Online social platforms (e.g. Facebook, Twitter) provide a built-in translation feature which gives users the option to have the content translated into their own language allowing them to interact with other users in other languages. However, content generated by users in these platforms is mostly written in colloquial ‘‘bad’’ language which could be described as ‘‘toxic’’ because it contains a substantial amount of profanity and obscenity. Properly handling said features of these informal texts is still beyond the capacity of the available MT systems. Therefore, this additional translation task poses a challenge to developers who find themselves in a situation where they feel their focus needs to be shifted from refining the current models in order to produce a better translation to training models that can handle these features. This project aims to address this issue by investigating what general users actually consider to be good quality translations of this type of texts by investigating what part of these texts is prioritized by the end user: receiving the message, but with bad language and associated sentiment unrendered; the correct translation of the bad language and sentiment, but at the expense of the rest of the message; or both are a must for complete understanding. In the event of the third option, the relevance of the elements of profanity, as well as their frequency, to the comprehension of the stance and emotion of the creator of the message has to be understood.
My PhD project involved interdisciplinary research across Translation Studies, Computer Science and Natural Language Processing with industrial impact in tool development and user-experience. It was focused on the study of Statistical Machine Translation, a branch of Natural Language Processing, using the predominant research-led open-source SMT toolkit, Moses, to show how linguists and translators without a background in SMT theory and application could develop sufficient knowledge to build their automated MT engines and use it in their translation work beyond what is typically taught in Translation Studies. I conducted a multi-method study, with qualitative data collection in different learning and teaching settings to test the first English into Arabic MT translation training open to Arabic-speaking master-level students regardless of where they had enrolled for their studies, in an attempt to ascertain in which way training on translation technology should (potentially must), change the translators’ role from mere receivers to actual creators (or at least adopters) of their MT engines. Drawing on what early studies achieved, the study emphasizes that further education-focused research should open new paths for a better and more viable human-machine integration.
In 2017, I joined the H2020-funded INTERACT Crisis Translation Network, led by Prof. Sharon O’Brien (Dublin City University). I was involved in Work Package 4 (Crisis Machine Translation), Work Package 2 (Crisis Translation Policy – Research), and in part to Work Package 5 (Citizen Translator Training). I have conduct research on crowdsourcing of translation and interpreting aid to people needing language mediation in crises: https://sites.google.com/view/crisistranslation/consortium/university-college-london
This project aimed to enhance pupils’ social and civic competencies. It was led by the British Council in collaboration with London School of Economics. I worked with Prof. Sandra McNally at LSE and my responsibilities included recruiting schools and communicating with those interested to involve them in the project’s activities
A project funded by the highly competitive UCL Grand Challenge of Cultural Understanding – Doctoral Students' Grants Scheme (2017-2018).
I developed and led, as the principal investigator, this cross-disciplinary research project in collaboration with a doctoral researcher at the time affiliated at UCL school of Medicine, funded by the highly competitive UCL Grand Challenges scheme. It investigated the linguistic and cultural communication skills of non-UK doctors for their successful transition into employment. The project had social and academic significance as it enabled, through tailored training, refugee doctors to practise in the UK and integrate into British society. The project’s activities:
Presentation: ''Voices of (Unemployed) Refugee Doctors in the UK: An Exploration of their Linguistic and Cultural Needs & Aspirations'', the V International Conference ‘Translating Voices, Translating Regions ‘Minority languages, risks, disasters and regional crises.
Two research papers presented at ''Migration and Language-Learning: Histories, Approaches, Policies'' workshop, Leeds and at the ''1st World Congress on migration, Ethnicity, Race and Health'', Edinburgh - 2018.
The project focused on multilingual communications in emergencies at the arrival of migrants in Sicilian ports and how people from diverse linguistic and cultural backgrounds communicate at such times. It involved collecting qualitative data by interviewing aid providers, mediators as well as asylum seekers in refugee camps.
I contributed to the Multilingual Healthcare Project, focusing on providing a multilingual aid to healthcare workers who are in contact with migrants, during my Erasmus+ internship at University of Ljubljana, under the supervision of Prof. Nike K. Pokorn. My contribution included adapting existing aids to online use - reviewing the Arabic version of their multilingual handbook for medical use to facilitate communication in healthcare settings, which was published in eight languages. I also participated in their training workshops aimed to train medical staff on how to handle patients from other cultures, Muslims in this case, for a better understanding and effective communication: http://multilingualhealth.ff.uni-lj.si/
Translation - 33,000 words - The Washington Group on Disabilities Statistics.
Translation – 4400 words – Global Health 5050 2019 Report.
Reviewing and proofreading Translation E-A – UCL Translation and Media Accessibility Services. The job is proofreading and quality checking of a website of e-learning courses for university students. Review: total words - 200,000. The work included: Cultural review of course in English to confirm that all the course is of relevance to the target market - Review glossary of common terms prior to full translation - Review of translated course - Re-review any corrections that were highlighted when reviewing website.
Kids in Museum – Manifesto.
TRANSLATION – 10,000 words – Financial Translation.
TRANSLATION A-E: 6000 words - Health and Safety at Work.
Translation A-E: 16000 words – Financial Translation.
Translation of a legal contract (4500 words).
Review of a Handbook for Medical Use.
Transcreation and Ads Editing.
Subtitling of videos.
Translation – 3500 words – Global Health 5050 Brochure and Report.