Showing posts with label MT. Show all posts
Showing posts with label MT. Show all posts

Sunday, 29 January 2017

Poems about Translation 27: Robo-Burns; The Orfeon Translates (from Bill Herbert)



This poem for Burns Night may be slightly cheating because Burns Night is just past,  but it cheered me up so much that I wanted to share it anyway. It's not strictly speaking a poem about translation, but the whole post can be considered a poetic reflection on translation, if I squint...

The three texts represented are the Scots poem by Bill Herbert 'Rabbie, Rabbie, Burning Bright'; an translation into Bulgarian by Kristin Dimitrova, and a back-translation from Bulgarian into English by Facebook Autotranslate.  They are in reverse order, so you might like to start by scrolling up from the bottom of the page.

P.S. For a previous Burns Night post in this series, see no.21, 'Robert Burns on the Translator as Assassin'.

Monday, 23 July 2012

Job: post-editor for technical documentation, JP>EN

Forwarded from Stephen C.:

*Post-Editor for Technical Documentation (JA>EN)*

Milengo Ltd. is looking for (engineering) students or graduates who are interested in freelance work as post-editors (Japanese into English). Post-editing is the process of modifying a machine-generated
translation with refined translation memories to insure a certain level of quality. Requirements: good computer skills, excellent technical understanding, bilingual or native speaker in English or Japanese;
excellent command of grammar, attention to detail, clear writing style and positive attitude to Machine Translation. If you meet these criteria and are interested in working with us, please contact us at: translators at milengo.com.

Feel free to circulate.

Thursday, 8 September 2011

Poems about translation 10: Miroslav Holub, 'Wings'


This isn't really a poem about translation but it's a poem which resonates strongly for me in relation to translation. Miroslav Holub's 'Wings' was translated from Czech by George Theiner (also available here):
We have   
a map of the universe
for microbes,
we have
a map of a microbe
for the universe.

we have
a Grand Master of chess
made of electronic circuits.

But above all
we have
the ability
to sort peas,
to cup water in our hands,
to seek
the right screw
under the sofa
for hours

This   
gives us
wings.
For me there is something there that's very close to the distinction between a piece of machine translation software and a human translator. We need the electronic circuits (and we have to learn to get along with them) but there are also screws to be sought under the sofa, and peas to be sorted (funny how in Holub's poem these activities sound much more artistic!) and we need human translators to sought them, sort them, etc..

I came across the poem in a 1967 essay on Holub by Al Alvarez. Another remark by Holub quoted in the same essay really appealed to me:
There is no deep difference between the scientific mind and the artistic mind: both include the maximal creativity with the maximal freedom. Science is both theoretic and experimental. Art is only experimental.
If for 'art' we read 'translation', this is a beautifully succinct expression of the truth that no matter how much theory, what critical frameworks, what models one applies to the production of a translation, in the end the only question worth asking is 'does the translation work'? It's only through experimentation, the trying-on and jettisoning of possible solutions, that a translation can achieve excellence.

Monday, 29 August 2011

post-doctoral applied research position in Machine Translation

There's a two-year post-doctoral position in applied MT research available at Dublin City University:

The post-holder will be responsible for the specification, design, configuration and evaluation of system components of DCU’s MATREX machine translation framework for use in pilot and contract research engagements with commercial collaborators. This work will include development as necessary of additional tools in the MATREX framework for pre-processing and post-processing of collaborators’ data in the translation process.
He / she will work closely within a team of researchers and software developers and with other partners at various academic and commercial locations. The post-holder will also have the benefit of interaction with the wider CNGL research team, including the significant body of researchers working on Language Technology and Machine Translation at CNGL.
The deadline for applications is 9 September 2011. Further details and contacts here.

Saturday, 23 July 2011

funded PhD studentship in translation technology, Dublin City University (28 July)

PhD Scholarship in Cognitive Translation Studies and Human-­Computer Interaction at Dublin City University
Funded by Symantec Ireland

Background
Recent work by Koehn (2010) has focused on the concept of monolingual post-­editing of Machine Translation output. Similar work is being conducted within the NSF-­ and Google-­funded Monotrans, Monotrans2 (Hu et al., 2011) and Paratrans projects in the US, and within the Language Grid project in Japan (Lin et al., 2010). While these projects have shown that in certain cases monolingual experts can produce post-­edited products that outperform those generated by bilingual translators, they have not explored the cognitive dimensions associated with this new type of activity. Indeed, the following questions have yet to be fully answered:
  • Are there any pre-requisites for monolingual editors (and indeed bilingual post-­editors) to produce reliable output (background, domain knowledge, task tolerance, personality profile, interest in supporting technology, philanthropy)?
  • Are there any criteria (such as source or raw output quality) that may affect the quality of monolingual (and bilingual) editing?
The PhD scholar will be expected to carry out empirical research on cognitive and contextual dimensions to ascertain what the success factors are for monolingual vs. bi-­lingual post-­editing.

Scholarship Details

Applications are invited for a three-­year scholarship to conduct research in the domain broadly outlined above. The scholarship will include annual fees for three years (approx. 5,000 euros for EU candidates) and an annual stipend of 18,000 euros. The successful applicant will register as a full-time PhD-track student in the School of Applied Language and Intercultural Studies, Dublin City University, under the academic supervision of Dr. Sharon O'Brien. The scholarship is fully funded by Symantec Ireland's Shared Engineering Services Research Group. Industrial cosupervision will be provided by Dr Fred Hollowood and Dr Johann Roturier (both of Symantec Ireland).

Expectations

This study is one part of a larger funded project, with multiple partners. The student will be involved in:

• Defining several rounds of user studies involving the post-editing of Machine-­Translated User-­Generated content by both speakers of the target language and bilingual speakers. The objectives of these studies will be to compare the user productivity and quality of the generated output given different set-ups, and to compare monolingual and bilingual post-editing as a process.
• Defining user requirements to help design an enhanced Post-­Editing environment (where monolingual and bilingual user edits can be collected).
• Conducting additional tasks in the area of Corpus Linguistics, such as:

o User Generated Content (UGC) corpus analysis to identify areas where input normalisation may be required (in terms of spelling, terminology, grammar, style)
o Defining a model to classify UCG (by identifying specific utterances such as questions and answers).

• As a stipended PhD student, the student may also be asked to assist with teaching duties at some stage during the scholarship duration.
• The student would be expected to publish his or her research findings in relevant conferences and journals. Where appropriate, this will include co-publication with supervisors.
• The student will have the opportunity to take relevant modules designed for graduate training at the university.
• The student will be an active PhD member of the Centre for Translation and Textual Studies at DCU.

Qualifications, Skills and Qualities Required

• Ideally a completed Master's qualification in Translation Studies, with a minimum of a 2.1 award or higher. A Master's in Human-Computer Interaction would also be considered.
• Excellent English skills, academic writing skills and research skills.
• One or more additional languages (especially French, German and/or Japanese, but other languages will be considered).
• Good knowledge of linguistics.
• Some knowledge of statistics is desirable.
• Good IT skills are essential.
• Experience in a translation production environment is a bonus.
• The candidate must be comfortable having one foot in an industrial environment and one foot equally in an academic environment. S/he will be required to spend a signficant amount of time at Symantec’s Dublin office on a weekly basis.
• The candidate must be able to juggle short-term, industry-set deliverables with longer-term academic milestones.
• The candidate should be comfortable taking feedback from multiple channels and working within a team environment.

How to Apply

• The strict deadline for receipt of applications is Thursday, 28th July.
• Only electronic applications will be accepted. Applications should be sent to the following e-mail address: DCUSYM at gmail.com
• The following should be compressed (zipped) into one file and attached to the application e-mail:

o A CV (about 3 pages)
o A current photograph
o A cover letter stating why you are interested and how your qualifications etc. match the requirements set out above. This should also contain the contact details of one academic who has agreed to be a referee for you.
o In addition to naming one academic referee in your cover letter, please ask a second academic referee to write a one-page letter of recommendation. The referee (not the applicant) should submit this to the e-mail address specified above, with the subject line: Academic Reference for [YOUR NAME], also by the deadline specified above. This referee must be different from the person named in your cover letter.

Following Application

Suitable candidates will be short-listed during August. We will endeavour to set up interviews (face-to-face or online) at the end of August/early September. Shortlisted candidates will be asked to supply electronic copies of university awards and marks at that point. We hope to make an offer by the end of September. Candidates must be available to start in early October.

Thursday, 21 July 2011

Webinar on developments in Machine Translation

A colleague just pointed this webinar out to me - might be of interest to some readers? If any of you do attend, I'd be interested to hear your thoughts.

Building Strategic Advantage and Revenue from Machine Translation

Tuesday July 26, 4pm, UK time (for other time zones see below)

Many translators and LSPs have been curious about Machine Translation (MT), but have yet to fully understand how they are able to build strategic advantage and increase revenues. Recent advances in MT have changed the landscape from being able to get the meaning (gist) of text in another language using MT to being able to publish the output with very few human edits. Industrial-strength customized MT now offers near-human quality and greatly improved translation throughput. With this increased productivity come new market opportunities for LSPs. During this webinar, we will explore the following:

Part 1: Understanding MT Technology and Data

● Overview of Machine Translation
    - Different machine translation approaches
    - Where MT makes the most sense
    - Where humans and MT meet

● Requirements for Building a High Quality Near-Human Quality Translation Engine
    - Understanding the data
    - MT and human translation are more similar than you think

● Incremental Improvement of Translation Quality
    - The importance of corrective feedback
    - Focused correction
    - Building terminology
    - Adding the right kind of data

Part 2: Business Models For Strategic Advantage and Long Term Revenue

● The Long-Term Benefits of a Good MT engine
    - Lower cost, greater productivity and speed
    - Business opportunities
    - TM or MT – should they be treated any different?

● Sustainable Revenue Models for LSPs
    - The importance of focus when selecting and leveraging a domain
    - Generating recurring revenues
    - Targeting multiple clients in the same domain

● Expanding the scope of translation – Strategic Advantage
    - Extending beyond the traditional human translation limitations
    - Customer support & conversations
    - Moving closer to real time

Times:
EDT 11:00 / 11AM (UTC – 4) Boston, New York
PDT 08:00 / 8AM (UTC - 7) Los Angeles, San Diego, San Francisco, Seattle
BST 16:00 / 4PM (UTC + 1) London, Dublin
CEST 17:00 / 5PM (UTC + 2) Amsterdam, Prague
EEST 18:00 / 6PM (UTC + 3) Bucharest

Space is limited.

Reserve your Webinar seat now at:
https://www1.gotomeeting.com/register/645356736

After registering you will receive a confirmation email containing information about joining the Webinar.

System Requirements
PC-based attendees
Required: Windows® 7, Vista, XP or 2003 Server

Macintosh®-based attendees
Required: Mac OS® X 10.5 or newer