Just as defining workflows and putting in place metrics are essential parts of an effective corporate strategy for post-editing machine translation (PEMT), so is the development of a post-editing guide for linguists. Z-Axis Director Uwe Muegge will present a simple approach to writing a PEMT guide. He will demonstrate how a PEMT guide has a positive impact on post-editing output, which in turn frees up linguistic QA resources for more important work.
3. TAUS MT Post-Editing Guidelines
• Four guides in one document
• Evaluating Post-editor Performance
• MT Post-editing
• Post-editing Productivity
• Pricing Machine Translation Post-
editing Guidelines
• Post-editing guide
• Defines quality levels
• Provides concrete recommendations
• Available for free from TAUS
website
Home>Academy>Best Practices>Post-
Edit>MT Post-Editing Guidelines
4. ISO 18587 Post-editing of machine translation
• Main part includes sections on
• Post-editing process
• Qualifications of post-editors
• Requirements of full post-editing
• Annex includes sections on
• Post-editor training
• Light post-editing
• Pre-editing
• Available for CHF58 from ISO
website
www.iso.org>Search ISO>18587
6. Require the use of a translation memory system
• Translation memory systems are a
superior editing environment
• Source and target segments are in
close proximity and in synch
• Simple completeness check (it’s hard
to miss a segment!)
• Automatic terminology recognition
• Automatic fuzzy matching and
manual concordance search ensure
consistency
• Many systems feature automatic
quality assurance functions
7. Require that terminology be managed
• In some post-editing projects, clients
may tolerate grammatical and stylistic
imperfections. Incorrect terminology is
almost always an issue.
• Having project-specific terminology
available in a translation memory
system dramatically improves post-
editing quality and productivity.
• Especially in large projects involving
multiple post-editors, managing
terminology can make the difference
between success and failure.
9. Identify issues by measuring post-edit distance
Use a tool for measuring post-edit distance (e.g. TAUS DQF, Post-Edit Compare,
Welocalize weScore) to identify major recurring problems.
10. Use your translation memory system to identify major recurring problems.
Note: This method is very labor-intensive and time-consuming!
Identify issues by inspecting translation memories
12. Start with one post-editing guide
per language.
If your post-editing projects vary
greatly, write separate guides
rather than including too many
rules in one guide.
Keep your
post-editing
guide simple
and short
Focus on the
biggest
issues
14. Stakeholder Alignment Maturity Plan
Level 1
INTRODUCTION
(Client stakeholder signs off on all assets)
Level 2
CLIENT STAKEHOLDER ALIGNMENT BUILDING
(Client stakeholder signs off on high visibility assets
only)
Level 3
CLIENT STAKEHOLDER ALIGNMENT MAINTAINANCE
(QA signs off on all assets)
Meet & Greet between Client Stakeholder & QA
Mgmt
1) Explain QA Mgmt role & 2) Set QA Mgmt as POC for
quality
Note: QA Mgmt should proactively look for the best
ways to offload Client Stakeholders from linguistic
review and maintain high quality of the translations at
the same time.
Proactive approach by QA Mgmt
Driving communication around resolving quality
issues, reaching out to Client Stakeholder delaying
sign off, supporting Client Stakeholder-QA
alignment/communication
Maintaining alignment with Client Stakeholder
Periodical meetings with Client Stakeholder, survey,
to check a satisfaction level from QA services
Maturity Levels Level Description Estimated Timeline
GROUP-1 LANG.
Avg. volume/quarter
> 300k words
FIGS, RU, PL, TR, NL, CZ, JA,
KO, CN, ES-LA, FR-CA, PT-BR
Starting point
After 3 months
After 6 months
GROUP-2 LANG.
Avg. volume/quarter
< 300k words
AR, SK, RO, FI, SE, NO, DA,
VI, TH, IN, UA, CRO, SL, BUL
Starting point
After 6 months
After 9 months