Tiṅanta-Rūpasiddhi Tool
K. Kabi Khanganba
School of Sanskrit and Indic Studies,
JNU, New Delhi 110067, India
Abstract
The paper presents to develop a rule-based tool which generates a Rūpasiddhi table of a particular Sanskrit verb paradigm. A traditional Rūpasiddhi table is the output of a verbal paradigm generation process which chronologically displays each and every rule along with their corresponding outputs. The system of Aṣṭādhyāyī processes a Dhātu or a verbal root through a number of rules. A Sūtra or a rule takes an input and produces an output which again becomes an input of another rule. The set of rules in Aṣṭādhyāyī keep processing until the Dhātu is terminated and the paradigm is produced. In this way there are a series of steps operated by the rules till the termination of a Dhātu in producing a paradigm. The complete algorithm of the tool is based on Panini’s system. And this produces a Rūpasiddhi table of a particular paradigm of a Sanskrit verb root technically called Dhātu.
Keywords: Aṣṭādhyāyī; Ᾱrdhādhatuka; Ᾱtmanepada; Dhātupāṭha; Gaṇapāṭha; Parasmaipadam; Rūpasiddhi; Sārvadhātuka; Subanta; Syllabic Modifiers; Tiṅ; Tiṅanta; Vācya
1. Introduction
A Sanskrit syntactic word is either a Subanta or Tiṅanta. They are primarily generated from a nominal stem and a verbal stem respectively which are stored in two lexicons called Gaṇapāṭha (GP) and Dhātupāṭha (DP). A paradigm having the terminal of anyone of the 21 Sup suffixes is called a Subanta while a paradigm having the terminal of anyone of the 18 Tiṅ suffixes is called a Tiṅanta. Rūpasiddhi is the process of paradigm generation through the system Aṣṭādhyāyī which is comprised of around 4000 Sūtra or rules.
A Tiṅanta Rūpasiddhi explains building of a Tiṅanta (i.e. a verbal paradigm) from its stem. Primarily it displays the steps and series of the Aṣṭādhyāyī Sūtras or Rules which process the stem, along with their corresponding operations in hierarchal order.
The tool is supposed to generate a table of Tiṅanta Rūpasiddhi. It is designed on the basis of Panini hierarchy.
1.1. Related Works
Professional guide books on Rūpasiddhi are available. They are written with a view to fulfil the requirements of school, college students. Remarkable researches on Pāṇinian morphology has been carried out. Linguists have established terminologies for defining Pāṇinian derivational process. Cardona2 postulates the following formula for a Pada or syntactical word formation.
(N – En) p • • • (V – Ev) p
where p stands for Pada[1] having either nominal affixes (En) or verbal affixes (Ev) which is following the respective bases. Girish Nath Jha3 who presents a computational morphology of Sanskrit case affixes. Under the research, a tool written on Prolog was developed, which generates all the 24 (21 + 3) forms (of a Prātipadika) in 8 cases and parses constituents and details of processing. Another tool[2] like this tool also has been developed on Java/JSP by the same author collaborating with Divakar Mishra, Subhash Chandra and Vertika Verma. This is live on http://sanskrit.jnu.ac.in/subanta/generate.jsp#results. One of the tools available on this site is “Sanskrit Verb Analyzer (संस्कृत-तिङन्त-विश्लेषक)” which parses a given verb. It is a part of M.Phil. dissertation of Muktanand Agrawal1. Muktanand discusses an analysis of Sanskrit verb forms on the basis of the Pāṇinian morphology somewhat in reverse direction. “Pāṇinian analysis identifies different morphemes in any given Pada and presents an analysis where he provides step-by-step methodology to derive a verb form from a given verb root in certain paradigms”1. Parsed output of the verb भवति (Bhavati), as an example is given here:
भवति { ( कर्तृवाच्य ) भू ( [ भ्वादिगण ] [ अनिट् ] [ अकर्मक ] ) ( [ लट् ] ) तिप् ( [ परस्मै ] [ प्रथम-पुरुष ] [ एकवचन ] ) }
A declension machine by Dhaval Patel, Sivakumari Katuri and Mr. Marcis Gasuns resembles the present work. It derives a given Dhātu in four Lakāras (i.e., Laṭ, Loṭ, Laṅ, and Vidhi-liṅ) in anyone of three voices/Vācyas. The Dhātu should be entered with its Anubandhas. It accepts Devanagari Unicode, IAST and SLP1 as writing scripts. It generates all the Rūpasiddhi steps in all persons and numbers of the selected Lakāra. It is written on Php. The source code is available on https://github.com/drdhaval2785/SanskritVerbtiGanta. The tool is live on http://lanover.com/lan/sanskrit/SanskritVerb/tiGanta.html with other Sanskrit tools.
Other online morphological generators and analysers are संस्कृतपदविश्लेषकः (TDIL, Govt. of India)[3], Sanskrit Verb Conjugation Rules (Green Message)[4] which provides output like aforementioned tools.
2. Sanskrit Verb Morphology – Panini’s View or Generative Perspective
Technically a Sanskrit verb (syntactic) is called a Tiṅantam in Panini’s terminology. In generative process of a particular paradigm, first of all a Dhātu is always followed by anyone of ten Lakāras. A Lakāra is a semantic constraint which carries the meaning of different syntactic relations, tenses and moods. Syntactically a Lakāra expresses either a subject or an object or the Bhāva (the root or the action itself). After the assignment of a Lakāra, that Lakāra is replaced by Tiṅ.
Tiṅ is an acronym of 18 suffixes. Basically they are comprised of two sets of suffixes (9 + 9 = 18) means they are divided into two groups, i.e. 1) Parasmaipadam (Pp) and 2) Ᾱtmanepadam (Ap).
There are three groups of Dhātus determined by the two sets of Tiṅ. First group is called Parasmaipadī. This group can be terminated only in Pp (not in Ap). Literally Parasmaipadī means that which has Pp. Second group is called Ᾱtmanepadī literally meaning that which has Ap. This group can be terminated only in Ap (not in Pp). The third group can be terminated either in Pp or Ap. And this is called Ubhayapadī or that which has both Pp and Ap.
The first three suffixes of both Pp and Ap are called Prathamapuruṣa, the second three are called Madhyamapuruṣa and third three are called Uttamapuruṣa. So, now it has three groups of three suffixes for each of Pp and Ap. The first suffix of anyone of the groups is called Ekavacanam, second is called Dvivacanam and third is called Bahuvacanam6,8.
Table 1. Tiṅ Suffixes
|
|
Ekavacanam |
Dvivacanam |
Bahuvacanam |
|
Prathamapuruṣa |
Tip/Ta |
Tas/Ᾱtām |
Jhi/Jha |
|
Madhyamapuruṣa |
Sip/Thās |
Thas/Ᾱthām |
Tha/Dhvam |
|
Uttamapuruṣa |
Mip/Iṭ |
Vas/Vahi |
Mas/Mahin |
3. The Architecture of the Tool
The tool has two main components – DP and Aṣṭādhyāyī. DP is the verbal stem lexicon of Panini grammar and it is applied as it is for the tool. The Aṣṭādhyāyī is the central generative processor (CGP) of the tool based on actual Aṣṭādhyāyī of Panini5,9.
3.1. Aṣṭādhyāyī — the Rules
It consists of 7 main blocks of rules9. They are:
A 1San introduces and names the technical terms which will be used in different operations by other rules, especially the 4Vid. It defines and classifies the terminologies and data in different categories. It is also the tagger in the system. For instance the 1San rules Vṛddhirādhaic and Adenguṇaḥ define the phonological terms Vṛddhi and Guṇa. The Bhūvādayo Dhātavaḥ names and classifies the list of verbal stems in DP as Dhātu. A 2Adh declares a domain, an area and scope of the succeeding rules that these rules must function in this definite context, in this definite scope etc. A 3Par is a general rule that is applicable through the entire the system. Iko Guṇavṛddhiḥ is an example. A 4Vid executes an operation in a particular scope. Just as Vartamāne Laṭ executes the Laṭ after a root in the domain suffixation. A 5Nis prohibits a condition previously defined by another rule. Halantyam classifies a last consonant of an Upadeśa as It. Na Vibhaktau Tusmāḥ prohibits the formulation class It on these consonants T, Th, D, Dh, N, S, and M of a nominal suffix. A 6Niy restricts an operation. A 7Ati allows an object to be stored the attribute of else object.
Fig.1. CGP of Tiṅanta-Rūpasiddhi Tool
3.2. DP
Panini’s DP is a well-structured verbal root lexicon5. This database is designed with phonetic keys called Anubandha (Anb) for making the roots easily interpretable to the rules. The Anbs are directly attached to a root. For instance, Gam (to go) is entered as Gamḷ in DP. An Anb either precedes or succeeds a root entry. In some cases an Anb can be attached to both sides of a root. Anb is a Paninian hierarchy which tags data as their lexical ID that handles many operations in the generative process. For example the sound Ṅ, Svarita and Anudātta accents (SAc & AAc) are some of the Anbs which are attached to roots. These Anbs are tagged as terminal ID of roots that decide which roots will be terminated in Pp or Ap. The rule Anudāttaṅitātmanepadam (03.01.021, Ast) interprets the roots tagged with Anb or ID of AAc and Ṅ will be terminated in Ap. And the rule Svaritañitah Kartrabhiprāye Kriyāphale (03.01.072, Ast) terminates roots tagged with Ṅ and SAc in Ap in the semantic environment of the result (fruit) of a root goes to the agent.
The roots in DP have 10 sections which means they are divided in 10 classes. A Dhātu is always followed by a Lakāra and terminated in anyone of the final terminals of Tiṅ.
Fig.2. Roots Terminal Layers
Lakāra is the first layer of three semantic terminals undergone in the scope of suffixation. The terminals are Kartṛ/agent-expression (KtE), Karma/object-expression (KmE), and Bhāva/action-expression (BvE). After terminating at any one of these three semantic terminals, it undergoes to next terminals of mood and tenses i.e. the Lakāras. From this, it goes to next two terminals of Pp and Ap. Next is three personal terminals of first, second and third persons. Then it is finally terminated in the numbers i.e., singular, dual, and plural.
3.3. Root Classes
One of the reasons for dividing roots in 10 classes is phonetic variances of the paradigms of roots. Technically a syllabic modifier (SM) of a particular class of a stem occurs when the stem undergoes primary terminal process. SM modifies the primary syllabic feature, mostly mono-syllabic, of a stem. Just as Bhū is modified as Bhava while it undergoes of primary terminal process of Tiṅ. Thus it becomes Bhava-ti > Bhū + ti. SM has no role of semantic modification. Thus Bhava still means Bhū. SM are 7 in number. They are listed below:
Table 2. Root Syllabic Modifiers
|
Class |
Root Syllabic Modifier |
|
Bhvādī (C1) |
a > śap |
|
Adādī (C2) |
--- |
|
Juhotyādī (C3) |
--- |
|
Divādī (C4) |
a > śyan |
|
Svādī (C5) |
nu > śnu |
|
Tudādī (C6) |
a > śa |
|
Rudhādī (C7) |
na > śnam |
|
Tanādī (C8) |
u |
|
Krayādī (C9) |
nā > śnā |
|
Curādī (C10) |
aya > ṇic |
Thus C2 and C3 have no SM which means stems of these two classes undergo no syllabic modification. Other classes also have limited SMs according to the Lakāra. Technically the limitation of SM occurrence is handled by classifying the 10 La into two groups as Sārvadhātuka[5] (SLa) and Ᾱrdhādhatuka[6],8 (ALa). An SM occurs only in the SLa of the non C2 and C3 classes, that also while the La expresses the agent, i.e. in the active voice. Laṭ, Loṭ, Laṅ, and Leṭ1 are SLa[7]. Liṭ, Luṭ, Lṛṭ, Liṅ1, Luṅ, Lṛṅ/ are ALa[8].
3.4. Repetition of Listing Roots in DP
A root is listed in more than one class just because (a) the root optionally has an SM or (2) it has more than one SM. For instance, Īḍ is listed twice in C2 and C10 and has the paradigms Īḍ-te (of C2) and Īḍ-ya-ti (of C10). First example has no primary syllabic change or it has 0 syllabic modification in root and second example has the syllabic modification in the primary root. Both the paradigms express the same meaning. /mən/ is another example of having more than two SM. It is listed in C4 and C6 so it has two different modifications in root. For example, Manute and Manyate have two syllabic modifications.
Homonymous roots are listed two more times in the same classes and in another classes. Bhū is listed once in C1 and twice in C10 in three different meanings 1) to be, 2) to get and, 3) to mix respectively.
4. The Tool Output – The Rūpasiddhi Table
According to its terminal potency, a Dhātu has a maximum of 396 Tiṅanta paradigms and a minimum of 110 Tiṅanta paradigms. The tool processes a Dhātu and creates a Rūpasiddhi table for each paradigm. Here is an example of a Rūpasiddhi table of the paradigm Bhavati of the Dhātu Bhū.
Table 1. Rūpasiddhi Table of Dhātu Bhū in Laṭ, Ppr, 1v
|
Rule |
Operation |
Output |
|
Vartamāne laṭ |
Assignment of Lakāra in present |
Bhū+laṭ |
|
--- |
--- |
Bhū+l |
|
Laḥ karmaṇi ca bhāve cākarmakebhyah |
Assign the semantic terminals, active, passive, BvE |
Bhū+l |
|
Tiptasjhisipthasthamibvasmastātāñjhathāsāthāmdhvamiḍvahimahiṅ |
Assign the two secondary terminals AP PP |
Bhū+Tiṅ |
|
Laḥ parasmaipadam |
Defines which much be Pp in Tin |
Bhū+Tiṅ |
|
Taṅānāvātmanepadam |
Defines which much be Ap in Tin |
Bhū+Tiṅ |
|
Śeṣātkartari parasmaipadam |
Assign Pp |
Bhū+tiptasjhisipthasthamibvasmas |
|
Tiṅastrīṇi trīṇi prathamamadhyamottamāḥ |
Divide Tiṅ into 9 sets having 3 suffixes each |
Bhū+tiptasjhisipthasthamibvasmas |
|
Śeṣe prathamaḥ |
Assign 3rd person |
Bhū+tip |
|
Tiṅśitsārvadhātukam |
Define LSd |
Bhū+tip |
|
Kartari śap |
Assign SM (syllabic modifier) |
Bhū+śap+tip |
|
--- |
--- |
Bhū+a+tip |
|
Sārvadhātukārdhadhātukayoh |
Modifies the syllable of root |
Bho+a+tip |
|
Ecoyavāyāvah |
Final Phonetic operation |
Bhav+a+tip |
|
--- |
|
Bhav+a+ti |
|
--- |
|
Bhavati |
Fig 3. A Screenshot of the Tiṅanta Rūpasiddhi Tool Running of JSP Web App
5. Conclusion
The tool is to be lived online for evaluation from the users. There are variances of a Rūpasiddhi table traditionally taught in Gurukuls and other traditional institutions etc. User comments and may improve the architecture of the tool more smart. Other external links and commentaries related to the rules can be upgraded.
Acknowledgements
I would like to show my gratitude to all those who supported and encouraged me to develop this tool. I thank my teachers, friends and juniors for their comments and suggestion in improving the tool.
References