Presentation
Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Lightning Talks
Do readers pre-activate phonology when reading for comprehension?
There is a Poster PDF for this presentation, but you must be a current member or registered to attend SNL 2023 to view it. Please go to your Account Home page to register.
Poster C122 in Poster Session C, Wednesday, October 25, 10:15 am - 12:00 pm CEST, Espace Vieux-Port
Eva Gutierrez-Sigut1, Manuel Perea2, Marta Vergara-Martinez2, Mairead MacSweeney3; 1University of Essex, 2University of Valencia, 3University College London
Making predictions about upcoming input is an essential part of language processing. Previous research has shown that highly predictable words and those associated in meaning are much easier to recognise due to the pre-activation of semantic category features (e.g., Federmeier & Kutas, 1999; Metusalem et al., 2012). However, the extent to which form features are pre-activated is still debated (e.g., Ito et al., 2016). The current experiment investigated whether readers also predictions based on phonological features in a sentence reading-for-comprehension task. We recorded the EEG of 22 skilled readers of English while they saw 224 high-cloze (>79%) probability sentences (e.g., “Pete broke his arm and had to wear a …”) one word at the time at the centre of the computer screen. The sentence endings were manipulated across four conditions: 1. Congruent (cast), 2. Semantically incongruent (wall), 3. Pseudohomophonic non-word (kast), and 4. Orthographic control non-word (yast). Low-cloze (<30%) probability congruent filler sentences were included, so only 30% of the items contained a pseudoword. Participants were asked to answer a comprehension question after each sentence. Mass univariate time-course analysis showed a widely distributed larger negativity for the semantically incongruent (wall) than the congruent (cast) word endings between 300 and 600 ms post-target onset. This finding shows pre-activation of semantic features. Our analyses also showed a larger negativity for the pseudohomophone (kast) than for the congruent ending, but only between 330 and 430 ms post-target onset, and in a small number of left hemisphere electrodes. In contrast, the orthographic control non-word (yast) was more negative than the congruent ending (cast) from 240 to 480 ms in most centro-parietal electrodes bilaterally. The larger N400 effects for both the pseudohomophone and the orthographic control conditions indicate that readers can accurately detect dissimilarities in form between the predicted and the presented items via bottom-up word recognition. However, the differences in the latency and scalp distribution of both effects show that pre-activation of phonological features modulates this bottom-up process in order to facilitate sentence comprehension. Moreover, the pseudohomophone condition showed a stronger post-N400 LPC than the orthographic control condition, indicative of monitoring and re-analysis. The semantically incongruent condition however did not show an LPC. Our findings demonstrate that skilled readers preactivate phonology in a reading task that does not require an explicit judgement on formal features of words. Our effects are congruent with the facilitation previously reported for non-words that were orthographically similar to predictable words (Laszlo & Federmeier, 2009; Kim & Lai, 2012), thus suggesting that skilled readers may preactivate specific lexical items during reading highly predictable sentences for comprehension (see Ito et al., 2016, for a different view). Finally, although readers can detect that the pseudohomophone differs from the expected word, they engage in further re-analysis, perhaps treating the pseudohomophone--but not the orthographic control--as a misspelling (LPC pseudohomophone effect).
Topic Areas: Reading, Meaning: Lexical Semantics