Corpora: enquiry about reliability in factor analysis

[Marina Santini]

I'm computing factor analysis on a small corpus of 230 texts
(i put the texts as rows in a table)
using 41 linguistic feautres (the columns of the table).
Is that true that to get reliable results  the number of columns
should be the square root
or the cubic root of the number of
the number of rows?

thank a lot

marina
Umist (UK)
santinim at grisnet.it
marina.santini at stud.umist.ac.uk