This website proposes a structure for the grammatical description of junction and complex sentence formation.

Basic concepts

The field identified as complex sentence construction in semasiological perspective corresponds to the domain of junction in onomasiological perspective (s. the two perspectives).

The notion of a complex sentence thus presupposes the notions of clause and sentence. A clause is a linguistic unit consisting of one predicate and its dependents. The sentence is the smallest linguistic unit which carries an illocutionary force.

The notion of junction presupposes the notions of proposition and of interpropositional relation. A proposition is the semantic representation of a clause. Its mental correlate is a thought. It consists of a combination of a predication with a reference. An interpropositional relation is a syntagmatic relation between two propositions. We will see that some syntagmatic relations are peculiar to propositions while others combine propositions as well as other entities (while yet others combine entities that are not propositions). In a dynamic view, a proposition may be expanded into a complex proposition, and conversely the latter may be condensed to a simple proposition.

Needless to say, the extensions of the two concepts ‘complex sentence’ and ‘nexion’ do not coincide (as structural and functional concepts in linguistics rarely coincide). As a consequence of incomplete overlap of the extensions of the two concepts, there are (structurally) complex sentences which do not code junction; and there are (semantically) complex propositions not expressed by a complex sentence. These are problems of delimitation of the concepts involved which will be taken up in the next section.

The “grammar of the complex sentence” therefore results from the combination of a certain area of structural grammar with a certain functional domain. Its two main sections are consequently

  1. Onomasiological approach: nexion
  2. Semasiological approach: complex sentence construction.


In either of the two approaches, problems of delimitation arise. At the language-particular level, functions are associated with structures. Thus, whenever a given structure is associated with a function that is outside the domain of junction, and whenever a certain function is not fulfilled by a complex sentence as defined above, the topic does not belong in the grammar of the complex sentence of that language. At the interlingual level, there are no fixed associations of functions with structures. Therefore such delimitation problems cannot be resolved at this level. The only thing we have at this level are prototypical and less prototypical function-structure associations.

a) In the semasiological perspective, problems of delimitation arise, at the high pole of the complexity continuum, between a combination of clauses into a sentence (i.e. a complex sentence) and a combination of sentences into a paragraph (a piece of text). A criterion deducible from the definition of ‘sentence’ that excludes a paragraph-level sequence of clauses from the status of (complex) sentence is the illocutionary force: if those clauses differ in illocutionary force, they do not belong to one sentence. If they have the same illocutionary force, structural criteria such as intonation may come in.

Rules of the linguistic system concern the grammatical levels up to the sentence level. At the text level, there are principles of coherence and conventions of structuring a text of a certain genre which have to do with logic and rhetoric, but not with the linguistic system. There are also principles of cohesion which concern semantics, but not grammar. Nevertheless, the border between the combination of clauses into a sentence and the combination of sentences into a piece of text is not sharp. The principles of cohesion, including importantly anaphora, are the same; and the sets of connectives overlap. To the extent that rules of the linguistic system reach out onto the text level, the domain of junction will be expanded to comprise them. Thus, no sharp borderline is intended at the high pole of the complexity continuum.

Information structuring is a distinct functional domain, which however generates syntactic complexity of its own, in the form of cleft-sentences, pseudo-cleft-sentences and left- or right-dislocated topics which may be clausal. Such constructions do not fall into the domain of junction. A cleft-sentence, for instance, satisfies the structural criteria for a complex sentence but not the semantic criteria of junction, as no interpropositional relation between its propositions is involved.1 More on this elsewhere.

At the low pole of the continuum, problems of delimitation arise for periphrastic verb forms and for verb series. A periphrastic verb form is only one verb form; consequently, no complex sentence is involved by definition. However, periphrastic verb forms arise by grammaticalization of complex sentences, so there are borderline cases. Verb series may be formed analytically, i.e. according to rules of syntax, out of full lexical verbs; then they do form complex sentences. However, a verb in a (binary) series may also be grammaticalized to a coverb and further into an adposition or function verb; and then the clause becomes a simple clause. Or again a verb series may be lexicalized into a compound verb, the clause thereby ceasing to be complex. In all such cases, the continuousness of grammaticalization and lexicalization renders a clear delimitation of structural concepts impossible (and, fortunately, unnecessary).

b) In the onomasiological perspective, problems of delimitation are presented by constructions that do involve two propositions but do not display syntactic complexity as defined. For instance, an attribute may take the stead of a predicate, as in the ‘ab urbe condita construction’, of which an instance appears above in the expression as a consequence of incomplete overlap. Here a structural technique (attribution) whose locus is in the domain of concept formation is exploited in the domain of junction.

Again, a proposition which depends on another may be reduced, by some operation of desententialization, to a phrasal (rather than clausal) clause component. Each of the two synonymous variants of comprises two propositions, connected by a temporal relator. The construction consequently falls in the domain of junction.

.a.That happened after Donald Duck was president.
b.That happened after Donald Duck's presidency.

However, only the #a version is a complex sentence. The #b version just contains a temporal adverbial. These examples suffice to show that the grammar of junction and the grammar of complex sentence construction do not coincide.

1 If the semantic analysis of clefting and topicalizing structures is spelt out (cf. Lambrecht 1994), it does involve more than one proposition even for such constructions that bear a clear paradigmatic relation to a simple clause. The semantic criterion by which such complex propositions may be kept out of the domain remains to be made precise.