Integral Part Of Natural Language Processing English Language

Essay add: 24-10-2015, 21:50   /   Views: 373

Abstract:-Machine Translation is one of the central areas of focus of Natural Language Processing where translation is done from Source Language to Target Language preserving the meaning of the sentence. Large amount of research is being done in this field. However, research in Machine Translation remains highly localized to the particular source and target languages due to the large variations in the syntactical construction of languages. Inflection is an important part to get the correct translation. Inflection is basically the adding of appropriate suffix to the word according to the sentence structure to obtain the meaningful form of the word.

This paper presents the implementation of the Inflection for English to Marathi Translation. The inflection of Nouns, Pronouns, Verbs, Adjectives are done on the basis of the other words and their attributes in the sentence. This paper gives the rules for inflecting the above Parts-of-Speech.

Keywords: Natural Language Processing, Machine Translation, Parsing, Marathi, Parts-Of-Speech, Inflection, Vibhakti, Prataya, Adpositions, Preposition, Postposition, Penn Tags.

INTRODUCTIONMachine translation, an integral part of Natural Language Processing, is important for breaking the language barrier and facilitating the inter-lingual communication. Marathi, an Indo-Aryan language derived from Sanskrit, is spoken by 70 million people in India. The script currently used in Marathi is called "baalbodha" which is a modified version of Devnagri Script [1]. While translating one language to another changing of the word order and its form according to the grammar of the target language is very important. For the scope of this paper the Source Language is English and Target Language is Marathi.WORD ORDERING IN LINGUISTICSThe syntactic structure of a language is determined by the word order.

Words are classified into 8 parts-of-speech (POS) [noun, pronoun, adjective, verb, conjunction, preposition, interjection]. The arrangement of these POS in sentences is determined according to the structure the language follows. English follows Subject-Verb-Object (SVO) structure while Marathi follows Subject-Object-Verb (SOV) structure [2].Along with Marathi other Indo-Aryan language like Hindi, also follow the SOV structure.IMPORTANCE OF ADPOSITIONS IN LINGUISTICSAdpositions are words which can occur before or after a phrase, word, or a clause that is necessary to complete the meaning of a given sentence.

Adpositions are mainly categorized as:PrepositionsPostpositionsCircumpositionsPrepositionsPrepositions are defined as the words placed before the complement [3]. Prepositions are used in English.Example:I value my family above everything else.PostpositionsPostpositions are words which come after the complement.[2] Postpositions are used in Marathi, Hindi, Urdu, Korean, Turkish, and Japanese.Example:Ekh ckdh izR;sd xks'Vh is{kk ekb&;k dqVqackyk tkLr egRo nsrks .CircumpositionsCircumpositions are words that appear on both sides of the complement. They are used in English, Dutch, Swedish, and French.Example:I will exercise regularly from now on.The languages which follow SOV Structure use postpositions.

Hence, while translating an English sentence.(SVO structure) to a Marathi sentence (SOV structure), we need to change the prepositions (of English) to postpositions (of Marathi). This is a major issue which needs to be resolved for inflecting the nouns, verbs and cases (Vibhakti).INFLECTIONGenerating inflection of a word is important to retain the correct form of the word in Marathi. Words can be classified in two types depending on the Inflection as [1]:Inflectional Words:NounPronounAdjectiveVerbNon-Inflectional Words:AdverbPrepositionInterjectionConjunctionThe words are inflected on the basis of changing Gender (Masculine, Feminine, Neuter), Multiplicity (Singular, Plural), Tense (Present, Past, Future), and Case (Nominative, Accusative, Instrumental, Dative, Ablative, Genitive, Locative, Vocative).Noun InflectionNoun inflection is performed on the basis of change in Gender, Multiplicity or Case (Vibhakti).

The inflection of a word can be determined from the word endings. Following table describes the word endings and its inflections.TABLE ITERMINATING VOWEL OF ITS ROOT [2].

Terminating Vowel of the Root

Plural Inflection

Masculine

Feminine

Neuter

vNo changevk bZ

,

vk

,

vk

…

bNo changeNo changeNo changebZNo changevk ;k

,

mNo changeNo changeNo changemwNo changev

,

,

…

vkbZZ,s

…

vk

…

vksNo changevk

…

vkSNo change

…

…

The above table will be found helpful in determining the plural form of a noun by terminating vowel of its root. For instance, the plural form of "ck;dks" must be "vk" making up "ck;dk" as "vk" stands opposite to the vowel "vks" in the column super scribed Feminine[2].The word "xk;" must be inflected to "xk;h" as "v" would be replaced by "bZ". Another example of a word-ending in "bZ", "ik;jh" would get inflected as "ik;j~;k" wherein ";k" is the word ending.Case is an inflected form of Noun by which its relation to other words in the sentence is indicated.For example:He drove the car."R;kus xkMh pkyoyh"TABLE IICASE TERMINATIONS [2].

Case (Vibhakti)

Singular

Plural

Nominative

---

---

Accusativel ] ykl ] ukInstrumentalus] f"kuh f"kDativel ] ykl ] ukAblativegwu mwugwu mwuGenitivepk]ph]pspk]ph]psLocativerrVocative

---

uksVerb InflectionWhen we translate the verb using Marathi Dictionary we get the gerundial form i.e it is given with the particle ".ks".Example:"[ksG.ks" - to playFor inflecting the verb, first we need to derive the verbal root (/kkrq) and then add personal endings to it, to indicate its relation to the noun. We can get verbal roots by dropping the particle ".ks" from the gerundial form Inflection of the verb depends upon the following particulars:The Gender (fyax): Masculine, Feminine and Neuter.The Number (opu): Singular, Plural.Person (iq#'k): First, Second and Third.Tenses (dkG): Present, Past and Future.Sometimes personal endings may also depend on moods (vFkZ), the constructions (iz;ksx), the participle and the verbal nouns (/kkrqlk/khrs) [2].The table for verb inflection is given below (Table no. III).TABLE IIIRULES FOR DHAATU TO VERB [1].Adjective inflectionAdjective is a verb which is joined to a noun to qualify it.

Inflection of adjective depends upon gender, multiplicity, attachment of postpositions to the noun modified by such objective. When genitive case makers or some prepositions are attached to nouns, it produces adjective [4].TABLE IVADJECTIVE TERMINATION FOR "vk" [4]

Changing part in masculine form

Feminine

Neuter

Oblique form

vkbzzZ

,

;kPronoun inflectionA pronoun is a word that can be substituted for a noun or a noun phrase.Pronoun inflection is similar to noun inflection but there are some special cases which need to handle seperately. For the verbs such as "like", "want", "will", "need" and "would", the pronoun inflection are different than general cases. Some sentences have the same structure alongwith the parse tree, gender, multiplicity, cases but have different translation of pronoun.

These sentences will typically include the above mentioned verbs.Following table determines the translation of pronouns if the sentence has any of the verbs "like", "want", "will", "need" or "would".TABLE VEXCEPTIONAL PRONOUN INFLECTION.English wordMultiplicityMarathiISeykWePvkEgkykYouSrqykYouPrqEgkykHeSR;kykSheSfrykTheyPR;kaukFollowing cases gives a comparison of different cases where the same pronoun is used but has different inflections depending upon the verbs used in the sentence.

Case 1 :( First person singular)

Example 1:

English sentence: I play football.

Parse Tree:

Marathi translation(After inflection):eh [ksGrks QqVckWy++-

Example 2:

English sentence: I like mango.

Parse Tree:

Marathi translation(After Inflection):eyk vkoMrks vkack-

Case 2:(first person plural)

Example 1:

English sentence: We are friends.

Parse Tree:

Marathi translation(After inflection):vkEgh vkgksr fe=-

Example 2:

English sentence: We need books.

Parse Tree:

Marathi translation(After inflection):vkEgkyk ikfgts iqLrd-

Case 3: (Second Person Singular)

Example 1:

English sentence: What do you do?

Parse Tree:

Marathi translation(After inflection):dk; rq djr vkgs

Example 2:

English sentence: What do you like?

Parse Tree:

Marathi translation(After inflection):dk; rqyk vkoMra

Case 4: (Second person plural)

Example 1:

English sentence: You should do this.

Parse Tree:

Marathi translation(After inflection):rqEgh ikghts djk;yk gs-

Example 2:

English sentence: You will get this.

Parse Tree:

Marathi translation(After Inflection):rqEgkyk feGsy gs-

Case 5: (Third person singular)

Example 1:

English sentence: He

Article name: Integral Part Of Natural Language Processing English Language essay, research paper, dissertation