No description available.

This thesis investigates the problem of unsupervised learning of natural language grammar in the context-free grammar formalism, and argues that linguistic notions are beneficial to the task. Like some recent approaches, this thesis employs distributional clustering, which is based on the linguistic notion of distribution. Although grammar induction is conceptually complicated as it involves both the demarcation between constituents and non-constituents and that between different types of constituents, it is shown in the thesis that these two tasks are actually the two sides of the same coin. That is, nonconstituents can also be classified into different clusters and these clusters are very easy to be separated from those of constituents, and therefore the real problem in grammar induction is how to identify constituents. This thesis provides a generic framework of distributional grammar induction for experimenting with the effect of different criteria for selecting clusters of constituents. Experiments show that a criterion based on the simple principle of minimum variance fails to learn plausible grammars from vast amount of complex data, and it also leads to inconsistency in syntactic analysis as well as flat parse trees. Another criterion is proposed on the basis of the fragment test, one of the constituency tests proposed in distributional linguistics. This criterion, augmented by a novel grammar rule rewriting mechanism, is shown to be successful in guarding against many frequently-occurred non-constituents, in learning very many types of constituents, and in removing redundancy in grammar and giving rise to highly hierarchical syntactic structure.

Language appears to be special. Well-rehearsed arguments that appeal to aspects of language acquisition, psycholinguistic processing and linguistic universals all suggest that language has certain properties that distinguish it from other domain general capacities. The most widely discussed theory of an innate, modular, domain specific language faculty is Chomskyan generative grammar (CGG) in its various guises. However, an examination of the history and development of CGG reveals a constant tension in the relationship of syntax, phonology and semantics that has endured up to, and fatally undermines, the latest manifestation of the theory: the Minimalist Program. Evidence from language evolution can be deployed to arrive at a more coherent understanding of the nature of the human faculty for language. I suggest that all current theories can be classed on the basis of two binary distinctions: firstly, that between nativist and non-nativist accounts, and secondly between hypotheses that rely on a sudden explanation for the origins of language and those that rely on a gradual, incremental picture. All four consequent possibilities have serious flaws. By scrutinising the extant cross-disciplinary data on the evolution of hominins it becomes clear that there were two significant periods of rapid evolutionary change, corresponding to stages of punctuated equilibrium. The first of these occurred approximately two million years ago with the speciation event of Homo, saw a doubling in the size, alongside some reorganisation, of hominin brains, and resulted in the first irrefutable evidence of cognitive behaviour that distinguishes the species from that of our last common ancestor with chimpanzees. The second period began seven to eight hundred thousand years ago, again involving reorganisation and growth of the brain with associated behavioural innovations, and gave rise to modern humans by at least two hundred thousand years ago. ii I suggest that as a consequence of the first of these evolutionary breakthroughs, the species Homo erectus was endowed with a proto-‘language of thought’ (LoT), a development of the cognitive capacity evident in modern chimpanzees, accompanied by a gestural, and then vocal, symbolic protolanguage. The second breakthrough constituted a great leap involving the emergence of advanced theory of mind and a fully recursive, creative LoT. I propose that the theory outlined in the Representational Hypothesis (RH) clarifies an understanding of the nature of language as having evolved to represent externally this wholly internal, universal LoT, and it is the latter which is the sole locus of syntax and semantics. By clearly distinguishing between a phonological system for semiotic representation, and that which it represents, a syntactico-semantic LoT, the RH offers a fully logical and consistent understanding of the human faculty for language. Language may have the appearance of domain specific properties, but this is entirely derived from both the nature of that which it represents, and the natural constraints of symbolic representation.

Cognitive poetics has largely drawn so far on psychological models, and only recently have researchers turned their attention to cognitive linguistics. Considering the insights drawn from systemic-functional models over the past few decades and the revolutionary analyses that cognitive linguistics has brought to the fore, several of the difficulties that arise in the stylistic application of Hallidayan, cognitive linguistic and narratological frameworks seem to be resolvable from the perspective of Langacker's Cognitive Grammar (Langacker 1993a, 1993b, 1999a, 1999b, 2001, 2007, 2008, 2009; see also Croft and Cruse 2004; Evans and Green 2006; Taylor 2002). Specifically, Cognitive Grammar offers a means of accounting for experiential organisation, the attenuation of experience and how it is simulated, in a literary reading. In this study I propose an extension of Cognitive Grammar towards a cognitive discourse grammar through the unique environment that literary stylistic application offers. Thus, this research aims to develop, almost from an incipient point, the application of Cognitive Grammar in literary analysis. Through the application of Cognitive Grammar concepts, a verifiable framework and methodological principles will be established, drawing upon points of contact with cognitive-narratological developments as well as upon elements of cognitive linguistics in the process. Such points of contact include but are not limited to similarities with text world theory, cognitive semantics, cognitive narratology, systemic-functional grammar, mindmodelling and structuralist approaches to narrative layering. Rather than suggesting CG as a replacement for these existing frameworks, this analysis instead involves providing a cognitive or CG-extension to their approach. As a data set for this application, I use examples of contemporary fiction, which both exemplify and challenge the central CG concepts.

This thesis is concerned with text typology. In this thesis, the written part of the British National Corpus World Edition, which is a one hundred million-word collection of British Enghsh, will be used to study the characterisation of text types by identifying their linguistic characteristics. By means of multivariate analysis, the variation of the iccurrence of selected linguistic features among genre categories or registers will be classified. A multivariate analysis of this sort holds out the promise of being able to systematize the genre categories in the corpus while also revealing the characteristic linguistic features of the groups classified. The criteria for classification will be based exclusively on the dimensions revealed as significant by the multivariate analysis. The groupings will then be interpreted linguistically. This thesis also deals with the study of sociolinguistic variables which are fairly systematically associated with region, class, ethnic group, age, and gender.

No description available.

This thesis aims to explore the interface between scientific and technical translation (STT) and cognitive linguistics (CL), placing particular emphasis on the translationally relevant phenomena of explicitation and implicitation. The two concepts are regarded as potential indicators of translational text-context interaction, which may be of specific importance in the knowledge-intense field of STT and which can be modelled within the CL framework. Parallel to the microscopic attempt to give a coherent account of explicitation and implicitation in STT from a CL perspective, the thesis follows a macroscopic approach that aims to highlight the wider potential which cognitive linguistics holds for the field of scientific and technical translation. Translationally relevant elements of the CL framework include a coherent and cognitively plausible epistemological basis that explains the stability of scientific knowledge, the concept of common ground, which can be used to model the shared knowledge of specialized discourse communities, the field of cognitive semantics, which has developed tools for modelling the organization and representation of specialized knowledge, and the concept of linguistic construal, which allows the description of various linguistic aspects of STT (explicitation and implicitation among them) from a cognitively plausible perspective. The first part of the thesis takes a macroscopic perspective, being concerned with scientific and technical translation, cognitive linguistics, the philosophical grounding of the two fields and their interface. The perspective is then narrowed down to the two specific phenomena of explicitation and implicitation, which are reconceptualized in cognitive linguistic terms so as to fit into the overall framework of the thesis. The interface between STT and CL is then illustrated in a qualitative corpus-based investigation of explicitation and implicitation as indicators of text-context interaction in translation. The qualitative discussion of the results of the corpus analysis then brings together the theoretical strands of the thesis.

No description available.