ETHNOLOCIY OF TKK fNDO-PACiriC 18I.JLM>3. 



13 



In some dialects the former are pro<luced by the union of roots 

 of llie let and 2nfl perpons. The liUter is represented by the dual 

 forma only. In the Malay n-Polynesian languajres the two pluraU 

 and also the ilttiil arc founds find ai:^ they are not now Malagasy, 

 allhou>;b ionntl in Semiiicn-Alricfin hmgiiages, they may be 

 Dravirien traits. In some languages the dual and relative plum! 

 are not distinguished. 



The geneml charact<»r of the most ancient Asonosian pronominal 

 system — as preserved in various degrees in the Australian languages, 

 in Tai-awnn, Vitism, Tanan, in Polynesian and in some of the less 

 ini[iovenshcd Indonesian htnguages— is similar to the Dravirian, 

 but it is more archaic, more complete and less concreted. The 

 different elements are more numerous and more freely and regu- 

 larly comhinable. In the Austraiian system we find not only all 

 the forms that are now extant in South Dravirian, as well as the 

 dual and the peculiar transition or agen to-objective forms of 

 Kol, but several others produced by the same power of compoun- 

 ding dements in which these originated. This power is much tepg 

 im pared in Austra!i:in and the id lied Asoneslan systems, and the 

 inference is that tn this, as in several other respects, they better 

 preserve the archaic ludo-Asonesian type, and may hence 

 snggest to us what the condilion of Dravirian itself was before its 

 forms had b<;conie dimitiished, confused and concreted as we now 

 lind them. In Australian the pronominal roots are compounded 

 with definitives, singular and plural, wiih the numeral '*two*' to 

 form duals, with mase. and feni, definitives in the 3rd person, and 

 in all the 3 j^ei'soiiS wiih each other, thus producing not only abso- 

 lute and relative plurals of the 1st person, but sevei-al other com- 

 plex plurab. The Vitt-Tarawan elements are still more freely 

 compounded and their forms of this kind are consetjuently more 

 numerous. The incorporation of numerals appears not to have 

 been confined to " two," for in some of the Papua nesian languages 

 a (rmal is foujid, and in Palynesian the same form has lost its 

 original meaning and become a generic plural. This highly 

 a«iglomerative hut crnde pronominal system has not been derived 

 fioni Mulygasiy, aful its pit-scnce in Asoncsia is attributable to a 

 prior formation^ of Indian origin, similar to the Dravirian but 



