[Résumé non disponible]

No description available.

No description available.

Automated classification of texts into genres can benefit NLP applications, in that the structure, location and even interpretation of information within a text are dictated by its genre. Cross-lingual methods promise such benefits to languages which lack genre-annotated training data. While there has been work on genre classification for over two decades, none has considered cross-lingual methods before the start of this project. My research aims to fill this gap. It follows previous approaches to monolingual genre classification that exploit simple, low-level text features, many of which can be extracted in different languages and have similar functions. This contrasts with work on cross-lingual topic or sentiment classification of texts that typically use word frequencies as features. These have been shown to have limited use when it comes to genres. Many such methods also assume cross-lingual resources, such as machine translation, which limits the range of their application. A selection of these approaches are used as baselines in my experiments. I report the results of two semi-supervised methods for exploiting genre-labelled source language texts and unlabelled target language texts. The first is a relatively simple algorithm that bridges the language gap by exploiting cross-lingual features and then iteratively re-trains a classification model on previously predicted target texts. My results show that this approach works well where only few cross-lingual resources are available and texts are to be classified into broad genre categories. It is also shown that further improvements can be achieved through multi-lingual training or cross-lingual feature selection if genre-annotated texts are available in several source languages. The second is a variant of the label propagation algorithm. This graph-based classifier learns genre-specific feature set weights from both source and target language texts and uses them to adjust the propagation channels for each text. This allows further feature sets to be added as additional resources, such as Part of Speech taggers, become available. While the method performs well even with basic text features, it is shown to benefit from additional feature sets. Results also indicate that it handles fine-grained genre classes better than the iterative re-labelling method.

How do people listen to soundscapes in the built environment? Current soundscape research within urban planning disciplines tends to focus on measuring outdoor spaces in the built environment by interviewing the people within. This thesis, by contrast, followed individual listeners, using a qualitative, Grounded Theory methodology, examining listening preferences and habits across multiple environments. This approach gave a broad range of reactions to different soundscapes, from homes to workplaces to bars, clubs, and places of worship. This thesis reviews various soundscape epistemologies, methodologies, and methods, and argues that we need a stronger theoretical understanding of all these elements. It questions what is being measured, and how people are measuring it. The thesis suggests some ways qualitative and quantitative research can work together more effectively, and move soundscapes from the current multidiciplinary research landscape to a truly interdiciplinary one. In defining the soundscape as ‘the listener's perception of their auditory surroundings’, I shift the focus from measuring people's evaluation of spaces, to evaluating people themselves. This leads to a radically new empirical approach and theoretical description of the soundscape, using social science methods to build thick description of listening habits. Twenty people were given audio recorders and log books, and asked to record their day-to-day lives for two weeks. They were then interviewed about their experiences. The main finding was that soundscapes are not noticed most of the time, with participants seeming to have a ‘noticing threshold’: affected by factors such as control, expectation, and activity. Soundscapes which were noticed fell into one of four categories: positive--loud, positive--quiet, negative--loud or negative--quiet, with different judgement criteria for each. Participants were also highly adept at using coping mechanisms, such as recorded music and TV, to counteract undesirable sound environments.

The aim of this study is to examine nine texts used as teaching material in the reading project A Reading Class in terms of three different text-analytical structures. The texts analysed are non-fiction texts intended for grades 1–3. The analysis focuses on the similarities and differences in the textual, ideational and interpersonal structure of the texts and how the progression in the texts is revealed.   The result of the analysis shows that there are considerably more similarities than differences between the texts as regards the analysed structures. The analysis cannot find any clear progression between the texts in certain respects, although the analysed material is intended for pupils in different grades, while in other respects it is possible to see some notion of progression. The result also shows that the concept of progression is problematic to apply in a text analysis alone, since it is affected by how teachers work in practice with the material in the classroom.

This thesis investigates how various text-talks, i.e. text-focused classroom discussions, may scaffold students’ reading of specialised texts in upper secondary school. The study consists of qualitative case studies based on classroom observations of two teachers’ History instruction, focusing on parts defined as text-talks. An intervention study was conducted where one teacher worked with two text-talk approaches. The research questions regard how students move in relation to the text in the text-talks and how text content is incorporated, what scaffolding structures emerge, and whether and how the text-talks differ. A secondary aim is to generate theories concerning the potentials and limitations of text-talk as a reading scaffold. Analyses were done in terms of text movability to show reading positions, intertextual cohesion to show relations between source text and text-talk, and scaffolding which includes peer scaffolding, teacher scaffolding and the text-talks as a scaffold per se. A methodological contribution is the development of a model for content-based analyses of authentic text-talks. The results show that in text-talks that work as a scaffold, students take the expected positions toward the text, and the talks are clearly related to the source text, by means of lexical and conjunctive cohesion that is often varied and built-out. For more demanding texts, the students show dynamic text movability and move between exploring contents, subject field and context. Other characteristics are either peer scaffolding showing dialogicity and negotiation of meaning, or teacher scaffolding enabling students to progress and develop tools for text reception. The intervention approaches seem to scaffold reading to a greater extent than text-talks within ordinary instruction where the framing is weak. In conclusion, the results suggest that both student- and teacher-led text-talks may scaffold reading, but they need to be well planned and prepared with a structured framing.

published_or_final_version / abstract / Linguistics / Master / Master of Philosophy

The topic of this thesis is the process of'normalisation' by which a translator accommodatesa sourcet ext to the norms of the target languagea nd culture. The researchi nvestigatest his phenomenonin two ways: first, through text analysis identifying features in the target text which potentially contribute to easy readability, and, subsequentlyt,h rough readersr espondingt o the translation and the sourcet ext. Computer tools were used to compare the source text and the target text, resulting in a list of eleven identifiable main features which were found to contribute to a normalising effect. One of the features, 'patterns of repetition', was examined in detail. The translation had less repetition in part because of systemic differences but also due to the translator's choice, whether conscious or unconscious, to use variant terms. A section of the thesis investigates negativity which is of considerable literary relevance in the case of A Hora da Estrela, and evidence is supplied that negativity is reduced in the translation. Other featuresi n the sourcet ext which are mostly associatedw ith spoken aspects of language were found to have shifted to a written style, contributing to a normalising shift from spoken to written register. Such features suggest that the translation hasb een 'domesticatedt' o suit the English-speakingr eader. The second way in which normalisation was investigated was via readers' responses. A number of studies were carried out, using non-specialists and'semi-specialised readers' (students of literature). Critical reactions to the various translations of the novel suggested that normalisation was perceived when the source and the target were compared, but not by readers of the target version alone. iii

No description available.