Machine translation has come a long way since the application of neural networks to this task in the middle of the last decade. At the same time, the science of localizing websites to extend coverage to additional markets has grown by leaps and bounds.
Localization is by now “settled science” with hundreds of service providers and dozens of decent tools designed to simplify the process of creating multilingual websites. We’ll take a look at your options to achieve this result and focus on how to achieve this goal with a minimum of coding.
The Main Approaches to Website and App Translation and Localization
If you are an old school, hands-off website owner, the safe way to go is to entrust the task of translation and localization to outside experts. The terms “translation” and “localization” are often interchangeable, so most agencies that provide translation can also deliver localization and vice-versa.
If you’re inclined to take this route, google a short-list of the recommended agencies, describe your localization task, get free quotes, and make your selection based on the responses.
Working via agencies is the easy way to get a job done, effectively outsourcing this task and therefore avoiding doing the work yourself. You will pay a premium for this privilege: the cost of an agency is significantly higher than that of a freelance translation or localization resource.
If you want to outsource to a freelancer, you can go to freelance marketplaces like Upwork and Freelancer and collect bids for your project, then select the one that looks capable of delivering the best bang for the buck.
For the full-stack hacker, there’s the API route. Google is one of the leaders in the machine translation game, and the company has invested heavily both in the core algorithms needed for translating among more than 100 languages to one extent or another.
The latter qualification reflects the fact that Google does not support some of the more advanced functionalities, like voice interpretation and camera translation, in all languages, but only among several dozen.
If you are dealing with an exotic language or have highly customized translation needs, you will probably need to go the API route and resort to “training” the “auto ML” engine to recognize and execute the translation pair you require.
If you are a hard-core hacker, this may be satisfying if challenging approach. But this is overkill for many developers, and clients, who simply want to deliver a working translation application for a commonly used language.
Hacker-Friendly Shortcuts using Off-the-Shelf Web Services
For Hackers who seek the middle ground of plug and play web services, there are a few attractive options. Integromat, Automate, and IFTTT offer highly integrated collections of tools and services which make it easier to connect one web module to another.
Yahoo’s Pipes has years ago been retired, but the metaphor of plumbing lives on to illustrate the essential nature of the task. Microsoft also remains in the automated web services with its flow product. But for our time and money, the way to go is with Zapier.
Based in Sunnyvale, but with a global footprint, the company has grown dramatically since its first release in 2012. Its collection of “Zaps” makes it a snap to complete a wide variety of web tasks without programming, and one of those tasks is automated translation. Let’s take a medium deep-dive and get our hands wet to demonstrate this feat.
Let’s Take a Shallow Dive into Zapier’s Translation Zap
Zapier offers an automated translator for your apps. Simply insert a translation step into any Zap and transform content, comments, messaging, and news into web pages into the language preferred by the individual user.
Whether it’s French or Mandarin or dozens of other languages, the solution delivers seamless translation without forcing users to resort to that annoying copy/paste experience on Google Translate or another external site or app.
Zapier’s solution is a SaaS webservice that connect other webservices, without programming. Essentially it is a workflow editor with thousands of potential paths among such services, each designed to achieve a specific task.
In the translation fields, Zapier offers two core translation functionalities, each supporting more than 100 languages.
- (Optionally) Auto-detect the language
- Translate from one language to another
With those two functions, there is a plethora of tasks which you can suddenly execute, without programming. For example, you can:
- Translate local weather into the user’s Slack channel
- Translate an RSS feed and then generate a new RSS in the translated language
- Translate new Twilio SMS messages and archive them on Google Sheets
- Create comments on non-English Zendesk support tickets with their translations
Since web services like these have taken over our digital world in the last decade, the possible permutations are virtually endless. Zapier provide an army of agents (Zaps) to connect them.
The common value is translating an input in one language to output in another. The benefit is bridging language gaps: customers and staff each sends and receives messages in the language with which they are most comfortable.
Let’s take a step-by-step look at a simple sample to give you an idea of the simplicity of the concept and its execution.
Simply log in to the source web service, in this case a WordPress account you control. Zapier has trusted relationships with hundreds of software service providers, so there the inter-app connections are secure and there are no OAuth problems.
When a new WordPress post is published, it triggers the Translate Zap, which can be set either to autodetect the post language, or you can skip this step and enter default source and destination languages.
When the translation is completed, that triggers the creation of an outgoing email in a webservice of your choice, such as Gmail or Office365.
The product is metered according to tasks, like the sample above, and according to the diversity of “Zaps” you employ.