Overview

We'll look at using a spreadsheet program, Microsoft Excel, and Microsoft Graph (part of MS Word).

Computing word frequencies with CONC

• Open CONC
• New...
• Options...Index (0 references)
• Build...Index
• Export Index as...
• Edit file
• Copy into Clipboard

Computing word frequencies with PERL

• Open wordcount
• Modify program (include code for output file, modify print statement)
• Run the program
• Edit file
• Copy into clipboard

Spreadsheet example: computing the Zipf constant

• Paste the two columns into Worksheet
• Determine number of rows (= number of types) using COUNT
• SORT: $B$10:$B$2000, by Rows, Descending
• Determine number of tokens (SUM over range)
• Compute Relative Frequency: F_i / N, expressed as a Percentage
• NB: Something like =B10/E1 is not robust when copied-and-pasted!
• Format Relative Frequency as a percentage
• Compute Z, Zipf's constant ( Z_i= i * f_i)
• Compute (for all Z_i)
  • Sum(Z_i)
  • N
  • Sum(Z) / N (i.e. mean)
  • AVERAGE (using Excel)
• Compute (for i = 1 to 200)
  • Sum(Z_i)
  • N
  • Sum(Z) / N (i.e. mean)
  • AVERAGE (using Excel)
• Compute X - mean(X)
• Compute (X - mean(X)) squared
• Divide by N - 1 (variance)
• Take the square root (standard deviation)
• Do the same thing with STDEV

Graphs

• Plot F_abs vs Rank (how does this variable look?)
• Plot Z_i vs. Rank as a scattergram (Is s.d. constant?)

• Plot F_abs vs Rank (how does this variable look?)
  
• Experiment with other plot types