Human Translation vs Machine Translation

This is part 3 of 4 on localization best practices, primarily targeted to developers or studios that are using or are considering using translators for their localization.

In this section, you’ll get a quick rundown on how to access your localization tools, configure your localization settings, and begin adding strings and automated translations.

Localization is the art of conveying concepts from a source locale to a target locale. Oftentimes, this goes beyond the translation of words – it involves communicating ideas and evoking emotions and reactions. The content may even change in translations.

This is especially true for highly creative content such as movies or games titles. For instance, the “Pirates of the Caribbean” was translated to “Curse of the Caribbean” (“Fluch der Karibik”) in German for a better flow.

Automatic Translation

Automatic Translation is a great tool that is practically instantaneous, that is low cost, and whose quality improves continuously thanks to machine learning. It is particularly useful for lower budget productions, for high volumes of string content, for venturing into a market that is not yet ready for more costly human translation, or for testing an experience for internationalization bugs.

However, Automatic Translation isn’t meant to replace human translators. There may often be gaps in context as well as linguistic nuances that can only be picked up by human translators. For instance, the word “pumpkin patch” may playfully refer to an avatar accessory. Machine translation would likely translate it as “field of pumpkins” whereas a human translator could correctly identify it as a pumpkin shaped eyepatch.

Pumpkin (Eye) Patch or Pumpkin Patch?

Human Translation

Human Translation, especially if conducted by qualified and/or professional translators, can produce high-quality output and provide people from around the globe the same great experience as native speakers of the source language. Human translators can take into account context (e.g. whether “Saw” refers to a sharp tool, the verb “see”, or the title of a horror movie), pick up linguistic intricacies such as formality (e.g. Spanish “tu” or “usted”), regionality, and even adapt the content to make it culturally relevant or appropriate for the target audience. This is especially true for highly creative content such as experience titles and descriptions, but also in-game content.

Three different “saw” images that translate differently in other languages:

Human translators can identify the correct meaning of a word where machine translation may struggle.

A Hybrid Strategy: Machine Translation Post-Editing

Machine Translation Post-Editing (MTPE) is a combination of both approaches, in which human translators review automatically translated output. Benefits of MTPE include that any newly added UI content is instantly translated by Automatic Translation. Additionally, human translators provide high quality translations during the review process and at higher speed that translating from scratch.

That concludes our overview on the various pros and cons and strategies for machine or manual translation. To continue onto our localization best practices topics, head over to Tips and Tricks for translators