Morphology For Marathi POS-Tagger Veena Dixit 11/ 10 /2005

1
MorphologyForMarathi POS-TaggerVeena Dixit
11/ 10 /2005
2

• Contents
• Word
• Morphology
• Marathi Morphology - definition of the task and
  difficulties thereto.
• Marathi Morphology - solutions to the challenges
• Different word classes
• Postpositions
• Particles
• Interjections
• Conjunctions
• Pronouns
• Adjectives
• Adverbs
• Verbs
• Nouns

3

• Words are the orthographical strings separated by
  spaces and some punctuation marks.
• To syntax, words make sentences and to
  morphology, word has internal structure and has
  different inflectional forms.
• Inflectional forms of a root word form a
  paradigm based on a principle.
• Root word is the form which is stored in lexicons
  / dictionaries.

4

• What is Morphology?
• Morphology is the study of forms of words in the
  language, especially the different forms used in
  declensions, conjugations, and word building. It
  deals with the morphemes.
• Morpheme is a term which refers to the smallest
  component of a word that (a) seems to contribute
  some sort of meaning, or a grammatical function
  to the word to which it belongs, and (b) cannot
  be decomposed into smaller morphemes.

5
Marathi MorphologyDefinition of the task and
difficulties thereto

• Morphological analysis of Marathi plays
  significant role in natural language processing
  because Marathi, a pan Indian Language, is rich
  in morphology.
• Marathi, being the language of the area situated
  centrally, gets influenced by almost all language
  groups of India.
• This makes the Marathi morphology more
  complicated.

6
Marathi Morphologysolutions to the challenges

• Morphological analysis is done category wise.
• Parameters for changes in the root word are
  identified.
• Rules are constructed in the tabular form to
  facilitate computation.

7

• Marathi Word Classes
• Nouns
• Pronouns
• Adjectives
• Verbs
• Adverbs
• Postpositions
• Conjunctions
• Interjections
• Particles
• Punctuation Mark

8

• Postpositions
• Postposition is the morpheme that follows the
  words and shows the relation between the word
  that is followed and other word in the sentences.
• Case markers and shabdayogi avyaya are classified
  as postpositions in Marathi because they show
  same behavior.
• (ref. Classification of Words, Veena Dixit,
  proceedings of 26th AICL, Shilong, 2004 )

9

• Postpositions (continued)
• In Marathi, postpositions are attached to all
  classes of words except interjection. examples
• When a postposition is attached to a stem it
  produces mainly adverb, but also, adjective and
  conjunction.
• Postpositions are handled along with other word
  classes.
• 5 subgroups of postpositions are identified on
  the basis of what is the possible order of their
  attachment and to which group of words they can
  be attached.

10

• Particles
• Strings like ?? hi_also, ? cha_only, ??????
  suddha_also, are
• sometimes attached to other words (e.g.. ????
  khaali _under ?????????? - khaalisuddhaa_under
  also / ??? - jhaaDa _ tree - ????????? -
  jhaaDasuddhaa _ tree also )
• or sometimes they are written separately (e.g..
  ???????? - jhaaDaakhaali_ under the tree
  ???????? ?????? - jhaaDaakhaalisuddhaa_ under the
  tree also).
• When such words are attached to other words, the
  word to which it is attached, does not get
  inflected.

11

• Interjections
• Interjections are identified from the lexicon and
  stored to produce the tag.
• Conjunctions
• Conjunctions are identified from the lexicon and
  stored to produce the tag.
• Morphology also plays a role in the case of
  conjunctions.

12

• Conjunctions (continued)
• When some of Marathi postpositions are attached
  to a pair of demonstrative pronouns, they produce
  a pair of conjunctions in some instances.
• ?? ????????? (jo jyaapaasuna --- which from
  which)
• ?? ????????? (to tyaapaasuna --- that from
  that)
• ????????? ??? ??????? ????, ????????? ?? ??????
  ??????? ?????? ???. jyaapaasuna kaala suruvaata
  keli, tyaapaasun aaja nakkicha suruvaata
  karaayalaa nako_One should not start from the
  (same point) from which it was started yesterday.

13

• Pronouns
• Number of inflected forms of a pronoun and the
  rules describing such inflection are almost equal
  in number.
• Number of pronouns and their respective inflected
  forms are finite and less when compared to verbs
  and nouns.
• All inflected forms of the pronouns will be
  stored to produce the tag for pronoun.
• Derivational morphology of pronoun is handled
  with rules.

14

• Pronouns (continued)
• Inflectional forms of pronouns act either as
  adjectives (???? maajhaa_my) or as adverbs (???
  malaa_to me ) or as conjunctions (??
  ????????? (jo jyaapaasuna --- which from
  which) ?? ????????? (to tyaapaasuna --- that
  from that)).

15

• Pronouns (continued)
• All together 29 pronouns have 526 inflectional
  forms, which are either words or stems.
• 21 paradigms are identified generating several
  rules.

16

• Adjectives
• Adjectives are mainly, inflectional and non -
  inflectional.
• Adjectives inflect for gender, number and
  attachment of postposition to the noun they
  modify.
• Adjectives in Marathi agree in gender and number
  with the nouns they modify.

17

• Adjectives (continued)
• All inflectional adjectives belong to one
  paradigm, which corresponds to several rules for
  generating inflectional and derivational forms
  from an adjective.
• Most of aa ending adjectives agree with
  masculine nouns and further get inflected
  according to the gender and number of the noun
  they modify. (????? / ????? / ????? /
  ???????_mokaLaa / mokaLi / mokaLe /
  mokaLyaa_empty)
• There are some exceptions to this rule, such as,
  (???? - jaada_extra, ???? naanaa_different,
  ???? vaayaa_wasted).

18

• Adverbs
• Adverbs are mainly, inflectional and non -
  inflectional.
• Adverbs inflect for attachment of postpositions.
• ???? (khaali_under -- ???????? khaalapaasuna
  _from the underneath)

19

• Verbs and Nouns will be discussed in next
  sessions.
• Thank you.
• Veena Dixit
• 11/ 10 /2005