A SYNOPSIS OF ROLE AND REFERENCE GRAMMAR 67(48) a. Lögreglan tók Siggu fasta.
the.police(N) took.3pL Sigga(A) fast(A)
"The police arrested Sigga."
b. Sigga var tekin föst af lögreglunni.
Sigga(Nf) was.3sG taken fast(N) by the.police(ö)
"Sigga was arrested by the police."Many apparently transitive verbs in Icelandic have "direct objects" which
are in the dative or genitive case, not the accusative, and as argued in Van
Valin (1991b), the undergoer of a transitive verb in Icelandic always
receives accusative case in active voice (simple) sentences. From this it fol
lows that none of the postverbal dative NPs in (49) are undergoers.
(49) a. Ég skiladi henni peningunum.
I(N) returned.ISG her(ö) the.money(D)
"I returned her the money."
b. Ég hjálpadi honum.
I(N) helped.ISG him(ö)
"I helped him."
Nevertheless, such a dative NP may function as a true syntactic pivot in a
passive construction; in this way Icelandic differs from German, in which
"dative subjects" in passives do not take on the behavioral properties of
nominative pivots.^42
(50) a. Henni var skilad peningunum.
Her(ö) was(iMPER) returned the.money(D)
"She was returned the money."
a'. * Peningunum hefur vend skilad henni.
the.money(ö) has(iMPER) been returned her(ö)
"The money was returned to her."
b. Honum var hjálpad (af mér).
He(ö) was(iMPER) helped (by me(ö))
"He was helped (by me)."
Since these verbs do not have undergoers, the correct statement of the
linking in the passive construction in Icelandic cannot be "U = PrP";
rather, it must be "~A = PrP". The characterization of the non-actor argu
ment is an important question, as the ungrammaticality of (50a') shows;
with a verb like skila "return, give back" which takes two non-actor argu-