Figure 4.

AFA of the usage frequency of the 1-gram ‘unemployment’ in the corpus of English books. The Hurst parameter, as obtained from the detrended data, is H = 1.32. (a) The variation of the usage frequency of ‘unemployment’ with time. The thin line depicts the original data, while the thick line depicts the estimated trend (using a window of length 101). (b) Detrended data, i.e. the difference between the thin and thick curves in (a). (c) Best fit to the F(w) versus w dependence for detrended data on a double log scale yields H = 1.32. (d) Best fit to the F(w) versus w dependence for original data on a double log scale yields H = 1.39.