DeepL MT, a game-changer for professional translators

On September 25, as part of the Proz.com celebration of International Translation Day 2018, I will present “Learn DeepL MT for beginners” (go to https://hubs.ly/H0dP7wp0). All sessions sponsored by ProZ.com on the 25th and the 26th are FREE to all participants.

I believe that at present, DeepL is one of the best tools available for translators. My presentation therefore centers on the mechanics of the platform and I shows you how easy it is to use. As with any tool translators may use, confidentiality of client’s information should always be a priority.

This platform uses Deep-Learning Neural Machine Translation. NMT uses a “large artificial neural network to predict the likelihood of a sequence of words, typically modeling entire sentences in a single integrated model.”  Deep learning NMT “processes multiple neural network layers instead of just one. It is a a completely new generation of artificial neural networks that … learn to grasp the subtle meanings of sentences,and conveys them in the target language in a way not seen before.”

I am recommending the use of this platform and API integration because I have found it to produce quality draft translations that are more fluent than other MTs while productivity may be doubled, tripled or more. It is very intuitive and extremely easy to use. Additionally, it relies on EU-level privacy protocols. At present it only services English, Spanish, German, French, Dutch, Italian and Polish, but they are expanding languages.

Come join us on the 25th and learn the mechanics of DeepL and other interesting tricks related to MT, plus enjoy all the other sessions offered for FREE by ProZ.com.

==============================================================

As part of the study materials I will be distributing, I will share some of my Quality Guidelines for Human-Quality PEMT (Post Editing of Machine Translation Output).

This is what I believe a Human-Quality PEMT should achieve:

Purpose:

·      Accurate (preserve intended meaning of the original)

·      Correct grammar, syntax, style

·      Correct in formal aspects of written language

·      Sensitive to regional variations (dialects and target language use)

·      Written (mostly) in plainlanguage

·      Culturally and Linguistically appropriate (terms and expressions must be understood appropriately and consistently by most people reading text in TL)

·      Follow glossaries when provided

·      Follow style guides when provided

·      Avoid literal translation (too formal, complicated, awkward texts, though semantically equivalent)

·      Aim for “comparable constructs”

Information:

  • Complete (no omissions, additions or changes)
  • Loyal (same message is conveyed)
  • Accurate (same information transferred)
  • Reliable
  • Consistent
  • Technical equivalence (technically and conceptually)
  • Conceptual equivalence (content) (Does it accurately reflect original meaning?)
  • Intent (Have connotations and denotations been transferred)

Essentials

  • Grammar (system and structure of language)
  • Style (word choice, point of view, tone, + syntax)
  • Syntax (rules governing sentence structure, including word order, to achieve particular effects)
  • Semantics (sense, reference, implication, logical form, word relations, cognitive structure)
  • Register (level of formality)
  • Readability Level (easier or more difficult to read)
  • Format and Layout (including document setup, arrangement, charts, graphics, illustrations)
  • Spelling
  • Punctuation
  • Capitalization

Correct and optimal use of language:

  • Language conventions
  • Dates
  • Numbers
  • Addresses
  • Acronyms (transfer)
  • Abbreviations
  • Names of people
  • Names of countries and places
  • Brand names

Others issues to ensure:

  • Fluency
  • Phrasing
  • Nuances
  • Absence of False cognates
  • Cultural Sensitivity (will it make sense to target audience or will it create confusion or concern)

Other recommendations:

CONSTANTLY COMPARE MACHINE TRANSLATION PLATFORMS

  • Select three or four to compare
  • Obtain translations
  • Review each individually
  • Compare changes
  • Select preferred
  • Repeat every three or four months (new or improved versions)

Rating

5         Excellent

4         Correct

3         Enhance

2         Doubtful

1         Incomprehensible

or

5         Top human translation quality equivalence

4         Normal HT equivalence

3         Beginner HT equivalency

2         Misinterpretations

1         Gross Errors

==============================================================

Screen Shot 2018-09-19 at 7.53.00 PM

Join my free session on DeepL Machine Translation on Sept 25 to celebrate Int’l Translation Day sponsored by ProZ.com @prozcom livestream #ProZTV  https://proz.com/tv/ITD2018 

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s