The role of affect and sentence processing is an understudied topic. In an event-related potential (ERP) language experi-ment, we investigated modulation of the P300 ERP component by dispositional affect. Using our previous ERP paradigm,we employed a 3x2 design where 32 participants read sentences presented in 1- and 2-word chunks (Berent et al., 2005;Patson & Warren, 2010). Sentences started with subject nouns that were either universally quantified or not, and continuedwith a direct object which was either indefinite, definite singular, or plural e.g., (i) Every kid climbed a tree/the tree/thetrees vs. (ii) The kid climbed a tree/the tree/the trees. Number judgments were required at tree(s), which was always pre-sented alone (and never final). Reduced P300 amplitudes were observed for the plural condition indicating interference;furthermore, low positive affect individuals showed responses sensitive to local high probability features associated withthe control singular condition.

In an event-related potential (ERP) language experiment, we investigated whether variability in the P600 component, amarker of syntactic anomaly, could be accounted for by dispositional affect. Sentences such as (i) The broker plannedto conceal the transaction * was sent to jail vs. (ii) The broker persuaded * to conceal the transaction was sent tojail (critical words are underlined) were read by 25 participants. These stimuli were adapted from Osterhout & Holcomb(1992), an influential early study on the P600 waveform. We expected to replicate previous findings, where P600 effectswere expected at to in (ii) vs. (i) and at was in (i) vs. (ii). The P600 effect at to did not replicate, whereas it didat was . Regarding affect, our results showed a significant positive correlation between positive affect scores and P600amplitude. Results are discussed in terms of the family of P600 components and affect.

We investigated the relationship between dispositional affect and the interpretation of sentences containing modal auxiliaries (must, should, might, would), that were embedded in factual vs. non-factual contexts. 49 participants read 2-sentence discourses followed by superficial questions, e.g., The art collector is admiring the statue. It would cost thousands of dollars. Question: The art collector is appreciating the statue. 1) True 2) False A negative correlation (r= -.355) was found for question-response accuracy scores and positive affect for must, should sentences vs. might, would. Thus, low positive affect individuals were more accurate after reading sentences with must, should vs. might, would. Questions were always about the context sentence; less attention at the second sentence would allow for higher accuracy rates. Given that must, should do not require restrictors for interpretation (Stump, 1985) whereas might, would do, low positive affect individuals paid less attention to these sentence types allowing for higher accuracy.

We used event-related brain potentials (ERPs) to investigate noisy channel models of sentences containing errors of deletion vs. errors of insertion. Ditransitives e.g., (i) The aunt mailed the letter to her niece by post were compared to deletion conditions (ii) #The aunt mailed the letter_ her niece by post. Furthermore, double object constructions (iii) The aunt mailed her niece the letter by post were compared to insertion conditions (iv) The aunt mailed her niece #to the letter by post. All sentences were followed by yes/no comprehension questions. The Bayesian size principle proposes that deletion errors are more likely to occur than insertion; accordingly, perceivers should be more likely to revise sentences with deletion, resulting in P600 effects. Instead, results (N=41) revealed long-lasting negative-going waveforms for deletion errors at niece and positivity at by for insertion errors. Results are interpreted in terms of perceivers' decisions to revise via search space considerations.