2010 Dec 17, 6:17
2009 Nov 23, 2:20
"The Atlas of True Names reveals the etymological roots, or original meanings, of the familiar terms on today's maps of the World, Europe, the British Isles and the United States. For instance, where
you would normally expect to see the Sahara indicated, the Atlas gives you "The Tawny One", derived from Arab. es-sahra “the fawn coloured ,desert”."
2009 Aug 7, 7:03
Starting with an English phrase, Translation Party converts back and forth between English and Japanese until the result no longer changes.
2009 Mar 16, 4:22
"This data set, contributed by Google Inc., contains English word n-grams and their observed frequency counts. The length of the n-grams ranges from unigrams (single words) to five-grams. We expect
this data will be useful for statistical language modeling, e.g., for machine translation or speech recognition, as well as for other uses." 6 DVDs for only $150 with licensing restri... ok nm.
2008 Jul 3, 11:50
Another translation error. At least they're upfront about it though.
2008 Feb 18, 3:05
A case study on the origins of a humorous mistranslation. FTA: "The really weird ones are apparently from dictionary look-up errors ... not just taking an unlikely choice from the correct entry, but
actually reading a different (but nearby) entry."
2007 Jul 4, 10:58Hackdiary
I really enjoy reading Matt Biddulph's blog hackdiary
. An entry some time ago talked about his Second
Life flickr screen
which is a screen in Second Life that displays images from flickr.com based on viewers suggested tags. I'm a novice to the Second Life scripting API and so it was from this
blog post I became aware of the llHTTPRequest
. This is like the XMLHttpRequest for Second Life code in that it lets you make HTTP requests.
I decided that I too could do something cool with this.
I decided to make a translator object that a Second Life user would wear that would translate anything said near them. The details aren't too surprising: The translator object keeps an owner
modifiable list of translation instructions each consisting of who to listen to, the language they speak, who to tell the translation to, and into what language to translate. When the translator
hears someone, it runs through its list of translation instructions and when it finds a match for the speaker uses the llHTTPRequest to send off what was said to Google translate
. When the result comes back the translator simply says the response.
Unfortunately, the llHTTPRequest limits the response size to 2K and no translation site I can find has the translated text in the first 2K. There's a flag HTTP_BODY_MAXLENGTH provided but it defaults
to 2K and you can't change its value. So I decided to setup a PHP script on my site to act as a translating proxy and parse the translated text out of the HTML response from Google translate. Through
experimentation I found that their site can take parameters text and langpair queries in the query like so:
. On the topic of non US-ASCII characters (which is important for a translator) I
found that llHTTPRequest encodes non US-ASCII characters as percent-encoded UTF-8 when constructing the request URI. However, when Google translate takes parameters off the URI it only seems to
interpret it as percent-encoded UTF-8 when the user-agent is IE's. So after changing my PHP script to use IE7's user-agent
US-ASCII character input worked.
Actually using it in practice is rather difficult. Between typos, slang, abbreviations, and the current state of the free online translators its very difficult to carry on a conversation.
Additionally, I don't really like talking to random people on Second Life anyway. So... not too useful.
2007 Jun 6, 4:59
A list of free web based language translation services including language guessing which I think is neat.
2006 Sep 14, 11:35 2005 Apr 9, 9:13 2005 Apr 4, 3:28