Standard concept

Two linguistic units are in complementary distribution if they share no context, so that their distributions do not overlap, but complement each other. Here are two examples:

1. Two fricatives are in complementary distribution depending on the preceding vowel:
   fricative₁ / __ V₁ ~ fricative₂ / __ V₂
2. Two sets of declension allomorphs are in complementary distribution depending on the declension class of the noun stem:
   paradigm₁ / noun_stem_decl.class.1 __ ~ paradigm₂ / noun_stem_decl.class.2 __

For the two sets of contexts to complement each other means that they form a homogeneous set of contexts which has a natural common denominator. Thus, the two contexts defined by the two classes of vowels in ex. 1 form a unitary context ‘after a vowel’. Likewise, the two sets of nouns of ex. 2 form a whole, which could be (if there are only two declension classes) the set of all noun stems.

Given this concept of complementary distribution, then the observation that two linguistic units U₁ and U₂ are in complementary distribution indicates that

• the context is the conditioning factor for the occurrence of one rather than the other unit, in other words, that U₁ and U₂ are combinatory variants
• U₁ and U₂ have the same meaning or function.
• at the next more abstract level, U₁ and U₂ are just (variants of) one linguistic unit U₀.

The standard concept of complementary distribution as defined here presupposes a narrow concept of context. This is defined by (classes of) linguistic units which are immediately adjacent to the unit whose distribution is analyzed, as shown in the above two examples. With these narrow conditions, the above-mentioned inferences are safe.

Exclusive distribution

Two units that are in complementary distribution exclude each other in a given position. Often, two units that occupy different syntagmatic positions of a given construction exclude each other in it. Two linguistic units are in exclusive distribution (or in mutual exclusion) if they do not cooccur in the same construction. A recurrent example of this situation is the distribution of noun phrases (including tonic pronouns) and clitic pronouns as verb dependents. Standard French is one of the languages showing this situation. The direct object of the finite verb may be represented by a lexical or pronominal NP or by a clitic pronoun. The former is in postverbal, the latter in preverbal position, as shown in .

In a given clause, the direct object is either an NP or a clitic. The two cannot cooccur in one clause (as they can in colloquial French and in other languages).

Such a distribution resembles complementary distribution in that the two subclasses of (pro-)nominal expressions exclude each other in a given construction. And similarly as in complementary distribution, they are, in fact, subclasses of a more abstract class. As a consequence, this situation has occasionally been subsumed under complementary distribution in the literature (e.g., in Kibrik 2011:187).

Exclusive distribution differs from complementary distribution in several respects:

1. The units in question do not occupy the same syntagmatic position. Consequently, they neither form a paradigm nor contract any paradigmatic relation.
2. Choice among the units in question is not conditioned by the context, so they are not combinatory variants of each other. Instead, there is free choice among them just as if they were in opposition.
3. Consequently, no inference towards functional equivalence is possible.

References

Kibrik, Andrej A. 2011, Reference in discourse. Oxford etc.: Oxford University Press (Oxford Studies in Typology and Linguistic Theory).