Sie sind auf Seite 1von 16

1.5. The Aa:dhya:yi: and Modern Linguistics. 1.5.1.

The Aa:dhya:yi:, which has no rival in the Indian tradition, has won the appreciation of western scholars right from the time they came to know about it. They have further elevated it to the level of the best one among all the grammars written so far for any language noting that it is far superior to the Greek and the Latin grammars of anti uity in both methodology and coverage. i The !anskrit language directly and the Aa:dhya:yi: indirectly helped the growth of the comparative grammar in the last uarter of the 1"th century and in the 1#th century. The role of the Aa:dhya:yi: in that great scientific discovery was to present to scholars a clear picture of the morphological structure of the !anskrit language on the basis of which they ac uired a thorough understanding of the morphological structures of Greek, Latin and other Indo$%uropean languages &the ancient Greek and the Latin grammars contained very little morphological descriptions'. ii (oncepts such as the distinction among roots, nominal bases and suffi)es, primary and secondary suffi)es, the analysis of compounds and the system of vowel gradation &ablaut' were taken into comparative linguistics from the Indian grammatical tradition &*ocher 1#+5,-. 1##/,1-0'. 1n the other hand, being a synchronic grammar, it naturally has a direct relationship with the descriptive linguistics of the present century. The linguistics of ancient India has been hailed as 2the direct germinal origin of the linguistics of the 3estern world of today2 &%meneau 1#55,1-5'. It is said that 4loomfield2s realisation of the value of descriptive grammar and the depth of insight that he brought to it arose out of his intimate study of 5a, ini &4loch 1#-#,#6, %meneau 1#"",+55'. 4loomfield praised the Aa:dhya:yi:, which was one of his bedside books &%meneau 1#"",+5+' as 2one of the greatest monuments of human intelligence2 and characterised it as 2an indispensable model for the description of languages2. 7s coming from a person who has contributed greatly for the development of linguistics of the modern era, his views on the Aa:dhya:yi: must be uoted here in full, The descriptive grammar of !anskrit, which 5a,ini brought to its highest perfection, is one of the greatest monuments of human intelligence and &what concerns us more' an indispensable model for the description of languages. The only achievement in our field which can take rank with it is the historical linguistics of the nineteenth century, and this, indeed, owed its origin largely to %urope2s ac uaintance with the Indian grammar. &1#/#,/8"' This scientific condensation, which places every feature into its proper setting, is one of the two virtues which make Indian grammar a model for us. The other is completeness. 5a,ini gives the formation of every inflected, compounded, or derived word, with an e)act statement of the sound$variations &including accent' and of the meaning 9 :or no language of the past have we a record comparable to 5a, ini2s record of his mother$tongue, nor is it likely that any language spoken today will be so perfectly recorded. &1#/#,/+-' This grammar, which dates from somewhere round 056 to /56 4.(., is one of the greatest monuments of human intelligence. It describes, with the minutest detail, every inflection, derivation, and composition, and every syntactic usage of its author2s speech. ;o other language, to this day, has been so perfectly described. &1#00,11' 4loomfield adopted 5a,ini2s style of conciseness and 2hatred of repetition2 in writing his book, Language, which is difficult for beginners and 2even relatively advanced students 9 because of 4loomfield2s concise, closely$knit, unemotional, uasi$mathematical reasoning and presentation, in which each sentence must be thoroughly understood and remembered before proceeding to the ne)t.2 &<all 1#-#,1/1. %meneau 1#"",+55' 5a,ini2s grammar can be considered as the immediate source for 4loomfield2s e)position of such topics as morphology, morphophonemics, suppletion, =ero,

analysis of compounds, form$classes and generality and specificity &*ogers 1#"+,1/#'. 3ith regard to morphology, 4loomfield &1#00,/6"' himself has remarked that 29 in the main, our techni ue was developed by the <indu grammarians2. The importance of =ero element has been emphasi=ed by him &1#00,/6#' in the following words, In other cases there is not even a grammatical feature, a single phonetic form, in the manner of homonymy, represents two meanings which are distinguished by means of a linguistic form, as, singular and plural noun in the sheep (grazes) : the sheep (graze). <ere the <indus hit upon the apparently artificial but in practice eminently serviceable device of speaking of a zero element, in sheep , sheep the plural$suffi) is replaced by zero $$$ that is, by nothing at all. 7lthough, at the time of writing his 1#/# article, 4loomfield was critical of 5a, ini for using the techni ue of ordered rules &see >1.-.1+', he must have realised the importance of it at a later time since e)plicit ordering of rules was resorted to by him in 2?enomini ?orphophonemics2 &1#0#' and 2%astern 1@ibwa2 &1#5"' &*ogers 1#"+,1/#'. 1.5./. %ven after 4loomfield2s time, the Aa:dhya:yi: with its 2surprisingly modern2 methods continues to attract the attention of scholars interested in linguistic theory. iii <ere we do not see any direct influence of 5a,ini but the e)istence of a number of parallels between the methodology of 5a,ini and that of these new schools has been noticed &see (ardona 1#+8a,/0/$08. *ocher 1#+5,08$ -5'. The originator of the transformational generative grammar, (homsky himself has observed, 23hat is more, it seems that even 5anini2s grammar can be interpreted as a fragment of such a Agenerative grammar" in essentially the contemporary sense of this term.2 &1#85,v'. 7lthough 4loomfield had categorically stated that 5a,ini2s grammar contained synta) &see the uotation from his 1#00 work in >1.5.1.', his statement was neglected by scholars for uite some time afterwards. Thieme &1#58,-' stated that 5a,ini2s grammar, 29 being defined as abda,nua,sana Aan instruction in &the formation of ' word$forms9A, is not concerned with sentences2. in addition, the absence of any statement on word order &noted elsewhere by Thieme' also seems to have contributed to such a view.iv That this is an erroneous view has been pointed out by scholars like (ardona &1#8+,/10, note 1#', Boshi &1#8#,18', van ;ooten &1#8#,/-/$-0', !taal &1#+8,116' and Ciparsky &1#"/,1'. v It is, however, remarkable that the syntactic component of 5a, ini2s work has been recognised and studied assiduously only after synta) received its due attention in general linguistics with the advent of (homsky. %ven in ancient times, some people seem to have entertained this wrong notion as is evident from Binendrabuddhi2s &+th century 7.D.' remark against it. vi !ince he was describing a language with comple) morphological structure, 5a,ini had to devote a ma@or part of his work to the morphological analysis. the natural preponderance of the morphology seems to be the root cause of the misunderstanding. The ka,raka theory &see (hapter "' and the treatment of comple) sentences with the infinitive are two of the important areas that vouch for 5a,ini2s concern for synta). 1.5.0. It is important to note that 5a,ini2s system cannot be compared in toto with any of the modern systems despite the e)istence of partial resemblances. !taal &1#88', Boshi &1#8#' and Ciparsky and !taal &1#8#' have noted similarities between the Aa:dhya:yi: and transformational generative grammar in areas like sentence structure and the derivation of compounds. 5a, ini, however, does not derive one sentence from another, nor does he employ devices like phrase structure rules, embedding and so on &note that many of these as well as transformational rules are abandoned in the latest version of transformational grammar'. :illmore2s &1#8"' case grammar closely resembles the ka,raka system of 5a, ini &see (hapter "' in that, in the former, case roles which are based on semantic criteria and hence belong to deep structure are manifested by surface

categories like sub@ect and ob@ect without there being a one$to$one relationship between the two classes &7nanthanarayana 1#+6'. Eet there are many areas of difference between the two of which the following may be noted, &i' In case grammar deep structure categories such as 2agentive2 and 2ob@ective2 are defined in purely semantic terms in an attempt to see that they are applicable to all languages while in the ka,raka system semantic and syntactic criteria are combined in defining some of the ka,rakas &(ardona 1#+8a,/00'. &ii' The concept of 2sub@ect2 as opposed to 2agent2 employed in case grammar is absent in 5a,ini2s system. &iii' In 5a,ini2s system even an inanimate ob@ect can function as karta, 2agent2 if the speaker wishes to attribute the uality of independence or self$ promptedness to it &cf. svatantra karta: 1.4.54, >".8.'. :illmore &1#8",/-' originally insisted that the agent must always be animate but (ook &1#+#', in his revised version of case grammar, admitted the possibility of an inanimate ob@ect functioning as the agent &!arma 1##5,10"'. The use of partially ordered rules in accounting for morphophonemic changes is a remarkable similarity between 5a,ini and modern linguistics &see >1.8.11.$.1-.'. The level of understanding the techni ue of 5a,ini grows with the level of development of various aspects of descriptive linguistics. 4ecause of its strong theoretical background, the Aa:dhya:yi:, though composed in the hoary past, has played an important role in the evolution of various branches of modern linguistics and will retain its attraction for linguists for a long time to come. The terms sandhi and ka,raka &the former current from very early times and the latter in recent times' and less commonly the names of compounds tatpurua &T.T. 16#' and bahuvri,hi &T.T. 186' that have entered into the terminology of linguistics bear elo uent testimony to the great relevance of 5a,ini2s work to modern linguistics. 3e may close this section with what Ciparsky &1##-,/#/0' has to say on the sub@ect under discussion, 2?any of the insights of 5a,ini2s grammar still remain to be recaptured, but those that are already understood constitute a ma@or theoretical contribution. Its impact on generative grammar was felt first in phonology &the %lsewhere (ondition, unmarked rule ordering', and more recently in synta) &Linking theory'. The rewards for the currently burgeoning study of morphology promise to be richer still.2 1.6. Pa:ini's principles o analysis and presentation. 1.8.1. 5a,ini2s grammar may, at the theoretical level, be visuali=ed as one that starts with semantics and ends with the phonetic realisation of the word, word group or the sentence, which is derived by the application of a series of rules. It should however be noted that there is no separate section either for semantics or for synta) and that phonetics is not discussed in the work but a knowledge of it &which is dealt with in detail in the ikas and 5ra,ti$a,khyas' is assumed on the part of the student of the grammar for the proper application of the morphophonemic rules &cf. stha:ne' ntarata!a 1.1.5". F>8.+.G and tulya:syaprayatna! savara! 1.1.#. see also >1.8.16'.vii ?orphology, which is not separated from but is interlinked with semantics and synta), is presented in the three 7dhya,yas, the 0rd to the 5th &>/.0.'. it is the focal point of the grammar. 7ll the suffi)es that occur in the language are given in this part of the work. Those of the suffi)es that can be added to a root are given in the section governed by the adhika,ra su,tra, dha:to $.1.#1 &>/.0.-.'. the &underived' roots are listed in the Dha,tupa,ha &>1.-.1.'. The suffi)es that can be added to a nominal stem are given in the section governed by the adhika,ra su,tra, ya:ppra:tipadika:t 4.1.1. &i.e. the -th and the 5th 7dhya,yas, >/.-.1.'. !ince underived nouns of !anskrit as those of any other language constitute an open class and since the inflectional morphology of nouns is more or less regular and the variations involved are largely predictable on the basis of the final sound of the nominal stems, the latter are not listed in the grammar &e)cept a few of them mentioned in the Ga as FT.T. "-G for specific

grammatical operations'. They have to be taken from other sources like a le)icon, which, however, includes in it both the underived and the derived ones, or one has to depend on his knowledge of the language for this purpose. The roots, underived nominal stems and suffi)es are the three kinds of basic elements that serve as the input for the grammar. In addition, the word &an inflected noun' also serves as the basis in the derivation of denominal verbs &>>+.1.0.$.8', secondary derivatives &>+.0./.' and compounds &>+.-.".'. The case suffi) in such formations is deleted by supo dha:tupra:tipadikayo %.4.&1. &7pp.I,1'. In fact, in deriving these morphologically comple) formations, 5a,ini accounted for their meanings and syntactic functions by constant reference to the phrases or sentences that correspond to them. The grammar generates the actual forms of the language by combining together these basic elements and making the necessary changes in their phonological shapes by framing appropriate rules. Hariations in morphological structure are accounted for by one of the three processes, &i' substitution &>1.8.5.', &ii' preference for e)ception &or blocking, >1.8.8.' and, &iii' augmentation &>1.8.".'. The process of substitution is adopted in describing inflection and morphophonemic changes. %ven the =ero element &with all its subdivisions' is introduced as a substitute &>1.8.+.'. *eduplication of the root is treated as a concomitant feature of adding the suffi)es li &>16././.', $sa&n' &>+.1./.' and $ya&' &>+.1.+.'. In morpho$phonemic rules, the substitution of one sound for a se uence of two sounds is also found &eka,dea FT.T. 8+G, >>1.8.16., /.8.5.'. The techni ue of assigning a suffi) as an e)ception to the general one is used in derivation. the convention to be followed in such cases is that the e)ceptional suffi) takes precedence over the general one under stated conditions. 7lthough 5a, ini has not defined the term va,kyam 2sentence2, he recognised it as a grammatical term & abdasa@Ia,, cf. vaco' a'dasa()a:ya:! &.$.6& 2the c of vac$ 2to speak2 is not replaced by k before $& 'ya&t' unless the derived word is a grammatical term2, e.g. va,c$ya$m 2that which can be said2 but va,k$ya$m 2sentence2' and made use of the concept in a number of su,tras to account for the facts of the !anskrit language &for further details, see va,kyam FT.T. 1#+G'. 7dhya,yas 8 to " take care of the changes that occur in the sounds &including those in accent' when a suffi) combines with a nominal stem or a root or when words combine together with one another in a compound or a sentence. 1.8./. 3hen 5a,ini assigns any suffi), he specifies the following five characteristics of it by general or particular su,tras &it is hardly necessary to point out here that some or all of these may not be present in the particular su,tra but are read into it through the techni ue of anuv tti FT.T. 1/G', &i' The meaning of the suffi) &>1.8.0.'. &ii' The particular class of the base to which it is added &i.e. nominal stem or root with specification of the subtype or the individual item when re uired'. &iii' The morphophonemic changes it causes in the base. These changes, which are chiefly conveyed by the indicator &see (hapter -' and which may sometimes affect the base and the suffi) concurrently include gua or vddhi substitution of the first or the final vowel &the first vowel of a stem commonly undergoes vddhi before a secondary derivative suffi), see >+.0.0.', substitution of one sound in the place of one or two consecutive sounds &in the latter case, the first sound is the final one of the baseJprevious suffi) and the second sound the first one of the suffi) that follows', deletion of a sound, addition of a new sound or sounds &i.e. augment FT.T. 05G' and the substitution of an entirely different base. In some cases, an entire suffi) is substituted by a new one under specified conditions. for an e)ample, see a:asiasa:! ina:tsya: &.1.1%* +>#.8.1.. also see >1.8.5.'. &iv' The position of the main accent in the derived word &this also is mostly conveyed by the indicator, see >>/.8.#.,-.0.1.&iv' '. &v' The syntactic behaviour &like cooccurrence restrictions' of the derived word &>1.8.-.'.

1.8.0. The meaning of a suffi) is taken as primary and the suffi) is assigned for the purpose of e)pressing the meaning that is conveyed by it. In the su,tra that assigns a suffi), the word that indicates the meaning is always given in the +th case, which, according to the !anskrit idiom sati saptami,, can be e)panded by adding an appropriate word in the +th case as abhidheye or va,cye as has been done by the commentators in giving the v tti &see note 05 for e)amples'. The construction then means 2when meaning ? is to be e)pressed2. Therefore, for e)ample, kar!ai dviti:ya: %.$.%. &>".".1.'' means 2the /nd case occurs &after a nominal stem' when the ob@ect &karman$' is to be e)pressed2, hetu!ati ca $.1.%6. &>+.1.8.' means 2when the causative sense &hetumat$' is to be e)pressed, the suffi) $&'i&c' is added to the root2 and varta!a:ne la $.%.1%$. &>16.1.1.' means 2the suffi) la occurs after a root when the present tense &vartama,na$' is to be e)pressed2. viii Thus the grammar takes the meanings as the starting point and generates the utterances that represent them much as a human being produces sounds that represent the idea he wants to convey. It may not be out of place to mention here that 5a, ini used semantics in two other ways. !emantic features, like phonological features, were used for identifying classes of words, roots or stems. :or e)ample, dvandva ca pra:itu:ryasena:ga:na:! %.4.%. says that a co$ordinative compound of nouns that denote body parts, parts of a musical group or of an army will optionally be in the neuter sg., e.g. pa,ipa,dam 2hands and feet2, ma,rdagikapa,avikam 2&a group of' players of &the musical instruments' mdaga and paava2, rathika,va,roham 2&an army consisting of' charioteers and cavalry2. In this the classes of words are identified by their meanings. In cases like pra:istha:d a:to la( anya,tarasya:! 5.%.#6. 2a stem denoting a thing found on the body of an animate being and ending in a, optionally takes the suffi) $la&c' to denote the sense of possession2 &e.g. cu, a,$la$Jcu, a,$vat$ 2one who has a crest2', the semantic criterion is combined with a phonological one. :urthermore, the meaning or the shade of meaning of the whole word to be derived is used by 5a,ini as a condition in deriving words as well as compounds. :or e)ample, according to a:' ci 4.1.#, the feminine suffi) is $&'a,&p' when the entire word refers to a Hedic stan=a &e.g. dvipad$a, k 2a Hedic stan=a with two lines2' but it is $& 'i,&p' otherwise &e.g. dvipad$i, devadatta, 2a woman by name Devadatta, who is two$legged'. also cf. vaco' a'dasa ()a:ya:! &.$.6&. uoted in >1.8.1. In the case of compounding, the whole compound may refer to a particular ob@ect, activity or person or it may carry additional overtones like censure $$$ factors like these are sometimes relevant for the grammar. This is illustrated by trikakut parvate 5.4.14&. trikakut with the change kakuda$ K kakut$ refers to a particular mountain with three cliffs while any other thing with three protrusions will be trikakuda$. 1.8.-. That the suffi) assigning su,tras carry, when necessary, syntactic information will be clear from the following e)amples. The four su,tras beginning with yu!ady upapade - 1.4.1"5 &see >16.6.+.', which were cited by the ancient commentator Binendrabuddhi to establish that grammar is concerned with synta) also &see >1.5./., note 00', are fairly simple e)amples inasmuch as they stipulate agreement between a nounJpronoun &e)pressed or understood' and the finite verb that is coreferential to it. 5a,ini2s concern for precision is evident in these su,tras in two ways. :irst, by stating in the first one of these stha,niny api 2even when it is absent2, which is to be read into the subse uent su,tras also, he overtly accounted for the absence of the nounJpronoun in the sentence. !econdly, he ensured that the correct finite verb suffi) is selected when the finite verb is simultaneously in agreement with a first and a second person pronoun as in aham ca tvam ca gaccha,va 2I and you &sg.' go2 by making as!ady utta!a 1.4.1"&. a later rule so that it will have precedence over 1.4.1"5. according to the metarule, vipratiedhe para! ka:rya! 1.4.%. &>8.06'.

dha:to kar!aa - $.1.&. &>+.1./.' assigns the $sa&n'$ suffi) optionally in the sense 2to desire2 to a root only when the action denoted by the root is the ob@ect of i $ 2to desire2 and when both the root to which the suffi) is to be added and the root i $ &which is understood' have the same person as the agent, e.g. devadatta kartum icchati L ciki,r$$a$ti 2Devadatta wants to do &something'2. In this e)ample, k$ 2to do2 took the $sa&n'$ suffi) because Devadatta is the agent for the actions e)pressed by both the roots k$ and i$ as stipulated by the su,tra. 1n the other hand, the sense of devadattasya bho@anam icchati ya@Iadatta 2Ea@Iadatta desires Devadatta2s eating2 cannot be e)pressed by ya@Iadatto bubhuk$$a$ti since in the former sentence the agent of bhu@$ 2to eat2 is different from that of i$. therefore, the latter sentence means only 2Ea@Iadatta wants to eat2. sa!a:nakartkayo pu:rvaka:le $.4.%1 says that $&k'tva, is added in the sense of 2earlier &action'2 to a root when the latter is associated in the sentence with another root provided that the agent for both the actions is one and the same, e.g. devadatto bhuk$tva, vra@ati 2having eaten, Devadatta goes2. The condition that both the actions should have the same agent &sama,nakart ka$' in the above su,tra prevents the use of the suffi) in a sentence like ya@Iadatte bhuktavati gacchati devadatta 2Devadatta goes after Ea@Iadatta eats2 in which the agents of the two actions are different. The use of a case form depends upon one of the following two factors, &i' The ka,raka role of the referent with respect to the action denoted by the verb &the ka,raka roles come under semantics to a great e)tent but in a few cases the semantic criterion is combined with a syntactic one, see >".6.5.'. &ii' The word &other than a verb' with which it is connected &see >".6.8.$.#'. It should be noted that the actual form of the verb in a sentence is decided first and the nouns take the cases according to it &see >".6.+.'. The second factor is purely a syntactic one. for e)ample, the word nama 2salutation2 re uires the noun connected with it to be in the -th case, e.g. agnaye nama 2salutation to 7gniM2 &see >".16.-.''. The use of a particular laka,ra &T.T. 1+8' in a sentence sometimes depends upon the presence of another word as shown by the following e)amples. 1nly lu 2aorist2 is used with the prohibitive particle ma,&', e.g. ma, ka,ri, 2Don2t doM2 &!a:i lu $.$.1&5. F>16.#.1.G'. li 2perfect2 &>16./.' is normally used to denote unseen past events that took place before the current day & paroke li $.%.115.' but la instead of li can optionally be used to e)press the same sense when the word ha &a particle of emphasis' or avat 2repeatedly2 is there in the sentence +haavator la ca $.%.116. F>16.+.1.G', e.g. iti ha N akarot &la'Jcaka,ra &li' 2heJsheJit did so2. 1.8.5. The variations in form that occur in the case suffi)es &sup FT.T. /-1G', finite verb suffi)es &ti FT.T. 110G' and in some others like the gerundial suffi) $&k'tva, are stated by treating the more commonly occurring form as the basic one and the others &including =ero, see >1.8.+.' as its substitutes. :or e)ample, for the 0rd case sg., -th case sg., 5th case sg. and the 8th case sg., the most common variants, namely, $&'a,, $&'e, $&'as&i' and $&'as respectively are taken as the basic ones and those that occur only after a stem ending in a short a, namely, $ina, $ya, $a,t and $sya respectively are given as the substitutes of the former & a:asiasa:! ina:tsya:h &.1.1%* er ya &.1.1$. F>>#.8./., #."./.. Table #.1.'. In the case of the derivation of the finite verb, an abstract suffi) called la&ka,ra' &T.T. 1+8' to be added after a root is set up &for details, see >16.6.'. 7ny one of the ten laka,ras is first substituted by one of a set of 1" suffi)es called ti &T.T. 110' and it, in turn, is substituted by some other one or undergoes change in shape depending upon the particular laka,ra that is the source for it and also upon the type of the preceding root. The class sign suffi), the passive suffi) and some others are intoduced with overtly stated conditions as occurring after a root before a ti suffi) &see >16.6.#.'. The gerundial suffi) $&k'tva, &sa!a:na - $.4.%1, see >1.8.-.' is substituted by $&l'ya&p' after a root that is compounded with a gati word &T.T. "5' or a preverb &T.T. 5"' but not with the negative na&I'$ & sa!a:se' na)pu:rve ktvo lyap &.1.$&.., e.g. pra$h$&k'tva, K

pra$h$&l'ya&p' K pra$ht$ya 2having hit2 &augment t&uk' added by hrasvasya piti kti tuk 6.1.&1. 2the augment t&uk' occurs after a short vowel before a primary derivative that is a p$it2'. If the form ultimately re uired can be obtained by changing or deleting one or more of its sounds, this method is also resorted to. :or e)ample, by ita a:t!anepada:na:! er e $.4.&#. F7pp. I,/G the final vowel &N consonant' of an a,tmanepada suffi) that is the substitute of a -it laka:ra is replaced by e and by ita ca $.4.1"". F>16.+./.G' the i of the parasmaipada substitute of a -it laka:ra is deleted. ;ote that in the 5a,inian system a substitute is identical with the substituend for all purposes e)cept the phonological ones &stha:nivad a:deo' nalvidhau 1.1.56. F>8.10G'. 4y virtue of the same metarule, the former also inherits the indicators of the latter. in the case of the gerundial form given above, for e)ample, the substitute $&l'ya&p' also is considered to be a k$it like its substitutend $&k'tva, so that the gua change & K ar' is prevented &cf. the e)ample bhuk$tva, in >1.8.-.. see >-.-.11'. In accounting for the allomorphic variations, 5a,ini observed economy in statement by putting together in a su,tra the substituends and the substitutes that occur under one and the same conditioning &e.g. a:asiasa:!ina:tsya: &.1.1%. F>#.8./G' and by capturing the generality involved in the change of sounds in more than one suffi) &e.g. ita a:t!anepada:na:! er e $.4.&#. F7pp.I,/G, er u $.4./6. which latter applies in the 0rd sg. and pl. parasmaipada forms of lo 2imperative2 F>16.8./.G'. 1ne, therefore, should not e)pect to find all the variants of a suffi) in one place or separate statements of variants for each of the suffi)es when they can be brought under one general rule. 1.8.8. 1n the other hand, the primary &>+./.' and the secondary derivatives &>+.0.' that have one and the same sense but occur in different environments, the feminine suffi)es &>+.5.' and even the class signs in the verb &>16.1.-.' are treated as separate suffi)es. In such cases, the more widely occurring suffi) is taken as the general one and the suffi)es that are of restricted distribution are treated as e)ceptions. The domain of the general one lies outside that of the e)ceptional one&s' &see >1.8.#.'. :rom this, it is clear that the relationship that obtains between a morpheme and its allomorphs in modern linguistics is represented by two different types in 5a, ini , substitutes &a,dea FT.T. 0"G' of the substituendum &stha,ni, FT.T. /-8G' and restricted suffi) &apava,da FT.T. 1"G' as opposed to the general one &utsarga FT.T. 5/G'. 5a,ini makes use of one of the three types of conditioning in stating the occurrence of a substitute or a restricted suffi), &i' phonological, &ii' morphological, and &iii' a combination of the above two, which may be called phono$morphological. 7n illustration for the phonological conditioning is that under which the variants $ina &of the 0rd case sg.', $a,t &of the 5th case sg.' and $sya &of the 8th case sg.' occur, namely, after a stem ending in a short a &a:asiasa:! ina:tsya: &.1.1%. F>#.8.1.G. for e)amples from the feminine suffi)es, see >+.5.+. &i'.'. 7n instance for morphological conditioning is the substitutions -th case pl. $bhyas K $abhyam, 5th case pl. $bhyas K $at and 8th case pl. $a,m K $a,kam which take place only after yumad$ and asmad$, the second and the first person pronominal stems respectively &>#.6.-.. for e)amples from the feminine suffi)es, see >+.5.+.&ii'.'. The various listings in the Ga apa,ha and the Dha,tupa,ha are all illustrations for morphological conditioning. The subtypes of it, apart from listing, include conditioning &i' by a neighbouring suffi) which may sometimes be referred to by its indicator so that other suffi)es with the same indicator can also be taken in, and &ii' by the particular type of compound &for e)amples, see >+.5.+.&ii'.'. There is ample evidence to show that 5a,ini used a third type of conditioning, namely, the phono$morphological conditioning, which should also be recognised in modern linguistics &!ubrahmanyam 1##6, >/1'. This type of conditioning involves a combination of a phonological feature with a non$phonological feature &like gender, meaning, pronominal status or @ust an e)clusive group of stems' and, as such, is different from the other two

types. The following stem classes which serve as the conditioning factors for accounting for the suffi) variants in nominal inflection are e)amples for this type of conditioning, &i' ;euter stems ending in a short a &after such a stem, 1st sg. $s&u' and /nd sg. $am are replaced by $am while, after the rest of the neuter stems, the suffi)es are deleted, >>#.1.,#.-.'. &ii' ;euter stems ending in a vowel, a plosive, , , s or h &such a stem takes the augment n&um'before 1st and /nd pl. $& 'i, e.g. vana,n$i 2gardens2 &stem vana$' versus catva,r$i 2four &neut.'2 &stem catur$' '. &iii' ;euter stems ending in i, u, &such a stem takes the augment n&um' before a suffi) beginning with a vowel, e.g. va,ri $e 2for water2 &stem va,ri$' versus manas$e 2for mind2'. &iv' :eminine stems ending in $a,&p'. i) &v' :eminine stems of the nadi, class &T.T. 1/-'. &vi' The ghi class &T.T. #6'. &vii' The a class &T.T. /1"'. &viii' 5ronominal stems ending in a short a &versus a non$pronominal stem ending in the same vowel'.) 1.8.+. 3e have noted above &>1.5.1.' that 4loomfield introduced the concept of =ero element into modern linguistics under inspiration from 5a,ini. 5a,ini used four different technical terms, namely, lopa &T.T. 1#6', luk &T.T. 1"/', lu &T.T. /1+' and lup &T.T. 1"5' in the sense of 2disappearance2 &adaranam. adarana! lopa 1.1.6"'. 3hile the first one is used for the disappearance of a sound &in sandhi' as well as that of a suffi), the other three are used only with reference to a suffi) &pratyayasya luklulupa 1.1.61.'. The =ero elements everywhere are given as substitutes for an actual, that is, overtly present form of the language $$$ either a sound &or a se uence of sounds' or a suffi). In other words, =ero is never an independent suffi) without being connected with another overt suffi). %)amples for the use of lopa for the deletion of a sound, ho he lopa /.$.1$ 2 h K O before h2, ro ri ".0.1- 2r K O before r2. )i %)amples for its use with regard to a suffi), halya:''hyo 6.1.6/ &>#.1./.' which deletes 1st case sg. $s&u' after a stem ending in a consonant or the feminine suffi) $&'i, or $a,&p' &other details are ignored here' and ehrasva:t sabuddhe 6.1.6# &>#.1.5.', which deletes the vocative &1st case' sg. $s&u' after a stem ending in e, o or a short vowel. The difference between the use of the word lopa on the one hand and the remaining three on the other is that while in the former case the deleted suffi) is considered to be present for all purposes &pratyayalope - 1.1.6%. F>8.1+G, a suffi) deleted by one of the three terms with lu will become ineffective to cause any change in the preceding stem &na lu!ata:gasya 1.1.6$ F>8.1"G'. %)amples for the three types of =ero with lu are given in >8.1". and under the respective technical terms. 1f the three, luk is more common than the other two. 3hile lu occurs only after the hu$ class of roots as the substitute of the class sign $& 'a&p'$ and causes reduplication of the root &>16.1.-.', lup occurs only in a few secondary derivative forms &the retention of the plural number and gender of the stem is a peculiarity associated with the lup deletion'. 1.8.". The e)tra meaningless phonological material that crops up in certain conte)ts when two morphs are @oined together poses a problem of analysis for the linguist. The treatment of such e)tra sound&s' &the !anskrit term for which is a,gama 2augment2, T.T. 05' by rules with the convention that an augment forms part of the element to which it is added is an important trait of 5a, inian grammar &see yada,gama,s tadgui,bhu,ta,s tadgrahaena ghyante Fno. 11 of the Paribha:enduekharaG '. 7n augment may comprise a single sound or a se uence of sounds and it, unlike a substitute, does not replace anything but is added to a sound or morph under stated conditions.)ii In the su,tra that assigns an augment, the element to which it is added is always put in the 8th case as in the case of a substitute. 7 $it augment is added at the beginning &>-.-./6', a k$it

augment at the end &>-.-.11' and a m$it augment is added immediately after the final vowel &>-.-./"' of the element given in the 8th case in the su,tra. 3hat is called 2empty morph2 &e.g. the r in children. note that the term is a self$contradictory one since a morph by definition cannot be empty, i.e. meaningless' by some scholars in modern linguistics corresponds to 5a, ini2s a,gama but 5a,ini did not give it the status of a pratyaya 2suffi)2. It sometimes matters in !anskrit grammar whether the augment is &considered to be' added at the end of the stem or at the beginning of the suffi). The advantages gained by the assignment of augments by general rules are the following, 1. (apturing the commonness in the allomorphic distribution of different morphemes. :or e)ample, the assignment of the augments a,& ', ya,&' and sya,&' to $it case suffi)es &see (hapter #, note "' captured the partial commonness of the suffi)es and at the same time avoided the multiplicity of allomorphy in each of the suffi)es. /. 7voidance of setting up of a number of partially similar allomorphs. :or e)ample, the assignment of the augments s&u ' and n&u' allowed 5a,ini to abstract the common element $a,m as the basic suffi) from the three variants of the 8th pl. suffi) $a,m, $sa,m and $na,m &>#.11.'. thus avoiding the multiplicity of allomorphs at the same time. 1.8.#. 1ne important principle that is normally followed by 5a, ini in presenting the grammar is to give the statistically predominant form or process as the general one &utsarga FT.T. 5/G' without any conditioning or restriction on it and give the less common one related to it as the specific or restricted one &apava,da FT.T. 1"G'.)iii This principle holds good in many areas of the grammar like &i' the assignment of suffi)es, &ii' morpho$phonemic substitutions, and &iii' the definition of technical terms. The general rule, although given first, will not apply automatically to all cases that could come under its purview but will apply only to such cases that are not covered by the specific rule&s'. This convention, which should be scrupulously followed in interpreting the su,tras, has been stated in the metarules &framed by the commentators', pu,rvam hy apava,da, abhiniviante paca,d utsarga, 2the restricted rules operate first and the general ones only later &that is, the latter apply in such cases not covered by the former'2, prakalpya ca,pava,daviayam tata utsargo2 bhiniviate 2the general rule operates avoiding the domain of the restricted one&s'2 &nos, 8/, 80 of the Paribha:enduekhara'.)iv %)amples in the case of suffi)es, kar!ay a $.%.1. 2a root having a complement whose meaning is the ob@ect of the activity denoted by the former takes $a& '2 is a general rule accompanied by the specific rule, a:to' nupasarge ka $.%.$. 2&under the same conditions' a root that ends in a, and is not preceded by a preverb takes $&k'a2. Therefore, in the case of n $pa,$, $&k'a alone occurs since the root pa,$ 2to protect2 ends in a, and is not preceded by a preverb &n $pa,$&k'a$ K n$p$a 2king, one who protects people &n$'2 Fa, K O by a:to lopa ii ca 6.4.64 2stem final a, K O before the i& ' augment and an a,rdhadha,tuka &T.T. -0' suffi) that begins with a vowel and is a k$it or $it2G'. $a&' occurs in cases not covered by this and some other specific rules, e.g. kumbha$k$a&'$ K kumbha$ka,r$a$ 2potter2 & K a,r by aco )iti &.%.115. F7pp. I,1+G'. 1f the ten class signs in verbs &>>16.6.#., 16.1.-.', $&'a&p' is the general one assigned without any specific conditioning on the root &kartari ap $.1.6/ 2$&'a&p' occurs after a root before a sa,rvadha,tuka suffi) &T.T. /0#' that denotes the agent2' and the others are assigned after specific classes of roots &e.g. diva:di'hya yan $.1.6# 2$&'ya&n'$ occurs after the roots of the div$ &i.e. the -th' class &other conditions being the same'2 '. Therefore, the former occurs in cases left out by the latter ones, that is, after a root of the first and the tenth classes and after a derived root &note that after the second and the third classes of roots also it occurs at the underlying level but is replaced by luk &$O$' and lu &$O$' respectively'. The operation of this principle on a large scale can be seen in the taddhita section in which subdivisions are mostly made on the basis of a handful of general suffi)es &see >>/.-.-. $ /.-.".'. %ach of those general suffi)es

applies only in cases which are not covered by the specific suffi)es &see >+.0.1.'. %)amples for morphophonemic substitutions, The general rule ho ha /.%.$1 2h K h before @hal &T.T. 160' and at the end of a word2 is supplemented by a few specific rules beginning with da:der dha:tor gha /.%.$% 2the h of a root beginning with d is substituted by gh &under the same conditions'2. In a root like dah$ 2to burn2 which begins with d, the h will be substituted by gh and not by h and h K h obtains only in the cases not covered by the specific rules, for e)ample, in lih$ 2to lick2 &see note -1'. In the finite verbs, er u $.4./6 by which the i of the basic suffi)es that are the substitutes of lo &>16.8.' is substituted by u is the general rule which is supplemented by two specific ones, ser hy apic ca $.4./& and !er ni $.4.// which teach the substitutions, /nd sg. &5.5.' $si&p' K $hi and 1st sg. &5.5.' mi&p' K $ni respectively. Therefore, $.4./6 will not apply to these latter cases but will apply only elsewhere. 7n e)ample in the realm of technical terms is as follows. The term bham &T.T. 181' applies to a nominal stem before a nominal suffi) &other than a sarvana,mastha,nam FT.T. /0+G' beginning with a vowel or y & yaci 'ha! 1.4.1/'. This is modified by siti ca 1.4.16 which says that before such a suffi) that has s as the indicator the preceding stem is a padam &T.T. 105'. The term bham will not be applicable in the latter case. 1.8.16. The morphophonemic rules occupy a large part of the Aa:dhya:yi:, namely, the 7dhya,yas from the 8th to the "th. !ome of them are given even in the earlier 7dhya,yas along with the suffi)es when they are individualistic and not amenable to any generalisation. :or e)ample, pu:takrator ai ca 4.1.$6 assigns the feminine suffi) $&'i,&p' after pu,takratu$ simultaneously replacing the stem final u by ai so that pu,takrata,y$i, 2wife of 5.2 is derived. The morphophonemic rules account for partial or complete change in the base as well as in the suffi)es and for the augments that crop up between them &>1.8.".'. The aga,dhika,ra &>/.8./1.' is specifically devoted to the changes in the nominal stems and roots when they come into contact with specific suffi)es or groups of suffi)es. In morphophonemic rules, the basic form, which undergoes change, is given in the 8th case &>8.8.', the substitute or the augment in the 1st case and the conditioning factor is given in the 5th case if it precedes &>8./6' and in the +th case if it follows &>8.1#'. 7s far as possible, the rules are framed so as to bring out the similarity in the behaviour of sounds that form a natural class. the pratya,ha,ra su,tras &see (hapter 0' have been designed only for this purpose. The allomorphic variations are normally accounted for by changes in some of the sounds of the basic form. where this is not feasible, the entire basic form is replaced by a new one. 3hile the substitution of one sound for another is the norm, there are cases in which a single sound occupies the place of two successive sounds. the latter are given in the single substitute section &8.1."-. $ .111, >/.8.5.'. There are certain su,tras like ho ha /.%.$1 &see note -1' which directly replace one sound by another. 4ut usually 5a, ini captures the generalisations involved in changes of similar sounds by stating changes in features utilising the concepts of sa,varya 2phonetic similarity2 &T.T. /0"' and a,ntaratamya 2ma)imum similarity2 &see >8.+.'. :or e)ample, (hayo ho' nyatarasya:! /.4.6%. provides for the optional substitution of h after @hay 2a non$nasal stop2 &T.T. 161' by a sound similar to the preceding stop, e.g. va,g hita, K &optional' va,g ghita, 2nice word2, a hala,ni K &optional' a

hala,ni 2si) ploughs2 tad hi K &optional' tad dhi 2that surely2. <ere the sa,var ya principle says that h should be replaced by a velar after a velar, by a retrofle) after a retrofle) and so on. The a,ntaratamya principle decides which one among the sounds of a series should be selected. !ince h has the phonetic features ghoavat 2voiced2 and maha,pra,a 2aspirated2, that sound is chosen which also has these features so that the final choice is limited uni uely to the voiced aspirate &gh, @h, etc.' of each of the series. 1.8.11. The presentation of the su,tras in a particular order in certain sections of the Aa:dhya:yi: is one of the features of that work that have won the appreciation of linguists. The correct forms are derived only when one keeps in mind the implications of such an ordering in applying the rules. The sections in which the su,tras are purposely arranged in the particular order in which they occur are &i' the ekasa@Ia,dhika,ra &>/.1.8.', and &ii' the Tripa,di, &>/.".0., T.T. 118'. The ekasa@Ia,dhika,ra is controlled by the metarule, vipratiedhe para! ka:rya! 1.4.%. so that, in cases of conflict, the technical term assigned by a later rule supercedes the one assigned by an earlier one &for further discussion and e)amples, see >8.06.'. <ere it may be noted in passing that ordering is significant also in the pratya,ha,ra su,tras &(hapter 0', and in listing the sup &T.T. /-1' and the ti suffi)es &T.T. 110', nominal stems in the Gaapa,ha and roots in the Dha,tupa,ha. The ordering followed in arranging the 1- pratya,ha,ra su,tras and in listing the sounds in each one of them cannot be changed since any such change will lead to wrong interpretation of the terms coined on the basis of them. The order of items even in the Ga apa,ha and the Dha,tupa,ha in some instances is very crucial. :or e)ample, certain items within the sarvana,maga a &T.T. /08' are referred to as tyada,di 2items beginning with tyad$2 &see tyada:di:na:! a &.%.1"%, etc.'. it is clear that the items that come under the latter group should be kept apart from those that do not by strictly retaining the order. !imilarly in the Dha,tupa,ha also, order is important not only in the arrangement of roots in the ten classes but also in cases involving smaller groups like 2the four roots beginning with v t$2 referred to in the su,tras &see na vd'hya catur'hya &.%.5#.'. 1.8.1/. 7s stipulated by pu:rvatra:siddha! /.%.1., a su,tra of the Tripa,di, section is not ready &asiddham' for the operation of a su,tra of the first seven 7dhya,yas or that of the first 5a,da of the eighth 7dhya,ya &sapa,dasapta,dhya,yi,m prati tripa,dy asiddha,'. similarly, even in the Tripa,di, section, a later su,tra is not ready for the operation of an earlier one &tripa,dya,m api pu,rvam prati param a,stram asiddham'. 3hat this means is that in a derivation, if a su,tra of the Tripa,di, is applied, we should not apply on the product of the former any su,tra that is earlier to it even if the conditions for the application of the latter are present. This is a classic case of 2rule ordering2 which occupies an important place in modern linguistics &see Ciparsky 1#"/,++$1/6'. The asiddha device is employed by 5a,ini in three places. the other two su,tras in which the term is used are atvatukor asiddha 6.1./6 &>1.8.10.' and asiddhavad atra:'ha:t 6.4.%%. &>1.8.1-.'. 7s e)plained by Ca,tya,yana &atvatukor asiddhavacana! a:dealakaa,pratiedha:rtha! utsargalakaa'ha:va:rtha! ca, va,rttika 1 under 6.1. /6, Maha:bha:ya FCielhorn III,85G', the asiddha device is useful in two ways, it &i' prevents a substitute from causing a change &a,dealakaa$pratiedha,rtham', and &ii' allows a change caused by the underlying form even when the latter had undergone substitution by something else &utsargalak aabha,va,rtham'. The asiddha restriction imposed only on the su,tras of some parts of the grammar indicates that in general there are no order restrictions on the operation of the su,tras in the grammar. In other words, we can say that the general principle sarvatra siddham &not e)plicitly stated but presupposed by pu:rvatra:siddha! /.%.1. and the other two su,tras which use the asiddha principle' holds good in

the entire grammar barring the places where such a restriction is imposed &see Ciparsky 1#"/,+#'. This means that, in general, one su,tra can operate on the product of another one in a derivation if the conditioning for the former is satisfied irrespective of the relative positioning of them in the grammar. %)amples for the asiddha principle, &i' asmai 2for him2 N uddhara 2liftM2 K asma,y uddhara &ai K a,y by eco' yava:ya:va 6.1.&/ F7pp.I,8G' K asma, uddhara &y K O optionally by lopa a:kalyasya /.$.1# F>1./.+.&viii' &c'G'. 1n this a:d gua 6.1./& &7pp.I,+' could have operated &resulting in a, N u K o ' but this is prevented since /.$.1# that causes the deletion of y is a Tripa,di, su,tra and hence is asiddha &considered not to have operated' with respect to 6.1./& which is a non$Tripa,di, su,tra. In other words, because of pu:rvatra:siddha! /.%.1., the deleted y between a, and u is considered to be present for the operation of 6.1./&, which, as a result, cannot operate. &ii' asau 2he2 N a,ditya 2sun2 K asa,v a,ditya &au K a,v by eco' yava:ya:va 6.1.&/' K asa, a,ditya &v K O optionally by lopa a:kalyasya /.$.1# F>1./.+.&viii' &c'G'. 1n the product of this Tripa,di, su,tra, aka savare di:rgha 6.1.1"1 &F7pp.I,16G, which is a non$Tripa,di, su,tra'* could have operated &resulting in a, N a, K a,' but this is prevented by /.%.1. as e)plained under e)ample &i'. &iii' ra,@an$bhis 2by the kings2 K ra,@a$bhis &n K O by nalopa pra:tipadika:ntasya /.%.&. F7pp.I,/"G' K ra,@a$bhi. !ince the stem after the deletion of n ends in a, ato 'hisa ais &.1.#. &>#.+.1.' could have operated on this &resulting in $bhis K $ais' but this does not happen since /.%.&. that deletes the n is a Tripa,di, su,tra and &.1.#., which is a non$Tripa,di, su,tra, cannot operate on the product of the former. &iv' suparvan$au 2two gods2 K suparva$au &n K by akupva:nu!vyava:ye' pi /.4.%. F7pp. I,/#G' K suparva,$au &a K a, by sarvana:!astha:ne ca:sa'uddhau 6.4./. F>#.0.1.G'. 4ut for /.%.1.* 6.4./. could not have operated in this derivation since according to the latter the vowel that precedes a stem final n &but not ' is lengthened and n had already been converted to by /.4.%. /.%.1. ensures that the change n K is ignored for the operation of 6.4./. The last e)ample is an instance for utsarga$lakaabha,va,rtham and the earlier three for a,dealakaapratiedha,rtham &note that O is the a,dea for yJvJn in the above e)amples'. 1.8.10. atvatukor asiddha 6.1./6 2the single substitute &taught in the section 6.1./4. $ 6.1.111. F>/.8.5.G' is not ready &asiddha' for the change s K and for the addition of the augment t&uk'2 uses the asiddha device to prevent the change of s K on the one hand and to ensure the addition of the t&uk' augment on the other. 7n e)ample for the non$operation of s K , In ko2 siIcat &here as in such other cases 2 has no phonetic value' derived from kas 2who2 N asiIcat 2sprinkled2 & KK ko N asiIcat K ko2 siIcat', o N a is replaced by the single substitute o by ea pada:nta:d ati 6.1.1"# 2the first sound is the single substitute for word$final eJo N a &short'2. ;ow, since o comes under the i &T.T. -8' group, there is a possibility of s to change to by a:deapratyayayo /.$.5#. 2the s which is a substitute or belongs to a suffi) is substituted by after i &T.T. -8' or a sound of the k series2.)v 4ut this is prevented by 6.1./6* which makes the single substitute, o, non$e)istent. /.$.5#. cannot apply since the se uence here has to be taken as oas and not as os. 7n e)ample for the addition of the t&uk' augment, In adhi,tya 2having studied2 &P adhi N i$&l'ya&p' Fi$ 2to go2G', t&uk' cannot be added to i$ since i N i K i, by aka savare di:rgha 6.1.1"1. &7pp.I,16' and the su,tra that assigns t&uk' stipulates that the vowel should be short &hrasvasya piti kti tuk 6.1.&1. F>1.8.5.G'. 4ut, by virtue of 6.1./6.* the single substitute i, is considered not to have come into e)istence and conse uently the way is cleared for the addition of the t&uk' augment.

1.8.1-. asiddhavad atra:'ha:t 6.4.%% also stipulates that the asiddha principle holds good between two su,tras of the section called 7,bhi,yaprakaraa &i.e. 6.4.%%. $ 6.4.1&5. F>/.8.//G'. %)amples, &i' e$dhi 24eM2 is the imperative /nd sg. &see >16.8.' of as$ 2to be2. in this, as$hi K ae$hi &s K e by ghvasor ed dha:v a'hya:salopa ca 6.4.11#. 2the final sound of the ghu &T.T. #1' roots and of as$ is substituted by e before /nd sg. $hi and the reduplicated syllable is deleted2' K e$hi &a K O by snasor allopa 6.4.111 2the a of the class sign $& 'na&m'$ and of as$ is deleted before a sa,rvadha,tuka &T.T. /0#' k$it or $it suffi)2' K e$dhi &$hi K $dhi by hu(hal'hyo her dhi 6.4.1"1 2$ hi K $dhi after hu$ Ato sacrificeA and a root ending in @hal &T.T. 160'2'. 4ecause of 6.4.%%, the e substitute &caused by 6.4.11#' is non$e)istent for the operation of 6.4.1"1 and the root is considered to be still ending in s, which is a @hal sound. &ii' a,$dhi 2*uleM2 is the imperative /nd sg. form of a,s$ 2to rule2. a,s$hi K a,$hi &a,s$ K a,$ by a: hau 6.4.$5' K a,$dhi &$hi K $dhi by hu(hal'hyo her dhi 6.4.1"10 see e)ample &i'.. 4ecause of 6.4.%%, the s that disappeared is considered to be present for the substitution in the suffi) which depends upon a preceding @hal consonant. &iii' @a$hi 2CillM2 is the imperative /nd sg. form of han$ 2to kill2. han$hi K @a$hi &han$ K @a$ by hanter (a 6.4.$6 2han$ K @a$ before $hi2'. In this form, ato he 6.4.1"5 2$hi K O after a stem ending in a short a2 could operate since $hi occurs after a short a but because of 6.4.%% the substitution han$ K @a$ caused by 6.4.$6 is considered not to have taken place with respect to 6.4.1"5 and, as a conse uence, $hi is not deleted. The change effected by one su,tra of this section is considered to be non$e)istent only with respect to another su,tra of this section but not with respect to a su,tra that is outside this section. :or e)ample, the nasal of raI@$ 2to be dyed2 is deleted by a su,tra of this section, i.e. gha)i ca 6.4.%&, and this holds good for the following su,tra which does not belong to this section, ata upadha:ya: &.%.116 F7pp.I,1#G according to which the penultimate &short' a of the stem undergoes v ddhi before a I$it or $it suffi). 7s a result, the penultimate a changes to a, in ra,g$a 2dye2 &with $&gh'a&I' '. If I were considered to be present, a in the above root will not be the penultimate and, as such, could not have undergone the change prescribed by &.%.116.

i (f. 2Indo$%uropean comparative grammar had &and has' at its service only one complete description of a language, the grammar of 5a,ini. :or all other Indo$%uropean languages it had only the traditional grammars of Greek and Latin, wofully incomplete and unsystematic.2 &4loomfield 1#/#,/+6'. 29 this grammar describes the entire !anskrit language in all the details of its structure, with a completeness which has never been e ualled elsewhere. It is at once the shortest and the fullest grammar in the world.2 &?acdonell 1#+1,)i' 23e shall see below that the people of India, aided by the great clearness of their language, had carried very far the analysis of the component parts of a word which convey meaning, and the investigation of word$formation. 4ut Greek and Latin were far from the transparency of !anskrit, and the Greeks and *omans did not attain anything like the Indian mastery of word$ analysis.2 &5edersen 1#8+,0' 2The Indian grammatical tradition is not only independent of the Greco$*oman but also earlier, more diverse in its manifestations and in some respects superior in its achievements.2 &Lyons 1#8",1#' 25a,ini2s grammar of !anskrit has fre uently been described, from the point of view of its e)haustiveness &within the limits which it sets itself, i.e. mainly with regard to the structure of words', its internal consistency and its economy of statement, as far superior to any grammar of any language yet written.2 &Lyons 1#8",/6.' 7lso see note 16. ii !cholars like 3hitney and 5edersen maintained that !ir 3illiam Bones was aided in his discovery of genetic relationship between !anskrit on the one hand and Greek, Latin &and some others' on the other by the 2peculiar structural clarity2 of the !anskrit language &*ocher 1#+5,-'. <owever, %meneau &1#55,1-#' has rightly pointed out that !anskrit is inherently no more transparent in structure than Greek or Latin but 5a, ini2s analysis of that language made it appear to be so. ?aster2s doubts &1#58,1"+' whether !ir 3illiam Bones had access to 5a, ini2s grammar are unfounded since, as later clarified by %meneau &1#+1,#86', even if Bones had atually studied only a later grammar &the Siddha:ntakaumudi: &>1.-.+' or the Mugdhabodha' it still is a representative of the 5a,inian system. The latter work by 4opadeva &10th century 7.D.' is a simplified version of the Aa:dhya:yi: with changes in the wording of the su,tras &see 4elvalkar 1#+8,"+ff.'. iii (f. 2:irst of all, the importance of the great grammarian is considerable, due to the synchronic view which he gives of the language of his time and due to the method which he uses in the description of this language. This method is in certain respects archaic, but often surprisingly modern, and always precise and rigorous. :urthermore, the techni ue of 5a, ini &and his school' is of interest for those who are working toward an a)iomati=ation of grammar. in this respect 5a, ini2s grammar represents a model that has no e ual in anti uity.2 &*enou 1#8#,-"1'.

iv In his 1#5+ article &p./80', Thieme made another controversial statement uestioning 4loomfield2s view that the Aa:dhya:yi: is a 2perfect description of a language2. !trangely enough, he went to the e)tent of claiming that it should not be called a 2grammar2 at all, 23hen speaking of 5a, ini2s work 9 as a 2grammar2, we are actually committing a little inaccuracy9 it is imperfect as a full description of speech usage. 5a, ini2s work is e)clusively concerned with defining & lakaya ' the procedures of regular word$formation &saska:ra' $$$ anything else does not fall within its scope. it is dealt with either sweepingly, as for e)ample the unanalysable nominal stems &0.0.1. u a,dayo bahulam', or not at all. There are, for e)ample, no rules concerning word order, no rules concerning gender unless it is a function of a particular suffi). There is no phonetic instruction.2 7 little reflection would reveal that the reasons adduced by Thieme in support of his view are e)tremely weak. ;o grammar can be e)pected to contain a list of unanalysable nominal stems with specification of their gender. it is only the le)icon that should contain them. 5a,ini has listed some of such stems in the Ga apa,ha when they show some grammatical peculiarity or other. *egarding phonetics, 5a, ini used the phonetic theory in his grammar &as in tulya:syaprayatna! savara! 1.1.#. FT.T. /0"G' e)pecting his students to know the details from the ikas and 5ra,tia,khyas. it is clear form this that 5a,ini considered phonetics as a discipline that is allied to but still different from grammar. <e even stated the positional variants &allophones' of the basic sounds &phonemes'. !ince there is no fi)ed order in !anskrit for the occurrence of the sub@ect, verb and the ob@ect nor for that of the ad@ective and the noun in a sentence, 5a, ini cannot be blamed for not giving any rules in this regard. <owever, it is not true to say that there are no rules at all concerning word order as can be understood from the following e)amples, &i' The position of a preverb &T.T. 5"' and a gati word &T.T. "5' with respect to a verb in a sentence is mentioned &te pra:g dha:to 1.4./" 2they occur before a root2, chandasi pare' pi 1.4./1 2in the Heda, they occur even after it2, vyavahita: ca 1.4./% 2in the Heda, they occur also separated from it2'. &ii' In k) ca:nuprayu(yate lii $.1.4" the perfect forms of k&I'$ 2to do2 are declared to be added a!ter &anuprayu@yate. but not be!ore' the root of the main verb with the suffi) $a,m &see >16./.0.'. &iii' kaa:diu yatha:vidhy anuprayoga $.4.46 stipulates that in the case of a primary derivative form with the $ &'am&ul' suffi) derived by the su,tras $.4.$4 to $.4.45 a verb from the same root to which the above suffi) is added !ollo"s it,

e.g. samu,laka,$am kaati 2plucks out withroots2. &iv' yatha:vidhy anuprayoga pu:rvas!in $.4.4 ensures that im e)amples like luni,hi luni,hi N ity eva N ayam luna,ti 2cuts vigorously &lit., saying 2cutM2, 2cutM2'2 the verb from the same root from which the repeated imperative is formed !ollo"s it. It is unfortunate that this ill$considered criticism of Thieme is highlighted by *obins &1#8+,1--', 29 it is not what would ordinarily be called a complete grammar of the !anskrit language.2. v The following statement of Ciparsky &1#"/,1' is representative of this view, 25a, ini2s grammar of !anskrit has been appreciated for its morphology and phonology ever since it became known in the 3est, but it was not fully reali=ed until recently that it also includes a detailed and interesting treatment of synta).2 vi The #ya:sa commentary under na pada:nta- 1.1.5/ &>8.15', kim punar idam ra,@aa,sanam $$$ padasaska,ra,yaiva abda,nua,sanam kartavyam itiQ atha a,straka,rasyaiva N ayam abhipra,ya iti cet, na. a,straka,rea hi Ayumady upapade sama,na,dhikarae stha,niny api madhyamaA iti yumada,dyupapade madhyama,di$puruavidha,na,d va,kyasaska,raprayuktam api a,stram etad iti su,citam. 2Is it a royal command that a grammar must be written only to account for wordsQ If it is said that this is the opinion of the author &i.e. 5a, ini' himself, it is not so. 4y assigning second person &verbal' suffi)es when the sentence contains a second person pronoun &in yu!ady upapade- 1.4.1"5 F>16.6.+G' and similarly other personal suffi)es, it was indicated by the author himself that the purpose of the grammar was also the correctness of sentences.2 vii There are only si) su,tras that deal with phonetic details in the entire work. 1ne of them, namely, !ukhana:sika:vacano' nuna:sika 1.1./ defines the term anuna,sika &T.T. 16'. 7nother one, u:ka:lo' ( (hrasvadi:rghapluta 1.%.%& defines the terms hrasva &T.T. /5/', di,rgha &T.T. 11"' and pluta &T.T. 155'. The remaining four su,tras define the three accents, uccair uda:tta 1.%.%# &T.T. 50', ni:cair anuda:tta 1.%.$" &T.T.#', sa!a:ha:ra svarita 1.%.$1 and tasya:dita -1.%.$% &T.T. /-#'. viii (f. the vttis 2e)pansions2 provided by the commentators for the su,tra, hetu!ati ca $.1.%6: &$a:ika:' preaa,dilakao hetuma,n, tasminn abhidheye dha,tor ic pratyayo bhavati 2hetumat$ means Acausatio and the likeA. when that is to be e)pressed, $&'i&c'$ suffi) occurs after the root2. &Siddha:ntakaumudi:' prayo@akavya,pa,re preaa,dau va:cye dha,tor ic sya,t. i) The peculiarities of a stem ending in $a,&p' in inflection are the following, & i' deletion of 1st sg. $s&u', &ii' 1st R /nd du. $au K $ &'i,, &iii' a, K e before 0rd sg. $&'a,, 8th R +th du. $os and vocative sg. $s&u', &iv' addition of augment ya,&' to a $$it suffi) &if, however, the stem is a pronoun, the augment is sya,& ' ', &v' addition of augment n&u ' to 8th pl. $a,m &but the augment is s&u ' if the stem is a pronoun', and & vi' +th sg. $&'i K $a,m & &v' and &vi' are found also after other feminine stems called nadi, &T.T. 1/-' and some individual stems'. ) The distinction pronominal versus non$pronominal in the case of stems ending in &a short' a is re uired to account for the variations in five case suffi)es as shown below &a stem ending in a is either masculine or neuter. the difference counts only in the case of 1st and /nd cases'. A!ter a pronominal stem A!ter a non pronominal stem 1st pl. $&@'as -th sg. $&'e 5th sg. $&'as 8th pl. $a,m +th sg. $&'i K $&'i, &a N i, K e' K $smai K $sma,t augment s&u' &plus a K e' K $smin &a N as K' a,s K $ya K $a,t augment n&u' &plus a K a,' &a N i K' e

)i 7n e)ample for /.$.1$ is li, ha$ 2that which is licked2 &from the root lih$ 2to lick2' derived as follows, lih$&k'ta$ K li h$ta$ &h K h by ho ha /.%.$1 2h K h before @hal &T.T. 160' and at the end of a word2' K li

h$dha$ &t K dh by (haas tathor dho' dha /.%.4" 2tJth K dh after @ha &T.T. 165' but this will not operate in the case of dha,$ 2to bear2 2' K li h$ ha$ &dh K h by una: u /.4.41 F7pp. I,0"G' K li$ ha$ & h K O by /.$.1$' K li,$ ha$ &i K i, by hralope pu:rvasya di:rgho' a 6.$.111 2the a &T.T. 8' vowel that precedes the deleted hJr is lengthened2'. 7n e)ample for /.$.14 is hari, ra,@ate 2<ari thrives2 from hari$s&u' N ra,@ate K hari$r&u' ra,@ate &s K r&u' by sasa(uo ru /.%.66. F7pp. I,01G' K hari ra,@ate &r K O by /.$.14' K hari, ra,@ate &i K i, by hralope - 6.$.111'. )ii In a few cases like the following &both of which are from the Hedic dialect', 5a, ini treated the e)tra phonological material that is found optionally at the end of a suffi) as augment. &i' as&uk' added optionally to 1st pl. $&@'as after a stem ending in aJa, &e.g. deva$&@'asas&uk' KK deva,sa 2gods2 besides deva, , a:( (aser asuk &.1.5"'. &ii' ya&k' added optionally to the gerundial suffi) $ &k'tva, as in da$tva,ya 2having given2 besides da$tva, &ktvo yak &.1.4&'. )iii This has been noted by 5ataI@ali & Maha:bha:ya, 5aspaa,hnika FCielhorn I,8G', kiIcit sa,ma,nyavieaval lakaam pravartyam yena,lpena yatnena mahato mahata abdaugha,n pratipadyeran.. kim punas tatQ utsarga,pava,dau. ka cid utsarga kartavya kacid apava,da.. katham@a,ti,yaka punar utsarga kartavya katham$@a,ti,yako2 pava,daQ sa,ma,nyenotsarga kartavya. tad yatha,. karmay a &0./.1.'. tasya vieea,pava,da. tad yatha,. a,to2 nupasarge ka &0./.0.'. 2The grammar that has general and specific rules should be composed so that large numbers of words can be obtained with a minimum of effort. 3hat is itQ General rules and e)ceptions. 1ne should be framed as the general rule and another as an e)ception. Then, what type should be made the general one and what type the e)ceptionQ The general rule must be made to contain no restriction. It is as follows, kar!ay a. The e)ception rule must be made with a restriction. It is as follows, a:to' nupasarge ka2 &for meaning and e)amples, see >1.8.#.'. )iv !ee also the Maha:bha:ya &under kartari kt $.4.6& FCielhorn II,1+"G', ea eva nya,yo yad uta,pava,dair utsarga, ba,dhyeran 2it alone is the convention that general rules are superceded by specific rules2. )v ;ote that the root in asiIcat is given as ic$ in the Dha,tupa, ha. K s by dha:tva:de a sa 6.1.64. 2the at the beginning of a root is substituted by s2. The s, therefore, is a substitute.

Das könnte Ihnen auch gefallen