next up previous contents index
Next: Constraints in the application Up: Methodology adopted in this Previous: Values pertinent to a

Logic relationships between values

Once the attribute-value pertinent set is circumscribed, another requisite of the provision of guidelines is the detailed description of the logic relationships between the values of the system pertinent to the language in analysis, in the cases in which the values belong to different levels.

This point is, in our opinion, particularly crucial. As already explained in the Morphosyntax document, in EAGLES all the lexicon specifications constituting the proposal have been collected together in a large repository, articulated on different levels of obligatoriness. They constitute a redundant inventory of the information which, in a bottom-up approach, have been found actually encoded in existing lexicons of different languages for different purposes. This set cannot be implemented as such, but each language has to pick up the relevant values, specifying its system and internal constraints.
The language specific applications presented in the EAGLES document are integrating part of the proposal, in the sense that constitute the required instantiation and implementation of the proposal itself. In these applications, the relations between values can be deduced on the basis of the semantics and the actual exemplifications. This, however, can cause misinterpretation.

In this Validation task, for each language, a formalized system readily has been provided. The aim is to obtain a structured inventory, specifying the logic relationships between the values, in order to make others able of consistently interpret their semantics itself.

Following a first suggestion carefully explained in a report by Teufel 1995 an example of how the tables display these relationships will be given here.

Let's assume that the Case system displayed in the EAGLES proposal is our universe and the values the objects whose relations are to be formalized.

The case system proposed for Nouns at the level EAG-L2 contains the values nominative, genitive, dative, accusative. For any other value presented at a different level, the relationships entertained with the objects of this universe should be specified, as the points below show.

The Case system proposed for the Pronoun PoS foresees the oblique value:

Let's consider another typical relationship. At the L1 of the EAGLES proposal, the Gender universe is split into three values masculine, feminine, neuter and the Number between singural, plural.

In the tables, these interrelationships can be expressed for each language, by using indices added to the values in analysis.

Practically, (see table 2.1 for Nouns) the value set masculine, feminine, common, highlighted as the Italian Gender universe (neuter is not relevant), will receive the following indices:



The values feminine and masculine are given a numerical index - 1 -; common, the value added to the EAGLES L1 at the language specific level L2b for Italian, receives the same numerical index pointing to the values it disjoints, and is further labelled with the index D for the disjunction relation.

We can use, for example, D for the disjunction relation, R for the replacement relation, A for the addition relation. Please note that this last does not need numerical indices, since the values are simply added to the system.



next up previous contents index
Next: Constraints in the application Up: Methodology adopted in this Previous: Values pertinent to a