POD is minimal. There is a good synopsis, and that's it. Code comments are in Portuguese. (Documentation=2)
This module seems to be focused on the author's specific problem: it requires input files to be called "1.txt", "2.txt" and so forth, and creates a seven column CSV file as output. Therefore is very difficult to reuse. (Ease of use=2)
From the synopsis:
&LATIN("4"); #3 (4-1) texts will be analised.
The interface needs refactoring. (Interface=1)
It exports a single method called LATIN (why is it UPPERCASE?). The method is exported by default, when it should have used @EXPORT_OK.
This method is 268 lines long and tries to do so much. The problem is the high complexity and deeply nested code structure.
The code use package variables.
The distribution needs more tests: the only test is if the module can be used (use_ok('Text::Statistics::Latin')).
Overall: the module is an important contribution but needs to be refactored.