An Experimental Investigation of the Semantics of Attitude Reports
|Event Name||An Experimental Investigation of the Semantics of Attitude Reports|
|Start Date||7th Mar 2014 3:00pm|
|End Date||7th Mar 2014 5:00pm|
This talk presents experimental evidence bearing on the analysis of a particular type of attitude known as an attitude ‘de se’. A de se attitude is one that is interpreted from the first personal perspective of the subject of the attitude verb. A pronoun that takes the subject of the attitude verb as its antecedent is typically (though not necessarily) interpreted de se: Doni thinks that hei is successful can be used to report a belief that Don would express using the first person pronoun (‘I am successful’).
We present a series of experiments testing the proposal that de se pronouns undergo covert movement to the left periphery of the embedded clause (Percus and Sauerland 2003).Percus and Sauerland note that this theory predicts that the available interpretations of pronouns in attitude reports should be constrained by syntactic conditions such as superiority: in a configuration where one pronoun ccommands the other, the lower pronoun cannot be construed de se, since movement across the higher pronoun would constitute a superiority violation
As evidence that this prediction is born out, Percus and Sauerland consider reports of scenarios where someone dreams that she is someone else, as in Carol dreamt that she was Sandra, and she was buying her book. They claim that this sentence cannot be used to report a dream where Carol buys Sandra’s book; since this reading would require the lower pronoun to be interpreted de se, this is taken as evidence for the movement-based theory. Yet it is difficult to introspect about the scenario and the sentence in question, and the judgments are very delicate
Our study provides evidence that the reading in question is indeed unavailable with dream, although it is available with believe – a fact first noted in (Anand 2006). We present a proposal concerning why we find this difference with these two verbs, and discuss some general consequences for the use of experimental methods in the collection of semantic judgments.