During the past few months, I've written articles exploring how to include multilingual page content in web sites to attract more visitors. In my last article, Virtual Keyboards: Input Multilingual Character Sets for the Web, I showed an easy way to create content in non-Western languages and scripts using a virtual keyboard. At the end of the article, I concluded that virtual keyboards work best when you have familiarity with the language in question! Otherwise, you still need someone to translate the text for you. Or do you?
Several
of the big search engine players, such as Google and Yahoo!, provide free language
translation services. An extension of Machine Translation (MT) applications,
this new breed of web-based translation services and tools promise impressive
results. Are they good enough to use in professional applications? That's the
question that I'll attempt to answer in my next couple of articles,
using Google's Language Translation services as the guinea pig. In doing
so, I'll also explore its many features. Now, if you are ready, let us begin
the experiment!
Origins of MT Applications
Using computers to translate from one language to another is almost as old as computers themselves. Case in point, SYSTRAN, the technology behind Yahoo!'s Babel Fish and Google's language tools until about 2007, was established in 1968. At their most basic, translation applications merely substitute words, relying on a database for performing lookups. At the other end of the spectrum, sophisticated MT software takes into account the complex rules of human languages and grammar and can even learn languages from scratch much like a child does.
How Google Language Translation Services Work
Unlike similar services that are built on longstanding technology such as SYSTRAN,
Google's utilizes its own translation algorithms. And why not? It was their
revolutionary search algorithm that started the Google revolution.
The main benefit of using a translation service such as Google's, as opposed
to a desktop application, is the ability to translate web pages in real time
using an interface that allows visitors to select their language preference
from a web control such as a dropdown.
Components of the Google Translation Services
Google's contribution to the translation services arena is really three separate services:
- Document translation
- Translated search
- Translator toolkit
Document Translation
The Translate page is just that:
it offers visitors a simple interface for translating text, web pages and documents.
Text or a web page URL can be inserted directly into a text field. Alternatively,
a document can be uploaded using a standard file upload web control. I couldn't
resist trying it out on my web page. I entered my site's URL and selected French
from the dropdown. For my quote at the top of the page:
A quality instrument is like a race car. When you get behind the wheel, you don't drive it around the city at 35 miles an hour. You take it out on the open road, open it up, and see what she can do. - Rob Gravelle
Google came up with this:
Un instrument de qualité est comme une voiture de course.
Lorsque vous prenez le volant, vous ne conduisez pas autour de la ville à 35 miles à l'heure. Vous le prenez sur la route ouverte, l'ouvrir et voir ce qu'elle peut faire. - Rob Gravelle
Impressively, it's an almost flawless translation!
A nice feature of the Google Translation Services is that positioning the mouse over a line of translated text will show the original in a bubble:
Image 1: Google Translation Sample
Upon copying the text to this article, I noticed that both languages were put onto the clipboard. Intrigued, I had a look at the source code to see what was happening. As I suspected, both the original texts are included "side-by-side" in the source code, with the original text enclosed between "google-src-text" <SPAN> tags:
Google passes the first test with flying colors!