Вы находитесь на странице: 1из 7

SOME S P O R A D I C C H A N G E S OF VOWELS I N MIDDLE INDO-ARYAN

by

L. A. SCHWARZSCHILD
Melbourne

Complaints have sometimes been made that the so-called "sporadic changes" which include assimilation, dissimilation, metathesis and epenthesis have not been studied as thoroughly as the regular sound-changes? This applies as much to Middle Indo-Aryan as to other languages. The Prakrit grammarians have simply listed such sporadic changes as exceptions to the general rules, and the modern grammars have taken over their lists with some additional examples. This is true for instance of the cases of sporadic change of the vowel u which are listed by Pischel. 2 This article represents an attempt at a further analysis of some of the problems involved. While Pischel simply lists the cases where the Sanskrit vowel u has been replaced by other vowels in Middle Indo-Aryan, some more recent grammars make confusing statements to the effect that u becomes a " b y dissimilation or assimilation". Some of this confusion comes from a failure to distinguish between two different processes, dissimilation and differentiation. Differentiation occurs when two similar sounds are in immediate contact and the speakers distinguish between the two sounds, thereby changing the phonetic characteristics of one of them. Meillet 3 and G r a m m o n t 4 consider differentiation as the reverse of assimilation: differentiation is a means by which assimilation is avoided and it helps to preserve the identity of the word which might be lost if assimilation took place. On the other hand dissimilation occurs when two identical sounds are close together, yet not in contact, and there is a tendency to avoid the repetition of the one sound. This might be considered as a more truly 1 For the most recent comments on this subject see R. R. Posner, Consonantal
Dissimilation in the Romance Languages (= Publications of the Philological Society,

XIX) (London, 1961), p. 2. R. Pischel, Grammatik der Prakrit-Sprachen (Strassburg, 1900), w 123-4 and w 126. 3 A. Meillet, "De la diff6renciation des phon6mes", MSL, XII, p. 14ft. 4 M. Grammont, Trait~ de Phondtique (3rd. ed. Paris, 1946), p. 229ff.

26

L. A. SCHWARZSCHILD

sporadic change than differentiation, it is often merely due to a slip of the tongue and is linked with the subconscious dislike of repetition: On the basis of this definition it is clear that cases of dissimilation of vowels are quite common in some languages as in French: Fr. selon < *soblongu < sublongum; secousse < *soccossa < succussa; devin < divinum. In these and many parallel cases the unaccented vowel has been altered before a similar accented vowel. In the history of Middle IndoAryan such developments are extremely rare. There seems to have been singularly little feeling against the frequent repetition of vowels: in fact in many of the onomatopoeic words where other languages sometimes have a variation of vowels (as in the famous examples of English tick-tock, French tic-tac) Middle Indo-Aryan generally repeats the same vowel. Words like misirniseman, e (Ndyddhammakahdo), and simisimiya (Pai~macariya 105.55) from Sanskrit simasimdya- 'crackling' and ki.dikid,idbh~te from k~tikit.ika- 'rattling', and kirikiria 'whispering' are familiar to readers of the Jain canon. The frequent use of such onomatopoeias is characteristic especially of Apabhra .m~a, where we find for instance jhalajhalai 'to splash', gulugulai 'trumpeting (of elephants)', kilikilai 'to shout for joy' and very many others. Even when the vowels are later altered, which is usual in onomatopoeias where the consonants are most expressive, the vowels still remain repetitive: jhirimiri 'drizzle' which occurs in the Old Gujarati of the Sthfdibhadraphdgu, gives modern Gujarati jharrnar; rimijhimi gives modern Gujarati rumjhum: In the ordinary words of the language, outside the realm of onomatopoeia, there is also very little evidence of a dislike of repetition of vowels and of dissimilation. There are in fact few clear examples of dissimilation of vowels in Middle Indo-Aryan. Some apparent examples have often been explained by other causes. Pkt. garua 'heavy' is now generally derived not from garu(ka), but from the 'gun.a' form of the word, which is found in the Sanskrit comparative and superlative gar~yas and gari.st.ha.7 The fact that the vowel a of the initial syllable is found only in the lengthened form garua and not in guru s does make it highly unlikely that this is a simple 5 For references to some of the theories in connection with dissimilation cf. R. Posner, op. cir., pp. 7-8. 6 ~rf-Sthfdibhadra-phdgu, ed. H. C. Bhayani in the Quarterly of the ~rf Karbasa Gujarati Sabha, Bombay, 1954. See for instance M. A. Mehendale, Historical Grammar of Inscriptional Prakrits (Poona, 1948), p. 7. 8 There are some very isolated instances of the use of garu, instead of guru, particularly in the Pali expression agaru 'not troublesome'. See F. Edgerton, Buddhist Hybrid Sanskrit Dictionary (Newhaven, 1953) s.v. aguru.

SPORADIC CHANGES OF VOWELS IN MIDDLE INDO-ARYAN

27

example of vocalic dissimilation and certainly points to the influence of the longer derivatives of the word which had the vowel a in the initial syllable, that is the forms gadyas and gari.st.ha, already quoted, as well as other derivatives, especially gariman 'weight' > Pkt garima, and garitra 'weight': the longer form garuka was associated with this group rather than with the simple word guru which survived in Middle Indo-Aryan as guru. The Prakrit word purisa, Pall porisa, from puru.sa 'man' is at least partially due to the substitution of the more usual suffix -isa (e.g. as found in words like erisa, harisa); -us.a was also replaced by -asa in pharasa < paru.sa 'rough'. The unusual development in the first syllable of the Pali form porisa has been explained by H. Berger on the basis of the generalisation of the metrically long Vedic form p~ru.sa. 9 One of the possible cases of dissimilation of the vowel i has been interpreted as an archaic form: sa.dhila 'loose', which occurs in Prakrit along with the more usual sidhila < ~ithila, has been explained from *SrthiIa 'loose', ~/2rath 'to become loose') ~ Some of the other words which could be considered as examples of vocalic dissimilation are rare words, or unusual or borrowed, in any case they are words that have no strong links with others; they do not belong to a big etymological group and therefore have little support in the language and are more liable to "sporadic" changes. Such a word is the Dravidian borrowing aguru, Prakrit agaru, agalu, aguru, which was obviously influenced by Skt guru and Pkt garua with which it was associated by popular etymology/1. Sometimes the origin of the word is so unclear that it is difficult to say which is the original vowel. This is the case for instance with the probable Munda borrowing which is represented by Vedic chubuka 'chin', Sanskrit cubuka, cibuka, Prakrit cibua. Equally uncertain are Sanskrit tumburu, Prakrit tumburu, tumbaru and t.imbaru, Pali timbaru 'a tree'; Sanskrit k17i~ja, Prakrit and Pali kila~ja; Sanskrit mucilinda, Prakrit mucilanda 'a tree' as well as Sanskrit jhillikd, Prakrit jhallikd 'an insect'. These words cannot be considered as clear examples of dissimilation of the vowels i and u, but it is probable that dissimilation did play a part in the unusual sound-changes undergone by these rare words. Sometimes the change of vowels is due to analogical influences rather than to dissimilation: thus the Prakrit .neura, n.iura < Sanskrit n~pura
9 H. Berger, "Pall porisa 'Mensch'", Wiener Zeitschrift fiir die Kunde Siid- and

Ostasiens, I, 1957.
10 R . L . Turner, Dictionary of the Nepali Language (London, 1931), s.v.d.hilo. 11 For aguru as a Dravidian loan-word see T. Burrow, The Sanskrit Language (London, 1955), p. 380, and for some of the further history of the word see Jean Fflliozat "L'Agalloche et les Manuscrits sur Bois dans l'Inde", J.A., 1958, pp. 85ff.

28

L. A. SCHWARZSCHILD

'anklet' has been shown already by Pischel to be a remodelling under the influence of keura < keyura 'bracelet'. The phonetic development of the Prakrit bhamaya, bhumaya 'brow' is most difficult. These words are connected with Sanskrit bhrftmukha > Prakrit bhumuha, bhamuha, but the suffix -maya was substituted for the original final of the word. This is indicated already by Hemacandra (II 167). The change of bhumaya to the alternative form bhamaya is probably due to the influence of the original word bhrftmukha > bhamuha 'brow'. The development of bhrgmukha to bhamuha represents what is perhaps the only clear example of vocalic dissimilation in Middle Indo-Aryan. It is typical of examples that abound in other languages, the change being regressive rather than progressive: the earlier of two sounds is altered rather than the later, because of the tendency of the speakers to think ahead. TM The word bhamuha had become completely dissociated from the original bhrft 'brow' and was therefore liable to further sporadic changes in Apabhram. ~a, where it appears as bh6ha. The scarcity of examples shows that dissimilation in Middle Indo-Aryan was very rare indeed, probably in the spoken as well as the written language. The infrequency of vocalic dissimilation in Middle Indo-Aryan, as opposed to the Romance languages for instance, may be explained by the more even accentuation of Middle Indo-Aryan: whereas particularly in French the stress accent was so heavy that vowels in unaccented syllables, preceding the stressed vowels, tended to be dissimilated. It would appear that differentiation on the other hand was not so infrequent in Middle Indo-Aryan, though it has not usually been recognised as such. The most common kind of differentiation is of the type mukula > muula > mai~la 'bud', where the first of the two vowels u is altered as much as it possibly can be and loses its major phonetic characteristics: it is unfounded as well as lowered, becoming a which for some time at least was pronounced in a separate syllable from the following u. There are a number of very similar cases: mukut.a 'diadem' appears as mai~.dain Ardham~gadhi as well as in later Jain texts, mukunda 'a kind of drum' becomes maunda, mukura becomes mai~ra. Some of these words appear in lexical Sanskrit with the vowel a in the first syllable, and sometimes also in Buddhist Hybrid Sanskrit, e.g. makut.a 'crest', 'diadem' (Divy~vad~na), makura 'mirror', 'bud' (lexical Sanskrit), and it is probable that the vowel a in the initial syllable of these forms is due to the 1~ The tendency for dissimilation to be regressiverather than progressive is discussed by R. G. Kent in LanguageXII, pp. 245-258.

SPORADIC CHANGES o r VOWELS IN MIDDLE INDO-ARYAN

29

influence of Middle Indo-Aryan. It is possible that kut~hala < kohala 'pleasure', 'curiosity' shows the same type of differentiation, as suggested by Pischel, and that kutf~hala was changed to *kafthala, whence kohala. The occurrence of the variant form kouhalla in Hemacandra's grammar and of kofthala in the Jain canon might suggest that in some areas perhaps the differentiation did not go so far in this word, and the u of the initial syllable was only lowered to o, and not lowered further and unrounded to a. It is however more likely that the word kut~hala ~ kohala does not represent a straight-forward example of differentiation, but shows the influence of the derivative abstract noun kautfthala ~ kofthala 'curiosity': this is almost certainly the case in the Sauraseni and Jain Sauraseni form kodflhala. The usual Prakrit and Apabhra .m~a word somdla 'tender' has been explained by Pischel as being parallel in development to kohala: sukumdra ~ suumdra, suumdla ~ sai~mdra ~ somdla, except that in this case the intermediate form saftmdra is actually attested in M~h~rS.s.tri (Setubandha). Professor Alsdorfa3 has suggested a different etymology for the word somdla, namely saumya + dla. Even if one is persuaded by the presence of the intermediate form sai~mdra that the old derivation from sukumdra 'tender' is probably correct, it is very likely that the word sukumdra > somdla was at least associated by the speakers with saumya 3> soma 'good', 'gentle' and this may account for the prevalence of the form somdla rather than saiimdra. Although the usual process of differentiation of u-u is thus to a//there are some other cases where it seems that the second of the two vowels has been differentiated to a. The noun duk~la 'a kind of cloth' appears in Middle Indo-Aryan as dufda (H~la), dugulla (Ardham~gadhi and Jain Mah~r~.st.ri) and dualla (Hemacandra). This last form is almost certainly due to the substitution of the suffix -alla for the final fda, -ulla; the influence of the two other words of similar meaning ambara > ambala (Pall) 'cloth', 'garment' and kambala 'blanket', 'woollen garment' was probably not instrumental in this substitution, as these two words do not show a double l in their final. Jugupsa > dugumcha, du(g)amchd 'disdain' is also an example of the complete remodelling of a word rather than simple differentiation; the initial syllable was obviously influenced by the pejorative prefix dur- ~> du-. There is thus little doubt that the usual differentiation ofu-u was to a/L There is a further and barely recognisable example of such a differentiation in sai~risa 'a good man' which occurs in Jain M~h~r~.st.ri and M~h~r~.st.ri. This is probably not always to be de1~ Hariva.mgapura.na yon Pu~padanta, ed. Ludwig Alsdorf (Hamburg, 1936), Glossary

s.v. s6m~la.

30

L . A . SCHWARZSCHILD

rived from satpuru.sa, but sometimes corresponds to supuru.aa, which appears also as suurisa and sf~risa (Hc. 1.8). The two vowels a/~ occurring in hiatus formed a most distinctive group: they were as far apart in pronunciation as possible, and because of this distinctiveness the grouping seems to have been favoured in Middle Indo-Aryan. The same applies to the vowels a~' in hiatus, though perhaps to a slightly lesser degree owing to the probable frontal articulation of the short a in Middle Indo-Aryan. 1~ These two groups, a~'and a/~ thus became very prominent in Prakrit and Apabhra .m~a. They were pronounced in many tatsama words from Sanskrit which had originally contained the diphthongs ai and au that had disappeared from the language by Middle Indo-Aryan. a/~ was quite particularly frequent, and only in some cases the variant semi-tatsama form with o was allowed by the grammarians; a~' was slightly less common and the variants containing e were used more freely.15 The groups a/~ and a~' were also introduced in some cases by differentiation, not only in the examples of u-u > a~ listed above, but also in a few words where there has been further differentiation from eu and el, by the lowering of e to a in contrast with the following higher vowels. This might account for the unusual form alan.a, from e~n.a, eg~n.a < ekona 'lacking one' which is found in numerical compounds in the Jain canon, as for instance in a~.nattisa 'twenty-nine' (Ndyddhammakahdo). The meaning of the word aiLna renders the derivation from agun.a, suggested by Pischel, quite impossible. 16 By a change very similar to that of ekona > eft.ha to ai~.na, Prakrit chaYlla (Gujarati chel) has replaced chei(l)la from Sanskrit chekila 'clever'. A most striking instance of the way in which the distinctive group of vowels a/~ was favoured in Middle Indo-Aryan is shown by the cases in which the u was due to the vocalisation of labial consonants. After vowels other than a such vocalisations do not seem to have occurred until very late in Apabhra .mga and in the modern Indo-Aryan languages; thus sapatn~ 'rival wife' became savatff and sai~tti in Jain M~hfir5.st.ri (Supdsa.ndhacaria), pravahana 'ship' became pai~hana, navati 'ninety' became nai~i. When following on vowels other than a, the labial consonants do not undergo such a development until considerably later, if at all, and we for instance always find devara 'husband's younger brother', and sevaya
1~ For the pronunciation of short a in Sanskrit see W. S. Allen, Sandhi (The Hague, 1962), p. 30, and P. Thieme, Pd.nini and the Veda (Allahabad, 1935), pp. 89ff. and l18ff. 15 R. Pischel, op. cit., p. 58. 1~ Cf. R. L. Turner, A Comparative Dictionary of the Indo-Aryan Languages, Fasc. II (London, 1963), s.v. ekona. The Old Hindi forms agunis etc. might be explicable by a contamination between aiina and eguna.

SPORADIC CHANGES OF VOWELS IN MIDDLE INDO-ARYAN

31

'servant' (apart from an isolated example of seuya < seveka in the Kalpasfctra) 17 and vevaya < vepaka 'shivering' in Middle Indo-Aryan, and forms such as deurdM < devardn. T 'husband's younger brother's wife',

and di~so (Nepali) < divasa 'day' are not usual till the modern languages. When the intervocalic consonants of Sanskrit weakened, and in the case of dentals, gutturals and y disappeared, the use of vowels in hiatus, practically unknown within Sanskrit words, became a prominent feature of the phonetic system. The groups of vowels with the maximum phonetic difference a/~ and to a slightly lesser extent at' were particularly favoured. The frequent use of ai' and a/~ was all-important in lending a certain stability to Apabhram. ~a, where a~ was the final of the nominative singular of all extended masculine nouns, and at' was the final of the third person singular of the verb. Apart form the cases where a/~ and at' resulted from 'regular' sound-changes there are also some words, as shown above, where they are due to differentiation, which therefore has a certain significance in the phonetic development of Middle Indo-Aryan. Vocalic dissimilation on the other hand is so rare as to be insignificant: the extreme scarcity of vocalic dissimilation is connected with the absence of a heavy stress accent in Middle Indo-Aryan.

17 deula<devakula has been omitted from this discussion, as it probably represents a development de(va)-fk)ula > deula, with elision of the syllable va as in deva > de, see Munshi lndological Felicitation Volume (Bombay, 1963), p. 216.

Вам также может понравиться