Each panel represents a subject (primary school children), and displays the autocorrelation function for that subject. For some readers, response latencies at lag 20 are still correlated.

R code for complexity-based ordering as discussed in Plag and Baayen, Language, 2009 is available here. The data set for this code is available here. If code and data file are available in the current working directory, the following lines of R code produces the graphs shown below.

  library(graph)
  library(RBGL)
  library(Rgraphviz)
  source("CBO.R")

  mOrig = loadData.fnc()  
      # Figure 1
  plotmat.fnc(mOrig)                          
      # Figure 2 
  res = analysis.fnc(mOrig)                 
  m = as.matrix(res$m)
  mG = as(m, "graphNEL")
  print(mG)
  isConnected(mG)
      # figure not shown
  plot(mG)                                     
      # Figure 3
  plotAndHighlightViolations.fnc(m, mG)         

---

Figure 1: the unordered adjacency matrix



Figure 2: the same adjacency matrix for complexity-based ordering



Figure 3: the corresponding directed graph for complexity-based ordering