CC BY
32
Artemij Keidan Compositional history of Panini's karaka theory
Резюме. В статье обосновывается гипотеза о текстологической неоднородности древнеиндийской грамматики Панини. В частности, обращается внимание на разительные несоответствия в грамматическом описании и терминологии, которые наблюдаются в сутрах, посвященных вопросу семантических ролей и падежей (прежде всего в секции karaka, а также в других отделах). Разница в теоретическом подходе настолько сильная, что кажется логичным заключить, что имеющийся сегодня текст — результат наслоения теорий нескольких авторов.
Ключевые слова: древнеиндийская грамматическая традиция, Панини, текстология Аштадхьяи, семантические роли и падежи.
1. Introductory remarks
The question of philology, textual analysis and compositional history of the Astadhyayi has provoked long-standing debates among Paninian scholars in the last decades. For a long time the opinion prevailed that Panini's grammar has been perfectly preserved in its original shape, while any conjecture of textual stratification or interpolation hypothesis was rejected in advance. Though a critical analysis of the extant manuscripts is still lacking, many authoritative scholars have claimed that the received text is virtually identical with the original one, and that the manuscripts do not exhibit relevant variations (e.g. Renou 1969: 484). This uncritical philological approach falls perfectly within the "myths of Paninian studies" recently discussed in an insightful paper by J. Houben (2003). However, already from a general point of view, the claim for textual integrity cannot be accepted: no text of Antiquity—not even in India—has preserved exactly its original wording, and Panini's grammar would hardly be an exception. Moreover, from very ancient times many instances of textual corruption in the official text of the Astadhyayi have been observed by the indigenous grammarians (see Iyer 1983 for a collection of such cases), though this fact was scarcely
considered significant for the preservation of the grammar. The traditional scholars, indeed, have always been confident of their capacity to restore the original wording of the Astadhyayi on the ground only of the knowledge of what was described in it—i.e., the Sanskrit language—and how the description would be shaped in order to describe it perfectly. Since every change in the text of the grammar would lead to mistakes in the description of Sanskrit, it seemed to them a simple task to emend the possible errors and to restore the right description.
At the beginning of modern Paninian studies only very few philologists (primarily F. Kielhorn) dealt with textual problems, mainly, but not only, from a methodological point of view; however, the common traditionalist view remained. Unfortunately, the most reliable edition of the Astadhyayi until now remains Bohtlingk's translated edition, and little, if any, textual criticism has been made on the manuscripts. Therefore, scholars can mostly postulate secondary hypotheses regarding the compositional history of the Astadhyayi based on some extra-textual ground (cf. Aklujkar 1983: 1). More recently, in such papers as Bahulikar (1973), Birwe (1955, 1958), Butzen-berger (1995), Joshi, Roodbergen (1983), Iyer (1983) and Keidan (2007) the dogma of textual integrity has been questioned, and several interpolations have been hypothesized; see also the criticism by Cardona (1999: § 2.1.2.3).
In the present paper I wish to apply the principles elaborated by preceding scholars (especially by Joshi, Rood-bergen 1983). I shall attempt to reconstruct the compositional history of the syntactic sections of the Astadhyayi, corresponding to the sutras describing the formation of the simple sentence starting from basic semantic categories and the morphological inflectional classes expressing them, namely the sutras 1.4.23 to 1.4.55 dealing with the definitions of karaka categories, and the sutras 2.3.1 to 2.3.73 dealing with vibhaktis 'case-from categories'. My main claim is that these sections of the grammar must be considered heavily interpolated, since several parts of them are so strongly inconsistent with one another from a theoretical point of view that they must be considered mutually exclusive.
2. On methodology
As a general principle, I have considered (following Bahulikar 1973: 80) the text of the Astadhyayi per se as the only relevant data, getting rid of all what we know from later discussions and comments made by ancient commentators and grammarians (which can serve, at most, as indirect evidence). Furthermore, I assume that Joshi and Roodbergen's (1983) criteria for the individuation of textual interpolations in the Astadhyayi are correct in essence, at least as a general guideline. These criteria are listed below in a lightly modified and expanded form. Thus, a rule may be considered possibly interpolated if:
(a) it presents theoretical inconsistencies with other parts of the grammar;
(b) it interrupts the logical order of the sutras, inserting unrelated material in an otherwise homogenous thematic section;
(c) it shows an inconsistent use of metalinguistic markers and terminology;
(d) it presents discrepancies in phrasing, style and mutual disposition of the sutras;
(e) it is formally or theoretically paralleled by other ancient Indian texts and/or schools, e.g. parallelism between the mimamsaka tradition and certain sutras of the karaka section suggested by Butzenberger (1995);
(f) it prescribes unattested Sanskrit forms, e.g. the sutras 2.3.57-60 dealing with case government rules that are unattested in Classical Sanskrit and found only in Prakrits, see Speijer (1886: 90);
(g) it mentions unattested lexical meanings—and ones even totally unintelligible without the purport of later commentaries—as a condition for the application of a rule;
(h) it is a paribhasa 'metarule of the grammar': such kinds of rules are mostly provided by later commentators, and only some of them entered the received text of the Astadhyayi;
(i) it is considered, by the ancient scholarship, as a quotation from a purvacarya 'preceding author' or as an emendable rule;
(j) it lacks a direct or indirect mention in Patanjali's Mahabhasya, being a later interpolation still unknown to this commentator when he composed his glosses; however, this argument is considered insufficient or totally inconsistent by some more traditional scholars (see Cardona 1976: 158).
For my concern here, the most important criterion of those listed above is the first one. My claim is that the theoretical contradictions within the sections dealing with karakas vibhaktis are so strong that we are forced to hypothesize a manifold authorship for them. This approach is not unprecedented. The most known example of the application of this criterion is Joshi and Roodbergen's (1983) hypothesis on the interpolated nature of the taddhita 'secondary derivation' and samasa 'compounding' sections of the Astadhyayi, both of which would have been added to the main text in a later period (at least later than the kit 'primary derivation' section, which is claimed to pertain to the original version of the grammar). The major evidence for such a hypothesis is a theoretical one, namely a difference in the way the semantic conditions for the application of morphological rules are stated (see below for a detailed analysis).
3. The functioning of the karaka-device
Generally speaking, karaka categories roughly correspond to what today's linguists call semantic roles or, to some extent, even to the macro-roles (for these notions see Van Valin and Lapolla 1997: chap. 4). The karakas are six in number and have essentially semantic definitions, though the most important karaka, i.e. the so-called kartr 'agent', could be considered, from a certain point of view, more similar to a syntactic entity, namely the subject, since it is defined as svatantra 'the independent one' (for a detailed discussion of possible modern equivalents for Paninian syntactic categories in the light of modern Functionalism see Keidan forthcoming b). Besides the
kartr, the other karaka categories, listed in the traditional order, are: apadana 'source', sampradana 'receiver', karana 'instrument', adhikarana 'location' and karman 'patient'. Each karaka, it is taught, has many possible morphological expressions. Interestingly, the finite verbal terminations are considered one such possibility, but we are more interested here in the other possible expression, namely the nominal case-form endings called vibhakti. The latter are similar to our grammatical cases, except for the fact that their nomenclature is purely formal and avoids any semantic reference. The vibhakti terms are based on their serial number within the standard sequence of cases: prathama 'first', dvitiya 'second', trtiya 'third', caturthi 'fourth', pancami 'fifth', sast! 'sixth' and saptami 'seventh', corresponding, respectively, to nominative, accusative, instrumental, dative, ablative, genitive and locative. Each karaka has a standard or "canonical" vibhakti realization: thus, the kartr is taught to be primarily expressed by the instrumental case, the karman by the accusative, etc. Notice, however, that, besides a certain privileged status of karakas' canonical vibhakti realizations, there is no one-to-one relationship between karaka roles and the vibhaktis: each karaka may be expressed by more than one case-form; similarly, each vibhakti expresses several semantic functions, not even limited to karakas.
Thus, Panini's most brilliant achievement in the field of syntax amounts exactly to this treatment of semantic categories as separate entities with respect to their morphological realization. Such a clear-cut differentiation of the two sides of the language—the plane of forms and the plane of functions—remains unparalleled in any other ancient grammatical tradition, and has been fully attained by western linguistics only in the last few decades, particularly, as far as the sentence arguments realization is concerned, after the formulation of the Deep Case theory by Ch. Fillmore (1968).
However, if we take into account not only the definitions of the karaka categories, but also the other rules of this section, where the primary definitions are emended and enlarged, the situation changes: the separation between the forms and the functions becomes far less accurate and rigid. The general tendency in this respect was to identify the karakas with their canonical vibhakti realization and, eventually, to replace
completely the former with the latter. This is what happened, according to Joshi and Roodbergen's (1983) analysis, to the taddhita section where the vibhakti terms are used as if they were equivalent to the karakas 'semantic roles', in order to indicate the semantic constraints on compound formation. In the taddhita section even the vibhakti terms seem to be disclaimed: instead, inflected pronouns are used in order to denote their own case-form category, and therefore the related semantic role, as a constraint to the application of certain secondary nominal suffixes.
It must be stressed that the abovementioned approaches to one and the same linguistic material—namely, the morphological coding of sentence arguments—are so antithetical as to induce the hypothesis of a possible difference in the authorship. The separation of the forms and the functions represents a revolutionary paradigm change in linguistics that should not be underestimated. A sudden abdication of this principle must be considered strong evidence for textual interpolation. Thus, my claim is that the present text of the Astadhyayi is to be analyzed in different theoretical strata put together only in a later period lying between Panini's time and the earliest extant ancient commentary, namely the Mahabhasya of Patanjali. Indeed, it seems impossible for one and the same author to have adopted so many contradictory positions inside one and the same grammatical treatise.
4. Theoretical layers in the syntactic rules of the Astadhyayi
What follows is a mere listing of the relevant theoretical layers. The listing itself does not imply very much regarding the relative chronology of the corresponding textual strata. However, the succession of the list entries is obviously not random, and could be taken as a preliminary approximation of the historical development of the Astadhyayi.
The first and most "primitive" layer, which can be called the "proto-karaka layer", corresponds to the karaka terms in their "etymological" or "everyday" meaning. Already the ancient tradition considers them borrowings that Panini inherited from some purvacaryas 'preceding authors' of whom
we know little. Genuine Paninian terminology, indeed, is usually made of meaningless marks and abbreviations. The etymological meanings of karaka terms vaguely recall the semantic definition of the relating karakas. Hypothetically, in a pre-Paninian period, the karaka terms were conceived as self-explanatory (as similarly the European case nomenclature), and were probably used for both semantic and morphological categories, i.e. semantic roles and grammatical cases, with no clear distinction yet between the two planes of the language.
The second layer corresponds to the standard karaka/vibhakti framework. Panini takes the pre-existing karaka terms, gets rid of their "etymological" interpretation, gives them some abstract, clear-cut, but still semantic definitions, and puts them into correlation with vibhaktis. The latter are given their "numbering" names, and put into a special section of the grammar where all their uses are listed. Thus, the opposition between semantics and morphology is, for the first time in history, clearly stated and observed (notice that such level of sophistication has been reached by western linguists only in the second half of the 20 th century).
Let us take as an example the definition of kartr 'agent' (or 'actor', if we accept the view of karakas as macro-roles), which is qualified as svatantra 'self-standing, independent'. This definition has generated long-standing debates among traditional Indian grammarians, but could simply be an attempt to provide a more general definition of this semantic role, going beyond the literal meaning of 'doer'. Only with such a broadened definition could some peripheral semantic roles, such as experiencer (i.e., the one who experiences a feeling), be subsumed into the greater category of the macrorole. The kartr is expressed by either krt 'primary suffixes', such as the suffix -aka, or finite verb endings, such as the 3rd person singular active -ti, or some vibhakti 'nominal case-form', primarily the trtiya 'third one', i.e. the instrumental case, but sometimes also by the sasthi 'sixth one', i.e. the genitive case.
Soon after the establishment of the karaka/vibhakti device the definitions of karaka started to be emended and enlarged, partly because the original general and abstract sense of the primary definitions was not understood anymore. Thus, the degradation of the theory of karakas began. Initially this
process did not result in a change of the theoretical approach, since the karakas were still conceived as semantic categories, not yet fused with morphological ones. However, already Patanjali noticed that such rules were useless, since the additional semantic definitions could be metaphorically deduced from the primary ones. For example, at some point, the ancient commentators disregarded the real sense of Panini's definition of karman as ipsitatama 'the most desired' (which, in its original formulation, was probably intended as something similar to what modern linguistics calls undergoer macro-role). They started to interpret this sutra in its literal sense, thus implying an excessively restrictive characterization of this macro-role. Therefore two sutras were added (1.4.5051), in order to provide a better definition thereof (it was stated, e.g., that the undergoer happens also to be 'undesired'). It has been convincingly suggested by Butzenberger (1995) that this interpolation was made under the influence of the mimamsaka tradition and its terminology.
The third theoretical layer surfaces in those rules where the karaka categories are identified with corresponding canonical vibhakti realizations. At the beginning it was only an implicit tendency, and the karaka terms were still in use, though the distinction between semantics and morphology was being blurred. At this stage, given a certain primary semantic definition of a karaka, some rules were added where the semantic definition of a karaka was changed and/or the karaka category assignments underwent a modification depending on a certain case-form governed by some verb that had to be accounted for. On the contrary, within the standard karaka/vibhakti framework, all such cases would have been treated in the vibhakti section, simply as alternative expressions of a given karaka.
Let us analyze the sutra 1.4.43 divah karma ca as an example. It describes the fact that the verb div 'to gamble' can optionally code with the accusative—instead of instrumental—the argument corresponding to 'what is gambled with'. In the view of the standard karaka/vibhakti framework, it would be simply a case of alternative coding of a karaka, to be taught in the vibhakti section. According to the commentators, instead, the sutra means that, with the verb div, what seems a karana
may also be optionally classified as karman. The implicit reasoning here seems to have been the following. a) The instrument of an action is primarily classified as karana by 1.4.42 sadhakataman karanam; b) the verb div exhibits an argument matching the semantics of the karana category; c) the canonical vibhakti realization of the karana is the instrumental case by 2.3.18 kartikaranayos trtiya; d) the karana argument of the verb in question, however, may also be coded by the accusative; e) the accusative case is taught to be the canonical realization of another karaka category, namely karman by 2.3.2 karmani dvitiya; f) ergo, the semantics of an instrumental argument must be classified, limited to the verb div, as belonging to the category of karman instead of karana. In other words, according to this new approach, whatever stands in the accusative case-form must be automatically considered a karman. Therefore, the privileged one-to-one relationship between a karaka and its canonical realization brought about a complete fusion of these two planes of the language.
It may be noted that many of the sutras pertaining to this theoretical layer also exhibit some of the features listed at § 2 above that substantiate their interpolated origin. Thus, exactly half of them are not mentioned in Patanjali's Mahabhasya. The sutra 1.4.38 uses a non-standard terminology: only here the term upasrsta is used to denote a 'prefixed verb' (Sharma 1990: 249). Sutras 1.4.34, 1.4.39 and 1.4.40 rely completely on otherwise unattested verbal meanings.
The fourth layer is the one to be found in the sections devoted to samasa 'compounding', and in some other rules, where the karaka/vibhakti distinction seems to be completely abandoned. While karakas are not mentioned at all, the vibhakti terms are used here in order to express semantic functions totally undistinguishable from the corresponding morphological classes. Effectively, every karaka is indicated by its canonical vibhakti realization. The vibhakti terms are used as such, or in composition with the word arthe 'in the meaning of'. Thus, the sutra 2.1.37 pancami bhayena teaches the formation of compounds whose first member is intended in the sense conveyed by the pancami 'fifth case' (i.e. ablative), hence the apadana 'source' semantic role (e.g. caurabhayam 'fear of thieves').
The fifth layer is the one attested in the taddhita section as well as in some of the paribhasutras 'meta-rules' of the grammar. Here, even the vibhakti terms are no longer in use. In order to refer to both semantic roles and case categories inflected pronouns are used: they indicate "iconically" their own case category, which, in its turn, refers to the corresponding semantic role (as in the preceding layer). A good example is that of the paribhasutras defining the metalinguistic case-forms. Thus, sutra 1.1.67 tasmad ity uttarasya defines the "left context ablative" (referred to as tasmad 'from that'); sutra 1.1.66 tasminn iti nirdiste purvasya defines the "right context locative" (referred to as tasmin 'in that').
Interestingly, this layer is paralleled in the western grammatical tradition. Inflected indicative or interrogative pronouns symbolizing the corresponding case-forms have been used as a descriptive and didactic means starting from Ancient Greek and Latin grammarians (see Belardi, Cipriano 1990: chap. III) up to modern times. Even today, the case-forms are sometimes taught as "answers" to some "questions", which the pupils are supposed to know by heart. Thus, B. Liebich (1886: 209, 215) glossed the karaka categories precisely in this way: «Das karman antwortet daher auf die fragen 'wen' oder 'was' und bei der bewegung auf die frage 'wohin'»; «Das sampradana antwortet daher auf die fragen 'wem' oder 'fur wen'», etc.
Regarding the post-Paninian developments, it must be noted that the idea of a radical karaka/vibhakti distinction was often partly or completely disregarded within the grammatical schools which arose outside the strictly Paninian tradition (see Butzenberger 1995: 54). For instance, in Candragomin's grammar, the vibhakti categories and terminology prevailed for both morphology and semantics (as in our fourth layer). On the other hand, in the Katantra tradition a more etymological interpretation of karaka terms was restored (similarly to our first layer). For example, the karman is defined here by the sutra 2.4.13 yat kriyate tat karma 'what is being done that is karman' (since the latter is a passive formation from the root kr 'to do'). Eventually, an intelligent synthesis of Panini's karaka/vibhakti device is to be found in the Saddaniti, an indigenous grammar of Pali. Thus, the sutra 551 defines the
patient role as follows: yam kurute yam va passati tam kammam 'what is done or seen is karman'. Here, the semantic definition of this meta-role is "demystified", resulting simply in a summation of its most frequent components: the patient and the so-called stimulus (something that is perceived by our senses).
The question of authorship may arise regarding the different theoretical layers listed above. It must be noted that their theoretical "quality" is varying: the second layer seems to be the most complex and sophisticated, while the subsequent ones represent a continuous degradation of the original theory. So where should we collocate Panini in this respect? Since we do not know much about Panini's dates (which is still an open question, notwithstanding all the hypotheses thus far proposed in this regard, see the "Introduction" chapter to Joshi, Roodbergen 1976), I feel justified in arbitrarily identifying his authorship with the abovementioned second theoretical layer, as homage to the great prestige Panini has had through many centuries in Indian grammatical thought.
S. Possible causes of the interpolations
A grammar can change mainly for two reasons: the evolution of the language that it describes, or the evolution of the theoretical ideas regarding the description itself. Both these circumstances may have appeared in the case under consideration.
Thus, on the one hand, the passage from the first to the second layer, when the karaka/vibhakti framework had been established presumably by Panini, represented a revolutionary advance in the description. The subsequent layers might be at least partly explainable by an increasing misunderstanding of Panini's original ideas (already starting, for instance, with the supplementary semantic qualifications added to the second layer in order to clarify some karaka definitions that no longer appeared intelligible).
On the other hand, the evolution of some important facts of language may also have played a role. Some grammatical phenomena that could be inexistent or excessively peripheral and therefore irrelevant in Panini's time, became more
pertinent in the subsequent periods. One such feature that strongly challenged the standard karaka/vibhakti device was the rise in Sanskrit of many subject-related phenomena, namely voice transformations and valency derivation. Now, an explicit delineation of the grammatical relations is completely lacking in the original karaka theory. However, the subsequent theoretical layers within the Astadhyayi, and even more some theories of ancient commentators, went clearly in the direction of a gradual and tacit rejection of the semantic-based karaka category in favour of a more syntacto-centric categorization similar, though not identical, to that of the grammatical relations (for a detailed account of how voice and derivation phenomena can explain grammarians' theory of the "agentness" of all karakas see Keidan forthcoming a). Furthermore, the later origin of the taddhita and samasa sections of the Astadhyayi, claimed by Joshi and Roodbergen (1983), may be substantiated if we think of the increasing importance assumed by compounding and secondary affixation in Late—i.e. post-Vedic—Sanskrit.
5. Concluding remarks
As pointed out by A. A. Zaliznjak (2007), the linguistic evidence can easily be underestimated or disregarded by nonspecialists. Indeed, the hypotheses I suggest in the present paper might not seem persuasive to many traditional indologists and specialists in Panini's grammar (as Cardona's critique against Joshi and Roodbergen's hypothesis shows). However, no linguist would deny that the distinction between semantic roles and case-form categories is such a revolutionary approach that a "backwards trip" would be virtually unimaginable for one and the same author. Therefore, I think that a theoretical stratification inside the received Astadhyayi must be considered, unless a possible critical edition of its text, based on the collation of the extant manuscripts, would disallow such a hypothesis.
Reference
Aklujkar 1983 - Aklujkar A. N. The Astadhyayi as a case in textual criticism. Part 2 // Proceedings of the International Seminar on Studies in the Astadhyayi of Panini / ed. by S. D. Joshi and S. D. Laddu. Pune: University of Poona. P. 1-10.
Bahulikar 1973 - Bahulikar S. Concerning the structure of Panini’s Astadhyayi // Indian Linguistics. Vol. 34. P. 75-99.
Belardi, Cipriano 1990 - Belardi W., Cipriano P. Casus interrogandi. Nigidio Figulo e la teoria stoica della lingua. Roma, Viterbo: Il Calamo.
Birwe 1955 - Birwe R. Interpolationen in Panini’s Astadhyayi' // Studia Indologica. Festschrift fur Willibald Kirfel zur Vollendung seines 70. Lebensjahres / ed. by O. Spies. Bonn: Selbstverlag des Orientalistiches Seminars der Universitat Bonn. P. 27-52. (Bonner Orientalistische Studien. Vol. 3).
Birwe 1958 - Birwe R. Variae lectiones in Adhyaya IV und V der Astadhyayi // Zeitschrift der Deutschen Morgenlandischen Gesellschaft. Bd. 108. S. 133-154.
Butzenberger 1995 - Butzenberger K. Panini's sutra.s on karma-karaka (Astadhyayi 1,4,49-51) and their historical development' // Berliner Indologische Studien. Bd. 8. S. 9-62.
Cardona 1976 - Cardona G. Panini: A Survey of Research. The Hague, Paris: Mouton.
Cardona 1999 - Cardona G. Recent Research in Paninian Studies. Delhi: Motilal Banarsidass.
Fillmore 1968 - Fillmore Ch. J. The Case for Case // Universals in Linguistic Theory / ed. by E. Bach and R. T. Harms. New York: Holt, Rinehart & Winston. P. 1-88.
Houben 2003 - Houben Jan E. M. Three myths in modern Paninian Studies // Asiatische Studien. Etudes Asiatiques. Vol. 57. P. 121179.
Iyer 1983 - Iyer V. S. On Variations in the Astadhyayi // Proceedings of the International Seminar on Studies in the Astadhyayi of Panini / ed. by S. D. Joshi and S. D. Laddu. Pune: University of Poona. P. 141-155.
Joshi, Roodbergen 1976 - Joshi S. D., Roodbergen J. A. F. Patanjali's Vyakarana-Mahabhasya: Anabhihitahnika (P. 2.3.1-2.3.17).
Poona: University of Poona.
Joshi, Roodbergen 1983 - Joshi S. D., Roodbergen J. A. F. The Structure of the Astadhyayi in Historical Perspective // Proceedings of the International Seminar on Studies in the Astadhyayi of Panini / ed. by S. D. Joshi and S. D. Laddu. Pune: University of Poona. P. 59-95.
Keidan 2007 - Keidan A. Panini 1.4.23: emendation proposal // Rivista di Studi Sudasiatici Vol. 2. P. 209-241.
Keidan Forthcoming a - Keidan A. Lability in Middle Indo-Aryan: theories, facts, and indirect evidence // Papers from the Workshop 'Typology of labile verbs: Focus on diachrony' (ISTAL 19) / ed. by L. Kulikov and N. Lavidas / Linguistics. Special Issue. Keidan Forthcoming b - Keidan A. Theoretical stratifications in Panini's karaka system // Papers from the panel 'The Indian traditions of language studies' (ICHoLS XI) / ed. by J.-L. Chevillard and E. Aussant.
Liebich 1886 - Liebich B. Die casuslehre der indischen grammatiker verglichen mit dem gebrauch der casus im Aitareya-Brahmana // Beitrage zur Kunde der indogermanischen Sprachen. Bd. 10. S. 205-233.
Renou 1969 - Renou L. Panini // Current Trends in Linguistics. Volume 5. Linguistics in South Asia / ed. by T. A. Sebeok. The Hague, Paris: Mouton. P. 481-498.
Sharma 1990 - Sharma R. N. The Astadhyayi of Panini. 2. English translation of Adhyaya one with Sanskrit text, transliteration, word-boundary, anuvrtti, vrtti, explanatory notes, derivational history of examples, and indices. Delhi: Munshiram Manoharlal. Speijer 1886 - Speijer J. S. Sanskrit Syntax. Leiden: Brill.
Van Valin, Lapolla 1997 - Van Valin R. Jr., Lapolla R. J. Syntax: structure, meaning and function. Cambridge: Cambridge
University Press.
Zaliznjak 2007 - Зализняк А. А. «Слово о полку игореве». Взгляд лингвиста (2е изд.). Москва: Рукописные памятники Древней Руси.
Summary
The main claim is that Panini’s Astadhyayi may be viewed as a result of several consecutive textual stratifications. Such strata are now primarily distinguishable on the theoretical basis, since there are some relevant dissimilarities in the analysis of the simple clause in different parts of the grammar. Particularly, the karake section as well as some other parts of the Astadhyayi connected with the role/case-form distinction are taken into consideration. The radical paradigm change in categorization and terminology that is observable in these sections could be explained if a change in the authorship is also assumed.
