Formal Description Of Arabic Syntactic Structure In The .

2y ago
30 Views
2 Downloads
238.79 KB
15 Pages
Last View : 15d ago
Last Download : 3m ago
Upload by : Milo Davies
Transcription

Formal Description of Arabic Syntactic Structurein the Framework of the Government and Binding TheoryHammo Bassam1, Moubaiddin Asm a2, Obeid Nadim 1, and Tuffaha Abeer11 KASIT, CIS Department,2 Department of Linguistics,The University of Jordan, Amman,Jordan{b.hammo, a.mobaiddin, obein}@ ju.edujo, atuffaha@ hotmail.comAbstract. The research focus in our paper is twofold:(a) to examine the extent to which simple Arabicsentence structures comply with the Government andBinding Theory (GB), and (b) to implement a simpleArabic Context Free Grammar (CFG) parser to analyzeinput sentence structures to improve some ArabicNatural Language Processing (ANLP) Applications.Here we present a parser that employs Chomsky'sGovernment and Binding (GB) theory to betterunderstand the syntactic structure of Arabic sentences.We consider different simple word orders in Arabic andshow how they are derived. We analyze differentsentence orders including Subject-Verb-Object (VSO),nominalsentences,nominalsentencesbeginning with inna (and sisters) and questionsentences. We tackle the analysis of the structures todevelop syntactic rules for a fragment of Arabicgrammar. We include two sets of rules: (1) rules onsentence structures that do not account for case and(2) rules on sentence structures that account for caseof Noun Phrases (NPs). We present an implementationof the grammar rules in Prolog. The experimentsrevealed high accuracy in case assignment in ModernStandard Arabic (MSA) in the light of GB theoryespecially when the input sentences are tagged withidentification of end cases.Keywords. Arabic syntax, Government and Bindingtheory,Arabicparser,Arabicnaturallanguage processing.1 IntroductionWords convey meaning. But when they aregrouped together based on grammatical structurethey convey larger meanings [15]. Identifying thestructure (syntax) is the first step towardsunderstanding the meaning of a sentence.Syntactic analysis (parsing) is a procedure thatrecognizes a sentence and discovers how it isbuilt (i.e., gives its grammatical structure).Recognition involves finding out whether thesentence under consideration belongs to aparticular language, i.e., whether it follows all therules that this language prescribes. Discoveringthe structure (parsing) involves identifying andmarking the various components of a sentence(i.e., phrases and individual parts of speech suchas noun, verb, preposition, etc.) [14].Parsingof sentencesis a necessarymechanism for many natural language processing(NLP) applications. Not all NLP applicationsrequire a complete syntactic analysis. Forinformation retrieval (IR), it is sufficient to findnoun phrases (NPs) and verbal phrases (VPs).However, for such applications as informationextraction (IE), text summarization (TS), questionanswering (QA), we are interested in ngful) roles such as agents, objects,locations, time, among others (who did what towhom, when, where, why, etc.).The key idea in rule-based parsing is thatgiven a gram m ar and a sentence, a parser willdetermine if a given sentence is well formedaccording to the grammar, and what a derivationtree would look like. Despite the fact that thedevelopment and maintenance of handwrittengrammars is a hard task, there is a strongadvantage of rule-based parsing as one caneasily modify and accommodate the parser tonew tasks. Diab et al. [7], suggested that rule-Computacion y Sistem as Vol. 18, No. 3, 2014 pp. 611-625ISSN 1 4 05 -55 46DOI: 10.13053/C yS-18-3-2017

612Hammo Bassam, Moubaiddin Asma, Obeid Nadim, Tuffaha Abeerbased parsers are implausible and that syntacticanalyzers (parsers) could be based on lexicalproperties and structure determining principles.and 4 are dedicatedits implementation.Incorporating the Government and Binding(GB) theory [3], [5] into a parser helps to eliminatemanygram m arrulesbecauseoftheirredundancy, as syntactic structures can bederivable using means other than explicit rules.Principles are general constraints on syntacticrepresentations (and not on rule application). GBenables linguists to replace many traditional rulesusing a small number of fundamental linguisticprinciples. GB principles are constraints over Xbar structures.1.1 Brief Introduction to ArabicThe significance of the principles is toconstrain the class of possible entation, withlanguage-specific rules,enables a parser to predict syntactic structure(s).Parsing Arabic texts is challenging becausethe Arabic language has rich morphology due toits highly inflectional nature, highly flexible wordorder and frequent use of clitics, which areattached to words [2], [20]. The emphasis in thispaper is on the use o f Government and Binding(GB) theory in analyzing the syntactic structure ofsomesimpleModernStandardArabic(MSA) sentences.We describe a parser that is based on GBgrammatical theory. W e shall use GB principlesand rules to describe Arabic-specific properties ormarked structures and to analyze the syntacticstructure of some simple Arabic sentences [4],[12]. W e consider different word orders in Arabicand show how they are derived. W e shall includean analysis of SVO, VOS, VSO, nominalsentences, nominal sentences beginning withinna (and sisters), and question sentences. Weuse this analysis to develop syntactic rules for afragment of Arabic grammar. Due to spacelimitation, we shall not present some importantcomputational steps and the structure of a lexiconnecessary to build the implemented system.This paper proceeds as follows: in the next twosubsections we present a brief introduction toArabic and give a brief presentation o f the GBtheory. In Section 2 we discuss the notion ofArabic syntactic analysis in light of GB. Sections 3Computacion y Sistem as Vol. 18, No. 3, 2014 pp. 611-625ISSN 14 05 -5 5 4 6DOI: 10.13053/C yS-18-3-2017totheparserandArabic is a Semitic language that has a richmorphology and a flexible word order. In thispaper we are concerned with Modern StandardArabic (MSA), which is used in modern writingand is understood by Arabic language speakers.Arabic gram mar distinguishes between two typesof sentences, verbal andnominal. Nominalsentences have two parts: a subject (m obtada’and a predicate (khabar j- ).When the nominal sentence speaks aboutbeing, i.e., if the verb of the sentence is ‘to be’ inEnglish, this verb is not given in Arabic. Arabicmorphology is based onroots and patternsthrough which words are derived. An Arabic wordmay be composed o f a stem consisting o f a baseroot and a pattern which defines its semantic andsyntactical role. Moreover, affixes and clitics areoften attached to words.Affixes include inflectional markers for tense,gender, and number. Clitics include epronouns. Here we present some of thecharacteristics and / orchallenges of theArabic language.1.2.3.It has a relatively free word order. It is notuncommon to find VSO, SVO and VOS wordorders within an Arabic text as in the followingexamples (see Table 1). All of the sentencesin Table 1 are grammatically well-formed andhave the same English meaning: “Theteacher read the lesson”.Arabic is a clitic or clitic-directed language.Clitics are morphemes that have the ly bound to other words (e.g.,coordinating conjunctions, the definite article,many prepositions and particles, and a classof pronouns that attach themselves either tothe beginning or the end of words) as in i t *:katabna (we wrote) which is made up of theverbkatab and the clitic to na which actsas the subject for the verb katabThe absence of diacritics (syntactic marks) inmost written Arabic texts is very common.

Formal Description of Arabic Syntactic Structure in the Framework of the Government and Binding TheoryTable 1. Examples of Arabic sentences withdifferent word orders2.OrderArabic example (reads right left)VSOSVOVOSto *I jua-dars-aal-mualim-uqar'-athe lessonthe teacherread Luto Ja-dars-aqar'-aal-mualim-uthe lessonreadthe teacherto * laJIjual-mualim-ua-dars-aqar'-athe teacherthe lessonreadjj JI3.4.5.6.Arabic is a pro-drop language. The subjectcan be omitted, leaving any syntactic parserwith the challenge to decide whether there isan omitted pronoun in the subject position.5.Homographs of words with/without the samepronunciation are often produced. They havedifferent meanings and usually different partsof speech (POS). For example, the Arabicword:thahab can be interpreted asthahab-a (a verb meaning “went”) or asthahab-un (a noun meaning “gold”).2.3.4.5.1.2 Brief Introduction to the Government andBinding Theory (GB)6.The GB theory [3], [5] is an approach to UniversalGrammar which includes rules and principles thatapply to all languages. However, while certaingrammatical principles and rules are universal,there is a lot of variation between differentlanguages such as different ordering for subject(S), verb (V) and object (O). It is agreed thatevery language has a basic word order, and allother word orders result from the movement ofsentence constituents and this movement isrestricted by some rules and principles. Wordsare organized hierarchically into bigger unitscalled phrases. Phrase constituents include:1.IP - Inflectional Phrase: a phrase headed byI/INFL. I/INFL stands for inflection, and itconsists of tense, number, and genderagreement (AGR) elements.CP - Complementizer Phrase: a phraseheaded by a complementizer C. C takes anIP(INFL Phrase) as its complement andheads the maximal projection CP.NP - Noun Phrase: a phrase headed by anoun (N).VP - Verb Phrase: a phrase headed by averb (V).AP - Adjective or Adjectival Phrase: a phraseheaded by an adjective (A).PP - Prepositional Phrase: a phrase headedby a preposition (P).The main principles of GB are:1.4.6137.8.Government, whichis concerned withsyntactic relations in a sentence and has itsmain application in case assignment.Theta Theory, which is concerned withdescribingthematicrelationsbetweenarguments and predicates.Predicates and arguments: arguments areparticipants minimally involved in the activityor state expressed by a predicate.Case Theory which is concerned with theassignment of abstract cases (nominative,accusative, and genitive) to words, based ontheir positions in a sentence.X-Bar Theory, which is concerned with phraseformation. It states that all phrases areheaded by a lexical head (noun, verb,adjective, or preposition).Complements combine with X to form X ’projections, adjuncts combine with X ’ to formX ’’. A specifier combines with the topm ost X ’to form the maximal projection X ’’/XP.D-structure and S-structure: all sentences arerepresented in terms of both forms, the Dstructure and the S-structure. The D-structureencodes predicate-argument relations andthematic properties of a sentence and it isbuilt upon the basic word order. The Sstructure accounts for the surface ordering ofthe sentence constituents.NP-Movement: GB assumes that thedifferent wordordersarise fromthemovement of sentence constituents. Hence, abasic word order is assumed, and all otherword orders are derived.Computacion y Sistem as Vol. 18, No. 3, 2014 pp. 611-625ISSN 1 4 05 -55 46DOI: 10.13053/C yS-18-3-2017

614Hammo Bassam, Moubaiddin Asma, Obeid Nadim, Tuffaha AbeerTable 2. Rules and examples of Arabic noun phrases U li!u iuNDet NNP NPNP Conj NPNP APjl»(il kitab-u el-bint-ial-layl-u wa anahar-ukitab-n mufid-nBookThe bookThe girl's bookThe night & dayA useful bookTable 3. Rules and examples of Arabic adjective phrases (AP)ArabicexampleSyntacticstructure-AP i AEnglishmeaningTransliterationmufid nUsefulA P i A APazrak-n dake-nDark blueA P i AP Conj APazrak-n wa dake-nDark & blueTable 4 . Rules and examples of Arabic verb phrases (VP)SyntacticstructureArabicexampleV P iVIjjsV P iV NPV i V PPV P iV P PPSci' cJEnglishMeaningTransliterationqar'-aHe readakal-a atufahat-aHe ate an applethahab-a ilaalmadrasat-iHe went to schoolwajadt-u alkitab-a alaI found the bookatawilat-ion the tableTable 5. Rules and examples of Arabic prepositional phrases (PP)SyntacticstructureP P i P NPArabicexample'2 Analysis of Arabic Syntax in Light ofGBIn th is section, we describe our analysisconcerning some of the Arabic grammaticalstructures in the light of the Government andBinding (GB) Theory. W e delve into the basics ofGB and attempt to apply it on some simplesentence structures in Arabic.Computacion y Sistem as Vol. 18, No. 3, 2014 pp. 611-625ISSN 14 05 -5 5 4 6DOI: 10.13053/C aningIn thelibrary2.1 Scope of Sentence Structure AnalysisOur implementation of Arabic syntactic analysis isrestricted to basic sentence structures thatinclude1.SVO, VOS, and VSO sentences where thesubject is an NP and the object is an NP.2.Sentences followed by a PP adjunct.

Formal Description of Arabic Syntactic Structure in the Framework of the Government and Binding Theory3.4.Nominal sentences made up from NP(s), orNP followed by PP, and nominal sentencespreceded by inna (and sisters).Question sentences staring with a questionword followed by a VSO order sentence.2.2 Constituent Structure in ArabicWe first analyze the lexical formation of thesmaller phrase constituents that make up asentence. The analysis includes a noun phrase(NP), an adjective phrase (AP), a verbal phrase(VP), and a prepositional phrase (PP).615IPSpecNPIVPVNPqar -areada-dars-athe lesson2.2.1 Noun Phrase (NP)An NP head is a noun and it can be representedaccording to the rules and examples presented inTable 2.2.2.2 Adjective Phrase (AP)An AP head is an adjective and it can berepresented according to the rules and examplespresented in Table 3.al-mualim-uthe teacherFig. 1. S-Structure of the SVO Arabic sentence“al-mualim-u qar’-a a-dars-a”2.2.3 Verb Phrase (VP)A VP head is a verb and can be representedaccording to the rules and examples presented inTable 4.2.2.4 Prepositional Phrase (PP)A PP head is a preposition and it can berepresented according to the rules and examplespresented in Table 5.2.3 Arabic Basic Word Order (SVO)Greenberg [10] claimed that languages whichexhibit a Verb-Subject-Object (VSO) word orderare a minority among the world languages. If sucha claim is valid, then a change in word order isexpected to be in the direction o f the morecommon SVO order. Classical Arabic can beconsidered as one of the VSO languages.According to El-Yasin [9], Colloquial JordanianArabic seems to exhibit SVO order judging by thefacts of subject-verb agreement and facts aboutthe number of topics allowed to precedesentences in this dialect of Arabic. In [9], theauthors concluded that Arabic would be anexample of a language changing from a VSOlanguage (in its classical form) into a SVOlanguage (in the case of Jordanian), thussupporting Greenberg's claim [10].We assume that the basic word order forArabic sentences within the framework of GB isSVO. In SVO order, I/NFL assigns a NOM case tothe subject at [Spec, IP] position (through thepercolation o f I/NFL to IP), and the verb whichheads the VP assigns an ACC case to its object.As an example, consider the following sentence:o j J i j j J (reads from right to left) (al-mualim-uqar’-a a-dars-a) “The teacher read the lesson”. Inthissentence“al-m ualim-u” (the teacher),receives a NOM case from I/NFL, and the noun“a-dars-a” (the lesson) is assigned an ACC casefrom its governing verb “qar’-a” (reads). As aresult of I/NFL's government, agreement isimposed. Here, we notice that there is a fullagreement in number and gender between theverb and the subject in SVO order as shown inFigure 1.Computacion y Sistem as Vol. 18, No. 3, 2014 pp. 611-625ISSN 1 4 05 -55 46DOI: 10.13053/C yS-18-3-2017

616Hammo Bassam, Moubaiddin Asma, Obeid Nadim, Tuffaha AbeerCPIPSpecNPIVPVSpec/\Vqar-’a a-dars-aread the lessonNPVNPal-walad-uthe boyal-mualim-uthe teacherya’kul-ueatstiFig. 2. Structure of the VOS Arabic sentence “qar’-a adars-a al-mualim-u”ya’kulueatsa-tufahat-aan appleal-walad-uthe boyFig. 3. (1) The D-structure and (2) the S-structure of theVSO Arabic sentence2.4 Other Word Orders and NP MovementIPThe other word orders are the result ofmovements applied on the basic word order.These include the word orders we discuss below.2.4.1 VOS Word OrderVOS results from the subject adjunction to theend of VP. Hence, it will receive a NOM case fromI/NFL. And to satisfy the EPP principle, we canassume [Spec, IP] to be occupied by PRO. Figure2 explains the VOS word order for Arabic.SpecNPIVPVNP2.4.2 VSO Word OrderIn the D-structure (Figure 3(1)), the subject at[Spec, IP] receives a NOM case from I/NFL, theverb's object receives an ACC from the verb. VSOorder is obtained by moving the verb to empty [C,CP] (Head to Head movement) leaving its co indexed trace (the accusative case of VP'sinternal NP is assigned through the verb's trace)[1] (see Figure 3(2)).Computacion y Sistem as Vol. 18, No. 3, 2014 pp. 611-625ISSN 14 05 -5 5 4 6DOI: 10.13053/C yS-18-3-20170's)m u fid -u na l-k ita b -uth e b o o kisu s e fu lFig. 4. The structure of a typical Arabic nominal sentence

Formal Description of Arabic Syntactic Structure in the Framework of the Government and Binding Theory2.5 Nominal Verbless Sentences2.5.1 Typical Nominal SentencesA typical Noun Phrase (NP) in Arabic containstwo nouns as in the following example: (al-kitab-u mufid-n) “The book is useful”.As mentioned earlier, we consider SVO to bethe basic word order for Arabic sentences. Toexplain the grammaticality of this type ofsentences, we can assume that there is a hiddenverb, such that this verb carries the meaning o f is ,and it occupies V. Accordingly, the phrase will betensed, which allows I/NFL to assign a nominativecase to NP at [Spec, IP].However, the hidden verb will fail to govern itsinternal argument. To solve this issue we willadopt the default case approach mentioned in[13] which says that NPs with no case assignerare possible in Arabic and they are assigned anominative case. Figure 4 shows the structure ofa typical nominal sentencei (al-kitab-umufid-n) “the book is useful”. “al-kitab-u” receivesa NOM case from I/NFL, and “mufid-n” isassigned the default NOM case.617walad-a”, and the verb “ya’kul-u” assigns an ACCcase to its object “al-tufahat-a” (see Figure 5 (2)).Thesameappliestoallotherinnasisters’ complementizers.2.5.3Question SentencesQuestions in Arabic usually start with a questionword such as(man) “who”, Wf (matha) “what”, ( m a t a ) “when” and(ayna) “where”.-Questions on Subjects or Objects. Both d?(who) and W? (what) can be used to ask aboutthe subject or the object. If a question is aboutthe subject, in the D-structure (Figure 6(1))the question word is placed at [Spec, IP] andin the S-structure (Figure 6(2)) it is moved to[Spec, CP].If a question is about the object, instructure the question word is placedobject's node under VP, and to producestructure, the question word is moved toCP]. Consider the following cases:-the Dat thethe S[Spec,Questions on Subjects. ? o j pJi b J u? (manqar’a a-dars-a?) “Who read the lesson?”2.5.2 Nominal Sentences with inna (and sisters)In the D-structure (Figure 6(1)), the base of thequestion word “man” is generated at [Spec, IP],Inna and sisters particles (i»4iclj 0J) can occur inbut in the S-structure (Figure 6(2)), it moves tonominal sentences. They include the particles o !inna, anna, layta, laalla. . Inna[Spec,and CP].sisters are complementizers that assign anQuestions on Objects. d K J I ij j Wf (mathaaccusative case to their noun governees.qar’-a a-talib-u?) “W hat did the student read?”Consider the following example: ojIn the D-structure (see Figure 7), the base of(inna al-kitab-a mufid-un) “the book is useful”.the question word “matha” is generated at theIn Arabic, in the existence of inna and sistersobject position. One can notice that this questionparticles at [C, CP], the NP at [Spec ,IP] isbegins with a question word followed by a verb. Inassigned an ACC case from the complementizerorder to explain the sentence grammaticality, weparticle. W e assume that inna and sistersassume that the question word is moved to [Spec,complementizers are stronger than tensed I/NFLCP], and the verb is moved to [C, CP] (seeand prevents it from assigning the nominativeFigure 8).case to NP at [Spec, IP]. In the above example,inna is a complementizer that assigns an ACCQuestions on VP Adjuncts. i(ayn-a)case to its governee “al-kitab-u” and the hidden“where”, ? m a t a ) “when”, a n d 'M 2 (k a y f-a )verb (is) fails to assign an accusative case to“how” are usually used to ask about a VP's“mufid-n”. Therefore, it is assigned the defaultadjunct. Consider the following example: ' d iNOM case as shown in Figure 5 (1).j -i i (ayna safar-a Ali?) “W here didNow, consider the sentenceojAli travel?”(inna al-walad-a ya’kul-u a-tufahat-a) “the boy iseating an apple”. In this sentence, theIn the D-structure, the base of the questioncomplementizer inna assigns an ACC case to “alword “ayn-a” is generated at VP's adjunctComputacion y Sistem as Vol. 18, No. 3, 2014 pp. 611-625ISSN 1 4 05 -55 46DOI: 10.13053/C yS-18-3-2017

618Hammo Bassam, Moubaiddin Asma, Obeid Nadim, Tuffaha AbeerCPCPNPVal-kitab-athe bookinnathatal-walad-athe boymufid-unusefulisy a’kul-uis at-aan applea-talib-uthe studentFig. 7. D-structure of a question on object with thequestion word mathaFig. 5. Two nominal sentences (1) with inna and ahidden verb, (2) with inna and a verbposition, and in the S-structure, we assume thatthe question word is moved to occupy [Spec,CP],and the verb is moved to [C, CP].CP2.5.4 Yes/No QuestionsMany Yes/No questions use the question word J(hal) “Did”. Here is an example to explain thiscase:J (hal safar-a Ali?) “Did Ali travel?”In this sentence, in the D-structure, weassume that the base of the question word “hal” isgenerated as an IP adjunct. To produce the Sstructure, the verb is moved to [C, CP] and thequestion word is moved to [Spec, CP].VNP3 The Arabic ParseremanWhoqar’-a a-dars-aread the lessontimanWhoqar’-a a-dars-aread the lesson@Fig. 6. (1) The D-structure and (2) the S-structure of aquestion on subject with the question word manComputacion y Sistem as Vol. 18, No. 3, 2014 pp. 611-625ISSN 14 05 -5 5 4 6DOI: 10.13053/C yS-18-3-2017Our syntactic parser (see Figure 9) takes asentence as an input and outputs whether it issyntactically correct also generating its bracketstructure. To assign part of speech tags (POSTs)to the sentence, we used the Aramorphtagger [4], which is the Java version ofBuckwalter's Arabic morphological analyzer.

Formal Description of Arabic Syntactic Structure in the Framework of the Government and Binding TheoryUnlike rule-based grammars that use a largenumber of rules to describe patterns in alanguage, the GB Theory explains these patternsin terms of more fundamental and universalprinciples [7], [12], [19]. A key issue in building aprinciple-based parser is how to interpret in aprocedural way the principles expressed asgrammar rules.SinceGBprinciples areconstraints over syntactic structures, one way toimplement the principles is as follows.1.2.3.Generate candidate structures of a givensentence that satisfy X-bar theory and sub categorization frames of the words in thesentence.Filter out structures that violate any one ofthe principles.The remaining structures are accepted asparse trees of the sentence.Once the gram m ar rules are compiled intoProlog, they receive a procedural interpretation,becoming a top-down, left-to-right, recursivedescent parser. In other words, by representingthe rules of gram m ar as axioms in Prolog hornclause logic, we can use Prolog theorem provingengine as a lib-uthe studentFig. 8. S-structure of a question on object with thequestion word matha3.1 Gram m ar RulesThe rules in the gram m ar base include a set ofArabic gram m ar rules derived from our analysis ofArabic sentences according to GB. They aredivided into two parts. The first part includes thesyntactic rules that do not account for words casemarks (diacritics) (see Tables 6 and 7) and thesecond part includes rules for case markedsentences (see Tables 6 and 8). In either case,the basic sentence phrase constituent syntax islisted first, followed by the rules for the syntax ofanalyzed sentences.4 System Implementation and ResultsAn input sentence to the parser is represented asa sequence of tags. The syntactic parser takesthe sequence of tags of the tagged sentence andreturns as output valid syntactic structure(s) of thesentence. W e adopted a top-down recursiveFig. 9. System architectureapproach; rules start at the sentence level (S),continue to rules for phrases (intermediate level)and finally to parts of speech (lowest level).We have opted to employ SICStus Prolog3.12.2 [21] to implement the parser since Prologcan be effectively used in natural languageanalysis due to the following: (1) Prolog is alogical programming language, which seemssuitable to express grammar rules and (2) wewere not aiming at testing the efficacy of theArabic gram m ar base. We have employed twofiles: one is consulted when caseless analysis isrequired by the user and the other is used whenthe user considers cases.Computacion y Sistem as Vol. 18, No. 3, 2014 pp. 611-625ISSN 1 4 05 -55 46DOI: 10.13053/C yS-18-3-2017

620Hammo Bassam, Moubaiddin Asma, Obeid Nadim, Tuffaha AbeerTable 6. Arabic grammar rulesRegardless of the caseS CP IPS S Conj SNP NNP N NPNP NPConj NPNP NP APNP PropNounAP AP AP APPP SpecSpecSpecAdjAdj APAP ConjPrep NP NP Pron DemRegarding the caseS CP IPS S Conj SNP(nom) N(nom)NP(acc) N(acc)NP(gen) N(gen)NP(Case) N(Case) NP(Case2)NP(Case) NP(Case) Conj NP(Case)NP(Case) NP(Case) AP(Case)NP Prop NounCase: a variable specifying NP's case.(Case2 Case or Case2 gen)AP(Case) Adj(Case)AP(Case) Adj(Case) AP(Case)AP(Case) AP(Case) Conj AP(Case)PP Prep NP(gen)Spec(Case) NP(Case)Spec(Case) Pron(Case)Spec(Case) Dem(Case)Table 7. Arabic sentence rules regardless of the caseOrderSVOVOSVSONominalSentenceQuestionsRuleIP Spec VPIP Pro VPVP V NPVP VP PPIP VPVP VP SpecVP VP PPCP vso VIP vsoIP vso SpecIP vso Spec VP vsoVP vso NP(acc)VP vso V P vso PPIP nominal Spec VP nominalVP nominal NPVP nominal PPCP Func word IP nominalCP Q word CP vsoAs mentioned earlier, the input to the Prologparser is a sequence of part of speech tags. Theparser matches the input list with sentence rules,breaking the sentence down and matching it withsmaller constituent rules until a match is found atthe level of tags. If there is a match, the systempoints out a possible syntax structure upon whichthe input sentence is built. However, if there is nomatch, the system returns a “No” result to indicatea mismatch with the sentence rules.The implemented system was tested on 500sentences. Sentences were tested againstComputacion y Sistem as Vol. 18, No. 3, 2014 pp. 611-625ISSN 14 05 -5 5 4 6DOI: 10.13053/C yS-18-3-2017Explanation(Pro: stands for a prodrop pronoun that is hidden)(object is an NP)(PP adjunct to VP)(PP adjunct to VP)(case of intransitive verb)(PP adjunct to VP)(comment is an NP)(comment is a PP)(inna and sisters starting the sentence )grammar rules for both cased and the caselesssentences.Table 9shows some samplesentencesalong with their resulting syntaxstructures.For allresults in the sample, boldsyntax structure signifies a correct result. Startingwith SVO sentences, Table 9 shows that for allanalysis both structures (caseless and cased) aresyntactically correct.For VSO sentences, the cased analysis for thefirst example is correct. The caseless analysis forthe same example produced two structures; thefirst one is correct while the second one is

Formal Description of Arabic Syntactic Structure in the Framework of the Government and Binding Theory621Table 8. Arabic sentence rules regarding the caseSentence orderSVOVOSVSONominal SentenceNominal sentence withinna and sistersRuleExplanationIP Spec(nom) VPIP Pro VPVP V NP(acc)VP VP PPIP VPVP VP SpecVP VP PPCP vso VIP vsoIP vso SpecIP vso Spec VP vsoVP vso NP(acc)VP vso V P vso PPIP nominal Spec(nom) VP nominalVP nominal NP(nom)VP nominal PPCP Func word IP nominal2IP nom inal2 Spec(acc) VP nominalincorrect. This is because the analyzer could notdifferentiate between the subject NP and theobject NP. The second example shows that bothstructures are correct.For VOS sentences, the cased analysis for thefirst sentence is correct. The caseless analysis forthe same sentence produced three structures; thefirst two are incorrect while the third one iscorrect. The second example shows a correctanalysis of the cased structure while it shows twostructures for the caseless syntax, and only thesecond one is correct.For the nominal sentences, the caseless andthe cased syntaxes were correct. Finally, for thequestion sentences, in the first example thecaseless and the cased analyses produced onestructure, which is correct in both cases. Thesecond example produced two structures for thecaseless syntax, but it produced only onestructure for the cased syntax.-it is obvious that the implementedparser based on the GB theory(when verb is intransitive)(PP adjunct to VP)(comment is an NP)(comment is a PP)Case 1: VSO VOS Nominal SentenceThis case is related to NP ambiguity becausethe parser has no clue on where a subject and anobject start and end.Case 2: VSO VOSIn this case the parser is not able to determinewhether an NP is a subject or an object.-By now,syntactic(PP adjunct to VP)sometimes produces multiple analyses for asentence. Mainly this is due to semanticambiguity, which is not expected to be covered bysyntactic parsing in the first

bar structures. The significance of the principles is to constrain the class of possible syntactic representations. The bound on syntactic representation, with language-specific rules, enables a parser to predict syntactic structure(s). Parsing Arabic texts is challengi

Related Documents:

113 70 0645 arabic letter meem 114 71 06ba arabic letter dotless noon 115 72 0646 arabic letter noon 116 73 0648 arabic letter waw 117 74 0624 arabic letter hamzah on waw . 121 78 0649 arabic letter alef maqsurah 122 79 06d2 arabic letter ya barree 123 7a 06be arabic letter knotted ha 124 7b a

ﺑﺮﻌﻟا The Beginner's Guide to Arabic GUIDE TO STUDYING ARABIC 2 WHY STUDY ARABIC 2 HOW TO STUDY ARABIC 3 WHERE TO STUDY ARABIC 4 WHAT YOU NEED BEFORE YOU START 4 THE ARABIC ALPHABET 5 INTRODUCTION TO THE ALPHABET 5 THE LETTERS 6 THE VOWELS 11 SOME BASIC VOCABULARY 13 RESOURCES FOR LEARNING ARABIC 17 ONLINE 17 RECOMMENDED BOOKS 18 OUR NEWSLETTERS 19 by Mohtanick Jamil . Guide to .

2. Syntactic Annotation of Quranic Arabic 2.1 Traditional Arabic Grammar ( ) Arabic is a morphologically rich language, and is highly inflected. One motivation for the historic development of traditional Arabic grammar has been to

representation of syntactic structures. Representation of syntactic structures can be presented in three ways: statements of the correct sequence of the parts of speech (syntactic categories), by series of transformational rules and by parsing diagrams. Syntactic categorie

0644 arabic letter lam 0645 arabic letter meem 0646 arabic letter noon 0647 arabic letter heh 0648 arabic letter waw 0649 arabic letter alef maksura 064a arabic letter yeh tashkil from iso 8859

Arabic Courses ARABIC 110 Elementary Arabic I Credits: 5 Fundamentals of the language, essentials of conversation, grammar, practical vocabulary, useful phrases, and the ability to understand, read and write simple classical Arabic. ARABIC 110 - MOTR LANG 105: Foreign Language I ARABIC 120 Elementary Arabic I

Classical Arabic to Modern standard Arabic Focusing on the main reason for changes within the Arabic language. Then it discusses the Arabic dialects focusing on the phenomenon of diglossia, which is the existence and use of two or more types of Arabic in an Arabic-speaking country, the reasons for its existence and its effect

American Revolution This question is based on the accompanying document (1-6). The question is designed to test your ability to work with historical documents. Some of the documents have been edited for the purposes of the question. As you analyze the documents, take into account the source of each document and any point of view that may be presented in the document. HISTORICAL CONTEXT: passed .