This map shows the relationship between the order of object and verb and the order of adjective and noun; these two features are shown on Maps 83A and 87A. This map is the third map showing the relationship between two typological features. Unlike the two previous maps, however, there is no correlation between the features or implicational relationship relating them: all four types defined on the intersection of these two features, the first four types shown in the value table, are common.
| Go to map | ||
|---|---|---|
| Value | Representation | |
 |
Object-verb and adjective-noun (OV&AdjN) | 216 |
 |
Object-verb and noun-adjective (OV&NAdj) | 332 |
 |
Verb-object and adjective-noun (VO&AdjN) | 114 |
 |
Verb-object and noun-adjective (VO&NAdj) | 456 |
 |
Languages not falling into one of the preceding four types | 198 |
| Total: | 1316 | |
The first type, OV and AdjN, languages in which the object precedes the verb and the adjective precedes the noun, is illustrated in (1) from Udihe (Tungusic; Siberia), (1a) illustrating the OV word order, (1b) the AdjN order.
(1) Udihe (Nikolaeva and Tolskaya 2001: 840, 487)
a.
mamasa
ule:-we
olokto-ini
old.woman
meat-acc
cook-3sg
O
V
‘The old woman is cooking meat.’
b.
uligdig’a-ŋku
moxo-ziga
beautiful-pl
cup-pl
Adj
N
‘beautiful cups’
The examples in (2) illustrate a language which is OV and NAdj, Kunuz Nubian (Nilo-Saharan; southern Egypt).
(2) Kunuz Nubian (Abdel-Hafiz 1988: 219, 283)
a.
id
ti:-g
kanni:r-ken
gos-s-u
man
cow-acc
knife-instr
slay-pst-3sg
‘The man slaughtered the cow with the knife.’
b.
e:n
duru-we:r
woman
old-indef
‘an old woman’
The third type is languages which are VO and AdjN. This type is illustrated by English, as well as by Coast Tsimshian (Tsimshianic; British Columbia); the VO order is illustrated in (3a), the AdjN order in (3b).
(3) Coast Tsimshian (Mulder 1994: 32, 162)
a.
yagwa-t
huum=[da
duus]=[a
hoon]
pres-3.erg
smell=[erg
cat]=[abs
fish]
‘The cat is sniffing the fish.’
b.
ʼwii
samn
great
spruce.tree
‘a big spruce tree’
(Note that the case markers in Coast Tsimshian are prenominal but encliticize onto the preceding word; for example the absolutive case marker =a, which marks the object noun phrase hoon ‘fish’ as absolutive, attaches phonologically to duus ‘cat’, but grammatically it goes with hoon ‘fish’; the square brackets indicate the syntactic constituents.)
The fourth type is languages which are VO and NAdj. This type is represented by Olo (Torricelli; Papua New Guinea); both VO and NAdj orders are illustrated in (4).
(4) Olo (Staley 1995: 31)
ki
k-alei
tifa
funi
1sg
1sg.subj-eat.3sg.m.obj
banana
ripe
‘I eat a ripe banana.’
The fifth type shown on the map involves languages not falling into one of the first four types. It includes languages lacking a dominant order of object and verb or a dominant order of adjective and noun, as well as languages lacking a construction in which adjectives modify nouns because the closest equivalent involves internally headed relative clauses (see Chapter 87).
Languages which are OV and AdjN are especially common in much of Asia, except in the Middle East and in an area stretching from northeastern India through Southeast Asia. Elsewhere in the world, this type is found scattered and in various small areas, though in many of these it is interspersed with OV and NAdj languages. These areas include (i) Ethiopia and Eritrea; (ii) two areas in Papua New Guinea, one in the eastern highlands, the other in the lower Sepik valley; (iii) an area in the southwestern United States and northwestern Mexico; and (iv) an area in Ecuador and adjacent parts of Brazil to the east.
Languages which are OV and NAdj are the dominant type in Australia and in New Guinea. They are also the dominant type in North America outside two well-defined areas in the Pacific Northwest and Mesoamerica; of the four main types, they are the sole type represented in eastern Canada and the United States as well as in the north, including Alaska, northern Canada, and Greenland. They are also the most frequent type in South America, though other types are common and the four types do not show clear areal patterning within South America. They are also found in (i) Burma and adjacent areas in India and China (all the languages in question being Tibeto-Burman); and (ii) various parts of Africa where OV languages are spoken.
Languages which are VO and AdjN are dominant in (i) an area in northern and eastern Europe; and (ii) the Pacific Northwest. There are also smaller pockets in (i) an area in central Africa centered in the Central African Republic; (ii) Taiwan, the Philippines; (iii) the extreme northern part of Northern Territory in Australia; (iv) an area in southwestern Australia; (v) two pockets in Mesoamerica, one in the northern part of this region, the other in the southeast. This type is also represented by the different varieties of Chinese, though these do not show up on the map as a well-defined area.
Languages which are VO and NAdj are the dominant type in (i) western Europe and around the Mediterranean, including the Arabian Peninsula; (ii) sub-Saharan Africa; (iii) a large area stretching from Southeast Asia eastward through the Pacific, except in the Philippines and New Guinea. They are also common in Mesoamerica, though there are also many VO and AdjN languages in this area, especially to the north and southeast.
The order of adjective and noun is often claimed to correlate with the order of object and verb. Dryer (1988a, 1992) argues against this, however, claiming that AdjN order occurs as often in VO languages as it does in OV languages. It has been known since Greenberg (1963) that many OV languages are NAdj. What Dryer (1988a, 1992) argues, however, is (i) that NAdj order is even more common among OV languages than had been realized, AdjN order being more common only in Eurasia; and (ii) that AdjN order is more common in VO languages than previously thought.
The first of these, that AdjN is more common among OV languages only in Eurasia, is reflected by the map. Nevertheless, the map somewhat exaggerates the frequency of OV and AdjN languages, for two reasons. First, the projection used shows areas further from the equator as somewhat larger than they really are. As a result, the area across northern Asia appears larger relative to equatorial regions than it actually is. Second, there are more languages and greater genealogical diversity in regions closer to the equator, so that there are a greater number of languages and genealogical groups (families and genera) with OV and NAdj languages than is reflected on the map. There are more languages in New Guinea, for example, than there are on the mainland of Europe and Asia combined. Similarly, there are more than twice as many languages in Africa as there are on the mainland of Europe and Asia combined.
In terms of raw numbers of languages, AdjN order is less common among VO languages than it is among OV languages. Similarly, in terms of raw numbers of languages, languages which are VO and NAdj are considerably more common on the map than languages which are VO and AdjN. But this is misleading. It reflects the fact that there are two very large language families, Niger-Congo and Austronesian, each containing about twenty percent of the languages of the world, in which the majority of languages are VO and NAdj. Almost half of the languages shown on the map which are VO and NAdj (151 out of 307) are in one of these two families: the vast majority of the VO and NAdj languages in sub-Saharan Africa are Niger-Congo, and most of the languages of this type in the region stretching from Indonesia eastward into the Pacific are Austronesian. If one counts, not number of languages but number of genera – genealogical groups comparable to the subfamilies of Indo-European (see the Introduction to the Genealogical Language List) – containing languages of the various types, then the number of genera containing languages which are VO and NAdj is indeed greater than the number that are VO and AdjN. However, the ratio of genera containing NAdj languages to those containing AdjN languages is only slightly greater for VO languages than it is for OV languages. The following table gives the number of genera containing languages of the type given:
(5)
OV&AdjN
61
OV&NAdj
112
VO&AdjN
40
VO&NAdj
78
The ratio of VO&NAdj to VO&AdjN is 1.95, only slightly greater than the ratio of OV&NAdj to OV&AdjN, which is 1.84. Furthermore, if one counts the number of language families containing languages of each of the four types, assuming the highest level of classification in the Genealogical Language List as a criterion for what constitutes a family, we get the numbers in the following table:
(6)
OV&AdjN
35
OV&NAdj
37
VO&AdjN
22
VO&NAdj
22
Here we find that the differences in terms of numbers of languages or in terms of numbers of genera essentially disappear: for both OV and VO, the number of families containing languages of the given type are about the same, and in particular there are as many families containing languages which are VO and AdjN as there are languages which are VO and NAdj.
The basis for the mistaken impression that many linguists have had that there is a correlation between the order of object and verb and the order of adjective and noun apparently derives from the fact that within Europe and Asia, there is the appearance of such a pattern. On the one hand, there is an overwhelming preference for AdjN order among the OV languages of Asia. Conversely, although there is a large area of VO languages in northern Europe with AdjN order, the VO languages of southwestern Europe are NAdj, as are a clear majority of the VO languages of Southeast Asia. In addition, the fact that the majority of VO languages within two very large families, Niger-Congo and Austronesian, are NAdj adds to the impression.
In light of the above discussion, the data provide no basis for thinking that OV languages have any stronger preference for AdjN order than VO languages do. However, some have suggested (e.g. Greenberg 1963) that AdjN order is less frequent in verb-initial languages (as opposed to SVO languages, which like verb-initial languages are VO). But although this is not shown on this map, the languages used for the map do not support this hypothesis either. The numbers in (7) give the number of genera and number of families for the two orders of adjective and noun in verb-initial languages, analogous to the numbers given above in (5) and (6):
(7)
No. of genera
No. of families
Verb-initial&AdjN
23
15
Verb-initial&NAdj
28
14
The data in (7) provides no reason to think that AdjN order is any less frequent in verb-initial languages. If anything, AdjN order is more frequent in verb-initial languages, since the ratio for numbers of genera of AdjN to NAdj in (7), 23:28, is higher than that for VO languages in general (40:78), or for OV languages (61:112), given in (5) above. Examples of verb-initial languages which are AdjN include Coast Tsimshian, illustrated above in (3), Batad Ifugao (Austronesian; Philippines), illustrated in (8), and Gude (Chadic; Nigeria), illustrated in (9).
(8) Batad Ifugao (Newell 1993: 15, 23)
a.
in-dat
Aligūyun
nan
dotag
ay
agī-na
theme.foc:pst-give
Aligūyun
def
meat
loc
brother-3sg
‘Aligūyun gave the meat to his brother.’
b.
nan
nappuhi-n
tibung
def
bad-link
wine.jar
‘the bad wine jar’
(9) Gude (Hoskison 1983: 90, 53)
a.
kə
kii
Musa
faara
comp
throw
Musa
stone
‘Musa threw a stone.’
b.
ma-gusə
minə
attr-short
woman
‘a short woman’
As discussed in Chapter 87, the term adjective is used here in a semantic sense, for words denoting properties, regardless of their word class in different languages. In many languages, adjectives are formally verbs, and hence as modifiers of nouns, they constitute a type of relative clause. This raises the question of whether distinguishing different languages on the basis of the word class of such adjectival words within the language might lead to new generalizations. Preliminary results suggest that this is not the case, but further investigation is necessary.