Linguistic assumptions, long (Rolf, also Dave)
kimmo.huovila at helsinki.fi
Mon Jul 3 12:35:35 EDT 2000
Rolf Furuli wrote:
> I think you share my view regaring the following points:
> 1) A dead language can never be fully understood.
Yes. (The same holds to living languages, if one goes deep enough.)
> 2) No description of Hebrew verbs is the final one.
> I agree there are problems with a strict differentiation between pragmatics
> and semantics. However, which theory or model is without its problems?
> Because of this, I try to avoid very theoretic and hingflying theories and
> models, in order not to force upon Hebrew models which are foreign to it.
> Instead I try to use more down-to-the-earth models, whose parts can be
> readily understood.
> The realization that punctuality is not a semantic property is important,
> because many wrong conclusions have been drawn of the basis that it is
> semantic. The realization that durativity, dynamicity, or telicity of verbs
> marked for one of these characteristics is semantic, may also be helpful.
Just to make sure we do not misunderstand each other by using different
terms: I assume that by semantic you mean uncancelable.
> These are simple pospositions which can be falsified. And note, there is no
> problem for the proposition that it sometimes is difficult to see if a
> particular construction is punctual or durative. And further, there is no
> problem that polysemy exists and that it is sometimes difficult to be sure
> whether or not a construction is durative, dynamic, or telic. When *some*
> constructions have these properties and they are cancelable in these
> constructions, the proposition is true.
I do not quite understand you here. When some constructions are
durative, dynamic or telic, and that semantico-pragmatic property is
cancelable in a contruction, the proposition (=durativity etc are
semantic) is true?
Also, I am not sure what kind of methodology you would use to falsify
(in principle) the proposition that punctuality is not semantic
(=uncancelable?), but durativity etc. are semantic.
> I am surpised that you are uncomfortable with the tense model of Broman
> Olsen (see my words below and your comments).
> CH:>>RF Regarding tense, Broman Olsen's scheme of semantic meaning is as
> >> Past tense: Reference time comes before the deictic point.
> >> Present tense: RT coincides with C.
> >> Future tense: RT comes after C.
> >> This relationship is uncancellable, and it can be applied to Hebrew. If we
> >> find one form with a particular morphology, to be a tense, its occurrences
> >> should have a uniform pattern as to the relationship between RT and C. No
> >> such form exists, and therefore Hebrew does not have grammaticalized tenses.
> >CH:I think here we are in danger of making the same mistake Porter made (in
> >my opinion) with his analysis of Greek. Yes, we find that there is no
> >time relation with any Greek 'tense' that could not be canceled (contra
> >Olsen, as argued by Decker). But that does not justify the idea that
> >Greek has no tense. Take English, for example. There are excpetions to
> >the past meaning of the imperfect (If I was to come...). So English does
> >not have tense either (you could find exceptions to any English tense).
> >Or Finnish, even more temporal language (though without an obligatory
> >future tense). The Finnish imperfect does code past tense almost always
> >(including conditionals). But there is a future or present referring,
> >rare use, based on presupposition. If I assume that you know something
> >already, I could, in some contexts, refer to it with the imperfect even
> >if the situation is in the future (the logic being that this WAS how the
> >situation would stand - as you know it from the past information, even
> >though the situation is in the future). If
> >the assumption is not shared, the expression can result in
> >misunderstandings. So the Finnish imperfect is no past tense either? But
> >in almost any context (with some exceptions, like the one I described
> >above) the native intuition is that the form codes exactly past tense,
> >and does not allow for other interpretations.
> The model is extremely simple and is not difficult to grasp for anyone with
> a basic linguistic knowledge, and no highflying theories are necessary. A
> discussion of whether this model fits the English verbal system (or the
> Hebrew for that matter) is also quite simple. So I do not see that your
> objections have much force (if I understand you correctly).
My comments were not meant to critique Olsen's basic tense theory, but
to point out problems that occur if the principle of uniformity is takes
as exceptionlessness (doesn't she push it so far, though? I ask because
I am not sure, but this is the impression I have from other people's
reviews of her work). Also, I take no issue with her suggestion that
Hebrew does not grammaticalize time, if what is meant by this is that
there is no one to one match between tenses and temporal reference (or
even a situation very close to that). However, I do think that it may be
possible (and I would use this as a working hypothesis) that temporal
reference is grammaticalized in Hebrew to a degree, but the expression
of temporal reference relates to aspect and mood as well (as perhaps
Sorry if I was unclear as to what I meant.
> The definition of tense is quite unproblematic, namely, "a
> grammaticalization of location in time." I would go so far as to say that
> if we cannot use Broman Olsen's scheme as one of reference for the study of
> Hebrew, because some languages have some "counterexamples", linguistic
> analysis of a language would be a waste of time, and we would have
> linguistic anarchy, anything would be fine for everything.
I have no problem with this.
> Nobody denies
> that verbs with past tense can be used in situations with non-past
> reference in many languages.
But people like Porter seem to me to be saying that if there are
exceptions to temporal reference of tenses, then they are not tenses. My
point was that this is inadequate. Now, if you disagree with (my
understanding of) Porter here, my comments missed the target.
> But that fact alone does not nullify Broman
> Olsen's scheme or her claim that the relationship between the deictic point
> (C) and reference time (RT) in the different tenses can be viewed as
Now, I am not quite sure what you mean by semantic here. It seems to
have a more narrow meaning than is often used. Uncancelable? If the
claim is that there are linguistic environments in which the temporal
meaning of a verbal form cannot be canceled, then this is exactly my
> Comrie discusses this problem (Tense 1985:19,20) His conclusions regarding
> English examples is as follows: "In order to abandon the characterisation
> of the English past as indicating basically past time reference, it would
> be necessary to show that there is some alternative characterisation of its
> meaning, from which past time reference, as well as politeness (and perhaps
> present counterfactuality) would all fall out automatically as special
> cases. Suggestions that have been made in the literature strike me as
> either incorrect (if interpreted literally) or as too vague to be
> testable." You even tell that your Finnish example is a special case, and
> special cases are not problematic for the scheme.
Yes, I agree with Comrie here, and I assume that I just expressed myself
too vaguely to be understood.
> I therefore think I am on sound linguistic ground when I compare all the
> Hebrew finite verbs regarding the relationship between C and RT, and say
> that because there is no uniform pattern, tense is not grammaticalized.
If what you mean by this is that the verbal forms do not grammaticalize
tense, I agree. However, I think that tense can be grammaticalized some
other, less simple, way. For example (to take an example of Hebrew
aspect rather than tense), in my view the Hebrew verbal forms cannot
grammaticalize aspect for the same reason as what you said for tense.
There is no uniform pattern (I assume you disagree here, which, if so,
goes to prove that we understand aspects differently). That does not
mean to me that aspect is not grammaticalized in Hebrew. I assume (as a
working hypothesis; if someone finds a counter-example, I would be
grateful) that YIQTOL in past contexts grammaticalized imperfective
aspect, but not in future contexts (where it is aspectually neutral).
Thus the pattern is not quite so simple as a one-to-one correspondence.
Something similar might be true for tense. And the above hypothesis
would imply that tense has grammatical reflexes, ie. is grammaticalized
to a degree.
> am open for the view that there may be exceptions in the material, so a
> group with a high score of uniformity (such as WAYYIQTOL) can have a
> uniform relationship after all. But in that case I demand that it must be
> explained why all these forms are exceptions (the way Comrie says).
Well and good, but sometimes a good explanation may be elusive. But in
principle I agree, though I would chance 'must' to 'should'.
> Something which is a half tense is in my view nonsensical; either a verb
> form is a tense ("grammaticalization of location in time") or it is not a
Does this apply to my hypothesis of the Hebrew verb above? If I
understand you correctly, you would rule it out also as nonsensical.
Perhaps I misunderstood you.
> (BTV: Has anyone come up with an example of a non-past use of Greek
Yes. Several. Almost any reference grammar gives counterfactuals and
polite uses where the temporal reference is not past.
> The advantage of the described approach, is that something
> which allways has been assumed to be semantic (the role of WAYYIQTOLs in
> narrative), can be reviwed, and can be analyzed for pragmatic
Yes, I am not quite convinced that WAYYIQTOLs must mean consecution. I
am very open to other explanations.
> Regarding aspect (I differentiate strictly between "asepct" (the
> non-deictic relationship between event time and reference time) and
> Aktionsart (the lexical meaning and nature of verbs- durative, punctual
> etc). I think I am on the same sound linguistic ground when I compare all
> the Hebrew verbs on the basis of the relationship between event time (ET)
> and RT, and I think this gives information about Hebrew aspect.
As a proponent of a view of nested aspects, I am tempted to warn about
missing other possible layers of aspect, if you just concentrate on what
you call Aktionsart and aspect (as just two layers of what I think is
the same phenomenon).
> I welcome the use of any scientific model for the study of Hebrew verbs; we
> can learn something from them all. But we should remember that the results
> are not better than their assumptions.
I agree wholeheartedly.
> One assumption which is fundamental
> for most models, is that Hebrew has four conjugations. If it has just two,
> as I claim, it is easy to see why many of the traditional conclusions are
> wrong. I have never seen a model with so few assumptions as Broman Olsen's
> model; it has a great explanatory force, and it is simple. This is the
> reason why I use it.
For now, I am not convinced about the two conjugation system, but if you
can give enough evidence I may change my mind. However, I assume that
YIQTOLs and WAYYIQTOLs are (at least very often) aspectually opposed in
past contexts. That just makes good sense to me in light of the data.
This view of mine may predispose me to thinking of four conjugations, as
well as the fact that I do not disregard Massoretic vowels as linguistic
evidence (though I am well aware that they are from a different time).
I hope this clarifies some issues where I may have expressed myself too
vaguely. I think that the major difference between us on a general
linguistic level is that we have a different overall theory of aspect.
More information about the b-hebrew