Chapter 3: Order of adjective and noun

Feature information for this chapter can be found in feature 3.

1. Feature description

Chapter 3 is about the position of the adjective relative to the noun it modifies. Here and in other APiCS features, the term adjective is defined purely in a semantic sense, as a word with a lexical meaning such as ‘hot’, ‘old’, or ‘blue’. For this reason, we disregard demonstratives, numerals, or words meaning ‘other’. In languages like English or French, adjectives belong to a distinct class of words but in other languages they are nouns or verbs. However, for the purposes of the present feature, a word is treated as an adjective as long as it denotes a property or quality, irrespective of its word class, thus including cases like

tall/strong person
See example 58-5

The conception of the present feature follows WALS Feature 87 (Dryer 2011a) and thus only considers adjectives modifying a noun (i.e. attributive adjectives), as in English the wicked man. Chapter 3 is not concerned with predicative adjectives in clauses where the noun is the subject and the adjective is the predicate, as in English The man is wicked.

2. The values

There are two possible orders of the modifying adjective relative to the noun: either the adjective precedes the noun (value 1) or it follows the noun (value 2):

Modifying adjective precedes noun343569
Modifying adjective follows noun73542

Value 1. The adjective-noun order (modifying adjective precedes noun) is found in e.g. Chinese Pidgin Russian:

small jacket (Perekhvalskaya 2013)

Value 2. The noun-adjective pattern (modifying adjective follows noun) is illustrated by an example from Ambon Malay:

small children (Paauw 2013)

This feature is a multiple choice feature in that a language can have both preceding and following adjectives. Thus, in Cape Verdean Creole of Santiago adjectives can either precede or follow nouns:

a poor old man (Lang 2013)
a poor old man (Lang 2013)

There is no perfect match with WALS Feature 87 (Dryer 2011a), which, in addition to the two APiCS values, has a third value “Both orders of noun and modifying adjective occur, with neither dominant” and a fourth value “Adjectives do not modify nouns, occurring as predicates in internally headed relative clauses”. WALS value 3 represents only a subset of the cases where APiCS records both preceding and following adjectives: WALS’s “no dominant order” means that both alternatives are about equally common (Romance languages therefore appearing as noun-adjective), while APiCS records all occurring orders and in addition gives information on relative importance. APiCS does not have a value corresponding to WALS’s fourth value “Adjectives do not modify nouns, occurring as predicates in internally headed relative clauses”, but two contributors note that in addition to attributive adjectives modification through relative clauses is regularly attested in their variety. This is found in Tayo (Ehrhart & Revis 2013) and in the mixed language Michif:

He was given a piece of meat that was big. or He was given a big piece of meat. (Bakker 2013)

3. Distribution

Numerical. Although noun-adjective is the prevalent order among the world’s languages (768 of 1005 languages in Dryer 2011a), a preceding adjective is clearly the dominant pattern among the (admittedly skewed sample of) languages represented in APiCS (69/76), with 34 varieties using this pattern exclusively and another 35 having preceding adjectives along with following adjectives. The noun-adjective pattern is found in 42 APiCS languages, of which only 7 are exclusively noun-adjective. These are Ambon Malay, Juba Arabic, Kikongo-Kituba, Kinubi, Lingala, Mixed Ma’a/Mbugu and Papiá Kristang.

Geographic/areal. While Dryer’s WALS map shows rather well-defined areas of noun-adjective and adjective-noun languages, the two patterns appear to be more or less evenly spread around the world as far as APiCS languages are concerned. This may well be due to the fact that the contact languages documented in APiCS often involved large-scale migrations and that the concept of areality in the traditional sense is therefore problematic in this case.

By lexifier/substrate. For most APiCS languages, Feature 3 appears to be a rather clear case of lexifier influence: with only two exceptions (Media Lengua, Papiá Kristang) the 27 Romance language-lexified contact languages allow both the adjective-noun and the noun-adjective pattern, just as in French, Portuguese or Spanish. A number of APiCS contributors note that similarities with the lexifier even extend to which adjectives may precede or follow nouns or to the meaning difference of the same adjective in pre- or postnominal position. Of the two exceptions mentioned above, the mixed language Media Lengua is exclusively adjective-noun, closely following the Quechua model, while Papiá Kristang’s exclusive noun-adjective order seems to have been the result of a strong sub- or adstratal influence of Malay. As to the Germanic lexifier languages, 25 out of 28 varieties in our sample have only preceding adjectives, just as in English and Dutch. Among the non-European lexified contact languages, we find the same trend of copying the lexifier order. For example, Kikongo-Kituba, Lingala and Mixed Ma’a/Mbugu (Bantu), Juba Arabic and Kinubi (Arabic) and Ambon Malay (Malay) copy the noun-adjective order of their lexifiers.

In some cases the lexifier and substrates have the same order of the adjective relative to the noun – as in Juba Arabic and Kinubi, whose Sudanese substrates are also noun-adjective – but the Central African Sango copies the adjective-noun model of its lexifier Ngbandi in spite of the fact that the pidginizers spoke either languages that were predominantly noun-adjective (speakers from East and West Africa and from the equatorial regions of the Congo River) or had both patterns (French-speaking Europeans; see Samarin 2013a). Fanakalo’s adjective-noun is a surprising exception, its main lexifiers Zulu and Xhosa as well as the Bantu substrates being noun-adjective languages. Here, the adjective-noun sequence is possibly modelled on the structure of the socially dominant English and Afrikaans-speaking pidginizers of Zulu and Xhosa.

The fact that Bislama, Kriol, and Tok Pisin (all spoken in the Australia-Pacific region) have both orders may also be due to substratal influence. Kriol in northern Australia is spoken in a region with a mix of languages allowing pre-, post and both pre- and postmodifying adjectives. Tok Pisin and Bislama are spoken in areas where the overwhelming majority of languages are premodifying, but there is historical evidence that both languages received a strong input from a pidgin spoken on plantations in Queensland, (eastern Australia), an area where both orders are found. In addition, Tok Pisin was influenced by Tolai, one of the few languages in Papua New Guinea that has no dominant order according to WALS.

The rule for Bislama and Tok Pisin is that adjectives modified by -fala/-pela precede the noun while the other adjectives follow it:

long beak
bad man

The case of the mixed language Michif (French nouns, Cree verbs) is interesting: Michif adjectives are of French origin and as in French, they either precede or follow the noun (Cree does not have adjectives but expresses noun modification through relative clauses or by noun-prefixation (Bakker 1997: 106). Compare example (5) above).

The three Malay-lexified varieties in APiCS show a wide variation with only Ambon Malay following the Malay lexifier noun-adjective order. Ambon Malay is spoken in an area of noun-adjective languages, just like Singapore Bazaar Malay, which however shows both orders. This is probably due to the fact that Indian and Chinese languages (predominantly adjective-noun) were involved in the genesis of Bazaar Malay (cf. Khin Khin Aye 2013a). Sri Lanka Malay seems to have adopted Tamil’s adjective-noun order (for the close contact between Malays and Sri Lanka Muslim Tamils (see Slomanson 2013a).