An Attribute-Sample Database System for Describing Chuvash Affixes

In the paper is described “KÜLEPEK” – a database system created by the author for description of Chuvash word-changing and word-forming affixes. The system is based on the attribute-sample model of affixes, which allows to describe affixes and their phonological, morphotactic and orthography rules. The system uses DBase IV database engine and was created in Borland Delphi 7.0 software environment, has a user friendly interface and can run under Windows 98/200/ME/XP. It is applicable for a large number of agglutinative languages with a finite-state morphology and can be used as a part of a morphological parser, and as an independent reference tool as well. 1. The Attribute-Sample Database System’s Common Structure and Description The attribute-sample model of morphology (Zheltov, 2003) is based on following principles: 1) division of affixes to types; 2) conforming to each type a number of patterns, which describe phonological, morphotactic and spelling rules. Chuvash language, being an agglutinative one, is mainly based on affixes and their interaction with stems and each other. In Chuvash morphemics are widely present such phonological variations (Sergeyev, 1992) as: 1) quality synharmonism “soft” (front vowel) stems agglutinate “soft” allomorphs, while “hard” (back vowel) stems agglutinate “hard” allomorphs. As a rule each affix has minimum two allomorphs – a “soft” one and a “hard” one: anne “my mother” + e (dat.-gen. affix) = annene “to my mother”; laşa “horse” + a (dat.gen. affix’s allomorph) = laşana “to horse”. But some affixes have only “soft” allomorphs (like 3rd pers. sing. possessive affix -ĕ/-i “his/her”), thus a distorting of. synharmonism is observed sometimes: šırăvĕ “his/her letter”. 2) interphonemes insertion: anne “my mother” +e (dat.-gen affix) = annene “to my mother/my mother”. 3) vowels reduction (elisia) a) in word formation: vat šın < vată šın – “old man”, purnăš < purănăš – “life”, from purăn – “to live” b) in word changing: vula (“to read”) + ăp (future tense 1-st pers. sing. affix) = vulăp “ (I) shall read”. 1 Chuvash language belongs to the Bulgar group of Turkic languages, together with extinguished Bulgar and Hazar and counts near 1,6 millions of speakers, their main part lives in Chuvash Republic and Volga region of Russian Federation. It is considered an endangered one and you can read more about the situation in (Zheltov, 2005). 4) consonants reduction: in ten verb stems, ending on -r, r is falling out in some verb forms: pır (“to come”) + -t(past tense affix) + -ăm (1-st pers. sing. affix) = pıtăm “(I) have come”. 5) consonants duplication in noun stems, ending on ă/ĕ, combined with final vowel reduction, when placed to dative-genitive case: tulă “wheat” + -a (dat.-gen. case affix) = tulla. sĕlĕ “oats” + -e (dat.-gen. case affix) = sĕlle. 6) final vowels alternation: u – ăv, ü – ĕv. šıru “letter” + ĕ (3-rd pers. sing. possessive affix) = šırăvĕ “his/her letter”, vĕrenü “studying” + ĕ (3-rd pers. sing. possessive affix) = vĕrenĕvĕ “his/her studying”. The n in the example above as well as in laşana is an interphoneme, placed when a stem ends on a vowel and the agglutinating to it affix also begins on a vowel. Interphonemes are also being used in Chuvash in some other cases. From this point of view Tatar language (a neighbour Turkic language of Kipchak group) has also two allomorphs of dative case: -a, -ə, while its others allomorphs -ga, -gə, -ka, -kə, -na, -nə are compound ones, decomposed to the allomorphs -a, -ə and interphonemes g-, -k-, -n-. But from the formal point of view the representation accepted by Chuvash linguists is more comfortable, especially for computer analysis. The database system has an interface structure, consisting of 6 tables that can be optionally filled for each affix: Affix Allomorphs Morphologic feature Type