w-soijid, nor to those of the western dialects, a predilecfcion 
for short Towels and the t-sound. As regards the structure 
of the noun, the homogeneity of the western group is dis- 
turbed by P. and G. The former, with its structure only 
slightly less simple than Hindi, lies between the intricate 
Sindhi and the no less intricate Kashmiri; while the latter 
also, only a little more full than Hindi, intrudes between Sindhi 
and Marathi. 

The comparatively simple structure of Panjabi, as compared 
with its neighbours to the north and south, is probably to be 
ascribed to the fact that the Panjab has been for numerous 
centuries the battle-ground of India, over whose plains have 
passed and fought Greek, Persian, Arab, Turkish, and Indian 
armies ; from aU, except perhaps the first, of whom the language 
has imbibed words and ideas, and has been led to reject much of 
its former complicated structure. Since the establishment of the 
Musulman on the throne of Delhi, it has been constantly ruled 
from the Hindi area, and so strong has been, and still is, the 
influence of Hindi, that the wonder is that anything should be 
left of a distinctly dialectic character. As it is, in the towns 
and more civiKzed portion of the country, Panjabi is fast dis- 
appearing, and will in all probability disappear entirely at no 
distant date. 

The simplicity of Gujarati has been frequently explained in 
these pages by a reference to the fact that it is a dialect of 
Hindi, separated at an early, but not precisely assignable date, 
from the parent language, and thus retaining a certain amount 
of archaism. 

Digitized by LjOOQ IC 


§ 80. With regard to the two eastern languages, much 
remains for fiiture research to make plain. In Orissa the 
traces of a large non- Aryan element in the population are 
patent to the most superficial observer^ and the same may be 
said of Bengal; but unfortunately neither of these proyinces 
has as yet been examined, either as to their language or 
ethnology, as thoroughly as could be desired. The Europeans 
resident in India are all busy men ; no one of us comes here 
except to work, and as a rule to work very hard. The machinery 
of Gk>yemment becomes more complex, and makes greater 
demands upon the time of members of the administration every 
year ; so that those whose previous training would best qualify 
them for the task of investigation have as a rule the least 
leisure for it. It is not possible for scholars in Europe to 
conduct minute inquiries, because the materials they possess 
refer rather to ancient than modem India ; and it is necessary 
that one should live in the country itself, and in daily com- 
mimication with its people, in order fully to breathe the spirit 
of their customs and institutions, and so to get an insight into 
the nature of them, such as may indicate the most promising 
line of inquiry to follow. There are more materials for Orissa 
than for Bengal, because the former province was more isolated, 
more homogeneous, more bound up in itself, more a nation in 
short than Bengal for many centuries ; and the records of the 
great idol-temple at Puri, together with others which are 
probably still to be found in various holes and comers of this 
most conservative of Indian provinces, will, when they can be 
got at and examined, probably yield a rich harvest of facts to 
the labourer in the departments of history and philology. 




I. ■ t.1- 

Digitized by VjOOQ IC 

Digitized by LjOOQ IC 

Digitized by LjOOQ IC 

Digitized by LjOOQ IC 

Digitized by VjOOQ IC