Patrick Sturt
University of Glasgow
Anaphor resolution is affected by discourse and semantics, but is also constrained by syntax, via binding theory. Thus, it offers a useful way of looking at interfaces between these levels.
Here we report two eyetracking experiments which examined the time course of the application of binding constraints during reading. The experiments aimed to determine whether binding constraints constrain the earliest stages of reference processing. The first experiment examined discourses such as the following:
1. Jonathan/Jennifer was pretty worried at the City Hospital. He/She remembered that the surgeon had pricked himself/herself with a used syringe needle. There should be an investigation soon.
In terms of binding theory, the reflexive anaphor in the second sentence can only refer to "the surgeon", and not to "Jonathan/Jennifer". Therefore, it is expected that "herself" will be relatively hard to process, because of the mismatch with the stereotypical gender of "surgeon" (stereotypical gender was used to avoid presenting ungrammatical sentences). The relative time-course of binding constraints can be determined by examining the point in processing at which this mismatch effect is found, in relation to any effect of the gender of the previously mentioned character (Jonathon/Jennifer), which is outside the binding domain of the anaphor, but which is highly focused in the discourse. The results show a reliable effect of the stereotypical gender of the local noun phrase (surgeon) on the first fixation time on the anaphor, suggesting that the configurational constraints of binding theory are operative from the very earliest stages of word identification. There was no trace of an influence of the binding-inaccessible noun phrase (Jonathan/Jennifer) in the first fixation data, but this did have an effect in the second-pass measure. This suggests that binding theory is operative from a very early stage, but that it does not necessarily rule out from consideration non-binding compatible antecedents at later stages of processing (cf. Badecker & Straub, 2001).
In order to rule out the possibility that the first fixation effect was due to lexical priming of the anaphor by the recently processed "surgeon", a second experiment was designed in which the linear positions of the first and second mentioned characters were reversed in the second sentence, but where their accessibility with regard to binding theory was kept constant:
2. Jonathan/Jennifer was pretty worried at the City Hospital. The surgeon who treated Jonathan/Jennifer had pricked himself/herself with a used syringe needle. There should be an investigation soon.
This experiment replicated the first fixation effect of the stereotypical gender of "surgeon" on the anaphor, with no reliable influence of the inaccessible antecedent in this measure or second-pass reading times. Note that the inaccessible character was less prominent in Experiment 2 than Experiment 1 (e.g., it is not a subject and does not c-command the anaphor).