1. Defining the values

This map shows the order of relative clause and noun. A construction is considered a relative clause for the purposes of this map if it is a clause which, either alone or in combination with a noun, denotes something and if the thing denoted has a semantic role within the relative clause. If there is a noun inside or outside the relative clause that denotes the thing also denoted by the clause, that noun will be referred to as the head of the relative clause. Headless relative clauses (like English what I bought at the store ) are not relevant to this map.

The two basic types shown on the map are languages in which the relative clause follows the noun , and languages in which the relative clause precedes the noun . The examples in (1) from English and from Maybrat (West Papuan) illustrate relative clauses following the noun.

(1)  

 

The example in (2) from Alamblak (Sepik; Papua New Guinea) illustrates a relative clause preceding the noun. 

(2) Alamblak (Bruce 1984: 109) 

 

The relative clauses illustrated in (1) and (2) occur with heads outside the relative clause; these can be referred to as externally-headed relative clauses . In some languages, the head is inside the relative clause; these can be called internally-headed relative clauses . These are illustrated by the examples in (3) from Mesa Grande Diegueño (Yuman; southern California and northwest Mexico); the fact that the head is inside the relative clause is clearest in (3a), in which the head ( gaat  ‘cat’) occurs between the subject and verb of the relative clause.

(3) Mesa Grande Diegueño (Couro and Langdon 1975: 187, 186) 

 

What determines whether 'ehatt ‘dog’ or gaat ‘cat’ is interpreted as the head in these examples is the presence versus absence of the subject relative prefix on the verb: its presence in (3b) signals that the head is the subject of the relative clause, namely 'ehatt ‘dog’, while its absence in (3a) signals that the head is something other than the subject, in this case gaat ‘cat’. Languages with internally-headed relative clauses are probably more common than the map suggests because, until recently, grammarians often failed to recognize them as such.

The fourth type shown on the map is correlative relative clauses , as in (4) from Bambara (Mande, Niger-Congo; Mali).

(4) Bambara (Bird and Kante 1976: 9) 

 

Correlative clauses are strictly speaking a subtype of internally-headed relative clauses in that the head noun occurs inside the clause, but they differ from those coded here as internally-headed in that the relative clause is outside the main clause and is connected anaphorically to a noun phrase in the main clause that corresponds to the head noun in the English translations. 

The fifth type shown on the map consists of languages with adjoined relative clauses . As with the preceding type, adjoined relative clauses are outside the main clause; they do not form a constituent with the head noun, which is in the main clause, and they may be separated from it. However, unlike correlative clauses, the head occurs in the main clause rather than in the relative clause. An example of a language in which relative clauses are of this sort is Diyari (Pama-Nyungan; South Australia), as illustrated in (5) (Austin 1981: 188).

(5) Diyari (Austin 1981: 210) 

 

The sixth type shown on the map is represented by a single language, Kombai (Trans-New Guinea; Papua, Indonesia), and is referred to here as a double-headed relative clause . As illustrated in (6), relative clauses in Kombai combine the features of externally-headed and internally-headed relative clauses in a single structure: they have both an external head noun and a noun corresponding to the head noun inside the relative clause. While the two nouns are sometimes the same, as in (6a), the external noun is usually more general than the one inside the relative clause, as in (6b), where the external noun is simply ro ‘thing’.

(6) Kombai (de Vries 1993: 78, 77)

 

The final type shown on the map involves languages which employ two or more of the preceding constructions, without one being dominant. For example, Gimira (Omotic; Ethiopia) allows the relative clause to either precede or follow the head noun, and there is no evidence for one order being dominant (Breeze 1990: 39). In most languages that allow both orders of relative clause and noun, there appear to be reasons for treating one as dominant (see “Determining Dominant Word Order” in Chapter 91 ). In Kapampangan (Austronesian; Philippine), for example, relative clauses can precede the noun only if they consist of a single word, whereas relative clauses of any length can follow the noun (Mirikitani 1972: 189). Kapampangan is thus shown on the map as having postnominal relative clauses. Some languages have both externally-headed and internally-headed relative clauses or correlative clauses, without one type being dominant. For example, Murrinh-Patha (isolate; Northern Territory, Australia) has both postnominal relative clauses, as in (7a), and internally-headed relative clauses, as in (7b).

(7) Murrinh-Patha (Walsh 1976: 289, 287) 

 

That the head kumukur ‘club’ in (7a) is external is clear from the fact that it bears an instrumental suffix indicating its role in the main clause, not the relative clause (where it would be in absolutive case). That the head mutyiŋga ‘woman’ in (7b) is internal to the relative clause is clear from the fact that it bears ergative case, reflecting its role in the relative clause, not its role in the main clause. Other examples of languages having more than one relative construction with none dominant include Panjabi, which has both prenominal relative clauses and a correlative construction (Bhatia 1993: 50, 53, 56), Ngiyambaa (Pama-Nyungan; New South Wales, Australia), which has both a correlative construction and an adjoined relative construction (Donaldson 1980: 297-299), and Kobon (Trans-New Guinea; Papua New Guinea), which has both prenominal and double-headed relative clauses (Davies 1981b: 29).

Note that languages vary as to whether relative clauses involve forms of verbs that also occur in main clauses. In English relative clauses like that in (1a), for example, the forms of the verbs are the same, as illustrated by am reading in (1a). In Kolyma Yukaghir (isolate; northeast Siberia), in contrast, the verb in relative clauses lacks the inflections associated with finite verbs and occurs with a general attributive suffix, as in (8).

(8) Kolyma Yukaghir (Maslova 2003a: 418) 

 

English also has nonfinite participial relative clauses, as in the man reading the book. Nonfinite relative clauses are sometimes not considered as relative clauses; however, since there are many languages where relative clauses are all nonfinite and since these constructions mean the same thing as finite relative clauses in English, such participial constructions are considered as relative clauses here.

In some languages, headless relative clauses are arguably the basic form of relative clauses, and relative clauses with a head can be analysed as nominal expressions in apposition to the head. Such an analysis is argued for by Curnow (1997) for Awa Pit (Barbacoan; Colombia and Ecuador); (9a) illustrates a headless relative clause, while the example in (9b) illustrates a relative clause followed by a head noun. 

(9) Awa Pit (Curnow 1997: 286) 

 

The fact that the headless relative clause in (9a) is itself a nominal expression is further brought out by the fact that it occurs with the postpositional accusative clitic =ta. Relative clauses like these are often called nominalizations, but are still considered relative clauses for the purposes of this map.

2. Geographical distribution

The geographical distribution of the two major types, i.e. externally-headed relative clauses with relative clauses preceding or following the noun, is quite clear. The overwhelmingly dominant type in much of the world is for the relative clause to follow the noun.  The exception to this is much of Asia, where except in the Middle East and Southeast Asia, the dominant type has the relative clause preceding the noun. Outside Asia, except for a scattering of geographically isolated instances, this prenominal type is found in only three relatively small areas: (i) New Guinea; (ii) Ethiopia and Eritrea; and (iii) southern Colombia and the adjacent area of Brazil.

Internally-headed relative clauses are scattered throughout the world, though two areas where they are more common are worth mentioning. One is North America: though they constitute only a minority here, they are more common than externally-headed relative clauses preceding the noun.  The other is a small area in West Africa where both internally-headed and correlative clauses are common. A second area where correlative clauses are found is South Asia. It must be stressed that the map only shows languages with a specific type if that type is dominant in the language.  There are more languages with internally-headed or correlative clauses than the map might suggest, because these often co-exist with some other type and are not dominant in the language.

There are only five languages shown on the map as employing adjoined relative clauses as the dominant type, four of them in Australia. Only one language is shown with double-headed relative clauses as the dominant type, in Papua (Indonesia). 

3. Theoretical issues

The concentration of languages in which the relative clause precedes the noun in much of Asia, but with low frequency elsewhere in the world, illustrates the extent to which a region as large as Asia can be a linguistic area. In fact, this area is more generally associated with a type in which all modifiers precede the noun, a type that is relatively uncommon outside of Asia. While most of the languages with this property are verb-final languages, it is clear that this type is found only among a minority of verb-final languages elsewhere in the world. The relationship between the order of relative clause and noun and the order of object and verb is discussed further in Chapter 96. Maps 122 and 123 show other features involving relative clauses.