| write.lexstats {UCS} | R Documentation |
Creates three data files in lexstats format, which can be used
to compare and LNRE model with other models from the lexstats
package and evaluate its goodness-of-fit by a multivariate
chi-squared test (Baayen, 2001, Sec. 3.3), using the lnreChi2
program (Baayen, 2001).
write.lexstats(model, file)
model |
an object of class "zm" or "fzm",
representing a Zipf-Mandelbrot (ZM) or finite Zipf-Mandelbrot (fZM)
LNRE model. The object must include observed word frequency data
(in components N, V, and spc), usually
because the model parameters have been estimated from the observed
frequency spectrum. |
file |
a character string giving the basename of the files that will be created |
This functions creates files in lexstats format with the
extensions .spc, .sp2, and .ev2, which are
required by the lnreChi2 tool (Baayen, 2001, 270).
In addition, the basename file is extended with the string
"_bZM" (for a ZM model) or "_bfZM" (for a fZM model),
so that the lnreChi2 tool can correctly identify the number of
degrees of freedom (reduced by two estimated parameters for the ZM
model, and three estimated parameters for the fZM model).
The full basename of the created files (obtained by adding a
model-specific suffix to the file parameter).
The combination of write.lexstats and the external
lnreChi2 program to evaluate the goodness-of-fit of a LNRE
model has been
superseded by the built-in lnre.goodness.of.fit
function (in the lexstats module). This function implements
the multivariate chi-squared test as described by Baayen (2001,
Sec. 3.3) in R without relying on external software.
Baayen, R. Harald (2001). Word Frequency Distributions. Kluwer, Dordrecht.
zm, fzm, EV,
EVm, lnre.goodness.of.fit