Luciana Peev, Lidia Bibolar, Jodal Endre *
A Formalization Model of the Romanian Morphology
By applying the codification system for the irregular verb "a
lua" we notice that
the verb displays three roots which determine the total alteration
of the root form "lu"/"ia"/"ie".
The method of variable value letters, used in the definition of
the alternations, is extended here and so the differing one is
no longer the letter, but the very root. The root can be considered
as a variation "l0", for which we define, just as presented
above, the following values:
- l0 ---> ia for Present Indicative I-s,
Present Subjunctive I-s, Imperative II-s
- l0 ---> ie for Present Indicative II-s, Present
Subjunctive II-s
- l0 ---> lu for the rest of the paradigm.
Once the root is defined, the corresponding generic flexionary
class can be established. The first level consists in establishing
the group: the verb "a lua" is part of the category
of I-st conjugation verbs, that is group A, and, if we take into
account the Gerund, which divides group A into two (luând),
it means that the corresponding flexionary group is A2; the second
level of classification is the one connected to the forms of the
verb in Present Indicative and Subjunctive, singular, III-p within
group A8. From now on we are no longer interested in whether the
verb is irregular or not, because the algorithm treating flexion
will interpret the root according to the corresponding grammatical
categories (mode, tense, person, number) and will pick up the
flective from the lot of class A1 and the subclass A8, according
to the grammatical categories.
Note: both the root and the flectives are established
according
to the classical theory and we consider that the lexico-morphological
thesaurus of Romanian language that we created can constitute
the basis of various applications dedicated to theoretical study
of Romanian language morphology.
3.5. The determining of data structures
The structure of an article is specific to the part of speech
it represents. In describing the structures, we shall mention
only those grammatical attributes that are valuable in solving
the problems of morphological analysis and generation. Each article
contains the element of identification in its primary structure,
represented by lemma, the coded root and the generic flexionary
classes.
| Noun representation:
| {lemma, root, ctg, clsflex, gender,...}
|
| Verb representation:
| {lemma, root, ctg, clsflex, ...}
|
| Inflexible parts representation:
| {lemma, root, ctg,..}
|
4. Dictionary generating
An interactive medium was projected in order to create the lexical
thesaurus on the basis of the enunciated principles. This medium
has integrated all the necessary functions in creating and verifying
the grammatical attributes put in, in exploiting and preserving
the primary data base, as well as in studying it with the means
of interrogating and leafing through interrogation and browse
functions. The basical dictionary of the resulted thesaurus does
not participate directly in any application. To perform a linguistic
application, one needs to define the problem and the necessary
resources. According to these, those dictionaries specific to
the application can be generated. This is realized by means of
the generators of dictionaries integrated in the work medium of
the computerized thesaurus of Romanian language that was realized
by our team.
For example, for an application of the recognizing type and of
morphological analysis for Romanian language there will be created
a specific dictionary in which for the verb "a absorb",
the roots and their corresponding flexionary classes will be generated:
absorb x1
absorb x2
In the case of a translation from another language into Romanian,
in which it is necessary to create a dictionary of correspondence
between the two languages, the verb "a absorb" will
appear this time in the following form:
abso0rb y1
The flexionary classes x1 and x2 in the first case and y1 in the
second one are generated according to the morphological data in
the thesaurus. The general diagram for the obtaining of specialized
dictionaries, according to the application, is the following:
|
| APPLICATIONS
|
| lexical thesaurus ---> dictionary generator --->
optimizer --->
| spelling checker
|
| morphological analyzer
|
| automatical translation
|
| ............
|
| other applications
|
75
