See FW, 123-4, 167, Van Valin (1987b) for discussion of the typological conse
quences of this contrast.
See FVV, sections 4.4, 4.5 and Foley & Van Valin (1985) for a detailed typologi
cal survey of passive (and antipassive) constructions.
All Icelandic examples are from Zaenen, Maling & Thráinsson (1985), Thráinsson
(1979), and Andrews (1982); see these references for detailed discussion and
exemplification of the pivot properties of the non-nominative "subjects".
The facts of Bantu applicative constructions are very complex, and a complete
treatment of them is beyond the scope of this discussion. That the "applied
object" is undergoer is shown by its taking on all (Kinyarwanda) or most
(Chichewa) of the properties of undergoers of underived transitive verbs.
For RRG analyses of idiosyncratic or "quirky" case in Icelandic and Latin, see
Van Valin (1990a) and Michaelis, this volume.
"Direct" here simply means "non-oblique", in the sense discussed in section 4.3.1.
Direct arguments are normally core arguments, except when they occur in the
PCS, a core-external but clause-internal position. Actors may be direct or oblique
(as in a passive construction), but undergoers are always direct arguments.
Oblique arguments may appear in the core, in the periphery, and in the PCS.
"Explicitly specified" means that the argument position in the LS is filled by a var
iable or a constant; for it to be unspecified, it would be filled by "0", as in (i).
(i) a. Max loaded the minivan.
b. [do'(Max)] CAUSE [BECOME be-in' (minivan, 0)]
b'. [do'(Max)] CAUSE [BECOME be-in' (minivan, the olives)]
The LS in (ib) does not specify what the theme argument is, unlike the one in
(ib'), and the theme is not specified in the sentence in (ia). Hence only (ib) is a
legitimate LS for (ia), because of the CC.
This is the default situation. Verbs like give, send, and hand are exceptions to this
rule, as their non-undergoer themes are not prepositionally marked.
In a syntactically ergative language like Dyirbal or Sama, these would be the
opposite: if active voice, then the pivot is the undergoer; if antipassive voice, then
the pivot is the actor.
It is important to distinguish between lexical and syntactic phenomena, on the one
hand, and lexical and syntactic rules, on the other; the two oppositions do not
necessarily map onto each other in a direct way. As the history of generative lin
guistics has amply demonstrated, all grammatical phenomena can be described by
means of syntactic rules, and virtually all can also be described by means of lexical
rules. LFG attempts to treat phenomena which other theories consider to be syn
tactic by means of lexical rules, while RelG treats phenomena which all other
theories handle with lexical rules syntactically, e.g. complex verb formation.
