Anaphora resolution in L1 English-L2 Spanish: evidence from the CEDEL2 corpus

Abstract

Previous experimental research has studied some of the factors that influence Anaphora Resolution (AR) in native and non-native discourse, that is, how null and overt pronouns and NP subjects co-refer with their antecedents in discourse. In this dissertation, AR is investigated in an L2 Spanish corpus, as this offers more natural and contextually rich texts. CEDEL2, an L1 English-L2 Spanish corpus, is analyzed in order to study a very specific context of AR: the Position of Antecedent Strategy (PAS), a purely structural strategy, where the syntactic fuction of the antecedent determines the form of the anaphor. Null pronouns tend to select subject antecedents, whereas overt anaphoric material tends to co-refer with non-subject antecedents. A sample of lower and upper advanced learners plus a native control group subcorpus was annotated following a fine-grained tagset, designed for the purpose of the study. The corpus data reveal that both lower and upper advanced learners behave similarly to the natives in terms of the PAS, as all groups use a null subject pronoun to refer to a subject antecedent. As for overt anaphoric forms (overt pronouns and NPs), LCR methodology reveals that overt personal pronouns are more neutral and do not show any biases towards a subject or non-subject antecedent, whereas NPs show a strong bias towards antecedents in non-subject positon, a finding which has been overlooked in previous experimental research, as only null vs. overt pronouns had been studied. Results also show that lower and upper advanced learners do not differ from the native control group in the co-reference patterns of null pronouns and overt material in terms of the PAS, a finding that goes against the Interface Hypothesis (IH), which postulates that L2 learners show deficits at the syntax-discourse interface even at very advanced stages of their development.