STC Translation Tools & Technology 2 — Questions and Answers
Question 1: What is 'post-editing' in the context of machine translation?
- Proofreading a human-translated document for typos
- A human translator correcting and refining the output of a machine translation engine (Correct answer)
- Adding footnotes and commentary after a translation is complete
- Reformatting a translated document to match the source layout
Correct answer: A human translator correcting and refining the output of a machine translation engine
Post-editing involves a human translator reviewing and correcting the raw output of a machine translation engine to achieve an acceptable level of quality.
Question 2: What is a 'fuzzy match' in a translation memory system?
- A translation that is phonetically similar to the source text
- A source segment that partially matches a stored TM segment but is not identical (Correct answer)
- An approximate match between two different target languages
- A segment flagged as potentially offensive by the QA tool
Correct answer: A source segment that partially matches a stored TM segment but is not identical
A fuzzy match occurs when a new source segment is similar but not identical to a stored TM entry, typically expressed as a percentage similarity (e.g., 75% match).
Question 3: What does 'leverage' mean in translation memory terminology?
- The negotiating power a translator has over project deadlines
- The savings gained by reusing previously translated matches from a TM (Correct answer)
- The ability to export translation files to multiple formats simultaneously
- A method for adjusting translation memory segment weights
Correct answer: The savings gained by reusing previously translated matches from a TM
Leverage refers to the cost and time savings achieved by reusing exact or fuzzy matches from a translation memory, reducing the volume of new translation needed.
Question 4: What distinguishes localization (L10n) from translation?
- Localization refers only to translating legal documents
- Localization adapts content culturally and functionally for a target market, beyond just converting language (Correct answer)
- Localization is a faster, less accurate form of translation
- Localization involves translating only numeric and date formats
Correct answer: Localization adapts content culturally and functionally for a target market, beyond just converting language
Localization goes beyond translation by adapting content—including cultural references, formats, images, and UI elements—to fully suit the expectations of a target locale.
Question 5: What is a '100% match' in a CAT tool?
- A segment where the translator has confirmed 100% satisfaction with their output
- A source segment that is identical to a previously translated segment stored in the TM (Correct answer)
- A perfect machine translation output that requires no post-editing
- A document that has been fully reviewed by a second translator
Correct answer: A source segment that is identical to a previously translated segment stored in the TM
A 100% match means the current source segment is identical to one already stored in the translation memory, allowing the translator to reuse the existing translation directly.
Question 6: Which of the following best describes neural machine translation (NMT)?
- A rule-based translation method using grammatical dictionaries
- An MT approach using deep learning neural networks trained on large bilingual corpora (Correct answer)
- A translation performed collaboratively by a network of human translators
- A system that translates speech in real time using voice recognition
Correct answer: An MT approach using deep learning neural networks trained on large bilingual corpora
Neural machine translation uses artificial neural networks—specifically deep learning architectures—trained on large amounts of bilingual text to produce more fluent translations than older statistical methods.
Question 7: What is the purpose of QA (Quality Assurance) checks built into CAT tools?
- To evaluate whether the translator holds valid professional certifications
- To automatically detect potential errors such as missing translations, inconsistencies, or terminology violations (Correct answer)
- To compress and archive finished translation projects for client delivery
- To generate word counts and invoices for completed projects
Correct answer: To automatically detect potential errors such as missing translations, inconsistencies, or terminology violations
Built-in QA checks in CAT tools scan translated segments for issues like untranslated text, number errors, tag problems, and termbase violations before final delivery.
STC Translation Tools & Technology 2
What is 'post-editing' in the context of machine translation?