Methodology for computer-assisted acquisition of reading abilities in L3

The lack of foreign-language skills, particularly reading abilities, is handicapping UK research, especially with the enlargement of the EU. Pedagogic methodologies are falling behind: they address general conversational skills, and fail to capitalise on advances in the semi-automatic analysis of large corpora. My research focuses on the foreign-language reading competence needed by researchers and translators.

The project aims at developing an effective corpus-based methodology for acquiring reading abilities in a third language (L3, here Polish and Ukrainian) based on the knowledge of a second language (L2, here Russian) by English (L1) native speakers, incorporating recent achievements in field of data-driven learning. The project focuses on models of text comprehension based on cohesive cues and purposeful reading strategies.

Towards a Linguistically Informed Kansei Engineering

My research aims to contribute to the understanding of the networks of concepts underlying descriptions of products and product components and assess the ways in which people actually vocalise their evaluations. Kansei engineering is a research tool for measuring people's feelings towards products and this is one possible beneficiary of the research. It has been established for 20-30 years and is centred around the use of word lists that describe a product. Little research has been carried out on the linguistic validity of these lists and so this is partly the aim of my research.

A Corpus-based Comparative Analysis of Proper Names in English and their Translations into Spanish and Italian.

In my PhD I am exploring solutions for the problem of translating culture-bound references, as well as the usability of novel corpus tools for the study of translated literary texts.

Proper names have traditionally been considered as single elements, independently from the rest of the text. However, this study shows that in the context in which a proper name occurs it offers a set of information to the source language reader that goes beyond the text, reaching into the culture of origin and the inside knowledge of the reader. This research considers the role that external elements and the translator play when it comes to transmitting the connotations hidden behind a proper name into a new target language. The analysis is undertaken using tools and a methodology derived from Corpus Linguistics.
The final outcome will contribute to a better understanding of the processes which lead from source to target name helping to establish whether the semantic content of a name can be transferred to a different target reader and culture.

Localising Pack Messages

This work makes two types of original contribution.

Firstly, it provides an in-depth study of localization in an under-researched genre, pack messages. The challenges of localization are approached by developing an account of cross-locale variation. This account is based on the analysis of existing comparable texts from two locales, Taiwan and the UK, as well as consideration of the factors that constrain pack design and consultation with people involved in the design process. It is hoped that the findings based on this rather extreme case will have relevance also for people working with other languages and domains.

Secondly, building on previous empirical research on multimodal discourse analysis, it develops and refines existing theoretical approaches, extends these to cross-cultural contrastive analysis and implements a suite of software tools to support the capture, annotation and analysis of data from static artefacts. It is the first corpus-based study of its kind. Novel accounts of multimodal texture and genre are presented. The role of typography in expressing language in graphic mode is central to both of these. This suggests a need for a fuller treatment of what might be called contrastive typography.

Corpus-based machine translation evaluation via automated error detection in output texts

My PhD focuses on the exploration of new automated methods for evaluating MT output. These methods need to be validated by finding correlations between human evaluation scores and those from new automated methods. Consequently, my research involves three main phases:

• Design and compile a multilingual corpus specifically for MT evaluation, as resources are currently lacking.
• Conduct human evaluations of MT output using machine and human translations from the corpus;
• Investigate new automated evaluation methods and validate by comparing scores with human evaluation results.

PhD, awarded 06/2006
Base: CTS

Acquiring Reading Skills in a Foreign Language in a Multilingual, Corpus-Based Environment

Due to a severe shortage of online resources for improving the acquisition of reading skills in several foreign languages at the same time, I intend to combine a wide range of resources into a new and effective environment suitable for teaching, as well as individual study.

The Development of Coherence and Quality of Performance in Conference Interpreter Training

Peer-Feedback for Conference Interpreter Training aims to explore how an interpreter trainee becomes a reflective practitioner, to explore and report the value of peer feedback for conference interpreting trainees and to organise the practicing criteria systematically in a user-friendly environment to enhance the effectiveness and efficiency of peer-feedback for conference interpreter training.

Restructuring the Source Text Information for Translation Training

The main focus of my study is that of manipulating information in the process of translation. I am referring to legitimate cases of information deletion and restructuring that occur because of different target audiences and publishing media.

Information Extraction Technology in Machine Translation

The goal of my project is to investigate the use of IE annotation for improving and evaluating the quality of MT, for ranking information relevance, explicating facts, motivating translation transformations and revealing the properties of legitimate variation in translation.