Program and Abstracts
Parallel Session 1A: Speech perception and raciolinguistic ideologies
Moderator: Professor Bob Bayley • Co-moderator: Brennan Gonering
Presenter 1: Tyler Kline (Doctoral student, University of California, Davis)
Title: Sociolinguistic Perceptions of Syntactic and Lexical Variation Among Persian-English Bilingual Speakers
Abstract: This study examines the relationship between sociolinguistic perception and variation of Persian syntactic and lexical features by three personae: (1) heritage speakers, (2) native speakers, and (3) L2 learners. Prior work has shown that preconceived notions about how speakers use language and what kind of language they produce can affect listeners’ perceptions (D’Onofrio, 2016; Hansen Edwards et al., 2019; Mack & Munson, 2012; Niedzielski, 1999; Strand, 1999; Staum-Casasanto, 2012). However, many questions remain unanswered in regard to how social meaning is applied in contact situations, especially among self-identified native and heritage speakers. Native and heritage Persian-English bilingual speakers completed an online study with the task of guessing which fictional characters (designed for the three aforementioned personae) they believed would say certain Persian sentences. This study was ultimately designed to measure social perceptions about language use in terms of how listeners apply social meaning to language use in contact situations. Current preliminary results suggest that non-standard forms of syntactic and lexical variation tend to be assigned to learner personae, while standard forms are generally perceived to be produced by native and heritage speakers. Results also suggest that Persian speakers tend to rely relatively more on lexical variation as a way to gauge speaker background. Future work can examine attitudes associated with these different perceptions as well as the relationship between sociolinguistic variation in production and social perceptual categorization.
Presenter 2: Jennifer Dibbern (Doctoral student, Northwestern University)
Title: Influence of raciolinguistic expectations on phoneme categorization in Spanish-English bilinguals
Abstract: Interactional context has been posited to influence bilingual language processing and control (e.g. Grosjean, 2001; Green and Abutalebi, 2013). The ideological nature of this context has, however, been understudied. While social expectations about a speaker’s identity have been shown to affect how their speech is perceived (e.g. D’Onofrio, 2019), this work has not extensively examined how ideologies related to race may influence language switching and perception in bilinguals. Here, we bring together these bodies of work to investigate how social information may be utilized in bilingual speech perception. In the United States, researchers have documented raciolinguistic ideologies around the use of Spanish, including that it is an indicator of Latinx identity (Rosa, 2016). To test whether racialized expectations of a speaker’s language background triggered language-specific processing strategies, we utilized a visually-primed phoneme categorization task. Thirty early or simultaneous Spanish-English bilinguals from the United States categorized bilabial stop continua (from /ba/ to /pa/) and provided social evaluations of the speakers. Results demonstrate that, while raciolinguistic evaluations appeared to influence bilingual speech perception, they did not work the same way for every voice. Both the photograph shown (p = 0.027) and voice heard (p < 0.001) significantly affected how the continua were categorized. Surprisingly, the effect of photograph ran opposite the expected direction. When participants were shown a photograph of an individual likely to be racialized as White, they exhibited a more Spanish-like identification boundary than when they were shown a photograph of an individual likely to be racialized as Latinx. Overall, these results suggest that bilingual speech perception may involve the integration of multiple cues (both acoustic and social), and that the influence of raciolinguistic ideologies on speech perception is not straightforward or homogeneous but instead contingent on complex aspects of the perceived speaker.
Presenter 3: Jeremy A. Rud (Doctoral student, University of California, Davis)
Title: How should a refugee sound? "Hearing" racialized personae in asylum proceedings
Abstract: Scholars have long studied the discursive origins of states’ and publics’ mistrust of asylum seekers (Daniel & Knudsen, 1996; Eades, 2005; Smith-Khan, 2017). Little research, however, has specifically sought to identify the role of the sociophonetic minutiae of asylum seekers’ narrative performances in these broader discourses of mistrust. In this study, I unearth micro linguistic details that can have life-or-death consequences for asylum seekers by investigating the following general research question: How do intelligibility and perceived origin of accented Englishes, as well as perceived emotional affect, influence perceptions of authenticity, credibility, and trustworthiness in asylum seeker narratives? To do so, I used sociophonetic perceptual tasks to elicit 25 listeners’ credibility evaluations of five audio-recorded asylum seeker narratives from YouTube: three performed by actors in Standard American, Standard Australian, and Arabic-accented English, and two performed by “authentic” asylum seekers in Arabic-accented English and English accented by an indeterminable African language. I then used qualitative discourse analysis to examine the relationships between intelligibility, affect, and raciolinguistic perceptions of national origin in the listeners’ responses, ultimately analyzing the constellation of social and perceptual cues that constitute “credibility” in asylum narratives. Preliminary results show that participants have more rigid sociophonetic expectations of authentic emotional performances when evaluating credibility of refugee narratives, than they have of the race or national origin of the speaker. Overall, this study extends the understanding of what asylum seekers are expected to say by determining how they are expected to sound and reveals the extent to which “credibility” in asylum seeker accounts is as much an issue of our ability to “listen” as it is about their ability to narrate, and as much an issue of us confronting our negative biases, as it about their tellings.
Parallel Session 1B: Psycholinguistics
Moderator: Hannah Brown • Co-moderator: Beverly Cotter
Presenter 1: Isabella Duan (Undergraduate student, Stanford University) and Bria Long (Postdoc, Stanford University)
Title: Do labels help children form more specific categories?
Abstract: Language learning requires infants to map words to categories. The word ‘dog’ can refer to both a dalmatian and a dachshund but not a cow (Markman, 1990). How does the developing child come to understand what falls under the ‘dog’ category, as opposed to the ‘cow’ category? Children may use labels, as opposed to other types of auditory signals, to form mental representations of categories (Balaban & Waxman, 1997). But do labels change the specificity of the categories children form? To examine this question, we first familiarized 9 to 18-month-olds to diverse photographs of objects from the same category while they heard labels (“Look at the toma!”) or tones. Afterwards, infants were shown more objects from this category next to objects from perceptually similar, somewhat similar, or very dissimilar categories, with no auditory accompaniment. Then, on word-learning trials, infants were shown these same object pairings again, while hearing labels. If labels helped infants form more specific categories, infants who heard labels should have looked more towards the novel categories. Using Lookit, an online data collection platform, we recruited 40 infant participants, 33 of whom had usable video data (M=13.3 months, Labels N=15, Tones N=18) and analyzed novelty preferences using linear mixed-effect models. In contrast to prior work, we did not find a significant effect of familiarization condition (Labels vs. Tones, t=0.53, p=0.598). Similarity of the familiar and novel categories had a numerically larger effect but was also not significant (t=-1.02, p=0.311). However, on word-learning trials, infants who had heard the labels during familiarization looked more towards the familiar category (t=2.2, p=0.03). Our ongoing data collection seeks to replicate these results using a larger sample size.
Presenter 2: Leonardo Zeine (Master’s student, University of São Paulo)
Title: Syntactic and semantic interference on the processing of disfluent sentences
Abstract: We evaluate through an extensive online self-paced listening experiment the processing of disfluent sentences such as I want you to put uhh drop the ball. Ferreira et. al (2004) presents a model for such sentences, by which the verbs before and after the disfluency are structurally overlaid, which results in a complex mental structure with properties inherited from both verbs. It follows from the Overlay operation that the previous sentence is less accepted than its grammatical counterpart, I want you to drop the ball, and also that the sentence I want you to drop uhh put the ball is more accepted than its agrammatical counterpart I want you to put the ball. Based on this model, we extend the experiment for Brazilian Portuguese sentences with verbs that differ in: categorial selection — such as to love/to sympathize (with) — prepositional lexical selection — such as to trust (in)/to sympathize (with) — and number of arguments, as the put/drop pair showed above. Also, for each of these three syntactic conditions, we evaluate two semantic conditions: consonant and dissonant meanings. By that, we want to analyze the disfluency processing when the two overlaid verbs have the same sentential meaning (that is, when the speaker wants to change the form of the sentence but not its content) and when the disfluency implies a complete change of meaning. Our results show that consonant meaning is a condition for the operation of Overlay to be placed, which raises questions about mental structures, pragmatic effects and semantic disposition of grammatical constructions. The intuition behind the extensive approach is that if two mental structures are overlaid during disfluencies, we can actually ask ourselves what is exactly that is being overlaid, i.e. which is the form of mental structures.
Presenter 3: Beverly Cotter (Doctoral student, University of California, Davis)
Title: The Impact of Working Memory and Language Experience on the Processing of Attachment Ambiguities
Abstract: Introduction. Working memory (WM) has been shown to influence how readers make decisions about syntactic ambiguity (e.g. Swets et al., 2007; James et al., 2018). In the 2007 study, analyses demonstrated that participants with low WM spans demonstrated an overall high attachment bias when processing relative clause (RC) ambiguities. However, there remains unanswered questions about language experience’s impact and whether such findings could extend to other forms of syntactic ambiguity such as the prepositional phrase (PP) attachment ambiguity. Thus, we aimed to conceptually replicate the 2007 study to explore these questions. Method. Across two studies, Spanish and Mandarin stimuli were presented to 225 native Spanish and 246 Mandarin speakers. In each study, participants completed the following tasks: two WM tasks (reading and spatial span), an attachment task (100 items: 20 experimental, 80 filler items; study 1 tapped into RC attachment preferences, study 2 tapped into PP attachment preferences) and a language history questionnaire (LHQ 3.0) as a self-reported measure of language experience. Results. In Study 1, logistic regression analyses demonstrated that although neither WM measure significantly predicted RC attachment response in the Spanish sample, both WM measures were significantly and negatively related in the Mandarin sample (β = -0.76, p = .034 ; β = -0.58 , p = .043). In addition, study 1 found that L1 proficiency was a significant predictor of high attachment response in the Spanish sample (β = .28 , p < .01). Lastly, our preliminary analyses demonstrate clear differences between RC and PP attachment across our two samples (see figures 1 and 2). Conclusions. The results from study 1’s Mandarin sample are consistent with the findings of Swets et al. (2007). Additionally, L1 proficiency appears to modulate high attachment for our Spanish speakers, supporting previously observed cross-linguistic differences in attachment preferences (e.g. Cuetos & Mitchell, 1988). However, study 2 brings forth potential differences that exist between RC and PP attachment processing. Potential variables that may have an impact on RC attachment processing may not extend to other forms of ambiguities such as the PP attachment ambiguity. Thus, the findings from our two studies underline the importance of further studying the PP attachment ambiguity and the differences between RC and PP constructions.
Parallel Session 1C: Language Teaching & Learning
Moderator: Lillian Jones • Co-moderator: Sophia Minnillo
Presenter 1: Ekaterina Sudina (Doctoral student, Northern Arizona University)
Title: Methodological synthesis of second-language anxiety and motivation questionnaires: A call for reform
Abstract: Unlike meta-analysis that typically focuses on aggregating effect sizes, methodological synthesis is a type of systematic review that surveys methodological practices rather than substantive results (Cooper, 2016). Among existing systematic reviews of second-language (L2) anxiety and motivation, meta-analyses appear to be prevailing (e.g., Al–Hoorie, 2018; Teimouri et al., 2019). To address concerns that have been raised about the psychometric properties of L2 questionnaires (Al–Hoorie & Vitta, 2019; Gu, 2016), the current study systematically examines changes in L2 anxiety and motivation scale quality over a decade by comparing scales from articles published in 2009–2014 with those published in 2015–2019. A total of 104 peer-reviewed articles (113 independent samples) that used 340 L2 anxiety and motivation scales and were published in five leading L2 journals in 2009–2019 were coded for 82 features pertaining to questionnaire quality. Results revealed that relative to measurement instruments that appeared in articles published in 2009–2014, scales from the 2015–2019 period were (a) slightly more often accompanied by a reliability coefficient (74.1% vs. 69.7%), (b) similarly reliable overall (M = .81 [95% CI: .80, .83] vs. M = .82 [95% CI: .80, .83]), (c) more often consisting of a single item (an increase from 1.3% to 2.7%), (d) less frequently evaluated by expert judges/panels (a decrease from 10.3% to 4.9%), (e) more often examined using factor analysis procedures (an increase from 35.5% to 50.8%), and (f) less frequently accompanied by a validity reference (a decrease from 12.9% to 11.4%). Overall, these mixed results demonstrate that there are multiple areas in need of methodological enhancement for psychometric instruments in L2 survey research. In light of methodological reform movement (Plonsky, 2014), suggestions for improving L2 scale quality and expanding efforts to examine the role of anxiety and motivation in L2 development are discussed.
Presenter 2: Hamoudi Benarba (Doctoral student, University of Béchar)
Title: The Role of Using Allegories in Enhancing EFL Learners’ Critical Reading Skills
Abstract: Proficient reading in any foreign language can be achieved once learners’ needs and interests are methodically met with an appropriate input. However, reading comprehension is usually given the lion’s share in classroom activities while more important skills such as critical reading are set to the periphery. Using literary texts such as allegories which can represent a more engaging type of reading input that can meet students’ needs and help them hone their critical reading skills and strategies has been also neglected. Therefore, a study has been conducted in an attempt to investigate students’ attitudes towards the use of literary allegorical texts in enhancing their critical reading skills and strategies. To carry out this study, a questionnaire was conducted and administered online owing to its practicality and convenience. It was addressed to 60 fifth-year students in the Department of English at the Ecole Normale Supérieure de Constantine_ Algeria. A quantitative and qualitative analysis of the data showed learners’ positive attitude towards allegories as reading materials with an overall consensus on such texts’ potentiality in triggering and fostering critical reading skills and strategies. The outcomes of this study may offer theoretical and practical insights with regard to incorporating literary allegorical works in EFL university reading courses.
Presenter 3: Sylvia Chaiyeon Lee (Doctoral student, University of San Francisco)
Title: Strategy-Based Instruction: The Effect of Curriculum-Integrated Explicit Strategy Instruction
Abstract: Language learning requires a set of appropriate learning strategies. Language learning strategies (LLSs) are specific thoughts and actions taken by language learners to improve the process of learning a language, consciously, subconsciously, or automatically (Cohen, 2014; Griffiths, 2007; Oxford, 2003; 2011). Language learning strategies (LLSs) provide students with tools for active and meaningful involvement in gaining language skills and reveal what students do in the process of learning a language, such as generating rules, learning from errors, and establishing mental schemata (Griffiths, 2014; Thompson, 2005). The use of language learning strategies helps students to learn a language better directly and indirectly and enables become more independent, autonomous, and lifelong learners by regulating and controlling their learning. Hence, the learning process is faster, more accessible and effective (Cohen, 2014; Oxford, 1990; 2011). Each learner has a unique set of general learning strategies that can be useful for their specific learning conditions. However, the successful strategy use depends on how and when learners use these strategies through the guidance of teachers rather than the mere knowledge of it. Strategy instruction can bring systematic scaffold and guidance into a language learning process, let learners discover the positive effects of language learning strategies, and promotes appropriate cognitive processing during learning. For this reason, incorporating learning strategy instruction into the curriculum has been gaining increasing recognition and used with growing frequency as a desirable learning and teaching method (Agee & Hodges, 2012). Grounded in cognitive psychology, Mayer’s (2014) select-organize-integrate (SOI) framework, this presentation suggests effective learning strategy instruction and how to implement them in ESL classrooms. Through the SOI process, learners not only actively engage in meaningful learning but also take ownership of their learning by employing appropriate cognitive and metacognitive processing during learning.
Most Innovative Research Panel - Undergraduate
Moderator(s): Skyler Reese & Lillian Jones • Panel Judges: Georgia Zellou, Fernanda Ferreria, Carlee Arnett
Presenter 1: Irene Yi (Undergraduate student, University of California, Berkeley)
Title: Using Sociolinguistic Factors to Improve Language Technologies in Predicting Mandarin-English Codeswitching Through Classification and Regression Trees (CART) and Random Forest Models
Abstract: Current research on computational modeling of codeswitching has focused on the use of syntactic factors and constraints as model predictors. This thesis demonstrates the value of incorporating sociolinguistic factors as predictors into computational models of codeswitching, as well. The study presented here is focused on Mandarin-English codeswitching, drawing on novel data collected from 12 bilingual speakers from Grand Rapids, Michigan. These speakers come from two generations, correlated with their age and immigrant generation. These speakers participated in sociolinguistic interviews that were designed to elicit codeswitching in narrative-style responses on a variety of topics, including family, school, and culture. Participants also answered metalinguistic questions about their own language practices and attitudes and completed a written Language History Questionnaire (LHQ) (Li et al. 2020), which asked for self-evaluations of language habits (proficiency, immersion, and dominance in the two languages). LHQ responses were then quantified into “scores” that served as sociolinguistic predictors in a Classification and Regression Tree (CART) model, along with age, education level, and sentence length. The CART model achieved an accuracy of 0.804 with the area under its ROC curve being 0.692. This is comparable to, if not more powerful than, previous computational studies (e.g. Li & Fung 2014) that trained models using proposed syntactic constraints as predictors. In the future, sociolinguistic factors could be used in conjunction with syntactic predictors in computational modeling to achieve a better understanding of codeswitching patterns. Such patterns found in this novel Mandarin-English data frequently, and potentially systematically, violate many of the currently proposed syntactic constraints on codeswitching (which mainly come from research on Spanish-English bilinguals), implying that the constraints may not be universal, and that new avenues should be considered for understanding the morphosyntax of bilingual codeswitching.
Presenter 2: Christy Lei (Undergraduate student, Reed College)
Title: Labeling emotions in a native and foreign language: An ERP study on emotion regulation in bilinguals via affect labeling
Abstract: Recent research on the Foreign Language Effect (FLE) suggests that bilinguals tend to experience a larger attenuation of emotion in their second language (L2) than in their native language (L1). Clinically, the use of L2 has been proposed as an emotional detachment tool for bilingual patients to distance themselves from negative emotions. However, to date, there is very little experimental evidence that compares the neural dynamics of L1 and L2 use in emotion regulation. The current study used the event-related potential (ERP) technique to investigate the regulation of negative emotional faces in L1 and L2 via “affect labeling,” an emotion regulation technique proved to down-regulate emotions. Chinese-English bilinguals completed the experiment under three conditions (passive viewing, labeling negative emotions in Chinese/L1, and labeling emotions in English/L2), while their brain activity was recorded and time-locked to the onset of displayed emotional faces. Early, middle, and late ERP components (N170, EPN, LPP) were analyzed and compared across conditions. Results revealed that, labeling emotions in L2, but not in L1, decreased the amplitude of the early components (N170, EPN) compared to passive viewing, suggesting a reduction in the automatic attentional allocation to emotional stimuli in L2. For the later component, L2 labels had no impact on the LPP amplitude, and L1 labels increased LPP, implying a heightened processing of emotions in L1. Further, frequency of L2 use in daily life (but not L2 proficiency), predicted ERP differences at the later stage, such that lower usage of L2 led to a more reduced LPP in the L2 label condition relative to passive viewing and L1 label condition. Overall, the current study confirmed the FLE phenomenon in non-dominant bilinguals’ emotion regulation. In particular, it suggests that L2 frequency of usage has an important influence on bilingual’s emotion regulation outcome.
Presenter 3: Amari Grey Johnson (Undergraduate student, Harvard University)
Title: From No-body to Un-bod(ied): Black Trans Social Death & Digital Grammar
Abstract: Highlighting popular Black trans media patterns and digital grammars, I consider how Twitter’s linguistic capabilities innovate Black oral traditions materially – how Black trans virtual, speech communities engage a fugitive space from the double “social death” of race and gender, the human and the body: Do queer speakers modify black English against racial and gender limitations? How does online communication transform these strategies? How does this virtual “translation” destabilize the body/gender, as a proxy for Language itself? Social death (Wilderson) is negative Blackness – perpetual fungibility, available for (re)inscription; genealogies of Language reveal social death as a linguistic void. The slave, stripped from speech, is forced into legibility as Black. In the development of black English vernaculars, the enslaved thwart linguistic disconnection, making Blackness fungible to itself; in sensory social spaces, this re-appropriation exploited audiovisual faculties such that black togetherness approaches a metalinguistic role, “frequency” (Campt). Blackness escapes Language limits, engaging the “semiotic-as-liberation”, turning heterogenous social arenas into embodied frequencies of transformation. Racialized social media spaces parallel discursive potentialities, extending black verbal semiotics into virtual fugitivity. The digital “profile” supersedes the “real” body, and operating in bodily absence, trans users mitigate dysphoria and perception. A semiotic strategy, the exploitation of racialized and gendered audiovisual cultures populates a novel lexicon. “Digital emphatic blackness” (Krystal A. Smalls), the black assemblage of rhetorical functions beyond analog speech, materializes “black Twitter” - a political, temporal subjectivity (Brock). The metalinguistic effort is intrinsically queer, agentively representing bodies and the centrality Black trans women’s cultural production. Assembling a Black trans “voice” induces gender affirmation, enjoying the non-bodied digital manifestation, and as users find agency navigating the public spheres, the non-bodied voice becomes the un-bodied, and gender affirmation approaches gender abolition. By destabilizing gender in the digital grammars, Black trans folks move from a sensory metaalanguage towards frequency, post-language.
Parallel Session 2A: Bilingualism
Moderator: Peter Torres • Co-moderator: Tyler Kline
Presenter 1: Meghan Elliott (Master’s student, University of Southern California)
Title: Self-rated daily language use, rather than self-rated bilingual language competency, moderates the effects of brain on cognition in a Hispanic cohort
Abstract: Background: Studies suggest that bilingualism may be a protective factor in cognitive aging, but varying definitions of bilingualism have led to conflicting results. To better understand the interrelations of linguistic competence, cognition, and the brain, we examined bilingualism as knowledge of two languages, and also as habitual practice patterns (i.e., predominantly English or Spanish, or daily use of both). Here, we investigated the effects of these language variables on cognitive outcomes of baseline and longitudinal executive function and episodic memory, incorporating brain measures, age, gender, and education, and also whether they moderated the relationships between cognition and brain integrity. Methods: A baseline Hispanic cohort (N = 100) with a single MRI scan and a longitudinal subset (N = 73) were sampled. Linguistic measures were self-reported bilingualism, and among bilingual participants, self-reported daily language use. Regressions modeled baseline and longitudinal executive and memory with brain signature regions and language variables, including terms for interaction of language variables and brain, covarying with age, education, gender -- separately for bilingualism in the entire cohort and then for language use restricted to bilinguals. Results: Bilingual knowledge itself did not significantly predict baseline or longitudinal cognition. For daily language use among bilinguals, English-only speakers showed significantly stronger association of baseline brain structures for memory compared to Spanish-speaking or multi-language bilinguals. Longitudinally, predominantly English-speaking bilinguals had significantly increased association of atrophy signatures with both cognitive changes compared to Spanish or speakers of both languages. Therefore, consistent language use appears to have a protective effect, but language knowledge does not. Conclusion: Results suggest a complex pattern of bilingualism’s effect on cognitive aging, depending on whether bilingualism was defined as knowledge or by daily language use. These results suggest that careful consideration of how bilingualism is defined is paramount in elucidating the role of bilingualism in cognitive aging.
Presenter 2: Lauren Salig (Doctoral student, University of Maryland)
Title: Modulating Bilingual Switch Costs with Cognitive-Control Engagement
Abstract: Research on the relationship between bilingualism and cognitive processing often compares bilinguals’ and monolinguals’ performance across tasks to draw conclusions about how a lifetime of language experience affects stable executive function traits. However, many researchers note drawbacks to the group-comparison approach: (1) treating bilingualism as a category rather than a continuum and (2) using tasks with low test-retest reliability to measure supposedly stable traits. In this study, we circumvent such issues by comparing bilinguals to themselves under different linguistic and cognitive contexts to ask how fluctuating cognitive states modulate language processing. Specifically, we ask whether engaging cognitive control—an executive function that helps resolve conflicting representations—improves bilinguals’ ability to interpret a code-switch (a mid-sentence switch between languages). Many studies demonstrate that bilinguals take longer to process code-switched versus single-language content (e.g., Altarriba et al., 1996; Bultena et al., 2015; cf. Johns et al., 2019), and some studies suggest that mixed-language or code-switched content promotes bilinguals’ cognitive-control engagement to resolve conflicting representations (Adler et al., 2020; Wu & Thierry, 2013). However, it remains unclear if bilinguals can reduce the longer processing time needed for a code-switch when they are already in a state conducive to resolving conflict. We dynamically manipulated Spanish-English bilinguals’ cognitive-control engagement (using Flanker-arrow trials) to observe the effect on their self-paced reading of a subsequent sentence that was either code-switched or in one language. By using this pseudorandomly interleaved Flanker-sentence design, we are able to make inferences about how bilinguals do or do not draw on cognitive states to regulate language comprehension from moment- to-moment. In preliminary results (N=57), cognitive-control engagement did not affect ease of code-switch processing. Yet, emerging trends in the data suggest the relationship may depend on code-switch type, highlighting the need to consider flexible recruitment of cognitive control that varies with language processing demands.
Presenter 3: Zunaira Iqbal (Doctoral student, University of California, Merced)
Title: How Speech-Shaped Noise Influences Phonetic Perception in Monolingual and Bilingual Listeners
Abstract: Speech perception in noisy situations is difficult, especially when trying to comprehend one’s second language in the presence of background noise. Here, we examined how background noise impacts speech perception along a continuum from /va/ to /ba/ in Spanish-English bilinguals and English monolinguals. The /v/ and /b/ phonemes are of key interest since Spanish phonology maps /v/ onto /b/, whereas English makes a clear distinction between these two sounds. This study also aimed to examine how individual differences in the bilinguals’ language profiles (e.g., age of English acquisition) modulate their speech perception along the /va/ to /ba/ continuum. Five consonant-vowel (CV) stimuli were extracted along a continuum from /va/ to /ba/. On each trial, one CV stimulus was presented at one of four signal-to-noise ratios (SNRs): Quiet, +6 dB, 0 dB, and -6 dB. Participants reported whether they perceived each stimulus as /ba/ or /va/. Due to differences in Spanish and English phonology regarding /b/ and /v/, we hypothesized that Spanish-English bilinguals would be biased towards perceiving /ba/, especially for the more /va/-like stimuli in noisier conditions, compared to English monolinguals. Although data collection is ongoing, preliminary results show that the presence of noise, even at +6 dB SNR, substantially affected the perception of /ba/ and /va/ in both Spanish-English bilinguals (n=19) and English monolinguals (n=5). Surprisingly, in the Quiet condition, the bilinguals perceived /va/ more often along the continuum, on average, compared to monolinguals. Moreover, in Quiet, the bilinguals who learned English before age 5 years had more robust /ba/ perception for the more /ba/-like stimuli, than the bilinguals who learned English after age 5 years. These preliminary results suggest that even when listening in quiet, the presence of the English phoneme /v/ may result in a phonetic recalibration process that shifts perceptual phonetic boundaries in bilingual listeners.
Parallel Session 2B: Language Identity & Policy
Moderator: Lillian Jones • Co-moderator: Daniela Cerbino
Presenter 1: Rebecca Ash-Cervantes (Master’s student, California State University, Los Angeles)
Title: Language Identity and Economic Outcomes in Spanish Bilinguals
Abstract: Studies analyzing bilingualism’s benefits have often looked at bilinguals through the individual facets of their lives, looking at their earnings separate from their persons as statistics (Shin & Alba, 2009), or looking at their language usage from a theoretical angle, with a disregard for concrete reasons, such as wage, that a speaker may consider when making the decision to use language (Amos, 2016). Research has also shown that bilingual Spanish speakers in the US traditionally earn less than their monolingual English-speaking counterparts (Gandara, 2015). This study aimed to provide holistic view of bilinguals, taking both wage and language identity into consideration when looking at the decision to use language in the workplace. This qualitative study looks at the relationship between language identity (Norton, 2019; Peirce, 1995) and wage among 11 Spanish bilinguals in 3 US cities in the United States. The study implemented chi-squared test to measure wage compensation and one-to-one interviews to elicit speaker’s perceptions of bilingual identity in the data collection process. Interview were transcribed and coded for themes using grounded theory and Nvivo. The research surveyed the speakers’ identity in the context of whether their language identity affected their inclination to use their language for perceived economic gain. Overall, the research found that Spanish bilinguals are often asked or inclined to use their bilingual abilities in the workplace without compensation due to need from the employer, a type of cultural burden (Padilla, 1994). In addition, speakers are moved to use Spanish in the workplace by a perceived usefulness from the identity aspect of belonging to the Spanish-speaking Latinx community. This research suggests that it is the combination of work and social pressures that leads to Latinx’ using Spanish in the workplace place, even though they are undercompensated. The presenter will describe the study and suggest further research.
Presenter 2: Laura Brass (Doctoral student, The University of British Columbia)
Title: Autoethnography as a Window into an Immigrant Female Teacher’s Identity
Abstract: Context. While a growing body of literature investigates issues related to ESL learners (Dagenais, 2020; Higgins, 2015; Kramsch, 2009; Norton, 2019), more research needs to be conducted on immigrant ESL female teachers in Canada (Amin, 2001; Walsh et al., 2011), which is important in terms of how they deal with tensions associated with relocating, competing for the same jobs as native-born professionals, and finding a sense of belonging and a meaningful place in mainstream society. To address these issues, I analyzed the lived experiences of a bilingual teacher identity development and transformation before and after immigrating to Canada. Drawing on new materialism (Barad, 2007; Bennett, 2010), which theorises that material objects are holistically connected to life and identity, and language teacher identity (Barkhuizen, 2017; Block, 2007), which conceptualizes the convergences between personal and professional identity, this study explored how past and imagined future experiences shape the teacher’s identity and how material objects orient one’s sense-of-self in unfamiliar environments. Methodology. I employed a qualitative research design (Wolcott, 1994) in the form of autoethnography (Canagarajah, 2012; Yazan, 2019) and generated data through stories of lived experience descriptions (Van Manen, 2016) and artifacts as identity texts (Cummins et al., 2015). First, I analyzed data chronologically (i.e., significant life history events), then thematically (i.e., significant identity transformations) (Menard-Warwick, 2008; Seidman, 2013), and I connected the teacher’s lived experiences to the larger Canadian social context (Creswell, 2014). Multimodal discourse analysis guided my open-ended line-by-line text analysis and interpretation (Bryman & Bell, 2019; Rose, 2016). Results. The proposed backward and forward cycle of identity brought lived experiences into reflective awareness, prompting a holistic awareness of the self. The teacher’s co-existence with (in)animate objects revealed tensions in her identity development across time and space. The stories mirrored three main stages in her identity transformation (e.g., Symbolic, Material, and Digital Objects), which speaks to an evolving relationship with the human and non-human world. Significance. The past-present-future cycle of identity is threefold: 1) as a research framework, it allows participants to (re)imagine their future identities; 2) as a pedagogical tool, it incorporates (auto)ethnographic data into educational programs, empowering students and teachers; 3) as a creative process, it reveals alternative ways of being in/with the world, which pertains to immigrant teacher and student identity development and transformation.
Presenter 3: Lillian Cheeks (Undergraduate student, University of California, Santa Barbara)
Title: Producing Humor by Portraying Asymmetries in Racial Common-Sense
Abstract: Early ethnomethodological and conversation analytic research demonstrated the systematic moral enforcement of common-sense knowledge both in relation to everyday action and inference in general, and more specifically with respect to category-based expectations, entitlements, and obligations. This study builds on this pioneering research, and on related studies of both race and humor, by considering how television comedy writers and actors exploit these features of social organization in the service of portraying humorous instances of racial ignorance. The recordings were collected from shows in which race is consistently topicalized (“racial comedies”) as well as those in which it is made relevant more fleetingly while not being a consistent topical theme. Through repeated viewing of the recordings, I identified a range of ways in which common-sense knowledge associated with particular racial categories was deployed and/or subverted in the portrayed interactions. The data set consists of an interactional exchange from Saturday Night Live’s “Lunch Run” sketch from Season 45 Episode 6 . My analysis uses a Conversation Analytic approach to examine how characters can be portrayed as acting, or intending to act, laughably incompetently on the basis of deficiencies in common-sense knowledge about people, places, and/or objects associated with particular racial categories. For example, in Saturday Night Live’s “Lunch Run” sketch a white British intern’s plans to retrieve lunch from a Popeye’s on Frederick Douglass Boulevard are quickly interrupted by his Black coworkers, who display horror at his lack of realization that a white person such as himself would be seen as being extremely out of place in a Black-identified restaurant on ahearably Black street. Thus, writers and actors of these scripted interactions construct humor by portraying asymmetries of category-based knowledge between characters of different racial categories. For viewers who recognize these types of asymmetries in knowledge from their own real-world observations and experience, their parodied portrayal on screen can offer ways of converting actions that might be problematic in real-world settings into occasions for humor. Conversely, these portrayals may serve as cautionary tales for viewers who lack awareness of their own similar deficiencies in category-based knowledge, thereby serving as a mechanism for educating uninformed viewers in relation to this knowledge.
Parallel Session 2C: Syntax
Moderator: Ana Ruiz • Co-moderator: co-moderator: Sophia Minnillo
Presenter 1: Amor Leal (Master’s student, California State University, Northridge)
Title: Agreement in Hixkaryana person portmanteaux: An analysis
Abstract: Person portmanteaux are morphemes that realize the person features of both arguments of a transitive verb simultaneously, spelled out postsyntactically in a single morphological unit. Person portmanteaux are an element of Hixkaryana, a Carib language spoken in the Amazon region in Brazil. Kalin (2014) proposes that Hixkaryana person portmanteaux agree via Specifier-Head (SPEC-HEAD) agreement (Chomsky, 1986) and that the person probes which agree with both subject and object are found on v. This study contends that SPEC-HEAD as a theory of agreement is inadequate as it fails to explain there existentials in English and quirky subject constructions in Icelandic (Schutze, 2019), and that AGREE, as propounded by Chomsky (2000, 2001) is a necessary improvement upon SPEC-HEAD. Therefore, AGREE is able to more sufficiently explain person portmanteaux agreement. Under an AGREE analysis, then, the person probes in Hixkaryana rather reside in finite Tense and search downward into their c-command domain to agree with subject and object, contrary to what Kalin suggests. Baker (2013) explains that if subject and object agreement in a language is a result of finite Tense agreeing twice with both arguments, as opposed to finite Tense agreeing with subject and v agreeing with object, then to distinguish such a language requires examining nonfinite or nominalized contexts where it is possible that object agreement was lost due to its being a function of finite Tense and not of v. Indeed, this is what we find in Hixkaryana. This study demonstrates that in non-finite/nominalized contexts, agreement between Hixkaryana verbs and their arguments in their respective clauses is lost, shown in the absence of portmanteau morphemes indicating subject and object agreement. This is apparent evidence that the person probes which lend to the structure of portmanteau morphemes in Hixkaryana do reside in finite Tense and not v.
Presenter 2: Cristina Ruiz-Alonso (Doctoral student, Universitat Autònoma de Barcelona)
Title: El preceding embedded clauses: a phenomenon of variation in Spanish and other Romance languages
Abstract: The aim of this paper is to present a phenomenon in Spanish that has not received much attention in the literature: el(que)ísmo. It consists in the optional insertion of the definite article el preceding embedded clauses (finite and non-finite ones). a. El que diga esas cosas es preocupante b. El decir esas cosas es preocupante theEXP that says-subj those things is worrying theEXP say-inf those things is worrying ‘The fact that s/he says those things is worrying’ ‘Saying those things is worrying’ In so doing, I describe the phenomenon and provide an overview of this phenomenon’s properties (Leonetti 1999; Serrano 2015; Picallo 2001) concerning the distribution (the preferred position is preverbal subject), the verbal mood (el is inflected in subjunctive), island effects and semantic constraints (some semantic classes of predicates reject el). Owing to the island status and the reluctance to show up in a first-Merge position, EL(Q) appears to be a spell-out of a Case morpheme, not a true determiner. I take such case to be of the inherent type. For concreteness, I assume the following structure, with “el” occupying the K head: (7) [K (el) [C[ . . . ]]] This work also shows the data in other Romance languages, where the phenomenon seems more restricted, but attested in non-finite clauses (8): (8) a. Il dire queste cose può risultare offensivo [Italian] b. El dir aquestes coses pot ser ofensiu [Catalan] ‘Saying those things may be offensive’ To conclude, this paper investigates EL(Q) in Spanish. It has been argued that the definite article is the spell-out of a Case projection, which I have taken to be inherent. The proposal makes a series of empirical predictions concerning the incompatibility between EL(Q) and complement (first-Merge) positions and accounts for this phenomenon in other Romance languages.
Presenter 3: Anjie Cao (Doctoral student, Stanford University) and Molly Lewis (Research Scientist, Carnegie Mellon University)
Title: Quantifying the syntactic bootstrapping effect in verb learning: A meta-analytic synthesis
Abstract: Syntactic bootstrapping refers to the phenomenon that young children can rely on the syntactic information in the sentences to infer the meaning of a novel verb (Naigles, 1990). For example, if a child hears the sentence “The bunny is gorping the duck”, she can rely on the fact that “gorping” is embedded in a transitive structure to infer that it refers to the action that the bunny acts on the duck. In the last decades, a range of empirical studies have investigated the syntactic bootstrapping phenomenon using a variety of testing materials on procedures on participants of different ages. Little is known, however, about the size of the effect and its robustness to various theoretical and methodology moderators. In the current work, we address this gap by using meta-analytic methods to synthesize across 60 experiences in the syntactic bootstrapping literature (N = 849) participants). We find that, despite its theoretical prominence, the syntactic bootstrapping effect is relatively small (d = .24[0.03, 0.44], does not strengthen over development (β = -0.01 [-0.03,<.001], SE = 0.01, z = -1.47, p = 0.14), and its present only for studies that use transitive sentences (β=0.24 [0.02, 0.46], SE = 0.11, z = 2.10, p = 0.04). None of the methodological factors examined had a significant influence on the effect size. Further, we compared the effect size of syntactic bootstrapping to other word learning phenomena. We find that the syntactic bootstrapping effect size is similar to cross-situational learning and sound symbolism, but appreciably smaller than mutual-exclusivity and gaze-following. Our findings have implications for verb learning theories.
Most Innovative Research Panel - Graduate
Moderator: Skyler Reese & Lillian Jones • Panel Judges: Georgia Zellou, Fernanda Ferreria, Carlee Arnett
Presenter 1: Andrea Ramirez Barajas (Doctoral student, University of California, Davis)
Title: Look it’s un Perro! Infant-Directed Speech With Spanish-English Bilingual Parents
Abstract: This study aims to understand how bilingual parents adjust their speech when talking to infants. Infant-directed speech (IDS) characteristics include higher pitch, wider pitch range, and longer duration compared to adult-directed speech (ADS). However, most IDS research has examined monolingual samples;thus, understanding of bilingual IDS is limited. We investigated acoustic characteristics of IDS and ADS with Spanish-English bilingual parents and their infants (8-20 months). 40 dyads participated in naturalistic tasks using 10 common objects. For bilinguals (N = 20), objects were labeled in Spanish and English, while for monolinguals only in English. The researcher and parent interacted with the same toys to collect an ADS sample.Using Praat, we extracted the fundamental frequency (f0), a measure of pitch, for all labeled objects (2835 tokens) from bilinguals and from monolinguals (2105 tokens). We examined whether bilingual and monolingual parents make similar acoustic adjustments in pitch when producing IDS versus ADS. For bilinguals, a within-subjects ANOVA of language (English vs. Spanish) and register (IDS vs. ADS) tested whether there was a difference in pitch IDS versus ADS across languages. There was an overall effect of register: IDS had higher pitch than ADS (p < .001), but no language interaction (English mean f0 = 23 Hz; Spanish mean f0 =22 Hz). Collapsing languages, an ANOVA examined language groups (Monolingual vs. Bilingual) and register (IDS vs. ADS) to test whether bilinguals and monolinguals differed in pitch. Across groups, IDS was higher than ADS and bilinguals’ pitch was higher than monolinguals’ pitch overall (both p < .001). However, there was no interaction with language groups. Ongoing analyses will examine moment to moment changes in pitch within each register (IDS vs. ADS) to explore if bilingual parents adjust pitch across repetitions of labels in each language.
Presenter 2: Jennifer Kaplan (Doctoral student, University of California, Berkeley)
Title: "I looked in the dictionary, it doesn’t even exist!": Young Nonbinary Montréalers’ Fight for Linguistic Recognition in a Prescriptivist World
Abstract: Overview: Though in recent years many scholars have written on non-binary English linguistic innovations, particularly neutral pronouns (Baron 2015 ; Bergman 2017; Feinberg 1998), comparatively little has been written on non-binary French forms. The binary quality of French grammatical gender has inspired innovative spoken structures used by non-binary Francophones, including neopronouns and neo-morphemes (Alpheratz 2018; Ashley 2019; Knisely 2021). However, non-binary French is still a nascent variety that has met resistance and at times mockery from the Francophone press. Focusing on the experiences of six non-binary French-English bilinguals in Québec, Canada, this ethnographic sociolinguistic study explores how prescriptivist language norms, gender essentialism, and institutional pressures affect when, where, and how non-binary Francophones feel comfortable expressing their identities in language. Methods: Interviews took place in Montréal during summer 2018. Participants answered questions about how they articulate their identities, as well as how different groups in their community (broadly conceived) react to their use of non-binary French. Building on sociologist Dorothy Smith’s work, I pay particular attention to the ways participants oriented their identities, and their expression thereof, in relation to the institutions (educational, governmental) they encounter in their daily lives. Major Findings: Above all, interviewees’ attempts at linguistic affirmation were undermined by prescriptivist attitudes operating at both personal and institutional levels, which have defined Montréalers’ relationship to the French language since British colonization. This linguistic prescriptivism is simultaneously being undermined and augmented by two different social forces. On the one hand, gender essentialism (Bornstein 1995) works to intertwine social conceptions of binary gender with binary conceptions of biological sex. On the other hand, participants agreed that old attitudes working to preserve linguistic and gender binaries in Francophone Québec are being shaken by young queer people, who are leading the charge in advocating for official recognition of French forms expressing gender diversity.
Presenter 3: Emilie Schiess (Master’s student, Ball State University)
Title: The Language of Women’s Migraine Pain: A Corpus for Improving Pain Assessment
Abstract: Although pain assessment is presented as objective, its language hosts biases and inaccuracies. Sussex (2019), citing Lascartaou and others, identify various faults in the McGill Pain Questionnaire (MPQ): its reliance on adjectives does not align with our natural use of nouns and verbs to describe the pain, and the selected modifiers lack semantic, cultural, and contextual distinction. Jaworska (2018) observes how this can negatively affect women, first by documenting how women’s pain has been dismissed or misinterpreted, and second by creating a corpus to identify how women describe pain differently from men for a variety of ailments. To replicate Jaworska’s approach on a more specific condition, this paper focuses solely on migraine pain given its prevalence and severity among women. Therefore, short narratives from 10 cisgender women about their migraine pain were coded for expressions of pain to determine differences with the MPQ. A previous pilot study with similar methodology confirmed recent literature that pain was described using metaphors of violence and with nouns and verbs more than adjectives and adverbs. The pilot study also suggested women in particular use narratives centered on inconvenience rather than intensity and used language that compared their pain to domestic violence or reflected victimization. This revised study seeks to replicate these results with more depth, adding new focus on the agency of sentences (do women feel in control of their bodies when feeling pain?), the topics discussed (do women emphasize inconvenience and emotion over physical symptoms?), and figurative expressions (do metaphors of violence reference war and weapons or domestic violence?). These results would not only demonstrate how gender shapes expressions of pain, but also indicate the flaws of modifier-based pain assessment. Preliminary trends in women’s language of pain assert this as a linguistic topic with direct social implications.
Parallel Session 3A: Text Processing & Computational Linguistics
Moderator: Sophia Minnillo • Co-moderator: Peter Torres
Presenter 1: Joy Peltier (Doctoral student, University of Michigan)
Title: Bare Nouns and Determiners in Kwéyòl Donmnik: A Corpus-Based and Experimental Study of Discourse-Pragmatic Features
Abstract: The syntax and semantics of Creole nouns has been a topic of intense study (see Baptista and Guéron 2007), but little research addresses the pragmatics of Creole bare nouns and determiners. This study examines Kwéyòl Donmnik (KD), an understudied French-/English-influenced Creole declining in use, focusing on speakers’ usage of bare nouns and of post-nominal determiners -la ‘the/this/that’, -sala ‘this/that’, and -lala ‘this here / that there’. Though typically characterized as a definite determiner or specificity marker (Corne 1999: 133), -la may have a deictic force akin to a demonstrative (Déprez 2007: 269, Christie 1998: 268). However, “no precise empirical test is provided to support this claim” (Déprez 1997: 269). Also, while generic, bare plural, and unique nouns are consistently bare in KD, bare nouns can “express all the meanings that non-bare NPs can” (Gadelii 2007: 234), suggesting that context informs speakers’ choice of nominal expression (Gadelii 2007: 252, Christie 1998: 277). Do speakers employ a bare versus a non-bare noun form based on whether the referent is specific and/or exemplifies a certain information status (Prince 1992)? Do -la and -lala have a deictic force akin to demonstrative -sala? If so, do specificity, information status, deictic dimension, or patterns in co-speech pointing inform speakers’ choice of determiner? Based on an analysis of 761 nominals produced by KD speakers engaging in conversation, picture book narration, and the gesture experiment “Stacks and Squares” (Cooperrider 2014, 2018), this study revealed that KD’s optionally bare nouns, as well as those accompanied by determiners, differ in their discourse-pragmatic features. Optionally bare nouns tended to have specific, discourse-/hearer-old referents, and -lala exhibited the features of a locative reinforcer. However, -la and -sala’s profiles overlapped considerably; both exhibited deictic force, with the only clear distinguishers being -la’s ability to mark inferable referents and -sala’s use in temporal expressions.
Presenter 2: Ioana Grosu (Doctoral student, New York University)
Title: A corpus analysis of children’s counterfactual production
Abstract: Counterfactual conditionals (e.g., “if cats had wings, they would fly”) describe a world state different from the actual world. Adults prefer to maintain similarity to the actual world when generating possible counterfactual worlds (Lewis 1979). Experimental studies show that children do not demonstrate this parsimonious similarity strategy and often give non-adult like responses when reasoning about counterfactuals. In this project, I consider the following question: What are children’s spontaneous counterfactuals like? I conduct an in-depth analysis of child production of counterfactuals (31 English-language corpora from CHILDES, ages 48 to 96 months). I code for twelve features of counterfactual scenarios, noted in prior experimental literature as impacting the difficulty of counterfactual interpretation (e.g., Byrne 2016, German & Nichols 2003). For example, children respond in a more adult-like manner to counterfactual worlds in which the antecedent is an immediate cause of the consequent, rather than counterfactual worlds which have multiple causal steps between the antecedent and consequent events. Additionally, additive counterfactuals (in which the antecedent counterfactually adds a non-actual event, rather than prevents an actual event from happening) are easier to reason over than their subtractive counterparts. Through this corpus work I assess whether factors which improved children's comprehension in experimental tasks are also corroborated by spontaneous production data. Preliminary results (n=95 counterfactual utterances; planned: ~500), indicate that these factors do appear more frequently in production: e.g., 95% of utterances had short causal chains, 95% were additive. I provide not only an in-depth, multi-factor analysis of counterfactual utterances on a large scale, but also a novel empirical survey on natural production of counterfactuals. I supplement theoretical and empirical claims about the ease of counterfactual comprehension, through an analysis of production data. This additionally elucidates whether linguistic and conceptual factors manipulated in experiments are well-supported in children's input (planned future study) and production.
Presenter 3: Maria Elizabeth Garza (Master’s student, CUNY Graduate Center)
Title: Spanish diacritic restoration
Abstract: A supervised machine-learning system was designed to restore diacritics to Spanish tokens like está/is that can be easily trained on any properly diacriticized text. This system could be used as a practical application to normalize Spanish text typed on devices that don’t easily have the means to restore diacritics. The most challenging part of Spanish diacritic restoration entails the disambiguation of mellizas, a term of art used in this presentation to describe tokens like esta/this-está/is that would look orthographically identical if they were spelled without diacritics. More specifically, mellizas are disambiguated by using Naïve Bayes classifiers trained on a small amount of labeled data, and by referring to a set of grammatical rules. In order to restore diacritics to invariantly diacriticized tokens like pañales/diapers, the system must perform a basic dictionary look-up in a dictionary built during the training phase, which contains entries of these token types that were extracted from the training data. Overall, the macro-averaged token accuracy for this system is 96.08%, with a relative error reduction of 14.24% over a baseline model which picks the most common class for each melliza. Error analysis reveals that while these classifiers are quite adept at disambiguating mellizas (accuracy=93.91%), surprisingly, the system struggles when it comes to restoring diacritics to invariantly diacriticized tokens like pañales (accuracy=71.90%). In a system like this, restoring diacritics to these tokens will always be a challenge because it is impossible to build a dictionary during the training phase that is robust enough to account for all such tokens that may appear in the test set. To lessen the system’s dependence on this dictionary, we outline a proposal for the design of a new classifier that will focus mainly on the vowel features of invariantly diacriticized tokens.
Parallel Session 3B: Sociolinguistics
Moderator: Xinye Zhang • Co-moderator: Daniel Cerbino
Presenter 1: Noelia Ayelén Stetie (Doctoral student, CONICET, University of Buenos Aires)
Title: Generic representations: the reliability of non-binary language in Spanish
Abstract: There is empirical evidence in different languages about how gender morphology affects the construction of mental representations on gender stereotypes (Bradley 2020, Prentice 1994, Prewitt-Freilino et al. 2012). However, there are not many analysis of the underlying psycholinguistic process in Spanish, especially studies that consider the two non-binary morphological variants usually used: [-x] and [-e], commonly known as inclusive language. Our aim is to analyze if: 1. the three morphological variants (-o, -e, -x) generate a generic representation; 2. there are any processing costs associated with non-binary forms; 3. the stereotipicality of the role nouns has an effect on processing costs. We programmed and performed a sentence comprehension task in PCIbex (Zehr et al. 2018). Sentences with simple noun phrases in the three morphological variants were presented, after which participants had to answer a multiple-choice question about the possible reference of that noun phrase (a group of women, men, or a mixed group). The task had a 3x2 factorial design: morphology (-o, -x, -e) and stereotipicality of role names (low, high). We evaluated 517 participants (age: M=34.42). For the sentence reading times, we find a stereotipicality effect –low stereotipicality role names were read faster than high stereotipicality ones–, but no difference depending on the morphological variants. Regarding responses, for non-binary forms, all participants consistently chose a mixed reference, regardless of the level of stereotipicality. For masculine morphology, we found that it could function as a generic reference only for low stereotipicality role names, but not for high stereotipicality. Furthermore, we found a morphology effect in the opposite direction to what we have hypothesized: phrases with masculine morphology required longer response times than those with either of the two non-binary morphological variants.
Presenter 2: Allison Taylor-Adams (Doctoral student, University of Oregon)
Title: Language documentation for talk about language: Strengthening research and practice
Abstract: The past two decades have seen a flourishing of language revitalization efforts across the world (Peréz Báez et al., 2019). Many linguists have shifted their research focus towards the support of these efforts; in particular, the subfield of language documentation has taken up this challenge, calling for primary language data that is conscientiously recorded and preserved, accessible to diverse audiences, and open to different analytical approaches and applied uses (Himmelman, 2006). Despite these important strides, it is increasingly clear that language revitalization also requires the support of secondary language data – for information about language practice (Penfield & Tucker, 2011). That is, while input for language acquisition is critical, revitalization efforts also need knowledge about pedagogical methods, about strategies for leveraging resources, about the experiences of individuals as well as the social contexts that create and sustain such efforts. One way to approach these diverse topics is through qualitative interviewing with language revitalization practitioners. In this talk, I demonstrate how a language documentation approach can improve interview data for qualitative research. Qualitative social scientists routinely call for attention to interview transcription as a disciplined and theorized practice (e.g., Mishler, 1991; Silverman, 2017). Using data from interviews I conducted with revitalization practitioners about their language learning motivations, I show how using documentation tools to produce time-aligned transcriptions answers these calls. I also show how a language documentation approach extends the potential of these interviews beyond the original research question to other future uses, for example as historical records of a community’s language efforts, or as inspiration and guidance for others in the global community of practice. By applying language documentation methods to new kinds of speech data, linguists can make important contributions to the rigor of social science practice while also becoming better equipped to support the multiple dimensions of language revitalization.
Presenter 3: Paul Kidhardt (Doctoral student, University of California, Davis)
Title: Lassoing Spanish Heritage Speaker Verb Mood Complexity
Abstract: Understanding the complexity underlying variation in the Spanish subjunctive among native and heritage speakers continues to challenge researchers. This study explores identity, family language, internal dialogue, religion, age and order of acquisition, and language dominance variables to determine what effects, if any, they have on the oral production of the “obligatory” subjunctive among Spanish HSs. Using videos with human actors and dramatic voiceovers to contextualize 14 predicates that normatively trigger the subjunctive, production data was collected on 95 HSs to predict the felicitous production of the subjunctive and reveal bilingual processes that may give rise to the subjunctive in this population. Along my analytical path I introduce, justify, and demonstrate the empirical power of inferential LASSO, a newly derived machine learning regression technique aimed at revealing causal connections. Results reveal that maternal grandmother language preferences, the language used in dreaming, self- directed talk and in church, and feeling like oneself when speaking Spanish are positive and meaningful predictors of felicitous subjunctive production among HSs. Finally, I demonstrate how age of acquisition and order of onset interact with language dominance to affect the production of the subjunctive. As Polinsky and Scontras (2020) assert, “we will never have a predictive model of HL competence without an understanding of the factors leading to, arising from, and constraining variation” (p. 50). Leveraging the power of the latest machine learning methods, I hope to reveal factors that lead to, arise from, and constrain variation with respect to verb mood choices among Spanish HSs in the United States.
Parallel Session 3C: Phonetics & Phonology
Moderator: Eugênia Fernandes • Co-moderator: Ana Ruiz
Presenter 1: Ashley Keaton (Doctoral student, University of California, Davis)
Title: What makes a voice memorable? Investigating voice recognizability through typicality?
Presenter 2: Dot-Eum (Rachel) Kim (Doctoral student, University of Georgia)
Title: Korean American English in GA: [+back] on the B/OW/T
Abstract: This study explores language as a resource for expressing ethnic identity by analyzing phonetic variation of Korean Americans in Gwinnett County, Georgia. The large and fast growing Korean American population in the Atlanta area, particularly in Gwinnett county (Census 2017), provides an excellent site, not only to examine the relationship of language and ethnic identity but also to document the development and diffusion of Asian American English in the Southeastern U.S. This paper presents an analysis on the production of the mid-back GOAT vowel by 12 college-age second generation Korean American speakers. The analytic focus is on the frontness and dynamic properties of this vowel. The linguistic behavior of Korean American speakers is compared with that of college peers of 10 European American speakers. Although the fronting of GOAT vowel is a well-known supra-regional phonetic feature of sound change attested in many varieties of English and among speakers of different ethnic backgrounds (Bauman 2016; Hall- Lew 2009; Labov et al. 2006; Thomas 2001, among many others), to my knowledge this vowel in the southern part of U.S. among non-Anglo/African American speakers, especially, Korean American speakers, is understudied. The results show that a backed GOAT vowel is used by Korean American speakers as a way to index ethnic identity in Gwinnett County, Georgia. In contrast, the dynamic properties of the GOAT vowel, as diphthongal or monophthongal, do not appear to vary according to speaker ethnicity or within ethnic subsamples. All in all, this study broadens the field of dialectal research in the Southern U.S. by shedding light on the linguistic behavior of this fastest growing, but linguistically understudied, ethnic population. Additionally, this study contributes to the overall understanding of the interactions between language and ethnic identity, and also enhances the awareness of how speakers creatively use linguistic resources in projecting their identities.
Presenter 3: Riley Stray (Undergraduate student, University of California, Davis)
Title: How Lexical and Phonological Variation Interact in Speech Perception Across American and British English
Abstract: Dialects differ both in how words are pronounced and how they are used in which semantic contexts; for example, American English speakers often use the word “rubber” to refer to a material while British speakers use it to refer to an eraser. Past studies have documented how listeners comprehend speech across dialects, but less research has considered how a word’s meaning can impact how it is perceived in distinct accents. The current study tests how phonological and semantic properties of words contribute to speech intelligibility across American and British dialects. Stimuli consisted of 24 dialectal “false friends,'' words with the same form but different meaning in British/American usages. For each false friend, we created two semantically-predictable sentences: one to cue a British and one to cue an American meaning of the word (e.g., British: “Even though the buses are cheaper I take the Tube” and American: “I asked if I could borrow a plastic poster tube”). We then generated spoken versions of our sentences by American and British text-to-speech (TTS) voices and mixed the sentences with speech-shaped noise (-6dB SNR). Participants (n=70) were instructed to listen to the sentences and write the final word they heard (the false-friend target word). Responses were coded for accuracy (correct = 1, incorrect = 0) and modeled with a mixed-effects logistic regression. Results showed lower accuracy for the British meanings of words but no effect of voice on accuracy. Furthermore, an individual’s experience with British English shaped accuracy based on sentence context (but did not interact with voice). Taken together, these findings suggest that semantic content of words can affect how they are perceived, but spoken dialect in TTS voices does not affect intelligibility.
3-Minute Thesis Competition Participants
Watch the participants' video submissions here on the Cluster for Language Research's YouTube Channel.
A glance at Portuñol from Generative Grammar, María Teresa Borneo and Nazira Günther (National University of Córdoba, Argentina)
There is a debate about the notion of Portuñol and its scope. Some define it as a pidgin, but it has also been defined as a variety of Spanish, as a variety of Portuguese, and also as a simple code-switching between the two languages. This study intends to begin to investigate how the alternations between two linguistic systems in speakers of Portuñol in the province of Misiones, Argentina are not random: they imply certain restrictions that allow us to observe the hidden design of language, common to all languages, behind the surface of the performance. For this, we will consider the assumptions of Generative Grammar, a theory that states that all members of the human species possess a mental faculty of language that allows them to develop and use one or more languages. Based on the idea, also put forward by Pinker (2001), that here is a single grammatical machinery in charge of manipulating symbols that, without exception, underlies all the languages of the world, we will seek to demonstrate that what appears to be a simple change of codes in the use of Portuñol, in reality obeys syntactic-functional rules that speakers seem to know perfectly well, beyond the simplification they make to help themselves be understood. Thus, based on the analysis of examples of concrete use taken from empirical studies by other authors, we hypothesize that behind the combinations and their restrictions there are structures formed by morphosyntactic principles that seem to be part of the knowledge and competence of the speakers.
Modifier Effects on Relative Clause Attachment Preferences in Turkish, Elifnur Ulusoy (undergraduate student, Boğaziçi University)
Turkish is no exception to the languages with relative clause attachment ambiguity. Previous research discovered that the attachment preferences in Turkish depend largely on the semantic relations, such as whole-part relations in genitive-possessive constructions (Kırkıcı, 2004). In this study, we looked at whether attachment preferences are affected by locality effects (Drenhaus, 2011). This effect arises when the distance between the noun and verb is increased, making the processing of RC verb difficult. For our purposes, we compared constructions with and without modifiers preceding NP2 and carried out a SPR study, followed by a 2AFC task to determine to what extent the preference gravitates toward NP1 attachment. Our prediction was that while the an NP2 with a modifier will not change preference in the NP1 and NP2 attachment conditions, but it will increase the NP1 preference in the ambiguous condition. Participants’ responses in the forced choice task revealed that N1 attachment in the complex N2 condition was 8 percentage points higher than it is in the simple N2 condition. The differences between the NP1 attachment preferences in NP1 and NP2 attachment conditions, however, were not statistically significant. In our reading times analysis, we observed a slowdown in the complex NP2 condition at the adjective and the remainder.
Bridging the Gap Between Belief and Practice: Social Justice in Community College Classrooms, Emily E. Packer (Doctoral student, University of San Diego)
70% of community college students do not earn certificates or degrees, and do not transfer, within six years of beginning their college journey, and a reason for this is not meeting critical milestones such as passing English courses (Moore & Shulock, 2010). This research looks at the gap between instructor beliefs and practices regarding social justice as its existence negatively impacts marginalized students. By analyzing this problem through a framework inclusive of postcolonial theory, critical race theory, and critical applied linguistics tenets, the question of how instructors can disrupt and replace their own inequitable practices in order to enact change within their classroom can begin to be answered. It was concluded that through a framework based on Freire’s (1968) pedagogy of love, critical reflection, critical dialogue, problematizing givens, and moving from othering to inclusion can minimize the gap between belief and practice. It is important to note critiques of the term “social justice” within education so future research can strengthen and unify this terminology. The findings show a need for further research into this issue specific to the community college context.
I’m Tawkin’ Here: Why don’t New Yorkers sound like Noo Yawkas anymore?, Jennifer Kaplan (Doctoral student, University of California, Berkeley)
Why aren’t most young New Yorkers tawkin’ like dis anymore? We know how their speech is changing: Some young New Yorkers are fronting their back vowels (namely, GOOSE and GOAT) as part of a sound change that originated in California and is now spreading across North America. The question investigated here, however, is why some young New Yorkers are changing their patterns of Back Vowel Fronting (BVF), while others are not. While BVF is widespread in North American English, the degree to which individuals in different regions back their vowels differs and may be locally distinctive. New York City English (NYCE) has historically only possessed mild to moderate BVF. However, new data collected as part of the New York City English Corpus (CoNYCE) project indicates that young people are now more likely to front more dramatically and increase their fronting in previously blocked environments, pointing to an ongoing regional linguistic shift. This study analyzes speakers’ attitudes, rating on a scale from 1-5 their self-identification with NYCE and their evaluations of New York and New Yorkers, to argue that it is young New Yorkers’ stances toward New York City and NYCE—especially their local versus non-local orientations toward the City—that determines whether or not they utilize styles with or without BVF.
**NOTE: This 3MT video (I'm Tawkin' Here) is not eligible for advancement to the finals. Please enjoy this wonderful talk and fascinating research, but votes for this video will not count.**
Examining Heteronormativity and Gender Representation in Korean Textbooks, Greece Peña (Master's student, California State University, Los Angeles)
Textbooks hold a powerful position in classrooms. Language students interact with these textbooks in order to create an identity suitable for the target language culture. If these textbooks hold ideologies that restrict specific identities and hold certain identities with more value than others, students will internalize these ideologies as they learn. The inclusion of heteronormativity and representation of gender are two ideologies that find themselves in the forefront of my study. Heteronormativity deals with the belief pf heteronormative relationships as the norm, while often excluding or ignoring other non-heteronormative relationships. Similarly, the manner in which gender is represented in these textbooks may follow certain stereotypes that limit actions, descriptions, occupations, etc. according to gender. This study will examine both heteronormativity and gender representation in a Korean language textbook series. This series, titled Integrated Korean, is the most used Korean language book in the United States for universities offering Korean language classes. This study will implement a mixed-methods research approach in order to provide a broader sense of how either ideology is found in these textbooks. The quantitative data in this study will provide strong support for an opening towards a discussion that looks through the qualitative data of this series.