An indefinite article is a morpheme that frequently occurs in noun phrases and signals that the referent is not uniquely identifiable by the hearer, as in We have a dog. The great majority of APiCS languages have not only a definite, but also an indefinite article. This chapter is modeled on the WALS chapter by Dryer (2005c).
Indefinite articles typically originate in the numeral ‘one’, and in the APiCS languages, this is almost universally the case: In the Romance-based languages, the indefinite article is derived from the Romance indefinite article un/um (ultimately from Latin unum ‘one’), and in the English-based languages, the indefinite article is derived from English a or one (both ultimately from Old English ān ‘one’). As a result of their origin in a numeral, indefinite articles are most often restricted to count nouns (cf. *We have a money) and to singular nouns (cf. *We have a dogs).
Two examples of indefinite articles are given in (1)-(2).
Just as definite articles often originate in demonstratives and are therefore often still synchronically identical to them, indefinite articles are often still synchronically identical to the numeral ‘one’. This is the case in Romance languages, e.g. in French, where un ami can mean ‘a friend’ or ‘one friend’. In Dryer’s (2005c) world-wide study, 204 out of 473 languages have an indefinite article, and of these, 91 have an indefinite article word that is distinct from the numeral ‘one’ (like English a), while 90 languages have an indefinite article word that is identical to the numeral ‘one’ (like French un).
Distinguishing article use from numeral use in semantic-pragmatic terms is not easy. In the three examples above, a translation with ‘one’ would yield a slightly different meaning, but not unacceptable sentences (‘One friend of mine came here.’, ‘He saw one fig tree.’, ‘We made one large garden.’). The basic difference is that the numeral underlines the cardinality, in implicit contrast with other cardinalities (‘one fig tree’, i.e. not two or three fig trees), while the indefinite article backgrounds the cardinality (even though it entails it as well). Thus, when a word that is also used as the numeral ‘one’ is used in a context where emphasizing the cardinality would be pointless, as in (4)-(5), we can be certain that we are dealing with an indefinite article.
This criterion is fairly vague, so in practice, the criterion that was normally employed by the contributors was whether the word is often used accompanying an indefinite noun where English would use its indefinite article, not its numeral one. Note that in many APiCS languages, the indefinite article is not obligatory, and the precise conditions under which it is used or omitted are usually unknown.
Four different values are distinguished:
|Indefinite article distinct from numeral ‘one’||20|
|Indefinite article identical to numeral ‘one’||46|
|No indefinite article, but definite article||1|
|Neither indefinite nor definite article||9|
Many pidgins and creoles are like English in that they have a special indefinite article that is distinct from the numeral ‘one’ (value 1). This is the case in quite a few English-based and Dutch-based languages where the English (and similarly Dutch) distinction between a and one has survived. Examples are given in (6).
But the indefinite article a has disappeared from many English-based languages, and has been replaced by the numeral one (mostly written wan in the pidgins and creoles), leading to identity (value 2, see below). In Jamaican and in Creolese, the indefinite-article use and the numeral use are said to be pronounced differently.
In some of the French-based languages, the indefinite article is occasionally differentiated from the numeral ‘one’ (un in French).
Another example of this type is Sango (see ex. 3 above), where the indefinite article does not derive from the numeral ‘one’. But note that a mere stress difference is not sufficient: In many languages, the numeral ‘one’ can be distinguished from the indefinite article by stress (e.g. Mauritian Creole en sát ‘a cat’ vs. én sát ‘one cat’). These cases are subsumed under the next value.
In the majority of our languages the indefinite article is identical to the numeral ‘one’ (value 2). This is the case in many Romance-based languages, where the nondistinctness of indefinite article and numeral was inherited from French/Spanish un and Portuguese um.
Nondistinctness is also found in many English-based languages, where wan (from one) is used in both ways (wanpela in Tok Pisin comes from one fellow):
When a language has two different indefinite articles, one of which is identical to the numeral and one of which is distinct, it is classified as value 1 (e.g. Nicaraguan Creole English, which has both a and wan).
The languages of the remaining two types have no indefinite articles. Value 3 comprises the one language that at least has a definite article (see Chapter 28), Yimas-Arafundi Pidgin. Finally, the languages of the last type (value 4) do not have any article; these are the same languages that have value 4 in Chapter 28. In (13), we see an indefinite and a definite noun phrase, and neither has an article.