an end-case analyzer of arabic sentences
TRANSCRIPT
J.King Saud Uni"., Vo1.8. Computer & Informalion Sci., pp.21-52 (A.H 141611996)
An End-Case Analyzer of Arabic Sentences
A. D. AI-Sawadi and M. G. Khayat
*KFUPM, Dhahran, Saudi Arabia. **Dept. of Electrical & Computer Engineering
King Abdul Aziz University. Jeddah, Saudi Arabia
{Received, 06 Sept. 1993; accepted for publication. 2] Sept. 1994)
Abstract. Natural Language Understanding (NLU) has been a growing area of research in computer science. Although morphology and syntax play an essential role in NLU, end-case
analysis in some languages, like Arabic, is an important component in the correct interpretation of a sentence.
In this research, an end-case analyzer (ECA) for Arabic sentences has been designed, implemented, and integrated within a Natural Arabic Understanding System (NAUS). ECA
consists of two main components; the end-case generator, and the invoker. The end-case generator determines the end-case analysis for the given sentence and its constituents according to Arabic end-case grammar rules. which have been encoded in ECA as IF-THEN
Prolog rules and predicates. The invoker was implemented as calls to the end-case generator inserted within the syntactic analyzer of NAUS. ECA was implemented in Prolog on a personal computer with Arabic support.
The modular design of ECA permits its use within NAUS in addition to other applications such as sentence generation, speech analysis and synthesis, translation
systems, teaching Arabic, and checking and correcting grammatical errors.
Introduction
The study of the Arabic sentence structure includes two main aspects: syntax and end-case. In Arabic, the determination of the syntactic identity and end-case
21
22 A. O. AI-Sawadi and M.G. Khayat
characteristics is called nirab". When parsing a sentence, end-cases play an essential role in determining the syntactic functions of the elements (words, phrases, clauses) of the
sentence. Although morphological analysis determines morphological characteristics such as tense, affixes, gender. number, person and root, it can not determine the
syntactic role or the end-case specially when the end-case is a diacritic, which is not explicitly specified in most Arabic text. Fig.! shows two examples, the first of which
contains two occurrences of a word that have the same exact morphological characteristics; yet their syntactic roles and end-case are both different. The second example shows a sentence where the end-cases are diacritics and the sentence has two
different interpretations depending on the end-cases.The example shows the sentence with and without diacritics in both cases. End-case analysis is essential not only for
understanding Arabic sentences and correct pronunciation of words but also for speech processing, translation systems, teaching Arabic, and checking and correcting
grammatical errors.
In this research work, we present and end-case analyzer of Arabic sentences. We first present the linguistic principles of end-case in Arabic, define the problem, and
survey the existing efforts in the field. Next, we present the proposed analyzer design, which is based on the linguistic principles of end-case, and the corresponding
implementation, and demonstrate its use within a Natural Arabic Understanding System (NAUS). Finally, we conc1ude by evaluating the proposed analyzer, and considering
different applications of the end-case analyzer.
Linguistic Principles of End·Case in Arabic
Words in Arabic are classified into three classes: nouns, verbs and particles [1 - 4]. The effect of end·case rules on a word can be an explicit or implicit mark. In the explicit
case, the mark can be a diacritic, suffix. or deletion of the terminal letter of the word. In the implicit case, the mark is a diacritic which can not be uttered (thus implicit) due to
phonological rules. In addition, end-case rules not only apply to words but also to subsentences (phrases, clauses and sentences). in which case the end-case mark is implicit. The end-case of a subsentence determines its syntactic function, thus contributing to the correct interpretation of the containing sentence.
Nouns can be in one of three irab cases: nominative, accusative, or genitive. A noun can be in one of these cases according to the end-case position that it occupies in a
sentence. Tables la, lb, and lc show the possible end-case positions (syntactic function) and corresponding irab cases with examples. From the end-case point of view, however,
End
-cas
e po
sitio
n
seco
nd p
an
of po
sses
ive
Exam
ple
1 co
nstr
uct
obje
ctor
ve
rb
subj
ect o
f ve
rb
Exam
ple
2
obje
ct o
f ve
rb
Fig
. 1. E
xam
ples
ahda
r ar-
raju
lu h
agaa
tib I a
l-mus
aafe
reen
I [
~,.".l
t.,.~ .. 1.1
, ~l
ia.~
,.,J
I ~l
The
man
bro
ught
the
lugg
age
of th
e pa
ssen
gers
gaba
la a
r-ra
julu
I al-
mus
aafr
een I
I ~~WII
~~I
Jtl
l Th
e m
an m
et th
e pa
ssen
gers
rata
lar
-raj
ulu I al-
wal
ada
.J"J
I I ~ .~·1
I1 c.S~
.J ~".
,I I~
,J.JI)
c.S'.J
Th
e m
an s
aw th
e bo
y
rata
lar-
raju
la I al
-wal
adu
~ ".1'1
~ ~I
I c.S~.
J ~,.
,JI I ~ )'1
c.S'.J
Th
e bo
y sa
w th
e m
an
~ f h § >
~ !.
'<
~ ... o .... i i'r
1(l
~
S :I ~ N
W
24 A. D. AI-Sawadi and M.G. Khayat
nouns in Arabic fall into one of two categories: variable and invariable [1 - 4]. Table 2 shows the two categories, the noun types in each category, and examples of each type.
A variable noun is a noun that experiences changes in its end-case mark when its
irab case in the sentence changes [1 - 4]. The end-case mark for this type of noun differs from one noun type to another. That is, for some of these types the end-case mark is a
diacritic, while for others it is a letter. Tables 3a. 3b, and 3c show a list, with examples, of these noun types and corresponding end-case marks.
An invariable noun in Arabic is a noun that does not experience any change in its end-case mark even when its end-case position in the sentence changes [1 - 4]. The end
case mark for this type of nouns is called, lTal-bena'a" mark. This mark differs from one noun to another according to its terminal letter, but it is fiexed for the same noun
whatever its irah case is. Since the end-case mark in this case is fixed, the noun's endcase is said to be implicit.
Verbs are classified into three classes: past, present, and imperative [2 - 4J. From
another perspective, a verb in Arabic is either perfect or imperfect [2 - 4J. A perfect verb is one that must be followed by a subject of verb ("fael"), which may be explicit or
implicit. An imperfect verb is one that is followed by a nominal sentence, which consists of a subject, (flmobtada"), and a predicate ("khabar"). When a nominal sentence
is preceded by one of these verbs, the subject is renamed as the verb's subject, and the predicate is renamed as the verb's predicate. While the verb's subject retains its
nominative case, the case of the verb's predicate becomes accusative. The third and fourth examples in Tables 1 a and 1 b respecti vely demonstrate the effect of the imperfect
verb. Past and present tense Versbs are also divided into active, and passive voice. From the end-case point of view, however, verbs in Arabic fall into one of two categories:
variable and invariable [l - 41·
A variable verb is a verb that experiences changes in its end-case mark according to changes in its irab case in the sentence. This change occurs if the verb is preceded by
a verbal particle. The present tense verb, that is not suffixed by the feminine "noon" or the energetic "noon" [5] is the only verb type that comprises this category of verbs. Variable verbs can be in one of three irab cases: indicative, subjunctive, andjussive [1],
[2 - 4]. If a verb is preceded by one of the subjunctive particles, then the verb is said to be in the subjunctive case. If it is preceded by one of the jussive particles, then it is said
to be in the jussive case. Otherwise, if none of these particles precedes a verb,then the verb is said to be in the indicative case. The end-case mark of a verb depends on its case,
An End-Case Analyzer of Arabic Sentences. 25
Examples eJ.;-I1 :~I.)-C-t I a.J 6.11
Nominative :Irab Case
f'4~ I~~~ 1~
I ar-rajulur kareemu subject
I~~ I~~I -J+I">
ar-rajulu I kareemu I JR<llcate
~~ I J...,Jq c)lS ~I"'l,J c)lS r--' = 0 -kana I ar-rajUIu I kareema
.Ij ~ subject of kana
..... tI'J 'l.-I ~~I ~~I c)!
~, -I ., • ~ "..J u. ~ -U -,. .. inna ar-rajula I kareer;J
tI'J
~ predicate of inna ~
U -I
1j. I ~~]~1 ~ ~l....lL:JI~ • ~ . - ~
~ 1a ahada I kareemu I predicate of negation la "E 0
i~I !.11"JI! ~~ JAU ~
daraba I al-waladu I al·kurata subject of active voicc verb
I ~ .. ·JlI ~ JAll~L:.
subject of passive voice
gutela I m:-Wl.llJ,! I verb
[~h~ ,.u (,$JI.i.
ya I moharnmadul designated proper name vocative
[ '.'JL) ~ iJ",.-i.. .~ (,$JL:a...
ya I mugeethul designated indefinite
substantive vcx:ative
I ~~II~.,JI .. 4- t ..... .,>l~U ja'a ar-rajulu lal-Jcareemu I dependent of a nominative
Table lao Nominative noun irab cases and the corresponding end-case positions
26 A. D. AI·Sawadi and M.O. Khayat
.. i.4·1 II :~I.,rc y I t.J WI Examples •
Accusative :Irab Case I~~ l4,.,..... J ~.".w. JU- J"..L
darabahul darban Ishadeeda absolute object
I~'.J~ 4..L.6~
mashaitu WI )1r-raseefa I object of accompaniment
[ L..L;d ~la~~ ~J..t..L kharaja khalidu lsa&lha 1 adverbial object § -
[1.a,J4 ]udJ..l1 ~L( . ~ . 'a ~ '" ~ .""", rIl 'lJ kana ar.tagsu ~ object of kana 0
Il.t -~p I J.t,.,JI) u! u! r--I
Go) ~.
~ ~ inna lar-rajula I kareemu subject of inna U --Tb -s. Jj.l11 ~ (.Iao11"' ~WUJI'i ~ . - r-
~ la' ahada I fi al-manzili predicate of ncgalion Ia "E
l-fiJf ~""I :&AI ~J,JAl..a ~
akhatha a1-waladu ~kurata object of verb
f~~J ~~J~l JI.. thahaba ar-rajulu I dahika I adverb of state
I u~1 ~~ J.a.1 Ju"-' akhatha ~ha. ~L qersha
_J distinction nouns
~lrJ····Ul~~ ~ _.a.".-La ~ .~ UJll..
ya ~ a1-clmi ejtahed undesignated indefinite substantive vocative
~I L.1"s!) ~."JJ rll 4.l.t-'i J,JAl..a
gama ar-rajulu Iiki3Il13iJ lahu object of purpose
loLl.".. " Y! ,) ........... $1 i~i ..JI u
exclusive adverbial ~a al-hodour illa i!I_-wal_aWI exception noun
~q'fiJl ":'~ -.,....:..L b . ~ daraba al-lrurata ~_kalo dependent of an accusative
Table lb. Accusative noun irab cases and the corresponding eodwease positions
An End-Case Analyzer of Arabic Sentence~. 27
Examples ~I :4.! .. ,I.,ret' UWI
Genitive :Irab Case
~fi l.Jf.u ~ ~L... 4...J1 ~w.... C -0 -:l, -" ...... ....."
sahibu 8 kareemu
...... 'lJ second part of the til
~ possessive construct -0 ~.
[l.:*-u~ ~! til ~ . ~t.J~.J..J~ ~
G.~ U -kharaja iIal al-maJa'ab I object of preposition .0 1.;
&l
~ ~
~~Ivl~~' .A- l;; 0 JJ . t:! ~
ar·rajulu fi ai-haiti I al-sagheel dependent of a genitive
Table Ie. Genitive noun irab cases and the corresponding end-case positioll8
28 A. D. AI-Sawadi and M.G. Khayat
Variable ("'="~) Invariable (~)
( •..• ~ • .M.:a.l ) ,..wI r--' ( ...• ~, , ~lJl ) J".-.,tll r-' Proper names Conjunctive (Relative) pronouns
(m , .,." • ~1 ) 2 .. , ·;11 ~L.......'il ( •.••• l.a • lolA ) ',JUt' ,......,' The Five nouns Demonstrative pronouns
(oo. , ~.J • J.1,J ) .1..>1..1.'1 ,....... Y 1 ( .... ~, 'll) J.~r-' Singular nouns Conditional pronouns
( ... • ~.J • t,j 1J.1,J ) ... ( ... , vA • ".. ) .,,:at........;J1 ~
Dual nouns Personal pronouns
( ... f JjL:a... • J4-.J ) -'!tuS';, ~ ( .... .;1, ~~) Ja,iJ1,........'
Broken plural nouns Verb exaggeration & sound nouns
( ... • u,J...,ll-.. , ~~) rJL... ~ ( ... • ~1, u... ) rtf;i ,,·1 ,.......1
Sound plural nouns Interrogative nouns
( ... • ~,J~) ~...,J.I ~ .I..u..J1 f'"""1 ( .... ~ ~ ) ~~I .l.u..J1 ,......,
Non-compound numerals Compound numerals
Table 2. Arabic noun types
~l~~l t..
o~J 4i
.::1
~1J
tJ6
JI
Irab
cas
es &
End
-Cas
e M
arks
Exam
ples
tl
,rll
Nom
inat
ive
L.:..a
~.,.J I
u I
S ~.,,,;II
kana
ar.
raj.t
i]hu
na
dam
ma
u~ ~."VI
4 ••
A.! ,
al-a
wla
4iJ y
ala'a
boon
da
mm
a
~~~I
~.,,;!J
al-b~ ya
la'ab
na
dam
ma
u.".
..,.
i1l\
. ~WI
JI.,J
I al
-mu
safQ
t m
usta
'edoo
n w
aw
~'il
J.;:J
.I ~"JI
al-w
alat
¥ fi
al-
alef
.fula
r-~ ..J -
"I ~ ~
J'
.,J ,
waw
a
ill
Ill-w
alad
i'
Tab
le. 3
a. V
aria
ble
noun
s an
d th
eir n
omin
ativ
e en
d~ca
se m
arks
~~It~
Nou
n T
ype
sing
ular
(.
d ~ 4
-i,.,.,u
I r--
'i I
,).:
",$
. III ~
brok
en p
lura
l
r-IW
J ~,jl' ~
soun
d fe
min
ine
plur
al
r-IW
1 ~.u
l ~
soun
d m
ascu
line
plur
al ~I
dual
~"
, ·;'1
1 _L
........
.'il
the
Five
Nou
ns
Q.. _. s:.>
n ::I.
(. ... _. n
V\
ell ~'l
~ .c-
g .(
~t
-(.
n.
(.
!'l
8- n .C
-~
.(
it I I I !
:>
~
tTl 5- n ~ (1
; :>
~
Co> .:;;
N ~
o ...., ~ n Vl
."
::::I ;;
::::I Ii '!'
t>J
\Q
~I.>£¥I L~,J ~Jj£1J t
J6J1
lrab
cas
es &
End
-Cas
e M
arks
f
,._
-. I,
Exam
ples
. A
ccus
ativ
e
Gj.J
.sJ1
"~I ~I
l .·
i;"
thaa
kara
al-w
alad
u ad~
falb
a
B.J~
I~~I
~ 1,
··;1
1 fa
tha
nasa
ha a
r-ra
julu
al-aw~
B'~b
J' ~t
.11 e
---1
2.,
,;;t
1
asm
a'a a
l-mub
adiru
a1-
hM~
fath
a
iw' t-
.JJ.11
..:.w~
1 .. W
I . ak
:ram
ati a
l-mad
rasa
tu a
l-m
ua'
auu
Ba
ya
{3~.
,J I r
-W-' ~
.. WI ..
nasa
ha a
l-mua
'allim
u al-w~
ya
.J"JI
[Y~
J.t:...
,JI cJ
~.J
t...1J
'J1
alef
ra
'a a
r-ra
iulu
ab
J al
-wal
adi
Tab
le. 3
b. V
aria
ble
noun
s an
d th
eir a
ccus
ativ
e en
d.oe
ase
mar
ks
~'iJ t~
Nou
n T
ype
sing
ular
~J.,:a4
.s.".
lll ~",
..Jo.: .
.. (;III~
brok
en p
lura
l
r-lW
I ~j
.J.J
~
soun
d fe
min
ine
plur
al
r-'W
J ~.uJ ~
soun
d m
ascu
line
plur
al ~J
dual
2
"1
·;11
,.t...
...,.,'i
f th
e Fi
ve N
ouns
Q. _. ~
n S: (t
~ V\
~l
. ~
.c-(I
I g -(
~t
I I I
_ t:.
I n
<.
I
!l.
8- n eC
-~
(D
.(
It
~
?-' ~
~ ~
Sl> ~ [ ~ p i a
~J~1J L~,J :
;.:!
I.>£
~I t
J6JJ
Irab
cas
es &
End
-Cas
e M
arks
Exa
mpl
es
~I
Gen
itive
[i-J
..J.
I ~ .u
.,J I
. ·:;
~Il}
al
-wal
adu
fi al-~
kasr
a, f
attra
~~iJ~
;;~I
yala
'aboo
n fi
al-
mal
aa'i
. ka
sra
g')
ub
Jl ~ ~WI
,.~
i~1
ia'a
an-
naas
u fi
al-h
afI1
:Jti]
ka
sra
8-:1lI
1 ~ '~I
• WI ..
al-lr
urat
u ba
yna aI-l
ati~
na
ya
~.)L!.l
.WI
asha
ara bi~i
.. ya
.WI
.I"J~
.A ~.)
.. th
ahab
a ila
dbil a
l-w
. ya
Tab
le. 3
c. V
aria
ble
noun
s an
d th
eir g
eniti
ve e
nd..c
ase
mar
ks
~'JI t~
Nou
n T
ype
sing
ular
«..
c; I ~ '-1
oJ..,.u
I t .....
,:~ I
,J.!
... <;III~
brok
en p
lura
l
rJU
t ~j
.l.1
~
soun
d fe
min
ine
plur
al
rJU
t ..s
111 ~
soun
d m
ascu
line
plur
al ~I
dual
~
· .... ·;
11 ,
.L....
....'il
the
Five
Nou
ns
0. -. Pol n g:
( •
~ ti\
~~
~ .c-
g e[
~t
_ L .
n
<-!~
~ eL
~
.(
It
> :::I ~ ~ n ~
Eo
~ '"' o ....., ~ o· ~
:::I n :::I ~ I..;
.) -
32 A. D. Al-Sawadi and M.G. Khayat
type, and terminal letter. Tables 4a, 4b, and 4c show the possible irab cases and the corresponding end-case marks with examples. The reader will find it useful to compare
the examples in Tables 4a, 4b, and 4c to observe the effect of the subjunctive and jussive verbal particles.
An invariable verb is a verb that does not experience any change in its end-case
mark even if it is preceded by any of the verbal particles that affect the end-case of a verb. These verbs include all the past and imperative tense verbs, and the present tense verbs that are suffixed by the feminine "noon" or the energetic "noon" letter. The end
case mark for this type of verbs ( called "al-hena'a" mark) differs for the same verb according to the suffix attached.
Particles are divided into two main classes: active and inactive [2 - 4, 6, 7]. The
active particles are those that affect the case of succeeding words (nouns and verbs). The effect depends on the particle's type and the type of the affected word as well. The
inactive particles have no effect on the succeeding word. The active particles are divided into three main classes: nominal, verbal, and nominal-verbal. The nominal articles put
the succeeding nouns in the nominative, accusative, or genitive case according to the type of the particle and the end-case position of the noun. The verbal particles on]y
precede a present tense verb and affect its case. The active verbal particles are divided into two main types: particles that put present tense verbs in the subjuntive case, and those
that put present tense verbs in the jussive case. The nominal-verbal particles precede verbs and nouns, and include the conjunction ("aI-a1f') particles. These particles assign
the end-case of the preceding word to the succeeding word. Fig. 2 shows the classification of particles and the types of particles under each class. Fig. 3 shows
sample particles of different types. From the end-case point of view, all particles in Arabic are of the invariable type. That is, the particle terminal diacritic does not
experience any change wherever it occurs in a sentence.
Sentences, from the end-case point of view, are divided into two classes: sentences with end-case position, and sentences with no end-case position. Tables 5a,
and 5b show these two classes, sentence types of each class, and an example of each type.
Problem Definition
The problem is to design and implement an End-Case Analyzer (ECA) of Arabic sentences. The analyzer is to be integrated into a Natural Arabic Understanding System
~I~t I L~,J :i
.: !I~~ I
tJ W
I J,a
..i.J
I t~
Irab
cas
es &
End
.. Cas
e M
arks
V
erb
Typ
e
E
1 tl
~1
e: xa
mp
es
Indi
.
~
(a ~ _
__
__
__
__
__
__
__
__
__
_ ~~==~~nv~e~
~. ~
1=1.
V\
g 'l
rl _
I
9-.r
-.
·,.,sJ
4 .A
.I"J
I~
~I
~~I C
.:' .
. 0 t.,,
)L..A
.L1
J,...
i.ll
~-(
yal
a'aG
-waJ
adu
bi-l
kura
ti da
mm
a th
ird-le
tter-
soun
d pr
esen
t ten
se v
erb
! 't
.Jt.,..
:J I ~ r~ I ~
i.J.
w1
L....
AJ
I ~\
' 4
,.,.a.1J
I ~ t
J L..
.AJ.
J Ja
.l.J I
ya£]
an-n
ujoo
ma
fi a
n-na
haar
i im
plic
itda
mm
a th
ird-I
ette
r-al
ef p
rese
nt te
nse
ver
b
.. !.
II
II
IlJl
n...
~ •
.wl ~I
I _I
L..
• " ~ p
L..A.
LI L
~ II
a
L. .;
.:--
r..r
.!JI'
l..j""
' _
.J
.J"..
•
.r--
t,..
,) ~
~
<.
yadc
&]a
n-na
sa il
aal-
khay
r im
plic
ilda
nuna
th
ird-I
ette
r-w
aw p
rese
nt te
nse
verb
~
1= ~I.
JJJI ~ i..
ll.J
~u..JI~
i.J.u
.J.1 ~I
_~~ ~~I ~ t.,.
JL....
aJ.1
Ja.i.J
1 ~-(
ya@
] at
-talib
u w
agta
hu Ii
aI-m
adla
sali
impl
icit
dam
ma
lhird
-lette
r-ya
pre
sent
tens
e ve
rb
~ 't
~1..,g
J.1 ~ v~ J
4-.J
1 o~J ~~
l " .. ;1
1 Jt
...lV
I ar
-rija
alu ya'~ fi
al-r
nasa
ani'i
!r
eten
tion
of n
oon
the
five
verb
s
Tab
le.4
a. V
aria
ble
verb
s an
d tb
eir
indi
cativ
e en
d-ca
se m
arks
;J;>
::l ~ ~ "' )- ::l ~
'< ~ o ..... ~ (;. CI
.l I w
w
~I~t' t..
.~J :l.::I~t'
tJb
JJ
Irab
cas
es &
End
-Cas
e M
arks
Ex
ampl
es ...
~ c:,
1..1',
Jl
araa
daan
~
r~1 ~':'1 ,
!b
.,
1
yast
atee
o'u
an Ya
el an-
nujo
oma.
UMUJr[}~ ~.
."a.
khar
aja l
i-ya
do'!
) an-
nasa
'!!_I
la n~:
. "
l.al
1 ..:t-.~u
.)
araa
da a
n ya~biI-haggi
~v~~
alay
bim
an
ya'
amaE
J
.',"
SUQ
iunc
tive
t.:W
1
fath
a
i.).&
111
4.a.
llJ J
impl
icit
fath
a
t..u
J.
fatb
a
~I
fath
a
~. -
.lla
.
dele
tion
of n
oon
Tab
le.4
b. V
aria
ble
verb
s an
d th
eir s
ubju
ncti
ve e
nd-c
ase
mar
ks
J,U
.J 't
":'
Ver
b T
ype
~)#I
~.: e
'«
t..)L
....iJ
.J Ja
.1.I'
thir
d-le
ner-
soun
d pr
esen
t ten
se v
erb
~4 .,
a.)" ~ t.
JL..4
.L1
J.....
UI
thir
d-Ie
tter
-aIe
f pre
sent
ten
se v
erb
.,f.
,J4
....
l'l ~ t.
)l..A
J.1
Ja.1.
I' th
ird-
Iett
er-w
aw p
rese
nt t
ense
ver
b
_WL
• '
ll I~
--~
l...AJ
.1 L
11
1 ..
. ~
v--
""'.)
...
...-.
thir
d-le
tter
-ya
pres
ent
tens
e ve
rb
4",0
;'1 J
l.t.l*Y
1 th
e fi
ve v
erbs
e: ~.
(l
g. ti\
c.
t
~ .C
'" ~
{
l t
r ~.
~'l
e .C
" ft
0(
l t
~
~ ~
~
,:n
~ e: § Q. a: ¢ ,.:: g a
~,~tJ t.o~J ~1~tJ ~l:
J'
Irab
cas
es &
End
-Cas
e M
arks
Exa
mpl
es ..
r~1
Juss
ive
~~rJ
cJ~1
lmya
kbn[
] sk
oon
r~1 0
rJ
-~la.
1m y€l
an-o
ujoo
ma
&
t..J~
l,)M
WI
t -r
J tl.
a.J I
r-1m
ya<
fJan
-nas
a de
letio
n of
~4~rJ
wea
k le
tter
1m yag
~ bi
l-hag
gi
...
~rY
u.,.:
J1 ~l..
lmya
'am
wg
dele
tion
of no
on
Tab
le. 4
c. V
aria
ble
verb
s an
d th
eir j
ussi
ve e
nd-c
ase
mar
ks
Jaj.
Jlt~
Ver
b T
ype
."A \'
I c:
.:' I. t
..J L.
AJ.I
J,.J
.J I
third
-lette
r-so
und
pres
ent t
ense
ver
b
~'i 4
~ ~ I
J:...
.. t.J
L..aJ
.1 J,
...iJ
I th
ird-
Iette
r-al
ef p
rese
nt te
nse
verb
.J 1 ~ 4
."A
Y I
J:i.a.
.. t.J
L-aJ
.1 Ja
.i.l1
third
-lette
r-w
aw p
rese
nt te
nse
verb
_WL
·'1
1 J:
.....
t ~I J
a..U
1 _
• .,
. .J
third
-Iet
ter-
ya p
rese
nt te
nse
verb
2 'I' .;1
1 JLU~I
the
five
verb
s
e:
(I ~. J:
f. Ii\
n
'l
" ::s ~ .r
~
-(
1-1
I~
n 1.
~ .r
~
,e l r
~ ~ W
>
::I e.. "'<! a o ... i. n I VJ
V\
u..Lc~
r~ .
..J.J~
juss
ive
~
verb
al ~.
..J.
J""
aa:u
sativ
e
Fig.
2. C
lass
ific
atio
n or
arab
i<:
part
icle
s
~.JJ
=-II
part
icle
s
:3:I.i-~f
nom
inal
-ver
bal
1
L..
.L.l&
ac
tive
~I
nom
inal
~I...J.J~
",'o
U ~
.J..Jo
I2Io
F ...J
.J.,.)
A ~ ~.J~
conj
unct
ion
voca
tive
func
tiona
l pr
epos
ition
~
? q ~ t 9: [ :: p 1': ! a
An End-Case Analyzer of Arabic Sentences. 37
inactive active
verbal nominal-verbal nominal
Iau ~ kay ~ wa .J inna . I 0 ..
itha 1m r.J thonuna ~ fi . III ~ ~ ..
gad .. ..u hatta ~ aw .,,1 ya La -
Fig. 3. Sample particles
Exa
mpl
es
{~4-a .1
1"Jf
a1-w
alad
u lY
ila'ab
u bi
I-kur
at[1
1··1,. to
; I .u
.,tlt ~
than
antu
al-w
alad
a I ya
cihllkU
I (L
Jl;¥J
t ~ ~~ r~1 ~
tath
haru
an-
nujo
omu
indm
a[ta
ght#
U a-
Sham
su I
I ttl
.: ..
u( ~ v-
m
n ya
jtahe
d ra
id ya
piih
u I
1~~I
~,..
.Jt
t.,A
khar
aja
ar-r
ajul
ul wYa
<l~o:.
.::'u;
'~ __
---,
ILJMw
l ~I~.J f'~
ja'a
raju
lun I y
atleb
U an
·nas
a ]
liU'il oL
wo
JJ .,
sflo
:! ~u..
Jf
at-ta
libu
yu
u w
a I ya
njah
u I
'-:' f.
.>£ 1
J &-0
J.:.
-.. lf
J c;:
J I ~ I
Sen
tenc
es w
ith
an E
nd-C
ase
Pos
itio
n
I~WI.,JJ ~I
pred
icat
e se
nten
ce
4~~WI~I~1
obje
ct se
nten
ce
~J lJ~ W
I.,.."
~I
seco
nd p
art o
f the
pos
sess
ive
cons
truct
sen
tenc
e
r,j~
.J.~ ~f~ W
I.,J
I ~I
end-
sent
ence
of a
con
ditio
nal s
ente
nce
with
a ju
ssiv
e pa
nicl
e
~la.WI~I~1
srat
c se
nten
ce
u...WI~I~t
mod
ifier
sen
tenc
e
~I..
>£'i
f 4;
A ~ 4
J ~ ~W, ~I
depe
nden
t sen
tenc
e o
f a s
ente
nce
with
an
end-
case
pO
sitio
n
Tab
le. S
a. E
nd·c
ase
clas
sifi
cati
on o
f ara
bic
sent
ence
s w
itb
end-
case
pos
itio
n
w
00
?'
~ ~
I til ~ 9: g Q. :: p ~ {
Exa
mpl
es
fi~4~_ ..
J~~
I al-
waI
adu
yaIa
'abu
bll-
ltura
liJ
~ ."
...1 r
-Iu-~--,.1
""-',
Y!
inna
hu "l
I ....
aw--ta~i~al:-amoo--naC'11 am
run
athc
emun
1 dllJl e+
;'lu1 ~! ~.Jl
awba
ina
ilaih
i ani
esn
a'm e
l-fU
lG I 1--'·-' "
-z""i Q~tl Il
! ith
a tb
a.ak
arta
I na
;8hta
I
~l...J ~i,n4.UI.J
wa-
llahi
IiD
nl m
oham
mid
un ia
S8(b
gun
I
1 sel
J~ ~
.lH
jI~
ja'a
ala
thi I
yago
ol81
-hag
ga I
I U-.J~ ~qJ ~ .1
1.,JI
al
·wal
adu
yala
'abu
wa
laI-b
intu
tadf
Usu
I
~~~¥I V
A ~ ~ ~ ~I ~I
Sent
ence
s w
ith n
o E
nd-C
ase
Posi
tion"
WI~~I t
i...:J
1 -
. .
pred
icat
e se
nten
ce
t.G~f~1
obje
ct se
nten
ce
L.,)
.,.
,li'
l tL
...aJ
, -
-.
seco
nd p
art o
f the
pos
sess
ive
cons
truct
sen
tenc
e
r j~ ~
..L..".
JJ 4'~ W
I"J I
t4J1
en
d-se
nten
ce o
f a c
ondi
tiona
l sen
tenc
e w
ith a
juss
ive
parti
cle
~ 41~ W
I."J
I ~I
stat
e se
nten
ce
J.,..
.".. ~~ t
I...
WI"
JI u
..Il
. m
odifi
er s
ente
nce
~I.J
oC11
U--
4J J
,.a....
oJ ~ ~UJI ~I
depe
nden
t sen
tenc
e o
f a se
nten
Ce
with
an
end-
case
pos
ition
Tab
le. S
b. End~e c
lass
ific
atio
n of
arab
ic s
ente
nces
wltb
no
end-
case
pos
ltlon
>
::I ~ ~ ~ q ~ o .... ~ 0"
V) " ::I fi
::I R
"..
W
\C
40 A. D. AI-Sawadi and M.G. Khayat
(NAUS) that is under development. Given the output of the morphological and the syntactic analyzers of NAUS, ECA is to produce the end-case analysis of the sentence under consideration. The morphological analyzer determines the characteristics of the words of the sentence; the syntax analyzer detrmines (he syntactic functions (e.g., subject of verb, verb, object of verb, ... etc) of the constructs of the sentence.
The purpose of ECA is to detennine the end-case characteristics of all the constructs in the sentence being processed. In addition, ECA is to determine the irab of the sentence as a whole. and the irah of the constituent subsentences, if any. The endcase characteristics [2 - 4] include the fonowing:
a) The end-case position, which represents the syntactic function of the construct in the sentence.
b) The explicit/implicit irab case of each construct of the sentence. c) The end-case mark for each construct.
Literature Survey
With the recent trend of computer Arabization, researcH has been directed towards analysis of Arabic as a natural language, However, most of this research mainly concenlrated on the fields of morphological analysis [8 - 13] and syntactic analysis [13,
15, 16J.
The research work in the end-case analysis field is very little. In [17], a system capable of performing lexical and end-case analysis of constructs is described. This system is a lexical analyzer with the first elementary stage of an end-case analyzer. The system can only analyze isolated single constructs; it does not determine the word's endcase position in a sentence. In another effort [18], a system was based on supplying the system with the grammatical rule in some special fonn before asking the system for an analysis that depends on that rule. The system is only capable of producing the end-case positions of words of simple non-imperative sentences. It does not produce end-case analysis of whole sentences. A third effort [19] presented an end-case analyzer of simple sentences only. The system was based on determining the compatibility of pairs of consecutive words. Although the system determines the end-case position and the endcase mark of the words in the given sentence. it is restricted to the analysis of simple sentences only, and does not address end-case analysis of sentences as whole units.
An End-Case Analyzer of Arabic Sentences. 41
Design of ECA
NAUS accepts an Arabic sentence from the user, passes it through the syntactic and morphological analyzers and hence forwards it along with the syntactic and
morphological output to ECA. The input to ECA consists of:
a) The sentence as a sequence of words. b) the morphological characteristics, which specify the type of word (noun, verb or
particle), person, number, gender, tense (past, present, or imperative for verbs),
type and identity of affixes (prefix, infix, or suffix), root and derivation pattern (for nouns and verbs).
c) The syntactic structure of the sentence_ The syntactic structure includes the general syntactic functions of the constructs of the sentence, and the type of the
sentence.
The output of ECA includes the following end-case characteristics for each construct of the given sentence:
1- The type (variable and invariable) of the construct, 2-. The end-case position of the construct,
3- The irab case of the construct, 4- The end-case mark of the construct, and
5- For sentences, the implicit end-case position of the whole sentence.
When developing an end-case analyzer as a component of a natural Arabic understanding system, the task can be divided into two main functions: invocation of the
analysis within the system and performing the analysis, which is an independent activity given al1 required input. Thus, ECA is designed to consist of two main modules: the
invoker and end~case generator.
The development of the invoker involves incorporation of invocations of the required rules of the end-case generator into NAUS. Since syntax and end-case in Arabic
are highly interrelated, the structure of the syntax analyzer greatly influences that of the invoker. As the syntax analyzer recognizes sentences, the corresponding end-case rules can be invoked.
Furthermore, since the syntax analyzer of NAUS is a top-down parser, compound
structures such as noun phrases are recognized as units and must be decomposed by the invoker into words in order to invoke the corresponding end-case rules of the end-case
42 A. D. AI-Sawadi and M.G. Khayat
generator. This approach is further imposed by the fact that the irab cases for some endcase positions, e.g., subject, predicate. verb, and sentence are context-sensitive (due to
the presence or absence of active particles and imperfect verbs). In this case, the end-case analysis invocation must be placed in the module that recognizes the context in which
the construct occurs. For the context-insensitive cases, e.g., subject of verb and object of verb (tlfa'el" and mafool"), The corresponding invocations for the end-case rules of the
end-case generator can be placed in the rule (in the syntax analyzer) that recognizes the syntactic functions of the construct, thus minimizing the number of occurrences of such invocations.
The end-case generator consists of four main modules for nouns, verbs, particles
and sentences respectively. Based on the irab cases of the noun, the noun module consists of three submodules (nominative, accusative and genitive) which detennine the
irab case and end-case mark of single words in accordance with Tables 1 a, 1 b, and I c. The verb module consists of three submodules: imperative, past tense, and present tense.
The imperative verbs are of the invariable type, but are of two kinds: perfect and imperfect. Since the end-case analysis of the verb itself does not depend on the fact that a
verb is perfect or imperfect, both types of verbs are processed in the same way except for determining the verb's type. Accordingly, only one module, with a parameter for the
verb's type, suffices to process all the imperative verbs. Past tense verbs are also of the invariable type and are further c1assified into two subclasses: active and passive voice.
Accordingly, we have two modules for the past verbs. The present tense dass is analogously classified into two subclasses: active, and passive voice. From the end-case
point of view, however, the present tense class is subdivided into variable and invariable verbs. Varable verbs are further divided into three subdivisions: indicative, SUbjunctive,
and jussive. Applying this division to the two c1asses of present tense, namely, active and passive voice classes, we have six modules for variable present tense verbs. For
invariable present tense verbs, there are two modules (for active and passive voice). Fig.4 shows the design structure of the verb module. Each of the eleven verb modules
encodes the corresponding Arabic end-case grammar rules for the corresponding type of verb. The module is designed to process any kind of particle according to its identity.
The input to the module is the identity of the particle and the particle itself. The output of the module includes: the type of particle, whether it is an active or a inactive particle, and its action in the sentence if it is an active particle. The rules for end-case analysis of
sentences as whole units are encoded according to Table 5.
Fig. 5 shows the overall design of the system. Fig. 6 shows the end-case analyzer embedded into NAUS.
Ver
b m
odul
es
Past
tens
e
Pass
ive
voic
e V
aria
ble
Act
ive
voic
e Pa
ssiv
e vo
ice
Subj
unct
ive
luss
ive
Indi
cativ
e
Fig.
4. V
erb
mod
ule
desi
gn
Impe
rativ
e
Act
ive
voic
e Pa
ssiv
e vo
ice
Juss
ive
~ ~ h ~ ~ '<
~ ... o ..., >
§. n' ~
A ~ R
'" ~
End
-Cas
e A
naly
zer
• In
voca
tion
Mod
ule
Con
text
-N
on-c
onte
xt
Ver
b Se
nsiti
ve
Sens
itive
M
odul
e M
odul
e M
odul
e
Fig.
5. D
esig
n st
ruct
ure
of
EC
A
• E
nd-C
aseG
ener
ator
Nou
n Pa
rticl
e M
odul
e M
odul
e
• Se
nten
ce
Mod
ule
~
~
;t> tl ::: en ~ 9: I:>
='
0
. ~ P
E;l ~ a
Use
r Int
erfa
ce
l •
•
1,
~
Sent
ence
Gen
erat
ion
--A
-Sy
ntax
Ana
lyze
r
Invo
catio
n M
aJuI
e ... --.
--
----
End
-Cas
e G
ener
ator
1~
i In
fere
nce
I~
Fig.
6. E
CA
em
bedd
ed i
nto
NA
US
---
Mor
phol
ogic
al A
naly
zer
.....
~ ...
..... ....
1~
1 K
now
ledg
e &
Sem
antic
s I
:>
;:I ttl
;:I W
[ '<
N
(1
) .., o -,
:>
.., §.
n"
In
n ;:I n ;:I
(")
(1)
In t;
46 A. D. Al-Sawadi and M.G. Khayal
Implementation of ECA
ECA was implemented in Prolog. The incoker was implemented by inserting the appropriate calls within the syntax analyzer. The number of rules used for decomposing
compound structures is 51. The number of rulo;;!s of the end-case generator is divided as follows: 132 for nouns, 52 for verbs and 55 for particles. Fig. 7 shows some examples
of calling the end-case rules for context-sensitive and context-insensitive constructs.The predicate bfb_feil of the syntax analyzer recognizes a verba: sentence consisting of a verb, subject of verb, and object of verb which are recognized by the predicates feit, fael,
and mfol respectively. As the end-cases of the subject of verb and object of verb are context-insensitive, their corresponding end-case calls are made in the predicates fael and
mfol (not shown in the example). The predicate bfm represents a verbal sentence consisting of an imperfect verb, recognized by the predicate kan, and a nominal sentence
(consisting of a subject and a predicate) recognized by the predicate bab. End-case analysis of the imperfect verb is performed in the context (not shown) in which the predicate bfm occurs since it is a context-sensitive verb. The end-case of the subject is nominative and is recognized by the predicate asma_marfo, and that of the predicate of
the nominal sentence is accusative and is recognized by the predicate asma_mnsb. Both cases are context-sensitive and the calls are made in the context in which they occur. The
predicate afm iI1ustrates a case in which the end~case of a sentence represented by. the predicate afb or btb is context-sensitive and is determined by the predicate irab-jrnlh as
the predicate of an imperfect verb.
The syntax analyzer of NAUS divides the sentences of Arabic into eight classes that cover the whole spectrum of sentence structures. ECA was run for all classes of sentences. Appendix A shows a partial sample of the results. With respect to single words, every irab case was tested (in sentences) and the correct analysis verified. Fig. 8 shows an output sample of ECA with corresponding translation and transliteration. The first line is the sentence being analyzed. The sentence is recognized by the predicate afm shown in Fig. 7. The sentence consists of the imperfect verb "awshaka" (recognized by the predicate raja_3wshk), the subject of "awshaka", (an implicit pronoun recognized by
the predicate rnbtda), the particle "an" (recognized by the predicate "an"), and the corresponding predicate (the verbal clause "yanzaliqa fi al-hofrati" recognized by one of the predicates ath or bfb). Note that the verb of the verbal clause is required (by predicate
xtrct) to be in the present tense. As the verbal clause is parsed, end-cases of the -particle "an", the prepositional phrase 10ft al-hofrati" and the subject of verb, an implicit
pronoun, are produced as shown in the second, third, fourth and fifth lines by the corresponding recognizing predicates, since they are context-insensitive. However, note
An End-Case Analyzer of Arabic Sentences. 47
that the end-case of the present tense verb "yanzaliqa" is not perfonned within the
h ~n example of calls for context insensitive constructs
bfb_feil([LIIL21,[Noml,No.2.Nom3,Nom41,Rem) :[~feil(Ll.[Typ,s~~~l$].NDml.Lsfxl) ~].
fael(Lsfxl,L2.Nom2.Jhmn2.L3.Lsfx2). mfol(Lsfx2,L3,Nom3,Jhmn3.L4.[.S]).
X examples of calls for context sensitive constructs
bfm([LIIL2J,CNoml,Nom21,Re.) :-kan(Ll,Noml,Jhl). bab(L2,Nom2,Rem). [_,t_,Y21,[_,Z2]]=Nom2. ~s.a_marfc(S~r~ Awt$.Y2), asma_mnsb(S~'~ ~,Z2).
~fm([LIIL2].[Noml.Nom2.Nom3.Nom4].Rem) :- raJa_awshk(Ll,Noml), mbtda(L2,Nom2,Jhmn2,[L3IL41).an(L3,Nom3). xtrct(2.Nom4'[[_'[_'_'_"~J'~I_]11_]). (afb(L4,Nom4.Rem);bfb(L4,N0m4,Rem», asma_.arfc(.~'~r ~~, AwtS,Nom2), [X,[[F2,V21:_]1-Ncm4. f_mansob(F2.Y2,_), lrab_j .. lh ($ ..... t..:.o....Jr ...1L_ .... Jr .r.-A ~ ...b...e .".... S, X).
Fig. 7. Example calls of the invoker within the syntax analyzer
48 A. D. AI-Sawadi and M.G. Khayat
1 ;,~I __ ..A....J~ ...:.IJ .s..:;.,,~
2 {.&. JI.itLo...JI ....l.L...a..JI...,.......:... ..,.........:.. ...... J-'>- : ...:.'t}
3 {~-.A~:..,....}
4 {Ji.Jo"'oi o-J.L ~..Alh_H ii>~1 ~f iM>~" J"~ ...... r : J3..P.-A ....... t ;,~t} :5 {..c. ...... J ....a.:..o .."...... ~ .,...- : ...h. t-.A: ",tl}
6 {.£o ... J....b.,.a.."..a ~ ~ : .....Jo1..:....Jt ...h.~1 --I : .,J&} 7 {Ji~i 6-1L ii>..A1h...JI .ii..o...:....if .,........:...JI iI...o:llJ;.".,.."........... .£oJ' ...... ...h. ...... =..:io..J~:..LJ.j-Jo
B {.......;at...:.....JI ...h...A....lt ~..,-..:.....b..-o .."..a ~;;""":'I ".lor ~ ... } 9 {.......::.....at u-iL ...,....:....,. ~t.A ~t...:. ...h.-.A: ~,,'t: ~""}
1 {.,.. r JL t..J r ..;...., r J&..J ...!:!uo '..J ;.",.!. t ~ t 4...J..G..;>: a.......E.,JA 4.."o..J.L....iI :a.."...:;. t 3.lo"}
1 Translation: "He w.as .about to fall in the dltch" 1 Transl1teratl0n" awshaka an yanzaliqa fi al-hufratl.".
2 {an: subjuncttve verbal particle} 3 {fl: prepositional genitive particle} 4 {al-hufrati: genitIve noun, end-case mark is kasra} 5 {subJect of verb (yanzaliqa)
impliclt pronoun 1n the nom1native case} 6 {subJect of imperfect verb (awshaka)
impllclt pronoun in the nomlnatlve case} 7 {yan~aliqa: present tense verb In the subjunctlve case,
end-case mark is fatha} 8 {slmple partlcle verbal sentence :
predIcate of imperfect verb. end-case impliclt f.atha} 9 {awshaka: imperfect past tense verb, end-case mark 1S fatha} 1 {compound partlcle verbal sentence :
Independent sentence with no end-case position}
Fig. 8. A sample output
An End-Case Analyzer of Arabic Sentences. 49
predicate afb or bfb because it is context-sensitivej. Next, the predicate asma_marfo proxuces the end-case of the subject of "awshaka" (line 6). The predicate Lmansob
produces the end-case of the verb "yanzaIiqa" in line 7. Note that the case of the verb is subjunctive because it is preceded by the subjunctive particle "an". Next, the predicate
irab-Jmlh produces the end-case of the predicate clause of "awshaka" (line 8). The last two lines (produced in the predicate that invoked the predicate l?fm ) in Fig. 8 show the
end--case analysis of the verb "awshaka" and that the whole sentence has no end-case because itis independent.
The source code size of the end-case analyzer, including documentation. was about ] 30 KB. However, memory usage by the analyzer, including the ca1ls inserted within
the syntactic analyzer, was about 84 KB only. The end-case generator consists of 290 rules. In order to measure the performance of ECA and avoid the effects of other modules
of NAUS, a driver was used to run ECA separately and simulated morphological and syntactic input was used. Although the analyzer was tun in interpreter mode, the
response time was a few seconds. Given the correct morphological and syntactic input, ECA always produces the correct end-case analysis, otherwise ECA returns failure.
Conclusion
In this research we have designed and implemented an end-case analyzer (ECA) of
Arabic sentences. The end-case analyzer encodes the Arabic grammar rules of irab and the effects of applying these rules LO the constructs of sentences. The analyzer has been built
to be used as a component of a Natural Arabic Understanding System (NAUS). ECA uses the output of the morphological and syntactic analysis modules of NAUS to
detennine the end-case position and the end-case mark for each construct in the sentence. ECA consists of two main components, the end-case generator and the invoker. The end
case generator encodes the rules of irab for all constructs of the Arabic sentence. The invoker initiates the relevant end-case rule from within the syntax analyzer of NAUS.
ECA interacts with the morphological and syntax analyzers to resolve ambiguities
and determine the correct analysis of sentence. ECA rejects the morphological or syntactic input if no end-case rule applies. In this case, the morphological and syntax analyzers are recalled (through backtracking) to produce another possible analysis. The
process is repeated until an acceptable analysis is produced or the sentence is rejected. The extent to which ambguily can be resolved is not comprehensive since ambguity can
be semantic which is another research problem.
50 A. D. AI-Sawadi and M.G. Khayat
ECA can be easily adapted towards any other related system or application by only restructuring the invoker to fit within the application under consideration. As an
immediate application of ECA, it has been slightly modified to serve as an end-case synthesizer (in addition to analysis) in NAUS for sentence generation. FOT future work, the analyzer can be integrated into some other Arabic applications such as text generation, speech analysis/synthesis, systems for teaching Arabic. translation systems, and systems for checking and correcting grammatical errors in written and spoken Arabic.
An End-Case Analyzer of Arabic Sentences. 51
Appendix A
A"ojS. ...J ~ I ~ I.MII -4.jo-1 1 {"~i .s-lL aJ-ilI.ID. .. .It 4.Gooa..JI ~J-Jt A..o~", .L"""JAI --.1 : 't~: 4,.....JI} {",.p.. i uJL iii J-iI I.I(W 1 a.....A...J 1 ~.j-J f a....o lI-L", .L".... ~ --. 1 : -'to 1..-. 34", iio..iwI +'" 1 .a}
{"~'i o-J.o:. 0,.4 fJi>..J r o~r ~f ....... If,L,,, J"'~ ....... r : ~I ... ,_ : ...J~I}
{~.p i .s-JL iiI,.A I iU f G..-..J 1 {!:-'"..,..J I u)l.e." L,t.JI ~ --. 1 :: ,r.oA: A"o J.!.} {..,.rJLf...Jr ..:...0 t~....u....o r...J ~'~r ~ ~ a.."...a...t}
.9£-..l,,> -4> ....... f ..,..a ~ < L..aJ ...b..o ....... ~ ~ : ....IL I ... : sJa} {"~'i ~ aJ-il'iUf~' ~ ....... r ...... ~", .L""'~ ....... r : ....IL r ... : 4,.....J'} { 1'.&, I 1 a.J~t G..-..Jt ~"""'t ....... ~" .L,t.JI.J-'> .L.JI- ....l.r. ..... : 3.L..l,,;>: 3L.l-::V { .. H .... ...,-.:. .....b..,o ....... ~r : ~ A".....1L ... } { ......:.......J t o-J.c. ~ -=-'..0 ....IL ..... :: -:4~: -:4~} {..,.I JL f...Jf ..:...0 fJU ...b...o f...J A"a..!.t,J..:.o...,>' ~ 1~.).AI ~..a}
, ......... , • .)...J", .. H ..:.I.£.
<".p i u.-JL a..;il 1 h_H iuwL.J, ~ .,..H .u lI-L" L".... .,.. --. r : ...:. r ji" ..-. I : .)...J"..-J r } {"~i u--1L it..;8fh..Jf ~I ~r ...... lJ.L",..,."..u..a --.t : ..:.I..s. ~ t..:...Jr.) { ~t uJL ~ ~, .... .a:at..:. ....IL ..... : ..:.r.£.: ":'I.£.}
{..,.t J.L t...Jt ..;..0 t..Q..J ~ t..J ~f ~f .cL...La..;>.:: a...,....!oJAI ~ ... }
..J.j..:.A...J I ."... ~ r", ~ ....... r { ~ ... ~ :."....a}
{~,.p..'i .s-JL bA'il...Jl b.)OUJLJ' .,.;....JI 0,.0." J"'''''''''''' __ , : J".P.>A ....... , : ...J~t}
{JI~'i .s-lL it.Arh..Jr <i..cuA..Jr ~"""'f ....... ~", .L".....JA ........ r .. ~: -4> ....... '} { ...abs. "',Po ,,}
{JI~'i .s-lL .. ,.4th..Jr A..o..A..JI ~"""'f .iLo~", .L-"......,.. __ I ..a~..o: ....tt.."..iA-f}
{~ .L--J ....bA +-iI ~ Ji.,.,oM>: J!J~'" J r 40 } <..,. r JL r...J f ..;..0 f JI...J ....bA , ..J iL:aJ. , ~ t.u..o..,. ~ ~, 0.,...:. r "..,'I}
....".$0 ~,;-J 1 ...:. 1 { ~ .L"" J ...... r..J..J ....-1...:. ....IL..a..Jr..,. ........... .::..-1...:. ... ,Po : ..:.r} (".J.i:..'i o-J.c. .. ,.4 1h..J 1 a......::......r ..,........:...t ....... ll.<."..,."....:...o ....... , : ":"1 ....... r : -4> ....... r)
{.a,.p..'i o-JL a"AfJiUt iL.GOa...Jt ~"""'t a..~" .L".....JA ....... 1 : ..:.1 ~ I ........ ."s.} {.,. 1 ~ r...J t ..:...0 r JI..J....bA r...J A"a..!. f ,J..:.o...,> t ~: ............ a".......' A.,,>.:>' ".l'l}
~ JI,.:..,.-'I.J ..".;,}....I t 4 ....... r {L'" J ...b...o +-iI ...J.A:....a ~ : ....l.r. I... :: .::.}
{ .,...a..:.. ...J.::..A +-iI ~ ~ : .a.,.. ...J~"" : JI}
< ..s.~ L"'J ~ Ji-JI...wl...J ...:.",.s.......Jr u-JL ~ ~r ... ....l.r...a:-:> .. J:,a..::..". .. .i} { ..,.I,JL I...JI ..:...0 fJU ~ r...J A.Joa ~ ~ iiI+-IL..it}
<.a,.p..'i d-lL iilA'JiUI ~r ~.j-Jt ....... ~", .s..";-JAI ....... 1 :: 'I~; ~ ....... t} {.L...iro J ...b..o ~ ...:.".s...w' t.S--4 ~ ...J"..."... ..-. 1 : A..AoIt: ..,.~ t }
<",.p..'i <l-1L a..A'JiU, A...e.A...Jf ~,.,..., AAI:.lL" .s.."...JA ...a.....1L __ t : ~: ~} {..,.',JL,...It ..:...0 t~ ~ f...J ~I~I ~ A..,a.£..JA -....-f}
52
~ { .......... 'i
{Ji.P~
{"~'i {.1l ........ 'i {...,.f..)S-t..Jt
A. D. Al-Sawadi and M.G. Khayat
iI.Pit.b.Jt.:....wuf (}--"j-JI ii...olU.~ .t."...a,.r4 ....... t .. ~: ~t} iJ.Pi r.b.J r 4..a.o.J r b-iI j-J I 4A lIL" .t. ~ JA ....... I ~ I ~) "' ...... ,.b.JI o...a.....Jt ~j-J' o...o:ll.t." "'-""'J-G ........ t .. ~ A..:a.".~f} it.;iith...Jt a......oJt ~~r ...... lIL" .L.!}JJjA ........ t ~; A....L"o-a:;..}
...:.... tJlJ ....b.A I..J iJ."....!.r~f iI..J...o.,;..: ~ ...".......t} ~t"...a.II:I.L ............ "t~f: ~ ;o",...o....r}
...:.... r Jl-.J ....b.A t..J A."..!.o t ...:..;......,. r iI..J..~.s-J.L D...iot~ ....
~I ..".... ..l~ -:!~t ~..)oG...Jt ........ ~ { J----4....a? : .."....} { Ji""':' j u..JL it ~ r A...J r D ~ t ~ r 0..0 ll.<." .J!J.A-4 ....... r : J ".A-4 __ r : ..:a.".-.J t} { .t.JJ.,J ....b...o .",..a ~ ..J+-~ : ....1L t.....: .,.ti} {JiJi..i u-tL "~tA...Jt iL-a.A..Jf ~~t o..ollL" .L!J-JIJA .t..)t.AAi ....l.c...a: .}~: ~~} { ...,..r JL r..Jt ...:....co tJi..J ~ r..J iI....Iao ~ ~ A."..:.t 3..)t ~ ...... }
{ ~, u,.....lL .".....:...,.. .,,- t _-0 ....1L...a : A.L~: ..<I.L':"'}
{"~'i ...s--J.t. iJ ...... 1.b.J I iL<>atI...Jl ~,J...lt 0...0:»£" .t.~..r4 __ , : --1£. 1...iI ; ..:a."...)ML.Jr} { A~ ~ .L ...... J ...h...o ~ ...J.L l....II..Jt.3 ....l.c....a...JI ...:.... ~r ) {~""'J ~ ."....... .:...~ f ~ ~ ..J.,..,,,.... ........ r : ........ f.')AI .. ~ .".~ t}
{..,.',JLf..Jt ~ r~..i:>...o f..J D.."..1.I~f,,~ A.,...SJ-G.:.."......u...a ~'3..)"}
~ ~I.>.:. .:...t { .t.JJ..) ...L:.,...o ~ ~ ~ : ..J.L r ...... : .;:w'I}
{ ~~ t ('" ~ t ...... p" A!JhA> .t..J I ........ ....1L..... • Ja,~ t ..J.L ..... : ~ r ~: .J.!o I ~) { .t. ...... .) ...L:.,...o ~ ~ J-::I-""" : ..J.L f.-a : .;:w'l}
{ ~ si---J r ('" J.::?-J I O"'O:./.L" .... .!J..F?-A.L oJ r........ .J.L..... • Ja,,.oW..J t ...,.. r -'""" :~:~) { ...,.. I JL I.J t ...:.... r.fl...J ..L...o I..J h.;.W..J r ...,.. f ~ o...J...a.:;. ~ D.."..JL.-a} { ..:....".....1 ..... A":p'-:> ...ojl-,;- h,Jii.J ..... ~ : .:...'}
{..,.I..J.L I..Jt ...:....0 IJl-.J -L;:....o I..J A."...i.or ~I ~: ~.r> ~...iI ~I "..llt}
An End-Case Analyzer of Arabic Sentences. 53
References
[1] Abdo, D. A. "A Course in Modern Srandard Arabic." Beirut:: Almaktaba Alhadeetha, 1962.
[2} Ghalayeeni, M. Jame'a At-Dorus At-Arabia. (in Arabic), 22nd Edition, Beirut : Almaktaba Alaasriyah, 1979/1409 H.
[3) Abo-Mesleh, K. AI-Kamelfi AI-Nalww wa AI-Sarf (in Arabic, Four Volumes). 2nd edition, Beirut: The Modem Bookshop, 1984.
(4J Dahdah, A. Mo'ajam Kawa'ed A/-Lugha Ai-Arabia. (in Arabic). 2nd Edition, Beirut::Maktabat Lebanon, 1985 .
[5J Cantarino, V. Syntax 0/ Modern Arabic Prose - The Simple Sentence. volume I. Bloomington: Indiana University Press, 1974.
(6] AI-Tunaji, M. Majam A/-Ad(lwat Al-Nalw.weiah. (in Arabic), 4th edition, Damascus: Dar AI-Feker, 1968.
[7] AI-Ansary, Ibn Hesham. AI-frob an Kawa'ed Ai-frab. (in Arabic, verified by A.F. Nabi\) 1st Ed., Riyadh:Riyadh University (KSU), 198111401 H.
[8) Helal, Y. "A Morphologica1 Analysis for Arabic Speech." A Workshop about Computerized Arabic Processing, Kuwait: (April (985).
[9} Gheith, M. and EI-Sadany, T. HAn Arabic Morphological Analyzer on a Personal Computer." Proceeding of The First K.S. U. Symposium on Computer Arabiration. (April 1987).
[10] AI-Bawab, M.; Meer-alam. y. and Al-Tayyan, M. " A Computrized Morphological Arabic System." TM Region«/ Conference on Itt/ormories and Ambization, Tunis, (19&&).
[II] AI-Authman. A. "A Morphological Analyzer. for Arabci." MS Thesis, King Fahd University of Petroleum & Minerals, Dhahran, Saudi Arabia, (1989).
[12] Al-Zarroq. A." A Morphological Anlyzer for Derived Arabic Words." The Second Conference on Arabic Computational Linguistics, Kuwait, (Nov. 1989).
[131 Beesley, K."Computer Analysis of Arabic Morphology: a Two-Level approach with Detours. " Proceedings o/The Third Annual Symposium on Arabic Lingui.rtics, University of Utah, (1989).
[14] AI-Saied A. N. "The Computerized Analysis of Arabic Language Systems- Part One: The Fonnation of Arabic Grammar as a Context Free Grammar." The 1st Conference for Computers. Kuwait, (March 1989).
[15] Gheith, M. and Aboul EIa, M."A Computer Based Arabic Syntax Analyzer." The llth National Conference, Dhahran, Saudi Arabia, (March 1989).
[16] Khayat. M. G. And AI-lahri. S. K."A Model for a Syntax Analyzer of Arabic." The 12th National Conference, Riyadh, Saudi Arabia, (October 1990).
[17] AI-Bawab. M. and Merayati. M. "A Computerized Morpho-Syntax System for Arabic." The 2'ld Conference on Arabic Computational Linguistics, Kuwait, (Nov. 1989).
S4 A. D. AI-Sawadi and M.O. Kha)'at
[18] AI-Saied, A. N. and AI-Garah. S. "The Computerized Analysis of Arabic Language Systems - Part Two: An Elementary Model for a Computerized End-Case Analyzer for Non-Imperative Nominal Arabic Senstences.· The 2nd Conference on Arabic Computational Linguistics, Kuwait. (NoY. 1989).
(19) AI-Sheshaini, H. and AI-Najjar. A. II Two Computerized End-Case Analyzers for Arabic Sentences." The 2nd Conferenc on Arabic Computational Linguistics, Kuwait, (Nov. 1989).
An End-Case Analyzer of Arabic Sentences. 55
• lt~ ';I~ ..w.-. J 1.$ J 1.,-1 1 • ~.I ,~;-JI¥ ~I ~~ ',-:-"t-JI .t.....t:.AJ ~1,h.rSJ.YI r-i It J ~/.;IJiJ1 '~.liA..JIJ JJr.U..tp..!1W1 ~~
~.,.r-JI ~.r1' ~I ,.~
. ,j'J1 ,--;-"Wl ~ ~ ~\ ..:..'14- ~llf; . --= .1.11 ..:..lAUl ~ 01 • ~I ~ ~ ~L> J ~~I~~10t)'1 '4.'" .,1.11 vlAlJl ~ 4--t ~ ~ ?IJ ~ ~10i ;r ~ )~J
. 4.L..::J1 iJ\,.) -4~ ~ W4 ~t J; ~ -* 4 ~~I ~I ~r1'4tJJ1 ~ rUi; JU,l"; ~;J ~Irl Jk-~J ~ JL.a...JIIJ... ~ ~~ ~I r ~J . ..l.!,l.:.IIJ ~I : W ~J;.r.~ If ~~.r! JL-JI ~~J • I U'" Y I ~ -4:k y!J1 ~} J.r. 4jJ .u.1.,A5 ~ ~ ifl ~I .u.1.,Al W J 4JL;.,>... J ~I '-;"'Ir! ~~
~G- ~-J ' I ~ Y I iUi:J ~,.,.::ll Jb-JI J ~ ~1~~.c.....1 ~ ..w.l.:.Il..r.!~ ~ ..LiJ
.~,rU i~ :lJ~ ~ d! ~l> j~ Js- 41'/)11 j1-J1 il~1 J IUW\'\..,· ..... ..;",.l.la; J<L..eI.L .... - .. 1~..:..\...L,:..-·<' ..... I...$o'1III- .11 1 Q 0;01
• • I ~ J, t.5 "--" .. ~ ~. ~ ~ ~,.. ..
;L\a;.:.\,\~ ... ~J 4 ?I~.J i~1 ~;J J.:.L-;.J~' .. l.!.;!J!.o 1S....->)l1 v l4.,klt
. ~.,ll1'