1. Pablo hubiera pusido un suéter púrpuro: Child heritage speaker' overregularization of Spanish past participles

• Elisabeth Baker Martinez, University of New Mexico

The current study investigates overregularization of Spanish irregular past participles (e.g., dicho ‘said’, regularized as decido) among 20 child heritage speakers of Spanish in New Mexico, ages 5;1 to 11;9 and 15 adults. Overregularization occurs when a child produces an irregular form analogously to its regular counterpart (e.g., eated instead of ate). Typically, children first produce the irregular form and then, after they have learned a morphological pattern, they overapply that pattern to the irregular form. Ultimately, children retreat from overregularization and once again produce the target irregular form. While there has been a wealth of studies on monolingual children’s overregularizations, very few have investigated this phenomenon in child heritage speakers.The children in this study were presented with a task designed to elicit past participles. Of the 20 participants, only 13 produced past participles (n = 233). The other children used forms like infinitives rather than participles. The past participles were analyzed to gauge the impact of age, Spanish language experience, Spanish morphosyntax proficiency, and lexical frequency on overregularization. Furthermore, in order to address how this structure is regularized across speaker groups and better perceive any input effect, children’s overregularizations were compared to those of an adult control group. The adult control group was comprised of 9 caregivers and 6 teachers whose ages ranged from 29 to 51. Results demonstrate that the children overshot the rates of regularized past participles found in adults' speech; children’s participles were overregularized at high rates (74%), resulting in forms like ponido and pusido (‘put’, cf: puesto). Meanwhile, adults regularized 16% of the time. Results from a regression analysis indicate that overregularization was more likely among the younger children, the children with lower morphosyntax scores, and with lower-frequency participles. Further, an interaction between morphosyntax score and lexical frequency indicated that children with higher scores overregularized with lower-frequency participles, but not higher-frequency ones. In contrast, children with low scores overregularized with both low- and high-frequency forms. In summary, child heritage speakers overregularize Spanish past participles at high rates, and the retreat from overregularization is tied to overall grammatical development and lexical frequency, suggesting that the acquisition of irregular participles is dependent on experiencing multiple instances of the irregular verb form.

2. Flexibility of complex syntax after puberty: The case of Turkish-American returnees

• Aylin Coskun Kunduz, University of Illinois at Urbana-Champaign
• Silvina Montrul, Ph.D., University of Illinois at Urbana-Champaign

This study addresses a fundamental issue at the heart of our understanding of language acquisition: how flexible and malleable are grammars after puberty? We know that heritage speakers (HSs) exhibit significant structural variability in their morphosyntax (Montrul & Polinsky, 2021). Intervention studies with explicit grammatical instruction have shown that HSs can recover aspects of a childhood language in adulthood (Muāgututiʻa, 2018), suggesting that linguistic knowledge acquired during the critical period, underused or underdeveloped throughout childhood, remains available when reactivated in adulthood (the Permanence Hypothesis, Bowers et al. 2009).We take this line of research a step further and ask whether interrupted acquisition before puberty in an immigration context can result in full nativelike attainment if HSs are tested in a naturalistic setting, when fully immersed in and using the HL in a majority language context (i.e., in the homeland). To this end, we test returnees: HSs born in an immigration context who returned to their country of origin in later years (Flores, 2020). We ask: To what extent do Turkish HS returnees show target-like acquisition of Turkish morphosyntax upon full immersion in Turkish after their return? If the Permanence Hypothesis holds, we expect returnees to pattern with monolingual Turkish speakers; if maturational effects play a more prominent role than input factors, post-puberty returnees would be less accurate than native speakers and more like Turkish HSs in the US. Thirty-twoTurkish-American returnees with varying age of return (AoR) to Turkey (before and after puberty) were compared to 30 Turkish HSs residing in the US and 30 Turkish-speaking adults in Turkey (i.e., the baseline group). We report on the production and the judgement data on three vulnerable structures in Turkish HSs: passives (Bayram et al., 2019), relative clauses (RCs) (Coşkun Kunduz & Montrul, 2022) and anaphoric reference (Gračanin-Yüksek et al., 2020) using a context-based Acceptability Judgement Task (AJT) and a Sentence Repetition Task (SrepT) with a total of 164 and 116 stimuli, respectively.Analysis of the data (binomial logistic regression models) indicated that returnees and monolinguals showed significantly higher accuracy than HSs in the SrepT and in the AJT, particularly with respect to RCs. Returnees also patterned with monolinguals in showing better judgement of passives than RCs, which in turn was followed by anaphoric references; in contrast, HSs showed similar performance for passives and RCs. Correlations between accuracy percentages, AoR to Turkey and length of residence (LoR) in Turkey of returneeswere not significant, suggesting that target-like acquisition of HL can occur after full reimmersion in the HL, regardless of the age of return to the homeland (i.e., before or after puberty). Our results also support the Permanence Hypothesis and suggest that complex morphosyntax is still nimble and malleable post-puberty. These findings provide a unique angle on the roles of age of acquisition and input factors in the acquisition and maintenance of a native language acquired in a bilingual context.

3. Heritage bilinguals' attitudes modulate their speech perception

• Maria Gavino, Ph.D., University of California, San Diego
• Matthew Goldrick, Ph.D., Northwestern University

The social information listeners are given about a talker affects their perception (Vaughn, 2019). For example, monolingual L1 English listeners are more likely to accurately perceive a talker when they are told the talker is a native versus non-native speaker, even though it is the same talker across the two conditions. This effect has been attributed to a number of factors, including listeners relying on prior experiences and language attitudes rather than the acoustic information, as well as listeners devoting less attention to speech when listening to stigmatized accents (Vaughn, 2019). Yet, limited work has tested how individual differences in these listener factors might modulate these effects, focusing instead on talker variation (e.g., Neuliep & Speten-Hansen, 2013). Heritage language bilinguals can offer a unique window into such effects as they exhibit significant variation in their language attitudes. This study investigates how heritage bilinguals’ language attitudes affect their auditory perception of talkers when they are given different social information (i.e., native versus non-native speaker) about a talker. One hundred and twelve Mexican-American L1 Spanish-L2 English bilinguals (dominant in English) transcribed speech in noise in English-only and Spanish-only blocks. In each block, there were two heritage Spanish-English bilingual talkers of different genders associated with native or non-native social information (with a total of 8 talkers, counterbalanced across conditions, distributed across the listeners). Participants then answered the following open-ended questions: “Do you think that Spanish and/or English are an important part of your culture? Of your identity? Why or why not?” A logistic mixed effects regression (accounting for random variation across listeners) showed that heritage speakers were significantly better at perceiving speech in their dominant language (i.e., English) than their non-dominant language (i.e., Spanish; β = 0.76, p < 0.001). However, there was no significant effect of the social information manipulation (β = 0.02, p > 0.05). A thematic analysis of participants responses showed that most participants (with the exception of 2) either considered both English and Spanish as part of their identity (N = 86) or only Spanish as part of their identity (N = 24). Participants who consider only Spanish as part of their identity showed an effect of social manipulation in the Spanish block (i.e., had higher transcription accuracy when they’re told a talker is a native versus a non-native speaker; β = 0.75, p < 0.01). However, participants who indicate both languages as part of their identity showed no effect of social manipulation (β = -0.11, p = 0.39).Overall, the results show language dominance effects for heritage bilinguals, however, effects of social information only emerge when language attitudes are taken into account. One possible explanation for this effect is that attitudes impact listeners’ language use (e.g., reducing interactions with non-native Spanish speakers) which then impacts listeners’ language processing. Consequently, it is important for future work to take these individual differences into account in order to avoid reaching inaccurate conclusions regarding the nature of bilingual language processing.

4. Investigating Korean heritage learners' peer interaction and the quality of written texts during collaborative writing tasks

• Ha Ram Kim, Ph.D., University of California, Irvine
• Eun Hee Kim, Ph.D., Northwestern University

While it is widely recognized that heritage language (HL) learners and second language (L2) learners have distinct learning needs, they are often grouped in advanced mixed language courses. This has prompted some task-based research exploring L2-HL interactions in such environments, revealing that these interactions may provide diverse learning opportunities (Blake & Zyzik, 2003; Bowles, 2011; Bowles, Toth, & Adams, 2014; Henshaw, 2015; Torres and Cung, 2019). Despite this, there remains a need for further exploration, particularly regarding how HL learner interactions, whether with other HL learners or with L2 learners, contribute to the development of written texts in less commonly taught languages, such as Korean. The current study aimed to address this need by examining the dynamics of peer interaction and the quality of written texts produced by HL and L2 pairs during collaborative writing tasks in an advanced Korean language classroom. To be specific, two HL and two L2 learners were paired with different partners to work on two separate collaborative writing tasks, each producing a single text based on a series of pictures. They interacted in various modalities such as audio and text during the outlining/drafting, and revising stages. In total, 8 audio-recordings of interactions between L2-HL, HL-HL, and L2-L2 pairs were qualitatively analyzed for language-related episodes, target language use, as well as the content of the talk. The rough and final drafts produced by each pair were also analyzed for syntactic complexity, accuracy, and length. The findings indicated that, despite the differences in their language and educational backgrounds, every pair participated in meaningful discussions, which resulted in texts that were both accurate and structurally complex. A notable observation from the interactions between L2 and HL learners was that the HL learners often relied on their intuitive understanding of the language, while placing trust in the grammatical knowledge of the L2 learners for accuracy. Furthermore, while both L2 and HL learners tended to avoid using the target language during the task, the cooperative writing activities still offered opportunities to enhance the learners' self-assurance in using the target language.

5. The Structure of Armenian Heritage Language Narratives

• Arus Movsesyan, University of Michigan-Dearborn

The differences between heritage and fully competent speakers have been studied extensively. However, most of the research has been confined to a limited number of languages. The present study aimed to identify the linguistic properties of 1.5- generation U.S.-based Eastern Armenian heritage speakers studying Western Armenian at a K-12 school. The study compared their oral narratives to those produced by Eastern Armenian Yerevan dialect speakers. It compared the narratives elicited from heritage speakers within and outside of their age group, and to themselves as they told the story in Armenian and English. The study also aimed to assess how second dialect acquisition played out in Armenian heritage speakers’ narratives. Applying the elicitation methodology by Berman and Slobin (1994), this study came with a twist; it used a different wordless book visually more appealing to the child participants in the study. The task replicated the original frog story elicitation mechanisms and instructions. The quantitative measures calculated in this study were replicated from Polinsky’s research on Russian heritage narratives (2008). The results of the current study suggested that Armenian heritage narratives mostly come in short utterances, rarely contain embedded structures, have an inconsistent use of tense forms, and tend to simplify the elaborate case system of Armenian by mostly relying on unmarked nominative case and accusative. In the case of accusative, however, heritage speakers fail to use the marked accusative when referring to animate objects which is unacceptable in the baseline Armenian. It is uncertain if such inconsistencies with the case system are attributed to the heritage speakers’ exposure to Western Armenian with its less demanding case marking or the contact with English, or maybe both. The study also examined the syntactic properties of the narratives. The results showed that SVO is the most frequently produced word order pattern among HL and fully competent speakers of Armenian. However, heritage speakers’ overreliance on SVO suggests that they have difficulties producing non-canonical word order patterns which are also atypical to their dominant language. Although the results of this preliminary research are promising, a much larger sample of narratives would render a more comprehensive and statistically testable dataset.

References

6. Sociolinguistically heterogeneous healthy Mandarin-English bilingual adults residing in California and their performance on a new Mandarin sentence comprehension assessment for aphasia

• Preeti Rishi, San Diego State University / University of California, San Diego
• Yusheng Wang, San Diego State University / University of California, San Diego
• Tracy Love, Ph.D., San Diego State University
• Henrike Blumenfeld, Ph.D., San Diego State University

Aphasia is an acquired communication disorder that frequently results in impaired sentence comprehension (Clark, 2011). When treating bilingual people with aphasia (biPWA), speech-language pathologists (SLPs) must thoroughly assess and understand their clients’ language profiles in order to maximize therapy outcomes (Centeno & Ansaldo, 2016). Several English assessments are available to evaluate sentence comprehension abilities in people with aphasia (PWA), such as the English version of the SOAP Syntactic Battery of Sentence Comprehension (SOAP; Love & Oster, 2002), but few exist in Mandarin. The available Mandarin assessments are validated on monolingual Mandarin speakers living abroad (e.g., Wang & Thompson, 2016). However, the sociolinguistic profiles of Mandarin-English bilinguals living in the U.S. are often heterogeneous and may differ from monolingual Mandarin speakers on dimensions that vary among heritage language speakers (Grosjean, 1989). Here, we think of heritage language speakers broadly and fluidly, including Mandarin speakers who were childhood bilinguals in the U.S., but also individuals who immigrated after childhood and are speaking Mandarin as a home language (Ortega, 2020). Currently in the U.S., and California specifically, Chinese dialects (including Mandarin) are the second most widely spoken non-English language cluster after Spanish (Dietrich & Hernandez, 2022; Migration Policy Institute, 2021), necessitating appropriate assessments for this population. The purpose of the current study is to ensure that a newly adapted and validated Mandarin version of the SOAP (M-SOAP; Rishi et al., manuscript in prep.) can be appropriately utilized across sociolinguistically diverse profiles of healthy Mandarin-English bilingual adults in California as a potential future clinical tool for SLPs to use with biPWA. An ideal measure would yield similar results in the context of sociolinguistic variation (Paradis & Libben, 1987). Thirty-five healthy Mandarin-English bilingual adults residing in California (mean[age]=42yrs, SD[age]=21yrs, range[age]=18–85yrs) completed the M-SOAP assessment, a language experience and proficiency questionnaire (LEAP-Q; Marian et al., 2007), and a picture-naming test. Principal component analysis revealed two factors that captured 71% of participants’ sociolinguistic variability: Current Proficiency and Exposure (λ = 4.45; includes self-reported ratings of Mandarin abilities, naming test scores, and current exposure) and Language Learning History (λ = 1.25; includes age of first Mandarin exposure, age of fluent attainment, and years residing in an English-speaking country). Multiple regression (F[2, 32] = 3.65, p < 0.05, R-squared = 0.43) showed that only Language Learning History significantly predicted M-SOAP accuracy (β = -0.02, t = -2.59, p < 0.05) while Current Proficiency and Exposure did not (β = -0.001, t = -0.12, p = 0.90). Thus, M-SOAP outcomes remain robust to differences in current proficiency and exposure but decrease as participants acquire and master Mandarin later and/or reside longer in an English-speaking country. The obtained results provide initial confirmation that the M-SOAP can be employed across sociolinguistically heterogeneous profiles of Mandarin-English bilinguals living in California. Caution should be taken when interpreting outcomes of this assessment with biPWA who learned Mandarin later in life or have experienced attrition of Mandarin skills, as weaker performance may reflect lower premorbid abilities as opposed to aphasia-induced sentence comprehension deficits.

References

7. Language Proficiency and Executive Control in Heritage Language Speakers and Monolingual Adults

• Gabriela Simon-Cereijido, Ph.D., California State University, Los Angeles
• Beatriz Barragan, Ph.D., A.T. Still University
• Lucía I. Méndez, Ph.D., University of North Carolina at Greensboro

Previous research suggests that bilinguals, compared to monolinguals, perform better in executive function (EF) tasks (Bialystok et al., 2012, 2016; Costa & Sebastián-Galles, 2014). It is hypothesized that the bilingual advantage is associated with reducing language processing conflicts between the two activated languages. However, this hypothesis has been challenged by data showing no significant differences between monolingual and bilingual groups (Paap et al., 2016), especially when controlling for language proficiency and degree of bilingualism (Rosselli et al., 2016). However, research with heritage speakers is complex because they are not a homogenous group, and their varying degree of bilingualism may impact their performance in language processing and executive control tasks. Research with monolingual and bilingual children reported differences in executive functions (EF) according to their language proficiency (Marton, 2008; Iluz-Cohen and Armon-Lotem, 2013), and high proficiency balanced bilingual young adults performed better on EF tests than low proficiency balanced bilinguals (Rosselli et al., 2016).The present study explores the relation between inhibition and shifting task performance and linguistic proficiency in monolingual (N=39) and Spanish/English bilingual (N=43) adults. Participants answered a linguistic questionnaire, completed a semantic verbal fluency test, and participated in receptive vocabulary assessments, including the ROWPVT, and the TVIP. Participants also completed a Flanker Task (FT) and a Switching Task (SWT). Language proficiency was calculated using a z-score combining the semantic and vocabulary tests in each language. No significant differences in EF performance were found between monolinguals and bilinguals. High-proficiency English monolinguals performed significantly better in the SWT than low-proficiency monolinguals (U=399, p=<.001), and a difference at the .1 level was found for the FT (U=123, p=.061). In bilinguals, no significant difference in EF performance was found between English proficiency groups, but there was a significant difference in the SWT performance between high and low Spanish proficiency (U=133, p=.023). When dividing bilinguals into graded proficiency groups (balanced high, balanced low, and unbalanced proficiency), a significant difference in the SWT performance between the balanced groups was found (H(2)=7.07, p=.029), with in-between results for the unbalanced group. Although our results showed no difference in EF tasks performance between monolinguals vs bilinguals, there was a significant difference in EF tasks performance between high- and low-proficiency monolinguals and high- and low-proficiency bilinguals. These suggest that a cognitive advantage could be more associated with the degree of linguistic proficiency than with bilingualism per se. In bilinguals, proficiency in Spanish, not in English, was associated with higher EF performance even though English was the dominant language for approximately 70% of the bilingual sample. Our data suggests that balanced high-proficiency bilinguals performed significantly better in EF tasks than balanced low-proficiency bilinguals. This finding has important implications for heritage speakers, suggesting that maintaining and improving the native language while learning and developing a second language reaching high proficiency levels in both, may have cognitive benefits.