In an applicative construction, the number of object arguments selected by the predicate is increased by one with respect to the basic construction. For example, in Tukang Besi (Austronesian; Sulawesi, Indonesia), the verb ‘fetch’ takes one theme object in the basic construction (as shown in 1a), but with the applicative marker it takes two objects, theme and benefactive (as shown in 1b).
Basic construction, two-place predicate
‘She fetched the wood.’
Applicative construction, three-place predicate
‘She fetched the wood (as a favor) for my mother.’
The term applicative is used to denote either the applicative construction or the verb in that construction. The term was first used by early Spanish missionary grammarians in the description of Native American languages, in particular Uto-Aztecan; it was later adopted by Bantuists and is now used for similar constructions all over the world. It is customary to restrict the designation applicative to those cases where the addition of an object is overtly marked on the predicate. Thus English pairs such as She baked a cake – She baked Oscar a cake do not count as basic-applicative alternation. The object added in the applicative construction is referred to as the applied object; if a verb without an applicative suffix takes an object, this latter object is referred to as a basic object. The verb from which the applicative is formed is called the base.
An applicative construction is a particular instance of a double object construction: in both, the two objects do not have to be distinguished by agreement morphology or case marking (see also chapter 105). However, the primary use of the double object construction is for the encoding of the event of transfer. The presence of an applicative construction per se often entails that a non-derived double object construction is also present. But the opposite is not the case; for example, Dutch and English have the double object construction but not the applicative construction. Some dialects of Tamil and some Australian languages seem to be an exception in having the applicative/causative but no double object construction (Polinsky 1995; Austin 1997; Tsunoda 1998).
Two main parameters in which applicatives vary cross-linguistically include the transitivity of the base and the semantic role of the applied object. The first map for this chapter reflects the values of these parameters.
With respect to the transitivity of the base, the main distinctions are between applicatives formed (i) from a transitive base only, (ii) from an intransitive base only, and (iii) from both bases (these distinctions are shown by different shapes of the symbols on the map). There are also constraints on the number of arguments of the base verb; they are not shown in the map because they are harder to determine on the basis of grammars.
With respect to the semantic roles of the applied object, the most common role of the applied object is that of benefactive. Accordingly, the map differentiates applicatives whose applied object (i) is limited to benefactive; (ii) corresponds to the benefactive and some other roles; (iii) corresponds to other roles to the exclusion of the benefactive. These distinctions are shown by the different colours of the symbols on the map.
Of nine logically possible values (three different shapes, three different colours), seven are actually attested. In addition, there are of course many languages without applicatives. The map thus shows the following eight values:
|Go to map|
|Benefactive object only; both bases||16|
|Benefactive object only; transitive base only||4|
|Benefactive and other; both bases||49|
|Benefactive and other; transitive base only||2|
|Non-benefactive object only; both bases||9|
|Non-benefactive object only; transitive base only||1|
|Non-benefactive object only; intransitive base only||2|
|No applicative construction||100|
Applicatives are common in three geographical areas: Africa (mostly in Bantu), the western Pacific region (Austronesian), and North and Meso-America (Salish, Mayan, Uto-Aztecan). The main generalization seems to be that applicatives are commonly found in those languages that have little or no case-marking of noun phrases in a clause and that have sufficiently rich verbal morphology to mark applicative formation on the predicate. The dearth of applicatives in Eurasia may thus be due to the widespread presence of rich nominal morphology in the languages of that area, and indeed, where applicatives are found is in languages like Abkhaz and Abaza (Northwest Caucasian; Georgia and Russia), which have little or no overt case marking of noun phrases.
Map 109A reflects the distribution of applicatives with respect to the semantic role benefactive, clearly the most common semantic role of the applied object. Other common semantic roles include location and instrument. The geographical distribution of these roles is shown on Map 109B.
|Go to map|
|Instrument and locative||12|
|No other roles (= Only benefactive)||36|
|No applicative construction||100|
Additional semantic functions that may be associated with the applied object include possessor, circumstance/event (time), comitative, and substitute (a participant on whose behalf the action is performed). Comitative applicatives formed from intransitives seem to be particularly common in the languages of Australia. Comitatives and substitutives are quite common in applicatives of intransitives, e.g. in Lai (2) and in Kinyarwanda (3):
‘He grew up with me.’ [comitative]
(3) Kinyarwanda (own field notes)
‘The man is travelling instead/on behalf of the woman.’ [substitute]
It is somewhat puzzling that some languages of Australia show a comitative applicative formed from an intransitive without a parallel benefactive applicative formed from a transitive. Such comitative applicatives may be marked by the same morphology as instrumental applicatives formed from transitives. It is possible that these Australian-type applicatives and applicatives elsewhere represent different phenomena.
Applicative verbs and constructions have generated a significant amount of research dealing with the morphosyntax of applicatives, transitivity, VP structure, distinct object relations, argument structure, and mapping from argument to syntactic structure. For an overview of theoretical issues, see Alsina 1996, Alsina and Mchombo 1993, Peterson 1999.
The intransitive base of applicatives is less common than the transitive base. This is quite clear from Map 109A, and there are only two languages in the sample that form applicatives from the intransitive base exclusively (Fijian, Wambaya). The overall tendency is that if a language has applicatives formed from the intransitive base, it also has applicatives formed from the transitive base.
The following explanation may be offered as to why the applicative is uncommon with intransitives. Adding an object to an intransitive amounts to creating a transitive verb. In order to transitivize an intransitive, languages typically use causativization, thus increasing the complexity of the event structure (V > CAUSE V); with the applicative, the event structure is not modified, just another participant is added. Under causativization, the argument added to the argument structure of the verb is agent; under simple transitivization, the added argument is theme (or patient). Note that applicative formation results in adding arguments other than agent and theme, and may thus be constrained by the general hierarchy of semantic roles: agent > theme (patient) > goal (recipient, benefactive) > location > other roles.
Although the constraint on applicative formation from intransitives seems not to be absolute, a particular subset of intransitives, namely unaccusative predicates (those whose subject originates as an object in the underlying structure), resist applicativization (see Baker 1998). However, even this generalization does not hold in some languages, for instance, in Halkomelem (Gerdts 1988a), Lai (Peterson 1999), and Sesotho (Machobane 1989), which suggests that it is just a strong tendency.
With respect to the formation of the applicative predicate, much discussion has centered on the question of which object, basic or applied, is structurally superior. Some researchers have proposed that the two objects may be structurally equal (Gary and Keenan 1977; Seiter 1980), but most analyses place the base object higher or lower than the applied object.
In some languages both objects, basic and applied, are accessible for passivization and relativization, can bind a reflexive, can trigger agreement on the verb, or can license coreferential deletion across clauses. Such languages are called “symmetrical”. In other languages, only one object, either applied or basic, can show the relevant grammatical behaviors, while the other object is syntactically quite inert. Such languages are called “asymmetrical” (Woolford 1993; Alsina and Mchombo 1993). The asymmetrical type seems to be more common.
The underlying structural superiority of the base object is typically claimed in the derivational account of applicatives which is based on the analysis in Baker (1988). According to Baker, applicative formation is the result of the incorporation of a preposition into a verb—thus, what appears to be an applicative affix starts out as a preposition introducing an oblique object. The oblique becomes an object following preposition incorporation.
The view of applicative formation as preposition incorporation has played an important role in approaches to applicative morphology. The basic idea is that morphological derivations must directly reflect syntactic derivations (and vice versa) (Baker 1985; Alsina 1999), and this isomorphism should be reflected in affix ordering (the so-called Mirror Principle). Crucial evidence for the Mirror Principle has been drawn from the interaction between applicative morphology and morphology of other morphosyntactic operations (passive, reciprocal) in Bantu.
Instances of multiple applicative marking and multiple applied objects are found in Koyraboro Senni, Bantu languages, Tukang Besi (and possibly other Austronesian languages), and Abaza. In Cahuilla, applicative formation is achieved either via prefixation (for adding a locative object) or via suffixation (for adding benefective/recipient), which suggests that the diachronic origin of these markers may be different. In those cases where identical applicative morphemes allow iteration, as in Bantu, it is unclear whether their number is constrained by grammar or by processing limitations.
It is sometimes hard to tell from grammatical descriptions whether benefactive is differentiated from recipient/goal, or other common roles associated with the applied object. It is also unclear how many languages merge benefactive and malefactive (the adversely affected object); impressionistically it seems more typical to keep them apart. In addition, the applied object can be mapped onto a possessor, whose place in the hierarchy of semantic roles is not quite clear. Benefactives, goals, and possessors are typically animate participants, which may create an impression that the applied object has to be an animate participant. Indeed, in some languages (e.g., Halkomelem) the referent of the applied object must be animate regardless of its semantic role (Gerdts 1988a; Gerdts 1993.) In some languages, e.g. Kinyarwanda, the linear order of the base and applied object varies depending on animacy (Kimenyi 1988; Polinsky and Kozinsky 1992; Polinsky 1995). Overall, animacy probably constitutes a separate dimension in the semantics of the applied object.