In the Prague Dependency Treebank, a part of the texts from the Czech National Corpus is being annotated on several layers, including the underlying (tectogrammatical) representations. The usefulness of such a treebank is briefly characterized and a large set of topics is discussed for which further monographical research appears to be necessary. The future discussion and elaboration of these topics can be carried out much more effectively with the use of the annotated corpus, and the results thus gained may then serve to an enrichment of the descriptive framework and of the annotation procedure.
Individual values of syntactic and morphological attributes used in the syntactic annotation of the Prague Dependency Treebank are discussed, as well as certain more general issues concerning the relationships between the underlying and the morphemic levels (word order, deletion and others).
The structuralistic point of view seems to be applicable in the sociolinguistics as far as the relation of language norms and varieties is concerned. Language norms can be classifiable as components of human consciousness, the function of which is to regulate language expectations and actions. These norms reflect social and language phenomena. Varieties are coherent collections of language elements which can be distributed according to geographic, social or functional criteria. The relation between norms and varieties is that between invariant and variant elements. These units facilitate to operationalize the sociolin-guistic perception of language-users. Invariant constructs with their sociolinguistic relevance (segments of norms) are realizable in various ways, which depends on various factors determining communicative situations. The author suggests to coin the term normeme for the realizable unit and allonorme for the realized variants (segments of varieties).