Translating
and the Computer 29 Conference
Day
One: 29 November 2007
Go to Day Two
08.30
Registration
09.15
Introduction by Chair: Olaf-Michael Stefanov, United Nations, Vienna
09.20 Translation Wikified: How will Massive Online Collaboration Impact the World of Translation?
Alain Désilets, National Research Council of Canada
Massively collaborative sites like wikipedia, YouTube and SecondLife are revolutionizing the way in which content is produced and consumed world wide. They are part of a broader wave of open collaboration that includes things like open-source software, blogs, Creative Commons licensing and free culture. These fundamentally collaborative technologies and paradigms are bound to have a profound impact on the way that content is not only produced, but also translated. In this talk, we will sample a number of translation related questions that naturally arise in this new frontier. For example, what does it mean to translate content that is constantly being edited collaboratively by a large community of anonymous authors? How might translators benefit from open, wiki-like translation resources? In a world where anyone can write and publish original content in their native language, will we need to cover more language pairs, and if so, how might Machine Translation technology help? Could massively collaborative technologies save freelance translators from extinction by allowing them to bid on parts of large contracts without giving up their autonomy and by allowing them to share expertise within the context of a world-wide community of practice? Will we see the emergence of a new breed of "amateur" volunteer translators and will this result in a de-skilling of the translation profession? How can organisations best leverage the collaborative energy of this new breed of translators? How do we ensure the quality of translations and translation resources in such a seemingly chaotic collaborative environment? Can massively collaborative technologies help save minority languages from extinction? Can teachers of translation take advantage of such technologies to provide students with real-life translation experience early on in their training?
We will illustrate questions like those with real-life examples of projects and trends that are currently happening. Using those examples, we will show how mass collaboration technology sometimes introduces new problems, while in other cases it helps solve them or even creates exciting new opportunities and niches that did not exist before.
10.00
Making a Sow's Ear out of a Silk Purse:
(Mis)Using Online MT Services as Bilingual
Dictionaries
Federico Gaspari, University of Manchester,
UK
This
paper reports (part of) a survey carried
out among 104 university students in the
United Kingdom investigating their usage
of free online machine translation (MT)
services. The study focuses in particular
on the widespread use of these MT tools
for a purpose that they were not designed
for, i.e. the translation of single lexical
items, as if they were bilingual dictionaries.
The 104 (from an original survey of 280)
had used web-based MT services in the past,
and 65 of these (62.5%) reported using them
for single-word lookup. This finding suggests
that designers and developers of online
MT services should seriously consider taking
a proactive approach towards raising the
awareness of users with regard to the most
(in)appropriate ways of using web-based
MT software. The paper argues that it would
be in the interests of those who have a
stake in offering and promoting MT in the
online environment (e.g. system designers,
developers, and ultimately the MT vendors
themselves) to manage the expectations of
naïve users.
10.40
Discussion
10.50
Coffee
11.20
Removing the Distinction Between a Translation
Memory, a Dictionary and a Parallel Corpus
Vincent Vandeghinste, KULeuven, Belgium
This paper presents a prototype MT system
which does not make the distinction between
a dictionary, a sub-sentential aligned parallel
corpus, and post-edited information (translators
output) like a translation memory. The system
is based on the METIS-approach (Vandeghinste
et al, 2006), and uses an XML-based dictionary
format in which not only simple word-to-word
translations can be included, but which
also contains complex dictionary entries,
including discontinuous entries, like idioms
and proverbs. The presented prototype is
a system that automatically adapts its dictionary
and target language corpus depending on
the post-edited output as made by the users
of the system, and will therefore have a
learning curve in its performance.
12.00
Discussion
12.05
A Dynamic Dictionary for Discovering Indirect
Translation Equivalents
Professor Anthony Hartley, University
of Leeds, UK
This paper
presents a free on-line service to help
translators find solutions to the translation
of difficult expressions from the general
lexicon. It is often the case that novel
non-technical expressions prove more problematic
than novel terms. The most original feature
of the tool is that it uses large monolingual
corpora in the source and target languages
-- rather than parallel, translated corpora
-- to propose and rank non-literal translation
solutions. In this way, the method extends
the range of existing, static bilingual
dictionaries in dealing with an open set
of multiword expressions. To do this, it
exploits automatically created thesauri
and tools for building concordances. We
present an evaluation of the coverage of
our tool and its usability by translators.
Currently the system works between English
and Russian, but it can be quickly extended
to other language pairs for which machine-readable
dictionaries are available. The system is
the result of the 2-year research project
ASSIST funded by EPSRC. The interface to
the system is available at: http://corpus.leeds.ac.uk/assist/v05/.
12.45 Discussion
12.50
Lunch and Exhibition (please note: coffee
will be served in the exhibition area)
14.10 Introduction by Chair: Daniel Grasmick,
SAP AG, Germany
14.10
A Free Terminology Extraction Suite
Antoni Oliver and Mercè Vàzquez,
Universitat Oberta de Catalunya, Spain
In this paper we will present a set of terminology
extraction tools that are distributed under
a Free Software License, so that users can
freely download, use, distribute and modify
them to meet their needs. The tools are
mainly programmed in Perl and they will
work under different platforms, such as
Windows or Linux. These terminology extraction
tools will help freelance translators, translation
agencies and companies to find the best
translation of a term or to build monolingual
or multilingual terminological glossaries.
Moreover, translators, correctors and terminologists
can use The Free Terminology Extraction
Suite to create a terminological database
for a specialist domain so as to automatically
obtain a list of domain-specific lexical
units (potential terms) with their equivalent
translations from bilingual corpora of domain-specific
documents.
14.50
Discussion
14.55
In Other Words: Using Paraphrases in
Translation
Pernilla Danielsson, University of Birmingham,
UK
The presentation involves three parts; Firstly,
a corpus consisting of multiple English
translations of the same source texts is
presented as the language resource used
in this study (The Transfer Corpus). Secondly,
a study focusing on the identification of
paraphrases in this corpus is presented
and the findings categorised, and thirdly,
the findings are included in a computational
tool to be used as translation aid. This
paper suggests that a useful starting-point
for the study of paraphrase is a set of
texts that can be identified as paraphrases
of each other on external rather than internal
criteria. Here, texts that have more than
one translation into English are compiled
into a corpus, the Transfer Corpus. In the
extension to this project, we envisage a
tool that will offer alternative words,
phrases or even grammar to that of the translator's
first choice.
15.35
Discussion
15.40
Break
16.10
Lost in Specialised Translation: an Inexpensive
and Under-Exploited Aid for Language Service
Providers
Gloria Corpas Pastor, University of Malaga,
Spain
In the paper and in the presentation Gloria
will discuss a translation assignment of
a specialised nature. After performing source
text analysis, she will demonstrate how
a step-by-step approach to building an adequate
and/or representative corpus from resources
in the Internet works in practice. Corpus
design criteria and qualitative issues will
be taken into account. Real examples will
be presented that show how to mine a corpus
and how to use it in order to meet translators'
needs as far as terminology, documentation,
target text conventions and other constraints
are concerned. She will illustrate how language
service providers can use freeware concordances
(Antconc, ParaConc, etc.) to search for
translation equivalents in comparable and
parallel corpora. This demonstration will
also show the clear benefits of using such
corpora over any type of dictionary as they
provide examples of how words or expressions
are used and translated in context. All
in all, she will demonstrate how corpora
(a) provide instant access to real usage,
(b) depict syntagmatic patterns and translation
equivalents unavailable in existing lexicographic
resources, and (c) facilitate guidance to
style and text conventions in both SL and
TL.
16.50
Discussion
16.55
End of Day One
Day Two: 30 November
2007
08.45 Coffee and Registration (for Day Two
delegates only)09.15
Introduction by Chair: Chris Pyne, SAP AG,
Germany
09.20
Medical Spoken Language Translation
Technology: What Do the Users Really Need?
Professor Harold Somers, University of
Manchester, UK
This paper focuses on the particular use
of emerging spoken language translation
(SLT) technology in the medical domain,
in particular to assist communication between
patients with limited English and healthcare
providers. The paper will first consider
the "pathway to healthcare" for
such patients noting how it extends beyond
the focal point of a doctor-patient dialogue
with a GP as it is conventionally portrayed,
to include interaction at various stages
with a range of medical specialists and
non-specialists. The translation needs (both
spoken and text) vary accordingly. The paper
will then critically review work done so
far on SLT in the medical domain, in particular
for its bias towards the doctor as the principle
user of proposed software, its often condescending
view of the patient's role, and the fact
that it is led by the technology (what can
we do?) rather than the users' needs (what
should we do?). Finally the paper will propose
a range of language technologies suitable
for different points along the pathway to
healthcare, and will discuss some of the
barriers to implementation, especially considering
the fact that in many cases the languages
spoken by the kinds of patients we are targeting
happen also to be languages of least interest
to commercial developers of language technologies.
10.00
Translation Quality Assurance Tools:
Current State and Future Approaches
Julia Makoushina, Palex Languages and
Software, Russia
The purpose of this paper is to review translation
quality assurance tools, to define where
they currently are, their advantages and
disadvantages and to visualise their future
capabilities and role in translation process.
The review is done from the viewpoint of
a practitioner with a software development
background and is highly focused on what
real features need to be implemented in
such tools in the near future. In order
to make the review more valuable we carried
out a survey among translation professionals
to identify the most popular QA tools, the
overall acceptance and common usage of such
tools, and their desired features and capabilities.
During this presentation we will briefly
examine how translation quality assurance
tools developed, consider what they have
in common and point out their unique features,
benchmark their performance and discuss
how the translation community accepts and
uses them, what translators expect of them
and what kind of future awaits them.
10.40
Discussion
10.50
Coffee
11.20
Rapid Development of RBMT Systems for
Related Languages
Jernej Vicic, University of Primorska,
Slovenia
This paper describes novel way of constructing
rule-based machine translation systems (RBMT).
RBMT systems are currently among the best
performing machine translation systems.
Most of the "big named" machine
translation systems (Systran, 2007)(Promt,
2007) belong to this category, but these
systems have a big drawback; construction
of such systems demands a great amount of
time and resources, and thus finance. The paper describes methods that
automate parts of the construction process.
The methods were evaluated in a case study:
construction of a fully functional machine
translation system of the closely related language
pair Slovene - Serb. Evaluation was conducted
on the functional machine translation system
and the results, presenting coverage using
referential corpus and selected evaluation
metrics, are shown. Objective and subjective
evaluation methods were used as only a correct
mixture of methods minimizes evaluation
bias. Translation quality evaluation was
conducted using subjective evaluation methods
where a set of native speakers scored translations.
The conclusions present strong and weak points
of this approach and explore grounds for
further work.
12.00
Discussion
12.05 Building a bilingual dictionary from movie subtitles based on inter-lingual triggers
Kamel Smaili, David
Langlois, Loria, France
Training statistical translation models requires a huge quantity of bilingual aligned corpora. Commonly, Statistical Machine Translation Systems use Canadian Hansard Corpus or corpora extracted from the proceedings of European Parliament. Then given this parallel corpora, translation probabilities are learned using Giza++ which is an implementation of the IBM's translation models. In this paper, a method is presented, based on DTW (Dynamic Time Warping), which automatically aligns subtitles. With this method, 40 parallel corpora of French and English subtitles are generated. Using this corpus to test this approach, based on inter-lingual triggers for learning translation probabilities, it was possible to automatically build a dictionary. Experiments showed that the resultant dictionary is well constructed and is suitable for machine translation. With the purpose of evaluating the actual contribution of this method, a real translation system was developed using the Pharaoh decoder. Evaluation has been achieved by using Bleu score.
12.45
Discussion
12.50
Lunch (please note: coffee will be served
in the exhibition area)
An opportunity to visit the exhibition stands
14.10
Introduction by Chair: Professor Ruslan
Mitkov, University of Wolverhampton, UK
14.10
E-learning and Employability in Translator
Training: Introducing
E-Portfolio and Personal Development Planning
at the University of Vienna
Notburga Rotheneder, University of Vienna,
Austria
The aim of this project is to introduce
the e-portfolio method to promote skills
development reflection and personal development
planning in the specific organisational
and didactic context of the Centre for Translation
Studies.
E-learning is regarded as a means to promote
skills that are becoming increasingly relevant
for translators, namely translation technology
and transferable skills. The e-portfolio
is an excellent tool for training reflection,
both about what relevant skills are and
how they are acquired, and also serves to
make skills profiles and evidence of achievements
visible to potential employers. These are
appropriate measures to improve the employability
of translation graduates. The focus is on
technology skills - such as translation
memory usage and management, software and
website localisation, terminology extraction
and management, corpus creation and exploitation,
on the one hand and on transferable skills
- such as organisational, communication
and team skills (project management) and
quality assurance - on the other hand.
14.40
Innovation and E-learning in Translator
Education: A Singapore Experience
Susan Xu, SIM University, Singapore
The last decades have seen the education
of translators shift to several new dimensions,
thanks to rapidly-developing technology
and globalization. There seems to be an
increasing perception that conventional
classroom teaching alone will no longer
equip translators-in-training with the wide
range of professional skills, knowledge
and competencies that are required by today's
fast evolving language service market. This
paper presents an overview of the innovative
development in devising integrated and detailed
e-Learning plans and activities to enhance
the design and delivery of the course content
in the BA in Translation and Interpretation
(BATI) programme. It examines how the blended
approach in intergrating e-learning environment
with on-site classroom changes the way translators
and interpreters are being trained and resolves
the problems in the training. Implication
and issues encountered in the process of
implementing e-learning
plans such as lack of technical experiences,
challenge of incorporating pedagogy into
the e-learning object, workload for teachers
and students, and conceptual restriction
of translation will be discussed. It is
concluded that the solutions to these problems
lie in developing greater control over technologies
and this poses challenges and opportunities
to software developers in the relevant field.
15.10
Panel Discussion on the future of Training
Translators
15.30
Tea
15.50
The Long Way to a Corporate Platform
for Multilingual Contents and Translation
Processes: Corporate Translation Management
in practice
Michael Leifeld, Polysius AG and Christian
Weih, across Systems GmbH, Germany
As Polysius AG, one of the world's leading
engineering companies in the field of equipment
for the cement and minerals industry, were
about to replace their existing translation
memory, it soon became clear to them that
they were renewing much more than just a
software application. The decision to implement
the concept of Corporate Translation Management
was made with a view to increasing efficiency
on various levels and this is now being
carried out step by step. Because of the
company's global activities, emails, reports
and above all plant-related information
such as manuals, clerical, technical and
legal texts are translated into a current
total of 10 languages. The majority of the
workload of approx. 150 translation jobs
per month, i.e. around 80%, need to be translated
into English and Spanish and are centrally
processed by the company's documentation
and translation department. With the new
system the documentation and translation
department now has a tool for effectively
managing projects that continually supports
individually defined workflows. The translation
memory and the terminology system form the
basis for both effective translation support
and the management of standardized corporate
terminology. The terminology system is available
by browser via the company's Intranet to
ensure simple access to both the company's
extensive terminology base and to the concordance
search for all staff members who need these
features. This paper will describe the decision
process the company undertook to develop
the new system.
16.25 Discussion
16.30
LTC Presentation on Workflow and Business
Information Management in the Language Industry
Adriane Rinsche, LTC, UK
LTC has almost a decade of experience in
helping corporate clients and language companies
implement business information and workflow
control tools. In this presentation we will
share some of our insights by explaining
various workflows and their objectives.
Attendees will be able to improve awareness
of how workflows are used and how they differ
depending on the desired outcome. They will
be given tips on how to describe and review
their own processes and workflows.
17.05
Discussion
17.10
Close of Conference
Please
note: Aslib reserves the right to make changes
to the publicised programme without prior
notice.
