Instantaneous Translation: Almost a Miracle

   Herewith a peculiar post. I want to point out and discuss something that is in front of the nose of every visitor to my blog. Accordingly, many will have noticed and followed up, while for others I may be bringing news.

   The appearance of the feature I want to discuss is misleadingly modest. Not far below the picture of my face (on the right side of the blog’s front page) is an invitation that says “Translate” and right below that you are told to “Select Language”—all of an inch’s worth of text. But that modesty ends when you click on Select. What then shows up is a list of 90 languages that this Google-driven feature is prepared to deal with. 

   Ninety languages! Who, among my readers, can name more than fifteen or maybe twenty? Yes, far more than 90 are in use around the globe, but among those for which there is written expression and is also used by more than a tiny population, those 90 constitute a pretty respectable number. Written, to be sure, in different scripts. More than ten certainly. While I went through them all, I can’t be sure of the total without a careful study that would involve printing out a whole bunch. E.g.: is the alphabet used for Serbian the same as that for Russian? There are more such cases that require distinguishing between similar looking alphabets and identical ones.

   If you haven’t tried it, the Translation feature is very simple to use and remarkably fast. With a text, presumably in English—a post, as the lingo has it—on the left side of the page, click first on the arrows in the box under “Translate” and then on the language into which you want your piece translated. In five seconds or less, in my several trials of not very long pieces, the text is transformed into the language you have chosen. And as a bonus, the list of labels that appear below the translation is also turned into the language you have selected.

   So far so good. A check of the of these rapid transformations reveals a quite mixed results. I only tested seriously translations into German. Since that is my best non-English language by far, I could judge the adequacy with which the job was done. Let me start by saying that you certainly don’t get good German prose, comparable to the English input. Can you make out what the English author had to say? Mostly, but by no means always. (It certainly helps if you linger over phrases and sentences, mulling, “what could he mean by that?” No doubt at times, a light bulb will go on.) Without having tested more than a few pieces of text, I have an additional observation. The more “concrete” the subject matter—my piece about big-time university sports for example (http://rhweingartner. blogspot.mx/search/label/Sports)--the more intelligible the translation. I compared it with a much more “abstract” essay, that entitled, “Art that is Heard is Not Like Art that is Seen” (http://rhweingartner.blogspot.mx/search/label/Music%20and%20Art) which  was harder to make out in the German translation.

   What to make of all that? The progress that has been made is quite remarkable. Years ago computer people at Carnegie Mellon told me that it would be ages before such translations would be possible, as they also thought the ability of a computer to "understand" speech would be in a distant future. Instead it has come to annoy all of us who have to jump through hoops before we get to talk to an actual  person on the phone. It is my guess, the speculation of someone largely ignorant of the world of computers, that what has been achieved is above all a function of phenomenal increases in memory. Somewhere in Googleland vast vocabularies of all these languages are stored, to be elicited when the button is pushed. Progress is not as far with the more complex matters of syntax and semantics, making the computer not a wholly skilled translator.  These guesses may well be wide of the mark. I hope that some readers of these remarks will comment and enlighten me.

   In any case, the citizens of the United States are hardly known for their knowledge of languages. We should therefore be thankful for what Google hath wrought and has presented to us—yea verily for free.