Prototype Theory and Hebrew Tense/Aspect

Matthew Anstey manstey at
Wed Dec 15 14:37:50 EST 1999

GDay all,

Too many interesting comments to go without comment:

>semantics and discourse structure. Maybe we should. He gives no
>further justifications e.g. for any extension into new fields, maybe
>his references do (I haven't read them). Anyway, if aspect is a
>semantic feature, are extensions being made? My justification would
>be, if it works, it's justified, but then I'm not a theoretical

Prototypical definitions of verb forms regarding their tense/aspect
meaning(s) are standard in linguistics, it is just that they are
normally discussed under semantics. So what you (and the Helsinki
thesis) is suggesting is not new, it is calling "syntactic" that is
novel (and I think incorrect).

>I'm interested in pursuing it; I would begin by suggesting that
>narrative sequence doesn't really qualify as a prototypical category
>because it is a discourse-level phenomenon, not a syntactic
>(clause-level) one.  I could go with past tense and perhaps
>perfectivity (though I am inclined to believe that aspect is a
>semantic feature, not a syntactic one) and would throw in realis
>mode, producing a prototype that looks something like
>[+past +perfective +realis -dependence -subordination]

I agree with sequentiality as discourse-level, and irrelevent to
wayyiqtol. Aspect and tense are naturally semantic categories, as you
say, so we are talking about a semantic prototype of wayyiqtol. The
+perfective feature I think is highly problematic. When you consider
how perfecitivity / imperfectivity works across languages, BH simply
fails to show similarities. So I personally (against the trend) would
drop this. The +realis feature is not part of the prototype, since
there is no morphological split between modal and non-modal forms in
Hebrew. All the tensed verb forms except wayyiqtol are secondary modal
uses, but they have to without any primary modal forms (ie there is no
subjunctive, optative, evidential, etc forms). Also past tense events
have a massive tendency towards realis interpretations for obvious
reasons. Your last two features traverse from semantics to clause
structure and cannot be part of a semantic prototype either. So this
leaves +past as the prototypical meaning.

>I don't know if fuzzy sets and logic have been applied to languages
>such as Hebrew, but there is no reason why it shouldn't.

I don't think it has, and having done some study in modal,
quantificational, and predicate logic I can see why. It would be
buying a lambourghini to purchase a loaf of bread. That is, it would
introduce too many distinctions and categories. It is like studies
that rely on the surface structure of a language to describe the
language. I heard of one person who had found over 900 different
syntactic surface structures for BH alone! The linguistic tools must
be appropriate for the task at hand.

>In our recurring discussions on whether or not the wayyiqtol
>encodes, to
>some degree, sequentiality, a question keeps coming up in my
>mind.  If
>the wayyiqtol *does not* include this feature, what is the source?
>After all, the normal case is that there are few, if any,
>adverbial type
>features (whether true adverbs or prepositional phrases acting as
>adverbs) to indicate the sequence of events, and there is no
>that carries this information, like the English, "then." Hence, I
rephrase my
>question: what objective markers are there in BH to show
>that events are
>sequential or non-sequential?

Great question Paul, and the answer to this is the answer to most of
our dilemmas in my opinion. Essentially there are four main ways
sequentiality is read into a text by the reader, for each I supply an

1) the way cognitive processing of narratives occurs, the principle of
minimal interpretative energy  (COGNITIVE LINGUISTICS)
2) world-knowledge  (PRAGMATICS)
3) semantic domains of words involved (SEMANTICS)
4) story-telling conventions (DISCOURSE ANALYSIS (really PRAGMATICS

Basically 1) means that we compose and process texts so that the
speaker and addresse has the least cognitive work to do. One cannot
give an 'example' of this, but one can do neuro___ tests to
demonstrate it.
Basically 2) means a shared pool of knowledge predominately about
causation. For instance, if I said "She gave birth to a boy. He was
conceived in a secret location" and "She gave birth to a boy. He was
whisked away to a secret location." you will assume that the temporal
ordering of the sentences is different for each sentence. This is
because we share a common belief that conception *causes* childbirth.
The second sentence relates to principle 1) and 4) and thus is taken
as chronological.
Basically 3) is like 2) but covers inter-linguistic associations
(rather than extra-linguistic). So if I said, "He annouced and said to
me, '...'" most people would take this as one event, not two in a row.
Or if I wrote, as poetry does, "She leaped over the hill; traversing
like a gazelle." again, most would take this as one situation, not
two. Other examples are more interesting. So in 2 Sam 12 we read, "The
poor man raised the lamb. He nurtered it." Both forms are wayyiqtol,
but who would argue they are sequential? The inter-relationship in the
lexicon of BH between the words (and the pragmatic world-knowledge)
lead us into a simulataneous interpretation. There are heaps of thses
in BH.
By 4) I am thinking of DRT (discourse representation theory) that has
a principle (creativelly called TDIP - The Discourse Interpretation
Principle!) that basically states that we assume each new predicate or
time adverbial updates the r-time. Obviously this is a hypothesis
justified either by empirical studies of stories, or by 1).

In other words, as Comrie argued over 20 years ago, languages do not
grammaticise sequentiality, it works rather by implicature, which is
why it is almost always cancellable. Furthermore, examples of all the
above types of sentences I have found in BH, all over the place,
except the clear presentation of events in reverse chronological
order, but I have not looked hard for this. Perhaps someone can find
an example such as "He fell over. John hit him." in BH for me, with
both verbs in the same form.

So the answer is there are no markers of sequentiality in BH. And
there doesn't need to be for most narratives to be read as a sequence
of events.

>It would be hazardous to attempt to apply discourse marking
>as information
>inherent in verb morphology.  ....<snip>... (once you put
>things in sequence
>the verb form becomes simple -- the only apparent exception
>being when
>giving a running commentary saying what has happened and a
>sequence is
>implied in the change of now).

>Does anyone know of a language that codes narrative
>discourse information
>into the verb rather than coding some aspect which incidentally is
>appropriate for the discourse? I know of none. I would think
>then that
>sequence is derived rather than principal to the verb form
>in this case as

I agree with your statement re discourse marking, but not with your
comments on English. But since this a BH discussion I won't get into
an argument about the English verb-system, which is notoriously

Regarding your question, this is a controversial issue among
linguists. It seems to be the case that certain languages have verb
forms whose primary "use" is to tell a narrative. (From memory I think
several are Bantu languages - perhaps Randy can enlighten us? I have a
more complete list in the room were my yaldah is sleeping!). However,
there are three issues to be resolved before thinking of wayyiqtol as
a narrative tense. Firstly, the controversy naturally is that this is
descriptive and not explanatory. That is why I say "use", since the
primary "meaning" of narrative tenses would probably be past-tense,
probably perfective. Also, other linguists question the analysis
altogether, and use the principles I explained to Paul to account for
these tenses. Secondly, this sort of description often omits
morphological and/or diachronic factors. In other words, the verbal
system of the language becomes a 'collection' of forms rather than a
'system' of forms (see Binnick 1981). A linguistic theory of tenses
however should be able more powerful in other words than simply
assigning a neat label to a verb form that is relatively fixed in its
use. For example, how do you explain the disappearance of wayyiqtol
*and* weqatal *if* wayyiqtol is a narrative tense? The drift seems to
be away from *waw-conversive*, not from just wayyiqtol.  Thirdly,
narrative tenses in languages that have them are much more severely
restricted in usage than wayyiqtol. They truly are "narrative" tenses.
That is their sole purpose in life, but wayyiqtol however has several
secondary uses.

Well, that's it for me for the week. I look forward to your responses.

With regards,
Matthew Anstey

More information about the b-hebrew mailing list