Online appendices for Gray, et al. Hahahahaha, Duuuuude, Yeeessss!: A two-parameter characterization of stretchable words and the dynamics of mistypings and misspellings. (pdf) (arxiv)

Online Appendix C

Balance Plots

Select a kernel to display the balance plot for that kernel. The vertical axis represents the length (number of characters) of tokens, and is broken into bins of lengths, with boundaries denoted by horizontal dashed lines, which increase in size logarithmically. For all the tokens that match the kernel and fall within a bin of lengths, the average number of times each character was stretched in those tokens was calculated, and is shown on the plot as the distance between two solid lines in the same order as in the kernel. For two letter elements, even though the letters can alternate within a given token, we still count the number of occurrences for each letter separately and display the average number of total repetitions in the same order as the letters appear in the kernel. For these balance plots, we stop plotting at the first bin with no tokens, even if later bins may be nonempty. See Sec. IIIB in the paper for more information.