Includes bibliographical references (p. 253-276). / The main theme of the thesis is language endangerment, which represents a subfield of enquiry in sociolinguistics. The language under investigation is Afrikaans in its setting in Rehoboth, Namibia. Afrikaans was maintained as mother tongue and has been the dominant language as well as lingua franca in Rehoboth for some hundred years but could be losing these positions to English because of function and status loss and changes in the demographic mix of the area after Namibia's Independence in 1990. The Afrikaans-speaking Rehoboth Basters were initially the primary research subjects but it soon transpired that the Afrikaans-speaking Coloureds of the town need focused attention too because their language behaviour is as vital to the future of Afrikaans in the area.

Mainstream twentieth-century linguistics, a segregational approach, cannot explain the most obvious characteristics of language. The reasons for this are investigated. It is concluded that linguistics suffers from an incoherent conceptual framework which is the result of influences from three major sources: 1. The desire to establish linguistics as a proper science which led to the acceptance of a mechanistic and positivistic view of science and a pre-quantum conception of matter. 2. The language myth: there are many notions about language and related issues which we have inherited from our ancestors and tacitly accepted without scrutiny. Contemporary ideas about language are biased by this inherited stock of 'knowledge'. 3. Saussure's theory of language, later adopted and adapted by Chomsky, in which the 'true object of linguistic investigation' is abstracted away from what we ordinarily view as language. Together these three sources resulted in a peculiar view of language which cannot explain the most obvious things about it. The proposed alternative view, an integrational approach, redefines language in the holistic terms of a complex-systems approach. Language is the outcome of the dynamic interaction between social and physiological systems -- particular attention is paid to consciousness. Neither language, society or culture is an 'object', but is created through the interaction between individuals in communicative situations. Language is not 'being', but results from 'becoming'. Meaning is not given in advance, but created in each event of communication. Meaning is not a static closed system, but an open system which is dynamically constructed from moment to moment. Concepts of mathematical topology (fractal geometry and catastrophe theory), non-linear, dynamic, open and complex systems, and of chaology are used as conceptual tools to break away from the stronghold our inherited view of language has on our contemporary thinking about it.

The South African Chain Shift involved the raising of the short front vowels KIT, DRESS and TRAP when compared to Received Pronunciation (Lass & Wright 1986). This raising wasparticularly evident in the speech of middle class white speakers of South African English, as well as coloured speakers in the Cape. Recent scholarship has suggested that this raising is being reversed in the speech of young white South Africans. In particular, Bekker and Eley(2007) and Bekker (2009) report the lowering and retraction of TRAP. Mesthrie (2012a)reports not only the lowering and retraction of TRAP, but also the lowering of KIT and DRESS. In addition, scholars such as Mesthrie (2010) have found post-segregation deracialisation of middle class South African English. This thesis therefore investigates the extent to which the reversal of the older South African Chain Shift exists in the speech of white and black middle class South Africans from Cape Town. It furthermore explores the potential merger between TRAP~STRUT and KIT~DRESS. In so doing, 53 participants in sociolinguistic interviews are reported on. The Forced Alignment and Vowel Extraction Toolkit was utilised for formant measurement and extraction. Statistical testing via R was performed, including linear mixed-effects modelling, random forest analyses, conditional inference trees, Euclidean Distance measures, Welch's Two Sample t-tests and Pillai Scores. The analysis finds evidence of the reversal of the South African Chain Shift in the speech of participants under the age of 30. In particular, speakers aged between 18 and 25 participate the most in lowering KIT, DRESS and TRAP. Moreover, the short front vowels are retracting in the speech of younger Capetonians, indicating that within the process of vowel lowering, further innovation occurs via vowel retraction. The Reverse Vowel Shift is found to be a combination of push and pull chains: the fronting of FOOT causes the lowering of KIT, and the lowering of TRAP causes the lowering of DRESS. The retraction of TRAP furthermore causes the backing and raising of STRUT, such that an anti-clockwise rotation of the short front vowels (barring LOT) is evidenced in South African English. The Reverse Vowel Shift evident in Cape Town is similar to trends observed in California, Canada, southeast England, Ireland and Australia. This illustrates the effects of globalisation on English in South Africa, though internal motivations are also responsible.

Includes bibliographical references. / The work presented in this thesis relies on language recordings gathered during thirty months of fieldwork in White City-Jabavu, Soweto. The data was collected from children between the ages of two and nine, following anthropological participant observation, and through the use of an audio recorder. Strong attention was given to the sociolinguistics and structure of the language collected. This thesis is interested in issues of slang use among children and language contact, as part of the larger field of tsotsitaal studies. It is interested in: sociolinguistic issues of registers, slang, and style; and linguistic issues regarding the structural output of language contact. The main questions answered in the thesis concern whether children in White City use the local tsotsitaal, known as Iscamtho; and what particular kind of mixed variety supports their use of Iscamtho. Particularly, I focus on the prediction of the Matrix Language Frame model (Myers-Scotton 2002) regarding universal constraints on the output of language contact. This model was used previously to analyse Iscamtho use in Soweto. Using methodologies from three different disciplinary fields (anthropology, sociolinguistics, and linguistics) as well as four different analytic perspectives (participatory, statistical, conversational, and structural), I offer a thorough sociolinguistic and linguistic description of the children's language. I demonstrate that the universal constraints previously identified do not apply to a significant part of the children's speech, due to stylistic and multilingual practices in the local linguistic community. I further demonstrate that style, slang, and deliberate variations in language, can produce some unpredictable and yet stable structural output of language contact, which contradicts the main hypotheses of universal natural constraints over this output formulated by the Matrix Language Frame model.

Includes bibliographical references (leaves 214-223). / The thesis applies a social constructionist framework and Foucauldian Discourse Analysis to demonstrate that while Tsotsitaal was perceived by many respondents as a language of gangsters and criminals, evidence suggests that it is actually part of an ongoing identity construction for young, black, primarily male urban township residents in South Africa, which is performed through a subcultural style. By applying Myers-Scotton's Matrix Language Frame model to questionnaire and interview data collected in two Cape Town townships, Gugulethu and Khayelitsha, the thesis identifies the syntactic framework of Cape Town Tsotsitaal as Xhosa.

Includes bibliography.

Includes bibliographical references (leaves 338-359). / This dissertation presents a case study of the demise of !Ui-Taa languages in South Africa during the 20th century, with particular attention to N|u, the last surviving variety. The geographic focus is on Gordonia and Postmasburg Districts in the Northern Cape province. Drawing on the work of Diamond (1998), the author argues that food producing peoples (agriculturalists / pastoralists / colonialists) typically penetrate the hunter-gatherer territory and break down the ecological setting that sustains the hunter-gatherer mode of subsistence. The changes in ‘language ecology‘ (Haugen 1972) trigger transformations in social relations and place the languages of hunter-gatherers at risk of rapid language shift. This theory is in contra-distinction to the argument put forward by Brenzinger (1992a, b), Brenzinger et al (1991) and McConveIl (2000) that changes in language attitudes are the primary cause of language death. The second aspect of the dissertation deals with the particulars of the identity of the N|u language and its speakers. The speakers of N|u cail themselves N||n≠e ‘home people‘ or Sasi ‘Bushmen / eland'. Drawing on original research and the oral history, the author argues that Bleek’s (1929. 1956) categorisation of two !Ui-Taa varieties, S2 (||Ng) and 82a (≠Khomani), should be reconsidered, as these are dialects of one another. Reconsidering the distribution of !Ui-Taa languages has implications for understanding hunter-gatherer demographics and social organisation in pre-colonial South Africa.

Bibliography: leaves 178-187. / The aim of this research is to provide some ground work in the study of Luhyia socio-linguistics. A fair amount of research on indigenous forms of English has been conducted in South Africa as well as West Africa. According to Schmied (1991), Nigeria is covered by several books and articles on English, but other areas of Africa are relatively blank. Schmied himself has produced primary work on English in East Africa. Studies of language maintenance and language shift have been undertaken by eminent scholars such as Brenzinger (1992), Eastman (1990, 1992). However, it is Myers-Scotton's pioneering research on code-switching among the Luhyia speakers undertaken in the 1980s that proyided the initial inspiration and further foundation for this thesis. An attempt is made here to build on Myers-Scotton's insightful observations on code-switching among Luhyia speakers. In addition this thesis explores the type of English in use among the Luhyia, and its effects on the indigenous language with which it has come into contact.

This dissertation explores the language that instructors of a second language (such as Spanish) provide their students. Given that input (i.e., communicative samples of the target language (TL)) is a fundamental concept to all theories of language acquisition, this dissertation examines the quantity, the quality, and the frequency of the input that instructors provide in the classroom. This dissertation reports on a study of instructors' input (N=14). Participants included instructors at four different levels of Spanish (Elementary I and II and Intermediate I and II), from different native languages (native speakers (NS) of Spanish, and non-native (NNS) speakers), and include both female and male instructors. By recording, transcribing, and coding the speech samples from four class periods per language instructor (which accounted for more than 2700 minutes of instructor speech), this study looked at how much TL (Spanish), and how much English, instructors use in the classroom, the contexts in which they use English, the types of input modifications they make, as well as the frequency of verb types (person, number, tense, and mood). Analyses of the classroom input data revealed that all instructors but one used the TL between 78-95% of the class time. Based on previous literature, it is suggested that instructors in this study were much more homogeneous in their TL language use, and used it an appropriate amount of time. Results also indicated that NS produced a higher number of TL words than NNS. However, NNS produced a higher percentage of TL input of their total percentage per class. Results also showed that female instructors produced a higher percentage of speech. Regarding the use of English, the speech samples showed that there was a high amount of code switching in the language classrooms. The contexts of English use followed the trends from previous literature. English was predominantly used to explain tasks, and to explain grammar, and to move faster in the curriculum and save time. However, by analyzing the samples, it was proven that the use of English for this matter was contra productive. Every time that the language was switched to English there were extra explanations that might have not even been needed. Results also showed that NS instructors used English to bond/empathy with students, significantly more than NNS instructors. The higher amount of English use was found for Spanish Elementary II. Results support the idea that topic familiarity leads to more comprehension in the L2, because the concepts in Spanish Elementary II were more abstract, unlike the grammar presented in Spanish I. Regarding the modifications of the input, results showed that instructors at all levels modify their input; however, instructors at beginning levels simplify their input more, by slowing their rate, simplifying their syntax and producing high frequency vocabulary words. All instructors but one made use of technology to enhance their input. Results suggest that media is becoming a fundamental tool in the language classroom, and without multimedia or visual supplements instructors might not be taking advantage of the potential contribution of the learners' capacity to process that material in the visual mode as well. Results also showed that the frequency of the verbs in the classroom is different from what has been shown in previous research and in language corpora, suggesting that the context in which communication occurs determines the type of input that is given. By conducting a survey among the instructors' participants about their perspectives of language use in the classroom, it was shown that instructors have a wider perspective of the amount of input they provide in the class and the modifications they do, suggesting the importance of having serial classroom observations periodically. Ultimately, the results from this research will inform practitioners and materials developers as to the kinds of input learners need outside of class in order to complement what they receive in class. Second, I propose to use the data compiled in this project to create the first classroom based input corpus (database) for Spanish and made it available to researchers in second language acquisition. / A Dissertation submitted to the Department of Modern Languages and Linguistics in partial fulfillment of the requirements for the degree of Doctor of Philosophy. / Fall Semester 2015. / December 2, 2015. / Includes bibliographical references. / Michael J. Leeser, Professor Directing Dissertation; Michael Kaschak, University Representative; Gretchen Sunderman, Committee Member; Lara Reglero, Committee Member.

The ability to comprehend temporal reference is fundamental to human language and cognition. Thus, skilled language comprehension requires sensitivity to grammatical cues such as verbal aspect (Bybee et al., 1994). For L2 Spanish learners, however, (im)perfective aspect has the reputation of being particularly difficult and late acquired (Montrul & Slabakova, 2002). Although the topic of L2 production of aspect has been heavily researched (Andersen & Shirai, 1996; Bardovi-Harlig & Reynolds, 1995), little is known about L2 comprehension of aspect. Consequently, this research study uses psycholinguistic methods to shed light on our understanding of L1 and L2 processing of aspect in Spanish. The present study reports the findings from three language tasks (N= 98; 30 intermediate L2 learners, 33 advanced L2 learners, and 35 native Spanish speakers). First, two offline cloze tasks (a story-in-context cloze task and an isolated sentence cloze task) were utilized to measure the participants' knowledge of perfective and imperfective aspect. Second, a forced binary choice sentence-picture matching task considered how aspect restricts the mental representation of the endpoint of a situation. Third, a self-paced reading task examined how native and non-native speakers detect aspectual mismatches between logical and illogical sentences in real time. By including these three tasks, the overall research design of this study allowed for a comparison between native and non-native comprehension of Spanish aspectual contrasts. As it was expected, all three groups performed best on the highly monitored offline cloze tasks. The results from the sentence-picture matching task suggest that there was a perfective facilitation effect for accomplishment verbs, therefore supporting the Aspect Hypothesis (Andersen & Shirai, 1996), which predicts that accomplishments are applied to perfective aspect early in development. The results from the self-paced reading task suggest that the L2 learners demonstrated some shallow processing (Clahsen & Felser, 2006), as they showed difficulties in comprehending morphosyntactic information. The implications of these results for L2 acquisition and L2 instruction are discussed in this dissertation. / A Dissertation submitted to the Department of Modern Languages in partial fulfillment of the Doctor of Philosophy. / Spring Semester 2017. / March 27, 2017. / (Im)perfective Aspect, Language comprehension, Psycholinguistics, Second Language Acquisition / Includes bibliographical references. / Michael Leeser, Professor Directing Dissertation; Michael Kaschak, University Representative; Gretchen Sunderman, Committee Member; Lara Reglero, Committee Member.