ATA ATA Translation Technology and CAT Tools 2 — Questions and Answers
Question 1: What does 'alignment' refer to in the context of building a Translation Memory?
- Aligning the margins and formatting of source and target documents
- The process of pairing source segments with their corresponding target-language translations to create or expand a TM (Correct answer)
- Synchronizing multiple translators working on the same project
- Matching source-text terminology with a termbase entry
Correct answer: The process of pairing source segments with their corresponding target-language translations to create or expand a TM
Alignment is the process of pairing corresponding segments from an existing source document and its translation to create new TM entries from previously completed work.
Translation Memory alignment is the process of taking an existing source document and its completed translation and pairing each source segment with its corresponding target segment to add to a TM database. Alignment allows organizations to leverage their existing translation archives without having to re-translate already completed work.
Question 2: Which file formats does SDL Trados Studio use for working files and TM exchange?
- .tmx only
- .xliff only
- .sdlxliff for working files and .tmx for TM exchange (Correct answer)
- .mxliff only
Correct answer: .sdlxliff for working files and .tmx for TM exchange
SDL Trados Studio uses the .sdlxliff format for bilingual working files and .tmx (Translation Memory eXchange) for TM data that can be shared with other tools.
SDL Trados Studio uses the .sdlxliff format as its primary bilingual working file format. For Translation Memory data exchange with other CAT tools and platforms, it uses the .tmx (Translation Memory eXchange) format, which is an open standard. Open exchange standards allow translators to share assets across different tools and reduces vendor lock-in.
Question 3: What is the XLIFF format and why is it important in translation technology?
- XLIFF is a proprietary format used exclusively by memoQ
- XLIFF (XML Localization Interchange File Format) is an open standard for exchanging localization data between tools and systems (Correct answer)
- XLIFF is a file format for storing machine translation models
- XLIFF is a format used only for translating software strings
Correct answer: XLIFF (XML Localization Interchange File Format) is an open standard for exchanging localization data between tools and systems
XLIFF is an open, XML-based standard for representing translation data, enabling interoperability between different CAT tools, localization platforms, and content management systems.
XLIFF (XML Localization Interchange File Format) is an OASIS standard that defines a universal format for representing bilingual translation data including source segments, target segments, metadata, and status information. Because it is an open standard, content management systems, localization platforms, and CAT tools from different vendors can exchange XLIFF files without proprietary conversion.
Question 4: What is Neural Machine Translation (NMT) and how does it differ from earlier Statistical Machine Translation (SMT)?
- NMT uses rule-based grammar algorithms; SMT uses statistical patterns
- NMT uses deep learning neural networks to generate fluent translations; SMT relied on statistical phrase tables and was more prone to unnatural output (Correct answer)
- NMT is slower but more accurate for technical texts; SMT is faster for all text types
- There is no meaningful quality difference between NMT and SMT
Correct answer: NMT uses deep learning neural networks to generate fluent translations; SMT relied on statistical phrase tables and was more prone to unnatural output
NMT uses neural networks (deep learning) to produce more fluent, context-aware translations than earlier statistical approaches, representing a major quality improvement in MT.
Statistical Machine Translation (SMT) worked by mining large bilingual corpora for phrase-level statistical patterns. Neural Machine Translation (NMT), pioneered around 2014 to 2016, uses encoder-decoder neural network architectures (including the transformer model behind systems like DeepL and Google Translate) to generate translations by learning deep patterns across entire sentences. NMT produces significantly more fluent output than SMT, though it can still make significant factual and terminology errors.
Question 5: How do professional translators typically handle client-provided Translation Memories that contain errors?
- Always accept TM matches as correct and do not modify them
- Flag and correct errors when encountered, and notify the client about TM quality issues (Correct answer)
- Refuse to use a TM with any errors
- Use the erroneous TM without comment to maintain client relationships
Correct answer: Flag and correct errors when encountered, and notify the client about TM quality issues
Professional practice requires correcting TM errors encountered during translation and informing the client, as flawed TM content will otherwise persist and propagate across future projects.
When working with a client's existing Translation Memory, translators will sometimes encounter segments that contain errors from previous translations. The professional approach is to correct these errors, flag them for the client, and recommend TM cleanup if the errors are widespread. Proactive communication about TM quality issues is a value-added service.
Question 6: What does 'transcreation' involve beyond standard translation, and when is it typically used?
- Transcreation refers only to adding citations to translated academic texts
- Transcreation adapts content creatively for the target market, often rewriting headlines, slogans, and marketing copy rather than translating literally (Correct answer)
- Transcreation is a machine translation technique for creative texts
- Transcreation means translating and then illustrating content with original images
Correct answer: Transcreation adapts content creatively for the target market, often rewriting headlines, slogans, and marketing copy rather than translating literally
Transcreation is used in marketing and advertising to recreate content for a target market, prioritizing emotional impact and brand voice over literal fidelity to the source text.
Transcreation (or creative translation) goes beyond linguistic equivalence to recreate the intent, emotion, and persuasive power of a source text in the target language and culture. It is most commonly employed for advertising slogans, brand taglines, marketing campaigns, and entertainment content. Transcreation professionals often have backgrounds in both translation and copywriting.
ATA ATA Translation Technology and CAT Tools 2
What does 'alignment' refer to in the context of building a Translation Memory?