Multiple text comparisons (for experienced users)
When a search
list has been developed which it is
considered can be applied reliably to a series of comparable texts,
this procedure allows their direct comparison simply by selecting the
file names and specifying the context
unit and other optional
settings to be applied. With a bit
of experience, this allows you to speed up the process of
comparing larger numbers of texts, without the need to examine each one
individually in detail.
As with the HAMLET Joint Frequencies
routine, statistics are first generated for individual and joint
frequencies of items in the search
list , expressed in the specified
unit of context, for each of the texts selected.
You will be prompted to supply these details when selecting one of the
analytical options:
Procrustean
Individual Differences Scaling (PINDIS) can be
applied to the series of configurations obtained by multidimensional
scaling of the matrices of joint frequencies
generated for each text,
Individual Differences
Scaling (INDSCAL) applied directly to the
collection of matrices. The MINISSA
scaling of the matrix for the first text entered above is
automatically used as the starting configuration for this
version of INDSCAL,
so that the stability of the results can be checked by varying the
order in which texts are entered.