ASU Electronic Theses and Dissertations
This collection includes most of the ASU Theses and Dissertations from 2011 to present. ASU Theses and Dissertations are available in downloadable PDF format; however, a small percentage of items are under embargo. Information about the dissertations/theses includes degree information, committee members, an abstract, supporting data or media.
In addition to the electronic theses found in the ASU Digital Repository, ASU Theses and Dissertations can be found in the ASU Library Catalog.
Dissertations and Theses granted by Arizona State University are archived and made available through a joint effort of the ASU Graduate College and the ASU Libraries. For more information or questions about this collection contact or visit the Digital Repository ETD Library Guide or contact the ASU Graduate College at gradformat@asu.edu.
Displaying 1 - 10 of 401
Filtering by
- All Subjects: Machine Learning
- All Subjects: Linguistics
Description
Sparsity has become an important modeling tool in areas such as genetics, signal and audio processing, medical image processing, etc. Via the penalization of l-1 norm based regularization, the structured sparse learning algorithms can produce highly accurate models while imposing various predefined structures on the data, such as feature groups or graphs. In this thesis, I first propose to solve a sparse learning model with a general group structure, where the predefined groups may overlap with each other. Then, I present three real world applications which can benefit from the group structured sparse learning technique. In the first application, I study the Alzheimer's Disease diagnosis problem using multi-modality neuroimaging data. In this dataset, not every subject has all data sources available, exhibiting an unique and challenging block-wise missing pattern. In the second application, I study the automatic annotation and retrieval of fruit-fly gene expression pattern images. Combined with the spatial information, sparse learning techniques can be used to construct effective representation of the expression images. In the third application, I present a new computational approach to annotate developmental stage for Drosophila embryos in the gene expression images. In addition, it provides a stage score that enables one to more finely annotate each embryo so that they are divided into early and late periods of development within standard stage demarcations. Stage scores help us to illuminate global gene activities and changes much better, and more refined stage annotations improve our ability to better interpret results when expression pattern matches are discovered between genes.
ContributorsYuan, Lei (Author) / Ye, Jieping (Thesis advisor) / Wang, Yalin (Committee member) / Xue, Guoliang (Committee member) / Kumar, Sudhir (Committee member) / Arizona State University (Publisher)
Created2013
Description
Many researchers have pointed out that sentence complexity plays an important role in language maturity. Using cohesive devices is a critical method to composing complicated sentences. Several grammatical researchers give cohesive devices different definitions and categories in the perspective of pure linguistics, yet little is known about the Chinese learners' acquisition situations of cohesive devices in the field of Teaching Chinese as a Foreign Language (TCFL). Combined with these definitions and pedagogical theories, the acquisition situations of four grammatical features of cohesive devices and eleven logical relations are discussed in this thesis. This thesis expects that through discovering different features of cohesive devices among different student levels, educators of Chinese will gain a more comprehensive understanding of the acquisition orders and features of conjunctive devices. In this study, I examine the teaching orders of cohesive devices in selected textbooks from first-year Chinese through fourth-year Chinese. Three groups of students were required to complete two essays based on the same topics and prompts. Twenty-eight valid writing samples are examined in total, including ten writing samples from fourth-year students, another ten from third-year students, and eight from second-year students. The results show that there are no obvious differences among the three levels of students in their use of certain grammatical features and logical relations of cohesive devices. Students in these three levels have difficulty understanding how to connect paragraphs together fluently and accurately in their compositions. Pedagogical implications include some suggestions about designing instructional writing assignments in order to give more clearly pedagogical instructions for teaching cohesive devices. In addition, comprehensible directions that explain which logical relations should be taught every academic year are proposed.
ContributorsHan, Jining (Author) / Spring, Madeline K (Thesis advisor) / Ling, Xiaoqiao (Committee member) / Oh, Young (Committee member) / Arizona State University (Publisher)
Created2013
Description
This work presents two complementary studies that propose heuristic methods to capture characteristics of data using the ensemble learning method of random forest. The first study is motivated by the problem in education of determining teacher effectiveness in student achievement. Value-added models (VAMs), constructed as linear mixed models, use students’ test scores as outcome variables and teachers’ contributions as random effects to ascribe changes in student performance to the teachers who have taught them. The VAMs teacher score is the empirical best linear unbiased predictor (EBLUP). This approach is limited by the adequacy of the assumed model specification with respect to the unknown underlying model. In that regard, this study proposes alternative ways to rank teacher effects that are not dependent on a given model by introducing two variable importance measures (VIMs), the node-proportion and the covariate-proportion. These VIMs are novel because they take into account the final configuration of the terminal nodes in the constitutive trees in a random forest. In a simulation study, under a variety of conditions, true rankings of teacher effects are compared with estimated rankings obtained using three sources: the newly proposed VIMs, existing VIMs, and EBLUPs from the assumed linear model specification. The newly proposed VIMs outperform all others in various scenarios where the model was misspecified. The second study develops two novel interaction measures. These measures could be used within but are not restricted to the VAM framework. The distribution-based measure is constructed to identify interactions in a general setting where a model specification is not assumed in advance. In turn, the mean-based measure is built to estimate interactions when the model specification is assumed to be linear. Both measures are unique in their construction; they take into account not only the outcome values, but also the internal structure of the trees in a random forest. In a separate simulation study, under a variety of conditions, the proposed measures are found to identify and estimate second-order interactions.
ContributorsValdivia, Arturo (Author) / Eubank, Randall (Thesis advisor) / Young, Dennis (Committee member) / Reiser, Mark R. (Committee member) / Kao, Ming-Hung (Committee member) / Broatch, Jennifer (Committee member) / Arizona State University (Publisher)
Created2013
Description
This study is about Thai English (ThaiE), a variety of World Englishes that is presently spoken in Thailand, as the result of the spread of English and the recent Thai government policies towards English communication in Thailand. In the study, I examined the linguistic data of spoken ThaiE, collected from multiple sources both in the U.S.A. and Thailand. The study made use of a qualitative approach in examining the data, which were from (i) English interviews and questionnaires with 12 highly educated Thai speakers of English during my fieldwork in the Southwestern U.S.A., Central Thailand, and Northeastern Thailand, (ii) English speech samples from the media in Thailand, i.e. television programs, a news report, and a talk radio program, and (iii) the research articles on English used by Thai speakers of English. This study describes the typology of ThaiE in terms of its morpho-syntax, phonology, and sociolinguistics, with the main focus being placed on the structural characteristics of ThaiE. Based on the data, the results show that some of the ThaiE features are similar to the World Englishes features, but some are unique to ThaiE. Therefore, I argue that ThaiE is structurally considered a new variety of World Englishes at the present time. The findings also showed an interesting result, regarding the notion of ThaiE by the fieldwork interview participants. The majority of these participants (n=6) denied the existence of ThaiE, while the minority of the participants (n=5) believed ThaiE existed, and one participant was reluctant to give the answer. The study suggested that the participants' academic backgrounds, the unfamiliar notion of ThaiE, and the level of the participants' social interaction with everyday persons may have influenced their answers to the main research question.
ContributorsRogers, Uthairat (Author) / Gelderen, Elly van (Thesis advisor) / Mailhammer, Robert (Committee member) / Adams, Karen (Committee member) / Arizona State University (Publisher)
Created2013
Description
This research conducts two methods of rhetorical analysis of State of the Union Addresses: 1. Computational linguistic analysis of all State of the Union Addresses from 1790-2007, and 2. Close-readings and rhetorical analyses of two addresses: one by President Truman and one by President Reagan. This research shows the following key findings: 1. I am able to see general shifts in the authors' approaches to the State of the Union Address through historical computational analyses of the content of all speeches, and 2. Through close readings, I can understand the impact of the author's ethos and the historical context on the addresses, something that would not be readily revealed in a computational analysis. This study starts with a historical computational linguistic analysis of all State of the Union Addresses between 1790 and 2007. The study follows with close-readings of two State of the Union Addresses from the early and late Cold War period in-context: 1. Harry Truman's 1951 Address and 2. Ronald Reagan's 1986 Address. The main conclusions drawn from this research are that close-readings of State of the Union Addresses cannot be replaced by computational analyses, but can work in tandem with computerized text analysis to reveal shifts in rhetorical and topical features. This paper argues that there must be more close analyses in coordination with large-scale text analysis in order to understand the complexities of rhetorical situations.
ContributorsWegner, Peter (Author) / Goggin, Maureen (Thesis advisor) / Boyd, Patricia (Committee member) / Goggin, Peter (Committee member) / Arizona State University (Publisher)
Created2013
Description
Despite the vast research on language carried out by the generative linguistics of Noam Chomsky and his followers since the 1950s, for theoretical reasons (mainly their attention to the mental abstraction of language structure rather than language as a performed product), historical linguistics from the start lay outside their research interest. This study is an attempt to bridge the gap between the formalism and theoretical constructs introduced by generative grammar, whose ultimate goal is to provide not only a description but also an explanation to linguistic phenomena, and historical linguistics, which studies the evolution of language over time. This main objective is met by providing a formal account of the changes hwæðer undergoes throughout the Old English (OE) period. This seemingly inconspicuous word presents itself as a case of particular investigative interest in that it reflects the different stages proclaimed by the theoretical assumptions implemented in the study, namely the economy principles responsible for what has become known as the CP cycle: the Head Preference Principle and the Late Merge Principle, whereby pronominal hwæðer would raise to the specifier position for topicalization purposes, then after frequent use in that position, it would be base-generated there under Late Merge, until later reanalysis as the head of the Complementizer Phrase (CP) under Head Preference. Thus, I set out to classify the diverse functions of OE hwæðer by identifying and analyzing all instances as recorded in the diachronic part of the Helsinki Corpus. Both quantitative and qualitative analyses of the data have rendered the following results: 1) a fully satisfactory functional and chronological classification has been obtained by analyzing the data under investigation following a formal theoretical approach; and 2) a step-by-step historical analysis proves to be indispensable for understanding how language works at the abstract level from a historical point of view. This project is part of a growing body of research on language change which attempts to describe and explain the evolution of certain words as these change in form and function.
ContributorsParra-Guinaldo, Víctor (Author) / Gelderen, Elly van (Thesis advisor) / Bjork, Robert (Committee member) / Nilsen, Don L. F. (Committee member) / Arizona State University (Publisher)
Created2013
Description
Complaints, characterized by LaForest (2002), are expressions "of dissatisfaction addressed by an individual A to an individual B concerning behavior on the part of B that A feels is unsatisfactory," (p. 1596) have been studied in the language of English speakers since the 1980's (Boxer, 1993a; 1993b; 1996; House & Kasper, 1981; Murphy & Neu, 1996; Trenchs, 1995; Vázquez, 2011; Wolfe & Powell, 2006). However, only a few studies on Spanish-language complaints have been carried out (Bolívar, 2002a; Márquez Reiter, 2005; Pinto & Raschio, 2008). Due to the lack of studies analyzing complaints among second generation Mexican-American Spanish-English bilinguals in the United States, role-plays were collected from 21 participants, ten males and eleven females, who interacted with a female interlocutor. The data was analyzed using Spencer-Oatey's (2005) Rapport Management in order to gain a better understanding of this population's politeness strategies used in complaining both in Spanish and English. In addition to acting out the role-plays, the participants were asked to fill out a Language Experience and Proficiency Questionnaire (LEAP-Q), in order to assess language proficiency. Upon completion of the role-plays, the participants completed a post role-play questionnaire, which evaluated their impressions of the interactions. The strategies used in the complaints included, but were not limited to: complaining/accusing, reason/explanation/ justification, threatening, suggesting/requesting/commanding, and providing information. The results showed that for the Spanish complaints the participants preferred the use of reason/explanation/justification, while they preferred suggesting/requesting/commanding in the English complaints. In addition, in both situations the participants chose to respect the association principle, however, this result was not statistically significant. With respect to face sensitivities, the participants chose to enhance the interlocutor's identity face in both the English and Spanish. It is concluded that these participants do not demonstrate a transfer of strategies from one language to another. Furthermore, no significant gender differences were observed. Moreover, the participants show a tendency toward positive politeness, which falls in line with other Hispanic cultures such as Cubans, Spaniards, Argentineans, Uruguayans, Peruvians, and Venezuelans. Although this study adds to the literature of Spanish in the U.S. pragmatics, further study of this population is needed.
ContributorsElias, Mariam (Author) / García Fernandez, Carmen (Thesis advisor) / Lafford, Barbara (Thesis advisor) / Cerron-Palomino, Alvaro (Committee member) / Arizona State University (Publisher)
Created2013
Description
This study investigates the effectiveness of the use of Concept-Based Instruction (CBI) to facilitate the acquisition of Spanish mood distinctions by second semester second language learners of Spanish. The study focuses on the development of Spanish mood choice and the types of explanations (Rule-of-Thumb vs. Concept-based) used by five students before and after being exposed to Concept-Based Instruction regarding the choice of Spanish mood following various modalities .The students in this study were presented with a pedagogical treatment on Spanish mood choice that included general theoretical concepts based on Gal'perin's (1969, 1992) didactic models and acts of verbalization, which form part of a Concept-Based pedagogical approach. In order to ascertain the effectiveness of the use of concept-based tools to promote the ability to use Spanish mood appropriately over time, a pre and post-test was administered to the group in which students were asked to respond to prompts containing modalities that elicit the indicative and subjunctive moods, indicate their level of confidence in their response, and verbalize in writing a reason for their choice. The development of these abilities in learners exposed to CBI was assessed by comparing pre and post-test scores examining both forms and explanations for the indicative and subjunctive modality prompts given. Results showed that students continued to rely on Rule-of-Thumb explanations of mood choice but they did expand their use of conceptually-based reasoning. Although the quantitative and qualitative analyses of the results indicate that most students did improve their ability to make appropriate mood choices (forms and explanations) after the CBI treatment, the increased use of conceptually-based explanations for their mood choices led to both correct and incorrect responses.
ContributorsBeus, Eric (Author) / Lafford, Barbara (Thesis advisor) / Beas, Omar (Committee member) / Cerron-Palomino, Alvaro (Committee member) / Arizona State University (Publisher)
Created2013
Description
Image understanding has been playing an increasingly crucial role in vision applications. Sparse models form an important component in image understanding, since the statistics of natural images reveal the presence of sparse structure. Sparse methods lead to parsimonious models, in addition to being efficient for large scale learning. In sparse modeling, data is represented as a sparse linear combination of atoms from a "dictionary" matrix. This dissertation focuses on understanding different aspects of sparse learning, thereby enhancing the use of sparse methods by incorporating tools from machine learning. With the growing need to adapt models for large scale data, it is important to design dictionaries that can model the entire data space and not just the samples considered. By exploiting the relation of dictionary learning to 1-D subspace clustering, a multilevel dictionary learning algorithm is developed, and it is shown to outperform conventional sparse models in compressed recovery, and image denoising. Theoretical aspects of learning such as algorithmic stability and generalization are considered, and ensemble learning is incorporated for effective large scale learning. In addition to building strategies for efficiently implementing 1-D subspace clustering, a discriminative clustering approach is designed to estimate the unknown mixing process in blind source separation. By exploiting the non-linear relation between the image descriptors, and allowing the use of multiple features, sparse methods can be made more effective in recognition problems. The idea of multiple kernel sparse representations is developed, and algorithms for learning dictionaries in the feature space are presented. Using object recognition experiments on standard datasets it is shown that the proposed approaches outperform other sparse coding-based recognition frameworks. Furthermore, a segmentation technique based on multiple kernel sparse representations is developed, and successfully applied for automated brain tumor identification. Using sparse codes to define the relation between data samples can lead to a more robust graph embedding for unsupervised clustering. By performing discriminative embedding using sparse coding-based graphs, an algorithm for measuring the glomerular number in kidney MRI images is developed. Finally, approaches to build dictionaries for local sparse coding of image descriptors are presented, and applied to object recognition and image retrieval.
ContributorsJayaraman Thiagarajan, Jayaraman (Author) / Spanias, Andreas (Thesis advisor) / Frakes, David (Committee member) / Tepedelenlioğlu, Cihan (Committee member) / Turaga, Pavan (Committee member) / Arizona State University (Publisher)
Created2013
Description
Wolfgang Haas is an award-winning Austrian author known primarily for his entertaining and quirky detective novels which follow the misadventures of Simon Brenner, an Austrian private investigator. These novels are notable for their subtle and not-so-subtle critiques of contemporary Austrian society and culture, their sometimes grisly content, and their unique and colloquial use of the Austrian variety of the German language. Haas has received numerous literary awards in the German-speaking world and attributes his success to the unique way he tells his stories, rather than the stories themselves. Of the seven Brenner novels that have been published thus far, only one is available in English translation, and he remains virtually unknown in the English-speaking world. This thesis includes a brief biography of Haas and an overview of his career, an analysis of his unique writing style and the problems they pose for a translator, and an English translation of the first two chapters of the novel Silentium! (1999).
ContributorsGeisler, Paul (Author) / Gilfillan, Daniel (Thesis advisor) / Ghanem, Carla (Committee member) / Hogue, Cynthia (Committee member) / Arizona State University (Publisher)
Created2013