A SYNOPSIS OF ROLE AND REFERENCE GRAMMAR 45with a verb which takes agent/effector and theme or patient arguments, the
agent/effector will always be actor. Transitive verbs which take experiencer
and theme arguments, e.g. see, or locative and theme arguments, e.g. have,
always have the experiencer or locative as actor, never the theme.^26 With
regard to the undergoer end of the hierarchy, the markedness relations are
reversed from the actor end. If a verb takes locative or experiencer and
theme arguments, e.g. ν and ζ with show and give above, the theme is the
unmarked choice for undergoer, even though these verbs allow both pos
sibilities. Evidence for this comes from a variety of sources. First, there are
well-known arguments for the markedness of the locative/experiencer as
undergoer form of the dative-shift construction in English, e.g. the restric
tions on pronominal arguments in this form. Second, nominalizations pro
vide evidence for the priority of theme over locative or experiencer: a
bookshower would have to be someone who shows books to people and a
studentshower could not be someone who shows things to students but
rather who shows students to people; the same interpretations are found
with flowergiver and girlgiver, which cannot mean someone who habitually
gives things to girls.^27 Third, in the survey of dative-shift constructions in a
wide range of languages presented in Foley & Van Valin (1985), the theme-
as-undergoer form is clearly the unmarked form in virtually all of the lan
guages; in most languages other than English, the base form of a verb of the
appropriate class can take only theme as undergoer, and a derivational
morpheme of some kind must be added in those languages which allow
locative or experiencer as undergoer. It is very significant that most lan
guages do not have dative-shift constructions at all; in them, the choice of
undergoer is lexically governed, and the vast majority of the relevant verbs
in these languages are like English put, donate and announce in having
theme as undergoer, not locative or experiencer. Thus with respect to the
undergoer end of the hierarchy, theme outranks both locative and experi
encer.
The hierarchy in (25) represents thematic relations along the con
tinuum presented in Figure 15. One additional distinction which is made in
FVV is the combination thematic relation of effector-theme, which corre
sponds to argument instrumentais in other theories. It arises in cases in
which the theme in a complex causal event plays a causal role, as rep
resented by its occurrence in the first argument of the primary CAUSE in the
LS, as with e.g. cut ([[do'(x)] CAUSE [BECOME be-at'(y,z)]] CAUSE [BE
COME cut'(y)]; cf. FW, section 2.6) as in Max cut the cake with a knife.