To quote this page: AAE Van der Geer 1998. The Bhasa Problem. A statistical research into its solution. PhD Rijksuniversiteit Leiden, the Netherlands

Back << >> Further

Back to my PhD

Combining The Methods

Taking all data together as they are would yield many groups each with a very limited number of elements, since only some data coincide. It is certainly not my aim to classify two plays each into a separate group on the basis of one single characteristic, especially if this characteristic occurs only once. It is therefore much more logical to weigh the results according to their relative importance and their discrimination ability. The resulting groups then will be a weighed average of the three tests.

I consider the outcome of the Coincidence Test the most important and the most discriminative, not only because its results are confirmed with the help of reference authors and by Pollock's (1977) results, but also because this test yields several characteristics, as there are overall percentages and types of pattern. In addition, this test can be applied to all metrical texts, independent of type of metre used, and to all languages, at least Indo-European.

Second in importance is the shloka pattern as found in the texts. Vipulâ distributions have already been used with success by earlier scholars, so it seems legitimate to rely on the data as obtained with this test. However, since the number of vipulâs is very low, due to the smallness of the works concerned, it would be too optimistic to give the outcome of this test priority. Therefore, I regard the shloka data as only additional.

And finally, third in importance are the Word Frequencies. These data are even less important, as the texts are too small to rely on the discriminators of Trautmann (1971). Of the remaining words, only a few words appear to be useful in discriminating works, such as the particles iva and na. In addition, some nouns can be used to indicate a non-common origin.

Now, starting with the groups according to the Coincidence Test, we get the following sequence. The play that remains in an uncertain position is the Pancarâtra with a pattern that is compatible to the first group, but with percentage between the first and the second group. The Dûtavâkya and the Bâlacarita both have a percentage according to the third group, but their rather similar patterns are very marked with a special prevalence for two forms only. Therefore, they can be traced back to the seventh or eighth century, as they must be post-classical according to their percentage, but pre-Madhyama with its still more pronounced pattern.

The following step is including the data from the shloka patterns. Excluding the three plays that appeared difficult to classify, namely the Karnabhâra and the two embassy plays, and placing the Pancarâtra and the Madhyama together, we get the following four groups. First, no preference (pattern 0:0:0): Cârudatta, Svapnavâsavadatta 1-5. Second, a low preference: Pratijnâ (0:2:1), Pratimâ (0:1.7:1), Avimâraka (0:1:1), and Bâlacarita (0:1:2). Third, again a low preference (pattern 0:4:1): Pancarâtra, and Madhyama. And fourth, a high preference (different patterns): Svapna-6, Ûrubhanga, and Abhisheka.

These figures fit very well with those of the Coincidence Test. The oldest group has a low or even no preference at all for the vipulâs, if we exclude the last act of the Svapnavâsavadatta. If we do not exclude this act, the Svapnavâsavadatta as a whole shows a high preference and must be classified along with a much later group, which is not very likely.

The patterns of the Pratimâ and the Avimâraka are very much alike, and are not very much removed from that of the Pratijnâ. The Bâlacarita might be close to this same group according to its preference, although its pattern differs. The same does apply to Pancarâtra and the Madhyama, both with an equal pattern and the same low preference for the vipulâ. It is remarkable that the same plays which are difficult to classify with the data of the Coincidence Test remain difficult to classify with the data of the shloka pattern test. The Pancarâtra has a low preference that fits to the oldest plays. The same does apply to the Madhyama, which in addition has a very marked Coincidence pattern. If we combine these data, both plays must originate from a rather late century, although they both share some archaic features. The Bâlacarita belongs to the older groups regarding its percentage, but its pattern is not exactly the same. Therefore, this play remains in an uncertain position.

The embassy plays occupy a middle position as to their preference. Their patterns, however, are very different. The pattern of the Dûtavâkya is most likely to set this play on a par with the Pancarâtra and the Madhyama. This is confirmed by its Coincidence Pattern, which inclines towards the extremely pronounced pattern of the Madhyama. Its Coincidence Percentage is according to the Ûrubhanga group. Therefore, it seems legitimate to classify this play after the 5th or 6th century, in the same century as the Pancarâtra. The Dûtaghatotkaca has an unique pattern, and cannot yet be grouped with one of the other plays.

Finally, the data as obtained with the distributions of the particles iva and na can be brought into the discussion. Simplifying the results by including the plays that occupy a middle position into the most likely groups, and non-splitting the Cârudatta, we get three groups. The first is the low-high percentage group, consisting of the Pratijnâ, Svapnavâsavadatta, Cârudatta, Pancarâtra, and the Dûtaghatotkaca. The second group is the low-low percentage group with five members, the Karnabhâra, Ûrubhanga, Dûtavâkya, Abhisheka, and the Bâlacarita. And the third group - the high-low percentage group - finally consists of the Pratimâ, Madhyama, and the Avimâraka.

Obviously, the low-high percentage group fits very well into the oldest group according to the Coincidence Test. Only the embassy play disturbs the picture, although according to the shloka Pattern Test it is close to the Pancarâtra, the play which seems to be old, but has marked patterns indicating a later age. This may then also apply to the embassy play, which fact then accounts for its belonging to the old group regarding the particle distribution. A more modern provenance is indicated by the occurrence of desideratives, and the Coincidence pattern and percentage.

The low-low percentage group coincides perfectly with the 5th-6th century group of the Coincidence Test. The embassy play and the Bâlacarita could not be classified with certainty, but seemed to belong to this same group according to their Coincidence Percentage. Their pattern, however, is of a more reduced type. The particle distribution confirms their belonging to this group or to be more precise, to the end of its period. This is confirmed also by the shloka pattern, according to which the embassy play belongs to the later centuries. The shloka characteristics of the Bâlacarita appeared not conclusive.

The high-low percentage group contains the 4th-5th century group of the Coincidence Test, plus the Madhyama, the play with the remarkable patterns which indicate a younger age. This younger age, however, is not confirmed by the outcome of the particle distribution. Why then is it found on the same plan as this early classical group, which certainly is much older? The only difference considering the distribution of all words between the Pratimâ and the Avimâraka on the one hand, and the Madhyama on the other hand is the occurrence of desideratives in the former two, and an intensive in the Avimâraka. However, the lack of desideratives is not only seen in the oldest group, but also in the latest plays, so perhaps the occurrence of a desiderative is not very helpful in analysing a play.

Finally, Schokker's list (1966:330-351 (appendix IV)) confirms the following three observations. Firstly, the last act of the Svapnavâsavadatta indeed may have been added in a later time. The vipulâ pattern, the na-distribution, and the occurrence of the early classical combination api nâma, are indicative of the later addition of this act. The Coincidence Test cannot be used to confirm this, as not even a single shârdûlavikrîdita-stanza occurs in this act, although this fact is in itself remarkable enough. Secondly, the Ûrubhanga on the one hand and the Karnabhâra and the Abhisheka on the other hand are written after the fourth century as the first contains an early classical word, and the latter two each a Kâlidâsan word. And finally, the greater part of the Trivandrum plays can indeed be dated before the seventh century, as no Bâna words occur.

All evidence taken together yields the following chronological table,