Basic concepts

The Greek syndesis literally means ‘binding together’. It refers to the use of connectives (see below for this term) that link clauses with each other. Asyndesis is the absence of such connectives, i.e. the mere juxtaposition of clauses. This section introduces these connectives. They are an integral part of the syntactic constructions in the domain of nexion. The constructions themselves are not dealt with here, but in the other subsections of the chapter ‘complex-sentence formation’. Many of those constructions are subdivided, at some lower level of the conceptual hierarchy, by the parameter ‘syndesis vs. asyndesis’. For instance, the treatment of the English object complement clause will mention that, under certain conditions, the conjunction that may be omitted, as in I know (that) you are right.

In other words, explicitness of linking – syndesis vs. asyndesis and the particular connectives employed – is a low-level parameter in the semasiological approach. Complex constructions are not primarily classified by the connective employed (as in many traditional and early structuralist treatments), but by the other parameters of complex-sentence formation. From a structural point of view, a connective is one of a set of slot-fillers in a construction. The generic concept of the construction involves the slot; its filling by the particular connective is just an instantiation of that construction defined by the choice of an element from a paradigm.

For instance, a semasiological description of German complex-sentence formation comprises the following construction:

C1. Sentence-level negative contrast

[ [ ... [Neg A]C ... ]S1 [ Conn ... [ B ]C ... ]S2 ]S

C1 formalizes a certain type of adversative coordination of two independent clauses which are structurally parallel, but differ in the lexical choice (A and B, resp.) for a certain syntactic component C. The paradigm of Conn comprises aber ‘but’ and sondern ‘but’. In that slot, aber implies that the content of S2 somehow makes up for the failure of the content of S1, as in .a. Sondern implies that claiming or presupposing the content of S1 was wrong, and S2 is right, as in .b.

.a.Erna hat nicht ihre Hausaufgaben gemacht, aber sie hat das Unkraut gejätet.
GermanLinda has not done her home-work, but she has cleared the weed.
 b.Erna hat nicht ihre Hausaufgaben gemacht, sondern sie hat das Unkraut gejätet.
It is not the case that Linda has done her home-work; she has cleared the weed.


In the narrow sense, asyndesis is the combination of clauses without connectives. In a wider sense, it is a coordinative or paratactic technique at any syntactic level between the combination of words into phrases and the combination of sentences into pieces of text.1 Here we will concentrate on the combination of clauses, with a side-glance at the combination of sentences.

A sentence (i.e. one that corresponds to the norm) is characterized by a unified intonation contour. To the extent that the latter's shape is part of the linguistic system, it is a basic structural means of clause linkage, one that remains even with asyndesis, i.e. in the absence of any other structural means coding the unity of the sentence. The intonation contour, however, does not code any specific kind of relation between the clauses; it simply guarantees the structural unity of the sentence.

The combination of clauses into a sentence normally presupposes some interpropositional relation between them. In asyndesis, this is by definition not coded. It is inferred from the meaning of the clauses, from the context and from world-knowledge. Thus, in a sentence like we know that the interpropositional relation involved is causal.

.I could not come earlier, the train was late.

What is important here is that the causal relation is not part of the meaning (i.e. significatum) of the (asyndetic) construction. A completely asyndetic construction can by definition not carry an interpropositional meaning. It is the sense construed in the speech situation which comprises the causal relation.


A word connecting two clauses is traditionally called a conjunction. A conjunction of a subordinate clause is sometimes called ‘subjunction’. Here, such formatives will be called connectives. Two things need to be noted about the term ‘connective’. First and needless to say, it means “overt connective”, or else it could never serve to make the distinction between syndesis and asyndesis. Second, although a connective is sometimes understood to be a word, this implication must be dropped here; cf. below.

Distribution of connectives

In semasiological grammar, the principal criterion for the classification of connectives is their distribution.

Level of combination

The first criterion is the level at which the connectives operate:

.a.Irvin is ugly. Nevertheless, Linda loves him.
b.Irvin is ugly, but Linda loves him.
c.Although Irvin is ugly, Linda loves him.

For clause level connectives, the next distinction is between coordinative and subordinative connectives, by the criterion of whether the clause they combine with is independent (.b) or dependent (.c). A viable operative test to ascertain whether a clause with its connective is dependent is the permutation test. It presupposes that the connective is a constituent of one of the clauses. This tends to be true even for seemingly symmetrical connectives like and, which are demonstrably the introductory element of the following coordinate constituent. Given this, the permutation test involves relocation of the clause which contains the connective. and illustrate its application.

.a.It was raining, but we left anyway.
b.But we left anyway, it was raining.
.a.Although it was raining, we left anyway.
b.We left anyway, although it was raining.

Given the syntactic relation of a connective with one of the connected clauses, then

It must be conceded, though, that certain types of subordinate clause (notably, consective clauses) do not easily permute, or even not at all.

Topology of connectives

Another criterion in the classification of connectives is the way and the sequential position in which they combine with one or both of the connected clauses. The subdivision is as follows:

Examples of (structurally!) symmetric connectives are not easy to come by. The English words and and or are symmetric in propositional calculus, but not in syntax. Examples like clearly prove that they belong to the clause they precede.

.a.Syntax is fun; and it is not difficult, either.
b.Syntax is fun; or do you think otherwise?

As for asymmetric connectives, examples for their different sequential positions with respect to their clause are:

Examples of paired connectives are:

These connectives occur in paratactic constructions. There may also be paired connectives in hypotactic constructions, like English if ... then, because ... therefore, although ... nevertheless. However, these particular English examples should probably not be analyzed as paired connectives, because the second part is optional, there is a (small) paradigm for it, and it is not even admissible if the clauses are inverted.

On the other hand, a paired connective may even be analyzed as a discontinuous morpheme if neither of the two parts can occur independently with the same (component) meaning. This is not the case for any of the paired connectives adduced here.

Correlative pronouns are a special kind of paired connectives. Examples are in the section on relative constructions.

Morphology of connectives

Morphological complexity and bondedness

Syndesis is a gradual phenomenon which ranges from full explicitness of an interpropositional relation by a connective phrase down to its minimal indication by a verbal mood or a change in intonation. Structural means that do not take the form of connective words positioned at clause boundaries are often not subsumed under syndesis; i.e. complex sentences exhibiting only such structural means of clause linkage would be regarded as asyndetic. In any case, syndesis is a gradient ranging from maximally explicit clause linkage down to asyndesis. The following example series illustrates the syndesis gradient:

.O estudante comprou um monte de livros especializados,
PortThe student bought a heap of specialized books
 [a fim de que o professor o tivesse por inteligente].
in order that the professor should consider him intelligent.
.[Nonostante l'opera fosse molto rumorosa],
ItalAlthough the opera was very noisy,
 mi addormentai nel secondo atto.
I fell asleep in the second act.
.[Postquam aurum abstulimus], in navem conscendimus.
LatAfter we had taken the gold away, we boarded a ship.(Pl.Ba.277)
.[Haec cum Crassus dixisset], silentium est consecutum.
LatWhen Crassus had said this, silence followed.(Cic. de or.1,160)
.A verdade é [que todos sairam].
PortThe truth is that they all left.
.Si vis [amari], ama.
LatIf you want to be loved, love.(Sen.ep.9,6)

Source: section 4.2.2 of:

Lehmann, Christian 1988, "Towards a typology of clause linkage." Haiman, John & Thompson, Sandra A. (eds.), Clause combining in grammar and discourse. Amsterdam & Philadelphia: J. Benjamins (Typological Studies in Language, 18); 181-225.

The series illustrates, in turn:

The order corresponds to increasing grammaticalization.

A systematic morphological classification of connectives might run as follows:

  1. phrasal connective
  2. one-word connective
    1. complex connective
      1. compound connective
      2. derived connective
    2. simple connective
      1. connective out of a paradigm
      2. universal subordinator
  3. bound connective
    1. affixal connective
    2. other

This classification combines the two criteria of morphological complexity and bondedness. Although these tend to correlate (inversely), it may be necessary to consider them separately.

Morphosemantics of connectives

In a semasiological perspective, the members of the paradigm(s) of connectives are analyzed morphologically as in the preceding section; and then each morpheme is analyzed as to its distribution and semantic variation. In their morphological makeup, connectives often embody syntactic constructions. Subordinative connectives often consist of a semantically specific relator and a morpheme that just subordinates the dependent clause (see below). By its origin, the semantically specific relator is often a noun or a case relator.

If the semantically specific relator is a noun, then there are two ways in which the subordinate clause can depend on it: it may be a substantive clause or a relative clause.

I'm going to do that when I've gotten there.(Swift 1963:94)

Interpropositional relators and case relators have many general properties in common; and a particular interpropositional relator like the causal one may be semantically similar to a particular case relator like the ablative or instrumental. As a consequence, many relators are polysemous between a use as case relator and a use as connective.

In many languages, the formatives that signal a subordinative interpropositional relation are a subset of, or otherwise formally related to, a paradigm of case relators, typically adpositions or suffixes. Then coordinative connectives may bear a regular morphological relation to semantically corresponding subordinative connectives. The following table lists a few examples from Cabecar:

whenthen/at that moment
because oftherefore
asthat way

The formatives in the left column are postpositions which also subordinate clauses. The comitative and superior morphemes are polysemous in having a simultaneous temporal and a causal sense, resp. D.Med is the demonstrative of medial deixis, which is the unmarked member of the demonstrative paradigm.

Connectives sometimes have a basic meaning (or etymology) that points to interactive discourse organization. For instance:

1 The use of the term ‘asyndesis’ with reference to the text level is current in stylistics and rhetoric.

2 The converse is possible, too: A complex case relator may consist of a connective plus a simple case relator, like English because of.