In the European languages, the same marker is used both for conjunction of noun phrases and for conjunction of verb phrases and clauses, as illustrated in (1) for Hungarian (and for English in the translation). Thus, Hungarian és and English and can conjoin both nominal and verbal/clausal constituents.
But in many non-European languages, different conjunction markers are used for different constituent types. An example comes from Dagbani (Gur, Niger-Congo; Ghana), where mini conjoins noun phrases (cf. 2a) and ka conjoins verb phrases (cf. 2b) and clauses (cf. 2c).
'this man and my father'
'He is bad and thin.'
'Lions eat meat and monkeys eat bananas.'
Since clauses generally contain a verb phrase, I simply use the term verbal conjunction for both conjunction of verb phrases and conjunction of clauses.
The main contrast is thus between identity and differentiation of nominal and verbal conjunction. However, it seems useful to single out one further type: languages in which there is no overt marker either for nominal or for verbal conjunction, so that conjunction is expressed exclusively by juxtaposition. Strictly speaking, this is a subtype of the first type, but it is treated separately here because of its interesting geographical distribution. Example (3) is from Nhanda (Kardu, Pama-Nyungan; Western Australia).
'our clothes and our rugs'
'You go in the front and I'll go behind.'
However, when just one of the two constituent types can be conjoined by juxtaposition, the language is classified as differentiating. This is actually quite common. In particular, many languages have an overt conjunction marker for nominal conjunction but use juxtaposition for verbal conjunction (e.g. Hausa).
Thus, the map shows the distribution of the following three language types:
|Nominal and verbal conjunction are largely identical||161|
|Nominal and verbal conjunction are different||125|
|Nominal and verbal conjunction are primarily expressed by juxtaposition||15|
In illustrating conjunction markers, I will sometimes use the variables A, B, etc. for nominal conjuncts, and the variables P, Q, etc. for verbal conjuncts (including clauses). For instance, Dagbani is said to have the markers A mini B and P ka Q.
It is of course not uncommon for languages to possess several ways of conjoining noun phrases, and/or several ways of conjoining verb phrases and clauses. This is unproblematic when the competing markers have exactly the same range of uses. For instance, Madurese (Madurese, Austronesian; Java, Indonesia) has the two conjunction markers ban and biq, and these are both used both for nominal and for verbal conjunction (A ban B, A biq B, P ban Q, P biq Q; Davies 1999: 25). Another example is Wardaman (Yangmanic; Northern Territory, Australia), where there are two different markers for nominal conjunction (A warrma B, and A B wayana ) and one marker for verbal conjunction (P gabani Q ) (Merlan 1994: 68, 87, 89). In contrast to Madurese, where there is complete overlap between nominal and verbal conjunction markers, Wardaman shows no overlap at all; hence Madurese is classified as "largely identical", and Wardaman is classified as "different".
But quite a few languages show partial overlap of their nominal and verbal conjunction markers. In such cases, the language is generally classified as "largely identical". For example, Southeastern Tepehuan (Southern Uto-Aztecan; Durango, Mexico) has two conjunction markers, gam and guio. Gam only links noun phrases describing entities of a similar nature, while guio links both noun phrases and clauses (Willett 1991: 216-217). And in Tinrin (Oceanic; New Caledonia), mwâ only conjoins clauses (verb phrases cannot be conjoined in Tinrin), whereas mê conjoins both clauses and phrases. The difference between mwâ and mê in clause conjunction is that mê is used when the two clauses are parallel and mwâ is used when they contrast in some way (Osumi 1995: 258).
However, cases of partial overlap are classified as "different" under two circumstances: (i) if the overlap concerns a marker which is a clearly minor alternative for at least one of the constituent types, and (ii) if the overlap concerns the juxtaposition strategy. Thus, Koromfe (Gur, Niger-Congo; Burkina Faso and Mali) was classified as showing different conjunction markers, although its nominal marker la (A la B ) can also be used for verbal conjunction (P la Q ). However, juxtaposition is by far the more common means of verbal conjunction in the language, and when it combines two clauses, la mostly means 'but' rather than 'and' (Rennison 1997: 88, 92). Thus, la was judged to be a minor alternative for verbal conjunction. Similarly, in Imonda (Border family; Papua New Guinea), the major alternatives for nominal conjunction are A-i B-i (for human conjuncts, e.g. ehe-i ka-i 'he and I') and A-na B-na (e.g. sa-na fo-na 'coconut and banana'). For verbal conjunction, the major alternative is juxtaposition, and there are two other markers: P-ie Q-ie, and P-mo Q. The element -mo also occurs in a minor nominal conjunction type, A-mo B-mo C-mo ..., which is used for enumeration (Seiler 1985: 68-70, 102, 196-197). Thus, -mo occurs both in nominal and in verbal conjunction, but not as a major option in either, so it was disregarded.
Juxtaposition is extremely common as an alternative option, especially for clausal conjunction, but also for nominal conjunction. It is therefore not very telling if a language uses it for both constituent types; hence if there are also overt conjunction markers, only these were taken into account. For example, in Trumai (isolate; Mato Grosso, Brazil), juxtaposition is a possible means of conjoining both noun phrases and verb phrases/clauses, but there are two other nominal conjunction markers (A B a, A B tam ) and several other verbal conjunction markers (e.g. P inis Q ) (Guirardello 1999a: 19-21, 367-370), so Trumai was classified as "different".
Conjunction is a subtype of coordination. We are dealing with a coordinating construction when there are two identifiable constituents which have the same semantic role and together form a larger constituent (Haspelmath 2005). Conjunction is the type of coordination that is translated by 'and', as opposed to disjunction ('or') and adversative coordination ('but').
There are usually few problems in recognizing nominal conjunction. Difficulties may arise when the conjunction marker is the same as the comitative adposition (or case affix) 'with', as is the case in a great many languages (see Chapter 63). Consider, for instance, the case of Babungo (Grassfields Bantoid, Niger-Congo; Cameroon), where nə̀ is translated as 'and' or 'with'. However, it is quite clear that nə̀ in (4) is a conjunction marker:
'Lambi, Ndula and Kometa went to the market.'
There are four reasons here why nə̀ 'and' cannot be simply identical with nə̀ 'with': It occurs within the subject noun phrase rather than after the verb (as adverbial expressions do); it occurs twice and links three NPs to each other; it need not imply that the referents carried out the action jointly; and finally, nə̀ 'with' no longer has the comitative meaning which must have been the original source of the conjunction meaning – it is now confined to the instrumental meaning 'with' (e.g. nə̀ fə̀ntɨ̄ 'with a stick'; Schaub 1985: 145). Although conjunction markers are often similar to comitative adpositions because they tend to develop from them diachronically, the two are in general fairly easy to tell apart. Whenever a marker is part of a noun phrase and is systematically translated as 'and' or described as a coordinating conjunction, it is regarded as a conjunction marker here.
Conjunction of clauses is more difficult to identify. On the one hand, it is not always clear whether we are dealing with a sequence of independent sentences or a complex sentence consisting of two conjuncts. If there are no overt markers, the best general criterion for recognizing conjunction is intonation, but since this is not described systematically in grammars, it is sometimes necessary to rely on the translation: When a clause-combining strategy is systematically rendered by 'and', then it is regarded as conjunction.
On the other hand, it is often difficult to distinguish between verbal/clausal conjunction and verbal/clausal subordination. In many languages, the most natural translation of 'A did X and B did Y' is by means of a special dependent verb form (sometimes called converb or medial verb) in one of the clauses. Examples come from Japanese and Tauya (Brahman, Trans-New Guinea; Papua New Guinea).
'Taro went to America, and Hanako went to France.'
'I went and he ate.'
The verb forms marked by -i in Japanese and -te in Tauya do not occur in independent clauses, so in that sense the first clauses in (5-6) could be regarded as subordinate, and one could claim that a more literal translation of (5) would be 'Taro having gone to America, Hanako went to France'. However, when other criteria for the suboordination distinction are taken into account, these sentences do not come out consistently as subordinate (see Haspelmath 1995 for general discussion, and Alpatov and Podlesskaya 1995 for Japanese). They thus show that the subordination/coordination distinction cannot be drawn in a straightforward way.
For the purposes of this chapter, a clause-combining strategy is regarded as a type of conjunctive strategy if it is consistently translated by 'and' and not by more concrete adverbial markers (such as 'after', 'while', or 'and then'). This is again a fairly crude criterion, but many grammars do not provide much more information. For a number of languages, no information about conjunction of verb phrases and clauses was available, but the nominal conjunction markers are described as "confined to noun phrase conjunction", so that it could be inferred that the nominal and verbal conjunction markers must be "different". Note that I assume that all languages have some way of expressing both nominal and verbal conjunction.
Some languages use a different marker for verb phrase conjunction and clause conjunction. For instance, Somali has three different markers: A iyo B for nominal conjunction, P oo Q for verb phrase conjunction, and P Q-na for clause conjunction:
'bread and fruit'
'He ate and drank.'
'The teacher is in the school, and the children are playing outside.'
Somali is straightforwardly classified as showing "different" nominal and verbal conjunction. But in a few languages, the marker for verb phrase conjunction is identical to the nominal conjunction marker, but different from the clausal marker. Such a language is Chamorro (Western Malayo-Polynesian; Guam).
'Juan and Maria'
'The woman is crying and laughing.'
'I wanted the book and I bought it.'
In such cases (which are quite rare), nominal conjunction is compared with clausal conjunction, so that these languages are classified as "different".
Differentiation of nominal and verbal conjunction markers is particularly widespread in African languages (as was noted by Welmers 1973: 305), but it is found in many parts of the world. Identity of the two markers is the only type found in Europe and Mesoamerica, but it is also found widely in the rest of the world. Juxtaposition is particularly common in Australia and South America.