Talk To Me About Translating Web Sites
December 12, 2014 6:48 AM   Subscribe

I'm working on a web site proposal that includes a subset of pages that need to be translated into 8 languages. How do I estimate the level of effort on translation?

I have a contact in a language services company, however they want word counts to to do a quote, which is kind of hard to provide on a web site that we haven't actually developed yet. Does anybody have enough experience with this to provide a rule of thumb? Assume I can find native speakers for all the languages in question. I'm hoping for something like, "Budget two hours per page per language," or whatever, that I can use in the estimate. The CMS will be Drupal, if that matters.
posted by COD to Computers & Internet (9 answers total) 2 users marked this as a favorite
Translation estimates are usually quoted on the basis of the number of words or pages (as in physical printed A4, not web pages). It is very rare to charge by the hour. Or did you want to know how long it would take? Most translators will work at around 300 words per hour. Obviously this will vary greatly depending on the source language, target language, the translator's experience, the difficulty of the text, and, frankly, the amount you pay (if you pay less per word then I need to do more words per hour to make the same amount; a fast translation is rarely a good one). This also doesn't take into account that the translator will likely have other projects, so you can't assume that 300 words would come back to you in 1 hour.

ProZ has aggregate data of the rates their members charge to give you an idea of the cost for different language pairs.
posted by tsh at 7:02 AM on December 12, 2014 [2 favorites]

How long is a page of a website? As soon as I hit the post button here, this page is getting longer.

I've translated websites before. It's immaterial how many distinct URLs fall under a domain name. What matters is the number of words or characters in the source material.

However, with a drupal website there's going to be a certain amount of "template content" (stuff you need .po files for), and a certain amount of "database content." The template content generally doesn't add up to all that much—menu titles, button text, error messages, that sort of thing. I would also point out that this kind of material is often provided to the translator free of context, which requires a certain amount of clairvoyance on the translator's part, because how you translate something will depend on the context in which it's being used. (e.g., if I see the single word "read," is that being used for a menu item or a button? Is it present tense or past tense?) I would try to provide screenshots or the dev website so that translators could answer these questions for themselves.

The database content is the meat of the website. Is your client providing that, or are your writing it from scratch?

Have you created a website of similar scope before? Can you strip out the text assets to get a sense of their total length? I could give you a ballpark estimate (for me, if I were doing the work, which I won't be, because I don't translate out of English) based on that, but I wouldn't give you a firm quote.
posted by adamrice at 7:41 AM on December 12, 2014 [2 favorites]

Seconding what tsh and adamrice said. I would not give an estimate without at least knowing the order of magnitude on word count (hundreds of words? thousands? tens of thousands?) and even then it would be correspondingly ballpark (days/weeks/months).

And to build on what adamrice wrote, if I were unable to provide screenshots prior to translation, I'd think about adding language proofing during user testing (which means adding time and money to your estimate), after the translation gets uploaded, especially for a more complex site.
posted by rebooter at 8:05 AM on December 12, 2014

I recently did a very large print job for a major travel agency. It was done in English, and then translated into Spanish and Portuguese.

Some lessons I learned: 1. Text length and layout can change drastically, especially starting from English. English is pretty compact compared to other languages. 2. Translations are tough; there is no one-to-one way to morph one language into another and keep the meaning.

I was hired to do layout. I wasn't involved in the pricing, but my impression was that no one involved was prepared beforehand for how difficult it ended up being. Also, the layout changes that were needed were not insignificant. Please keep these kinds of things in mind.
posted by Benny Andajetz at 8:21 AM on December 12, 2014

Response by poster: To answer questions above - this is actually a fairly small website for us. It's about 60 pages at the moment - static HTML, and we are proposing to put it into Drupal because the client needs a way to update and manage content without developer help (we are a Drupal shop). Not every page will be translated - the client is leaving that decision until the design firm is hired and we can spend some time with the client and with analytics data to determine what pages need to be available in multiple languages. My guess is no more than 10 pages on the site will be available in multiple languages, partly due to the content, and partly due to the budget. It's a consumer focused website, which needs to be more skimmable than it is now. I'd be surprised if there are 500 words on the average page after the redesign.

We've done multilingual sites before, but the client always took responsibility for the translated content. I've never had to subcontract that out - which is why I'm asking. And I do plan to asterisk the hell out the translation portion of the proposal, that it's all relative to what the final site looks like, so whatever estimate I'm providing is a rough estimate at best, and likely to change.
posted by COD at 8:42 AM on December 12, 2014

I developed a site in which almost all of the content is available in sixteen languages. Since all of the translating was done by volunteers, I can't help you with the costs of translation, but I will say that if any of your target languages are RTL, that will be a major time sink for you. It will also be an ongoing headache for the client—editing RTL languages in most any word-processing application or even plain text editor is endlessly frustrating.

Even LTR non-Latin alphabets can cause problems if a) there are no Web fonts that include the glyphs for that language or b) the translator's deliverable uses a non-Unicode font.
posted by bricoleur at 9:13 AM on December 12, 2014

What's the client's plan for updating the translated content? Drupal fixes the technical problem, but not the language problem. Are they going to contract with the same shop for updates?
posted by instamatic at 4:07 PM on December 12, 2014

Response by poster: //Are they going to contract with the same shop for updates?//

In a word, yes. Contract goes through 2017 and we'll be responsible for updating as needed. We'll probably just bill them our usual rate then sub it out as needed. I don't think updates will be real frequent, most of the content is reference type info that isn't changing often. The blog, where they'll do most updates, is English only.
posted by COD at 5:09 PM on December 12, 2014

Can you put in that section the price per word? Rather than asterisk it all to hell, get a few quotes on price per word, average it out, or take the highest/worst case quote, and base it on that. Or get a quote on 250 words and give a price that says "$X per 250 words" or something like that.
posted by looli at 8:51 PM on December 12, 2014

« Older two travel booking questions   |   Best approximation of listening to 90s radio? Newer »
This thread is closed to new comments.