> On Wed, Jan 20, 2010 at 10:33 AM, tayibet erzen <[email protected]>
> wrote:
> Thanks for your attention. I want to share an example of i'rab of verse
> Al-Fatihah (1-1) from Ruhu-l Kuran. We make the i'rab by dividing the words
> in the verse like in the example, that you can find as attachment.
Assalamu Alaykum Tayibet Erzen,
Welcome to the Quranic Arabic Corpus discussion list. I have looked the
attachment that you supplied previously, I hope I can provide some feedback on
your approach, as I imagine that is what you were after when you posted your
message :-) From what I can see, you have developed a draft scheme to divide
and group the words in verses of the Quran, and you have shown the first phrase
in the Quran as an example:
"bi ism allahi al-rahman al-raheem"
I've noticed that you inserted the implied pronoun "huwa" before this. This is
part of taqdeer (reconstruction) in i3raab, and there is nothing wrong with
this in general as part of traditional grammatical analysis, although there are
a few different opinions as to the reconstruction for this particular phrase in
the Quran (e.g. should the taqdeer/reconstruction use an implied pronoun or an
implied verb?)
You've also used what is usually termed something along the lines of a
"constituency phrase structure grammar". I've put your diagram into a more
easily accessible text-based form here below to show your hierarchy for this
verse:
(nominal-sentence
(mubtada huwa)
(predicate
(preposition bi)
(genitive
(antecedent ism)
(complement
(term allahi)
(adjective al-rahman)
(adjective al-raheem)
)
)
)
)
As a small suggestion, you might perhaps want to rename "mubtada" to "subject",
so as not to mix English and Arabic terms for grammar to keep things more
consistent. I've kept your original terminology intact as per the diagram, I've
just formatted things for textual display, as often shown in other tagged
corpora (e.g. the Penn Treebank) using Lisp-style s-expressions. I think that
your analysis raises a good point - should we apply constituency phrase
structure to Quranic Arabic, or use dependency grammar?
The Quranic Arabic Corpus uses dependency grammar, which is generally
considered a bit more suitable for languages with free word order. However,
using a hierarchical form of syntax for Arabic can also work (see the Penn
Arabic Treebank for example) BUT you may run into difficulties for other more
complicated verses with this approach, where things are harder to fit into this
kind of hierarchical pattern. The recent trend in tagged Arabic corpora has
been to move towards dependency grammar for Arabic (e.g. the Prague and
Columbia treebanks for Modern Standard Arabic).
You may also be interested to compare your analysis above with the dependency
approach adopted by the Quranic Arabic Corpus. For the same verse (1:1), this
is shown online in visual format here:
http://corpus.quran.com/treebank.jsp
In general, it is not unusual that when tagging is completed for a treebank,
for some automatic conversion to take place between different syntactic
representations. It may be possible to convert the final dependency grammar
used in the Quranic Arabic Corpus semi-automatically to a constituency phrase
structure approach as per your above analysis. As an example of this approach,
the Link Grammar parser does a similar conversion from dependency grammar to
phrase structure grammar as part of its syntactic analysis for English
sentences.
I would be interested to know how far you got with your approach - is this just
a proof of concept, or have you actually already annotated a large part of the
Quranic text? If so, it would be great to compare notes. Are you also using a
standard reference (i.e. an existing traditional grammar of the Quran in
Arabic) to drive your analysis? You may be interested to review some of the
reference texts listed on the Quranic Corpus bibliography page that have been
used to resolve differences of opinion for the more grammatically complex
verses:
http://corpus.quran.com/bibliography.jsp
I think your syntactic analysis looks quite interesting - keep up the good work
and let us know how you get on!
-- Kais Dukes
Language Research Group
School of Computing
University of Leeds
http://corpus.quran.com - The Quranic Arabic Corpus
[email protected] - Computional Quranic Arabic discussion list
