Coordination is a particular kind of sociation such that the relata are of equal rank in the following sense:
- The semantic relation between the relata is additive, and the structural process expressing it is iterative.
- The construction functions as a composite whole vis-à-vis its context.
- In terms of constituent structure, the relata belong to the same category, and the coordinative construction again belongs to the same category. In other words, each member can represent the construction grammatically.
This, however, requires more precision:
- What is at stake in principle #3 are generic syntactic categories. This is the level where an adjective and a relative clause may belong to the same category (Adjective Phrase), and an NP with nominal head and a nominalized clause may belong to the same category (Nominal Constituent). This generic notion of category is determined by the parallel syntactic function of the coordinate elements.
- Since the relation is additive, the resulting composite whole may count as grammatically plural even if its members are singular. Principle #3 must be restricted correspondingly.
In the appropriate subsection on nexion, we will come across ‘coordination’ as a kind of logical relation between propositions. The term is unfortunately ambiguous between a structural sense as defined here and a semantic sense as defined there. The latter concept will be called ‘logical coordination’. The distinction between a semasiological and an onomasiological perspective makes us see that not every logical coordination takes the form of coordination in structure. In languages such as Japanese and Yucatec Maya, Irvin and Linda is translated as something corresponding to Irvin with Linda (or Linda with Irvin). This construction does not fulfill the above criteria for coordination.
Coordinative constructions are described along the following parameters:
- levels of coordination,
- number of coordinates,
- explicitness of linking.
Levels of coordination
In principle, syntagms at all the syntactic levels can be coordinated. We thus get coordination of
- phrases: diverse categories,
- subphrasal units: words, clitics, stems, affixes.
There are structural differences among coordinative constructions of these levels, which mainly result from the fact that a sentence may be an independent utterance and a clause may be a sentence, while phrases and subphrasal units are by definition constituents of clauses. In principle, coordinative constructions are described independently for each of these levels. The question of whether coordination at some lower level may be analyzed as a reductive variant of coordination of some higher level is an empirical one to be answered for every language. In some languages like English, coordinative constructions at the various syntactic levels are maximally alike, involving always the same connectives, while in other languages such as Japanese constructions and connectives are specific to each syntactic level.
The question just mentioned arises on the basis of examples like :
|.||a.||[ Linda is intelligent] and [she works hard].|
|Eng||b.||Linda [[is intelligent] and [works hard]].|
In a semasiological grammar of English, an analysis of E1 recognizes that E1.a involves coordination of two clauses, while E1.b involves coordination of two predicates, as indicated by the bracketing. So far, the two sentences, although being in a paradigmatic relationship, have a different syntactic structure. One may wish to analyze E1.b by some kind of zero anaphora, i.e. as a variant of E1.a in which the pronoun she has been omitted, but otherwise with the same syntactic structure as E1.a. This is taken up in the section of conjunction reduction and zero anaphora.
Number of coordinates
Since coordination is by definition iterative, the potential number of coordinates is infinite. Many languages have different rules for just two coordinates vs. more than two coordinates; cf. e.g. the section on topology of connectives. This differential treatment results from the fact that a binary coordinative construction is symmetric, while an n-aryn>2 coordinative construction is not.
Explicitness of linking
The following is an application of the parameter ‘explicitness of linking’ to coordinative constructions.
In asyndetic coordination, coordinates are serialized, and the construction is interpreted as coordination.
|.||veni vidi vici|
|Lat||I came, I saw, I conquered|
In (the most celebrated example of an asyndetic construction), the interpropositional relation is one of immediate temporal sequence (“no sooner did I arrive than I saw ...”). The sequential order of the clauses iconically reflects the order of the events they designate. This is not so in a sentence like I could not come earlier, the train was late (cf. the section on asyndetic subordination). Nevertheless, the temporal order of events is part of the linguistic meaning of . It can be inferred on the following basis:
- the clauses have the same subject,
- all verbs are in perfective aspect,
- there is no explicit indication of a different temporal order.
Under such circumstances, the inference that the events occurred in the order of the clauses is cogent (it would also be the default interpretation if the subjects differed). The conciseness effect of asyndesis is partly based on principles of iconicity.
Classes of connectives are introduced in the section on syndesis. The present section describes the constructions in which they are used. In constructions with more than two coordinates, syndesis and asyndesis are often combined by a principle according to which the first to the one-but-last coordinate units are juxtaposed asyndetically and only the last one is linked by syndesis, as in .
|.||a.||Linda woke up, brushed her teeth, took a shower and prepared her breakfast.|
|b.||Do you want to play football, go for a walk or stay at home?|
Reduction phenomena in coordination are dealt with in the section on fusion of clauses.