Having completed the study of the first volume of “Introduction to Sanskrit”, part 1, by Thomas Egenes, although I will need to come back to it again and again, I made a first attempt at some form of poetry in sanskrit, if I may say so. On this page are the devanāgarī text, the transliteration and the metre pattern of the verses. The poem is entitled प्रिया पत्नी (priyā patnī).
It taps only into the grammar and vocabulary from Egenes’ first volume, apart from the word शिव (Śiva), the greatest god, whom he writes about in the second volume for that matter.
The poem, so to speak, is made up of 4 stanzas. Each stanza is made up of four lines (pādas) of 8 syllables (akṣaras). This metre (chandas) is called anuṣṭubh, better known as śloka, common to the Bhagavad-Gītā. So it seems I have written my first four ślokas.
However, I am aware of the existence of a whole science of composition in sanskrit literature, called kāvyas, whose surface I have only skimmed so far. The point I want to make is that, although my verses may look or sound like a śloka (but I doubt it), it is nothing like the standard śloka, of course, and I do regret it.
From what I have gathered, the end of each verse or line, called pāda, is marked by a pause or cæsura called yati. This is true for any pāda in any kind of chandas. In a śloka, in particular, there are two remarkable pauses. A stronger one at the end of the second pāda, marked by a single bar |, and an even stronger pause at the end of the fourth pāda, marked by a double bar ||.
pāda a) 1 2 3 4 5 6 7 8
pāda b) 1 2 3 4 5 6 7 8 |
pāda c) 1 2 3 4 5 6 7 8
pāda d) 1 2 3 4 5 6 7 8 ||
I did not write with the same pattern of pauses for my first poem in sanskrit, because I write within the limits of my knowledge of the language, obviously, not because I do not want to. I wrote with more pauses and with different positions for the pauses.
But this is not the main issue. The issue is important — for one, just because I like to abide by the rules of greater men than me. But the problem rather is to know where to cut a sentence (or a clause) so as to make 2 pādas out of it. This cannot be done wherever one likes.
pādas and meeting the conditions for a yati
The idea is the following. All words in a spoken sequence flow without a break. In sanskrit, this is even expressed on the paper, where written words are concatenated or sometimes merged into a continuous chain of letters where words sometimes lose their original forms. This verbal flow can and must be paused or interrupted in certain conditions. The one who writes pādas wants to use this possibility to artistically compose pādas while lawfully abiding to the rules of the language that permit or induce pauses or stops.
We want to compose a certain kind of pādas. I chose the verb “compose” because poetry is all about sounds, although it usually is written. Let’s stick with the 8-syllable example. We must have a pause, a yati, at the end of a pāda. This pause is permitted, or not, by three conditions: syntax, as in other languages; prosody, that is, roughly, rhythm; and what is called sandhi rules, specific to sanskrit, as far as I know. The decision-making sequence should be first syntax, then sandhi and finally prosody or first prosody, then syntax and finally sandhi.(1) Whatever the conception of poetry that prevails behind the final choice, if there is one actually, the decision is to discuss sandhi in a third and last point because, among the three conditions, it is the one that is the most specific to sanskrit, which is our focus here.
Condition 1: syntax
I will take the example of the pāda scheme above. Between a and b on the one hand, and c and d on the other hand, there is some syntactical unity, although each pair of pādas does not necessarily form a whole sentence; it could be a clause. For prosodic or other reasons they have been divided into 2 pādas. For example:
She went to the house
Near the Ganga river
Because she was tired
After all the long walk.
This whole stanza or paragraph is actually 1 sentence. The first 2 lines form a clause and the other 2 form another, coordinated, clause. These clauses have each been cut into 2 lines because, as such, they are syntactically comprehensible. Of course there also exist enjambements used for the particular purpose of poetic expression.
Condition 2: prosody
Metre, rhythm and intonation also permit, or not, to have a pause at certain point in a sentence. Since we want to compose pādas with a specific number of syllables, rhythm and intonation must express themselves within the constraint of this number of syllables. The rhythm does not determine how many syllables there should be, rather the number of syllables influences the degree of freedom in rhythm expression or the nature of rhythm expression. This should not be seen as a limitation though, but as liberating creative discipline.
For my first sanskrit poem, I have indicated the rhythm patterns. However, at this stage of my studies, I do not have much control over prosody. Rarely a metre was deliberate, most often these patterns were imposed by the sentences I could afford to make.
Condition 3: sandhi
This is the modification of an original sound in a word due to the other sounds around it. This happens for articulatory reasons. In English, for example, the original sound -b- may be pronounced -p- in a word like absorb, because of the following -s- sound. Here the transformation is a matter of economy, or even maybe aesthetics, because the -p- and the -s- belong to the same family of unvoiced (non-vibrating) consonants, and this requires changes to the mouth configuration which are less effortful. This is called assimilation, more precisely regressive or anticipatory assimilation. The sandhi system however encompasses more than these assimilations. Moreover, this sort of changes and others happen in a regular and systematic basis in sanskrit.(2)
These sandhi rules apply to certain combinations of sounds, a lot of them, but not all. Where these rules apply, they state that a change must occur (and which one) or that a sound must stay as is. So, even if a sound does not change, it does not mean that no sandhi rules have been applied to it. It could be that it did not change because of sandhi rules.
What does sandhi have to do with the possibility of pauses and splits in a sentence?
There are 2 main cases.
A) The sandhi rules do not apply. This can happen. A final vowel meeting an initial consonant are out of the scope of sandhi rules. Thus they are not modified (if so, maybe for other reasons, which I do not know right now) and a pause or split here would be possible, regardless of customs, if syntax and prosody demand so. This seems to be the case for most of the extracts from the Bhagavad-Gītā to be found in “Introduction to Sanskrit”, part 2, of which I have read only some parts about poetics, by Thomas Egenes (Point Loma Publications, Inc. 1995).
Such a configuration can be seen or heard in my own pādas 1–2, 13–14 and 16 (standing alone), but pādas 1, 13 and 16 are whole sentences in themselves anyway, so they come with natural pauses (marked by a single bar |), and sandhi rules may allow splitting the lines further if need be. So this was a wasted opportunity and evidence of poor writing on my part.
B) The sandhi rules do apply. There are then 2 other cases.
Case 1: the sandhi rules cause the final sound of a word to merge with the initial sound of the next word. 2 sounds become one, for example when a final short -a- meets an initial short -a-, they become 1 long -a- sound. Obviously we cannot have a pause and cut the sentence here.
This situation does not occur in my own pādas. That is the least I could do.
Case 2: the sandhi rules do not cause the final sound of a word to merge with the initial sound of the next word. 2 sub-cases occur.
Sub-case i: the sounds may be modified individually, the first one most often, the second one sometimes, or both. Splitting the sentence here is possible, regardless of customs, but with some caveats. First it is not convenient for the reader or listener who perceives a sound change in a word at the end of a pāda but has to move forward to the next pāda to understand the reason for this change. Second, it is not convenient for the composer or writer because he or she may forget to apply sandhi if the cause for it lies in the next pāda. Yet, at this point of my study, I have read such verses in sanskrit. On p. 241 from “Introduction to Sanskrit”, part 2 (ibid.), Thomas Egenes gives the example of two pādas where the first ends with a visarga (ḥ), transformed into an -s- because of the -t- in the first word of the next pāda. This is sandhi. So, a pause and a split were pemitted here.
This case occurs in my own pādas 11–12. The original final syllable or akṣaras -धन् (dhan) became ध्ंस् (dhaṃs) because of the initial त् (t) in the next pāda, number 12.
Sub-case ii: neither the final sound nor the initial sound is modified. In such a situation, regardless of customs, a pause and a split would not be impossible, if syntax and prosody urge one to appear.
This happens in my own pādas 6–7 and 9–10. These are pairs without the stronger pause at the end of the first of the two pādas. pāda 6 ends with a bare consonant (devoid of vowels, marked by the virāma symbol that you can see at the bottom of final characters). The next pāda, number 7, starts with a consonant (not bare). This gives the {-t s-} consonant combination. It is the same situation for pāda 9 and 10 where the consonant combination is {-t t-}. The particular meetings here do not induce any changes for either the final or initial sound.
Conclusion
The point of all this discussion was to show (but was that necessary?) that my so-called ślokas are a mere awkard attempt at sanskrit poetry. They do not abide by all the codes of sanskrit composition. I did so because I write within the confines of my limited knowledge of the sanskrit language and literature. The next step, apart from further sanskrit study, is to write a poem that would be more respectful of the sanskrit literary forms, assuming I understand them correctly. This one would have to meet two requirements. First it should honour the median and final pauses or yatis in the 2nd and 4th pādas of a śloka. Second, it should abide by the laws of the land, where a yati occurs at the end of a pāda according to the strict respect for the permissible or recommended interruption of utterance flow, which happens ideally in case (A) or, as a second best, in case (B2ii), both above.
I hope I am not too mistaken.
ॐ नमः शिवाय
Monday, 21st January 2019
Jean-Michel LEON-FOUN-LIN
Notes
(1) The problem of the decision-making sequence deserves a separate, extended discussion. Whether the poet, in his composition process, should start with syntax, then sandhi and finally prosody or start withprosody, then syntax and finally sandhi, might sound like the question by someone with no purpose in life. It does not seem realistic or relevant to poetry to think that any set of words can be matched to any music to produce a poem. Nevertheless it is important to think it over because of the philosophy about poetry behind the choice.
In the first case, meaning seems to be a priority and prosody an afterthought whereas in the second case music seems to be the placeholder for words of empty chatting (like here). Poetry is probably a balance between meaning and music. You might object that this balance is usually called a song. And I could retort that in poetry the word is the music.
Let us consider the sentence “The boy adored the pretty girl he saw.” As the lyrics in a song, it can be performed with different rhythms and melodies. You can indeed play around with pauses and syllable lengths to achieve any rhythm and you can play with pitches to achieve any melody by choosing to sing any vowel in a syllable lower or higher. As a verse in a poem, this line has the fixed metre x_x_x_x_x (x is unstressed), which determines the rhythm (but does not equate to it) in association with the word distribution in the sentence (the rhythm would be different if all the words were monosyllabic). It is all the truer in Sanskrit where syllables are inherently long or short, heavy or light (guru or laghu). As for the variation in pitches, it exists in a poem too but it does not depend on performance choices. Rather, the difference comes from the inherent relative height of the different vowels, such as can be heard between an -o- (low) and an -i- (high), whether these two vowels are said by a tenor or a baritone. Think in terms of colours if you prefer, where a yellow colour may look lighter than a brown colour even if they come from surfaces with the same reflectance. And we are not even talking about tonal languages like Chinese or Yoruba (and I could not at this stage). So, if our sentence is considered as the line in a poem, its melody and its rhythm are induced by the words themselves. Whether our sentence actually conveys music by itself, with its inherent rhythm and melody, will qualify it as poetry or not, provided that, in addition, this music suits the meaning of the words or the mood of the poet (because, let’s face it, an iambic pentameter does not necessarily refer to pretty girls, unfortunately). It follows that any sentence can be turned into a song as long as you can distort the words so that they comply with the music. But not any sentence can claim to be a poem line if it does not contain any specific internal rhythm and melody which must also be expressive.
Hopefully we have demonstrated now that, really, in poetry the word is the music.
Of course, this is too superficial a discussion for such a serious matter. 50 years from now, when I may have understood what poetry really is, perhaps I will come back here and laugh at my own impudence.
(2) Another fundamental difference with other languages like, say, French is that these spoken changes must be written down because in Sanskrit 1 letter corresponds to 1 sound. It is true that the letters of the Sanskrit alphabet do not represent all the sounds of the language. These extra sounds are modified versions of the original ones already mapped by the alphabet. Signs which are not part of the alphabet, rather than combinations of existing letters, are used to indicate these modified sounds. The anusvāra and visarga are such signs which allow the representation of extra vowel sounds. They let the language avoid possibly ambiguous combinations of letters like [in], [an], [on] to write new vowels not found in the French alphabet. Conversely, in Sanskrit 1 sound is symbolised by 1 letter. Again, it is true that the sound -a- for instance may be represented by two signs. But these two signs are just two forms of the same letter which have the same position in the alphabet and do not increase the number of letters. The outcome of this bijective function is that spoken sandhi changes must be echoed onto the paper and must be so with unique transcriptions.