[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

re: grammar tagging systems, why not two?

Clayton Bartholomew wrote:

>This question has been rankling for a few years without an answer. In 
>reviewing the tagging systems that are used for Greek NT, LXX and Hebrew 
>OT I have never seen one that specifically addresses the *functional* 
>level of syntax separated from morphology. Why not? 
>
>There are systems like the one used for the CCAT LXX, which are strictly 
>limited to morphology. There are systems that are primarily morphology 
>but mix in a little function like the Friberg NT. Then there are systems 
>that mix up morphology and function rather completely. 
>
>Two simple examples to illustrate the difference:
>
>Friberg Tags:   CC, CH, CS    include a second element that shows 
>function
>CCAT Tag:        C  does not show function
>Friberg Tags:   PG, PD, PA    include a second element that shows 
>function
>CCAT Tag:        P  does not show function
>
>Has anyone developed a rather full, robust *functional* tagging system 
>for the GNT which divorces it from morphology? Has it been applied to 
>the whole NT and if so is it available?
>
>I would appreciate any information you might have on this topic. Even if 
>the tagging system has not been applied to the entire NT. Just to have  
>a robust functional tagging system would be a benefit in doing analysis 
>of the NT.


I guess I should take this one on, since I'm an editor at some level for
the GRAMCORD MorphGNT (General), Westminster MorphHeb (Co-), and CCAT
MorphLXX (Associate).  I'm cross-posting this to bhebrew as well...

I wish I could give a nice simple answer; as well as one that won't upset
anyone; nevertheless, here goes...

At the outset let me say that NO Morph database is perfect or complete at
this stage of their development. They ALL have a long way to go in order to
be all that they can be in providing the researcher with tools for
research. On the other hand, their current status (aside from errors,
typos, etc.) do allow the researcher to do many, many things today, which
in past generations would have taked days or months to do...and so were in
general never done.

First and foremost, to understand the nature of the tagging schemes
currently being used you must understand that the original purpose of the
GRAMCORD and CCAT tagging schemes was both "morphology first" and "for use
in an electronic medium".  The Friberg system was constructed with neither
of those goals directly in mind. Indeed, Parunak and Whitaker used as their
guiding principle for the Hebrew "code what is written, not what is meant".
 As a consequence, the apocopated forms and the forms with the -AH suffix
are tagged as jussive and cohortative ON PURELY FORMAL GROUNDS, not on
functional grounds AT ALL (as opposed to the tagging scheme used by Todd
Beall in his _OT Parsing Guide_). Many people make the mistake of thinking
that the jussive/cohortative tags in the MorphHeb refer to the functions
not the forms.  This is certainly a limitation in the database, which will
be corrected in subsequent releases. 

On the other hand, the Friberg database was constructed for the printed
medium first of all (_Analytical GNT_) and for Wycliffe translators with
certain linguistic functions in mind. Consequently, things which work just
fine in a book don't work so well in the electronic form; eg., Participles
which the Fribergs have deemed to be Imperatival in function have a
completely separate tag from the other participles; Pronouns are subdivided
under Nouns and Adjectives rather than having their own separate tag as is
normal in grammatical discussions; Adverbs are listed under Adjectives (as
if all Adverbs derive from Adjectives); etc.  None of these points
mentioned is to say that the Friberg data is bad or wrong per se (and folks
can argue about the validity of their choices if they choose); it just has
built-in presuppositions which, IMHO, don't lend themselves well to use in
the electronic medium directly, unless the software publishers take these
approaches into account when constructing their search routines (you can
check your own program to see if they have or not; I'd also suggest you
check Harry Hahne's further discussion on the topic on his "Chorus" Web site).

With respect to Clayton's question about "functional" level tagging, as
I've tried to show above, one runs into problems when one attempts to
provide both the morph and function tagging in the same tag in the
database. Not all "function" tags are the same, and if added in the same
manner will skew the results. Two contrary examples would be (its still
fresh in my mind since I recently added these to the GRAMCORD GNT
database); Coordinating/Subordinating Conjunctions versus Nominative forms
used as Vocatives. In the first case, coordinate versus subordinate
"function" is fairly simple to determine and tag. However, in the second
case, if the Nom as Voc Nouns receive a tag which specifies that they are
Nominative in Form but Vocative in function (ie., N=Nom, G=Gen, D=Dat,
A=Acc, V=Voc, X=Nom used as Voc), it could potentially skew the results of
someone who is looking for all Nominative FORMS, since a search for N=Nom
would miss all the X=Nom used as Voc forms. What I chose to do was to add a
SECOND, Alternate/Used As tag to the GC database which tagged these as
Vocative, with the first tag--the pure morph one--being Nom. In this manner
the researcher will be able to find both. As I've wrestled with this
problem I've become increasingly convinced that the second, alternate
tagging approach is the way to go. To give another example; if we were to
tag all Jussive functioning forms (which includes apocopated as well as
Imperfect forms) as Jussive, then the researcher who is looking for
exclusively apocopated forms would not be able to find them...in fact
several people are researching this exact thing at the moment (some have
written me on this topic and said that it would be great to have the
functional tags ADDED, but that the formal tags had to be preserved). Thus
a wholesale change to, for example, Beall's system would make it impossible
for researchers who are looking to show that position in the sentence
determines whether a verb is a "volitive" not its form, to find the
information needed (those on bhebrew will recognize this discussion going
on at present among Hebrew scholars).

The other problem raised by Clayton's question/statement has to do with the
choice of the functional tags themselves. The Friberg and GRAMCORD
databases both have significant third and forth level tags for the function
of Conjunctions and Particles already; but *someone* had to decide on those
tags !  As a simple example, you will find that alot of NT Greek Profs
disagree with the Fribergs' tagging of certain Ptcs as Impv's, and you'll
find some who agree. So who's correct...it doesn't matter!! The problem is
that with only one tag (as is normal for the Conjunctions, etc.) the
researcher is faced with the decision of one person.  I think my Greek is
pretty good, but I DON'T WANT THAT RESPONSIBILITY!!  Witness the recent
discussion on bgreek about Eph 5:18; I'm absolutely convinced in my own
mind that the EN PNEUMATI is instrumental (because of the grammar with
passive verbs [or middle], the use of EN PNEUMATI in Eph, and the
meaning/use of PLHROW/PLHRWMA in Eph), but you can see that others are not
quite so convinced. Such tagging would need alternates...

I think that eventually the tagging schemes will get there, ie., with full
syntax tagging, but it should be a slow deliberate process done by experts
(I don't mean to be condescending here, but I've seen the work of grad
students...I'm fixing their naive errors all the time), so that the text
doesn't reflect and then forever fix in electronic concrete the quirks of
some transiently popular theory or theologically motivated individual.
Some folks are on my case, saying that I should go faster with the various
revisions I'm doing right now; but I see the results of such approaches
every day as I work in these texts, and I'm determined not to repeat these
errors (indeed there have been some people saying that they are going to
revise these COPYRIGHTED databases; I'd simply say--as gentlemanly as
possible--that there is only ONE OFFICIAL REVISION of these databases going
on at present, and any other work being done by others has no official
sanction or approval of the copyright holders.  Whatever changes are made
to these databases is *by contract* the property of the copyright holders
NOT those who make the changes, and cannot be released under the guise of
being an updated version of these texts, but rather must be submitted to
the editors for their review and consideration before being incorporated
into the text; but above and beyond that, such "changes" will only serve to
make a mess out of the orderly upgrading of these texts, since they will no
doubt adopt principles which are not in harmony with the direction the
editors wish to go with these texts [eg., the form versus function jussive
discussion above, which in fact is one of the things which has naively been
suggested]. Moreover, copyright law is very specific about changing
someone's else's work; can you imagine how Zondervan would respond if
someone decided to "fix" the NIV and put out their own version ?!?  How
about someone taking your book/article, making some changes in areas where
they didn't like what you said, and then releasing it under the same name
??!!.  I will endeavor to keep you all appraised of the progress on the
Morph database upgrades).

One of the interim steps in moving towards a truly Morph + Function encoded
database is now being taken for the MorphHeb by Vince DeCaen and Kirk
Lowry, namely "Syntax Encoding" --
http://www.chass.utoronto.ca:8080/~decaen/hsei/intro.html.  This material
will need to be incorporated FIRST and VERY CAREFULLY into the MorphHeb,
before we can begin to start talking about classifying the uses of the
Perfect, Waw-Consec Perf, Waw-Conjunctive Perf, Perf in poetry, etc., etc.,
etc.  Again, this is NOT a job for amateurs or grad students.

Finally, I'm mystified by Clayton's statement that he'd like to see a
"...*functional* tagging system for the GNT which divorces it from
morphology?". How does one construct a function system which is not
predicated on the forms of the words ??

I could go on, but that's probably more than most of you wanted to know
anyway...

XAIREIN...


***********************************************************************
Dale M. Wheeler, Th.D.
Research Professor in Biblical Languages        Multnomah Bible College
8435 NE Glisan Street                               Portland, OR  97220
Voice: 503-251-6416    FAX:503-254-1268     E-Mail: dalemw@teleport.com 
***********************************************************************