A SYNOPSIS OF ROLE AND REFERENCE GRAMMAR 155
- All Acehnese examples are from Durie (1985, 1987). In the clause structure
schemata on the right, "NPA " = Actor NP, "NPy" = Undergoer NP, "A-" =
Actor proclitic, "-U" = Undergoer enclitic. - The undergoer clitic cannot occur if the verb is immediately followed by the
undergoer NP, unless the NP is marked by the focus particle di. - The passive translation does not indicate that the Acehnese construction is a pas
sive; it is not. Durie (1985, 1988) shows that this construction is an active sentence
with a postposed actor dokto "doctor"; note that the actor is still cross-referenced
on the verb in the usual way by a proclitic. Aneuk agam nyan "that boy" is the
actor of tern "want" and the undergoer of peuréksa "examine". - See Van Valin (1981), FVV, Foley & Van Valin (1985), Durie (1987), Andrews
(1985), among others, for detailed discussion. - "Syntagmatic" is used here as a general term for structural relations between ele
ments in a construction. It is neutral with respect to whether the relations are syn
tactic (grammatical) or semantic in nature. - The RRG analysis of these constructions will be given in sections 6 and 7; it does
not involve either deletion or movement. Cf. also FVV, section 6.5. - An example of a non-subject pivot in English is the obligatorily omitted argument
in a purpose clause like He brought that book (for his sister) to read (*it). The
obligatorily omitted argument is the pivot, and it is postverbal, not preverbal;
therefore it cannot be the subject of the dependent core. See Cutrer (1987, this
volume) for detailed discussion of this construction. - This is something of an idealization, as in many languages direct core arguments
which are not undergoers may become pivot in a passive; cf. section 4.5, Van
Valin (1991b) for a detailed examination of this situation in Icelandic. This ideali
zation does not affect the larger point at hand. - Barai, another Papuan language, and Eastern Pomo, a Hokan language of
California, are exceptions to this generalization (for different reasons). See Olson
(1978, 1981) and McLendon (1978) for detailed discussion. - For extensive discussion and cross-linguistic exemplification of this distinction, see
FVV, chap. 4, Foley & Van Valin (1985), Van Valin (1981, 1987b). - There are two ways in which focus structure can be grammaticalized in clause
structure. The first is in relational clause structure, yielding [+pragmatic influ
ence] pivots, and the second is the grammaticalization of the predicate focus pat
tern into non-relational structure, yielding a VP category (cf. section 2.5). These
two possibilities yield a typology of four potential language types with respect to
this grammaticalization: (1) none at all, e.g. Lakhota, Warlpiri; (2) relational
structure only, e.g. Dyirbal; (3) non-relational structure only, e.g. Nanai, Ulcha
(Manchu-Tungus); and (4) both, e.g. English, Malagasy.