If a term database for spoken language is to provide both onomasiological and the semasiological query perspectives, the database system has to fulfil basic constraints with respect to the at least the following two general criteria:
The design of EAGLET incorporates the premise that all terms are uniquely accessible through a commonly shared key attribute . In general, the key attribute is defined as the orthographic representation; however, conceptually the two attributes are distinct. In cases of homography, for example, two entries may be required, each with the same orthography; if the orthography were simultaneously the key, then the key would not uniquely distinguish the entries.
The key is defined more generally as a key root which is identical to the orthographic representation, and a key extension, which is taken from a set of numerals (usually consecutive numerals starting with `001').
The definition of EAGLET as a relational database system follows.