| |
Abstract:
It has been argued that the preference to attach relative
clauses high in German N1 N2 RC constructions is mainly due to a
preference to attach the relative clause to a noun phrase whose
case matches with that of the relative pronoun (e.g., Sauerland
& Gibson, 1998). Most experiments published for German used
complex NPs in subject position with a nominative marked head noun
and subject relative clauses with a nominative marked relative
pronoun. So this looks like a viable alternative to solutions
proposed by, e.g., Hemforth, Konieczny, & Scheepers (in press).
If no case matching principle can be applied, recency is assumed to
determine the attachment preference, explaining why PPs in
comparable constructions are preferentially attached low.
We conducted to questionnaire studies with materials like (1) to
(4) with complex NPs either in subject position or in object
position and with subject or object relative clauses.
(1) Der Sohn des Lehrers, der den Arzt besuchte, traf
einen Freund.
"The son-NOM (of) the teacher-GEN who-NOM visited the doctor met a
friend."
(2) Der Sohn des Lehrers, den der Arzt besuchte, traf einen
Freund.
"The son-NOM (of ) the teacher-GEN who-ACC the doctor visited met
a friend."
(3) Ein Freund traf den Sohn des Lehrers, der den Arzt besuchte.
"A friend met the son-ACC (of) the teacher-GEN who-NOM visited the
doctor."
(4) Ein Freund traf den Sohn des Lehrers, den der Arzt besuchte.
"A friend met the son-ACC (of) the teacher-GEN who-ACC the doctor
visited."
According to the case matching principle there should be a
preference to attach the relative clause high in (1), where the
head noun of the complex NP and the relative pronoun are nominative
marked, and in (4) where both are accusative marked. Case matching
does not apply in (2) or (3) where none of the potential hosts
matches in case with the relative pronoun (the "low" NP is always
genitive marked). Therefore the RCs should preferentially be
attached low in these case.
The "syntactic/anaphoric-dualism" by Hemforth, Konieczny, and
Scheepers predicts a preference for high attachment across the
board for German NP1 NP2 Modifier-constructions whenever an
anaphoric component plays a role in the attachment process -- and
that is exactly what the questionnaire studies show.
|
