Eye-Tracking Data on Reading English Relative Clauses by Farsi and Dutch Speakers
Description
In two eye-tracking experiments, we explored whether there is a cross-linguistic influence of L1 Farsi and L1 Dutch on the processing of L2 English subject and object relative clauses. Moreover, we assessed whether these L2 learners rely on the morphological cues in that-clause and who(m)-clause conditions or whether they can also process the structural cues in English relative clauses. Forty-two Iranian Farsi-speaking students and thirty Dutch students at the University of Groningen participated in the experiment. They were tested individually in the eye lab at the University of Groningen. Eye movements were recorded using a Tobii 1750 Eye tracker, interfaced with a laptop. The sampling rate was 50 HZ. The materials for this experiment were taken from the items used by Staub (2010) and Staub et al. (2017). The present study used 36 sentences to create four types of relative clauses per item (in total 144 relative clause sentences); subject and direct object relative clauses with either the relative pronoun ‘that’ or the relative pronouns ‘who/whom’. We divided the English relative clause sentences into six regions: the antecedent, relative pronoun, embedded noun phrase in the relative clause, embedded verb phrase in the relative clause, matrix verb, and post-matrix verb. For each region, four reading time measures of first-pass reading time (FP) or gaze duration, First-pass total reading time, Regression path duration (RPD), and Total reading time (TRT) were compared. Statistical analysis was conducted using linear mixed-effects models of the reading times. We used R (R Core Team, 2019) and the lme4 package (Bates et al., 2015) to predict the times of the eye movement measurements (first-pass, first-pass total, total reading time, regression path duration) based on Clause Type (subject versus object) and Type of Pronoun (that-clause versus who(m)-clause). As random effects, we included intercepts for subjects and items as well as by-subject and by-item random slopes for the effect of Clause Type and Type of Pronoun and their interaction. First, we created the model with the main effects and then compared it with the model with the interaction. In cases where the model did not converge, we first removed the interaction from the random part, and if necessary, in some cases, also the random slopes for the main effect. In the models, the subject relative was the baseline for Clause Type and that-clause condition was the baseline for the Type of Pronoun. In the case of the interaction of Clause Type and Type of Pronoun, we ran extra analyses to test the effect of Clause Type in the two Types of Pronoun conditions separately.
Files
Steps to reproduce
The data were produced by an eye-tracking method. While reading relative clause sentences, the eye movement measures of L2 speakers were recorded.
Institutions
- Rijksuniversiteit Groningen