BACKGROUND Studies of resolution of Chinese overt subject pronouns (ta) and null subjects (e) report two main findings: in adults, while null subjects allow coreference to the subject antecedent in both forward and backward configurations, overt pronouns do so only in the forward configuration ((1) and (2)) (Huang and Lin 2021; Lust et al. 1996; Su 2020). In contrast, children allow both the null element and the overt pronoun to corefer with the subject antecedent in both contexts. Various accounts have been proposed for this, such as a Chinese specific non-coreference rule on overt pronouns (Huang and Lin 2021). Children at age 6-8 do not seem to have acquired this rule yet (Lust et al. 1996), since they do not distinguish between null and overt subjects.

CURRENT STUDY This study tests the development of these two types of anaphora in two different age groups of children: 24 children aged 4-6 (4;0-6;11, M=5;4) and 24 aged 8-10 (8;3-10;11, M=9;4) and 48 adult controls. We hypothesized that the younger children do not display distinct interpretations of overt vs. null subjects in forward vs. backward anaphora; in the case of older children, we expect adult-like interpretations, being at a stage of development in which they are claimed to acquire the difference between null and overt subjects, as well as the non-coreference rule. We conducted a picture selection task (24 experimental items + 24 controls, across 4 lists, Figure 1). As expected, adults show an asymmetry in forward and backward anaphora (Figure 2). This pattern is found in the older children, but not in the younger children: in this case, no difference emerged between null and overt subjects. A logistic mixed regression model on participants’ selection of the subject as the antecedent (Table 1) showed that coreference with the subject depended on Anaphora and Pronoun Type, being higher in the case of null subjects, and in the case of forward anaphora. Moreover, these factors modulated adults’ preferences more than children, especially the younger children, as indicated by the significant three-way interactions (a separate model on child data with Age (centered) as predictor showed a significant PronounAnaphoraAge interaction, p = 0.002).

DISCUSSION As predicted, the younger children do not yet show the distinction between null and overt subjects. However, two unexpected patterns emerge in null subjects. First, in the younger children, the null subject shows a far lower rate of coreference to subject antecedent (both forward and backward) than older children (and adults), contra previous observations. Second, the older children do not yet show adult-like interpretations, which contrasts with the results for overt subjects. This raises a question as to whether null and overt subjects are acquired in tandem at age 6-8. Even though our current study does not test this question, we explore a possible explanation in the nature of null subjects. Instead of being minimally different null exponents of overt pronouns, null subjects might be a null exponent of reflexives instead, as proposed by Barbosa (2019) and Zhao (2012).