Presentation
Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Lightning Talks
Effect of Working Memory Load in Resolving Filler-Gap Dependency in Natural Sentence Reading- evidence from the simultaneous recording of Eye Movements and Fixation-Related Potentials
Poster E121 in Poster Session E, Thursday, October 26, 10:15 am - 12:00 pm CEST, Espace Vieux-Port
Jou-An Chung1,2, Jie-Li Tsai4,5, Chia-Ying Lee1,2,3,5; 1Taiwan International Graduate Program, Academia Sinica, Taipei, Taiwan, 2National Central University, Taoyuan, Taiwan, 3Academia Sinica, Taipei, Taiwan, 4National Chengchi University, Taipei, Taiwan, 5Research Center for Mind, Brain, and Learning, National Chengchi University, Taipei, Taiwan
The filler-gap dependency within Relative Clause (RC) structure provides detailed information about the specific person or thing, but resolving the filler-gap dependency can impose large demands on working memory (WM). Prior event-related potential (ERPs) studies have demonstrated the effect of WM load imposed by the filler-gap dependency on sustained negativity or P600 in the word-by-word sentence reading with a fixed presentation rate. This study aims to investigate the effect of WM load in natural reading with the simultaneous recording of Eye-Movement (EM) and Fixation-Related potentials (FRPs), which provides complementary EM and ERPs measures to capture the cognitive operations in natural reading. Three levels of WM load are determined by the varying distance between filler and gap. The Low WM load (LWM) condition comprises the sentences with the simplest Chinese Subjective Relative Clause (SRC) ( [FILLER] RC verb (協助/help) + RC noun (居民/resident) + 的(DE) + Head noun [GAP] (志工/volunteer) ). As for the medium and high WM load conditions, one constituent- adverbial modifier (ex: 熱心/enthusiastically) and two constituents- a locative preposition and an adverbial modifier (ex: (在社區/in the community)+(熱心/enthusiastically)) were added prior to the RC verb, respectively. Thirty-six native Chinese speakers were invited to read 81 sentences at their own pace while their EMFRPs were simultaneously recorded. The improved masking empirical mode decomposition (iMEMD) method was applied to correct the oculomotor artifacts introduced by ocular motion. Additionally, the overlapping correction analysis was used to deconstruct the temporal overlap of FRPs signals elicited by the successive fixations. The result shows the effect of WM load on EM and FRPs. The EM patterns and FRPs on Head Noun where the ambiguity is resolved were compared among three conditions. EM indices include early (e.g., First-Fixation-Duration, Gaze Duration…etc.) and late measures (e.g., go-past time, regression). ERP components such as N100, P200, N400 and sustained negativity were examined as to their amplitude, latency, and topographical distribution. Compared to the LWM load, sentences with HWM load exhibited longer First-Fixation-Duration and Gaze Duration on Head Noun. The cluster-based permutation test shows that they also exhibited an increased N100 ( p = 0.004) and a reduced P200 amplitude (p = 0.009). These findings indicate the effect of WM load on the early stages of information processing. Readers are more attentive as the WM load increases. Moreover, sentences with HWM showed longer go-past time, higher regression-in rate, and greater negativity of N400 ( (p = 0.006) on the head noun. These findings suggest the need for reanalysis and processing difficulty in resolving the longer filler-gap dependency. Sentences with MWM showed similar EM and FRPs patterns to the sentences with HWM load, indicating that both types of WM load might impose similar amount of demand on readers. In sum, EMFRPs yield more complete evidence for the effect of WM load on natural reading from the early stages of information processing, sentence integration and reanalysis.
Topic Areas: Reading,