New ISO Standard for Translation Services in Place

26 05 2015

ISO 17100:2015 was published on May 1, 2015, and the content looks a lot like its predecessor, EN 15038:2006.

ISO 17100The core of the standard is established via definitions of various translation processes: translation, revision, review, proofreading, and final verification are delineated, as they were in EN 15038. Advancing slightly beyond EN 15038, ISO 17100 also requires routine practices for handling client feedback. Unfortunately there is no attempt to set quality metrics: a missed chance, according to some, as quality standards would strongly support translation industry excellence.

Another criticism of the new standard is that it does not address such influential new industry elements as crowdsourcing, Agile/on-demand and cloud functions, linguistic QA methodologies, Machine Translation (MT) and post-editing for MT. A GALA blog references data* predicting that the global machine translation market will grow nearly 25% by 2019; and yet the ISO 17100:2015 abstract specifies that “the use of raw output from machine translation plus post-editing” is outside of its scope. Happily, the upcoming ISO 18587 is intended to deal with MT and related issues; it is currently “under development” with no release date projected yet.

ISO standards are intended to provide transparency within a recognized set of controls from start to finish for all parties to any translation job, thus guaranteeing high quality and creating trust. Language Service Providers (LSPs) will have the option to: declare conformity with the standards without any external review, register without any external review, or meet specifications and be certified by accredited certifiers.

Skrivanek Group is among those LSPs who are already in compliance with 17100 specifications, and will achieve certification promptly.

  J. McShulskis



QA Tools Enable Unparalleled Accuracy

25 02 2015

Before translation agencies present final drafts to their clients, Quality Assurance (QA) tasks are performed – some by human beings alone, some by humans with the assistance of QA software.

Quality Assurance ToolsActivities such as proofreading and double-checking terminology with clients are QA methods that contribute to the linguistic excellence of the final draft and cannot be replaced by automation. But there are numerous problems that QA software is adept at highlighting for translators, eliminating hours and hours of human scanning and comparison, and improving accuracy where tedium can lead to errors.

The most commonly used QA tools have varying capabilities, but features often include the detection of:

  • inconsistently translated phrases
  • word/phrase translations that don’t match up with glossary definitions
  • untranslatable words or phrases
  • missing segments
  • punctuation and spacing differences
  • numbers and symbols that don’t match-up

The following QA tools are some of the more common: ApSIC Xbench (which is free), Wordfast Quality Check feature, D.O.G. ErrorSpy, Verifika, Yamagata QA Distiller, and SDL Trados Terminology Verifier and QA Checker. These vary in user-friendliness, effectiveness, and complexity.

To give you an idea of the parameters of QA software capabilities, every mistake is detected by comparison with software databases built into the program or those glossaries and translation memories that you create within it. Limitations exist in these programs’ ability to ‘understand’ different grammatical rules between the source and target texts; nor are they able to compensate for a misunderstanding the translator incorporated into the text, nor spot a poorly written phrase. But they have become indispensible to professional translation agencies, especially for large projects.

  J. McShulskis