Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Always Adding Alliteration: An investigation of word form in sentence processing using alliteration
Poster Session B, Friday, September 12, 4:30 - 6:00 pm, Field House
Emily Robles1, Marta Kutas1; 1UC San Diego
Despite suggestions of wordform pre-activation during silent reading, there is little consensus on which factors determine the impact (nature & timing) of orthographic and phonological feature activation in sentence comprehension. Specifically, there is continued debate over the interaction of wordform and semantic preactivation with prediction by production accounts suggesting semantic prediction is required prior to wordform preactivation. In two experiments, we used sentences comprising alliterative words that set up moderate expectancies for a critical word that either continued the alliterative sequence or did not. Critically, these sentences are low-constraint, thereby unlikely to generate a strong semantic prediction. Event-related brain potentials (ERPs), especially the N400 and post N400 positivity triggered by the critical word, served as an indices of feature preactivation and integration, respectively. In Experiment 1, 26 participants read 80 low-constraint sentences one word at a time. The first five alliterative words were followed by one of two equally low cloze critical sixth words: alliterative (AL) or non-alliterative (NA); e.g., Becky’s bewildered beagle barked before beggars/workers entered the tunnel. There was a small but significant reduction in N400 amplitude (300 - 500 ms) for alliterative vs non-alliterative critical continuations suggesting preactivation based on orthographic and/or phonological wordform and not semantic features. In Experiment 2 this alliterative sentence context paradigm was used to address whether orthographic features, phonological features, or both were pre-activated during silent sentence reading. Forty participants read 120 low-constraint sentences whose first four alliterative words were followed by one of three possible critical continuations. The Orthographically Related sentence set (60 sentences) included critical words that were alliterative in sight and sound (AL), non-alliterative (NA), and orthographically- but not phonologically-alliterative (OR); e.g., Carly’s curious cat carried cardboard (AL)/toothpicks (NA)/chewtoys (OR) around the house. The Phonologically Related sentence set (60 sentences) likewise included AL and NA critical words as well as phonologically- but not orthographically-alliterative critical words (PR); e.g., Sam sipped some sour sangria (AL)/beer (NA)/cider (PR) at the new bar. Replicating Experiment 1 findings, there was a significant reduction in N400 amplitudes to fully alliterative (AL) compared to NA critical words. In addition, there was a significant reduction in N400 amplitudes for the OR vs NA critical words (p=0.0006) but only a marginally significant reduction for the PR vs NA critical words (p=0.056). These results suggest that both phonological and orthographic information can be pre-activated, to a degree influenced by individual differences (still under investigation). Together these experiments suggest that wordform processing and preactivation may be more readily utilized than production accounts (which rely on strong semantic feature preactivation) have proposed.
Topic Areas: Reading,