Luciana Peev, Lidia Bibolar, Jodal Endre * A Formalization Model of the Romanian Morphology



The Level 2 sub-class is defined by the following grammatical categories:

The Level 2 Sub-class is characterized by a name and the sum of flectives associated to the grammatical categories.

In order to comprise all the auxiliaries and irregular verbs, specific flexionary classes were created. These do not regard the irregularities of the root but only the irregularities of the flectives in relation to the existing classes. The number of specific classes is very small and it refers to auxiliary verbs (a fi to be, a avea to have, a voi to want) and the irregular verbs together with their derivates (a sti to know, a da to give). Due to this hierarchized classification, a verb considered irregular such as a lua (to take) becomes a common verb from group A, to which the class of verbs with [-ând] Gerund ending corresponds. This classification took into account only the flexionary part of the verb: the root presents itself as more or less regular, so it does not have any part in the classification. The problems of the root are presented under 3.4.

This classification was used to create a database of the Romanian language and it helped us in working with the verbal derivatives. We do not deal with this problem because of the lack of space.

After the study of suffixes and prefixes of the Romanian language, the computerized thesaurus will be enriched with the classes corresponding to them.

3.3. The determining of flexionary classes for names

Gender is a criterion taken into account in grouping names. The flexionary classes within the name groups are determined by the flectives that can be associated to a name at a certain moment.

The flexionary class of the name is defined by the following grammatical categories:

and are characterized by a name and the lot of flectives associated to grammatical categories. The grammatical category of gender is explicitly associated to each name and it is not part of the grammatical categories defining the flexionary classes. This approach allowed us to solve the problems of those nouns displaying forms of masculine and feminine.

3.4. The determining of roots coding

In establishing the flexionary classes, the variation of the root was not taken into account, but it can suffer partial variations due to phonetic alternations or to its irregular character.

Phonetic alternations can be consonantic:

e. g.:

d/z: brad/brazi; cad/cazi
z/j: obraz/obraji;

and vocalic:

e. g.:

e/ea: ales/aleasa;
e/a: fiert/fiarta.

Although there are very precise linguistic rules referring to the alternations, there is a considerable amount of exceptions. In order to approach equally all forms of variations, we adopted a system of root coding in a unique, condensed root. This codification system can determine all the forms of the roots, starting from the condensed root and the corresponding flexionary classes.

Ex.: the verb "a absorbi" displays two roots:

Note: in some linguistic applications there might be necessary only the condensed information, that is a coded root and a generic flexionary class (e. g. the generation of flexionary forms or automatic translation). The codification system of the root is based on an idea taken from the research of Acad. Gr. Moisil [7].

Ex.: the root of the verb "a absorbi" displays an alternation corresponding to the letter "o" (o/oa) and is coded as follows: "abso0r" where: "o0" represents the letter of variable value:



74

Previous Index Next