abstract
-
Research on eye movement control during first language (L1) reading has long since established that (i) words are read most efficiently when the first saccade into the word lands near its center, (ii) words are refixated more often when landing positions deviate from the center of the word, and (iii) relatively proficient readers' saccades land closer to this center position. Eye-tracking studies of second language (L2) reading tend to compare participant groups based on their language background (L1 vs L2) rather than L2 proficiency. As of yet, there has been no comparison of these approaches. This study reports a comparative analysis of the Multilingual Eye-movement COrpus (MECO), which contains data on English text reading and its component skills from 543 participants representing 12 different L1s. Our analyses of the distributions of initial landing positions and refixation probabilities establish that the gradient measure of proficiency in English (as L1 or L2) has a greater explanatory power than categorical contrasts between language backgrounds. We also found that English proficiency has a gradient effect on efficiency of saccadic targeting: more proficient readers landed their initial saccades closer to the word's center. However, more proficient readers of English were also less accurate in their saccadic targeting, showing greater dispersion of initial landing positions. We link this puzzling finding to the observation that landing in a suboptimal position comes with a much higher processing cost (refixation probability) for less proficient readers. This paper discusses theoretical and methodological implications of the novel findings for reading research.