500B words can't be wrong!

I just discovered Google Labs' Books Ngram Viewer.  It presents a fascinating view into the words and phrases occurring in a corpus of scanned books, showing how their usage frequencies have evolved over time and displaying the results graphically.  Try, for example, this query comparing references to different communications media.  I thought it was illuminating to include both "telephone" and "phone" there; while both terms persist and describe the same thing, use of the shorter one accelerated sharply as of the 1960s.