A Formal Description of Arabic Syntax in Definite Clause Grammar

1. formulate a problem 2. provide some examples 3. number these examples and when referring to them 4. draw syntactic trees of these examples 5. explain the process and how it might be complicate (refer to Chomsky or other sources) 6. it doesn't have to provide a solution 7. don't make it as a book report. A Formal Description of Arabic Syntax in Definite Clause Grammar Hisham E1-SllisMny IBM Scientific Center, 56, Gameaat El-Doual EI-Arabeya St. Mohandesseen, Cairo, Egypt Abstract Arabic has some special syntax features which lead to complex syntax structures. We have developed a formal description of Arabic syntax in Definite Clause Grammar. This grarnrnar is characterized by its high descriptive power due to its dual formulation in terms of functions and in terms of grammatical categories. The developed grammar has a high coverage of Arabic language and has context sensitive capabilities. It is suitable for the advanced applications of natural language processing. 1o Introduction Arabic has some features which lead to comlpex syntax different from those of the European languages. Moreover, Arabic grammar exists cmly in a descriptive form and there is no comprehensive formal rep~ resentation for it so far. The few attempts to give a formal mode~ for Arabic sentences were based on transformational generative grammar - [111, [2], [3] and others-but some linguists adopted more recent linguistic models such as Lexical Functional Grammar [4], dependency grammar [5] and functional grammar 16]. Selecting a suitable grammar formalism for Arabic and the formulation of the grammar itself is the subject of a big debate in the Arab countries nowadays. Our descriptio:n of Arabic syntax in Definite Clause Grammar is different from the attempts undertaken so far. 2. Deft,ire clat,se gram,,ar for Arabic syntax In this grammar, the arguments of non terminals are used to hold special features of Arabic words such as the definitely or indefinitely determined feature of norms which is essential to recognize many structures. Also, the ability of Definite Clause Grammar to provide a general treatment of context sensitivity through the proper use of arguments of non terminals makes it easy to account for agreemeats such as number, gender and person agreements. Another important characteristic of l)efinite Clause Grammar, which suits Arabic, is the possibility of imposing extra conditions on the constituents of a phrase which must be satisfied for a rule to be valid [7]. 3. The develol,ed /ormal g,'ammar Due to the fact that there is IIO single basic word order for Arabic sentences, three basic sentence types were defined: a. Nominal sentence: a sentence that does not contain a verb or contains a verb which follows the subject. b. Verbal sentence: a sentence that contains a verb which precedes the subject. c. Sentences with special structures such as vocative sentences. Sentences are further classified functionally and according to their modalities. Larger sentences are also described in this grammar by conjoining sentences or embedding simpler sentences. We have used the grammatical categories together with the functional roles to define the syntactic structures. Grammatical categories alone were considered not sufficient to describe all the structures 345 since word categories in Arabic are globally classified into verbs, nominals and particles only. Nouns, adjectives and adverbs for example are differentiated in tt'te sentence according to their functions. This dual formulation increases the descriptive power of the grammar. In order to use this dual formulation to define the syntactic structures, some non standard definitions of grammatical categories were used. A nominal sentence (ns) is defined formally as composed from a 'mobtadaa' phrase, mbp, and a predicate phrase, predp: ns(ns(MBP, PREDP)) -~ mbp (N,G,m, -., MBP), predp (N,G,n,PREDP). A mbp is defined as a noun phrase (np) that can have 'motaalkat' which are either a prepositional phrase (PP) or adverbial. A predp can be one of the following: • a noun phrase that can have 'motaalkat' • a prepositional phrase or adverbial - a sentence which can be : - a verbal sentence - a nominal sentence (under certain conditions). Verbal sentences (vs) can be followed by either a subject phrase, sp, or a subject phrase and a complement phrase, cop: vs(vs(VP))-~ vp(T,V,S,I',VP). vs(vs(VP,SP,COP)) + vp(T,a,S,P,VP), sp(-,-,S,-,P,SP),cop(M,COP),{M=<T} cop(o,cop([ 1)) -~ I 1. For verbs in the passive voice the sp is replaced by 'naabfael' phrase (nsp), which can be either a np or pp. For non-terminal argument definitions the reader is referred to sections 3.1.1., 3.1.2 and 3.1.3. Tlle values of the arguments indicated above are: a for active voice, n for indefinite determination, m for definite determination and - for acceptance of any value. Verb phrases (vp) are defined as follows: vp --~ (cop), (particle), verb, (cop). whereas noun phrases (rip) are defined as follows: np -~ (particle), noutL (postmoditiers). Subject phrases are noun phrases whereas complement phrases modify the verbs and are one or more noun phrases (such as tile direct object) and/or prepositional phrases. The cop category is used to cater for the cases where the sp separates the verb from some of its modifiers, which is a characteristic of Arabic syntax. Sentences with special structures haw,, different structures which vary from one case to anotller. Sp and cop can be relative pllrases and in some cases the predp can precede the mbp. Categories with no equivalent in English are indicated between single quotes and categories inside the brackets are optional. 3.1. Non Terminal mgu,tents Non-terminal arguments are used to hold features of Arabic words necessary for recognizing some structures and for allowing agreements between the different costituents of the sentence. Arguments are introduced in the lexical entries of words and are inherited by the phrase in which the words are constituents, h't this grammar, arguments are defined as fol- ]OWS: 3.1.1. Arguments associated With verbs T transitivity, V voice, P person, S semantic feature. 3.1.2. Arguments associated with nouns N number, G gender, D definite or indefinite determination, I:' person, S semantic feature. 3.1.3. Arguments associated with pronouns N number, G gender, P person. M is an argument associated with senre non-terminals to test the possibility of modifying a verb witll objects. 3.2. Use of semantic features to reduce ambiguity in Arabic senstences. Semantic features art, used with verbs 346 and nouns in order to test the subject verb semantic agreement. The use of the described semantic features in the developed grammar helps in the diifferentiation between the subject and the object in Arabic sentences, since it is not possible on purely grammatical basis (except in some exceptional cases) to differentiate between them. 3.3. Examples In this section, examples of the syntactic structures of a nominal sentence and a verbal sentence according to the developed grammar are given. 3.3.1. Example of a nominal sentence: Which means: the big dog is in the garden. In arabic, the definite article is a prefix, there is no copulative verb and direction of writing is from right to left. SIS /\ pF p 7 p pp np ~ p ad} noun I I I I the garden in big the dog 3.3.2. Example of a verbal sentence: Which means: the boy drinks the water in the house v,~ cop sp .- cop I PP /\ np np np p I r I the water the child the house in v L drinks 4. Conclusiml This grammar has been implemented in a syntactic analyzer [8] developed in Prolog on a 1'C/XT-286 and tested on a large number of Arabic sentences. It is hoped that the developed grammar will contribute to the efforts undertaken recently in the Arab countries towards the development of a comprehensive, agreed upon, formal grammar for Arabic syntax. References [1] Bakir M. 'Aspects of clm,se structure in Arabic: a study of word order variation in literacy Arabic', Ph.d., Indiana University, 1980. [2l AI-Khuli M.'A contrastive transformational grammar: Arabic and English', Leiden: Britl, 1979. }3] Ayoub G.'Structure de la phrase verbale en Arabe standard', Ph.D., in: Analysis/Theories, 1981. 14l Fehri F. 'Complementation ct amophore en Arabe moderne: t,ne approche lexlcale fonctionnelle', These de Doctorat d'Etat, Univcrsitc dc Paris 3, 1981. [5] Owens J.'Structure, (:lass and Dependency: Modern linguistic them y and the Arabic grammatical tradition', in: Lingua 64, 198,1. [6) Mout(makil A. 'Pragmatic functi(ms in a {unctional grammar of Arabic', Dordrecht: Forts Publicatiolls, 1989. [71 Pereira F., Warren D 'Definite clause grammar for language analysis- A survey of the formalism and a comparison with transition networks', Artificial Intelligence, Vol. 13, pp. 231 - 278, 1980. [8] Hisham E1-Shishiny. 'A syntactic analyzer for Arabic sentences', IBM-CSC Technical Report # 32, 1989. 347