Chapter Order of Subject and Verb

by Matthew S. Dryer

1. Defining the values

This map shows the dominant order of lexical (or nonpronominal) subject and verb. The primary types are languages which are SV (in which the subject precedes the verb), a type represented by English and by Turkish, as illustrated in (1a), and languages which are VS (in which the subject follows the verb), exemplified by Welsh, as illustrated in (1b).



Turkish (Kornfilt 1997: 90) 









‘The water boiled.’ 


Welsh (Williams 1980b: 165) 










‘The man came.’ 

Values of Map 82A. Order of Subject and Verb
Subject precedes verb (SV) 1193
Subject follows verb (VS) 194
Both orders with neither order dominant 110

As with the other maps showing word orders, there are two ways in which a language can count as SV.  In some languages, SV is the only order permitted by the grammar. In other languages, both orders are grammatical, but SV order is dominant (see “Determining Dominant Word Order”). Languages in which both orders occur and in which neither order is dominant are classified as the third type on the map.

In languages in which the position of subjects in intransitive clauses differs from that in transitive clauses, the map shows the order in intransitive clauses. The question of what should be considered a subject in different languages and the question of whether all languages have subjects have generated considerable discussion in the literature over the years. However, these theoretical questions are largely irrelevant to this map: a language is considered SV if the single lexical argument in an intransitive verbal clause more commonly precedes the verb, but VS if such an argument more commonly follows the verb. Even languages for which people question the relevance of a notion of subject can be classified according to this criterion.

To a large extent, the distribution of SV on this map corresponds to three types on Map 81A, namely SOV, SVO, and OSV.  Conversely, VS on this map largely corresponds to the three types VSO, VOS, and OVS. There are a variety of ways in which these correspondences can fail to obtain. One is that there are some languages which are shown as lacking a dominant order on Map 81A, because both orders of object and verb are common, but which are shown as SV on this map.  German and Hungarian are languages of this sort.  Another way in which these correspondences can fail to obtain is that there are languages in which both VSO and VOS order are common, and which are thus shown as lacking a dominant order on Map 81A, but in which VS is dominant. Boumaa Fijian (Austronesian; Dixon 1988: 243) is an instance of such a language (see example in Chapter 83).

Another major source of differences is the fact that where subjects of transitive clauses and subjects of intransitive clauses differ in their position, this map shows the word order in intransitive clauses, while Map 81A shows word order in transitive clauses. For the vast majority of languages, the position of subjects is the same in intransitive clauses as in transitive clauses. However, there are a number of types of languages in which it is not.

The first type are languages in which word order follows an ergative pattern (see Chapters 98, 99, and 100), and here there are two subtypes. First, there are languages which can be described as Ergative-Verb-Absolutive. An example of such a language is Muna (Austronesian; eastern Indonesia), illustrated in (2).

(2) Muna (van den Berg 1989b: 150, 163) 












‘The dog ran again.’ 















‘The crow stole dried fish.’ 

Languages like Muna are shown as VS on this map, but as SVO on Map 81A. Languages of this type are common in Mesoamerica; examples include Huave (isolate; Mexico; Stairs and Hollenbach 1981: 335), Huastec (Mayan; Mexico; Edmonson 1988: 565), Tepehua (Tucanoan; Mexico; Watters 1988: 12), and Michoacan Nahuatl (Uto-Aztecan; Mexico; Sischo 1977: 313). This type is also represented by Iaai (Austronesian; New Caledonia; Tryon 1968: 62-63) and Paumarí (Arauan; Brazil; Chapman and Derbyshire 1991: 163-166).

There are also languages which are the mirror image of Muna, in which the order can be described as Absolutive-Verb-Ergative: these languages are shown as OVS on Map 81A and as SV on this map. In fact, three of the six OVS languages shown on Map 81A are of this type: Päri (Nilotic; Sudan; Andersen 1988), Mangarrayi (Mangarrayi; northern Australia; Merlan 1982) and Ungarinjin (Wororan; northwestern Australia; Rumsey 1982).

In some languages whose word order can be roughly described as ergatively based, the situation is often more complex than this characterization implies. For example, in Muna (illustrated above in (2)), subjects more often precede intransitive verbs if there is a prepositional phrase in the clause, as in (3). 

(3) Muna (van den Berg 1989b: 152) 














‘The dried fish fell to the ground.’ 

Hence the word order is sensitive to the presence of another nominal element (noun phrase or prepositional phrase) in the clause. Somewhat similarly, Salinan (Salinan; California) employs SVO order in transitive clauses containing a lexical subject and object and VS order in intransitive clauses, but again the pattern is not really an ergative one, in that transitive subjects also follow the verb if the object is pronominal and is realized only by the verb morphology (Dryer 1989). Hence, in Salinan, the order of subject and verb depends, not on the transitivity of the clause, but on whether there is a lexical object in the clause. Despite these complexities, both Muna and Salinan are shown on Map 81A as SVO and on this map as VS.

There are also languages in which there are positional differences between intransitive subjects and transitive subjects, but where the difference is not so categorical.  These include languages in which the dominant order for transitive subjects is SV, but where neither order is dominant for intransitive subjects. This type is found among a number of languages of Europe (e.g. Spanish, Bulgarian, Latvian). There are also languages in which intransitive subjects more commonly follow the verb, but in which neither order is dominant for transitive subjects. Examples of such languages include Domari (Macalister 1914), and Ostuacan Zoque (Mixe-Zoquean; Mexico; Engel and Longacre 1963: 335-336).

Among languages in which both orders of intransitive subject and verb are common and neither is dominant, a few exhibit what can be described as a split intransitive pattern, where more agentive intransitive subjects pattern with transitive subjects in preceding the verb while less agentive intransitive subjects pattern with objects in following the verb. This is illustrated in (4) for Arawak (Arawakan; Suriname); (4a) illustrates SVO order in a transitive clause; (4b) illustrates SV order in an intransitive clause with a more agentive subject; and (4c) illustrates VS order in an intransitive clause with a less agentive subject. 

(4) Arawak (Pet 1987: 108, 109, 161) 
















‘Yesterday, a man gave cassava bread to me.’ 














‘The man went home yesterday.’ 










‘The man is/was happy.’ 

The split in Arawak is more specifically governed by the stativity of the verb, rather than by some factor like volitionality: subjects of verbs denoting events precede the verb while subjects of verbs denoting states follow the verb. Thus the postverbal subjects occur with verbs that correspond semantically to adjectives in English. In Mokilese (Oceanic; Micronesia), subjects of intransitive verbs denoting events obligatorily precede the verb, while subjects of intransitive verbs denoting states may either precede or follow the verb (Harrison and Albert 1976: 299-300). 

Note that while this map is based on the order of subject and verb for intransitive clauses in the case of languages where it is known that there is a difference between transitive and intransitive clauses, some sources do not discuss differences between transitive and intransitive subjects.  It is possible that the behaviour of intransitive subjects in some languages is sufficiently different that further data would show that they are not coded correctly on this map.

2. Geographical distribution

SV order is clearly much more common than VS order and is widely found all over the world. But it is worth noting two areas which are overwhelmingly SV, in which both VS languages and languages lacking a dominant order of subject and verb are infrequent. One of these is mainland Eurasia, where the exceptions are primarily in Europe and in eastern Siberia. The other is New Guinea. Because VS is less common, I will concentrate here on describing where VS order is found.

Despite the instances noted above in which transitive and intransitive subjects differ in their position, the overall geographical distribution of VS order is rather similar to that of VSO and VOS languages on Map 81A. For most of the ways in which they differ, the differences involve languages which are shown as VS languages on this map, but not as VSO or VOS on Map 81A, so that there are more VS languages on this map than languages which are shown as VSO or VOS on Map 81A. There are two areas in Africa in which VS order is common: one is an area in eastern Africa extending up into Sudan, representing languages in various branches of Eastern Sudanic within Nilo-Saharan, including a number of Nilotic languages; the other is in North Africa, represented by a number of Berber languages. It should be noted that in earlier times, VS order was more widespread in North Africa and the Middle East, being the dominant order in many ancient Semitic languages and in Ancient Egyptian. There is a small pocket of VS Celtic languages in northwestern Europe; otherwise VS order is not attested in Europe or mainland Asia. VS order is widespread among Austronesian languages, though mixing with SV in a complex pattern. The areas among Austronesian languages in which VS is common include: (1) an area in and around the Philippines, extending north to Taiwan and south to the northern part of Borneo; (2) northern Sumatra and islands off the west coast of Sumatra extending north to include the non-Austronesian Nicobarese languages (which are Mon-Khmer); and (3) most Polynesian languages across the Pacific. I am aware of only one VS language among all the non-Austronesian (“Papuan”) languages of the New Guinea area: Kuot, spoken on New Ireland (Chung and Chung 1996: 1-4). There are a few VS languages in Australia, mainly in the southeast.  VS order is very common in two well-defined linguistic areas in North America: the Pacific Northwest, including the northwestern corner of the continental United States and the southwestern corner of Canada; and Mesoamerica, extending from southern Mexico into Guatemala, though not all languages of Mesoamerica are VS. It is scattered around South America, with the clearest concentration found in and around Bolivia.

There are two areas in which languages tend to lack a dominant order on this map that are roughly coterminous with areas in which languages tend to lack a dominant order on Maps 81A and 83A: these are Australia and North America. But there is one small area in which a number of languages are shown as lacking a dominant order on this map, where a number of these languages are shown as having a dominant order on Map 81A: this area is around the Mediterranean, including both Indo-European languages in Europe and Berber languages in Africa. It is also worth noting that languages lacking a dominant order of subject and verb are more common around regions in which VS order is common than around regions in which SV is common.

3. Theoretical issues

One theoretical question that arises is why SV order is so much more common than VS. A popular answer to this question is that it reflects a tendency for subjects to be topics in some sense (Tomlin 1986). In fact, as noted above, there are two subtypes of SV languages shown on this map: those which are fairly rigidly SV, and those with flexible order in which SV order is more common. One might surmise that in languages of the latter sort, SV order is more common because of a tendency for subjects to be topics, and that this is simply grammaticalized in languages of the former sort. However, evidence in the tradition of Givón (1983) shows that many languages seem to place what can be described as topical elements late in clauses, casting doubt on the notion that there is a universal pragmatic preference for topics to occur early in sentences. The question of why SV is preferred thus has not yet been satisfactorily answered.