The sourcing, preservation and documentation of emigrant letter collections is now gathering pace, with the Internet providing a significant new forum for the dissemination of long-hidden archives. Most existing digital letter collections consist of unannotated versions of original manuscripts. The digitisation process has made the letters more accessible and has also increased their searchability. However, relatively few emigrant letter projects have moved beyond the digitisation stage to exploit text content and enhance usability and searchability through the use of digital technologies. This thesis explores some of the opportunities and challenges of working with digitised historical emigrant letter collections. Essentially, the thesis does two things: first, it uses digital technologies (specifically corpus and computational methods of analysis) to explore the language of emigrant letters, building on the existing body of research – primarily by historians – to offer another way into migrant correspondence; second, it proposes a system of markup for capturing metadata relating to emigrant letters – metadata which can then be used to interconnect resources enabling users to carry out more nuanced and sophisticated searchers. I argue that my proposed system could be widely applied to emigrant letter collections, facilitating much greater interdisciplinary and collaborative analysis of such material than has been undertaken hitherto.

This thesis appears in three parts: Modules I, II, & III. The purpose of these units was to argue that Generation 1.5 (G1.5) learners are a distinct group of English language learners with unique ways of representing themselves in academic writing, and to identify those salient linguistic differences among G1.5, traditional ESL, and NS student writers. Using multiple methodologies, the text explores the discourse patterns of G1.5 students in their academic writing. Elements in each section include: Module I: o A discussion and literature review of research on Generation 1.5 students and design criteria for an extended corpus study. Module II: o A pilot study of early results from a corpus study comparing G1.5, ESL, and NES student academic writing, with a focus of pronoun and modal use. Module III: o A study involving surveys and interviews to evaluate what both students and instructors consider good academic writing and expect of student essays. o Corpus data from G1.5, ESL, and NS student corpora to determine lexicogrammatical and syntactic patterns in G1.5 student writers and how they differ from both ESL and NS students. Salient features are analyzed using a framework where features are mapped onto an adapted version of Halliday‘s (2004) three macrofunctions of language, allowing for an analysis of semantic and lexico-grammatical features in terms of ideational, interpersonal, and textual positioning. o Case studies of three essays to test corpus results and a framework of selfrepresentation against individual performance. The resulting text concludes that G1.5 students‘ self-representation in writing is distinct from other student writers, and manifests in their semantic choices, narrative style, and elements of a hybrid of academic and personal/interpersonal writing.

This thesis investigates four factors that can significantly affect the positive-negative semantic prosody of two high-frequency verbs, CAUSE and HAPPEN. It begins by exploring the problematic notion that semantic prosody is collocational. Statistical measures of collocational significance, including the appropriate span within which evaluative collocates are observed, the nature of the collocates themselves, and their relationship to the node are all examined in detail. The study continues by examining several semi-preconstructed phrases associated with HAPPEN. First, corpus data show that such phrases often activate evaluative modes other than semantic prosody; then, the potentially problematic selection of the canonical form of a phrase is discussed; finally, it is shown that in some cases collocates ostensibly evincing semantic prosody occur in profiles because of their occurrence in frequent phrases, and as such do not constitute strong evidence of semantic prosody. Finally, register-specific examination of grammatical patterning of CAUSE shows that both register and patterning can affect semantic prosody. This study shows that although positive-negative semantic prosody is an important aspect of meaning, it is potentially problematic, and any claims that a word or phrase has a positive or negative semantic prosody may require taking these factors into consideration.

This thesis analyses the process of construction of linguistic meaning from a diachronic perspective. The analysis is based on the theories of intertextuality and social construction and applies a corpus analysis of collocations and paraphrases of key notions discussed in Darwin’s correspondence. In particular, the focus of the analysis is on the terms \(species\) and \(varieties\) with the aim to observe how the meanings of these terms are formed in a process of social negotiation. The thesis analyses the difference in the meanings of these terms, but also focuses on the diachronic dimension of their use in the correspondence, which allows for the observation of how different interpretations of meanings emerge in discourse. Thus far, diachronic studies in the field of corpus linguistics focused on comparing different historical corpora rather than observing the diachronic change of the immediate contextual environment of particular terms. The results presented in this thesis show that the meanings of terms are not only formed of different interpretations in discourse, but that these interpretations can be specific to particular temporal spans in discourse.

As writing and reading are interconnected activities and as learners are at the centre of learning process, this thesis employed a reader response activity in English classes of two disciplinary different college-level students to examine the SL learners’ roles in the writing process. This study was motivated by three main research questions in the areas of the effect of learner variables on the choice of reading topics and the length of learners’ responses, the textual characteristics of the responses, and writer’s self-representation and reader engagement strategies. A sample of 600 student texts was analysed using referential statistics for addressing the question on learner variables, and text analysis, both manually and through a corpus tool, for the other two questions. The findings showed these students differed in their choices of reading topics and that the topic, the students’ linguistic ability and discipline affected their responses. The text analysis revealed that most responses belonged to the levels 2 and 3 of cognitive engagement and that their generic structuring consisted of three main rhetorical moves. It also illustrated that personal pronouns were used for self-representation and reader engagement and had various rhetorical functions. The findings have wide-ranging pedagogical implications.

Nominalization and agentless passives have attracted sustained attention in critical linguistics and Critical Discourse Analysis (CDA), where, it is argued, they 'mystify', i.e., reduce reader comprehension of, the role of social actors in depictions of events, particularly in news media discourse. Yet the capacity of readers to generate inferences automatically from textual cues and background information has not been adequately reflected in CDA accounts of reader cognition. The question of whether particular instances of nominalization or agentless passives actually reduce reader comprehension of social actors' agentive roles was put to an empirical test by asking volunteer readers to identify social actors deleted from newspaper editorials by the addition ofnominalization and agentless passives. While readers accurately inferred the missing actors in a majority of cases, textual constraints and background knowledge appeared to affect inference accuracy in ways generally consistent with the predictions of the idealized reader (IR) framework presented in O'Halloran (2003). It is argued that robust models of reader cognition should be incorporated more widely into CDA studies to prevent researchers from overestimating the capacity of textual features to mystify social actors to readers.

This thesis is an attempt to characterize the utterances in conversation. Following the principles of Sinclair & Caulthard (1975), it proposes a descriptive framework which is based on the concepts of 'class', 'structure' and 'system'. Chapter One argues against the position that utterances are multi-functional and the illocutionary forces they carry are largely indeterminate, hence they are not describable in categorial terms. It points out that such a position is a misconception arising from the lack of consistent criteria when characterizing utterances. It then examines studies in three major areas which would give insight to the setting up of a descriptive framework: speech act theory, conversational analysis and discourse analysis. Chapter Two gives an overall account of the descriptive framework. Its basic theoretical assumption is that conversation is describable in terms of a hierarchical rank scale, consisting of acts, moves, exchanges, sequences and transactions. Utterances are characterized as different primary classes of acts according to which element of structure of an exchange they operate at . Three primary classes are identified: those operating at the head of an Initiating Move are Initiating Acts, those operating at the head of a Responding Move are Responding Acts and those operating at the head o:f a Follow-up Move are Follow-up Acts. For each primary class, subclasses are identified according to their predictive assessment of what follows. The choices of subclasses which are available at each element of structure are presented in the form of a system. Chapters Three to Six discuss the four subclasses of Initiating Act, Elicitations, Requestives, Directives and Informatives respectively. Chapter Seven discusses Responding Act and its subclasses; and Chapter Eight discusses Follow-up Act and its subclasses. In Chapter Nine, the entire descriptive framework is applied to a piece of conversation. Its merits and limitations are discussed.

This thesis aims at developing an analytical model for differentiation of translation correspondences and for grouping lexical items according to their semantic similarities. The model combines the language in use theory of meaning with the distributional corpus linguistics method. The identification of translation correspondences derives from the exploration of the occurrence of lexical items in the parallel corpus. The classification of translation correspondences into groups is based on the substitution principle, whereas the distinguishing features used to differentiate between lexical items emerge as a result of the study of local contexts in which these lexical items occur. The distinguishing features are analysed with the help of various statistical measurements. The results obtained indicate that the proposed model has advantages over the traditional approaches that rely on the referential theory of meaning. In addition to contributing to lexicology the model also has its applications in practical lexicography and in language teaching.

This study examines the appearance of evaluative language and strategic adherence to the principle of objectivity in hard news reporting. While these concepts have traditionally been treated as distinct from each other, they are related. The study highlights a key relational tension between taking an evaluative stance and adhering to the requirement for objectivity. This relational tension is pointed out by Richardson (2007:87), who argues that news reporting is a value-laden process, and that journalists make language choices to express those values while remaining 'journalistically objective'. This demonstrates a two-sided tension that journalists strategically handle in news reporting. In this thesis, I examine this important aspect of the characteristics of hard news reporting, that is, how evaluative language and objectivity concurrently appear in the news. This is dealt with by considering 16 hard news articles from the Daily News and Mmegi. The news articles cover the 2011 nationwide public sector workers' strike in Botswana. In this study a combination of quantitative and qualitative approaches is used to compare how the two newspapers use evaluative language, and how they mitigate such evaluative language in order to remain objective. Four research questions are addressed in this study: 1. How frequent and varied is the use of evaluative language in the Daily News and Mmegi? 2. Given the legal requirement for press objectivity in Botswana, what strategies are used by the Daily News and Mmegi to mitigate such evaluative language? 3. Are there significant differences in the strategies employed by the two newspapers to use and mitigate evaluative language? 4. Can any differences in the strategies of evaluation and mitigation be related to the newspapers' political positioning or the nature of the event covered? Evidence from analysing comparable news articles indicates that, when studied within the context of hard news reporting, evaluation and objectivity are not mutually exclusive concepts, but that the variety of linguistic resources employed in news articles affords journalists success in expressing evaluative content while maintaining the objectivity ideal.

The broad aim of this study is to investigate the discourse of 'alibrāliyah' in the Saudi socio-political context. 'alibrāliyah', corresponding to liberalism in English, is a loanword to Arabic that started to be contested recently amongst opposing groups who attempt to charge it with their own ideological meaning. Due to the lack of studies that investigate the ideological battles between the groups in contesting movements in Saudi context, it is found significant to examine the recent movement of 'alibrāliyah' as identified by different groups, mainly conservatives and progressives, in their battle to gain or maintain hegemony. This thesis designs a framework for examining the diachronic construction of 'alibrāliyah' at macro and micro levels. It combines Corpus Linguistics methods with Discourse-Historical Approach (DHA) and Discourse Theory (DT). The thesis employed the developed model by examining the discourse of 'alibrāliyah' in a corpus of Saudi newspaper articles that discuses mainly the issue of 'alibrāliyah'. The macro corpus analysis of 575 articles has revealed that 'alibrāliyah' is mainly associated with the discourse of religion and the discourse of human rights in which it shifts from being associated with the religious discourse to associating it with the discourse of freedom and human rights at the end of the period. The micro discourse analysis of a sample of texts shows in general that the construction of 'alibrāliyah' shifts from being secular and contradictory to Islam into being compatible to non-fundamental Islam in which 'alibrāliyah' can be adopted within Islam to guarantee the freedom of individuals. It is concluded that this shift in meaning represents the success of liberals to establish their identity and to have power in Saudi society. It is also concluded that the developed model for this thesis can help by intricately investigating the construction of ideological movements and the relation between the groups struggling for hegemony.