Verbal Valency and Three-Place Verbs in Ainu

Ainu clearly belongs to the language type of head-marking languages according to Nichols(1986) framework. Nichols refers to subject, direct object, and indirect object as the most important factors in verbal coding in head-marking languages. Other things being equal, valency in Ainu is also supposed to be determined by this verbal coding. However, in Ainu, only subject and direct object can be encoded in a verb: indirect object cannot be marked. So it appears the maximal verbal valency in Ainu automatically amounts to two. In fact, this is not the case. Since nominal derivational prefixes and incorporated nominal bases may also be coded in a verb occupying the position for indirect object, we should extend the notion of verbal coding in head-marking languages by Nichols to include such indirect nominal elements in considering Ainu verbal valency. For the moment we name this special notion of a revised version of the valency for head-marking languages such as Ainu as “argument slot”. If we follow the notion of argument slot, the maximal valency of Ainu is supposed to be three, not two. It should be noted that “argument slot” is here a verb-internal abstract property just as gender for a noun and that it does not mean any surface positions in verbal morphology. And we have to assume that there are two further distinct frameworks for verbal valency in Ainu. First, we assume that the person marking is very important and that it corresponds to the concept of M-transitivity advocated by Role and Reference Grammar by Van Valin. According to M-transitivity, the maximal verbal valency of Ainu is not three, but two, namely, Actor and Undergoer. Furthermore, since verbal valency can be reduced by derivation and incorporation, the verbal valency on a surface level may not always match the argument slot valency or the M-transitivity valency. Therefore, in addition to these two valencies, it is necessary to recognize “morphological verbal valency”. Such a valency framework with three different verbal valencies, though very complex, is very useful for explaining examples involving an exceptional valency shift in three place verbs in Ainu. Chiri (1974[1936]) cites an example of wakka-ta-re-an ‘we make someone dip water’, which should take subject, causee, and theme. Since ‘to make someone dip something’ incorporates wakka ‘water’, it should appear as a two-place verb. However, it is actually realized as a one-place verb. Such irregularity in valency-shift can be explained by simultaneously considering three different verbal valencies, such as argument slot, M-transitivity, and morphological verbal valency. The verb ta-re is a three-place verb according to argument slot. However, the action ‘to make someone dip water’ is common in Ainu culture, and it is likely that usually there is little need to encode the causee. So it would be better to consider that the interpretation by argument slot is modified and that the valency is reduced to two for the pragmatic reason. Ta-re is then interpreted as a two-place verb and is inflected by M-transitivity. Then that is further converted into a one-place verb by the incorporation of wakka ‘water’. In addition, even if it is converted into a one-place verb, the nature of a three-place verb by argument slot is supposed to remain, so the meaning of causee still exists with pragmatic implications. Another similar case is iyomare-an ‘we pour liquor for someone’, which is also an exception to valency-shift for a pragmatic reason and seems to support the analysis proposed here.