I see a several issues and it seems like there are patterns emerging in the responses too. One, you need to have a strategy for dealing with alternative language users. Yes, the topic is huge, but that just makes it more important. Language support is the #1 accessibility issue on the internet, too many times we just ignore that fact. You should at least have a strategy, even if its got holes in it. Do the best you can is definitely superior to being intimidated and not doing anything.
There are three categories you need to break down your content into. There’s stuff that needs translation, stuff that would be nice, and stuff that you’re not going to do. Are you going to translated your sales pitch? Your about us page? Your blog? Your blog’s comments? You don’t have to have the same strategy for different types of content. If you don’t have the resources to keep up with translating a daily blog, then that’s not going to be part of your strategy. Maybe you have the resources to translate all of your blog comments, but you don’t see the value is there to make it worthwhile. It all comes down to knowing your site and knowing your audience. I completely agree with the all of the folks who’ve said “Is there any reason you can’t do a quick poll?” These are the kinds of issues that user experience design is all about.
That also includes deciding which languages to support in which way. I’m sorry, but I work religiously in Unicode. There are literally dozens of language codes recognized for the web and Unicode supports encodings for alphabets, syllabaries, etc. in languages even beyond that. I don’t think anyone would seriously advocate for ensuring accurate translation of your page into Middle Persian? I mean, it’s a dead language. Come on. Is there enough value to translating your blog posts into a Native Canadian language that’s spoken by 50,000 people in Nunavut (that’s a pretty successful indigenous language, many have fewer than a 1,000 speakers)? Are you going to translate into Chinese if you don’t have shipping to China? There are a lot of Chinese immigrants around the world, but is it going to make more money than you’ll spend producing (and maintaining) the translations? Are you going to translate your shopping cart into other languages? How many? Which ones? How are your users going to access alternate languages? Are you going to search their HTTP headers for their preferred language or does your shopping cart plug in not support that? What’s your progressive enhancement strategy? When you sit down with a website, its usually pretty obvious where there’s translation value and where not. It’s pretty obvious which areas are going to be relatively static and you can contract out some translation, vs. where you’ll be making frequent changes and translation would mean having someone in-house or out-sourced for a long term situation. When you’ve done your research (study, analytics, user testing), then you can start making some informed decisions.
And I have to disagree with the folks who’ve said “well, its better than nothing?” The “slave” translation is just one example where a bad translation can do serious damage to your website. This is especially true for translations into languages that no one affiliated with the website speaks, because there is no official means of fact checking those translations. Problems can be public for long periods of time, undermining a tremendous amount of marketing, customer service, and operations dollars that are trying to make the website / organization look good. It is one thing for a visitor to run your website through a translation program and try to interpret the results. It’s another for you to offer it that way. People are going to have a higher expectation that you are vouching for the content, and in that case, I would recommend you not do that for text that you have no real idea what it says. There may be portions of the site where it’s less critical and providing an automated translation does have value, but that should be part of your strategy. I think it’s hopeful to say that the automated translation is better than nothing, but the technology has not reached that point yet. Machine grammar and human grammar are fundamentally different paradigms. Also, there is currently no compiler capable of declaring the words coming out of your mouth to be illegal syntax. If the task of automated translation is possible, it has clearly not met any relevant end user criteria yet, so think about what you do and don’t want translated on your site.