The Determiner in Bulgarian - definite/indefinite article in DATR
Velislava Stoykova
52, Shipchensky proh. st., bl.17
Bulgarian Academy of Sciences,
Institute for Bulgarian Language
1113 Sofia, Bulgaria
e-mail: vili@bas.bg
The presentation is focused on the use of DATR (language for encoding lexical knowledge) to represent the grammar category of definiteness/indefiniteness in Bulgarian language. The programming realization is made, and is available for demonstration using QDATR 2.0 (the University of Duesseldorf's DATR implementation).
The grammar category of definiteness in Bulgarian has two aspects - syntactic and semantic.
The syntactic aspect of definiteness is concerned with its formal morphological marker, the so-called definite/indefinite article, represented in form of different endings for the words, depending on their number and gender. This grammar category in Bulgarian includes: the nouns, the adjectives,
the possessive pronouns, the reflexive-possessive pronouns, and the numerals (both cardinal and ordinal).
The syntactic realisation of definiteness is primarily concerned with its semantics, mainly to denote objects. It is approved also by the fact that the morphological form of the Bulgarian definite/indefinite article originates from the Old Bulgarian demonstrative pronouns. Hence, it can define the context
when there is lack of other context-definable conditions (for example - in sentences without verb).
The main syntactic function of the definite article in Bulgarian is to denote the subject in the sentence since Bulgarian language has a relatively free word order, and there is no other way to denote the subject when it is not at the first place.
At the same time, the use of the definite/indefinite article is bound by some lexical semantic feature of some classes of the noun like proper and common nouns. Some other noun-classes like singularia tantum, pluralia tantum, and communia have a special morphological representation.
The representation of the Bulgarian noun - definite/indefinite article in DATR is based on the division of the nouns into 10 classes according to the formal criteria of gender and number.
The first criterion - gender was taken into account combined with the suffix into which a particular noun ends in plural. The evaluable DATR theory includes both the suffixes for definite/indefinite article for the genders and plural, and all suffixes for plural plus rules for their relevant morphological combination. The nouns themselves are given as a knowledge about this theory consisting only of information about their stems. The possible query to be evaluated is information about either singular, plural, gender, singular definite or singular indefinite.
The above representation is to be of use not only for language generation tasks but the reverse algorithm could be used for language understanding tasks as well.