RESEARCH
My Research
In this section I've included some of the projects I have worked on in the past. Most of the research dates back to my days in Dublin City University while studying for my BSc. in Applied Computational Linguistics + German.
CALL, Corpus Linguistics, Forensic linguistics, Machine Translation, Psycholinguistics, Speech Processing
CALL - studied under Dr. Christine Appel
Nowadays, Computer-Assisted Language Learning (CALL) is an extremely important aspect of language learning. Because of the widespread availability of the World Wide Web, more and more people are finding it easier to use it to its full potential. As a student of computational linguistics, I am able approach the area of CALL with both linguistic and computational knowledge, allowing me to not only understand how a person learns a language, but also giving me an opportunity to code the various ways and means by which the student can learn and improve upon that particular language.
Kevin Graham, Denis Spohr and I designed an online language learning environment for the teaching of German; Taking into account the three different phases in the recent development of CALL (Behaviorist, Communicative, and Integrative approaches), our project implements several aspects of each of these approaches. Full details of the project are available in word or in html format.
Corpus Linguistics - studied under Dr. Dorothy Kenny
For our second year at university we completed a module in Corpus Linguistics. Kevin O'Shea and I worked on a project entitled: "A study of Hiberno English", where we recorded several conversations between Irish natives, transcribed their speech and analysed their sentences with respect to so called standard English.
This study is a bit rough around the edges but it is quite interesting to see how the Irish language has affected the way the Irish people speak English. Full details of the project are available in word or in html format.
Forensic Linguistics
I am currently reading a book by John Olsson entitled: "Forensic Linguistics: An Introduction to Language, Crime and Law". It's basically an introduction to the subject, with case studies such as the Harold Shipman case and the Salem Witch trials being analysed among others.
Most of us have probably already read "The Da Vinci Code" by Dan Brown, but were probably not aware that there is a massive court case going on between himself and Lewis Perdue over the alledged infringement of Perdue's 1983 novel "The Da Vinci Legacy" and 2000 novel "Daughter of God". Well, I'm not sure as I haven't had the time to read Perdue's two novels yet, but you can see some get an idea of why Perdue is so upset from the following two links:
http://www.thetext.co.uk/davinci_infringement/index.html
http://www.davincilegacy.com/Infringement/Justice.html
Machine Translation - studied under Dr. Andy Way
As an assignment for the 4th year module 'Machine Translation 1' in DCU we had to choose a MT system and try to figure out what was going on internally by means of comparing input and output in an iterative manner. I chose Softissimo's Reverso as my system for analysis. Details of the investigation can be found here.
Without placing any restrictions on the type of text used as input for an MT system, the chances of achieving high quality MT are significantly reduced. By tailoring an MT system to a domain specific language, it is possible to use the characteristics of this 'sublanguage' to our advantage. Another technique used is to simplify the source language, thus making the translation task easier for the system. You can view my essay on in "Why it is better to use an MT system for (a) a restricted sublanguage and (b) a controlled language".
Psycholinguistics
I took a course in Psycholinguistics when I was on my year abroad in Universität Stuttgart. Every week a set of students would have to present a certain topic related to Psycholinguitics. I, along with Barbara De Bernardi, Elena Galli, and Kevin Graham chose "Language and the Brain" as our presntation subject. To accompany the presentation we gave a handout to each student in the class.
Speech Processing
Here's the abstract from my 4th year Thesis, based around Child-directed Speech Synthesis.
It has been noticed that there are extreme differences in the speech we use towards young children compared with our eveeryday speech. Higher pitch and longer duration of phone segments are two salient prosodic features present in such child-directed speech (CDS).
CDSMorph is a tool that changes speech to be child friendly. Based on research in the area of prosodic features present in child-directed speech (CDS), morphing functions are calculated, which, when given an utterance, apply changes to the pitch and duration of the source to make it sound as if it were directed to a young child. Such a tool could be invaluable in the synthesis of children's stories. At the moment the system has been configured for an extract from the children's novel: Irish legends: St. Patrick (Tarantola Books, 1997) . It can however be configured to your needs by acquiring your own training data.
Click here to view the full details of the thesis. |