What to count (was: Hair-splitting...)
Paul S. Dixon
dixonps at juno.com
Fri Jul 16 12:10:17 EDT 1999
On Fri, 16 Jul 1999 12:53:08 +0200 Daniel =?iso-8859-1?Q?Ria=F1o?=
<danielrr at mad.servicom.es> writes:
>Paul Dixon wrote:
>>Would you not agree,
>>however, that in some cases a numbers count alone could yield
>>significant and telling information? If we are trying to determine,
>>for example, whether the aorist tense in the indicative mood is
>>increasingly assuming the functions of the perfect and pluperfect
>>in the same mood, shouldn't we expect to see this reflected in
>>the data? If not, then could you be a bit more specific as to why
>>not in this particular situation? Are you suggesting the change
>>in meanings of verbs could skew the data? If so, how?
>>To put it another way, if the data does not suggest this hypothesis,
>>then what possible basis would we have for suspecting it?
>In my opinion, there are too many variables to trust in raw numbers
>alone: some of such variables have been already pointed out: defective
>verbs, periphrastic constructions, different nuances of meaning, very
>different levels of language, etc. What I am suggesting is not that the
>quoted statistics may not be really indicative of the changes in the
>system: rather, what I suggest is that methodologically we can not trust
>such statistics before proceed to a much more detailed analysis of, at
>least, a significant part of the corpus.
Yes! I heartily agree. A raw numbers count may indicate a trend.
In fact, such a numbers count may be the basis for further scientific
analysis. The two are not mutually exclusive. Indeed, one wonders
if a raw numbers count (or at least, a suspicion of it) does not
suggest the hypothesis, then just what is the derivation of such
thinking. Which gets me back to questioning the idea that in
narrative literature the aorist is increasingly assuming the functions
of the perfect/pluperfect, if the raw data is not there to suggest it.
>There is another sense in which you can not rely on the statistics
>I gave "per se" for such an study: The volume of the corpus is very
>different from author to author, and that makes absolute numbers
>useless: you have to rely on percentages, but you can not use the chi
>test with percentages. Therefor, in order to test if the differences are
>really significative, you must calculate (roughly) the instances of
>and perfects in a corpus that is exactly the length of the NT. Moreover,
>you must study and compare the differences in the use of aorist and
>perfects in every book of the NT: if you find the differences between
>books of the NT are greater (??) or approximately the same as the
>differences between one of the NT books and, for instances, the Attic
>then you can not explain the differences only in terms of linguistic
>(I am not suggesting that such is the case: what I say is that you must
>control the validity of the experiment with such techniques).
Get the Internet just the way you want it.
Free software, free e-mail, and free Internet access for a month!
Try Juno Web: http://dl.www.juno.com/dynoget/tagj.
More information about the B-Greek