8/2/2019 7th ICCIT-2004_113

1/6

Context-Sensitive Phrase Structure Rule for StructuralRepresentation of Bangla Natural Language Sentences.

Mohammed Moshiul Hoque

Dept. of Computer Science and EngineeringChittagong University of Engineering and Technology,Chittagong-4349, BangladeshEmail: moshiulh@yahoo.com

Muhammad Masroor Ali

Institute of Information and Communication TechnologyBangladesh University of Engineering and Technology,Dhaka-1000, BangladeshEmail: mmasroorali@iict.buet.ac.bd

ABSTRACT:Phrase-structure rules are used in context-sensitive

grammars to describe a given language /s syntax. Thisis accomplished by attempting to break language downinto its constituent parts. Phrasal categories includethe noun phrase, verb phrase, and adjective phrase;

lexical categories include noun, verb, adjective, adverband many others. This paper proposes a set of context-

sensitivephrase structure rules to parse the all kinds ofBangla natural language sentences including simple,complex and compound sentences. The proposed rulescan parse the all types of Bangla sentences. This paperalso describes a technique to decompose a complexsentence into a dependent and independent clause and

a compound sentence into a simple sentencerespectively. The inflection of Bangla verb phrasecalledauxiliary can have different forms depending onthe tense, the person and the class of the subject of theverb. In this paper, we also presented a decomposition

procedure for Bangla verb phrase into severalsubparts and then extracting necessary informationfrom the auxiliary part.

Keywords: Phrase structure, SR, Parsing,Decomposition, Clause.

INTRODUCTION:Natural language processing (NLP) are developed bothto explore general theories of human language

processing tasks such as providing natural languageinterfaces or front ends to application system. Thesubject of NLP is an immense field with many

potential applications, including translation from one

language into another, retrieval of information fromdatabases, human/computer interaction, and automaticdictation. A language understanding program musthave considerable knowledge about structure of thelanguage including what words are and how theycombine into phrases and sentences. It must also knowthe meaning of the words and how they contribute tothe meaning of the sentence and to the context withinwhich they are being used. In this paper, we represent aset of context-sensitive grammar (CSG) rules to parsethe all kinds of Bangla sentences. To interpret anylanguage requires determining the sentence structure ofthat language. Phrase structure (PS) grammars define

how the basic components of symbol strings, thesymbols themselves, can be aggregated into phrases,

and how these phrases can themselves be aggregatedfinally into sentences. A grammar captures the legalstructure in a language and thus allows a sentence to beanalyzed [2]. By using the Context-free grammar(CFG) rules, it cannot be possible to generate all thesentence patterns in Bangla languages. To generate allthe sentences of Bangla, we have to use CSGs. TheCSG PS rules says that how parts of speech can puttogether to make grammatical sentences.A parser module can used the CSGs phrase-structurerules and also can find out how words in sentencerelated to each other. Therefore, parsing a sentence byusing a set of PS rules produces underlying structurefor sentence. The result is usually a tree or structuralrepresentation (SR) that can be used as input for aMachine Translation Engine. Machine translation (MT)[1] [2] is an emerging paradigm for processing naturallanguages. Now a day, MT and NLP are being used bylanguage industries for translation one language toanother and allow people to interact with computers ina natural human language that will save a lot of timeand cost of money.

1.1 Previous work and Contribution of thisPaper:Parsing of Bangla sentences by using CSGs rules is inrudimentary stage still now. Very few research workhave been conducted regarding parsing of Banglasentences but a significant number of researchactivities have been conducted on the recognition ofBangla characters. Among them [4] describe the

parsing of Bangla sentences for simple sentence, while[5] describes detail analysis of Bangla phrases and

different types of sentences by using theTransformational Generative Grammars (TGGs).Syntax analysis for Bangla sentences is presented in[7]. Besides these, [6] focuses on designing a Banglaconversion processor using natural language

processing. Parsing methodology for Bangla simplesentences by using context-free grammars (CFGs) is

presented in [8]. This paper is an extension to schemepresented in [8]. The extensions and refinementspresented in this paper over previous work are asfollows: A set of context-sensitive PS rules for simple

sentence.

A sets of context-sensitive PS rules forcomplex sentences.

8/2/2019 7th ICCIT-2004_113

2/6

A sets of context-sensitive PS rules forcompound sentences. Decomposition procedures for auxiliary ofverb phrase. Decomposition technique for complex andcompound sentences into simple sentence and

parse these sentences as simple sentences. SR has been used to represent the complexand compound sentences respectively.

2. CONTEXT-SENSITIVE GRAMMAR (CSGs)FOR BANGLA:Contex sensitive grammars (CSGs) are the collectionof a set of PS rules in which, at least one PS rule iscontext- sensitive [3]. Applications of PS rules arerestrictive by the context that is if certain conditionsregarding the context are fulfilled then they will beactivated. In CSGs, auxiliaries are depended onsubjects. The category of person + class for both the

auxiliary and subject should be the same. Now,according to the above discussion, we can represent aset of context sensitive phrase grammar rules forsimple sentence in table 1. Form the table 1, we findthe structure of the verb phrase (VP) and verb form(VF).3. STRUCURE OF BANGLA VERB PHRASE(VP):Every sentence in Bangla must have a verb phrase. Thecompulsory part of a VP is the verb Form (VF).The original part may contain a NP or an AP or a NPand AP. As a rule VP can be expressed as follows-VP (NP) (AP) VF

3.1. Verb Form (VF):Verb Form is the most important word in a sentence. Itis the word used for stating something about a personor thing. The Bangla verb form can be segmented intotwo parts: These are- i) Verb Root (VR) ii) Auxiliary(AUX)

4. AGREEMENT:In every language, there is an agreement between thesubject and verb form. Similarly, there is also anagreement of person + class between verb form andsubject in Bangla. The auxiliary changes for tense,aspect and person + class. For example-Tini Boi Porchen(ae hC fs-Re). Here, --------------

VP Boi Porchen (eB coQb)VF Porchen (coQb)VR Por (co)AUX Chen (Qb)AUX Aspect + Tense + ConcordChe = Ch + + en

= Continuous + Present + Concord of Third PersonHonorific for Present Tense

That means, in Bangla, the person + class of AUXfollows the person + class of subject NP. For example,if subject is Apani( Bfe) then person + class ofTable 1. CSGs PS rules for Simple Sentence of Bangla

Natural language.

RuleNo.

CSGs Rule

01.02.03.04.05.

06.07.08.09.10.11.12.13.14.15.16.17.18.19.

20.21.22.23.24.25.26.27.28.29.30.31.

SS NP VPVP (NP) (AP) VFNP NNP N (DET)NP NP (Biv) (NP)

NP (DEMO) (SPR) (AP) NNP N PMN PERAP AD/ADSDEMO (DD) (DO)SPR QFR (PP)VF VR AUXPER FP / SP/ TPFP ami / amara (Avwg/Avgiv)SP SPH / SPNH / SPPSPH apani / apanara (Avcwb/Avcbviv)SPNH tumi/tomara (Zzwg /Zvgiv)SPP tui/tora (ZzB/Zviv)TP TPH / TPNH

TPH tini/tara/shikak (wZwb/Zviv/wkK)TPNH se/tara/pakhi/boi (m/Zviv/cvwL)DET ti (wU)QFR ek/dui/bahu/anek (GK/`yB../e/AbK)PP ti/ta/khana/jon (wU/Uv /Lvbv/ Rb)AD khub/bhalo/sundor (Lye/fvj/ my`i)PM era/ra/der/samuha (Giv/iv/`i/mg~n)DD ai/sei/oi (GB/mB/H)DO prothom/ ditia (c_g/wZxq.)Biv ke/der/re/kar (K/`i/i/Kvi)VR likh/por/chal/ja (wjL/co/Pj/ hv)AUX e/che/bhe/bo/chhen (G/Q/e/e/Qb)

auxiliary should be en( He) or for Tumi(aj), it

should be o(A) or e(H) are accepted and so on.Such as- Apani Boi Porchen (Bfe hC fs-Re) [Apani(Bfe)+ Boi (hC) + Por (fs) + chh (R) + en (He)]. Butit is not possible to expressed Apani BoiPorcho(Bfe hC fs-R) [Apani (Bfe) + Boi (hC) +Por (fs) + chh (R) + o (A)], Because o(A) does notagree with the subject Apani(Bfe). Now, we will tryto shown in table 2, 3, 4, 5 how the auxiliary part ofverb form (VF) can relate or agree with the various

person and classes of the subject and with the varioustenses of the verb.

4.1. Subcategorization and updating the

Grammar Rules:Subcategorization expresses the constraints that apredicate (verb for now) places on the number and typeof the argument it wants to take. Therefore, fromtheabove analysis, we can concluded that the inflectionof Bangla verb called AUX can have different formsand hence subcategorized depending on the tense, the

person and the class of the verb. The forms aredescribed in table 2, 3, 4 and 5. After a thoroughobservation of table 2, 3, 4 and 5, we proposed a set ofupdated Context-sensitive grammar rules to furtherdecompose and extract information from the verbofasentence, with the following PS rules of table 6.

8/2/2019 7th ICCIT-2004_113

3/6

Table 2: Agreement of Person + class with the verb auxiliary and subject for Past Tense:Person and Class

FP SPH SPNH SPP TPH TPNHSubject ami/amara

(Avwg/Avgiv)apani/apanara(Avcwb/Avcbviv)

tumi/tomara(Zzwg/Zvgiv)

tui/tora(ZzB/Zviv)

tini/tara(wZwb/Zviv)

se/tara(m/Zviv)

IndefiniteIam =l o+am(jvg=j+Avg)

kor+lam= korlam (Kijvg)

len = lo + en(jb=j+Gb)

kor + len= korlen (Kijb)

le = lo + e(jb=j+G)

kor + le= korle (Kij)

li = lo + i(wj=j+B)

kor + li += korli (Kiwj)

len = 1o + en(jb=j+Gb)

kor + len =korlen(Kijb)

lo (j)kor + lo

= korlo(Kijv)

Continuous

chholam = chh + i+ lo + am (wQjvg= Q+B+j+Avg)kor + hhi- alm =korchhialm(KiwQjvg)

chhilen = chh + i+ lo + en(wQjb =Q+B+j+Gb)kor + chhilen =korchhilen(KiwQjb)

chhile = chh + i+ lo + ewQj =Q+B=j+Gkor + chhile =korchhile(KiwQj)

chhili = chh + i+ lo + i(wQwj=Q+B+j+B)kor + chhili =korchhili(KiwQwj)

chhilen = chh+ i + lo + en(wQjb=Q+B+j+Gb)kor + chhilen= korchhilen(KiwQjb)

chilo = chh + i+ lo(wQj=Q+B+j+A)kor + chilo= korchilo(KiwQj)

Perfect

echhilam =e + chh+ i + lo + am(GwQjvg=G+Q+B+j + Avg)

kor + chhilam= korechhilam(KiwQjvg)

echhilen = e + chh+ i + lo + en(GwQjb=G+Q+B+j+Gb)

kor + echhilen =korechhilen(KiwQjb)

echhile = e +chh + i + lo + e(GwQj)kor + echhile =

korechhile(KiwQj)

echhili= e +echh+i+lo+i(GwQwj)=kor + echhili

= korchhili(KiwQwj)

echhilen=e+chh+i+lo+en=(GwQjb)

=kor + echhilen= korechhilen(KiwQjb)

echilo=e+chh+i+lo(GwQj =G+Q+B+j+A)

kor+echilo= korechilo(KiwQj)

Aspect

Imperative Not applicableTable 3: Agreement of Person + class with the verb auxiliary and subject for Future Tense:

Person and ClassFP SPH SPNH SPP TPH TPNH

Subject ami/amara(Avwg/Avgiv)

apani/apanara(Avcwb/Avcbviv)

tumi/tomara(Zzwg/Zvgiv)

tui/tora(ZzB/Zviv)

tini/tara(wZwb/Zviv)

se/tara(m/Zviv)

Indefinitebo (e)kor + bo =(Ki+e =

korbo (Kiev)

ben (eb)kor + ben =(Ki+eb=Kieb

)korben

be (e)kor + be =(Ki+e=Kie)

korbe

bi (we)kor + bi =(Ki+we=Kiwe)

korbi

ben (eb)kor + ben(Ki+eb=

Kieb)= korben

be (e)kor + be(Ki+e

= korbeKie)

Continuous

te thakbo (Z _vKe) te thakben(Z _vKeb)

te thakbe(Z _vKe)

te thakbi(Z _vKwe)

te thakben(Z _vKeb)

te thakbe(Z _vKe)

Perfect

e thakbo (G _vKe) e thakben(G _vKeb)

e thakbe(G _vKe)

e thakbi(G _vKwe)

e thakben(G _vKeb)

e thakbeG _vKe)

Aspect

Imperative Not applicable ben (eb)kor + ben=(Ki+eb= Kieb)korben

o (A)kor + o =(Ki+A=Kiv)koro

ish (Bm)kor + ish =(Ki+Bm=Kwim)korish

ben (eb)kor + ben =(Ki+eb=Kieb)korben

be (e)kor + be =(Ki+e= Kie)korbe

Table 4: Agreement of Person + class with the verb auxiliary and subject for Habitual Past:

Person and ClassFP SPH SPNH SPP TPH TPNHSubject ami/amara

(Avwg/Avgiv)apani/apanara(Avcwb/Avcbviv)

tumi/tomara(Zzwg/Zvgiv)

tui/tora(ZzB/Zviv)

tini/tara(wZwb/Zviv)

se/tara(m/Zviv)

Indefinitetam=to+am(Zvg = Z+Avg)kor + tam =(Ki+Zvg=KiZvg)kortam

ten = to + en(Zb = Z+Gb)kor + ten =

(Ki+Zb = KiZb)korten

te = to + e(Z= Z+G)kor + te =(Ki+Z=KiZ)korte

ti = to + i(wZ=Z+B)kor + ti =(Ki+wZ=KiwZ)korti

ten = to + en(Zb=Z+Gb)kor + ten=korten(KiZb)

to (Z)kor + to =(Ki+Z=KiZ)korto

Continuous

te thaktam(Z _vKZvg)

te thakten(Z _vKZb)

te thakte(Z _vKZ)

te thakti(Z _vKwZ)

te thakten(Z _vKZb)

te tha kto(Z _vKZv)

Perfect

e thaktam

(G _vKZvg)

e thakten

(G _vKZb)

e thakte

(G _vKZ)

e thakti

(G _vKwZ)

e thakten

(G _vKZb)

e thakto

(G _vKZ)

Aspect

Imperative Not applicable

8/2/2019 7th ICCIT-2004_113

4/6

Table 5: Agreement of Person + class with the verb auxiliary and subject for Present Tense:Person and Class

FP SPH SPNH SPP TPH TPNHSubject ami/amara

(Avwg/Avgiv)apani/apanara(Avcwb/Avcbviv)

tumi/tomara(Zzwg/Zvgiv)

tui/tora(ZzB/Zviv)

tini/tara(wZwb/Zviv)

se/tara(m/Zviv)

Indefinite

i (B)

kor + i =kori(Ki + B =Kwi)

en (Gb)

kor + en =koren (Ki+Gb)

o (A)

kor + o =koro (Ki+A)

ish (Bm)

kor + ish =korish (Kwim)

en (Gb)

kor + en =koren (Kib)

e (G)

kor + e = kore(Ki)

Continuous

chhi/chchhi =wQ/wQkor + chhi (Ki + wQkorchhi (KiwQ)

chhen/ (Qb)chehhen (Qb)kor + chhen= korchhen(KiQb)

cho/ (Qv)chchho (Qkor + chho= korchho(KiQv)

chhish/ (wQm)chchhishkor + chhish =korchhish(KiwQm)

chhen/ (Qb)chchhenkor + chhen =korchhen(KiQb)

chhe/ (Q)chchhe (Qkor + chhe= korchhe(KiQ)

Perfect

echhi/(BwQ/BwQ)echchhikor + echhi= korechhi(Ki+BwQ=

KiwQ)

echhen/(GQb)echchhenkor+echhen= korechhen(KiQb)

echho/ (GQv)echchhokor+echho= korecho(KiQv)

echhish/(GwQm)echchhishkor+echhish =korechish(KiwQm)

echhen/(GQb)echchhenkor+echhen= korechhen(KiQb)

echhe/ (GQechchhekor+echhe= korechhe(KiQ)

Aspect

ImperativeNot applicable un (Db)

kor + un =korun (Kib)

o (A)kor + o =koro (Kiv)

Nullkor + Null= kor (Ki)

un (Db)kor + un =korun (Kib)

uk (DK)kor + uk =koruk (KiK)

5. TYPES OF SENTENCES:Structurally, there are three (3) types of sentence inBangla [3]. So, a PS rule for different kinds of sentencesof Bangla can be expressed as table 7:

Table 7: PS rules for Bangla Sentence.Rule No. PS Rule for Simple Sentence01.02.03.04.

S SS / CS / COMSSS Simple SentencesCS Complex SentencesCOMS Compound Sentences

5.1. Structure of the Simple Sentence:A simple sentence is formed by an independent clauseor principle clause. A principle clause in Bangla can berewritten as (or decomposes into, or consist of) a NPfollowed by VP. For example - se kall ashbe (pLm Bp-h) or Rahim eshkole jae (lqj -m ku).Therefore, a simple sentence can be represented by thefollowing PS rule-SS NP VP

The structure of sentences that are allowed by grammarFig. 1: SR for the simple sentences Ekti Chele boiPorche (HLV R-m hC fs-R).

trees or an SR. By using the context-sensitive PS ruleof table 1, 6, and 7, we can parse the simple sentenceshown in figure 1.Table 6: Context Sensitive PS rules for Bangla with

Subject -Verb and Person + Class agreement.RuleNo.

CSGs Rule

01.

02.03.04.05.06.07.08.09.10.11.12.13.14.16.

17.

18.

19.20.21.22.23.24.25.26.27.28.

29.

VP (NP) (AP) VF

VF VR AUXVR kor (Ki) /por (co) /other verb rootsAUX ASP T CONT Pr/Pt/FuPr NullPt lo (j)/ il (Bj)Fu b (e)/bo (ev)ASP I/Con/P/IMP/HabI NullCon chh (Q)/ chchh (Q)P echh (GQ)IMP NullHab to (Z)NICON NICONFP/

NICONSPH/NICONSPNH

NCON NICONSPP/NICONTPH/NICONTPNH/

ICON ICONSPH/ICONSPNH/ICONSPP/ICONTPH/ICONTPNHNICONFP i (B)/am (Avg)/Null/o (A)NICONSPH en (Gb)NICONSPNH Null/e (G)NICONSPP ish (Bm)/I (B)NICONTPH en (Gb)NICONTPNH e (G)/NullICONSPH un (Db)/en (Gb)/ben (eb)ICONSPNH o (A)ICONSPP ish (Bm)/Null/i (B)ICONTPH en (Gb)/un (Db)

ICONTPNH e (G)/uk (DK)/be (e)

NP

SPR N

QFR PP

VP

ASP T CON

VR

NP VF

NPer

PerTPNH

TPNH Con Pt NICON

NICONTPNH

eNullchhporboiCheleek ti

S

8/2/2019 7th ICCIT-2004_113

5/6

5.2. Structure of the Complex Sentence:

A complex sentence is formed by one principle clauseand one or more subordinate clause. For example -ami jani je tumi ashbe (Bj Se k aj Bp-h). Inthis sentence, Principle Clause: ami jani (Bj Se),and Subordinate Clause: je tumi ashbe (k aj

Bp-h). This two-sentence ami jani (Bj Se) and jetumi ashbe (k aj Bp-h) are simple form butsecond part is the subordinate of the first part. Clausesare combined with subordinator and/ or correspondingsubordinator complement.Some subordinator and their respective subordinatorcomplement used in Bangla are given in table 8.

Table 8: Some Subordinators and their respectiveSubordinator Complements.

Subordinator (SUBORD) Subordinator Complement(SUBCOM)

jadi (hw` )jadi (hw` )

jadi-na (hw`-bv)jekhane (hLvb)je (h)jemon (hgb)jehetu (hnZz)jara (hviv)

tahole (Zvnj)tobe (Ze)

tahole (Zvnj)sehkane (mnZz)se (m)temon (Zgb)sehetu (mnZz)tara (Zviv)

From the above analysis, proposed a set of PS rules forparsing the complex Sentences of Bangla. Apply therules of table 1, 6, 7, and 8 for parse the complexsentence is shown in fig. 2.Table 9: A set of Rules for parsing Complex Sentence.

5.3. Structure of the Compound Sentence:A compound sentence is formed by two or more

principle clause joined by an indeclinable orconnectives.

5.3.1 Connectives or Indeclinable:Indeclinable means no changes i.e. word in thesentence is unchanged. The part of speech, which isalways, unchanged form and which makes the sentencemore meaningful and also used, as a connective formore than one part of speech, clauses or sentences iscalled indeclinable. As a PS rule Connectives can berepresented as follows:Conj O (J), Ebong (Hhw), Fale (g-m), Sutorang(Bl palw), Kingtu (L).

5.3.2 Decomposition Technique of CompoundSentence:

To analyze the compound sentence, breaks thesentence into several independent clauses then

analyzed the independent clauses separately. Ifindependent clause is same as simple sentence thenanalyze that clause as simple sentence or if theindependent clause is same as complex sentence thenanalyze that clause as complex sentence. For example -Tomora manush noo, ar jader chalaoo taraoo manush

noo(ajl jeo eJ, Bl k-cl QmJ alJ jeo eu). Itis a compound sentence and it consists of twoindependent clause- (i) Tomara manush noo (ajl

jeo eJ) and (ii) Jader chalaoo taraoo manush noo(k-cl QmJ alJ jeo eu). These two independentclauses are combined a connective ar (Bl). Firstindependent clause is a simple sentence and secondindependent clause is a complex sentence. In thesecond independent clause is the combination of twoclauses: (i) Principle clause: Tomora manush noo(ajl jeo eJ) and (ii) Subordinate Clause: jaderchalaoo (k-cl QmJ). These two clauses are simplesentence. Therefore, first two independent clause are

separated and first one is analyze as simple sentenceand second one is analyze as complex sentence andthen the two clause under the complex sentence areanalyze as simple sentence separately. Compoundsentence can be constructed in several ways [8]. Suchas-

Fig. 2: SR for the Complex Sentence Ami Jani JeTumi kal ashbe (Bj Se k aj Bp-h).

(i) Simple + Simple:Consider an example- "se soth kingtu tar bhai osoth"(p pv L al iC Apv). In this sentence, "se soth"(p pv) and "tar bhai osoth" (al iC Apv) are principalclause or simple sentence and kingtu (L) is aconnective.(ii) Simple + Complex:Consider an example- "se soth kingtu je bonduti tarsange ashechilo se osoth" (p pv L k hV al

p- H-pRm p Apv). In this sentence, "se soth" (ppv) is a simple sentence and "je bonduti tar sangeashechilo se osoth" (k hV al p- H-pRm pApv) is a complex sentence.

(iii) Complex + Complex:

C DC ICCS IC DCDC NP (SUBORD) VPDC (SUBORD) SSIC NP (SUBCOM) VPIC (SUBCOM) SSSS All Simple SentencSUBORD Jadio, Jadi, Jekhane, JehetuSUBCOM Tahole, Tobe, Sekhane, Sehetu

Rule No. Rule for Complex Sentence

01.02.03.04.05.06.07.08.09.

IC DC

SS

VPNP

NPer

FP

VF

VR AUX

ASP T CON

I Pr NICON

SUBORD SS

NP VP

N

Per

SPNH

NP VF

N

Per

TPNH

VR AUX

I Fu NICON

ASP T CON

NICONSPNH

ami jan

Null

Null

i je tumi

kal ash

ebNull

CS

S

8/2/2019 7th ICCIT-2004_113

6/6

Consider a long sentence as an example -"jadi jantechaoo se keno asheni ta hole bolbo ami jani Na, ar jadi

jante chaoo ami keno jaini ta hole bolbo amar echheholona Tai" (kc Se-a QJ p Le B-pe a q-mhm-h Bj Se e, Bl kc Se-a QJ Bj Le Seaq-m hm-h Bjl C-R q-m e aC). In this example,

two complex sentences are connected by a connectivear (Bl). In this similar manner, compound sentence canbe constructed as simple + simple, simple + complex,complex + simple, Complex + Complex etc. Now, werepresent a set of rules to express the Compoundsentence in table 10.

Table 10: A set of rules for Compound Sentence.

Now, we apply the rule of table 1,6, 7, 8 and 10 toparse the compound sentence of Karim Danio loksutorang Tini Jakat Diben in the following fig. 3.

Fig. 3: SR for the Compound sentence Karim Donilok sutorang Tini Jakat Diben (Llj de

mL palw, ae kLa c-he).

6. FUTURE IMPROVMENTS:Our proposed CSGs PS rules can be parsed the allkinds of Bangla sentences in syntactic patterns. Thereare lots of researches scopes on the filed of Bangla

NLP. The possible future development may mention inthe following: CSGs for semantic analysis of Bangla sentences. CSGs for morphological analysis of Bangla

sentences. Efficient algorithm for Bangla NL parser to solve

the problem of ambiguity.

CSGs rules to handle various types of Banglapunctuation symbols.

The concepts of Voice, narration, composition, ofwords and inflection of noun or pronoun should befurther analyzed.

Pragmatic analysis on complex and compound

sentences may be carried on.

7. CONCLUSION:Parser is one of the most important parts of an NLPand an MT systems. Parsing a sentence then involvesfinding a possible legal structure for sentence andfinally gets an SR. Our proposed context-sensitive PSrules can be assigned all types of Bangla sentences intoan SR. To represents a complex and a compoundsentence into an SR, we uses the decompositiontechnique. An SR generally graphic object and thisrepresentation cannot be deals with computer. Thestandard representation of an SR is a list that are one of

the data structure that can be implemented andmanipulated very easily within a computer. Furthermodify and extending the CSG rules (with the additionof sentences consisting of idioms and phrases, doubleword, change of voice, and narration), we can able torepresents the all kinds of Bangla sentences which areused as an input of an MT engine to produced otherequivalent sentences.

8. REFERENCES:[1] A. Trujillo, Translation Engines: Techniques forMachine Translation, Springer-verlag, London, (1992).[2] D. Arnold, L. Balkan, S. Meijer, R. L. Humphreys,

and L. Sadler, Machine Translation An IntroductoryGuide, Ncc, Blackwell Ltd., London, (1994).[3] A. Humayan,Bakyattava, 2nd ed., The University ofDhaka, 1994.[4] M. M. Murshed, Parsing of Bengali NaturalLanguage Sentences, in Proc., InternationalConference on Computer and Information Technology,ICCIT 98, (Dhaka, Bangladesh), PP. 185-189, (1998).[5] M. R. Selim and M. Z. Ikbal, Syntax analysis ofPhrases and Different types of Sentences in Bangla, inProc., International Conference on Computer andInformation Technology, ICCIT 99, (Sylhet,Bangladesh), pp. 175-186, (1999).

[6] M. I. A. Khan, A. K. M. A. Hossian and R. C.Debnath, A Bangla Conversion Processor usingNatural Language Processing, In Proc. InternationalConference on Computer and Information Technology,

ICCIT02, (Dhaka, Bangladesh), PP. 262-266, (2002).[7] L. Mehedy, S. M. Arefin, and M. Kaykobad,Bangla Syntax Analysis: A Comprehensive Approach,In Proc. International Conference on Computer andInformation Technology, ICCIT03, (Dhaka,Bangladesh), PP. 287-293, (2003).[8] M. M. Hoque and M. M. Ali, A ParsingMethodology for Bangla Natural Language Sentences,In Proc. International Conference on Computer andInformation Technology, ICCIT03, (Dhaka,Bangladesh), PP. 277-282, (2003).

COMSCOMSCOMSCOMS

SS Conj SSSS Conj CSCS Conj CSCS Conj SS

SS

CSConj

All Simple SentenceAll Complex SentenceO, Ebong, Noile, Kintu, Fale,

Notuba, Sutorang,...

RULE NO. RULE for COMPOUND SENTENCE

01.02.03.04.05.

06.07.

SSSS

VPNP

N

Per

TPH

NP VF

NP VP

NP VF

N

Per

SPNH

VR AUX

I Fu NICON

en

ASP T CON

NICONTPH

COMS

S

AP N

N

Per

TPH

bNulldijakattini

sutorang

imp-licit

Null

lokKarim

doni