Kiel University, Kiel, Germany, 1996
In this study, control of various grammatical operations such as coreferential deletion, reflexivization, coordination, subordination, pronominalization and relativization are studied in the language of one small part of the Pali canon, i.e., three chapters (78 pages) of the Mahavagga of the Vinaya Pitaka, thereby limiting the study as much as possible to the language of one particular time and place and avoiding later influences. The basis of the study are the syntactic-semantic primitives A (transitive agent), S (intransitive subject) and O (transitive patient), as well as experiencer/experiencee. Also, Dixon's (1979, 1994) and Foley/Van Valin's (1985) somewhat differing concepts of 'pivot' are employed. It was found that all verbal categories, with the exception of the finite passive, treat A, S and O the same, irrespective of their case. The results can be summarized as follows:
- Reflexivization and equi-NP deletion operate on a SEMANTIC
agentive S and A);
- Pronominalization is triggered by the TOPIC (whether this be a core or non-core noun);
- Subordination involoving clauses in the locativus absolutus construction operates on a purely SYNTACTIC PIVOT (i.e., S/A);
- Main clause coordination operates along a PRAGMATIC PIVOT (core NP as topic occupying clause initial position), in which S and A may only delete a coreferential S or A, and O only a coreferential O, regardless of case;
- Subordination involving a verbum absolutivum operates along a SYNTACTIC PIVOT (S/A) which is restricted to the PRAGMATIC PIVOT (core NP in clause-initial position of the clause it appears in).
Also, it was found that word order in the Pali of our text sample is of grammatical significance, as it plays the decisive role in clause combining.
The results also call the theory of the development of the modern-day ergative constructions from an originally passive one into question, as it was found that the transitive patient controls NONE of the grammatical operations under study which it should be expected to do in a passive construction. Also, as the periphrastic perfect in Pali marks S and O in the nominative and differently from A, this is already an ergative pattern in Pali, with the exception that the nominative is not unmarked morphologically, i.e., it is not identical with the absolutive. Hence, the periphrastic perfect is considered an "ergative-like" construction.
Concerning the development of the periphrastic perfect: The PPP (and the gerundive) is considered in this study to have originally been a resultative verbal adjective, not (necessarily) passive. On the basis of this, and using the theory of grammaticalization as this applies to the respective verbal categories, a different development for the periphrastic perfect is given, one in which the development of the perfect of transitive verbs in many ways parallels that of the languages of western Europe. The functions assumed in these languages by the formally transitive verb 'have', i.e., above all possession and concomitance, are expressed by nominal phrases in Indo-Aryan. These two constructions directly influenced the syntax of the perfect of transitive verbs in Indo-Aryan. In the first construction, the PPP agrees with O in case, number and gender and A appears in the genitive, the same construction used to denote possession. In the second, more common, pattern, A appears in the instrumental, which was originally used alone to denote concomitance. In the perfect of intransitive verbs, S appears in nominative and the PPP agrees with it in gender, number and case.
As of February / March, 1997 copies may be obtained from:
Dekanat der Philosophischen Fakultaet
Christian-Albrechts-Universitaet zu Kiel