[B-Greek] Fwd: Imperfect and Aorist Tense-Aspects

Kimmo Huovila kimmo.huovila at helsinki.fi
Thu Sep 21 11:09:16 EDT 2006


On Thursday September 21 2006 09:17, Rolf Furuli wrote:
> To apply modern thories to dead languages are problematic, since we at the 
> outset know so little about these languages. Therefore we do not know if 
> generative grammar, cognitive grammar, Montague-grammar, and so forth can 
> give good results or whether they read into the dead languages values that 
> are not there.

If we test the approach with a good sized sample of living languages, and if 
the approach is found to be sound and good, then we should be able to apply 
it to dead languages, unless of course, we assume that dead languages are of 
a different species from living languages. I doubt that. Languages change, 
but the kinds of languages that human use don't change.

Methodology should be language-independent. If it is not, it is not a good 
general linguistic methodology (except perhaps for the study of some limited 
phenomenon or language). If the methodology is found to work irrespective of 
the living language, why would we assume it does not with dead languages? 
That's special pleading.

> I mentioned informants to 
> illustrate my point that as long as there are no living beings speaking a 
> particular language, we cannot know whether modern linguistic theories can 
> be applied to it.

That does not follow. If the theory is validated for language in general, it 
works for dead languages, too, as long as they are languages.
(KH)
<snip>
> > It is a matter of constructing the best
> > theory that explains the facts the best, including both attested and
> > unattested sentences.
(RF)
> 
> Your explanation above is to the point, but I disagree with your last 
> sentence. I do not see any purpose in predictions of what we should expect 
> to find in a dead language if this or that is true. We should only deal with 
> the corpus we have.

I was still speaking of the corpus we have. If in some situation some idea is 
expressed (in the corpus) hundreds of times and close to 100 %, and never in 
som other way, we can be pretty sure that the idiomatic way of expressing it 
is not the other, unattested idiom, but the attested idiom. If the linguistic 
theory does not explain why we never see (or see only very rarely) a verbal 
form in a certain context, the linguist's job has not yet been finished. 
That's true, by the way, regardless of whether the language is living or not.

> Further, I do not see any need for constructing a theory  
> for explanation. For example, when I analyse 997 examples of Hebrew 
> imperfect consecutive (WAYYIQTOL), there is no need for a theory to conclude 
> from this that imperfect consecutive does not represent grammaticalized past 
> tense. The crucial point is whether my analysis is correct, whether most of 
> the imperfect consecutives occur in normal contexts and we can see that 
> their actions occur after the deictic center. I do not need a theory for 
> this either, but I need to state what I view as normal conditions, in order 
> to give the reader my points of reference and an opportunity to test my 
> conclusions.

Perhaps we use the word theory differently. You need some theory of morphology 
to have a form called imperfect consecutive or WAYYIQTOL. You need a theory 
to have a meaningful definition of a grammaticalized past tense. You need a 
theory of deixis to define deictic center. And your analysis is a way of 
explaining the data. That is, it is a theory. Someone else may explain it 
with a different theory or analysis. Which one is to be preferred? One quite 
evident criterion is how well it explains the data, in this case the 
distribution of WAYYIQTOLs (where they are found and where not). Of course 
this kind of analysis goes much deeper than just the question of whether it 
is a past tense, but until it is done, the linguist's job is not finished. 
Before that you do not have a complete theory of WAYYIQTOLs. You may find 
that you need to take both tense and aspect into account to explain the 
distribution (cf. Östen Dahl's book on aspect and tense systems).
> 
> Your words indicate that we agree in the following points:
> 
> 1) The past reference of a Greek verb can indicate that the verb form itself 
> represents grammaticalized  past tense, or the past reference is caused by 
> the context, to the effect that the form itself does not  represent 
> grammaticalized past tense.
> 
> 2) In order to find out whether the aorists represent past tense or not we 
> must analyse the aorists in our corpus in order to see whether there is a 
> reasonable number of examples with non-past reference.
...that cannot be explained as systematic deviations of the rule. For example, 
the reasonable number of English sentences of type "if you went there" 
exemplify a group of exceptions, and it does not in itself prove the English 
imperfect is not a past tense. Same with Greek. 

> Since you speak of the necessity  of "a theory that explains the facts the 
> best" I am not sure if you on the basis of a reasonable number of aorists 
> with non-past reference (presuming a balanced analysis of them) without 
> reservation will conclude that the aorist does not represent grammaticalized 
> past tense. I for one will draw such a conclusion without seeking recourse 
> in any theory.
That is because of your language-theoretical commitments. In other words, 
using English as a test case of your methodology, imperfect is not a past 
tense, since there is a reasonable number of counter-examples. I think the 
same applies to Norwegian also.

Or to put it another way, you prefer a theory that concludes the aorist is not 
a past tense, if there is a reasonable number of cases with a non-past 
reference, REGARDLESS of whether there is another theory (analysis) that can 
explain the distribution of aorists better. This is a theoretical commitment 
before any analysis.
> 
> The point now is: Let us perform this analysis and see what we get instead 
> of entertaining preconceived ideas!
Neither of us will, can, or should approach the matter without preconceived 
ideas or without a theory of language to work with. Our linguistic theories 
seem to differ, and that will affect the analysis. You may be fully 
consistent within your theory of language, but if it is a bad theory, the 
results may not be any better than your theory you start with. The same is 
true, of course, for my theory. It is better to be aware of one's theoretical 
presuppositions than to deceive oneself into thinking that they are not 
there. At least then there is a possibility of testing some of the 
presuppositions if one is aware of them.

Kimmo Huovila





More information about the B-Greek mailing list