A Comparative Grammar of the Modern Aryan Languages of India v1.c7: Chapter IV

A Comparative Grammar of the Modern Aryan Languages of India v1.c7

Language: English

v1.c7.3 — 5 parts

Chapter IV

v1.c7.1

CONTENTS.—§ 72. ANCIENT AND MODERN ARTICULATION.—§ 73. THE NEXUS—ITS THREE CLASSES.—§ 74. THE STRONG NEXUS.—§ 75. "SAPTAN" AND ITS DERIVATIVES.—§ 76. "UD" COMPOSITION.—§ 77. THE MIXED NEXUS.—§ 78. NASALS IN A MIXED NEXUS.—§ 79. TREATMENT OF क्.—§ 80. THE SIBILANTS IN A MIXED NEXUS.—§ 81. TREATMENT OF त्.—§ 82. SIBILANTS IN COMBINATION WITH DENTAL AND CEREBRAL STRONG LETTERS.—§ 83. SEMI-VOWELS IN A MIXED NEXUS; र AND ल.—§ 84. THE SAME; य AND व.—§ 85. य, र, AND व IN COMBINATION WITH DENTALS.—§ 86. THE WEAK NEXUS.—§ 87. NASAL WITH NASAL.—§ 88. NASAL WITH SEMIVOWEL.—§ 89. NASAL WITH SIBILANT.—§ 90. SEMIVOWEL WITH SEMIVOWEL.—§ 91. SEMIVOWEL WITH SIBILANT.—§ 92. ह IN COMBINATION WITH OTHER WEAK LETTERS.—§ 93. RELATIVE STRENGTH OF THE WEAK LETTERS.

v1.c7.2

§ 72. THE treatment of compound consonants involves some curious considerations on the subject of the power of articulation possessed by the human organs of speech in ancient and modern times. Sanskrit in this respect resembles all the early languages of its family, inasmuch as it is rich in combinations of consonants in groups of two, three, and even occasionally four, without any intervening vowel. We find the same state of things in the Semitic family also; the Aramaic group, which is now generally admitted to represent the oldest type, clusters its consonants together in a way which at times almost defies pronunciation; the Hebrew, which stands second in point of antiquity, is less addicted to this practice, having several

v1.c7.3.1

contrivances, such as *sh'va*, for avoiding too great an accumulation of consonants, while in the Arabic, a group of a much later type, to use Renan's expression, "the consonants swim in a sea of vowels"; comparing Aramaic forms, such as *ktalt, ktaln, mektlan, khnun*, with the Hebrew *kĕtaltâ, kâtalnû, kâtêlenû, anakhnu*, and Arabic *katalatâ, katalnâ, inkiâlanâ, nakhnu*, the gradual increase of vowels is clearly seen. This instance is adduced merely as an illustration of what is meant by the power of articulation. To utter a series of consonants, each of which requires a separate contact, without allowing any vowel to intervene, is an effort of the organs of speech which can only be accomplished when those organs are vigorous and strong. Thus we find the Dravidians of the south of India unable to pronounce Sanskrit words containing a compound consonant without the insertion of one or more vowels. For instance, in Tamil, *istiri* = Skr. *strí*, *tânam* = Skr. *sthânam*, *kiruṭṭinan* = *kṛishṇa*, *appirayośana* = *aprayojana*, *tatuva* = *tatwa*. "The languages of the Indo-European class are fond of combining clashing consonants and welding them into one syllable by sheer force of enunciation; and it is certain that strength and directness of character and scorn of difficulties are characteristics, not only of the Indo-European languages, but of the races by which those languages are spoken.

v1.c7.3.2

On the other hand, the Dravidian family of languages prefers softening away difficulties to grappling with them; it aims at ease and softness of enunciation rather than impressiveness: multiplying vowels, separating consonants, assimilating differences of sound, and lengthening out its words by successive agglutinations; it illustrates the characteristics of the races by which it is spoken by the soft, sweet, garrulous effeminacy of its utterances."¹ Within the domain of the Indo-European family, however, there are many degrees of power in this respect. In the highest grade are to be placed the old Aryan, Latin, and old German. In the next Zend, which is a degree weaker than Sanskrit; Greek, which can bear less consonant-clashing than Latin; and Sclavonic, which interpolates vowels with greater frequency than its Teutonic neighbour. Weakest of all are the modern Romance languages, and the Indian Prakrits. And, considering the subject chronologically, each stage of every language is, as a rule, less capable of elaborate articulation than that which preceded it. And this brings us back to our immediate subject: just as the Prakrits effeminately soften or omit single consonants from Sanskrit words, so, in the case of compound consonants, they have recourse to several devices for avoiding the effort of uttering them.

v1.c7.3.3

Here again, however, the modern Indian languages are stronger than the Prakrits, in many cases exhibiting a greater power of consonantal articulation, though always weaker than the original Sanskrit. In modern Tadbhavas indeed, in which no trace of Prakrit influence is to be expected, the amount of softening or weakening is not very great, not more so than in modern English as compared with the English of a thousand years ago. Take for instance a passage from King Alfred's translation of Gregory's Pastoral:¹ "Ne sceal he noht unalyfedes dón, ac thæt thætte othre menn unalyfedes doth he sceal wepan swæ swæ his agne scylde, hiora untrymnesse he sceal throwian on his heortan, and thæs godes his nihstena he sceal fagenian swæ swæ his agnes." This would run in modern English thus: "Nor shall he nought unallowed (unlawful) do, and that that (that which) other men unallowed do, he shall weep so so (as) his own guilt (Germ. schuld, Dan. skyld, our guilt for sguilt), their untrimness (infirmity) he shall dree in his heart (sympathize with), and of the good of his next (neighbour) he shall fain (rejoice) so so (as) of his own." The words in parenthesis are the modern way of expressing the word which they follow, which is the etymological equivalent of the old English.

v1.c7.3.4

Now the modern Tadbhavas in the Indian languages differ very little more from Sanskrit, and sometimes not so much as ‘unallowed’ from ‘unalyfedes,’ ‘dree’ from ‘throwian,’ ‘next’ from ‘nihsta,’ or ‘fain’ from ‘fagenian.’ It must be remembered that the earliest of modern Tadbhavas, dating from the wane of Buddhism, are about contemporaneous with Alfred, while early Tadbhavas are of course far more ancient. It is necessary to bear this in mind, because some writers have asserted that the modern Indian languages have deviated far more widely from their primitive type than the European languages, and have argued from thence that the former are now more hopelessly corrupted and confused than the former. It may be replied that the rate at which development has proceeded seems to be the same in both cases, and that climatic influences, while they determine the direction of that development, do not accelerate it. When we find an Indian word much more extensively changed than we can produce a parallel for in European languages, the reason is generally to be found in the fact that the process has been going on for a much longer time in India than in Europe. Certain it is that some of the Prakrits exhibit at times extreme forms of corruption, but reasons have been given for doubting whether these Prakrits were ever spoken;

v1.c7.3.5

and from the example of the Bhâgavatî, we may conjecture that when more texts come to light, they will confirm what is now only a suspicion, that in the two leading points of hiatus and assimilation the real spoken languages of those times did not go so far as the literature would make us believe. The hiatus caused by the elision of single consonants is an attempt to reduce the labour of speaking, and quite parallel to it is the process of assimilation, which will now be explained, both processes being exemplifications of the fact that as a rule modern articulation is simpler and less harsh than ancient, and that especially in the case of consonants, whether compound or single, there is a tendency to reduce their number and simplify their enunciation.

v1.c7.4

¹ Caldwell's *Dravidian Grammar*, p. 139.

v1.c7.5

¹ P. 50, ed. Sweet, Early English Text Society, 1871.

v1.c7.6

running through all the languages of the Indo-Germanic family.

v1.c7.7

§ 73. For the purpose of expressing the combination of two consonants without the intervention of a vowel, as in *kt*, *pl*, *sn*, the term "nexus" is employed. Three kinds of nexus may conveniently be distinguished:—

v1.c7.8

(1.) The strong nexus, composed of strong letters only, as क्त, प्र, ञ्ज, न्र, त्य, ग्ध.

v1.c7.9

(2.) The mixed nexus, composed of a strong and a weak letter, as च्र, न्द, घ्न, झ्र.

v1.c7.10

(3.) The weak nexus, composed of weak letters only, as म्र, य्र, ह्र, र्म, ष्ण, ल्व.

v1.c7.11

By strong letters are meant the first four letters of the five vargas:

v1.c7.12

क, ख, ग, घ. च, छ, च, झ. ट, ठ, ड, ढ. त, थ, द, ध. प, फ, ब, भ.

v1.c7.13

By weak letters are meant the nasals ङ, ञ, ण, न, म; sibilants श, ष, स; semivowels य, र, ल, व; and ह. The combinations च and झ, so often treated as single letters, I include under the mixed nexus.

v1.c7.14

The laws of these three kinds of nexus are briefly as follows:

v1.c7.15

(1.) In the strong nexus the Prakrits assimilate the first letter to the last, and the modern languages, except Panjabi and Sindhi, reject the first of the two letters and in general lengthen the preceding vowel.

v1.c7.16

(a.) Where the second member of the nexus is an aspirate, Prakrit changes the first member into the corresponding unaspirated letter.

v1.c7.17

(β) Where the nexus is followed by a long or heavy syllable, the modern languages do not always make the compensatory lengthening of the preceding vowel.

v1.c7.18

(2.) In the mixed nexus the weak letter, whether preceding or following the strong letter, is in the Prakrits assimilated to it; and in the moderns one of the two letters is rejected, with the usual compensatory lengthening of the vowel.

v1.c7.19

(a) When the weak letter is a nasal and precedes the strong letter, it is changed to anuswâra; in Prakrit without, and generally in the moderns with, compensatory lengthening of the vowel.

v1.c7.20

(3.) In the weak nexus three processes take place:

v1.c7.21

(a) When the two letters are of unequal strength, the weaker is assimilated, as in the mixed nexus.

v1.c7.22

(β) When they are of equal or nearly equal strength, they are divided from one another by the insertion of a vowel; or,

v1.c7.23

(γ) They are combined into a third and entirely different consonant.

v1.c7.24

A table of the relative strength of the weak letters will be found in § 92.

v1.c7.25

Certain and accurate laws for all these three processes can hardly be expected at the present initiatory stage of the inquiry, but the examples to be hereafter given will fairly substantiate the tentative rules now laid down.

v1.c7.26

The whole of this part of the development of the modern languages seems purely Aryan, and may be paralleled by corresponding changes in the other languages of the family, as, for instance, the modern Romance languages of Southern Europe. As none of the non-Aryan families in India are prone to the use of compound consonants, their influence can hardly be looked for in this branch of the inquiry. It is, however, held by some writers, who are never easy unless they can drag in some hint of non-Aryan influence to account for changes which require no such explanation, that the weakness of Dravidian enunciation, which forbids the use of any complex accumulation of consonants, is parallel to the weakness which led the Prakrits to assimilate *kt* into *tt*. As, however, the Italians do precisely the same, it is not evident why non-Aryan intervention should be suggested. There is a process in Prakrit, carried on into the moderns, which certainly does resemble Dravidian customs, namely, that of splitting up a nexus by the insertion of a vowel; when the custom is discussed it will be seen how far this supposition is true; at any rate it has become of late years quite a nuisance, this perpetual suggestion of non-Aryans here, there, and everywhere; one will soon have to believe that the Aryans did not know how to speak at all till the Dravidians taught them the use of their tongues, and that the Vedas are a mangled copy of some ancient Tamil liturgy!

v1.c7.27

§ 74. THE STRONG NEXUS.—This form of nexus is not so common as the others. The number of groups of two letters each, which can be formed by combinations of twenty letters, is in itself small, and is further in practice limited by the euphonic laws, which forbid the union of a media with a tenuis, and the placing of an aspirate as the first member of the nexus. Some organs, also, are incapable of combination with others; and of other combinations which would be euphonically permissible, the Sanskrit language affords no examples. The actual instances, therefore, of the strong nexus are comparatively few.¹

v1.c7.28

¹ Portions of this section on the strong nexus appeared in the Journal of the Royal Asiatic Society, vol. v. p. 149. Having, however, been led by further research to modify some of the views which I expressed in that article (written in 1868), I have had to remodel a great part of it, so that the actual reprint includes very little more than the examples.

v1.c7.29

Vararuchi’s way of handling this process is rather extraordinary. He considers that of two consonants in a nexus the first is elided, and the second doubled. He also omits all mention of च, ज, ट, and ब. In spite, however, of his uncritical way of putting it, it is clear that he means to lay down the same law as that given above. ब he confuses with व, which being a weak letter would, in any case, whether preceding or following, be assimilated to the other member of the nexus. This is clear from his examples. Thus, Skr. लुब्धक lubdhaka = Pr. लोद्धओ loddhao, and Skr. पक्व pakwa = Pr. पिक्कं pikkam; Skr. लुभ्य lubdha = Pr. लुद्धो luddho.—Var. iii. 51.

v1.c7.30

As to ट, it so seldom occurs as the first member of a nexus (except in combinations where षष् “six,” constr. षट् forms the first element), that Vararuchi probably thought it useless to mention it. Similarly I am not aware of any cases where the palatals form the first member of a nexus. By Sandhi they would migrate into gutturals in composition; e.g. त्वच् twach + दोषिन् doshin = त्वग्दोषिन् twagdoshin, वाच् vâch + य ya = वाक्य vâkya. So that probably no instances exist of palatals as initials in a nexus. Vararuchi’s examples are: भक्त = भत्तं, सिक्यक = सित्यओ, मुग्ध = मुद्धो, स्निग्ध = सिण्हो, खड्ग = खग्गो, षड् = सञ्जो, उत्पल = उप्पलं, etc.

v1.c7.31

Lassen (p. 239) repeats Vararuchi for the most part, but shows some consciousness of the difference between a strong and a weak nexus. He gives (p. 235) five classes of nexus, among which is No. II., “Complexus consonantium quaternarum ordinatarum,” which corresponds to my “strong nexus.” His list of examples is different from Vararuchi’s, but does not add anything to the general rule already obtained from the Indian author.

v1.c7.32

Hoefer, de dialecto Prakritâ, p. 90, § 73b, gives a list of words in which the Sanskrit form of the nexus is retained in Prakrit, but this list has no value, as it rests upon incorrect readings. The principal instance is समिक्त्सु in Vikramorvaśi (110, 2, Calcutta edition), which Lassen concludes to have arisen “ob negligentiam editorum,” or, perhaps, rather from the proclivity of the scribe towards Sanskrit forms, a fertile source of bad readings in the plays. We should undoubtedly read सभिक्क°. Hoefer's work does not appear to be one of much value.

v1.c7.33

If now we turn to Pali and the Prakrit of the Jain writings, we find the same principle carried out. Instances from the Dhammapadam and Bhâgavatî are, पृथग्जने = पृथुज्जणे, ज्ञप्तं = णत्तं, पुत्रलं = पुग्गलं (or in Bhâg. पोग्गलं), सप्तन् = सत्त, सप्तष्टि = सत्तट्ठि, कुब्ज = खुय्य (कुब्ज is probably for कुभ्ज, hence the aspirate).

v1.c7.34

Pali seems to have the best claim to be considered as pure Magadhi, and in the Mahawanso, one of the chief works in that language, we find the same rule, e.g. वक्तव्यं = वत्तब्बं, प्राप्त = पत्त, युक्त = युत्त. Numerous instances may be collected from every page of the Mahawanso, or any other Pali work. The rule is apparently absolute and without exception, so much so, that any deviations from it may at once be set down as errors of copyists.

v1.c7.35

There seems to be no good ground for supposing that the last member of a strong nexus is ever assimilated to the first. The examples given by Hoefer rest chiefly on incorrect readings, as shown by Lassen, and no such proceeding can be discovered in any of those works of which original MSS. are available.

v1.c7.36

In the modern languages a good many of the examples given by Vararuchi occur:—

v1.c7.37

खड्ग *khaḍga*, "a sword," Pr. खग्गो, Var: iii. 1. H. 1st, खग्ग, Chand, Prith. 15, 24, लग्गयौ खग्ग उडि हत्थ तें "He attacked, raising sword in hand"; 2nd, खाग (but only used for the horn of the rhinoceros, for which also खड्ग is used); 3rd, खड़ग (*kharaḍg*); 4th, खांड (*khâṇḍ*). M. 1st, खड्ग (rare); 2nd, खंडा; 3rd, खांडा. P. *deest*, S. खन्नो, G. खाडुं, B. खड़ग, O. खंडा, deriv. खंडाइत "swordsman."¹

v1.c7.38

¹ The forms खांड and the like are probably connected with खण्ड. We cannot consider them as derived from खड्ग.

v1.c7.39

मुद्ग mudga, “a sort of pulse”¹ (Phaseolus mungo), Pr. मुग्गो, Var. ib., H. मूंग, M. मूग (also as a Tatsama मुद्ग, but rare), S. मुङ्गु (muṅgu), O. मूग (mūgo), G. deest, P. मुंग or मुग्ग, B. मुग (mugŏ).

v1.c7.40

¹ Erroneously in all the dictionaries as a kidney-bean, which it in no way resembles; it is a sort of millet.

v1.c7.41

दुग्ध dugdha, “milk,” Pr. दुद्ध, H. दूध, M. and G. दूध, O. दूध, P. दुद्ध, S. डोड्धि, B. दुध, दुद्ध, दुदु. M., G., and B. have also the Tatsama form in pretty common use.

v1.c7.42

मुद्गर mudgara, “hammer,” Pr. मुग्गरो. This word is used in two senses in the modern languages. (α) Late Tadbhava, = an athlete’s club for exercising the muscles. H. मुग्दर (मुग्दर), M. मुद्गल, S. deest, G. मगदल, O. and B. मुग्दर and मुद्गर, P. मुगदर. (β) Early Tadbhava, = a washerman’s mallet for beating clothes. H. मोगरा and मोगरा, M. मोगर ०रा, and ०री, S. मुङ्गिरो, G. मोगर and ०री. The latter form is clearly the older of the two.

v1.c7.43

शब्द çabda, “a word,” Pr. सह्दो, Old H. साद, H. सबद and शब्द, M., S., G., and B. शब्द (not very frequent, except in B.), P. deest, O. सबद (śŏbdŏ).

v1.c7.44

कुब्ज kubja, “crookbacked,” Pr. खुज्जो and खुय्यो. H. 1st, कुब्जा; 2nd, कुब्जा; 3rd, कुबड़ा (kubṛā, commonest form). M. कुबड़ा (in the Wāḍi dialect, खुबड), G. कुबडो (both in M. and G. खुबडा is “the snail,” i.e. humpback), S. कुबो, P. कुब्जा and कूवा, O. कूजा, B. कुज, कुंज, and कुब्ज.

v1.c7.45

मुग्ध mugdha, “foolish,” Pr. मुढो. मुग्ध occurs in M. and B., but in the others the form in use is मूढ़, from the Skr. मूढ, which is the other form of the p.part. of मुह.

v1.c7.46

भक्त bhaktam, “boiled rice,” Pr. भत्तो, H. भात, M. and G. id., S. भतु, B. and O. भात, P. deest.

v1.c7.47

सिक्थक “beeswax,” also generally any soft wet substance. Pr. सित्थओ, H. सीठ “indigo refuse,” but सीथ “boiled rice,” M. सीत “boiled rice,” P. सेट, सेट “expressed sugar-cane.”

v1.c7.48

The class of words containing the nexus क्त *kt* is rather numerous, but does not present many interesting features; a few examples follow:—

v1.c7.49

भक्त *bhakta*, “a devotee,” Pr. भत्तो, H., M., and B. भगत and भक्त, G. भगत, S. भगत्, P. भगत.

v1.c7.50

मुक्ता *muktá*, “a pearl,” Pr. मुत्ता, H. मोती, M. मोती, G. मोती, B. and O. मोति, B. also मुक्ता, मोत्तिका, and मति, S. मोती, P. *id.*

v1.c7.51

रक्त *rakta*, “red,” Pr. रत्तो, H. राता, M. *id.*, when it means “red,” but रक्त when meaning “blood,” S. रतो, G. रातु, P. *deest*, O. and B. रक्त in both senses.

v1.c7.52

In the above examples the duplicate consonant of Prakrit is retained in P. The lengthening of the vowel with rejection of one consonant occurs nearly always in the rest. S., though it writes only one consonant, pronounces two, as P. does; this accounts for the short vowel being retained.

v1.c7.53

The Sanskrit words शक्ति, व्यक्ति, पक्ति, भक्ति, मुक्ति, पङ्क्ति, occur as Tatsamas in nearly all the languages, but with slight modifications in some. They do not, however, offer much that is noteworthy.

v1.c7.54

§ 75. The most common and most varied exemplification of this kind of nexus is to be found in the word *saptan* “seven,” and its numerous derivatives. The whole of them are given here in full as a good example of the harmony of the whole seven languages, a harmony all the more striking on account of the slight local peculiarities, “not like in like, but like in difference.”

v1.c7.55

सप्तन् *saptan*, (oxytone),¹ “seven,” Pr. सत्तो, H. सात, M. and G. सात, B. and O. सात (pronounced *shātō*), S. सत, P. सत्त. B. and M. also Tatsama, but rarely used.

v1.c7.56

¹ But see Bopp, *Vgl. Acc. System*, p. 45, § 29; the modern languages appear to have treated it as paroxytone.

v1.c7.57

सप्तम *saptama* (oxytone), “seventh,” Pr. सत्तमो, H. 1st, सतं (Chand, Prith. 1, 5); 2nd, सातमौ (Braj); 3rd, सातवां *m.*, वीं *f.* (modern). M. same as H. 3rd, G. सातमो—मी—मुं, S. सतों—तीं, P. सत्तवां—वीं, B. and O. सप्तम.

v1.c7.58

सप्तदशन् *saptadaśan* (accent on first syll.), “seventeen,” Pr. सत्तरस. H. सत्रह (*satraha*), modern सतरह (*satarah*), M. सत्रा or सतरा, G. सतर, S. सवँहें, P. सतारાં, in respect of the long vowel it stands quite alone; B. सतेर, O. सतर.

v1.c7.59

सप्तदशः *saptadaśaḥ* (id.), “seventeenth,” H. सत्रहवां or सतरहवां—वीं, M. सत्रावां—वीं—वें, G. सतरमो—मी—मुं, S. सवँहें, P. सताह्रवां and ॰मां, B. and O. सप्तदश, or B. सतेर (same as the number for “seventeen”).

v1.c7.60

सप्तति *saptati* (oxytone), “seventy,” Pr. सत्तति, सत्तरि, हत्तरि, H. सत्तरि (Braj), सत्तर (mod.), M. सत्तर, G. सीतेर्, S. सतर, P. सत्तर्, B. सत्तर, O. सत्तोरि, in which it stands alone (M., B., and O. also indulge in their favourite pedantry by using सप्तति occasionally).

v1.c7.61

सप्ततितमः *saptatitamaḥ* (oxytone), “seventieth,” H. सत्तरवां—वीं, M. and P. *id.*, G. सीतेर्मो—मी—मुं, S. सतर्यो (*sataryo*). B. and O. the Skr. form.

v1.c7.62

It is next necessary to observe the curious modifications which the word for “seventy” undergoes when it occurs as the last member of a compound word.

v1.c7.63

“Seventy-one.” Skr. एकसप्ततिः *ekasaptatiḥ*. H. इकहत्तर *ik + hattar*. M. एकाहत्तर *ekā + hattar*. G. इकोतेर् *ik + oter*. S. एकहतर *eka + hatari*. P. इकहत्तर *ik + hattar*. B. एकात्तर *ek + úttar*. O. एकस्तोरि *eka-stóri*.

v1.c7.64

“Seventy-two.” बहत्तर *bahattar*, and so on. बाहत्तर *bāhattar*, etc. बोहोतेर् *bohoter*, etc. बोहतर *bohatari*, etc. बहत्तर *bahattar*, etc. बाहत्तर *bāhâttar*, etc. बास्तोरि *bástori*, etc.

v1.c7.65

Also the still greater transformations of "seven," when it is the first member of a compound, *e.g.*:

v1.c7.66

v1.c7.67

v1.c7.68

² Or सांयताब्त्रिंश.

v1.c7.69

| | "Sixty-seven." | "Seventy-seven." |
| Skr. | सप्तषष्टिः *saptashaṣṭiḥ*, | सप्तसप्ततिः *saptasaptatiḥ*. |
| H. | { सरसठ *sarsaṭh*. (सतसठ *satsaṭh*. | सतहत्तर *sathattar*. |
| M. | सतसष्ट *satsaṣṭ*. | सत्याहत्तर *satyāhattar*. |
| G. | सडसेठ *saḍseṭh*. | सीतोतेर *sītoter*. |

v1.c7.70

¹ Bengalis and Oriyas pronounce the स as *sh*, so that *shātḍiś* would be perhaps a better transliteration, and so throughout the series.

v1.c7.71

v1.c7.72

v1.c7.73

The series of "seven" and its derivatives belongs to the class of early Tadbhavas, in so far as it has everywhere lost its first member प्. This is only natural; a numeral being, next to a pronoun, the most commonly used description of word, and hence not to be changed by authoritative interference, such as I have assumed in the case of modern Tadbhavas.

v1.c7.74

The variations may thus be analyzed:—

v1.c7.75

सत्त° *satta-*, in the words "seventy," "seventieth."

v1.c7.76

सात° *sát-*, in the words "seven," "seventh."

v1.c7.77

सत° *satá-*, in the words "twenty-seven," "fifty-seven," "eighty-seven," "ninety-seven."

v1.c7.78

सैं° *sain-*, in the words "thirty-seven," "forty-seven."

v1.c7.79

सत्° *sat-*, in the words "seventeen," "sixty-seven," "seventy-seven."

v1.c7.80

And in Gujarati साड°, सूड°, सड°, सीत°. In M. सत्ते°, सत्य°; S. सते°.

v1.c7.81

Of these, सत्त° is simply the assimilated form of the Prakrit period, which in the compounds modulates into हत्तर, a form which is found even in Prakrit. This word is noticeable as the solitary instance in Hindi of a change of the initial स into ह, though in Panjabi the change is so common as to be regular (see Chapter III. § 66).

v1.c7.82

सात is the regular form of the early Tadbhava class, like भात and others. In सत्ता we have rather a curious phenomenon which requires fuller analysis; the words are सताईस, सतावन, सतासि, सतानवे, etc. In Prakrit we do not uniformly find the long â, e.g. sattavîsatimo, "twenty-seventh," but sattânave, "ninety-seven." But there is a remarkable uniformity in all the seven modern languages. We find the long â in all of them for "twenty-seven"; in all but P. and S. for "fifty-seven"; in all for "eighty-seven" and "ninety-seven."

v1.c7.83

Judging at first from सताईस only, and the Hindi chiefly, I thought this abnormal long â arose from the व of विंशति amalgamating with the inherent a of सप्त, first from av (sapta-vinsati) into au (sattauis), and thus, according to the usual Hindi analogy, into â. But this suggestion broke down in two ways: 1st, Because the Marathi has both the long â and the v in सत्तावीस; so also have the Sindhi and Gujarati: and 2nd, Because we have the long â in words where there existed originally no v to coalesce with it into au.

v1.c7.84

The opinion I am at present most inclined to is that the presence of the long â is due to the accent of the Sanskrit. Saptan is undoubtedly oxytone, and though I have not been able to find in the dictionaries or other published works any instance in which the accent of saptavinsati is given, yet I have no reason to doubt that, as in Greek, the accent remains in the compound word just as it stood on each member when separate. In Greek the rule which forbids us to place an accent further back than the ante-penultima obscures the analogy; but in Sanskrit we have no such restriction, and sáptavinsáti, with the minor accent on the penultima, and the major on the pre-antepenultima, is quite possible. Assuming this, and seeing that the two last syllables *ati* have disappeared, the word would fall into the form of a paroxytone, and the accent on the syllable *ta* being the only one left, would assume such prominence as to insure its remaining long throughout the ages.

v1.c7.85

We next come to the curious form *तैं*, only found in “thirty-seven” and “forty-seven.” In these two sets of numbers we find that all the odd numbers except the first take this form in *ain*, thus:—

v1.c7.86

| 33 | तैंतीस *taintís*. | 43 | तैंतालीस *taintális*. |
| 35 | पैंतीस *paintís*. | 45 | पैंतालीस *paintális*. |
| 37 | सैंतीस *saintís*. | 47 | सैंतालीस *saintális*. |

v1.c7.87

The other languages offer only faint indications of similarity. P. has *सैंची*, B. *सांईत्रीश*, O. *संइतिरिंश*, but none of the others. I think it has arisen from a sort of rhyming propensity sometimes observable in Hindi. From *ते* or *तै* the transition to *तैं* was easy, the insertion of the nasal in season and out of season being a weakness of that language. In the word for “thirty-five” they had at first *panchatriṅśat*, which soon got altered into *panchtis*; but here they had a strong nexus, *च + त*. The *च* therefore dropped out, and the resulting double *त* was reduced to a single, and the preceding vowel compensatorily lengthened, so that the form *पांतीस* arose; to make which rhyme with the word for “thirty-three” would be irresistible. So also the “thirty-seven,” which was probably *सात्तीस*, was dragged into the scheme, and became *सैंतीस*.

v1.c7.88

The form *सत* is found in “seventeen,” “sixty-seven,” “seventy-seven.” In “sixty-seven” the word was originally *saptásháshṭiḥ*, with two accents close to one another, the latter of which, so to speak, kills the former, so that when by the usual Prakrit process the first member is modified into *satta*, the second *a*, being atonic, falls out; and though the first *t* also falls, yet, as it is in the beginning of a word with a heavy termination, no compensatory lengthening can take place.

v1.c7.89

In "seventy-seven" we ought to have सतात्तर, and probably such a form did once exist, but was crushed into sat by the weight of the double tt of hattar so close after it. Only Marathi retains a form satyáhattar. Similarly the form for "seventeen" was originally sattárah, and is so still in Panjabi; perhaps some obscure idea of its similarity in sound to sattar, "seventy," may have operated in the minds of the people to shorten it.

v1.c7.90

The G. forms with ड are to be accounted for by the already described proclivity of dentals to turn into cerebrals; but why this tendency should exhibit itself in this particular instance it is difficult to say, the only thing at all cognate to it is the र in H. सरसठ, which would be more correctly written सडसठ; in this place it is easy to see that the cerebral arises from the influence of the ठ in सठ. G. has accordingly सडसठ. What makes the matter more obscure is that the cerebral letter occurs precisely in those two series where we should least expect it, namely, in those which begin with a dental, and where, accordingly, it would be natural that the त of सात should assimilate itself to the following homogeneous consonant. It must be remembered that Gujarati has emerged from the position of a Hindi dialect closely akin to Marwari, and that it still retains traces of its rustic origin in the sometimes inexplicable irregularity of its forms.

v1.c7.91

§ 76. Another rich class of examples of this nexus is formed by the words to which the preposition उद् or उत् is prefixed. This word having a consonantal termination coalesces into a nexus with the first consonant of the word following, giving rise to such forms as the following:

v1.c7.92

Skr. उद्गार "spitting," H. उगाल, P. उगाल्हणा, S. उमारण, G. उगाज and ओगाज, B. उगर, O. Tatsama.

v1.c7.93

Skr. उन्नमन “going up,” H. उगना (from उगौना), G. उगवुं, P. उग्गमणा and उग्गणा, S. उगणुं.

v1.c7.94

Skr. उज्ज्वलन “heating,” H., P. उबालना “to boil over,” G. उभारवुं, S. उबिरणु, उबारणु, M. उबाळा, B., O. उबलन.

v1.c7.95

Skr. उद्धारण “unloading,” H. उभारना, P. *id.* and उभराना, S. उभारणु, M. उभारणें, B. उभरण, O. *id.*

v1.c7.96

Skr. उत्थान “arising,” H. उठना, and so in all.

v1.c7.97

The noticeable point in this class of words is that, though they have universally rejected the double consonant which arose out of the assimilation of the first member of the nexus to the last, they have nevertheless not lengthened the preceding consonant. Thus, उद्धार would become in Prakrit उग्गार, and by rule should be in H. ऊगार. The reason of the absence of the usual compensatory lengthening is that in verbal nouns of the forms given above the accent rests on the syllable of the root; so in *bhāraṇam*, *jvālanam*, etc., it would be on the first syllable, and consequently, when preceded by a preposition, the vowel thereof must needs be short. Hindi is excessively sensitive on this subject: vowels are shortened when followed by long or accented syllables, or by such as contain a heavy mass of consonants, or they are lengthened when conditions the reverse of those occur. As a rule, Hindi, when rejecting one of two assimilated consonants, always, if it can, lengthens the preceding vowel. If it does not, we may be sure there is some cause for its not doing so; and if this cause cannot be found in a succeeding long vowel or a heavy termination, it will probably be discovered in the accent.

v1.c7.98

The other languages follow to a certain extent the same course as Hindi, Bengali, and Oriya; especially the former leaning towards long vowels, and Panjabi and Sindhi markedly towards short ones.

v1.c7.99

§ 77. THE MIXED NEXUS.—This nexus differs from the strong

v1.c7.100

nexus in this important particular—that whereas in the latter the first consonant is always assimilated to the last, in the former the weaker consonant of the two, whatever its position in the nexus, is assimilated. It is singular that the letters which I have classed as weak, and which are really weak when they occur in a nexus, should, as shown in the last chapter, be, when standing alone, more tenacious of existence, and less liable to elision than the strong letters. The cause of this apparent anomaly has been suggested as lying in their comparative softness, which renders them less liable to be omitted from euphonic causes, as they do not stop the liquid flow of Prakrit words by any harsh or decided check, like the strong letters; and this very softness renders them more liable to be passed over or treated as merely strengthening the strong letter with which they are combined, just as has been pointed out in the case of modern Bengali pronunciation in Chap. I., § 26.

v1.c7.101

§ 78. NASALS IN A MIXED NEXUS.—(a) When the nasal is the first member of a nexus it is generally elided, but the second

v1.c7.102

This nexus must be treated in detachments, because each of the subdivisions which it contains, though following in its principal features the general law laid down above, yet exhibits certain special peculiarities of its own, arising in the main from organic properties, such as the aspiration which lies hid in the sibilants, and the cerebral nature of the semivowel र. Such also are the volatile nature of the dentals and the extreme tenacity of palatals and nasals.

v1.c7.103

It is interesting also to notice that all these letters when in a nexus exhibit the same organic tendencies as when alone; the peculiarities which I have mentioned above will strike the reader as being the same as those which he has just noticed in Chapter III. Thus, again, we get harmony out of apparent confusion, and grasp laws out of the midst of a seeming chaos.

v1.c7.104

letter is not doubled, as in a strong nexus. Instead, anuswâra is inserted in the place of the lost nasal. It might at first appear that this substitution of anuswâra for the organic nasal was a mere compendium scripturæ—a simpler form of writing introduced to avoid the necessity of using the clumsy and intricate Sanskrit letters, such as ङ्, ञ्, ण्, न्, etc., and it might further be urged in support of this view that as several of the languages have abandoned the characters ङ्, ञ, ण्, the only course open to them was to use the nasal dot. But this supposition is rendered untenable by several considerations. In the first place, as will be seen from the examples, the substitution of anuswâra for the nasals of the several organs is almost always accompanied by lengthening of the preceding vowel, which is indicative of compensation for some loss. In B. and O., where the Sanskrit method of writing with the compound consonant is still retained, there are often found two forms side by side,—the learned or Tatsama form with the compound character, preceded by a short vowel, and the vulgar or Tadbhava form with the single strong consonant, preceded by anuswâra and a long vowel. Secondly, this anuswâra is in most languages pronounced as a distinct anunâsika or nose-intonation, as in the French *bon*, *son*; this is peculiarly the case in B. and O., where both forms are in existence side by side. And, thirdly, in many cases the insertion of the anuswâra is purely optional, which would not be the case were it another way of writing the nasal; nay, in some instances it is entirely omitted.

v1.c7.105

ङ्.

v1.c7.106

Skr. कङ्कण “bracelet,” Pr. कंकणं, H. कंकन, P., G., M. कंगण, S. केंगणु, B. कांकन, O. *id.*

v1.c7.107

Skr. जङ्घा “leg,” Pr. *id.*, H. जांघ, P., S. जंघ, G., M. जंघा, B. जांगी, O. Tatsama.

v1.c7.108

ञ. Skr. गञ्जाविका “hemp-plant,” H. गांजा, and so in all. Skr. चञ्चु “beak,” H. चोंच, G. चांच.

v1.c7.109

ण. Skr. कण्टक “thorn,” Pr. कंटओ, H. कांटा, M., G., *id.*, P. कंडा, S. कंडि, O. कांटा, B. *id.* Skr. शुण्ठि “ginger,” H. सोंठ, P. सुंठ, S. सुंढि, B., O. शुंठ, G. सुंठ, M. शुंठि.

v1.c7.110

न. Skr. चन्द्र “moon,” Pr. चंद, H. चांद, and so in all, but P. चंद, S. चंडु. Skr. स्कन्ध “shoulder,” Pr. खंधो, H. कांधा, P. कंधा, O., B. कांध.

v1.c7.111

म. Skr. कम्पन “trembling,” Pr. कंपनं, H. कांपना, P. कंबणा, S. कंबणु, G. कांपवुं, M. कांपणें, B., O. कांपन. Skr. जम्बु “rose-apple,” Pr. जंबू, H. जामुन, P. *id.*, G. जांबु, M. जांब, S. जमु, B. जाम.

v1.c7.112

In Marathi may be found very many exemplifications of the same process; thus, in words compounded with सम्: सांकड, सांगड, सांच, सांड, सांध, सांपडणें, सांप्रदाय, सांबर, for Skr. सङ्कट, सङ्घट्ट, सञ्चय, षण्ड, सन्धि, सम्पादन, सम्प्रदाय, and शम्बर; in all of which the change from the organic nasal to the anuswâra is accompanied by a compensatory long vowel. It cannot be said that we have here an instance of the Prakrit tendency to preserve the original quantity of the Sanskrit syllable. It is true that in Prakrit the syllable might have its vowel lengthened, and probably would have, if to the ears of those who spoke Prakrit it should appear that the removal of one of two consonants following a vowel was likely to render the syllable a short one, seeing that having a short vowel it had previously been long only by position. But this is not the case here, as is evident from the fact that Prakrit does not lengthen the vowel, but keeps it short, as in चंद्र, कंपण; and, moreover, the modern languages are not sensitive about keeping the exact quantity of Sanskrit words, as has been shown in Chap. II., § 40.

v1.c7.113

It is not easy to trace the history of this process; the early Hindi writers belonged to the western portion of the Hindi area, and Chand especially, being a native of Lahore, is peculiarly western in his pronunciation. In these old writers, therefore, there is a strong tendency to approximate to the Panjabi habit of retaining the short vowel. As there are no extant specimens of the speech of the eastern portion of Hindustan, or of Bengal, at that period, we can only guess from the general analogy of the languages that they, at least, used the long vowel.

v1.c7.114

In the case of जम्बु it is not, as at first sight might appear, the ब which has dropped out in जामुन, जाम, and जमु, but the ब, after absorbing the म, has itself been softened into म; this is shown by the transitional forms जांबु and जांब. This is just the reverse of the process explained in Chap. III., § 65, where म is resolved into व; here ब, preceded by anuswâra, has passed into म. It is only one more instance of how cognate processes work backwards and forwards in these languages.

v1.c7.115

The word कुम्भकार “a potter,” is peculiar. It is in Pr. कुंभआरौ, H. कुम्हार, P. id., O. कुम्हार and कुम्हार, B. कुमार, and in vulgar Bengali they say कुमोर. Prakrit follows the general rule, and in this it is followed by the western languages M., G., and S., which all have कुंभार. H. goes a step further, and rejecting the organic portion of the भ, according to rule (§ 68), retains only ह; the anuswâra is then written as म. This is a solitary instance, and may be explained by the existence of the Tatsama कुम्भ “a pot”; there would naturally be a desire to assimilate the two words, which would lead to the retention of the म, because if written as anuswâra it would sound kunghâr, and not kumhâr, by which all resemblance would be lost.

v1.c7.116

In कहार, the name of a caste in Bihar whose profession is to carry palanqueens, and whose name is derived from कंधधार, the double ध, together with the anuswâra, has disappeared, leaving only the ह, without any nasal at all. It is natural that such words as these, being in daily and hourly use in the mouths of the people of the lowest classes, should undergo abrasions and contractions of the extremest and most abnormal character.

v1.c7.117

A striking exception to the general rule is afforded by a small class of words in Sindhi derived from Sanskrit words having ण्ड :

v1.c7.118

Skr. रण्डा “widow,” S. रन.

v1.c7.119

Skr. लण्ड “penis,” S. लनु, P. लन.

v1.c7.120

Skr. षण्ड “bull,” S. सानु.

v1.c7.121

Skr. खण्ड “piece,” S. खनु, Old H. खण (written घण), B. खानि.

v1.c7.122

Skr. पिण्ड “cake,” S. पिनु.

v1.c7.123

On the analogy of these words, S. खनो “a sword,” would be derivable from Skr. खण्ड, and not from खड्ग (see § 74). In all these words the cerebral letter has disappeared, and the nasal has been dentalized. There are, as noted, a few parallel instances in other languages, though these words follow in the main the general rule, e.g.—

v1.c7.124

Skr. रण्डा, H. रांड, pronounced Rāṇṛ, vulgarly spelt Raur, and by our countrymen sounded sometimes so as to rhyme with “roar,” “soar”; at others so as to rhyme with “hour,” “sour” (!).

v1.c7.125

Skr. षण्ड, H. सांड (Sāṇṛ).

v1.c7.126

Panjabi has a little group of words somewhat analogous to the Sindhi list, in which it treats न्ध by dropping the organic portion of the ध and doubling the न; thus, from

v1.c7.127

Skr. अन्धकार, H. अंधेरा and अंधियारा, P. has अन्हेरा annherā.

v1.c7.128

Skr. संधि “a hole,” H. सेंध, P. सन्ह.

v1.c7.129

Skr. स्कंध “shoulder,” H. कांधा, P. कन्‍ह. Skr. बन्धन “binding,” H. बांधना, P. बन्‍हणा.

v1.c7.130

Putting together this list, the Sindhi list in ङ, and such Hindi words as कुम्हार, जामुन, etc., there results a tolerably strong body of words which do not follow the rule which has been taken as the general one. These words are too numerous to be traced as mere exceptions, and their existence must be accounted for by the peculiar tenacity of the nasal, which, as we have frequently seen before, is a great favourite with the people, and is more often inserted where it has no business than elided where it ought to be so according to rule.

v1.c7.131

(β) When the nasal is the last member of a nexus, it is regularly assimilated to the preceding letter, and in the moderns (except P. and S.) one of the two letters is elided with or without, but more often with, the compensatory vowel lengthening. Examples: Skr. अग्नि “fire,” Pr. अग्गि, H. आग, M., G. *id.*, P. अग्ग, S. अगि.—B. आगुन, and O. anomalously गिअ. Skr. लग्न “attached,” Pr. लग्गो, H. लगा, लाग, and लग, P. लग, M. लाग, and so in the rest. Skr. नग्न “naked,” Pr. णग्गो, H. नंगा, P. *id.*, S. नंगो, G. नागुं, M. नंगा, नग्गा, B. नेंटा (nengṭā), O. नंगला (diminutive). Skr. तीक्ष्ण “pungent,”¹ Pr. तिक्खहं, H. तीखा, and so in all, but P., S., and G. have the short vowel.

v1.c7.132

¹ For our present purpose क्ष् must be regarded as equivalent to ख.

v1.c7.133

It is not correct to suppose in the case of नंगा, as Trumpp does, that there has been an inversion of the nexus from *gn* to *ng*. The *gn* first becomes *gg*, and remains so still in one Marathi form, and in G., and the anuswâra is subsequently inserted for the sake of euphony more than anything else. As there existed a long period in which the word was regularly spoken naggo, it is impossible that there should have remained in the public mind any consciousness of the lost nasal, and consequently, not knowing of its existence, they could not insert it.

v1.c7.134

Reference may here be made to the remarks in Chap. I. § 30, with regard to the immense interval of time which separates Sanskrit as a spoken language from the modern vernaculars, and it should be remembered that the lower classes of India are not philologists; they are not constantly haunted by the presence of Sanskrit; the great panorama of change and development is not spread out before their eyes as it is before ours, and they are in consequence not led by the clearness of the connexion between the various links of the great chain into forgetting the lapse of ages through which it rolls out its immense length. The word नग्न, when resuscitated as a Tatsama in modern times, has not changed at all, and stands side by side with the ancient नंगा, नग्गा, and नागुं, without interfering with them. Bengali in this word exhibits an archaic diminutive, which I have written in Devanagari characters नेंटा, but this does not give the full force of the B. character ঙ, which is a very strong nasal ng. In its own letters the word is written নেংটা, and perhaps a more accurate transliteration into Nagari would be नेंगटा, although this even is hardly quite satisfactory.

v1.c7.135

The compensatory lengthening of the vowel has not been made in H., because Skr. nagná is oxytone, and to preserve the accent the last syllable is lengthened, giving nangâ; where the long vowel following, according to regular practice, forbids the lengthening of the preceding syllable. In लग, which is used as a postposition in the sense of “near,” “up to,” the excessive frequency of its use has probably resulted in its assuming a shortened form, though here also lagná is oxytone, so that the first vowel should be short. P. has the verb लंगणा “to approach,” “to pass by,” but no argument can be founded on this, as P. seldom lengthens a vowel as compensation.

v1.c7.136

§ 79. The compound letter ज्ञ = ज + ञ demands separate notice. It most frequently occurs in the verb ज्ञा "to know," and its numerous derivatives and compounds. Vararuchi has two rules on this subject. In iii. 5, he prescribes the regular course of assimilating the ञ, as in savvajjo, ingiajjo, for sarvajna, ingitajna.¹ The commentator adds that this process is followed in all compounds of jna, "to know"; but in this he is certainly wrong, as the next illustration shows. In iii. 44, the elision of the j is enjoined in all words having jn, and one at least of the instances is taken from the root jnâ,—they are जणो, विआण for yajna, vijnâna. Further on (iii. 55) he adduces सण्णा for sanjnâ, and directs that the n shall not be doubled when preceded by the preposition â, as in आणा, आणात्ती for âjnâ, âjnapti. From his xii. 7, it may be inferred that in the Sauraseni Prakrit, jn was inverted to ञ nj, and examples are given विञ्जो, जञ्जो for विज्ञ, यज्ञ, while in the next sûtra (xii. 8) the above-quoted words sarvajna, ingitajna, preserve the n and reject the j, becoming सव्वण्णो, इंगिअण्णो, instead of, as in Maharashtri, savvajjo, etc. Lastly, in viii. 23, we are told that this very root ज्ञा becomes जाण, as in जाणइ for जानाति "he knows"; this is nothing more than the usual substitution of ण for न, as in the special tenses in Sanskrit the form ज्ञा is substituted for ज्ञा, and the न is merely the conjugational sign of verbs of the krî class.

v1.c7.137

¹ I deem it unnecessary to put dots or dashes about this n, as it must be evident that the palatal n is meant; and dots or dashes are a nuisance alike to printer and reader.

v1.c7.138

In the Saptaśataka the root ज्ञा is regularly जाण, and perhaps in one or two instances आणु, though this is doubtful. ज्ञातं becomes खाअं, and विज्ञात = विआओ. आज्ञा = आणा. यज्ञ = जण्णो. The same treatment is observable in the Bhagavati.

v1.c7.139

There are thus four separate processes in the Prakrits:—

v1.c7.140

(1.) The regular and common one of rejecting the ञ and doubling the ज.

v1.c7.141

(2.) Rejection of ज, change of ञ to ण, and doubling it.

v1.c7.142

(3.) Single ज्ञ after आ. (4.) Inversion, peculiar to Sauraseni, into ज्ञ. The change in the special tenses of ज्ञ into जाणा comes under 1, as it is a rejection of ज्ञ, and the doubling of ज is precluded on account of its initial position.

v1.c7.143

In the modern languages considerable diversity of practice occurs. In all but G., S., and M., this letter is regularly pronounced *gy*, so that आज्ञा is sounded *âgyâ*, ज्ञान *gyân*. B. and O. retain the spelling ज्ञ, but H. and P. write as they pronounce. M. pronounces *dny*, G. as *jn* or *dn*.

v1.c7.144

Examples:—

v1.c7.145

Skr. आज्ञा “command,” H. आज्ञा, P. आगिआ, S. अग्ग्या.

v1.c7.146

Skr. यज्ञ “sacrifice,” H. जाग, Old H. जज्ञ, M., O., B. जाग, P. जगम, S. जगु.

v1.c7.147

Skr. यज्ञोपवीत “sacred thread,” H. जनेउ, P. *id.*, S. जज्ञो, G. जनोइ, M. जानवें. Not used in B. and O.

v1.c7.148

Skr. राज्ञी “queen,” Old H. रागिनी, H. रानी. In the rest राणी.

v1.c7.149

Skr. ज्ञान “knowing.” (α) As a verb, “to know,” H. जानना, P. जाणना, S., G., M., O. जाण°, B. जान°. (β) As a noun, “knowledge,” H. ग्यान, P. *id.*, B., O. *id.*, M., G. जाण, S. जाणु.

v1.c7.150

It would appear that the pronunciation *gy* is more frequent in Modern Tadbhavas, like *âgyâ*, and that in earlier times either the Maharashtri rule of rejecting the *n* was followed, as in Old H. जज्ञ, and in जाण° जान°, or the Sauraseni practice of rejecting the *j* as in जनेउ. So entirely, however, have the Prakrits abandoned their old geographical limits, if they ever had any, that we find Sauraseni characteristics in Marathi, and *vice versâ*. As to the time when, and the reason why, the pronunciation *gy* arose, there is nothing certain to be said. It would seem, however, from a passage in the Ain Akbari that the pronunciation *jn* had not quite died out in those days, for the writer, in mentioning the town of Jâjpur, in Orissa (यज्ञपुर

v1.c7.151

= “city of sacrifice”) writes it جشپور jashpur, which, he adds, “they also call جشنپور jashnapur.” This note seems to indicate that the town had begun to be called Jajpur (for I suppose he means to indicate the soft j, almost like a z, of Eastern India, by his ش), but that the older and more correct pronunciation, Jajnapur, was still retained by some. It is curious that the town should have retained the form jaj-, while throughout Orissa the letter ज्ञ is regularly and universally pronounced gy. The transition to this sound must have been in force much earlier in Western India, for Chand writes रगिनि for राज्ञी two centuries and a half before Akbar. The pronunciation probably arose, as Chand’s word would indicate, from splitting the nexus, from the fact of the nasal being the palatal ञ, which seems always to have had the sound ny or Spanish ñ; then followed the regular rejection of n and hardening of the remaining ज्य into ग्य. The whole process, however, is one entirely without a parallel in any of the languages.

v1.c7.152

§ 80. SIBILANTS IN A MIXED NEXUS.—It has been seen in Chap. III., § 66, that the sibilants when standing alone in a word are affected in two ways; on the one hand they are weakened into the aspirate, while on the other they are thickened into palatals. They exhibit precisely the same tendencies when forming part of a nexus; when they precede the other letter the transition to the aspirate is most frequently met with; when they follow it, that into the palatal.

First 156 of 1,261 passages. Read on

Cites (12)

Prākṛtaprakāśa — Vararuci, Sanskrit
Gāhāsattasaī — Hāla, Māhārāṣṭrī Prakrit
Dhammapada — Pāli
Pṛthvīrāj Rāso — Cand Bardāī, Hindi
Kṣetra — Sanskrit
Mahāvaṃsa — Pāli
Vikramorvaśīya — Kālidāsa, Sanskrit
Ṛgvedasaṃhitā — Sanskrit
Dṛṣṭi — Sanskrit
Viyāhapannatti — Sudharman, Ardhamāgadhī Prakrit
Lalitavistarā — Haribhadra Sūri, Sanskrit
Hastin

A Comparative Grammar of the Modern Aryan Languages of India v1.c7: Chapter IV

Contents

Chapter IV