Fig 1.
Expt 1: Proportion of continuations about the subject.
Fig 2.
Expt 1: Rate of pronominalization, by Verb Type and Re-mentioned NP.
Fig 3.
Expt 1: Item-by-item quantitative model evaluation for Experiment 1.
Table 1.
Expt 1: R-squared, MSE, and ACE of the three models.
Fig 4.
Expt 2: Proportion of continuations about the logical object.
Dashed lines illustrate the NP1 next-mention rate that would be predicted for passives under an alternative model that is identical to the Bayesian model except that the prior is replaced with the active-voice prior.
Fig 5.
Expt 2: Proportion of continuations by coherence relations in Free Prompt condition.
Fig 6.
Expt 2: Next-mention biases in explanation and result coherence relations by Verb Type and by syntactic construction.
Fig 7.
Expt 2: Coherence relations by Verb Type, syntactic construction, and next mention.
Fig 8.
Expt 2: Rate of pronominalization the the free prompt data (Collapsing null and overt pronouns).
Fig 9.
Expt 2: Proportion of continuations about the syntactic subject by Verb Type and Prompt Type.
The data in this figure are identical to those in Fig 4, but are plotted here in terms of proportion of continuations to NP1, to facilitate visualization of the effect of the likelihood.
Fig 10.
Expt 2: Item-by-item quantitative model evaluation for Experiment 2.
Table 2.
Expt 2: R-squared, MSE, and ACE of the three models.
Fig 11.
Expt 3: Proportion of continuations about the subject, by Verb Type, Prompt Type and syntactic construction.
Fig 12.
Expt 3: Distribution of coherence relations in free Prompt condition.
Fig 13.
Expt 3: Next-mention biases in explanation and result coherence relations by Verb Type and by syntactic construction.
Fig 14.
Expt 3: Rate of pronominalization, by Verb Type, Re-mentioned NP and syntactic construction.
Fig 15.
Expt 3: Item-by-item quantitative model evaluation for Experiment 3.
Table 3.
Expt 3: R-squared, MSE, and ACE of the three models.