L1 morphosyntactic attrition at the early stages: evidence from production, interpretation, and processing of subject referring expressions in L1 Spanish-L2 English instructed and immersed bilinguals

Abstract

Research on bilingualism and second language acquisition has extensively investigated how a bilingual's first language (L1) influences their second language (L2). While there is ample evidence of L1 influence on the L2, there has been limited research on the opposite direction, specifically L2 to L1 influence within the context of L1 attrition. Most studies on L1 attrition, especially in morphosyntax, have focused on long-term immersed bilinguals who have lived in an L2-dominant naturalistic environment for an extended period of time (Chamorro, Sorace, et al., 2016; Gargiulo, 2020; Gürel, 2004; Kaltsa et al., 2015; Tsimpli et al., 2004). Consequently, a key question that has been largely overlooked is whether intensive instructed exposure to the L2 can also lead to attrition. The potential attrition effects on bilinguals exposed to the L2 in instructed contexts, even though limited research has suggested bidirectional influence in these bilinguals (Cook et al., 2003; Długosz, 2021; Requena & Berry, 2021), remains largely unknown. As a novelty of this dissertation, we will explore L1 attrition in L1 Spanish-L2 English bilinguals who receive frequent L2 exposure and use in formal instructed contexts and who live in an L1- dominant environment, i.e., Spain, and compare them with immersed bilinguals in a naturalistic setting. Additionally, previous research has focused on the later stages of L1 attrition (i.e., after prolonged exposure to the L2), thus leaving a gap on the early stages of L1 attrition, which, as Schmid and Cherciov (2019, p. 273) argue, has been ‘a period more often than not completely neglected by attrition studies’. In order to fill that gap, special attention will be paid to the production, interpretation, and processing of subject referring expressions REs. In particular, we will pay attention to (null and overt) pronominal subjects and noun phrases (NPs), which are employed as a cohesive mechanism as they (can) corefer with an entity in prior discourse. Thus, reference management proves to be an essential and pervasive aspect of human communication. The importance of investigating subject REs also lies in the fact that they have been identified as a vulnerable domain in bilinguals, both in L2 acquisition and L1 attrition, as hypothesised by theories such as the Interface Hypothesis (Chamorro & Sorace, 2019; Sorace, 2011, 2012; Sorace & Filiaci, 2006), the Activation Threshold Hypothesis (Paradis, 1993, 2004, 2007), and the Pragmatic Principles Violation Hypothesis (Lozano, 2016, 2018). Therefore, to fill gaps in the current literature, the aim of this thesis is to investigate the production, interpretation, and processing of 3rd person singular subject REs in two groups of advanced L1 Spanish-L2 English bilinguals, i.e., instructed and immersed bilinguals, who will be compared against Spanish functional monolinguals. Thus, data from the same participants and from different domains will be triangulated. Data from 207 L1 Spanish-L2 English participants (33 functional monolinguals, 80 advanced instructed bilinguals in Spain, and 94 immersed bilinguals in the UK) were collected. The participants completed three methodologically distinct tasks: two naturalistic production tasks (corpus-based oral video-retelling tasks), an offline experimental task (a picture selection task that measured interpretation preferences), and an online experimental task (a self-paced reading task that measured reaction time in milliseconds). While the two production tasks explored the distribution of 3rd person singular subject REs in topic continuity (TC) contexts as a vulnerable domain in bilinguals (Lozano, 2009; Martín-Villena & Lozano, 2020), the interpretation and processing tasks tested the predictions from the Position of Antecedent Strategy (Carminati, 2002), which claims that null pronouns tend to select subject antecedents and overt pronouns tend to bias towards object antecedents. In addition, a background questionnaire, the Bilingual Language Profile (Birdsong et al., 2012) was used to collect data and provided a continuous dominance score for each participant, as well as a working memory task, and a placement test. The production tasks analysed all 3rd person subject REs (N = 9225) in TC and used a fine-grained tagset implemented in the UAM Corpus Tool (O’Donnell, 2009) and the results were reported using χ2 statistics. Both the picture selection and self-paced reading tasks were modelled after Tsimpli et al. (2004) and Kaltsa et al. (2015), respectively, and were analysed using (generalised) linear mixedeffect models in R (Bates et al., 2015). The analysis included all relevant fixed effects as well as their interactions and the random-effect structure that was allowed by the design (Barr et al., 2013). The results from the two corpus-based production tasks showed that advanced L1 Spanish-L2 English bilinguals significantly produced more pragmatically infelicitous overt subject REs (both overt pronouns and NPs) in TC than functional monolinguals overall, with significant differences between the two bilingual groups. These differences were exclusively attested in the most cognitively demanding task, that is, Task 2, as it included several same-gender and different-gender antecedents. Additionally, the results from the picture selection task indicate that 1) the two advanced bilingual groups differed from functional monolinguals only in the overt pronoun condition, and 2) that the more L2-dominant instructed and immersed participants were as measured by the BLP, the more flexible their interpretation of the overt pronoun was, biasing more towards the subject (Chamorro, Sorace, et al., 2016; Kaltsa et al., 2015; Tsimpli et al., 2004). Finally, the results from the self-paced reading task evidence a lack of processing cost when overt pronouns are forced to bias towards subject antecedents in bilinguals, a finding that is not replicated in functional monolinguals. In sum, the findings support the predictions from the Interface Hypothesis (Chamorro & Sorace, 2019; Sorace, 2011, 2012; Sorace & Filiaci, 2006), the Activation Threshold Hypothesis (Paradis, 1993, 2004, 2007), and the Pragmatic Principles Violation Hypothesis (Lozano, 2016, 2018). Notably, bilinguals have been shown to differ from functional monolinguals in the interpretation and processing of overt pronouns, arguably since they have an L2 competing element, in line with the ATH, and the former are comparatively more redundant in production so as to avoid potential ambiguity as predicted by the PPVH. Overall, these results shed new light on L1 attrition by exploring the overlooked initial stages of this phenomenon in immersed bilinguals by triangulating data from different domains (i.e., production, interpretation, and processing). The findings also provide new evidence of variability in the L1 of advanced instructed bilinguals, a population that has been under-researched within L1 attrition studies. Additionally, the results from this dissertation call for the use of continuous measures to investigate gradience in bilingualism to more subtly characterise variability in bilinguals leaving aside dichotomous characterisations which prevent the field from moving forward.