minimum sample

CWestf5155 at aol.com CWestf5155 at aol.com
Thu Mar 4 11:30:30 EST 1999


Dear Daniel,

In a message dated 3/3/99 4:00:02 PM Mountain Standard Time,
danielrr at mad.servicom.es writes:

> 	Thanks Cindy: you are very right that a sound answer to my question
>  probably depends upon the kind of grammatical analysis you apply to the
>  text. I was thinking on a syntactical analysis, and more specifically an
>  immediate constituents analysis.
>  	True, everybody uses different communicative strategies in
>  different levels of discourse, but we can admit as a principle for any
>  study in grammar that one uses one syntax and only one when writing a text
>  at a given time (sure, a novelist can introduce several characters of very
>  different social extraction, education, age, etc, each using a somewhat
>  different norm, or even different systems; and there are societies where
>  the language among men is radically different from the language used among
>  women, etc, but that's another problem).
>  	I think that when we analyse very large relatively homogeneous
>  written text corpus from an author of the past (pause for a breathe) we
>  should be able to describe the syntax the author is using with a very high
>  precision before parsing the whole corpus. Or, in other words, most of
>  *what we can know* of the syntactical system (and norm) he is using (since
>  we can *not* ask him directly to answer our linguistic texts) we can know
>  before parsing the 100% of the corpus. An example could be Ioannes
>  Chrysostomus: I never took the pain to count the words of his preserved
>  writings, but most probably they are over the million: I don't think that
>  after parsing the second five hundred thousand words your conclusions will
>  be very different from the conclusions you could draw after parsing the
>  first five hundred thousand!

Right! I was thinking so much about my own methology that I don't think that I
read your signals right.  

I'm chasing down some recommendations from a friend that works in the
computerized analysis of Hellenistic Greek, and may have some recommendations
for sources in New Testament studies  in a day or so.  As far as a general
source, I like Douglas Biber's approach to Corpus Linguistics: <<Variation
across Speech and Writing>> (Cambridge: Cambridge Univ. Pr, 1988)  and Biber,
<<Corpus Linguistics>> (Cambridge: Cambridge Univ Pr, 1998). In <<Corpus>> on
pp. 222-226, he discusses analyses of author's style, but I don't think that
it will answer your question directly.

Size of sample is an important judgment call.  I feel that analyzing more (or
all) is best, even in constituent analysis--at least there are definite
parameters on the corpus (a stopping point). 

The reason I say this, is because an individual author's style is effected by
genre and register variations (I'm sure you know what register is, but for
others, register involves the tenor [relationship between the participants],
the field [the message or content] and the mode [written, spoken, orally
transmitted then recorded, etc.].  Any changes in the genre or register effect
choices. Even if you can assume that the genre, mode and field are the same
(and as far as the field is concerned, that would be a big assumption indeed),
the differences in tenor can significantly effect not only the vocabulary, but
the most basic choices.

Cindy Westfall
PhD Student, Roehampton

 



More information about the B-Greek mailing list