A model is presented for the interactive generation of discourse in a clearly defined area of verbal behaviour (of interest in its own right) in procedural rather than purely descriptive terms. It is built up from an examination of the relation between form and function in the recorded data (fairly informal games of 'Twenty Questions' in English and three other contrasting languages) for evidence of - and justification for - the some sixty discrete discourse acts involved. The latter are then treated as complex input-output processing units called 'demons' operating within context formalizations termed 'frames'. The relationship between this procedural approach and descriptive discourse analysis and speech act theory is continually assessed. This discourse component is subsequently integrated with components dealing with sentence generation and game strategies to produce a complete model (incorporating syntax, semantics and pragmatics) whose operation is illustrated diagramatically for selected games. It is compared with a simplified but working program for Twenty Questions, and an attempt is made to test some of the replies embodied in the model. The methodology behind the thesis is a synthesis of approaches from Artificial Intelligence and Cognitive Psychology as well as from within Linguistics itself. The belief is that such an integrative project is a step towards a theory of context-specific Pragmatics meaningful to all three disciplines.

This thesis presents accounts of a range of linguistic phenomena in an extended categorial framework, and develops proposals for processing grammars set within this framework. Linguistic phenomena whose treatment we address include word order, grammatical relations and obliqueness, extraction and island constraints, and binding. The work is set within a flexible categorial framework which is a version of the Lambek calculus (Lambek, 1958) extended by the inclusion of additional type-forming operators whose logical behaviour allows for the characterization of some aspect of linguistic phenomena. We begin with the treatment of extraction phenomena and island constraints. An account is developed in which there are many interrelated notions of boundary, and where the sensitivity of any syntactic process to a particular class of boundaries can be addressed within the grammar. We next present a new categorial treatment of word order which factors apart the specification of the order of a head's complements from the position of the head relative to them. This move has the advantage of allowing the incorporation of a treatment of grammatical relations and obliqueness, as well as providing for the treatment of Verb Second phenomena in Germanic languages. A categorial treatment of binding is then presented which integrates the preceding proposals of the thesis, handling command constraints on binding in terms of relative oliqueness and locality constraints using the account of linguistic boundaries. Attention is given to the treatment of long distance reflexivization in Icelandic, a phenomenon of interest because of its unusual locality behaviour. Finally, a method is developed for parsing Lambek calculus grammars which avoids the efficiency problems presented by the occurrence of multiple equivalent proofs. The method involves developing a notion of normal form proof and adapting the parsing method to ensure that only normal form proofs are constructed.

This thesis seeks to empirically examine six contemporary theories of language acquisition by considering the acquisition of French word order by instructed English speaking learners. French and English differ in terms of surface word order with respect to negation, adverbs and object clitics. These differences are shown in the table below. Structure S-V-Neg-X S-aux-Neg-V-X S-Neg-V-X S-V-Adv-X S-Adv-V-X S-CI/Pro-V S-V-CljPro French elle ne regarde pas la tel€ *elle n'est pas regarder la tele *elle ne pas regarde la tele elle regarde souvent la tele *elle souvent regarde la tele elle la regarde *elle regarde la English *she watches not TV she is not watching TV *she not watches 'IV * she watches often TV she often watches TV *she it watches she watches it Table 1: Word order differences between French and English Pollock (1989) argues that these different word orders are due to one single parametric difference between the two languages - namely verb placement. Negation, adverbs and clitics are in fixed positions in the underlying structure. In French the verb undergoes movement whereas in English it does not. The difference between the two languages is argued to be the result of French having a strong uninterpretable Tense feature which requires verbs to move whereas English does not (Chomsky & Lasnik, 1995, Lasnik, 2007). The learnability issue for the English speaking L2 learner of French is acquiring this different Tense feature. In this thesis I will investigate the acquisition of these structures (negation, adverbs and object clitics) and will also consider the use of subject clitics to investigate potential parameter re-setting. This study seeks to empiricaliy test between three theories of the Initial State of L2 acquisition and three theories of L2 development. The Initial State theories tested are Minimal Trees/Organic Grammar (Vainikka & Young-Scholten, 1996, 2005), Full Transfer/Full Access (Schwartz & Sprouse, 1996) and Modulated Structure Building (Hawkins, 2001). These three theories all make different empirically testable predictions about the level of Ll transfer and the underlying structure of the Initial State. The theories of development tested are the Missing Surface Inflection Hypothesis (Prevost & White, 2000), Representational Deficit Hypothesis (Hawkins & Chan, 1997) and Feature Reassembly (Lardiere, 2008). Again these theories make different predictions concerning possible parameter re-setting in L2 French. This study is therefore framed by the following research questions: A. What is the initial state in L2 learners of French? B. How do functional features develop in these learners? C. What is the role of the Ll feature settings in this development? I examine data from five groups of 15 instructed English speaking learners of French who have all been taught in the British school and university system. The beginner group (aged 12-13) has received one year of instruction, the low-intermediate group (aged 15-16) have had four years, the high-intermediate (aged 17-18) have received 6 years of instruction. The low-advanced group (aged 19-20) are in their second year of an undergraduate French degree and the high-advanced group (aged 21-23) are in their final year of undergraduate study and spent at least 5 months in a French speaking country. Ten native speaker controls were also tested. Results from two elicited oral production tasks, a comprehension task and an acceptability judgement task are presented and the theories of the Initial State and development mentioned above are evaluated in light of these results. The results show significant levels of Ll transfer in the Initial State and a gradual development of sentence structure. I argue that these results provide evidence against Full Transfer/Full Access and Organic Grammar and in support of Modulated Structure Building. In terms of development there are significant correlations between the use of verb raising with negation, adverb placement, object clitics and subject clitics for both the oral production task and the judgement task. This would support the view that parameter re-setting is possible supporting Feature Reassembly and counter Representational Deficit Hypothesis. There is also partial support for the Missing Surface Inflection Hypothesis but further research is required. This thesis concludes that parameter re-setting is possible for instructed English speaking learners of French. However, learners build their syntactic representation gradually and transfer their knowledge of English at each stage before re-setting the parameter to the French values.

The research described in this thesis is concerned with some of the problems encountered in the processing of sociolinguistic data. Different methodologies are seen as different sets of strategies for coping with the problems which arise from investigations of sociolinguistic variability within any speech community. One early approach to the analysis of sociolinguistic variation (that of Labov: 1963, 1966) is discussed, and some of the difficulties raised by this approach are indicated. One investigation of sociolinguistic variability in a British urban setting (Trudgill: 1974) is also described (Trudgill' s study is based on Labov's (J 966) general methodology). FN The Tyneside Linguistic Survey (T.L.S.) is offered as an alternative approach, which overcomes some of the problems inherent in Labov's methods.

No description available.

The thesis reports on the linguistic accommodation of 39 university students from Moravia (the eastern half of the Czech Republic) living in Prague, Bohemia (the western part of the Czech Republic). In Bohemia, the informants' highly-localized native dialects and Standard Czech (SC) - a semi-artificial, archaic and primarily nonspoken standard with no native speakers - are both stigmatized, although for different reasons. Consequently, it has been 'hypothesized' that speakers of Moravian dialects living in Bohemia quickly reduce the frequency of or avoid stigmatized variants of their localized vernaculars and converge towards the host dialect, Common Czech (CC). Although a non-standard variety, CC is a semi-prestigious koine that is socially unrestricted throughout Bohemia and parts of western Moravia and is, according to some linguists, assuming the role of a national vernacular. However, the 'contact hypothesis' is based solely on introspective data and is ideologically driven, insomuch as it is the product of linguists who promote the social and geographical spread of CC. The present study is the first attempt to systematically describe the results of dialect contact between speakers of CC and Moravian dialects and to test the above hypothesis. To my knowledge, it is the first systematic variationist account of language variation in the Czech Republic. The study combines a quantitative analysis of six linguistic variables with both qualitative and ethnographic research and it identifies to what extent speakers of Moravian dialects living in Prague assimilate CC forms, what route their accommodation takes, and which variants of the host variety are most likely to acquired or rejected. A primary aim of the study is to describe the impact of a set of independent social variables on speakers' assimilation of CC forms. Special attention is accorded to speakers' sex, region of origin, length of residence in the host community and network integration.

Dialogue systems are artefacts that converse with human users in order to achieve some task. Each step of the dialogue requires understanding the user's input, deciding on what to reply, and generating an output utterance. Although there are many ways to express any given content, most dialogue systems do not take linguistic variation into account in both the understanding and generation phases, i.e. the user's linguistic style is typically ignored, and the style conveyed by the system is chosen once for all interactions at development time. We believe that modelling linguistic variation can greatly improve the interaction in dialogue systems, such as in intelligent tutoring systems, video games, or information retrieval systems, which all require specific linguistic styles. Previous work has shown that linguistic style affects many aspects of users' perceptions, even when the dialogue is task-oriented. Moreover, users attribute a consistent personality to machines, even when exposed to a limited set of cues, thus dialogue systems manifest personality whether designed into the system or not. Over the past few years, psychologists have identified the main dimensions of individual differences in human behaviour: the Big Five personality traits. We hypothesise that the Big Five provide a useful computational framework for modelling important aspects of linguistic variation. This thesis first explores the possibility of recognising the user's personality using data-driven models trained on essays and conversational data. We then test whether it is possible to generate language varying consistently along each personality dimension in the information presentation domain. We present PERSONAGE: a language generator modelling findings from psychological studies to project various personality traits. We use PERSONAGE to compare various generation paradigms: (1) rule-based generation, (2) overgenerate and select and (3) generation using parameter estimation models-a novel approach that learns to produce recognisable variation along meaningful stylistic dimensions without the computational cost incurred by overgeneration techniques. We also present the first human evaluation of a data-driven generation method that projects multiple stylistic dimensions simultaneously and on a continuous scale.

The complex multiracial/multilingual situation of Malaysia poses challenges for local professionals, such as speech and language therapists, who work with children. The present cross-sectional study investigated ethnic Chinese children's simultaneous phonological acquisition of English, Mandarin and Malay, which are the three major local languages for the Malaysian Chinese population. The aims were to provide preliminary normative data on phonological acquisition for this population, as well as to investigate processes underlying multilingual phonological acquisition. Sixty-four pre-school children aged between 2;06-4;05 were recruited. A single-word naming test, a word consistency production subtest and an intonation imitation sub-test were devised for each of the three languages. Particular attention was paid to the characteristics of the local adult speech varieties as the benchmark for assessing and analyzing the children's responses on the tests. This sociolinguistic dimension has often been neglected in previous research with similar populations, where non-local, e.g. "standard" adult varieties have been taken to be the language model for the children being studied. The children's phonological acquisition was analysed in term of consonants, vowels, syllable structures, word production consistency, intonation and tones (Mandarin only). Overall, significant developmental trends were evident for all three languages. Most phonological components under study were acquired by 4;00-4;05. Similar phonological milestones were achieved as those reported in the literature for monolingual and bilingual peers acquiring the same languages, though some qualitative and quantitative differences were observed. Overall, the patterns of phonological development that were identified reflect the interaction of common cross-linguistic tendencies with the specific characteristics of the three ambient languages. As well as having clinical implications, the present findings contribute to the development of theory and models for multilingual phonological acquisition. The reliability and validity of the test battery indicate that it will prove a valuable tool for speech and language therapy practice and for future research.

My study investigates aspects of the idiolect of Queen Elizabeth I using a sociolinguistic framework. My source material for Elizabeth's idiolect is an electronic corpus which I have compiled from transcripts of the best-authenticated examples of Elizabeth's own compositions in letters, speeches and translations. My investigation analyses nine morpho syntactic variables in the corpus and I chart their distribution and development diachronically. I also provide the first detailed analysis of Elizabeth's spelling, assessing developments in forms and the level of consistency in their use. For each linguistic feature, I compare her idiolect with macro-level linguistic data in order to contextualise her usage within previously established statistical norms of Early Modern English. I conduct a detailed analysis of social, stylistic, interactive and systemic factors to assess their potential influence on the frequency patterns in her idiolect. There are three key outcomes of my study. Firstly, I offer a new perspective on Elizabeth's writing by considering how the linguistic developments in her idiolect reflect and relate to her biographical experiences. I re-evaluate the emphasis historians have placed on her accession, and identify other biographical events that appear to have an impact on her language-use. I also consider how far Elizabeth's role in language change (where she is often a leader and frequently comparable in usage to her male contemporaries) affects current accounts of her socio-political role as a female monarch. Secondly, I evaluate the applicability of my idiolectal data to questions of authorship in the canon of Elizabeth's writings, considering the theoretical merits of morpho syntactic and spelling data before testing its application with four case studies. Thirdly, I reflect on the role of the idiolect in historical sociolinguistics, and demonstrate how my findings can test existing sociolinguistic accounts, and help to expand our understanding of the processes involved in language change.

No description available.