Hiatt,
Mary P. "The Feminine Style: Theory and Fact." 1978. On Writing Research: The Braddock Essays 1975-1998. Ed. Lisa Ede. New York: Bedford St. Martin’s, 1999. 77-83.

Hiatt’s 1979 Braddock essay begins with a question about pattern and style in
large batches of text: Is mass style observable? Hiatt pursues this
question using empirical and computational methods (involving punch cards) to
analyze one hundred books; fifty by women and fifty by men. A 200-word sample
was pulled, at random, from each book in an effort to analyze "common
characteristics in the writing of certain groups" (77). Given that language
about style is problematic (often ambiguous, grounded by the lifespan of terms),
Hiatt sought a more reliable set of measures for isolating data from which she
might ascertain gendered stylistic differences in the sampled prose.

Hiatt proceeds with a report on the findings from her statistically-based
discourse analysis of the samples. This is an early example of
computer-based discourse analysis, and despite its literary leanings, the
project demonstrates an impressive range of analytics in light of the available
processes. Conclusion: " There is, in other words, clear evidence
of a feminine style and sound justification for the theory of group style"
(82).

"If one is attempting to discern stylistic differences between two
sets of 100,000 words each, one can, of course, try to read all these words and
note the occurrence of such stylistic matters as sentence-length and complexity,
inserts, types of modifications, and so on. One can try to do this, but
no one should. The human mind is often an inaccurate perceiver, and
errors inevitably occur. A mechanical mind is not accurate. Hence,
the only objective and accurate way to deal with such a vast amount of text is
to use a computer" (78).