Panini in BNF: Reply to comments

Hans Nilsson

unread,

Aug 10, 2010, 5:49:42 AM8/10/10

to sams...@googlegroups.com

Dear Mr Abhyankar and Mr Vaidya,

Thank you for the valuable comments. Here are a few replies :

1. Regarding karakas, I have tried to make a distinction between karaka = syntactic categorisation and vibhakti = morphological case endings. I have now added a statement for sup-vibhakti. I have however changed the English term form sampradana to ”indirect object”. I have also added <upasarga-vibhakti> which is the only term I can find for posession (genitive).

2. Regarding base, root and stem: OK, in order to avoid confusion, I will go back to the conventional terminology as follows:

Dhatu = verbal root (both primitive and derived??)

Pratipadika = nominal stem (both primitive and derived ??)

Very often, I see mentioned both ”verbal root” AND ”verbal stem” as two different concepts. Is the difference then the following ??

- verbal root (dhatu) = bhU

- verbal stem (???) = bhav (i.e. dhatu+ vikarana )

If so, what is then the correct sanskrit term for this second concept of « verbal stem » ?

This actually leads me to the following more detailed description of the derivation of a tinanta :

<tiṅanta > ::= {<upasarga>} <dhātu> <vikaraṇa> (<lakāra> => <tiN>)

; According to Panini’s derivation method, lakāra is first added, and then later replaced by tiN

where <vikarana> is a marker of insertion for conjugation class, tense or mood (for more details, see the updated document.)

Furthermore, I have tried to incorporate the term dhātu-sādhita (verbal derivate). I am not sure I have got it right

3. I have added the karaka lET (subjunctive)

4. Trying to analyze your example, I come to the following response :

pat : verbal root (dhatu)
paat : verbal stem, with added vikarana causing guna/vrddhi

pranipaat : new stem containing two added upasargas (?)

pranipaatena : verbal derivative (dhAtu-sAdhita)

5. I have added back sup, tin and vikarana to the list of pratyayas. After all, they ARE pratyayas. Butt hen I can not any longer use these statements since sup and tin are pratyayas:

<pada> ::= <prakṛti> [<pratyaya>] [<vibhakti>]

<subanta> ::= < prakṛti> [<pratyaya>] []

Instead I have avoided the use of the term pratyaya, and made more explicit definitions as follows:

<tiṅanta > ::= {<upasarga>} <dhātu> <vikaraṇa> <tiN>
<subanta> ::= <prātipadika > []

which should be fine, since both dhātu and prātipadika can themselves both be primitive and derived. Please also note that I have introduced upasarga as an optional prefix.

6. Regarding derivatives, I have illustrated it with the following table. Please correct me if something is wrong.

Derivation
From:	To: prātipadika	To: dhātu
prātipadika	taddhitanta, strīanta	nāmadhatū
dhātu	kṛdanta	sannanta etc

7. Regarding the relation to modern/English categories: I just wanted to show the correspondence between Panini word classes and modern terminology. Please note: I am only talking about “Lexical categories” like noun, verb, adverb etc. I am NOT talking about “phrasal categories” like subject, predicate, adverbial etc (which according to modern grammatical terminology are called Noun Phrase, Verb Phrase, Adverb Phrase etc).

From this point of view, adverb is a lexical category. So there is one “tomorrow” which is an adverb, and another homonym “tomorrow” which is a noun, just like the adverb “hard” is homonymous with the adjective “hard”.

In sanskrit, as I understand it, an adverb is always a subanta, but belonging to one of several sub-categories. It can be contained as a primitive in the list <svaradi> or be derived through as <taddhitanta>. However, I had difficulties expressing this in BNF, and so I wrote the following, which is a bit misleading (it looks like all svaradi can be adverbs):

<ADVERB> ::= < svarādi > ; as listed in gaṇapāṭha
| <kriyā-viśeṣana> ; derived, e.g. by taddhita

I would be happy for any suggestion of better expressing this. Or maybe it should be left out altogether, regarding it not as “lexical category” but a “phrasal category”, which is not covered in this description.

The new BNF version is inserted below. The real document, which is more nicely formatted can be found at the following addresses: http://www.hansnilsson.se/BNF.doc and here: http://www.hansnilsson.se/BNF.pdf

Hans Nilsson

---------------------------------------------------------------------------------------------------------------------------------

1. Lexical categories

1.1 Paninian view

<pada (word)> ::= <subanta (nominal)> | <tiṅanta(verb)> ; A 1.4.14

<avyaya (indeclinable)> ::= <svarādi (sun etc.)> ; A 1.1.37-41
 | < nipāta (particles) > ; A 1.1.56
 | <upasarga (preverb)> ; A 1.4.58
 | <gati > ; A 1.4.60 (=upasarga)

<subanta> ::= <prātipadika> []

; Panini has only two main word classes. He includes avyayas in subanta, adding endings and then deleting them again in A 2.4.82! This is described above by being optional.

; Yāska, who preceded Panini, made the following categorization of word classes:

<padajāta> ::= <ākhyāta (finite verb)> | <nāman (nominal> | < nipāta > | <upasarga (preverb)>

1.2. Relation to modern/English lexical categories

< nipāta> ::= <CONJUNCTION> | <PREPOSITION>

<ADVERB> ::= < svarādi > ; as listed in gaṇapāṭha
| <kriyā-viśeṣana> ; derived, e.g. by taddhita

; P does not clearly differentiate between adjectives (viśeṣaṇa) and nouns. Only in A 2.1.57 does he talk about viśeṣaṇa (qualifying) and viśeṣya (qualified). Adjectives were not recognized as a separate lexical category until very late. Even in latin, they were not really separated.

; According to Gombrich (”He cooks softly”: Adverbs in Sanskrit Grammar), « words which appear to as to be adverbs, or used adverbially, are heterogeneously classified by P. All come into the category of subanta. Certain simple adverbs are listed as avyaya in the Gaṇapāṭha. » The term kriyā-viśeṣana is first used by Patanjali, and seems to be referring to derived words.

2. Morphology

<pada> ::= <subanta> | <tiṅanta> ; A 1.4.14

<prakṛti (base word)> ::= <dhātu> | <prātipadika>

<vibhakti> ::= | <tiN> ; A 1.3.104

<pada> ::= <prakṛti> [<pratyaya>] [<vibhakti>]

; tiN, sUP and vikaraṇa are formally pratyayas according to Panini.

<dhātu-pratyaya (verbal affix)> ::= <kṛt> | <tiN> | <vikaraṇa> ; can be added to verbs

<prātipadika-pratyaya (nominal affix)> ::= <taddhita> | | <strī> ; can be added to nominals

< dhātu-sādhita (verbal derivative)> ::= ; can be derived from verbs
<kṛdanta> | <sannanta> | <yaṇanta> | <nijanta>

< prātipadika-sādhita (nominal derivative)> ::= ; can be derived from nominals
<taddhitanta> | <strīanta>| <nāmadhatū >

Derivation
From:	To: prātipadika	To: dhātu
prātipadika	taddhitanta, strīanta	nāmadhatū
dhātu	kṛdanta	sannanta etc

; Yāska gives the following general morphology:

; This means that <śabda> == <prakṛti> ??

2.1 Tinanta morphology

<tiṅanta > ::= {<upasarga>} <dhātu> <vikaraṇa> (<lakāra> => <tiN>)

; According to Panini’s derivation method, lakāra is first added, and then later replaced by tiN

<dhātu (verbal base)> ::= < mūla dhātu (primitive)> | < sanādyanta dhātu (derived)>

< mūla dhātu> ::= <bhūvādayaḥ dhātu> ; from dhātupāṭha, A 1.3.1

<sanādyantāḥ dhātu> ::=
<sannanta (desiderative)> | <yaṇanta (intensive)>
| <nijanta (causative)> | <nāmadhatū (verbal noun)>

<upasarga> ::= ; A 1.4.59
 “pra” | “parā” | “apa” | “sam” | “anu” | “ava”
 | “niḥ” | “duḥ” | “vi” | “ā” | “ni” | “adhi” | “api”
 | “ati” | “su” | “ut” | “abhi” | “prati” | “pari” | “upa”

<vikaraṇa (infix)> ::= ; conjugation class, tense and mood markers
 | ”ŚaP” | ”LUK” | ”ŚLU” | ”ŚyaN” | ”Śnu” | ”Śa” | ”ŚnaM” | ”u” | ”Śnā” | ”NiC” ; A.3.1.68 ff
 | ”CLI” | ”sIC”| ”sya”| ”yaK” | ”yāsUṬ”| ; for lUṄ, lŖṄ etc

<tiN> ::= "tiP" | … | "mahiN" ; A 3.4.77

2.2 Subanta morphology

<subanta> ::= <prātipadika > []

::= "sU" | ... | "suP" ; A 4.1.2

< prātipadika (nominal base)> ::= < mūla prātipadika (primitive nom.base)>
| <vyutpanna prātipadika (derived nom.base)>

< mūla prātipadika)> ::= < gaṇapāṭha prātipadika> ; A 1.2.45
 | <sarvanāman (pronouns) ; A1.1.27
 | <saṅkhyā (numbers)>

<kṛdanta> ::= <dhatū> <kṛt>

<samāsa > ::= ; A 2.1.3
 <avyayibhāva (indecliable)> ; A 2.1.5
 | <bahuvrīhi (unmentioned)> ; A 2.3.23
 | <tatpurusa (case-determining)> ; A 2.1.22
 | <dvandva (pair)> ; A 2.2.59

3. Other grammatical elements

<kāraka > ::= ; A 1.4.23 syntactic-semantic categories
 <kartā (agent)> ; A 1.4.54-55
 | <karma (object)> ; A 1.4.49-53
 | <karaṇa (instrument)> ; A 1.4.42-44
 | <sampradāna (indir.object)> ; A 1.4.32-41
 | <apādāna (separation)> ; A 1.4.24-31
 | <upapada-vibhakti (posession)>
 | <adhikaraṇa (location)> ; A 1.4.45-48

Eddie

unread,

Aug 11, 2010, 4:12:30 PM8/11/10

to sams...@googlegroups.com, Eddie Hadley

Gentlemen,

Such a clarification of these grammatical elements is very much to be desired. Such glosses are there are, appear somewhat confused, not to mention, confusing.

In this matter, I have found Paul Kiparsky's paper On the Architecture of Pāṇini's Grammar, among the more acceptable ( http://www.stanford.edu/~kiparsky/Papers/hyderabad.pdf ).

His categorization of dhātu, prātipadika, pada etc., together with their 'basic' and 'derived' varieties, which he discuses side by side with the numbered sūtras, do make sense to me.

Would anyone like to proffer an opinion as to how closely they (the English words) conforms with the Saṃskṛta.

Thanks,

Eddie

--
You received this message because you are subscribed to the Google Groups "samskrita" group.
To post to this group, send email to sams...@googlegroups.com.
To unsubscribe from this group, send email to samskrita+...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/samskrita?hl=en.

No virus found in this incoming message.
Checked by AVG - www.avg.com
Version: 9.0.851 / Virus Database: 271.1.1/3060 - Release Date: 08/09/10 07:35:00

Piergiorgio Muzi

unread,

Aug 12, 2010, 1:38:42 PM8/12/10

to sams...@googlegroups.com

Dear Hans, of course, I agree with the suggestion of Eddie about the long research and study made by Kiparsky et alii in the field of synchronic general linguistics and its generative interpretation of Sanskrit grammar. Since the beginning this school worked in co-operation with computer science and formalization of natural languages. I think that it can match your interesting and difficult work. From my point of you, not been too much a follower of Chomsky, I would consider other points of view: above all not to forget the enormous and productive effort made by the eighteenth century's German and British philologists.

BTW, I have considered your issue about bhU/bhav. I don't believe that you need consider the change of the apophonic degree from weak (basal or zero) bhU to guNa (middle degree) bhav/bho as an application of vikaraNa. I would say vikaraNa for the application of the thematic vowel "a" to have bhav-a-ti. This vowel characterizes all the classes of 1st conjugation (bhuAdi, dIvAdi, tudAdi, curAdi). But some verbs of first and 6th class have the root in weak degree amplified with "cch": gacchati, yacchati, icchati, Rcchati, pRcchati. I would consider also this a vikaraNa which precedes "a", but I am not sure. About the definition of this suffix by means of Panini's grammar terminology we'd better question pandit Abhyankar, who will certainly enlighten us. This suffix is the same you find much oftener in Old Greek and Latin and arrives through this language up to the modern western languages, including English (-ish, in finish, distinguish, perish...). This was originally sk gave an inchoative or intensive sense to the root.

Apart from this, how would you define the change of the root by means of nasal infix, n or na (sic>sincati; yuj>yunakti, yunjanti...)?. And what about the formation of a new stem by means of duplication (3rd class, but also some verbs of 1st class: tiSThati, pibati, sIdati<si-sdati...; and in perfect tense, as well as in desiderative and intensive secondary conjugation)?

Best wishes for your research!

Piergiorgio Muzi

धनंजय वैद्य <deejayvaidya@yahoo.com>

unread,

Aug 12, 2010, 12:10:51 PM8/12/10

to samskrita

I agree - the Kiparsky paper is very good.

Dhananjay

Hans Nilsson

unread,

Aug 13, 2010, 2:23:06 AM8/13/10

to sams...@googlegroups.com

Eddie,

Thanks for reminding me of Kiparsky’s Hyderabad paper. I read it some time ago, but it was a bit hard for me at that time. Now that I am deeper into Panini, I will really re-read it.

Glancing through it now, I found an interesting paragraph on vikaraNa, which relates to what Piergiorgio wrote here on the list the other day:

“BTW, I have considered your issue about bhU/bhav. I don't believe that you need consider the change of the apophonic degree from weak (basal or zero) bhU to guNa (middle degree) bhav/bho as an application of vikaraNa. I would say vikaraNa for the application of the thematic vowel "a" to have bhav-a-ti. This vowel characterizes all the classes of 1st conjugation (bhuAdi, dIvAdi, tudAdi, curAdi). But some verbs of first and 6th class have the root in weak degree amplified with "cch": gacchati, yacchati, icchati, Rcchati, pRcchati. I would consider also this a vikaraNa which precedes "a", but I am not sure”.

Kiparsky writes (on page 40):

“The vikaraNas bear grammatical markers which among other things have effect on the form of the rott, particularly on its accent and strong vs weak grade.”

So when I write the following definition:

<tiṅanta > ::= {<upasarga>} <dhātu> <vikaraṇa> <tiN>)

It means that (and the other entities) should not just be seen as a sequence of string, but a sequence of operators, that imply both string insertions AND operations on the preceeding string (anGa) like guna/vrddhi change.

Regards

Hans

Piergiorgio Muzi

unread,

Aug 16, 2010, 12:36:21 PM8/16/10

to sams...@googlegroups.com

Dear Hans, thanks for the clarification about vikaraNa. So, I suppose that's it needs to contain also reference to insertion of nasal (n or na). Do you think so?

I share your appreciation of Coulson's grammar (Teach Yourself Sanskrit). I have used it for about 5 years with my beginning students in Singapore. The only criticism I could do is about its notational system for sandhi and samAsa, a bit harsh...Two precious points in Coulson's:

1) The sandhi of det. "sa", which is explained clearly thanks to reference to I.E. comparison (Old Greek, but we could add also Gothic sa, saa, thata).

2) The clear explanation of the alternation an/a (e.g. Instr.S. brahman-aH / Instr.P. brahma-bhyaH; or han-ti / ha-taH, etc. as well for am/a..), due not to dropping of n or m, but to the historical development of n or m, when a vowel, into a, like ar into R or o into u, etc.

If you like Coulson's, probably you would like more Goldman & Goldman's DevavANIpraveshikA, which associates the vernacular Indian technical terms systematically and has a rich chapter about derivations of nouns, etc.

BTW, you haven't used the common term taddhita, for pratyaya starting from prakRti, including kRdanta, to form names or adjectives (subanta) called taddhitAnta, contrasting with kRdanta. According to Goldman, we should have:

1) dhAtu+ kRt-pratyaya > kRdanta

2) prakRti (including both original nAman and kRdanta) + taddhita-pratyaya > taddhitAnta

3) recursively, taddhitAnta + taddhita-pratyaya > taddhitAnta

A note: It is very common the usage of the expression "past passive participle" or p.p.p. for shrutaH, mataH, gataH, jAtaH... (the same usage is in Latin, etc.). But this participle is passive only with a transitive verb. mataH and shrutaH are passive, but gataH is normally active (=who has gone). In this case there is no difference in meaning between gatavAn and gataH, while there is one between matavAn and mataH.

At last, I allow myself to remind you the existence of perfect participles; Atm. cakRvAn, Par. cakrANaH, not so common in classic Sanskrit, except for some verbs: vidvAn is common enough in philosophy.

I hope that I can be useful as well as your reflections are useful to my study.

Namaste,

Piergiorgio

----- Original Message -----

From: Hans Nilsson

To: sams...@googlegroups.com

Hans Nilsson

unread,

Aug 19, 2010, 4:46:00 AM8/19/10

to sams...@googlegroups.com

Dear Piergiorgi,

Thanks for your additional comments. Yes, I already have also Goldman&Goldman which is also very good. When I get stuck in one book, I switch to another and so on. I have too many books, and too little time to read… Maybe it’s time to re-read G&G again…

But I do use taddhita/taddhitanta in my BNF document. I am attaching the final version.

Kindest regards

Reply all

Reply to author

Forward