Wow ! is making a comback in 2006.... Google recently launched their "Google Trends " site which enables the user to enter a few search terms and view their corresponding frequency distributions over time. Most frequency related data for words is usually restricted to usage counts within a particular text, or throughout a set of texts without any temporal attributes factored in.

One of the nicest examples of non-temporal visualization of word frequency data is the WordCount project which utilizes machine readable text data from the British National Corpus . They have also released QueryCount which tracks how people interact with the WordCount tool using the same Flash based interface as WordCount. "Word Clouds", or "tag clouds " as they have become known, are popular and are popping up all over the place (an example can be found at Newzingo ). They are a low intensity visualization approach, essentially utilizing type-face size as an indicator of frequency. I don't find these clouds of much use, they do provide a crude approximation of "popularity", but they don't even do that very well.

Of all the sites which utilize temporal data, probably the most well known is the Baby Names Wizard's NameVoyager. It tracks the popularity of various male and female names taken from U.S. Social Security Administration data and a great tool for naming that next baby, car, boat, or pet. There is another site... but the name slips my mind... if i remember, i'll try to add it. If anyone else has any suggestions, please let me know...

Marketers love this kind of data. Sites already exist for advertisers to find the most frequently used terms within a given period as an aid to more profitable keyword inclusion on their websites. Yahoo has a unique "buzz" index tracking popular culture terms using a stock market like metaphor. Daypop (the original site doesn't seem to be up anymore, but this seems to work ) has a word "burst" feature which shows the most frequent terms in a set of popular blog postings. "Burstiness" got quite a burst itself in 2003 with this New Scientist article on the work of some Cornell University researchers (the research site can be found here ).

Back to Google Trends.... there are currently no quantities associated with the Y axis to incidate actual frequency numbers. Google states that this measure is simply "how often" the term has occurred, but with no scale, one can't estimate an actual figure. The site allows up to 5 terms, but not all terms work (not enough data to generate the graph).

For example, this graph clearly shows that one is in the lead (as i suspected), followed by two, then a neck and neck race of desparate semantic relevancy between three, four and five.

 

The poet Dan Farrell did a great piece a few years back by taking certain words and plotting their frequency (i think using data from LexisNexis). If anyone knows where these can be found on the web, please let me know, i remember seeing them once upon a time. A Google Trends version would look something like this, showing conclusivley that "spring" occurs more in Spring, then in Summer, Fall, or Winter. Who would have thought ?

 

 

 

And finally, the correlation between fear and terror is similar to that of bread and butter: