The performance of automatic speech recognition (ASR) degrades significantly in natural environments compared to in laboratory assessments. Being a major source of interference, acoustic noise affects speech intelligibility during the ASR process. There are two main problems caused by the acoustic noise. The first is the speech signal contamination. The second is the speakers' vocal and non-vocal behavioural changes. These phenomena elicit mismatch between the ASR training and recognition conditions, which leads to considerable performance degradation. To improve noise-robustness, exploiting prior knowledge of the acoustic noise in speech enhancement, feature extraction and recognition models are popular approaches. An alternative approach presented in this thesis is to introduce eye gaze as an extra modality. Eye gaze behaviours have roles in interaction and contain information about cognition and visual attention; not all behaviours are relevant to speech. Therefore, gaze behaviours are used selectively to improve ASR performance. This is achieved by inference procedures using noise-dependant models of gaze behaviours and their temporal and semantic relationship with speech. `Selective gaze-contingent ASR' systems are proposed and evaluated on a corpus of eye movement and related speech in different clean, noisy environments. The best performing systems utilise both acoustic and language model adaptation.

Argument diagramming can scaffold the process of argumentation but only a few studies have investigated its impact on the quality of argumentative writing. This research contributed to this direction with two studies. An exploratory study investigated the impact of argument diagramming, applied as a paper-based or a computer-based method, on the quality of argumentative text. The computer method increased the refutations and overall quality of essays. The study highlights the significance of writers’ argumentative ability for interpreting improvement. A qualitative study looked into the impact of argument diagramming on the process of writing cognition through analysis of online process data, diagrams and essays of sixteen undergraduate students. Writers with myside bias schema used the method to increase counterarguments and refutations. Writers at lower level of pseudo-integration adopted more advanced strategies like weighing, and writers at middle level of pseudo-integration formed positions with qualifications. Needs at higher levels of argumentative ability are not met. The support of writing planning processes through argument diagramming affects mainly the semantic aspects of the text while the support of linearization processes affects mainly the rhetorical aspects. The analysis of interviews revealed that interacting with argument diagramming can improve awareness of argumentation schema, hence, a writer can progress from unaware, to aware-and-lost and aware-but-oriented. Improvement is signified as being sensitised to limitations, gaining knowledge of writing processes and the ability to self-regulate.

Policy, training and research reflect the importance of patient involvement in decisions about their care. Adoption demands certain skills though, may result in conflict, or be too threatening for patients. Using an iterative process, politeness theory was used to analyse the linguistic management of these threats and challenges in videos of GP consultations. The collaborative nature of GPs’ positive politeness had persuasive effects, whilst their negative strategies gave rise to examples of ambiguity causing confusion. Patients’ negative politeness demonstrated discomfort when presenting potentially contentious decisions whilst their use of positive politeness acted as a means of promoting cooperation. GPs used positive politeness when supporting patients’ decisions, offering reassurance and redressing damage to face, conversely disagreement was conveyed by the absence of such strategies and lack of reparative work. Difficulties were identified in the way in which space for patient participation was created and managed, and the strategies used to convey information. The contrast in GP responses to patients’ decisions highlighted how subtle barriers to participation can be. These findings demonstrate the complexity of language and meaning and the need for a more sophisticated understanding of language use in communication skills and related training, as well as associated research.

This thesis presents a multi-layered study of multilingual classroom discourse, with two teachers, in a primary school in Timor-Leste. The wider context for the study was a major shift in language-in-education policy – to the use of Portuguese and Tetum as media of instruction – on the independence of Timor-Leste in 2002. This is the first study in this context to use linguistic ethnography to investigate the ways in which teachers are navigating the policy shift and to analyse the links between multilingual classroom interaction and wider policy processes and language ideologies. Fieldwork for the study was conducted in 2012. It included classroom observation, note-taking, audio/video-recording of classroom interaction, interviews with teachers and with policymakers. The data analysis presented here centres on talk around Portuguese textbooks, in Tetum and Portuguese. The findings were as follows: (1.) teacher-pupil relationships were discursively co-constructed as strict and asymmetrical; (2.) code-switching practices evoked beliefs associated with hegemonic ideologies about bilingual education; and (3.) teachers mediated textbooks language and content by building bridges between textual knowledge and local knowledge. The study foregrounds teacher agency in language policy processes, but also makes connections with powerful political and academic discourses about language tied to nationhood and culture.

This thesis investigates the mechanisms that underlie the interaction of gesture and speech during the production and comprehension of language on a temporal and semantic level. The results from the two gesture-speech production experiments provide unambiguous evidence that gestural content is shaped online by the ways in which speakers package information into planning units in speech rather than being influenced by how events are lexicalised. In terms of gesture-speech synchronisation, a meta-analysis of these experiments showed that lexical items which are semantically related to the gesture's content (i.e., semantic affiliates) compete for synchronisation when these affiliates are separated within a sentence. This competition leads to large proportions of gestures not synchronising with any semantic affiliate. These findings demonstrate that gesture onset can be attracted by lexical items that do not co-occur with the gesture. The thesis then tested how listeners process gestures when synchrony is lost and whether preceding discourse related to a gesture's meaning impacts gesture interpretation and processing. Behavioural and ERP results show that gesture interpretation and processing is discourse dependent. Moreover, the ERP experiment demonstrates that when synchronisation between gesture and semantic affiliate is not present the underlying integration processes are different from synchronous gesture-speech combinations.

In this thesis, I have examined the screen translation of Japanese politeness into Cantonese dubbing as well as Chinese subtitling at three difference levels: (a) face threatening acts; (b) frequently used politeness markers and (c) discernment aspect of politeness. It is not difficult to find equivalents in the target languages for the politeness strategies exploited in dealing with face threatening acts. However, the indirectness expressed through certain commonly adopted politeness markers (such as negative interrogative) in Japanese cannot be conveyed into our target languages easily. Translators also encounter some difficulties when they deal with the discernment aspect of Japanese politeness (i.e. the distinction between plain, formal and honorific form). Both target languages are able to distinguish politeness of two levels instead of three as observed in the Japanese original. Finally, the screen translation, especially the dubbed version, of the two films that I examined demonstrates the dual role of a translated text as not only a reproduction of the original text but also a text which has its function in the target culture.

This thesis investigates the impact of an intensive programme of English for academic purposes upon the second language writing development of postgraduate students at the University of Birmingham. The study uses a 300,000 word corpus (EAPCORP) of essays from the beginning and end of the programme covering two separate years, in order to identify and measure written linguistic feature development. A multidimensional investigative approach underpins both of the two main analytical tools applied to the EAPCORP, with the basic premise that it is possible to identify register differences between different types of language by the assemblage and analysis of a large number of textual features. Firstly, Coh-Metrix is a programme employing a range of algorithms applied to a series of data bases to analyse the linguistic structure of texts. Secondly, MAT (Multidimensional Analysis Tagger) employs algorithms developed by Douglas Biber and uses an automated text tagger. The analyses suggest strongly that there has been progression from the initial production of a high frequency of features characteristic of speech to that more typical of academic writing. The results emphasise the importance of well-designed EAP programmes especially in uncertain economic contexts.

Language teacher cognition research has been a growing area of research interest in recent times. However, the majority of studies of teacher cognition in relation to teacher change have focused on pre-service and novice, rather than experienced teachers. Additionally, most research has been carried out in contexts other than the public secondary school EFL settings where most English education goes on in the world. The present study aims to redress this tendency by deeply investigating the beliefs and practices of three experienced Thai EFL teachers working in a public secondary school in Thailand. Also investigated are the sources of their beliefs and how these have developed over time throughout the teachers’ lives. Using a range of methods including semi-structured interviews, stimulated recall, classroom observation, Snake interviews, repertory grid interviews and research diary, the present study focuses on Thai EFL teachers’ beliefs concerning appropriate methodology, a notion of importance both at the macro level of educational change and at the micro level of individual classrooms, where teachers can be viewed as key change agents. Adopting a personal construct theoretical perspective, the study reveals that the teachers’ beliefs and practices have been particularly informed by their core constructs, which have deep roots in their personal biographies and are tightly tied to their sense of self. Furthermore, the findings cast light on the extent to which individual teachers may or may not change according to their core constructs. These results carry important implications for those involved in change management, teacher education, and teacher cognition research.

This research uses literary resources as evidence against the argument that names are potentially semantically meaningless entities. A secondary goal is to highlight and discuss the value of onomastics from both a literary and linguistic perspective. The thesis proposes a methodology for the assessment of literary sources based on genre, arguing that names, and genre in turn, may be defined through their respective engagement with thematic considerations, providing a relevant critical structure by which to assess the application or construction of names within fiction. The proposed methodology is first used to assess the placenames within dystopian literature, taking Orwell’s 1984 (1949), Huxley’s Brave New World (1931), and Zamyatin’s We (1924) as exemplar texts for the genre. The emblematic themes identified within the onymic patterns (propaganda, classification and regulation) all share a common thematic root: power and control. In order to assess the validity of this approach, the fictional worlds depicted in a selection of other dystopic texts are also examined. A special study is made of terrapsychology and fictional ontology, as well as of three distinct subgenres of the gothic. Case studies of the latter are each focused around a different ontological mode (fictional, part-fictional, and non-fictional placenames), covering the fantastic world of Peake’s Gormenghast setting (two texts published in 1946 and 1950), Lovecraft’s variant New England county (six texts, 1922 to 1936), and the representative contemporary setting of Brook’s World War Z (2006), respectively.

Studies have shown that permanent childhood hearing impairment can have detrimental consequences for spoken-language development. It is widely accepted that early detection and intervention may improve outcomes for profoundly hearing-impaired children. However, few studies evaluate the influence on the families or give particular attention to children with mild-to-severe hearing impairments. This research used spoken language and communication to focus on a range of factors that may influence outcomes for children with permanent sensorineural hearing impairments. In two studies, children with a range of hearing impairments, aged 32 to 85 months (mean = 63, s = 14) were audio- and video-recorded at home interacting with a major care-giver. Measures of spoken language for the children and their interlocutors were derived from transcripts. Controlling for the age of the child, spoken-language outcomes were evaluated in relation to factors such as the severity of the child's hearing impairment, age of intervention and the language addressed to the child during the interaction. The first study indicated that hearing severity, excluding profound hearing impairments, may not be the most important influence on spoken language. However, earlier intervention corresponded to better language performance. The second study failed to replicate these findings but suggested that a complex interaction of factors - including earlier referral for hearing assessment - may influence spoken language production for hearing-impaired children. Questionnaires revealed the families' attitudes and feelings towards the diagnosis of their child's hearing-impairment, showing that parents often experienced negative emotions at the time and that intervention provisions often fail to take parental attitude into consideration at this time. Results also suggested that earlier and prompt intervention for childhood hearing-impairment is viewed as beneficial and essential. This may have important implications for habilitation and intervention programmes for hearing-impaired children and their families. The conclusions suggest that further studies - which evaluate and detail the potential long-term benefits of very early intervention for hearing-impaired children - need to be conducted.