Translation technology plugging the gaps_ecpd

r

Welcome to our webinar

Translation Technology:

How can we plug the gaps and fix what’s wrong?

with Jost Zetzsche

29 January 2014

Twitter hashtag: #WhatsMissing

www.ecpdwebinars.co.uk

How to interact with us

www.ecpdwebinars.co.uk

Translation Technology How can we plug the gaps

and fix what’s wrong?

© Jost Zetzsche

Twitter: @jeromobot

Web: internationalwriters.com

How did we get here?

• Prior webinar two weeks ago to collect ideas

and concerns where translation technology

falls short

• Around 100 participants

• Additional suggestions by email and Twitter

What did we come up with?

• Voice recognition

• Access to external resources

• Termbases

• Translation memories/corpora

• Machine translation

• General user-friendliness

• Exchange standards

• Other

Who did we send it to? • Across

• AIT

• Alchemy Software

• Atril

• Heartsome

• Jive Fusion

• KantanMT

• Kilgray

• Lingotek

• Lingua et Machina

• Lionbridge

• MadCap

• Maxprograms

• Memsource

• MetaTexis

• Microsoft

• Multicorpora

• OmegaT

• Pangeanic

• SDL

• STAR

• Tauyou

• Terminotix

• Text United

• Tilde

• Western Standard

• Wordbee

• Wordfast

• XML-INTL

Who did not respond? • Across

• AIT

• Alchemy Software

• Heartsome

• Lingotek

• Lionbridge

• MadCap

• Maxprograms

• MetaTexis

• Microsoft

• Pangeanic

• Tilde

• Western Standard

Who did respond?

• Atril

• KantanMT

• Kilgray

• Lingua et Machina

• Memsource

• Multicorpora

• (OmegaT)

• SDL

• STAR

• Tauyou

• Terminotix

• Wordbee

• Wordfast

• XML-INTL

What did we get?

• A lot of respect and surprisingly honest

answers (most of the time)

• Indications of which of our suggestions will be

implemented by which tool vendor

• A little bit of marketing and "show-and-tell"

• Also: Different attitudes between "old" and

"new" tools

Please send an email after the webinar to

[email protected]

so that I can send you the spreadsheet

What to consider

"Our development plan is quite full. I would be lying if I

say it's not possible to turn it around now.

"Therefore I can only say “we do it” for a very small

part of your list. But all points seem “logical” and stuff

to be done in the mid- or long-term."

Voice recognition

• SDL Trados: Great project for some Open-

Exchange developer (the "TEnP" concept)

• Déjà Vu: fairly extensive testing

• Terminotix: hired expert

• MemSource/XTM/Wordbee: will do more

• Peer-to-peer support

Access to external resources

• Trados: through OpenExchange

• Wordfast Anywhere, Transit: already have

access to user-definable resources

• Logiterm: has a large number of resources

and can integrate them right into termbase

• memoQ: access to a good number of

resources, upcoming: TaaS

• Others: will work on it

Termbases

"Comparing apples and oranges"

• Trados: realized because of OE apps that

many users use MultiTerm in a more simple

manner than it was built for

• Multitrans, Logiterm, Transit, XTM: Some

morphology support

• Wordbee: We're not a terminology tool,

MultiTerm is great

TMs/corpora

• Déjà Vu, Multitrans, Wordfast Pro, Trados/

OE: will develop better QA processes for TMs

• Déjà Vu, Wordfast Pro, MultiTrans, Transit:

use fuzzy repairs/MT repairs (Transit: some

users don't like it)

• memoQ, XTM, MemSource: will implement

fuzzy repairs/MT repairs

• Wordbee: "None of our clients has asked us

to implement fuzzy match repairs yet."

Machine translation

• Enhanced user-friendliness key component

for Tauyou and KantanMT

• One of KantanMT's technologies: "Instant

Segment Retraining"

• Tauyou: "In some cases, [instant training] is

possible, while not in others (or too dangerous)."

"Can developers be persuaded that, although they are

IT experts, we freelance translators have varying

degrees of expertise in computing packages and

troubleshooting IT problems? Their assumption often

seems to be that we understand computer

programming as well as translation."

"Conversely, could translators learn to report technical

problems with precision and *a lot* of screen shots?"

User-friendliness

• User role-based interface:

• Transit already has one

• MemSource is planning one based on user

experience

• Trados, memoQ, Wordfast Pro will develop one

• regex-based controls:

• Virtually all (desktop-based) tools use it but don't

want to give it up to not lose flexibility and power

• Solutions: wizards, storing of expressions,

some UI, many examples in docs

User-friendliness

• The semi-WYSIWYG approach is the (old

and) new standard

Segmentation

• Widely acknowledged that there are

limitations and that those limitations have

been reached (exception: Transit's virtual

joining of target segments)

• Document-specific standards through SRX

Standards

• XLIFF is widely seen as disappointing

• memoQ and XTM: package standard

• Standards for word counts: about evenly split

between GMX-V and MS Word standards

• We need to define more clearly what I meant

by

"exchange protocols for

server-based processes"

Standards

• Keyboard shortcuts

• Generally embraced, but . . .

Standards

• Possible for QA exchange?

Quality control of tools

• Virtually every vendor said that they are doing

more and more testing but that it's virtually

impossible to foresee every scenario

Peer-based workgroups

• Wordfast and MemSource/XTM/Wordbee:

offer it

• memoQ: just launched it

• Trados: priority, through GroupShare

• Déjà Vu: open to it

What we have achieved

• Learned about some capabilities in tools

• Extended the "user group" concept

• Communicated to MT vendors that we want

them to adjust tools to our needs as well

• Initiated development

• (Maybe) resurrected standards (SRX, GMX-V)

• Learned that we need to communicate better

What's next

• ?

• Follow-up call next year?

• See what progress has been

• See whether there are new items

What else?