Scopal ambiguity preferences in German negated clauses

Scopal ambiguity preferences in German negated clauses Barbara Hemforth (barbara.hemforth@lpl.univ-aix.fr) Laboratoire Parole et Langage, UMR 6057, Univ. de Aix en Provence 29, av. Robert Schuman, 13621 Aix en Provence, France Lars Konieczny (lars@cognition.uni-freiburg.de) Center for Cognitive Science, IIG, Univ. Freiburg Friedrichstr. 50, 79098 Freiburg Abstract When following a negated matrix clause, adverbial clauses (ACs) like “because it was paid very well” in (1) can be interpreted as residing within the scope of the negation (1b), or outside of it (1). IP AC neg (1) a. Peter did not quit his job because it was paid very well b. Peter did not quit his job because it was paid very badly. VP because it was ... Depending on the scope of the negative, the interpretation differs dramatically: Whereas Peter did in fact not quit his job in (1a), he did so in (1b), but for yet unknown reasons. It has been shown for English (see Frazier & Clifton, 1996), that there is a preference to interpret the adverbial clause outside of the scope of the negation so that (1b) appears fairly odd. This observation challenges recency based processing principles, such as late closure, since the high attachment (to IP) appears to be preferred over low attachment (to VP) (see figure 1). In this paper, we will present evidence on German equivalents of (1a,b), varying the order of the negative and the verb (Experiment I), the context in which the ambiguity appears (Experiment II), and the position of the adverbial in relation to the clause boundary where the negation of the main verb is restricted or even retracted (Experiment III). None of these variations reduced the preference substantially. Only an explicit alternative cause reduced it but even this variation did not eliminate the difficulty of the inside scope interpretation. We will argue that incremental interpretation as well as immediate attribution of prosodic structure determine the interpretation of the adverbial clause. Figure 1 However, Frazier and Clifton (1996) found a clear preference for an interpretation of the AC outside of the negation scope. In their experiments, this preference does not show up in early stages of processing, but only in later off-line measures. In the framework of Construal Theory, the authors argue that only argument like or primary relations are attached to the phrase marker of the sentence immediately. Only primary relations are subject to syntactic attachment principles like Minimal Attachment or Recency. As a non-primary relation, the adverbial clause is only construed as part of the maximal projection of the preceding thematic domain (i.e. the IP). All kinds of factors (syntactic, semantic, pragmatic, or prosodic) jointly determine the final interpretation of the clause. In this paper, we do not want to dispute the immediacy of the attachment or the interpretation of adverbial clauses. We are much more interested in the question of which factors are driving this final preference, and how general or universal it is. This means testing this preference in different constructions, in different environments, and across languages. Therefore, we have been looking at German versions of Frazier and Clifton’s materials. In four experiments, we tried to find a way to override the interpretational preference, by putting the negation in focus position (Exp I), varying contextually given presuppositions (Exp II), and restricting the negation (Exp III). Since our main interest does not lie in the question of when these factors come into play, but only whether they do have an effect at all, in all of our experiments, we applied an off-line acceptability judgment task. Introduction Whereas the scope of quantifiers has been the subject of substantial research (e.g., Ioup, 1975, Johnson-Laird, 1969; Kurtzman & MacDonald, 1996; for an extensive discussion see Frazier, 1999) this is much less the case for the scope of negations. In this paper, we will look at sentences like (1) where an adverbial clause can be interpreted as either being within the scope of the negation in the matrix clause or outside of it. Structurally, the adverbial clause has to be attached to the VP if it is interpreted as residing within the scope of the negation whereas it has to be attached to IP if it is interpreted outside the scope of the negation. Locality based principles of syntactic attachment as they are assumed in most theories of human sentence processing (e.g., Frazier, 1978; Gibson, 1991) predict a preference to attach the adverbial clause to VP, and thus a preference to interpret it inside of the scope of the negation. Experiment I In Experiment I, we wanted to test whether the preference for high attachment of the adverbial clause which has