OS_20. Sentence and text processing
Saturday, October 01st, 2011 [10:50 - 11:50]
TALKS
OS_20.1 - Are case markers like postpositions? The brain can tell the difference
Zawiszewski, A. , Erdocia, K. , Santesteban, M. & Laka, I.
University of the Basque Country
Several ERP studies show that morphological violations elicit a late positivity (P600), often preceded by LAN or N400) (Coulson et al., 1998; Molinaro et al., 2008). Here we investigate in further detail the generality of the neurocognitive processes underlying morphological processing. To this end we compare case marking morphemes and postpositions, debated in Linguistics as to whether they belong to one or two distinct categories (Mahajan, 2008). We conducted an ERP experiment in Basque with three conditions: (i) ergative case (ERG), (ii) dative case (DAT) and (iii) postposition “to” (TO). Morphologically well-formed nouns (irakaslea-k 'teacher-ERG', neska-ri 'girl-DAT' and klase-ra 'classroom-TO') were compared to ungrammatical counterparts (irakaslea, neska and klasea), all of them lacking the case marker or postposition required by the sentence context. 23 native speakers participated in the study. ERPs were registered while participants read grammatical and ungrammatical sentences (word-by-word) and performed a grammaticality judgment task. Results showed significantly higher accuracy and faster response times to case than to postposition violations. All violations elicited a P600 component, but only case violations generated an N400. Our results suggest that there are fine-grained distinctions in morphological processing related to grammatical category, where case and postpositions are distinct.
OS_20.2 - Information-theoretic measures of cognitive processing effort predict word-reading times
Frank, S.
Department of Cognitive, Perceptual and Brain Sciences. University College London. London, United Kingdom
In the Computational Psycholinguistics literature, it has been argued that the amount of information conveyed by each word in a sentence is a measure of the amount of cognitive effort needed to process the word. Two complementary formalizations of word information have been proposed: surprisal and entropy reduction. These quantify, respectively, the extent to which a word’s occurrence was unexpected, and the word’s effect on the uncertainty about the rest of the sentence. The goal of our study was to investigate whether both these information measures indeed predict processing effort, as observed in word-reading times. A recurrent neural network was trained on 700,000 sentences (comprising 6.9 million word tokens; 7,754 types) from the British National Corpus. Next, the network generated surprisal and entropy reduction estimates for 5,043 word tokens of 361 sentences, selected from three novels on www.freeonlinenovels.com. Reading times on the same words were collected in a self-paced reading task involving 54 native English speakers. Mixed-effect regression analyses showed that both surprisal and entropy reduction are positively related to word-reading time. This supports the hypothesis that more cognitive effort is required to process words that convey more information, and suggests that both unexpectedness and uncertainty reduction quantify information content.
OS_20.3 - Why some surprises are more surprising than other surprises
Foster, M. & Keane, M.
University College Dubin
The contrast-hypothesis of surprise (Teigen & Keren, 2003, Cognition) argues that cognitive responses of surprise occur when there is a high contrast between the surprising event and the default expected alternative. The integration hypothesis (Maguire, Maguire & Keane, 2011, JEP:LMC) has countered that surprise varies with the ease with which the surprising event can be integrated into a developing model of the discourse. Previous tests have shown the main difference between these two views is likely to lie in how they handle the impact of prior knowledge and different processing demands. Two experiments are reported that looks at one possible knowledge impact (whether the surprise scenarios, are within range of predictive schemas) and one type of processing demand (an intervening distractor task between comprehension of the scenario setting and the presentation of the surprise outcome). The results do not show any evidence that this processing demand impacts surprise, but both experiments show strong effects of prior knowledge. Specifically, we found that some surprising outcomes, were less surprising, than others if they fell within the "predictive range" of prior experience. The extent to which these two views can explain these results are discussed.