[b-hebrew] Hebrew grammar, (was Zech 6:8)

Herman Meester crazymulgogi at gmail.com
Thu Dec 1 12:37:26 EST 2005


Peter (and all others interested),
Thank you for your elaborate response to all the several issues
related to my obsession (wayyqtl) ;)
I will treat the issues in the order you come up with them in the mail
I reply to.
It took me a while to write this all down.
I hope the length won't discourage you and others. I want to thank you
all in advance for carefully weighing my arguments, and criticizing
them.

***
I'll be clear on the definite article:
Forms such as והקדש we-haqqodesh are a little problem to my theory.
Reason: if there is no original [ha-], why not וקדש waqqodesh?
All I could guess is, this is because of the semi-vowel character of
[we-]. I admit, I don't know the answer (yet).
Then again, forms such as כהיום ke-hayyom are rare. So rare, that they
in turn are i.m.o. a little problem to my opponents.
I still believe that, at least in the definite article (if we
disregard wayyqtl here) gemination is primary: it is such a nice
theory, it fits with Arabic too (see below), and it perfectly explains
why we almost always have כדבר kaddavar, לדבר laddavar, בדבר baddavar.
It also respects the general idea of gemination happening
autonomously, like in a pi'el verb, in a word like גנב gannav, and the
like. I feel gemination as a morphosyntactic marker is usually taken
for granted but never really taken seriously. It may be only about
double consonants we usually don't hear when we read the MT ourselves,
but gemination is just as "real" as a suffix, a prefix, a stress shift
or vowel change! Take a look at Arabic for instance, how very much
relevant a double consonant can be.

Talking about Arabic, I think you are the victim of my erroneous
writing on that subject.
I am very sorry, the error harms my own theory.
Here the correction: Egyptian Arabic has *more* consonants that are
not *dis*similated in the "definite article".
The process is dissimilation, and Egyptian Arabic has kept the
original geminated forms in more consonants than MSA.
Example:
iGGawaami` "the mosques"
iKKursi "the chair".
MSA would have -LG- and -LK-, respectively.
(It is not surprising that Egyptian, and not Arabian Arabic for
example, kept this older practice by the way, because Egypt is
sedentary and agrarian, which may explain why its language changes a
little slower, phonologically, than some other dialects. But it's not
my field of expertise)
Sorry for the confusion.

You quote אלתולד in Jos 15,30. That's an interesting idea indeed.
However, it can mean two things: one is that it represents what was
*heard* by the editor of the text, another is that is represents what
was *read* by the editor. I don't know how old the Arabic spelling
with ل laam is, but I can imagine the latter may be the case too: an
old Arabic spelling that crept into the Hebrew as well.
I admit, it looks like an argument against Ullendorf (who I think came
first with the idea that ال [al-] is in fact only an original
gemination). That's something I must take seriously.

You also mention that if we *don't* accept that the definite article
is actually this original gemination, we have "some inconsistencies in
assimilation rules". That's right, I would say.
Those are all resolvable, you say, "by more careful consideration of
the full phonetic context and history."
Well, I have a nice example for you showing that the tendency is often
*dissimilation* in stead of *assimilation*!
The example is in Job 3,7.
As some people believe (including one of my teachers Karel Jongeling,
who wrote the dictionary of NWSemitic inscriptions with Hoftijzer) the
text of Job is rather Eastern as far as language is concerned. For
example, it sometimes uses ל when את would be expected (before the
object), a little Aramaic flavour. Some hapax words in Job can be
looked for in the East: possibly "Aramaic" or even "Arabic" words (or
rather, a Canaanite dialect that is somewhere in the middle between
Hebrew and Aramaic, cf. Deir Alla).
Anyway, Job 3,7: גלמוד Galmud.
What is the root? The word is usually understood "hard, infertile,
stony"; we have the Arabic root جمد GMD "be solid, frozen"
(Köhler/HALOT 185, H.Wehr 158). It looks convincing that what we have
in גלמוד is dissimilation of גמוד Ga(m)mud to גלמוד Galmud.
Where I may add, the same applies to the "definite article" ال [al].
To prove my position even more, for your convenience I quote my own
little unpublished article, which I here correct btw. (most important
is the second paragraph):

"(...)
In Arabic, too, the definite articleال‬(a)l is a secondary development
out of C1 gemination of a noun. It could be argued that in Arabic, an
original definite article ال [al] (visible in writing) assimilates to
many consonants, but in fact dissimilation of the gemination consonant
(a)CC1 to (a)l-C1 is the case in a certain group of consonants.

This can be proved when we compare verbal stem VIII verbs of a root
with C1=laam, to definite nouns with C1=taa:
(i)ltaqā التقى (stem VIII), root ل ق ى where -lt- in (i)ltaqā is not
assimilated to *tt;
(a)ttarika التركة (a noun), where -tt- in (a)ttarika is not
assimilated out of *lt; on the contrary, -tt- is an original geminate,
producing the definite noun.

If there was a phonetic rule ltV > ttV in Arabic (for consonants in
initial position), one of the above examples would be impossible.
Furthermore, in Egyptian Arabic, the geminated C1 remains
undissimilated to more consonants than is the case in Standard Arabic,
which may well point to a conserved older practice.
(...)"

***
(We-yiqtol/wayyiqtol)
I totally agree with you on the phonological side of the story: we
cannot have both [we-yiqtol] AND [wayyiqtol] if we continue to believe
that both have the same "Vorlage" *wa-yaqtul(u/a/-).
That's the essence of Hatav's idea: there is "something" between w(a)-
and -yiqtol. I believe it is just the gemination, she has her
reservations, but tends to think so too.

By the way I would be very cautious regarding what you said:
"it might be worth looking at forms traditionally understood as
interrogative he + YIQTOL to see if they could be reanalysed as
HAYYIQTOL."
I've been trying similar things, but I tend to basically trust the
Masoretes, who handed down to us the Wayyqtl tense in the first place.
To reanalyse the vocalisation is something that could have no end. By
the way, I prefer to convince others of my theory on the basis of
'unamended' material only, because we all know that when you prove
something but you need to amend words in MT for this proof, there will
always be objections raised regarding your method. Then I'd rather not
use that evidence at all.

As I explained to Joel I believe, I am not convinced that something
like *Hayyiqtol is needed at all. Hebrew being "VSO", it is already
very common for verbs to open the clause, only to be preceded usually
by short adverbs or copula: such as w(a)-.
(BTW, do I have to argue here (as some people seem to require) why I
think Hebrew is VSO, and include a 500 page monograph in this mail? I
don't think so, please allow me here to use the communis opinio that
Hebrew is VSO.)
When Yiqtol, then, received its gemination: yiqtol ~> *yyiqtol, it
simply met the *w(a)- that so often opened a clause, and they merged:
~> Wayyiqtol. No need of any *hayyiqtol here.
Said differently, when Yiqtol received its gemination, ~> *yyiqtol, it
opens with a 2-consonant-cluster ~> it needs an auxiliary vowel ~>
*ayyiqtol; and see who *ayyiqtol had already seen so many times coming
before it? W-, so the match was easily made ~> Wayyiqtol.
Here too, no need of *hayyiqtol.

I have been thinking about the *hayyiqtol absence, and one scenario I
can put forward, conjectural of course, is that *(a)yyiqtol was
innovated in the Hebrew/Canaanite language at some point in time (this
must be the case of course, if proto-Semitic didn't have it already)
and very soon or immediately after that, it merged with [w(a)-] to
Wayyiqtol, and it started to be looked upon by its speakers as
something that you can't just chop the [wa-] off anymore: one tense.
This then made Wayyqtl not acceptable after a noun (in clauses with
VSO order), something I also talked about in the mail to Joel.

I do appreciate the sympathy (at least I interpret your word
"promising" that way ;) ) you seem to have toward the theory I defend
here. You seem not convinced yet, but that is not the issue: the issue
is to try to look at the model first, and see how much it "fits"
better, in comparison to older models of explanation. Then we can fill
in the details, of which I have said before, Hatav's/my theory in the
end produces less "collateral damage" than other theories that want to
explain the same phenomena.
As I predicted, once tasted, you'll start to look at the article &
wayyqtl with fresh eyes, and see how much sense it makes and how
simple it is.

***
I agree with you on Az Yashir, which in my eyes is a rather independent issue.

***
Then you talked about the apocopation etc., in other words, the
make-up of the yiqtol base. I have to say, this issue is obviously
potentially so complicated that it is beyond me. All I can say is that
it may be that for the Wayyqtl "tense", a certain yiqtol base was
preferred. But what I say, essentially, is that after Wayyqtl received
its gemination, all this didn't matter so much because the "anchoring"
event had already happened to this Yiqtol form that received
C1-gemination.
I mean, all this percentage magic, which doesn't convince anyone of
anything he didn't already believe in (is my impression), is
apparently unclear *because* it became rather irrelevant once the
C1-gemination happened.
Whether we explain the apocopation by the gemination, or we say
"apparently one preferred to apply C1-gemination to the shortest
yiqtol base available", it doesn't matter, because Wayyqtl now has its
own, specific, non-modal, indicative "aorist/passé simple"-meaning.

I think this explanation I give in the above lines is clear enough. Do
we really need 600 page dissertations on aspect and tense in the
Semitic verbal system, on the yiqtol, on the reason for the
apocopation, looking at Ugaritic, Phoenician, Akkadian, etc., before
we are allowed to say anything regarding Wayyqtl? Or do we simply say,
"what we observe is, when we look at the meaning of the Wayyqtl tense
(in prose), it clearly differs from all the modal and aspectual stuff
that all sorts of yiqtol can embody."
If we observe this, let's just describe what we see: the most
important event in the Wayyqtl *is* C1-gemination. This can not be
really satisfactorily explained by any phonological argument (I mean,
that's been tried for hundreds of years, and is there a consensus
yet?) so let's just accept that the gemination is primary.

You also wrote:
"I think we can hold that there is "some kind of principle that
wayyiqtol always has the apocopated form", at least in the third
person singular. There is certainly a high enough correlation to
suggest, if not prove, that non-jussive YIQTOL and WAYYIQTOL have
different origins."

About the character of the yiqtol inside the wayyqtl, in the analysis
of which you may well be right, I believe, again, that this may be
diachronically relevant, but synchronically it hardly is. What we have
in biblical Hebrew is largely yiqtol X, yiqtol Y, maybe yiqtol Z,
*and* wayyiqtol. The analysis of all different yiqtols that we may
discover (jussive, imperative, imperfective, and the like) are not so
relevant to the analysis of wayyqtl, because our diachronical analysis
should follow, not preceed, our synchronical analysis. In the latter
analysis, we already concluded that Wayyqtl differs from all Yiqtol
types in one or more respects.
Why then the preferred apocopation? As I said before, I can imagine,
language being economic, people being lazy ;), that if a verbal form
is getting heavier in the front due to the gemination, it might as
well lose some weight at the back. In a yiqtol verb, presence or
absence of an apocopated ending, or a short -a or -u short vowel in
older phases of Hebrew would have given the yiqtol a certain meaning,
but Hebrew lost its short end vowel anyway, and wayyqtl gets its
meaning out of the gemination. The loose yiqtol is still available in
the short jussive version and in the longer version.

All we can do then, is to conclude that BHebrew apparently had a large
enough repertoire of verb forms to convey all desired kinds of
nuances.
For those who want to suppose there was also a preterite yiqtol, I
won't forbid that, but even if this had been the case in proto-Hebrew,
proto-NWSemitic or proto-Semitic, it is not so relevant to the
biblical stage of Hebrew that we find in our MT. Why not? If the
(simple) past tense that Wayyqtl basically is, can be satisfactorily
explained by means of the primary gemination, then "Ockham's razor"
forbids us to *also* try to explain it by means of some proto-Hebrew
preterite yiqtol, which is after all a reconstruction.

One concluding remark on word order: this is best studied by writing a
syntax of only those passages in BHebrew prose that clearly represent
direct speech.

Best regards,
Herman Meester

2005/11/29, Peter Kirk <peter at qaya.org>:
> I didn't get to look at this thread until today, so I am now responding
> to many points in one message.
>
> On 26/11/2005 08:28, Herman Meester wrote:
>
> >...
> >
> >Why do I believe in this theory?
> >First, it meets a very important criterium in science:
> >"The simpler, more elegant, and more surprising theory, that can
> >account for *more* phenomena with *less* hypotheses, or only one
> >hypothesis, is to be preferred." ...
> >
> >
>
> Herman, I agree with you that this hypothesis is atttractive in many
> ways. But I do have some comments, as below.
>
> >... The problem that remains when we try to
> >explain both wayyiqtol out of *wa-yaqtul as well as wɔyiqtol out of
> >*wa-yaqtul(u), ...
> >
>
> (If you are trying to write a (phonetic) schwa here, you should use ə
> (U+0259), rather than ɔ (U+0254, open O). The latter might be the
> appropriate phonetic symbol for short qamats, on some phonetic
> reconstructions of Hebrew, but it is not suitable for Hebrew sheva.
> U+0259 also has the advantage that it is found in recent versions of
> Microsoft core fonts and so is more likely to be readable in e-mail.)
>
> >...
> >If a noun in the definite state is preceded by the prepositions כ ב or
> >ל, the noun's geminated C1, preceded by its auxiliary vowel, does not
> >need an auxiliary consonant other than כ ב or ל already there. No
> >cases of, for example *להקּדש have been found in old phases of Hebrew,
> >because no word ה existed. ...
> >
> >
>
> But words of the form והקּדשׂ WHQ.D$, i.e. as above with initial vav
> instead of lamed, do exist, e.g. the first word of Genesis 1:2. Your
> theory requires that vav can be added straight to the geminated
> consonant in WAYYIQTOL verbs. So why doesn't the same happen with
> geminated definite nouns, giving a form וָאָרֶץ WF)FREC rather than the
> attested וְהָאָרֶץ W:HF)FREC? This is certainly a complicating factor for
> the hypothesis, which makes it not quite as simple as you claim.
>
> I see that Joel spotted this issue too, although he did not analyse it
> according to your theory. Your answer to Joel confuses the issue because
> your example using MELEK is untypical, because of the rule that before
> mem the conjunction vav is pronounced as u- rather than ve-. But the
> problem is lessened by Joel's observation that forms like כְּהַיּוֹם
> K.:HAY.OWM are attested, although rare - and the suggestion that if
> forms like וָאָרֶץ WF)FREC had originally occurred they would very probably
> have been reinterpreted at some later stage as something like וְאָרֶץ W:)FREC.
>
> >...
> >Furthermore, in Egyptian Arabic, the geminated C1 remains
> >unassimilated to more consonants than is the case in Standard Arabic,
> >which may well point to a conserved older practice.
> >
> >
>
> Surely this would suggest the opposite. If Egyptian Arabic has more
> unassimilated consonants (presumably explaining the place name "Sharm
> el-Sheikh" rather than "Sharm esh-Sheikh") and is the conserved older
> practice, that would suggest that the process is assimilation rather
> than dissimilation. There is also the evidence that in written Arabic
> going as far back as the Qur'an at least the article is always written
> as ال i.e. al- or el- with an explicit L, even in cases where the
> classical pronunciation is geminated. While this spelling just could
> have been an incorrect standardisation by someone who believed that the
> variations were caused by partial assimilation rather than partial
> dissimilation, it seems more probable that it represents an actual
> spoken version of the language in which the L was always pronounced.
> There might be evidence for such a pronunciation in ancient
> transliterations. One possible example in the Hebrew Bible is the name
> Eltolad in Joshua 15:30, 19:4, which may have a preserved Arabic article
> complete with L in a context where standard Arabic assimilates, cf.
> Tolad in 1 Chronicles 4:29 which is probably the same place.
>
> All of this suggests to me that the original definite marker was not
> simply gemination, but a prefix morpheme something like al- or hal-, in
> which the final L has become completely assimilated in Hebrew and
> partially assimilated in post-Qur'anic Arabic. But I accept that this
> means some inconsistencies in assimilation rules, which may be
> resolvable by more careful consideration of the full phonetic context
> and history.
>
> In reply to Herman, Rolf wrote:
>
> >The WE- and WAY- of the so-called consecutive forms are the conjunction WAW,
> >and the gemination and patah of the WAY- prefix are caused by phonetic rules
> >and the stress position.
> >
>
> These phonetic rules would suffer from similar problems of inconsistency
> that Herman has noted concerning the Arabic article. There are clear
> cases where the distinction between WEYIQTOL and WAYYIQTOL forms cannot
> be accounted for on the basis of the stress position as heard by the
> Masoretes (which we know from the Masoretic accentuation) or any other
> purely phonological factors. This strongly suggests that there is a real
> morphological distinction, corresponding to a different derivation and
> probably a semantic difference. Herman's theory seems a very promising
> way of explaining this morphological distinction. Also, I agree with
> David K that
>
> >I don't think wayyiqtol and yiqtol are semantically similar at all.
> >
> - and if they are not semantically similar (and I don't want to get into
> a new argument with Rolf about that), that strengthens the argument for
> a morphological distinction. I note that the distinction which Rolf does
> not recognise between WEYIQTOL and WAYYIQTOL is phonetically identical
> (and on Herman's theory also morphologically identical) to that between
> indefinite LEQODE$ and definite LAQQODE$, which everyone accepts is a
> real distinction because the non-prefixed forms QODE$ and HAQQODE$ are
> also attested. The problem with WAYYIQTOL is that there is no attested
> form like HAYYIQTOL without the vav prefix. Although come to think of
> it, it might be worth looking at forms traditionally understood as
> interrogative he + YIQTOL to see if they could be reanalysed as HAYYIQTOL.
>
> Rolf also wrote:
>
> >As far as I know, it is
> >unprecedented in any language that a conjunction alone, or in combination
> >with another element prefixed to a verb form causes or signals that this
> >verb form has the very opposite meaning of the same form without the prefix.
> >
>
> But here he is attacking a straw man, for Herman and Hatav are not
> proposing that the underlying verb form is the same, rather they are
> following the admittedly unproven hypothesis that WAYYIQTOL (apocopated
> where possible, with a handful of exceptions) and YIQTOL (not
> apocopated, when not jussive) are based on two different underlying verb
> forms.
>
> Joel wrote:
>
> >YIQTOL for past-tense is not always preceded by VA-, but only VA- forces dagesh in the next letter.  For example, "AZ YASHIR MOSHE," not "AZ YYASHIR...."  It seems to me your theory wrongly predicts that the form ought to be "AZ AYYASHIR," or "AZ HAYYSHIR," or something along those lines.
> >
>
> Surely a YIQTOL in this position is imperfective i.e. marking a
> continuous or repetitive action, and so is semantically distinct from
> WAYYIQTOL. This is quite sufficient to explain the lack of the definite
> marking.
>
> Later, Herman wrote:
>
> >If we would be dealing with some kind of principle that
> >wayyiqtol always has the apocopated form, I wouldn't disregard it so
> >easily, but wayyiqtol doesn't always have the short yiqtol.
> >Apparently, while the distinction between the jussive form and the
> >long yiqtol form disappeared, C1 gemination in wayyiqtol was still
> >very relevant.
> >
>
> Rolf had written:
>
> >Regarding WAYYIQTOL I found that 98.7% of
> >3.p. s. m. were apocopated, but only 33,1% of 1. p. s., and 25,0% of 3. p.
> >pl. m.
> >
>
> But with first person jussives there is a tendency to add an extra he
> ("cohortative") which tends to cancel out any apocopation in the 1st
> person, and it seems likely that the same phenomenon was found in
> WAYYIQTOL. In the plural apocopation becomes a rather different
> phenomenon because of the suffixed vav. So, in the basic third person
> singular form we find that 98.7% of WAYYIQTOLs which can be apocopated
> are. The remaining 1.3% can include grammatical and textual errors, as
> well as cases where the Masoretes pointed WEYIQTOL as WAYYIQTOL in
> error. And perhaps there was some loss of apocopation in very late texts
> (I think I remember Rolf pointing out some in Daniel). But other than
> this, I think we can hold that there is "some kind of principle that
> wayyiqtol always has the apocopated form", at least in the third person
> singular. There is certainly a high enough correlation to suggest, if
> not prove, that non-jussive YIQTOL and WAYYIQTOL have different origins.
>
> Concerning archaisms, Herman wrote:
>
> >It's an interesting problem. Are there some notorious passages in the
> >mentioned corpus where people say it is archaic artificially?
> >
>
> The book of Esther is commonly mentioned in this regard. It is clearly
> post-exilic, but its syntax is closer to the Pentateuch and the
> Deuteronomistic History than to the other clearly post-exilic books
> which are sometimes called "late biblical Hebrew". The potential problem
> for those who date most of the historical books to before the exile is
> resolved if Esther is taken to be written in deliberately archaic style.
> Whereas for those who date the historical books after the Exile, there
> is another issue: if the distinction between standard and "late"
> biblical Hebrew is not diachronic, what is it?
>
> Concerning Phoenician, Rolf wrote:
>
> >Nobody would say that
> >this infinitive absolute has some intrinsic completedness of an intrinsic
> >past tense. The infinitive absolute simply presents the verbal idea of the
> >root without making visible the beginning or end or anything else.
> >
>
> It seems that there are a lot of assumptions here. One is that the verb
> form in question, found in presumably unvowelled inscriptions, is indeed
> an infinitive absolute and not some other form with the same consonants.
> Another is that a verb form which is etymologically an infinitive cannot
> acquire tense or aspect as a result of semantic shifts.
>
> --
> Peter Kirk
> peter at qaya.org (personal)
> peterkirk at qaya.org (work)
> http://www.qaya.org/
>
>


More information about the b-hebrew mailing list