The linguistic architecture defines the basic objects of the model and their relations. It also specifies the general terminology which is common to the whole standard and used to talk about the dictionaries, their components and the interaction of these.
If the lists of features for each linguistic level and each language are not rigidly restricted (as is the case in MULTILEX), the linguistic architecture defines relations between blocks of features instead of between individual features.
Because most of the words described in a dictionary are quite regular and can be accounted for within a consistent framework while a considerable number of words are quite complex and variable, the linguistic architecture defines a robust basic structure and provides the means of building the additional apparatus when required.
Finally, the linguistic architecture guarantees that the model is consistent at the top level whatever the language. This idea is very important: in database technology, a distinction between meta-schemata, schemata and instances of schemata is made. The meta-schema defines general well-formedness conditions for schemata. Schemata then define the logical format of language-specific and level-wise linguistic descriptions. Instances are then individual lexicons for which there is a translation relation expressed between the individual format of the instance and the ``type'' defined by the schema.