1. Spanish heritage speakers' robust knowledge of periphrastic and se-passives: Evidence from within-language structural priming and receptive judgments

• Erin Mauffray, University of California, Los Angeles

The study investigates effects of target language frequency, dominant language transfer, and task type in Spanish heritage speakers’ (HSs) knowledge of two passives: the periphrastic passive (1), which also exists in English, and the se-passive (2), which only exists in Spanish but is eight times more frequent than the periphrastic passive (Green, 1975).

(1) a. Los libros fueron publicados __.
the books be.PST.3PL published.PTCP.M.PL

b. Los libros se publicaron__.
The books SE publish-PST.3PL
‘The books were published.’

Recent work suggests that Spanish L2ers can be primed to produce both structures but show instability in their receptive knowledge of se-passives (Mauffray & Mateu, 2023). In this study, we ask, (1) Can HSs be primed to produce periphrastic and/or se-passives? And if so, (2) Do priming effects reflect target-like structural representations? If these structures are robust, we expect HSs to perform equally well on both types in the receptive task. If HSs perform better on periphrastic passives, this could be attributed to dominant language transfer and/or attrition of the form not attested in the dominant language. Lastly, if target language frequency is the main predictor in the maintenance of a HL structure, HSs will show an advantage with se-passives. This study constitutes the first to investigate both passive types, in both receptive and productive tasks, in both English-dominant, heritage Spanish speakers and Spanish-dominant native speakers (NSs). Experiment 1, a structural priming production task, includes three phases: (1) a baseline picture description task; (2) a priming phase with 32 prime/target pairs; and (3) a post-test. In Experiment 2, a timed acceptability judgment task (AJT), participants judge sentences as acceptable or not across passive items and controls. Ungrammatical test items show a mismatch in verbal agreement, which participants may incorrectly accept if they do not have target-like representations of these structures. This choice was based on a self-paced reading study in which HSs failed to notice agreement mismatches in se-passives [Rodriguez, 2018]. If HSs are ignoring se in our receptive task (and positing a null subject with which the verb agrees), they may fail to reject ungrammatical sentences with se+1PL marking on the verb.

Participants include 27 intermediate-advanced adult HSs, with 32 adult NSs for comparison. Results from Experiment 1 analyzed with logistic regressions show that HSs can be primed with both passives in both immediate and post-test phases (p<.003), patterning with the NSs. In the AJT (Experiment 2), unlike L2ers in previous studies, HSs successfully reject ungrammatical sentences to the same degree with both passives, comparably to NSs (no main effects or interactions in a logistic model).

In this study we find that HSs’ performance and production patterns with both passive types converge with NSs, despite se-passives not existing in their dominant language, and periphrastic passives typically emerging after HSs’ language dominance shift (Pierce, 1992; Oliva & Wexler, 2018). Given L2ers’ deviant performance on se-passives in previous studies, this suggests that early language experience offers an advantage for HSs in the morphosyntactic domain, contrary to some previous findings (Polinsky, 2011; Montrul, Foote, & Perpiñán, 2008).

References

2. Predictive language processing in Russian Heritage Speakers: Task effects on morphosyntactic prediction

• Olga Parshina, Ph.D., Middlebury College
• Nina Ladinskaya, National Research University Higher School of Economics (HSE) University, Moscow, Russia
• Lidia Gault, University of Wisconsin-Madison
• Irina A. Sekerina, Ph.D., City University of New York (CUNY) Graduate Center

The existing research provides evidence that prediction plays a crucial role not only in monolingual but also in bilingual language processing (Kaan & Grüter, 2021). The prediction-by-production theory (Pickering & Gambi, 2018) suggests that comprehenders pre-activate linguistic input based on production phases, starting with semantics, followed by morphosyntax, and concluding with phonology/orthography. Accordingly, a bilingual speaker with fluent production skills should be able to complete all stages and accurately predict both lexical and morphosyntactic features. However, the utility account of prediction (Kaan & Grüter, 2021) adds that this ability is influenced by various factors, including task demands: Bilinguals, even with a strong command of oral language, may not exhibit prediction in cognitively challenging tasks.In this study, we employed two tasks with different task demands (Exp.1: cloze reading task with production involved, and Exp.2: eye-tracking reading task focusing on noun-adjective gender agreement) to investigate whether heritage speakers of Russian (RHSs, N=29) demonstrate differential prediction of morphosyntactic features based on the task. We also compared morphosyntactic prediction abilities of RHSs to second language learners (RL2ers, N=29) and monolingual Russian controls (N=43). In the production task (Exp. 1), we anticipate that RHSs would predict and produce morphosyntactic features (word class, noun/adjective gender, case and number) due to their good command of oral Russian and the concentration of cognitive resources on the prediction task. According to the utility account, we expect that RHSs would not engage in morphosyntactic prediction in the more challenging reading task (Exp. 2), as heritage speakers are known to struggle with reading due to limited experience with written heritage language (Polinsky & Kagan, 2007). The findings from this study will contribute to our understanding of the factors that influence prediction in heritage language processing.

3. The use of Spanish in Emerging and Hispanic Serving Institutions

• Alicia Muñoz Sánchez, Ph.D., University of California, San Diego
• Lorenzo Sianez Jr., University of California, San Diego & San Diego State University
• Alejandro Rodriguez Aispuro, University of California, San Diego

Over the past decade, many universities have created programs to increase the number of Latino students. Having the designation of HSI (Hispanic Serving Institution) has become a goal and a source of pride. While the presence of Latinos has increased at many universities (NCES 2022), their presence is not always linked to a sense of belonging on the part of this minority group (García, G. 2019). In this study, we examine the use of Spanish in the linguistic landscape of two universities in San Diego, UC San Diego, Emerging Hispanic Serving (EHS), and San Diego State University, HSI, and analyze what the different uses of Spanish index in various spaces from the institutional perspective and by the Latino population on campus. The study also compares and analyzes data collected from over three hundred students who identify as Latinx in both institutions. The data examines the use of Spanish and the feelings of belonging to campus by the Latinx population. Our results show mostly monolingual English signage in spaces, but also reveal the use of bilingual practices, particularly in places that foster inclusion and pride in culture such as identity centers and some campus food stores. It also reveals that the Spanish language has been relegated to a cultural function, since its informative use is scarce. Institutional investment in bilingual signage is more evident in the HIS that we studied when compared to the emerging HIS.In their use of Spanish, Latinx students report feeling relatively comfortable using their heritage language on campuses, but prefer doing so in spaces where we we found more bilingual signage. Feelings of belonging to the campuses are lower in the EHS institution. However, instances of Latinx students feeling excluded or discriminated by their language use or ethnicity are still pervasive in both institutions, with an average of 17% students reporting cases of exclusion. Our study suggests that goals of institutional plurality must be accompanied by multilingual linguistic landscapes and institutional practices that truly demonstrate that institutions value students’ linguistic capital, not just as a cultural representations, but also as an official vehicle of transmitting knowledge. The study will also share current efforts to influence linguistic practices at the institutional level.

4. Factors Impacting Bilingual Preschoolers' Heritage Language Development in English-Only Instruction

• Gabriela Simon-Cereijido, Ph.D., California State University, Los Angeles
• Simona Montanari, Ph.D., California State University, Los Angeles
• Jieru Bai, Ph.D., California State University, Los Angeles

Language attrition, the decline in language abilities over time, is a significant phenomenon observed in immigrant populations experiencing language shift. This process is particularly relevant in the United States, where English holds high societal status and is essential for professional and educational advancement. Spanish-English heritage language learners, who constitute a large segment of the student population, are at risk of language attrition, potentially affecting their overall language and literacy development. This study aims to investigate the changes in Spanish lexical and grammatical skills among young Spanish-English heritage language learners in Los Angeles, California, after formal exposure to English begins in preschool. The study examines whether the children's Spanish vocabulary and grammatical system show growth or attrition over one year as they are raised and schooled in an environment that favors English. Additionally, the study explores the impact of variables such as gender and parental acculturation on the language attrition process. Contrary to previous studies documenting a loss of Spanish skills in bilingual children entering preschool or elementary school, our findings suggest that entering the school system and receiving English-language instruction does not lead to a complete shift or attrition of Spanish, at least within one year and in a context with a significant number of Spanish-speaking immigrants, such as Los Angeles. We observed a significant increase in Spanish productivity at Time 2, regardless of age, gender, or mothers' cultural orientation. The analysis of the proportion of grammatical utterances and the number of omitted words revealed developmental progress at Time 2. The qualitative analysis identified persistent nonstandard productions in the children's speech samples. While errors in articles and copula verbs decreased over time, errors in verbs, clitic object pronouns, and prepositions remained consistent. Gender significantly influenced lexical diversity, mean length of utterance, and the proportion of grammatical utterances, favoring girls. Maternal cultural orientation predicted the proportion of grammatical utterances, with both enculturation and acculturation levels being significant predictors. In conclusion, our study demonstrates that young Spanish-English dual-language learners maintain productive use of Spanish with a diverse lexicon and show growth in grammatical abilities over one year despite English-language instruction in preschool. However, their Spanish productions still show nonstandard features, indicating ongoing challenges in the acquisition of certain grammatical structures. The findings underscore the need for continued support and intervention to foster Spanish language development in this population, especially for boys and those with lower levels of maternal cultural orientation to Latinx and American culture.

5. Conversational Agents Supporting Heritage Language Literacy: A Neurocognitive Study of Sociolinguistic Factors in Rural America

• Xiang Zhang, Ph.D., The University of Alabama Tuscaloosa
• Yanyu Xiong, Ph.D., The University of Alabama, Tuscaloosa
• Aparna Bhooshanan, The University of Alabama, Tuscaloosa
• Addison Miller, The University of Alabama, Tuscaloosa

In recent years, the linguistic stability of HL has become an important question, specifically “What parts of the language are stable across NL speakers and HL speakers and why?” HL reading and writing can be indicators of such linguistic stability. Our eyes lie on HL literacy study in the large territory of rural America, which has very little research in literature. The two largest heritage languages are Spanish and Chinese in Alabama (USA Census, 2022). Despite some parental investment and little governmental support in HL programs and community-based schooling, learners have a disparity between oracy, literacy, and overall HL proficiency (Baker, 2011; Zhang & Petrovic, 2024).To remedy this discrepancy, we explore how AI (Artificial Intelligence) provides Spanish Heritage Language (SHL) and Chinese Heritage Language (CHL) learners with tailored real-time feedback and observe how their literacy changes over time in collaboration with a variety of social constituents. By examining the influences of social and cultural factors on HL learners, we aspire to augment our comprehension regarding the efficacy of pedagogical modalities in equipping them to acquire advanced skills. This study revolves around the following research questions: 1) How do language backgrounds, proficiency, gender, socio-economy, grammatical correctness, structure styles, and recognition of logography, influence the literacy accuracy of LLMs like ChatGPT 2) What is the dynamic relationship between the effectiveness of reading feedback from learners and technological constraints? 4) Is there a noticeable cognitive correlation between sociolinguistic factors, literacy levels, and the use of AI technology in HL learning? To answer the above questions, we first issue surveys to assess HL learners’ profiles and proficiency throughout different developmental intervals. Using question banks from various standardized language exams, we fine-tune a neural network model to create an automated, interactive feedback system for students learning and perceive how their reading/writing transformed during 6 months. We employ language corpora to establish an automatic reading feedback system by training Bidirectional Encoder Representations from Transformers (BERT). Then we explore the influence of learner and linguistic variables on the system's performance. For instance, we will test whether integrating the prompt engineering technique in our guided interactive protocol will improve the performance of the interactive feedback system (White et al, 2023). After evaluating the feasibility of the system during interaction with the learners, we examine if it increases literacy and whether there are changes in the neural correlates underlying the increased language competence by using the Electroencephalogram (EEG) technique to observe how participants’ neural changes evolve alongside their linguistic development). Overall, we want to produce an easily accessible resource for SHL and CHL students and teachers to improve reading/writing skills, and overall confidence. We hope to expand this project to encompass other major HL groups in the American South, and, especially, Rural America, including languages like Gujarati, Korean, German, Russian, and Vietnamese, which have almost no school system or community support OUTSIDE of major metropolitan cities.

6. Sociolinguistic competence in children and young heritage language speakers: subject personal pronoun expression

• Xinye Zhang, University of California, Davis
• Xue Yin, Missouri State University
• Han Chang, University of Pennsylvania

Learning a language means not only mastering the grammatical structures in a target language but also becoming a legitimate member of the target speech community. It requires knowledge about how to recognize and produce contextually appropriate language, or sociolinguistic competence, which has been examined by measuring the native-like patterns of sociolinguistic variables. (Bayley et al., 2022; Bayley & Regan, 2004; Lyster, 1994; Regan et al., 2009). Previous studies have shown that besides internal linguistic constraints, external factors including age, gender, language proficiency, and language contact (e.g., studying abroad) all may affect learners’ choice between different linguistic variants (Edwards, 2011; Eisenstein, 1982; Kennedy Terry, 2022; Pozzi, 2022; Preston & Bayley, 1996; Regan et al., 2009; Romaine, 2003). However, language learning often differs in typical L2 learners and heritage language learners due to input, language environment, and learner background. It may be reasonable to suspect that these factors also contribute to the development of sociolinguistic competence. This study investigates the subject personal pronoun (SPP) variation in Chinese Mandarin by young adults and children who were Chinese Heritage Language (CHL) learners and explores the development of sociolinguistic competence in early childhood and early adulthood. The presence of SPP is optional in Mandarin and to some extent can reflect the sociolinguistic competence of the speaker: the omission of SPP in some cases may cause misunderstanding while the overuse of SPP may sound inappropriate.Data were collected from 15 CHL students (aged 18-27, 11 females) and 27 children (aged 3-5, 12 girls) via sociolinguistic interviews. In total, 10451 tokens were elicited and coded according to the presence of the SPP and intra-linguistic, social, and discursive constraints on it. Specifically, intra-linguistic factors include person and number of the subject, clause type, and the semantic classification of the verb. Social factors are age, gender, and heritage generation (US-born vs. early arrival) of the speaker. Discursive factors contain referent continuity and priming. Generalized linear mixed-effects models have been adopted for statistical analyses (Johnson, 2009; Plonsky & Ghanbar, 2018; Speelman et al., 2018). Statistical results show that intra-linguistic constraints (i.e., person and number, clause type, and semantic type of verb), social factors (i.e., heritage generation, gender and age), and discursive constraints (i.e., referent continuity and priming) all significantly affect SPP variation in heritage Mandarin. Overall, the patterns observed in linguistic and discursive constraints resemble previous findings in subject pronominal variation in Mandarin and other languages (Chang & Zheng, 2018; Kim, 2000; Li, 2014; Li et al., 2012; Nagy et al., 2011; Paredes Silva, 1993; Shin, 2016; Valian & Eisenberg, 1996; Wang et al., 1992; Zhang, 2021). CHL children demonstrated a higher SPP rate (78.7%) than CHL undergraduate students (73.9%). Moreover, compared with children, students are more sensitive to differentiated gender roles regarding SPP usage. Lastly, the effect of language contact may result in (1) US-born speakers tending to use more SPPs, and (2) with age increases, the SPP rate by both children and young adult students rising as well.