Chapter Determining Dominant Word Order

by Matthew S. Dryer

Please note: This text is related to Chapters 81, 82, 83, 84, 85, 86, 87, 89, 90, 91, and 94.

1. Determining Dominant Word Order

A number of maps in this atlas show the dominant word order of various sets of elements, in most cases pairs of elements (like adjective and noun) but in some cases sets of three elements (like subject, object, and verb). For any set of elements, there are some languages in which only one order is permitted and other languages in which more than one order is permitted. Among languages of the latter sort, one can further distinguish languages in which one order is used more frequently than others from languages in which this is not the case. For example, with respect to the order of adjective and noun, there are languages which only employ adjective-noun order, others that only employ noun-adjective order, and still others that allow both orders. Among languages that allow both orders, there are some in which adjective-noun order is more frequent, some in which noun-adjective is more frequent, and some in which both orders occur with comparable frequency.

Where a language is shown on one of the word order maps as having a particular order as the dominant order in the language, this means that it is either the only order possible or the order that is more frequently used. The maps do not distinguish these two possibilities, because it is often not possible to obtain reliable information from descriptive grammars on whether a particular order which is not the most frequent order is grammatical or not. While a grammar may say, for example, that the order of adjective and noun in a language is adjective-noun, it often turns out that the alternate order is possible, either in special discourse contexts or in special grammatical contexts, so it is rarely possible to conclude with confidence that only one order is permitted.

The expression dominant order is used here, rather than the more common expression basic order, to emphasize that priority is given here to the criterion of what is more frequent in language use, as reflected in texts. The reason for assigning priority to this criterion is that for most languages, this is the only criterion for which we have any relevant information. When a language allows both orders of adjective and noun, for example, grammars will often mention this but describe one order as the normal order or the more frequent order. For some languages, the classification of a language in this atlas is based on actual text counts. The rule of thumb employed is that if text counts reveal one order of a pair of elements to be more than twice as common as the other order, then that order is considered dominant, while if the frequency of the two orders is such that the more frequent order is less than twice as common as the other, the language is treated as lacking a dominant order for that pair of elements. For sets of three elements, one order is considered dominant if text counts reveal it to be more than twice as common as the next most frequent order; if no order has this property, then the language is treated as lacking a dominant order for that set of elements. Of course, unless one examines a large number and a broad variety of texts, one cannot be sure that differences in frequency may not occasionally reflect the idiosyncratic properties of a particular set of texts. It is likely that in some cases, further text counts would lead to classifying a language differently.

For some languages, the classification on the map is based on a claim in the source that some order is basic or that it is pragmatically neutral. In the absence of evidence to the contrary, I assume that these are also the dominant orders. Occasionally, however, such claims are at odds with frequency data provided by the author. For example, Abbott (1991: 25) characterizes OVS order (object-verb-subject) in Macushi (Carib; Brazil) as basic, and says that SOV order is used to highlight the subject. However, she cites text count data that show that OVS and SOV order are about equally common. I base my classification of Macushi here on the frequency counts, and since no order is more than twice as frequent as the next most frequent order, I treat this language as lacking a dominant order of subject, object, and verb.

For some word order features where more than one order is possible, such as the order of object and verb, the order will generally be determined syntactically or by extragrammatical factors. But for other word order features, it may be largely determined by specific lexical items. For example, in languages with both prepositions and postpositions, it is generally the case that each adposition is either always a preposition or always a postposition. In such cases, the classification of a language as prepositional or postpositional is based here on a combination of whether the number of prepositions outnumbers the number of postpositions (or vice versa) and which adpositions express basic meanings and are thus likely to be used more frequently. For example, in Koyraboro Senni (Songhay; Mali), there are over a dozen postpositions but only three prepositions, and among the postpositions are a number with apparently higher frequency of usage, including one marking indirect objects, one marking locatives (covering meanings of ‘at’, ‘to’ or ‘from’), and one meaning ‘on’, while the prepositions tend to have more specialized meanings (‘since’, ‘until’, and ‘during’, though also ‘with’) (Heath 1999a). Because this suggests that postpositions are more frequent, Koyraboro Senni is classified here as postpositional. 

Similarly, in Korowai (Trans-New Guinea; Papua, Indonesia), all adjectives can precede the noun, as in (1a), but a few, like the adjective meaning ‘big’, can also follow the noun, as in (1b).

(1) Korowai (van Enk and de Vries 1997: 69) 








‘a bad teacher’ 








‘a very big person’ 


Again, it is assumed from this description that adjective-noun order is more frequent, and Korowai is treated on Map 87A as adjective-noun.

Some grammars will describe a particular word order as more contrastive. It is assumed from statements of this sort that the more contrastive order is used less frequently; hence the language will be coded according to the noncontrastive order. For example, in Asmat (Trans-New Guinea; Papua, Indonesia) adjective-noun order is described as contrastive, while noun-adjective order is neutral, as in (2).

(2) Asmat (Voorhoeve 1965b: 140)








‘good people’ 








‘good people (in contrast to bad people)’ 


The situation is similar in Ilocano (Austronesian; Philippines) except that the situation is reversed (Rubino 1998b): in Ilocano, the neutral order is adjective-noun, while noun-adjective order is contrastive. 

If a grammar indicates that both orders of a pair of elements are possible, without stating that one is more common or without any comment suggesting that one order is more common, then the language will be shown on the map as having both orders without one being dominant.

Similar considerations apply to determining the primary method of coding semantic or functional categories in chapters by Matthew Dryer. For example, in Chapter 69 (“Position of Tense-Aspect Affixes”), a particular morphological method of coding tense-aspect is considered primary if it appears to be at least twice as common as other methods in terms of text frequency.