Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
7294948 | International Journal of Psychophysiology | 2018 | 8 Pages |
Abstract
Inference generation is a crucial skill in language comprehension. Recent research suggests that readers use both the contents from prior written text and their background knowledge, stored in long-term memory, to generate predictive inferences about what will come up next in a sentence. We recorded Event-Related Potentials (ERPs) to examine the reader's ability to make online inferences even in the presence of pseudowords (orthographically legal, but meaningless letter strings), that is, in the presence of referents with no a priori match to vocabulary stored knowledge. As expected, a large and sustained negativity (250-900Â ms) was elicited by the target word 'fly' when preceded by the pseudoword 'Sias' in the sentence 'Sias fly.' relative to when preceded by 'Birds' in the sentence 'Birds fly'. However, when readers were provided with an initial statement inviting to make an inference: 'Sias have wings', the word 'fly' in 'Sias fly' only elicited a negative voltage deflection over 100Â ms period (250-350), rapidly falling down to baseline. This result indicates that participants rapidly generated online inferences even with a hindered access to a referent's meaning (i.e. not knowing what 'Sias' are). Remarkably, brainwave traces to the access to a word's meaning in long-term memory (access to a well-known fact such as 'Birds fly') only diverged from ERPs for an inferred-from-reading knowledge ('Sias fly') for 100Â ms. We conclude that a fundamental search for across sentence coherence drives fast inference making processes in reading tasks. This pattern of brain response is critical to understand the rapid acquisition of new vocabulary when learning first and second languages.
Related Topics
Life Sciences
Neuroscience
Behavioral Neuroscience
Authors
Pablo RodrÃguez-Gómez, Natalia MartÃnez-GarcÃa, Miguel A. Pozo, José A. Hinojosa, Eva M. Moreno,