The web isn’t a country– it’s all countries. Only a quarter of internet users’ primary language is English. Chinese, Spanish, and Arabic account for another third of the world’s netizens, and there are literally hundreds of languages spoken online.
A global audience brings new challenges to communication, and automated translation can only get you so far. The Hanlon Creative Blog reached out to a translation partner we’ve developed a great relationship with to discuss the state of the industry today.
Myriam Siftar is President and Founder of MTM LinguaSoft, a language services partner with expertise blending language, culture and technology. She explained in depth the advantages of robust, human translation:
What makes for a great translation partnership?
Communication. Translation is a service, not a commodity. As long as there are no typos or grammar mistakes, you could have ten different translations and they’d all be slightly different, and they’d all be correct. So, in our industry, a translation is judged according to how well it meets the client’s needs.
We need to spend time understanding those needs: what is the translation used for and who will read it? Is it for consumers or experts? Admins or engineers? Patients or doctors? Is it for print or online publication? In order to give clients exactly what they need, we need to ask them a lot of questions, especially during our first project together.
How are cultural assessments made? Who determines what is culturally relevant or offensive and how, and what are the hallmarks of ‘culturally neutral’ graphics?
Cultural consultants are bilingual subject matter experts who are located in their native countries and familiar with the client’s target audience. For example, if you are marketing medical supplies in Italy, you want someone who knows the industry to review your content and make recommendations for changes that would make it more appealing.
Regarding graphics, one thing to keep in mind is that colors have different meanings in other parts of the world. In some countries, white connotes death, while for others the gloomiest color might be black or light blue. In China, yellow flowers have negative connotations. Numbers are also important – for example, you wouldn’t want to depict four of anything in China because four is an unlucky number.
As far as “culturally neutral” graphics are concerned, it’s easier to generalize about what NOT to do, because you can’t predict whether something is going to look weird in the context of a different culture. In general, you want to avoid showing hand gestures. Things we consider normal, like using the index finger to point, could come off as offensive. Feet can be touchy as well. Some Arabic cultures have strong feelings about feet – it’s rude to display the sole of your shoe, or even to stand with your toe pointing at someone.
What is required to manage a translation memory database?
Professional translators use a type of software called a Computer-Aided Translation (CAT) tool. CAT tools create translation memories (TMs): matched pairs of source and target text that can be leveraged when content is updated or duplicated. If a client’s document includes content that was translated for an earlier project, the CAT tool mines the TM to provide full or partial translations and reduce both the translation workload and the cost to the client.
TMs are particularly useful for content with frequent updates, for example legal documents, machine manuals, and insurance policies. Bilingual glossaries can be also loaded into CAT tools to make auto-suggestions while a translator works.
What is lost when you rely on automated translation versus human translation? Are human and automated translation mutually exclusive, or can both be used together?
The “man vs. machine” rivalry isn’t what it used to be. Translation technologies are improving by leaps and bounds. We’re already using CAT tools, and some global businesses are currently using customized machine translation software for internal-use translations (for example, for user generated content). But for publication-ready translation, even the best automated translation requires bilingual proofreading, or “post-editing.” A single misplaced “not” can lead to disaster!
Some content is not appropriate for automated translation at all. If you want to persuade, spark curiosity, encourage learning, or invoke an emotional response, you need translation by a human who knows how to write copy in their native language. We call this specialized service “transcreation.” The biggest mistake a marketer can make is to carefully craft English-language marketing copy and then expect a quick, cheap translation to be just as effective.
How effective are automated translations? Are paid services significantly better than free services like Google Translate?
If you have a document and you don’t know what it is about, Google Translate can help you get the gist of it. But you’ll need bilingual proofreading at the very least if you are going to use the translation for anything important. It’s a bad idea to use free services to translate from English to other languages, especially Asian and Middle Eastern languages – the tools aren’t reliable enough.
What parts of a site should get priority for translation, and which can be deferred?
The home page is important, of course, and some clients with limited budgets start with localized landing pages or microsites that highlight key offerings for a particular region. For e-commerce sites, clients are tempted to skip the translation for product descriptions if they are selling in a market with a lot of English fluency, but this is a mistake. Another thing that clients sometimes forget about is the hidden text in a website – tags, page titles, etc. – that aren’t visible to the user but which are important for search engine optimization (SEO). These also need translation.
Our lead developer and digital project manager, Kevin Harter, and his development team have had great success integrating and maintaining websites using the WordPress Multilingual (WPML) plugin. Does your team have first-hand knowledge of WordPress and WPML?
Yes. WordPress has become a go-to CMS for many businesses so we encounter it a lot. WPML is a great plugin that can simplify and improve the translation workflow. It also makes it easy to capture content that may be buried in widgets or plugins that is otherwise not easily exportable.
It does require some know-how to set up properly and has a lot of moving parts, so we recommend that clients give themselves enough lead time and have someone on-hand familiar with WordPress configuration. To make sure things are set up properly, we sometimes do what’s called a pseudo translation, where the exported English text is “translated” using random dictionary entries from the target language. If certain content was not properly captured in the export, we’ll see it in English and can make the necessary adjustments before the actual translation starts.
There are lots of different content management systems for websites. How does a website’s CMS impact the translation process?
Your CMS is the software system used for creating and maintaining your website and other digital content. Popular CMSs include WordPress, Drupal, and Joomla. Or a business might use a custom-built CMS. Ease of localization depends on two issues: language encoding and content export.
A translation-friendly CMS publishes web pages using UTF-8 language encoding and stores content in a Unicode compliant backend database. If these aren’t true of your CMS, you might need to change the encoding before localizing.
For translation, it’s usually best practice to export content files so that the localization team can take advantage of computer-aided translation (CAT) tools. The easiest format for CAT tools to work with is called an XLIFF (XML Localization Interchange File Format) but most CMSs provide other types of XMLs. These require some consultation with our project managers to confirm translatable and non-translatable segments.
If you expect frequent updates on an ongoing basis, you might consider a more direct, dynamic connection to your CMS that can be provided by working with a proxy service.
What’s the best domain structure for a localized site?
This is partly an SEO question because the structure of the site impacts how search engines interact with it.
If you are translating into Spanish for North America (localizing for language within an established market), creating a subfolder of your main site (www.yoursite.com/es) is a good option. The Spanish version will benefit from your established authority and rankings with Google. Subfolders are less expensive to create and maintain.
If you are translating into Chinese for mainland China (localizing for a new region), Google rankings are irrelevant because Baidu is the leading search engine there. It favors sites with a country code top-level domain (www.yoursite.zh). The down side is, your Chinese site needs to build its authority from scratch. Also you need to jump through more regulatory hoops to register a Chinese ccTLD, and maintaining this type of site is more labor-intensive. But it makes better sense in the long run.
Your domain structure depends on your overall business plan. Variables include your budget, how frequently you expect to update the site content, the regulations in effect in your target market, and whether your brand already has a strong presence in the region.
How has the rise of responsive websites affected designing for multiple language websites?
People around the world are using tablets and smartphones more than ever. For some regions, it’s not even worth localizing a site unless it’s responsive. If you are updating your site for mobile, now is a good time to figure out what adjustments you can make to ease the localization process later, if you think you’ll translate it eventually.
Text expansion and differences in syntax can cause problems for responsive development. Some languages, like Spanish, take up a lot more space than English. Others require less space (Chinese, for example). Japanese is unpredictable – it can get bigger or smaller. German words tend to be longer than English words, and Chinese words can change their meanings altogether if a line break is misplaced. We work with our clients to solve text expansion issues after the content has been translated but before the site is launched.
Which languages are you seeing the most growth in?
We’ve always seen high demand for the European languages of trade: French, Italian, German, and Spanish (FIGS). We also handle a lot of marketing, government and social service translation into Spanish for US residents. Chinese is on the rise for all our subject domains, and luckily our project manager Ken speaks fluent Mandarin. Demand for Korean is also rising. Brazilian Portuguese saw a business translation boom a couple of years ago which is still apparent, though it’s slowing down a bit.
In the past few years we’ve worked on projects requiring bulk translation into rare languages like Uzbek, Uyghur and a variety of African and Southeast Asian languages. These were for government-sponsored research into language processing. Uyghur was particularly interesting – it’s spoken by ethnic Arabs living in Northwest China and Southeast Siberia.
The fun thing about working in language services is that there are always new things to learn about culture, language, and technology. It’s never boring!