There was evidence of a complex interaction between world knowledge, markedness, order of mention and syntax when selecting an antecedent, in an antecedent choice task. In Chapter 3, we looked at what information is used in off-line referent choice. These results combined suggest that the constraint-based model should be modified in such a way that it can incorporate some kind of “task-engagement” parameter.
![bryce hirschberg look alike actor bryce hirschberg look alike actor](https://hollywoodzam.com/wp-content/uploads/2021/08/Bryce-Hirschberg-1.jpg)
There was no difference between ambiguous and unambiguous pronouns in the task condition that required no disambiguation.
![bryce hirschberg look alike actor bryce hirschberg look alike actor](https://i.pinimg.com/originals/13/46/87/1346878ec837423e57543312ce378b34.jpg)
Specifically, there was evidence of ambiguous items taking longer than unambiguous ones at the pronoun in the Antecedent Selection and the Plausibility Rating condition. In Chapter 2, the results from the self-paced reading experiments suggested that in the comprehension of pronouns all sources of information are used in parallel, since the constraint-based model best described processing in the cases where participants made an effort to interpret the stories as soon as possible. This was addressed by using a diverse set of psycholinguistic response measures (self-paced reading, antecedent selection and ERP (Event-Related brain Potentials) registration). One of the key issues in this type of investigation is how multiple sources of information are combined by the language processor. This thesis looked at anaphora resolution, concentrating in particular on the influence of markedness of anaphoric expressions (in terms of prosodic and/or morphological realization) and plausibility of the antecedents (based on world knowledge). In contrast, when one of the two antecedents was a much more plausible antecedent than the other, antecedent choice was almost wholly dictated by plausibility, although reference form prominence did significantly attenuate the strength of the preference. When the pronouns were marked due to accentuation (English) or use of specific morphology (Spanish), preference switched to the alternative antecedent, the grammatical subject of the previous clause. Listeners generally resolved toward the (parallel) grammatical object of the previous clause. We looked at sentences where both antecedents were equally plausible, or where only one of the antecedents was plausible. We manipulated the form of the anaphoric expression in terms of accentuation (English: Experiments 1a and 2a) and morphology (Spanish: Experiments 1b and 2b). We investigated how listeners resolved object pronouns that were ambiguous in the sense of having more than one possible antecedent.
![bryce hirschberg look alike actor bryce hirschberg look alike actor](http://i2.cdn.turner.com/cnnnext/dam/assets/150701105239-michael-b-jordan-nick-cannon-split-super-169.jpg)
In this paper, we argue that of these factors, as tested here, plausibility is the most important, but interacts with form markedness and structural parallelism. Previous research on pronoun resolution has identified several individual factors that are deemed to be important for resolving reference.