Multilingual efforts over the years
As the pandemic slowed down, a new awareness of the ways in which the world is interconnected brought increasing concern with the plans for Gutenberg Phase 4 — Multilingual Support.
Multiple languages were not by any means a new phenomenon for WordPress. “WordPress has always been localized,” says Francesca Morano. “Localization of WordPress and the wealth of plugins we have is the big factor in making WordPress so popular.” A look at the history of WordPress shows that localization gathered momentum as time went on.
Multiple language packs, the technology that allows automatic translation on WordPress sites, were available as early as 2007. The first Rosetta site—WordPress.org websites in non-English languages—was launched in 2008. It was in Bulgarian, largely because a Bulgarian contributor put in the time. While this was a classic example of progress where a highly motivated individual was involved, WordPress continued to move toward better multilingual performance every year.
2013’s version 3.7 featured “better global support,” including automatic translation for version updates. 2013 was also the point at which the mission and work of the Polyglots team became more structured, Francesca says.
In 2014, version 4.0 included the capacity to install WordPress in a variety of languages, leading to what Aaron Campbell called “explosive growth” of installs in other languages and outside the United States. By the end of the year, non-English WordPress reached an impressive milestone, with downloads outpacing downloads of the English version.
In 2015’s State of the Word presentation, Matt announced that all plugins and themes would support language packs. The plugin and theme directories were localized and became available in more languages that year. In 2016, WordPress added the language switching feature for users on a WordPress site. Users could choose their site’s admin language from a drop-down menu. 2017’s 4.7.2 release brought in numerous additional language packs. More continued (and continue) to be added in nearly every release. Plugins and themes continued to be required for complete translation, but the changes to the software were speeding up.
All these steps forward centered on translating the interface of the software. Pascal Birchler, a core contributor, wrote in 2019, “WordPress is still a platform that does not offer us a comfortable and unique solution to have a multi-language website. We have several plugins that solve many of the problems we are encountering, but it’ll be a while until WordPress becomes a fully multilingual platform.” This comment reflected the growing desire for a platform that supported multilingual content as well as provided multilingual access to the tools.
Increasing global connections were reflected in some other WordPress events around this time.
In 2020, WordCamp Spain took place online, welcoming Spanish-speaking visitors around the world. “We have consciously given a push to unite both sides of the pond thanks to this common interest: the growth of WordPress in Spanish,” wrote Pablo Moratinos, the organizer of the WordCamp. Matt favors more language-centered WordCamps, too.
In 2021, a language switcher was added to the login screen in version 5.9. Erica Varlese proposed to translate learn.WordPress.org, and by the end of the year, the training site was available in 21 languages.
Changing the language of the admin area of a WordPress website is just a matter of choosing from a drop-down menu of nearly 70 languages, with the site automatically adjusting languages. Over the life of the software, more and more languages have been included. However, even for common languages that have been at the 100%-translated point for years, many elements of most websites are still in English.
Some languages pose extra challenges. Naoko Takano was one of the earliest translators of Japanese. She points out that how Japanese is written means that word counts differ between Japanese and English. Japanese words might be automatically divided by the software in unnatural ways. Languages may use different characters, be written in different directions, and have other special characteristics that must be handled in the programming.
And American English is the automatic default language. If a section of a given language has yet to be translated, English will display as the default. For Lao language users, showing Thai would make more sense. South American Spanish speakers would be better off with Spanish from Spain than English. The Preferred Language plugin allows users to set fallback languages, but WordPress doesn’t do so natively.
In fact, plugins and themes picked up the slack for multilingual websites.
Jeff Paul pointed out that a high level of stability is required before it makes sense to translate software. Still. “Internationalization and accessibility are go-to areas,” he says. “Those are areas where the product could do better.”
WordPress continues to improve in this area with time, but challenges remain.
The content of websites was controlled by the owners and producers of the websites, but an increasing interest in WordPress support of multilingual sites grew over the years.
Organizing a multilingual website can be done in several ways: multisite, all the languages on one site, with or without connections among the various translations, and so forth. These decisions are determined by the site owners and producers. WordPress doesn’t currently offer automatic translations or processes for building multilingual sites, though many plugins do.
“It’s tricky,” says Matias, because there are many different solutions.” Building the collaborative processes in Phase 3 will give greater insights into the needs of multilingual sites and the people who use those websites. “We’re going to see,” says Matias, “the things they wish would exist.”
Jeff Paul mentioned that people in the community would like to see multilingual support (planned as Phase 4) come before collaboration (planned as Phase 3). Matt mentioned that he gets that request often, and Courtney Robertson admitted that she’s one of the people who makes that request. A desire for global diversity is on the rise in the WordPress community.
However, from a technical point of view, the collaborative tools of Phase 3 will provide a foundation for Phase 4. The order of the two is not based on importance or urgency but on the infrastructure required.
Matt points out that localization of content goes beyond language alone. “You might want to have different imagery, or different testimonials, or endorsement celebrities,” he said. “I’m excited about supporting it.”
Localization is the process of making a web page accessible to speakers from another locale or language community. Localization includes translation but also covers things like local currency, laws, and images that may carry different connotations in different cultures.
Internationalization is the process of making a website available for localization. WordPress technology keeps internationalization in mind as part of the project’s commitment to diversity.
Both localization and internationalization are essential for a global web solution.
The Polyglot team comprises volunteer contributors working to translate WordPress into their native languages. Languages are further divided by locales: French, as spoken in Belgium, is separate from French spoken in France.
Currently, 208 locales are at some stage of preparation. Contributors range from one for Icelandic to 3,959 for French (France). “It’s amazing WordPress is available in so many languages,” Yvette Sonneveld points out, considering that all the translation is done by volunteers.
“Most spoken languages in the world have a localized release at the same time,” Francesca reports. She found great satisfaction in working on localization as a volunteer and then as a sponsored contributor. “Contributing to my local community and to the global community,” she said, were among the primary benefits of her work with WordPress.
Automatic translations are already available in browsers like Google Chrome. WordPress website builders and users often use Google’s automatic translating tools for their websites. People can now choose what language to use to access any website. So could artificial intelligence take the place of human translators?
Matt expects that browsers will continue to improve automatic translations to the point where they can do a lot of the heavy lifting. But he also recognizes that AI is a tool to augment human creativity, not to replace it.
Polyglots like Francesca question whether there will ever be satisfying AI solutions for translation. So many local details go into the localization of any page, and automation can’t be aware of all of them. “My heart is a lot more into people,” she says. “I make sure the context makes sense to Italians.”
AI has improved in translation to an impressive degree, and people have also become more tolerant of machine translations. Matias suggests that newer generations of users might have a different attitude to automatic translation. “There’s some outdated mental modeling” of AI translations, he says. “We’re coming from a time when that wasn’t ubiquitous and we know the limitations, we know that it’s rough, we know we’re going to get the best experience in the original language.”
Yvette agrees that people tolerate automatic translations but points out that there are consequences. “When it comes to branding,” she says, “the only way to really build trust is to provide the content with the best user experience.” Matt pointed out that WordPress already uses AI in the form of Akismet. Automation, in general, is used to replace human effort in dirty, dull, or dangerous work. Handling spam comments is a good example. Translation is not. Matt doesn’t see AI as a threat to human translators. “Humans will be around for a long, long time,” he laughed. As the 20th anniversary of WordPress neared, the software and the community were poised to extend the global language capacity of WordPress along with its global reach.