AbstractComputerizedword frequency data from a 104,737 word sample of Associated Press wire news are used to test Zipf's “law” of vocabulary balance, an equation predicting statistical regularity between word frequency and word ranking in written English. The wire news word frequencies are plotted in logarithmic coordinates and compared with the slope produced by a newspaper language sample of 44,000 words used by Zipf to illustrate his model. The two slopes follow roughly the predicted slope of 45 degrees, but the computerized AP news data fit the standard slope less well than the smaller sample. This suggests the need for further research using computerized frequency lists from large samples of news prose and mathematical modifications of Zipf's equation before stochastic models of language behavior are developed from his work in vocabulary bala