Complex nodes are nodes representing autosemantic lexical units (nouns, adjectives, verbs, adverbs, numerals and pronouns). Autosemantic lexical units express morphological categories (in contrast to synsemantics, which are represented by nodes of other types). To be represented properly, autosemantic lexical units need (apart from a t-lemma and functor) grammatemes (the tectogrammatical correlates of morphological categories, see Section 5, "Grammatemes"), and possibly other attributes. The fact that complex nodes have grammatemes sets these nodes apart from all other types of nodes.
The t-lemma assigned to a complex node (representing a word present at the surface level) is usually the basic form of the relevant lexical entry (i.e. a sequence of graphemes representing the basic form). An exception to the rule are complex nodes representing personal and possessive pronouns: these make use of the t-lemma substitute
#PersPron (see Section 4, "T-lemma substitutes").
Also (newly established) nodes that are copies of other nodes (their t-lemma is identical to the t-lemma of the node that is being copied) and (newly established) nodes with the
#PersPron t- lemma, as well as those with the t-lemma substitute
#EmpNoun belong to the class of complex nodes.
Newly established nodes with the
#PersPron t-lemma (which is not different from the t-lemma corresponding to a regular pronoun, present at the surface level) occupy the position of non-expressed (deleted) personal pronouns (usually in the argument position). The information brought about by the morphological categories (of the pronoun), some of which are reflected in the verb's form at the surface level, has to be encoded in the value of the relevant grammatemes; this is mainly the information regarding the number and gender of the null pronoun.
Newly established nodes with the
#EmpNoun t-lemma are inserted as the mothers of nodes corresponding to adjectives that cannot have a nominal function themselves. Such a new node contains the information concerning some of the morphological categories of the dependent adjective (the values of its number and gender, see Section 6.1.2, "Definite pronominal semantic nouns: demonstratives"). Newly established nodes with the t-lemma substitutes other than
#EmpNoun belong to the class of quasi-complex nodes (see Section 8, "Quasi-complex nodes").
Complex nodes are divided into four basic groups, so called semantic parts of speech, which are further classified (see Section 2, "Semantic parts of speech"). The information about the semantic category of a complex node is encoded in the
sempos attribute (see Section 3, "Attributes superior to grammatemes"). It follows from the membership in a semantic (part-of-speech) category which grammatemes are relevant for a given node (i.e. by indicating the value of the
sempos attribute the set of relevant grammatemes is unequivocally determined).
Complex nodes can be modified by other complex nodes or by all other kinds of nodes with the exception of the technical root of a sentence (see Section 1, "The technical root node") and nodes with the
FPHR functor (see Section 5, "Nodes representing foreign-language expressions"). The order of the nodes dependent on a complex node is subject to the rules for the deep word order (see Section 3, "Communicative dynamism").