The article contends that it is hardly defendable to identify the syntactic function of a dependent clause according to the semantics of its conjunction (as is the practice of the Latin handbooks currently used at the universities in the Czech Republic). For instance, the dum-clauses are generally taken for temporal (adverbial) clauses according to the temporal value of the lexeme dum; with the verb exspecto, however, these clauses function as a direct object and are not temporal (adverbial).
Similarly, the prototypically conditional si-clauses and prototypically temporal cum-clauses can function as the subject or object of a sentence. Moreover, a dependent clause should be taken not as a syntactic function (subject, object, adverbial) at all, but rather as a construction which can itself have a syntactic function in a sentence, but can also be only a component of a syntactic function (e.g. quam + dependent clause) or can be a part of the so-called nonsententials, i. e. structures, that lack the traditional sentence functions as object, subject, etc. at all (e.g.
Quid + dependent clause?).