Theses and Dissertations from UMD
Permanent URI for this communityhttp://hdl.handle.net/1903/2
New submissions to the thesis/dissertation collections are added automatically as they are received from the Graduate School. Currently, the Graduate School deposits all theses and dissertations from a given semester after the official graduation date. This means that there may be up to a 4 month delay in the appearance of a give thesis/dissertation in DRUM
More information is available at Theses and Dissertations at University of Maryland Libraries.
Browse
58 results
Search Results
Item DO MEASURES OF INDIVIDUAL WORDS AND FORMULAIC SEQUENCES TAP INTO THE SAME TRAIT: THE PERSPECTIVE OF ASSESSMENT AND THE CONTRIBUTIONS OF PHONOLOGICAL SHORT-TERM MEMORY AND EXPOSURE(2024) Deng, Zhiyuan; Hui, Bronson; Second Language Acquisition and Application; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Nativelike language use is characterized by a high level of formulaicity (Pawley & Syder, 1983; Sinclair, 1991), and formulaic sequences are often believed to be building blocks of language acquisition (Christiansen & Arnon, 2017) and crucial to language fluency (Saito, 2020). Although they consist of multiple words and are analyzable, some researchers argued that the knowledge of formulaic sequences is largely lexical in nature, i.e., stored and processed holistically without recourse to analysis (Wray, 2002). Wray (2008) further proposed a heteromorphic view of mental lexicon, pushing the boundary of vocabulary to encompass not only individual words but also larger-than-word units such as formulaic sequences. The main purpose of the present study was to empirically test this proposal from the perspective of assessment, i.e., see if measures of formulaic sequences tap into the same latent construct underlying measures of individual words. In addition, the present study also investigated the contributions of phonological short-term memory (PSTM) and exposure to the knowledge of formulaic sequences and individual words. The study was carried out in an English as a Foreign Language (EFL) context, and 136 Chinese participants of intermediate to advanced proficiency completed a battery of nine linguistic measures assessing their receptive and controlled productive knowledge of collocations, phrasal verbs, and individual words. In addition, their capacity of PSTM was measured by a non-word span test, and their engagement in various types of English-medium activities was measured by an exposure questionnaire. Confirmatory factor analysis and model comparisons were conducted to examine the factor structure of nine linguistic measures, and a bi-factor solution with a single latent trait factor underlying all nine linguistic measures and a method-specific grouping factor for all six receptive measures was selected as the best-fitting model in terms of fit and parsimony. In addition, structural equation modeling revealed that PSTM, exposure, and length of learning English were all significant predictors for the knowledge of formulaic sequences and the knowledge of individual words. The three predictors combined explained about 33.4% of variance in the knowledge of formulaic sequences and 30.9% of the variance in the knowledge of individual words. However, the contributions of PSTM and exposure to the knowledge of formulaic sequences and to the knowledge of individual words were not significantly different in magnitude. The results provided psychometric evidence supporting the legitimacy of conceptualizing a heteromorphic mental lexicon showing that measures of formulaic sequences and individual words tapped into the same latent trait.Item THE CROSS-LANGUAGE ACTIVATION OF FIRST LANGUAGE (L1) HOMONYMS TRANSLATIONS IN SECOND LANGUAGE (L2) PROCESSING: AN INVESTIGATION OF WHETHER L1 TRANSLATION ARE ACTIVATED IN L2 SENTENCE CONTEXT(2024) Alsalmi, Mona Othman; Jiang, Nan; Second Language Acquisition and Application; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)A present study aimed to investigate the role of a first language (L1) translation on a second language (L2) word processing in a sentential context by relatively advanced Arabic learners of English. The focus is on cases where a homonymous word in the L1 is realized by independent words in the L2, (e.g. Arabic قرش realized by English shark and coin). Using the visual-world paradigm, Arabic-English bilinguals and English native participants were auditorily presented with English sentences that are predictive of a target word (e.g., “shark” in Scuba divers saw the sharp teeth of a giant shark yesterday) while looking at a visual screen. The screen contained one of the three critical objects: a target object whose English name corresponded to the target word (shark; Arabic: قرش) in the target condition, an Arabic competitor object whose Arabic name shared the same Arabic translation with the target word (coin; Arabic: قرش) in the Arabic condition, or an object that was unrelated to the target word (drums; Arabic طبل) in the control condition.Compared to native speakers of English, relatively advanced Saudi learners of English made more fixations on the critical objects in the Arabic condition compared to the control condition. This study supports the potential automatic activation of L1 translations when processing sentences in L2, even in relatively proficient learners and suggests evidence for the verification model in L2 word recognition.Item Popular Backlash to Language Assimilation Regimes(2024) Derks, John William; McCauley, John; Government and Politics; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Do assimilationist restrictions on a minority language lead to greater national unity or a more rebellious minority population? Under what conditions might short-term backlash to language assimilation evolve into greater national unity in the long term? While much of the literature on ethnic politics implicitly treats language simply as an identifying feature of ethnic groups, this dissertation contends that salient language identities and grievances can serve as a source of meaningful division. I examine when and why the costs of pursuing linguistic homogeneity exceed its practical benefits. Just as minority individuals must conduct a cost-benefit analysis to determine whether they should acquire or have their children acquire the dominant language of a host state, so too must governments consider the likelihood that an assimilation program will succeed or fail. I highlight an interaction between three key mechanisms that lead to a distortion of political leaders’ cost-benefit analysis when deciding on the nature of their desired language assimilation program. This distortion leads host states governed by the dominant language group to systematically overestimate the willingness of minority individuals to assimilate voluntarily and underestimate the likely level of subsequent backlash to severe language restrictions. From this theoretical framework, I argue that more severe language restrictions increase minority backlash and that the intensity of this backlash is influenced by the presence of exclusionary political and economic policies targeting the minority group. To this end, I conduct five comparative historical case studies on the language assimilation programs imposed on the South Tyroleans in Italy, Amazigh in Algeria, Azerbaijanis in Iran, Mayans in Mexico, and the Anglophones in Cameroon. The overall findings show that the use of more severe language restrictions and exclusionary political and economic policies are very likely to result in intense backlash responses. More often than not, this elevated backlash response will inflict considerable long-term damage to a state’s national unity.Item UNDERSTANDING HOW AFRICAN AMERICAN ENGLISH-SPEAKING CHILDREN USE INFLECTIONAL VERB MORPHOLOGY IN SENTENCE PROCESSING AND WORD LEARNING(2024) Byrd, Arynn S; Edwards, Jan; Huang, Yi Ting; Hearing and Speech Sciences; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)This research examined how linguistic differences between African American English (AAE) and Mainstream American English (MAE) impact how children process sentences and learn new information. The central hypothesis of this dissertation is that these linguistic differences adversely impact how AAE-speaking children use contrastive inflectional verb morphology (e.g., was/were, third person singular -s) to process and comprehend MAE sentences, as well as to infer word meanings when they depend on dialect-specific parsing of sentence cues. To test this hypothesis, this dissertation conducted three experiments on how linguistic mismatch impacts spoken language comprehension and word learning in school-age MAE- and AAE-speaking children. The first study examined how children used the auxiliary verbs was or were to comprehend MAE sentences in an offline spoken language comprehension task. In contrast, the second study asked the same question in an online sentence processing task. The final study examined how children used inflectional verb morphology (i.e., third-person singular -s, was/were) to infer information about novel verbs. Each study examined how participants’ dialect, either MAE or AAE, predicted performance on listening tasks produced in MAE. Furthermore, each study examined how individual differences such as age, dialect density, and vocabulary size influenced children’s performance.Across all studies, results demonstrated that when there were redundant linguistic cues that were not impacted by dialect differences, AAE- and MAE-speaking children used available linguistic cues to process and comprehend spoken language and infer verb meanings in a similar manner. However, when linguistic redundancy was decreased due to perceptual ambiguity, there were group differences in how AAE- and MAE-speaking children used inflectional verb morphology on spoken language tasks. The second study showed that AAE-speaking children were sensitive to contrastive verb morphology in real-time processing, but they were less likely than their MAE-speaking peers to use it as an informative cue to revise initial parses when processing spoken language. The results of the final study indicated that individual characteristics such as age and dialect density influence how dialect impacts a learning process. These results demonstrate that linguistic mismatch can affect spoken language processes. Furthermore, the findings from this research highlight a complex relationship between the effects of linguistic mismatch and individual differences such as age and dialect density.Item The Learning and Usage of Second Language Speech Sounds: A Computational and Neural Approach(2023) Thorburn, Craig Adam; Feldman, Naomi H; Linguistics; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Language learners need to map a continuous, multidimensional acoustic signal to discrete abstract speech categories. The complexity of this mapping poses a difficult learning problem, particularly for second language learners who struggle to acquire the speech sounds of a non-native language, and almost never reach native-like ability. A common example used to illustrate this phenomenon is the distinction between /r/ and /l/ (Goto, 1971). While these sounds are distinct in English and native English speakers easily distinguish the two sounds, native Japanese speakers find this difficult, as the sounds are not contrastive in their language. Even with much explicit training, Japanese speakers do not seem to be able to reach native-like ability (Logan, Lively, & Pisoni, 1991; Lively, Logan & Pisoni, 1993) In this dissertation, I closely explore the mechanisms and computations that underlie effective second-language speech sound learning. I study a case of particularly effective learning--- a video game paradigm where non-native speech sounds have functional significance (Lim & Holt, 2011). I discuss the relationship with a Dual Systems Model of auditory category learning and extend this model, bringing it together with the idea of perceptual space learning from infant phonetic learning. In doing this, I describe why different category types are better learned in different experimental paradigms and when different neural circuits are engaged. I propose a novel split where different learning systems are able to update different stages of the acoustic-phonetic mapping from speech to abstract categories. To do this I formalize the video game paradigm computationally and implement a deep reinforcement learning network to map between environmental input and actions. In addition, I study how these categories could be used during online processing through an MEG study where second-language learners of English listen to continuous naturalistic speech. I show that despite the challenges of speech sound learning, second language listeners are able to predict upcoming material integrating different levels of contextual information and show similar responses to native English speakers. I discuss the implications of these findings and how the could be integrated with literature on the nature of speech representation in a second language.Item A MULTILEVEL ANALYSIS OF LANGUAGE PROFICIENCY GROWTH(2023) Mackey, Beth; Gor, Kira; Bolger, Donald; Second Language Acquisition and Application; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)The U.S. Military Services employ thousands of servicemen and women in language-related positions that are critical to the nation’s national security. These positions require personnel with high-level capability in various languages and dialects (Asch & Winkler, 2013). A complex accession and training system that begins at local recruiting stations across the nation leads to worldwide placement of language professionals who serve multiyear tours in the U.S. Air Force, Army, Navy and Marine Corps. High levels of cognitive ability, as measured by two cognitive aptitude batteries, one general (ASVAB) and one language (DLAB), are required for selection into these positions. Following significant investments in basic levels of training, the jobs themselves demand high level skills, and the service members find themselves constantly challenged to grow their skills. Traditional research on the effectiveness of the accession and training processes focuses on learning outcomes, rather than growth. This research used a longitudinal design to investigate how general aptitude, language aptitude, non-cognitive and language distance measures impact language proficiency growth. Hierarchical linear models and hierarchical generalized linear models were used and the significant findings were similar. The study found that overall, while language test scores followed a drop-and-recover pattern, there was very little growth overall. Three aptitude subtests, one from ASVAB (Mechanical Comprehension) and two DLAB subtests (Part 3 and Part 4) were found to constrain initial growth in the listening modality. Language distance was found to constrain initial and subsequent growth in listening and reading.Item INVESTIGATING INDIVIDUAL DIFFERENCES’ PREDICTION OF LANGUAGE PROFICIENCY OUTCOMES: A LATENT GROWTH CURVE MODELING APPROACH(2023) Rhoades, Elizabeth Rogler; Gor, Kira; Clark, Martyn; Second Language Acquisition and Application; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Although decades of research within the field of second language acquisition have been dedicated to investigating the impact of individual differences on second language learners’ success, longitudinal research focused on individual differences and their impact on adult second language acquisition is extremely limited. Additional longitudinal research on individual differences is necessary to further our understanding of the nature of the process of adult second language acquisition. This area of research is also critical to the U.S. Government and the Department of Defense as thousands of military service members work in language-related positions, and these service members’ maintenance of high levels of language proficiency is critical for our nation’s national security. The current study used a longitudinal design to investigate the impact of individual differences such as general cognitive ability, language aptitude, and attitude toward learning assigned second language (L2) on military service members’ language proficiency outcomes. Latent growth curve modeling (LGM) was used to model participants’ initial proficiency levels and growth trajectories, and measures of cognitive ability, language aptitude, and attitude toward learning assigned L2 were used to measure the impact of these individual differences on language proficiency outcomes. Additional variables including GPA, age, education level, number of language training hours, billet type, and sex were also included in the analyses. The results from the four phases of analyses support the conclusion that the predictive value of individual difference factors on language proficiency outcomes differ not only by DLI Language Difficulty Category, as suggested by previous research, but also by language and even language modality.Item ISOLATING EFFECTS OF PERCEPTUAL ANALYSIS AND SOCIOCULTURAL CONTEXT ON CHILDREN’S COMPREHENSION OF TWO DIALECTS OF ENGLISH, AFRICAN AMERICAN ENGLISH AND GENERAL AMERICAN ENGLISH(2023) Erskine, Michelle E; Edwards, Jan; Huang, Yi Ting; Hearing and Speech Sciences; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)There is a long-standing gap in literacy achievement between African American and European American students (e.g., NAEP, 2019, 2022). A large body of research has examined different factors that continue to reinforce performance differences across students. One variable that has been a long-term interest to sociolinguists and applied scientists is children’s use of different dialects in the classroom. Many African American students speak African American English (AAE), a rule-governed, but socially stigmatized, dialect of English that differs in phonology, morphosyntax, and pragmatics from General American English (GAE), the dialect of classroom instruction. Empirical research on dialect variation and literacy achievement has demonstrated that linguistic differences between dialects make it more difficult to learn to read (Buhler et al., 2018; Charity et al., 2004; Gatlin & Wanzek, 2015; Washington et al., 2018, inter alia) and recently, more difficult to comprehend spoken language (Byrd et al., 2022, Edwards et al., 2014; Erskine, 2022a; Johnson, 2005; de Villiers & Johnson, 2007; JM Terry, Hendrick, Evangelou, et al., 2010; JM Terry, Thomas, Jackson, et al., 2022). The prevailing explanation for these results has been the perceptual analysis hypothesis, a framework that asserts that linguistic differences across dialects creates challenges in mapping variable speech signals to listeners’ stored mental representations (Adank et al., 2009; Clopper, 2012; Clopper & Bradlow, 2008; Cristia et al., 2012). However, spoken language comprehension is more than perceptual analysis, requiring the integration of perceptual information with communicative intent and sociocultural information (speaker identity). To this end, it is proposed that the perceptual analysis hypothesis views dialect variation as another form of signal degradation. Simplifying dialect variation to a signal-mapping problem potentially limits our understanding of the contribution of dialect variation to spoken language comprehension. This dissertation proposes that research on spoken language comprehension should integrate frameworks that are more sensitive to the contributions of the sociocultural aspects of dialect variation, such as the role of linguistic and nonlinguistic cues that are associated with speakers of different dialects. This dissertation includes four experiments that use the visual world paradigm to explore the effects of dialect variation on spoken language comprehension among children between the ages of 3;0 to 11;11 years old (years;months) from two linguistic communities, European American speakers of GAE and African American speakers with varying degrees of exposure to AAE and GAE. Chapter 2 (Erskine [2022a]) investigates the effects of dialect variation in auditory-only contexts in two spoken word recognition tasks that vary in linguistic complexity: a) word recognition in simple phrases and b) word recognition in sentences that vary in semantic predictability. Chapter 3 [Erskine (2022b)] examine the effects of visual and auditory speaker identity cues on dialect variation on literal semantic comprehension (i.e., word recognition in semantically facilitating sentences). Lastly, Chapter 4 [Erskine (2022c)] examines the effects of visual and auditory speaker identity cues on children’s comprehension of different dialects in a task that evaluates pragmatic inferencing (i.e., scalar implicature). Each of the studies investigate the validity of the perceptual analysis against sociolinguistcally informed hypotheses that account for the integration of linguistic and nonlinguistic speaker identity cues as adequate explanations for relationships that are observed between dialect variation and spoken language comprehension. Collectively, these studies address the question of how dialect variation impacts spoken language comprehension. This dissertation provides evidence that traditional explanations that focus on perceptual costs are limited in their ability to account for correlations typically reported between spoken language comprehension and dialect use. Additionally, it shows that school-age children rapidly integrate linguistic and nonlinguistic socioindexical cues in ways that meaningfully guide their comprehension of different speakers. The implication of these findings and future research directions are also addressed within.Item Examining the Disproportionate Representation of Bilingual Children in Special Education(2022) Ortiz, Jose A; Cummings, Kelli D; Special Education; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Chapter 2: Nonword repetition has been endorsed as a less biased method of assessment for children from culturally and linguistically diverse backgrounds, but there are currently no systematic reviews or meta-analyses on its use with bilingual children. The purpose of this study is to evaluate diagnostic accuracy of nonword repetition in the identification of language impairment (LI) in bilingual children. Using a keyword search of peer-reviewed literature from several large electronic databases, as well as ancestral and forward searches, 13 studies were identified that met the eligibility criteria. Studies were evaluated on the basis of quality of evidence, design characteristics, and reported diagnostic accuracy. A meta-regression analysis, based on study results, was conducted to identify task characteristics that may be associated with better classification accuracy. Diagnostic accuracy across studies ranged from poor to good. Bilingual children with LI performed with more difficulty on nonword repetition tasks than those with typical language. Quasiuniversal tasks, which account for the phonotactic constraints of multiple languages, exhibited better diagnostic accuracy and resulted in less misidentification of children with typical language than language-specific tasks. Evidence suggests that nonword repetition may be a useful tool in the assessment and screening of LI in bilingual children, though it should be used in conjunction with other measures. Quasiuniversal tasks demonstrate the potential to further reduce assessment bias, but extant research is limited. Chapter 3: The disproportionate identification of language-related disorders in schools, including communication disorders and specific learning disability, is an ongoing problem for bilingual children, with evidence of both over- and underrepresentation. Previous research has uncovered distinct identification patterns for emergent and English-proficient bilinguals, as well as differences in identification rates across grades. However, there is limited information about disability identification for different groups of bilinguals across grades. Thus, the purpose of this study is to examine the prevalence and incidence of language-related disorders in emergent and English-proficient bilinguals in elementary school. Using a nationally representative, individual-level, longitudinal data set, this study examined representation in language-related disorder categories, as well as identification rates by year. This study also examined individual- and school-level predictors of disability identification for bilingual children. Results indicate that emergent and English-proficient bilinguals exhibit distinct patterns of language-related disorder identification. Emergent bilinguals experienced a disproportionate increase in disability identification rates in third grade, resulting in significant overrepresentation in subsequent grades. By fifth grade, emergent bilinguals experienced approximately twice the odds of being identified with a language-related disorder, compared to monolinguals. English-proficient bilinguals, on the other hand, were underrepresented in language-related disorder categories in early elementary school grades, but experienced identification rates similar to monolinguals by fifth grade. Outcomes from this study provide insight into patterns of language-related disorder identification for bilinguals that have not been addressed in previous research. The implications for education practice and policy are discussed. Chapter 4: The disproportionate representation of bilingual children in special education is an ongoing issue in US schools, with evidence of both over- and underrepresentation. Identification rates of language-related disorders, including communication disorders and specific learning disability, are particularly relevant for bilingual children given the challenges associated with differentiating language difference from disorder and the possibility of misidentification. School-based speech-language pathologists are well positioned to address the issue, but many do not engage in practices that may reduce disproportionate disability identification. The purpose of this practitioner paper is to provide school-based clinicians with an evidence-based model for addressing disproportionality in bilingual children, with a focus on prevention. This paper provides a review of the literature on the topic and integrates information from relevant studies to provide a clear depiction of the nature of the problem. In addition, this paper describes a model of disproportionality prevention, and provides a set of evidence-based methods that clinicians can employ. Topics include, pre-referral intervention, early identification, parent engagement, and collaboration. By adopting the methods described in this paper, school-based speech-language pathologists can strengthen their ability to meaningfully address many of the issues that contribute to over- and underrepresentation of bilingual children in special education.Item Transfer Learning in Natural Language Processing through Interactive Feedback(2022) Yuan, Michelle; Boyd-Graber, Jordan; Computer Science; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Machine learning models cannot easily adapt to new domains and applications. This drawback becomes detrimental for natural language processing (NLP) because language is perpetually changing. Across disciplines and languages, there are noticeable differences in content, grammar, and vocabulary. To overcome these shifts, recent NLP breakthroughs focus on transfer learning. Through clever optimization and engineering, a model can successfully adapt to a new domain or task. However, these modifications are still computationally inefficient or resource-intensive. Compared to machines, humans are more capable at generalizing knowledge across different situations, especially in low-resource ones. Therefore, the research on transfer learning should carefully consider how the user interacts with the model. The goal of this dissertation is to investigate “human-in-the-loop” approaches for transfer learning in NLP. First, we design annotation frameworks for inductive transfer learning, which is the transfer of models across tasks. We create an interactive topic modeling system for users to find topics useful for classifying documents in multiple languages. The user-constructed topic model bridges improves classification accuracy and bridges cross-lingual gaps in knowledge. Next, we look at popular language models, like BERT, that can be applied to various tasks. While these models are useful, they still require a large amount of labeled data to learn a new task. To reduce labeling, we develop an active learning strategy which samples documents that surprise the language model. Users only need to annotate a small subset of these unexpected documents to adapt the language model for text classification. Then, we transition to user interaction in transductive transfer learning, which is the transfer of models across domains. We focus our efforts on low-resource languages to develop an interactive system for word embeddings. In this approach, the feedback from bilingual speakers refines the cross-lingual embedding space for classification tasks. Subsequently, we look at domain shift for tasks beyond text classification. Coreference resolution is fundamental for NLP applications, like question-answering and dialogue, but the models are typically trained and evaluated on one dataset. We use active learning to find spans of text in the new domain for users to label. Furthermore, we provide important insights on annotating spans for domain adaptation. Finally, we summarize the contributions of each chapter. We focus on aspects like the scope of applications and model complexity. We conclude with a discussion of future directions. Researchers may extend the ideas in our thesis to topics like user-centric active learning and proactive learning.