Chapter Nonperiphrastic Causative Constructions

by Jae Jung Song

1. Defining nonperiphrastic causative constructions

The causative construction is a linguistic expression which denotes a complex situation consisting of two component events (Comrie 1989: 165-166; Song 2001: 256-259): (i) the causing event, in which the causer does or initiates something; and (ii) the caused event, in which the causee carries out an action, or undergoes a change of condition or state as a result of the causer’s action. The following Japanese sentence is such a linguistic expression.

(1) Japanese

Kanako

ga

Ziroo

o

ik-ase-ta

Kanako

nom

Ziro

acc

go-caus-pst

‘Kanako made Ziro go.’

In (1), the causer (Kanako) did something, and as a result of that action the causee (Ziro) in turn carried out the action of going. Map 111A shows the geographical distribution of nonperiphrastic causative constructions.

Nonperiphrastic causative constructions are causative expressions with the following three properties. First, the expression of the causer’s action (e.g. -ase in (1)) and the expression of effect (e.g. ik- in (1)) must both be contained in one and the same predicate, which may consist of one or more verbs (e.g. one verb ik-ase- in (1), or two verbs me ŋò in (6) below). To put it differently, such causative expressions must be monoclausal. Second, the causer noun phrase must occupy a grammatically more ‘prominent’ position (e.g. the subject in (1)) than the causee noun phrase (the object in (1)). Third, the expression of the causer’s action, be it an affix or a separate verb, should be without specific meaning. In (1), the expression of cause -ase, as opposed to the expression of effect ik-, lacks specific meaning; all that is expressed by -ase is the pure notion of causation.

Causative expressions such as (2), taken from Manam (Oceanic; Papua New Guinea), are not considered here, because the causer’s action and the effect are expressed by different predicates. These predicates in turn appear in different clauses. (Causative expressions such as (2) are the topic of Chapter 110.)

(2) Manam (Lichtenberk 1983: 447)

wása

ʔúsi

i-emaʔ-í-be

i-moaʔúsu

wind

cloth

3sg.real-cause-3sg.obj-and

3sg.real-move

‘The wind made the loin cloth move.’

Monoclausal causative expressions such as (3), taken from Tuvaluan (Polynesian), are also not included in the present discussion, because the causer’s action, if expressed at all, is not contained in the predicate (although the notion of effect is).

(3) Tuvaluan (Besnier 2000: 325)

te

paala

e

tasi

ne

sao

ia

Aaifoou

the

kingfish

npst

one

pst

escape

because.of

Aifou

‘The kingfish got away because of Aifou.’

2. Defining the values

As indicated in the feature-value table, there are two types of nonperiphrastic causative construction: the morphological type and the compound type. Thus four values are represented on the map:

Values of Map 111A. Nonperiphrastic Causative Constructions
 ValueRepresentation
No morphological type or compound type 23
Morphological type but no compound type 254
Compound type but no morphological type 9
Both morphological type and compound type 24
Total:310

The morphological type involves a morphological process which applies directly to a basic verb (Song 1996: 21-26; Dixon 2000: 33-34). Typically, this morphological process consists in affixation: in the morphological type the predicate — or the causative verb — is made up of a basic verb and a causative affix (Comrie 1989: 167-168). In Japanese, for example, the causative suffix -(s)ase applies to basic verbs to derive causative verbs, as illustrated in (1) above. Compare (1) with (4), the noncausative counterpart of (1).

(4) Japanese

Ziroo

ga

ik-u

Ziro

nom

go-pres

‘Ziro goes.’

The causative element can take the form not only of suffixes (e.g. -(s)ase in Japanese), but also prefixes (e.g. r- in Abkhaz), infixes (e.g. -y- in Lepcha) and circumfixes (e.g. a- … -ineb in Georgian) (Song 1996: 21-28).

The morphological process may also involve an internal change in vowel or consonant quality (e.g. Lahu dɔ̀  ‘drink’ vs.   ‘give (i.e. cause) to drink’), repetition of a consonant in the basic verb (e.g. Egyptian Arabic mawat ‘die’ vs. mawwit ‘kill’), internal vowel lengthening (e.g. Kashmiri marun ‘die’ vs. mārun ‘kill’), reduplication of the basic verb (e.g. Korana xa  ‘learn’ vs. xa-xa  ‘teach’) or even a tonal change (e.g. Lahu   ‘eat’ vs. cā  ‘feed’). There is one other variation on the morphological type worth mentioning here. In Russian, for example, it is noncausative, not causative, verbs that bear an additional morpheme, i.e. the anticausative suffix -sja  (e.g. lomat’  ‘to break [something]’ vs. lomat’-sja  ‘[for something] to break’). For the purposes of Map 111A, the Russian causative verb also will be regarded as an instantiation of the morphological type: in comparison with the noncausative verb, it can be thought to contain a zero causative morpheme. In the sample, there are only a small number of languages which display such variations on the morphological type like those found in Lahu, Egyptian Arabic, Kashmiri, Korana and Russian.

In the compound type, the causer’s action is expressed by a separate verb instead of a morphological element, but that verb must appear next to a basic verb so that other elements are not able to intervene between the two. Thus the two verbs behave as a single predicate (i.e. a compound causative verb), very much as the basic verb and the causative suffix in (1) are a single predicate. Romance languages such as French and Spanish are described as languages having the compound type. In French, for example, the two verbs in question, albeit separate lexical items, must appear next to each other, so that the causee noun phrase (and the object noun phrase of the basic verb) is (are) prevented from breaking up the unit, as illustrated in (5). The negative element pas or adverbials can in fact intervene between the two verbs, but this does not detract from the fact that normally nothing comes in between them. (For further discussion, see Comrie 1976b: 296-303.)

(5) French

Je

le

lui

ferai

ire.

I

it.acc

her.dat

make.fut

read

‘I’ll make her read it.’

In Eastern Kayah Li (Tibeto-Burman; Myanmar) the verb of cause and that of effect must abut on each other, as in (6). Note that the causer noun phrase ʔa  and the causee noun phrase phúcè  appear before and after the sequence of both verbs.

(6) Eastern Kayah Li (Solnit 1997: 69)

ʔa

me

ŋò

phúcè

he

do

laugh

child

‘He made the child laugh.’

It is not merely the physical adjacency of the verbs of cause and effect that allows one to recognize the compound type, but also the ability of these two verbs to form a grammatical unit. This is demonstrated, for example, by the difference between Tamil (Dravidian; southern India and northern Sri Lanka) and Kobon (Madang; Papua New Guinea). The Tamil causative sentence in (7) is not regarded as being of the compound type in spite of the fact that the verb of cause and that of effect are adjacent to each other.

(7) Tamil (Asher 1985: 155)

naan

avane

veekamaa

ooʈ-a

vacceen

I

he.acc

quickly

run-purp

cause.pst.1sg

‘I made him run quickly.’

The reason why (7) is not taken as an example of the compound type is that the verb of effect, ooʈ-a, actually contains purposive marking, -a. The presence of this purposive marking in the verb of effect indicates that the two verbs do not form a unit, and that ooʈ-a  is part of a subordinate clause of purpose. Thus (7) should instead be regarded as an example of the periphrastic causative construction. In fact, the two verbs in (7) end up next to each other mainly because of the verb-final word order of Tamil. The verb of effect, being in the final position of the subordinate clause, appears right before the verb of cause (i.e. the verb of the main clause). Kobon, also a verb-final language, has the verb of cause and that of effect next to each other, but unlike in Tamil, there is evidence that the two verbs form a kind of grammatical unit, which is characteristic of the compound type. Consider:

(8) Kobon (Davies 1981b: 164)

a.

mab

dudu.g-öp

 
 

tree

be.bent-perf.3sg

 
 

‘The tree is bent.’

b.

yad

mab

dudu.gɨ

yu-bin

 

1sg

tree

be.bent

throw-perf.1sg

 

‘I bent the tree.’

(8a) is a noncausative sentence, whereas (8b) is a corresponding causative sentence. (Note that in (8b) the verb of cause yu-  has lost its original meaning of throwing; recall that the expression of the causer’s action should be without specific meaning.) In (8b), the causer’s action is expressed by a separate verb yu-, which appears immediately after the verb of effect dudu.gɨ. In (8b), the single aspect marker on the verb of cause has its scope over the whole unit, supporting the view that the two verbs form a compound causative verb.

Languages such as Modern Greek, Kilivila, Krongo and Maybrat use neither the morphological type nor the compound type. Thus readers may wonder how the causative situation, as defined at the beginning of section 1, is expressed in these languages. They depend on periphrastic causative constructions. Languages like these are a minority, however.

When reading Map 111A in conjunction with Map 110A, readers will notice that many of the languages appearing on the former are not represented on the latter. As mentioned also in the companion chapter, this is largely because the primary sources for many of these languages discuss only the morphological type without even indicating whether or not a periphrastic causative construction is in use as well. (This explains why Map 111A displays more languages than Map 110A does.) Moreover, it is well known that the morphological process involved in the nonperiphrastic causative is rarely completely productive (e.g. Nedjalkov and Silnitsky 1973; Song 1996: 170-172). In view of this, it is very likely that most of the languages shown as employing only the morphological type may also have other means of expressing causation, i.e. the periphrastic causative. These points must be borne in mind when interpreting the two maps together. Still, a good number of languages are identified on the maps as making use of both the morphological type and the periphrastic causative.

3. Geographical distribution

In view of the prevalence of the morphological type in the world’s languages, it is not inappropriate to describe its geographical patterning in negative terms, that is to say, with respect to its absence. There are languages lacking the morphological type in all major geographical areas: Africa, Eurasia, Southeast Asia-Oceania, Australia-New Guinea, North America and South America. Within Eurasia, languages without the morphological type are found only in Europe, namely Modern Greek and Irish. In each of the major geographical areas, however, languages without the morphological type are too few or scattered to reveal clear geographical patterns, except for possible clusters of such languages in Southeast Asia (Mandarin, Hmong Njua, Kayah Li, Mulao, Thai, Vietnamese and Yay) and northwestern Australia (Djaru, Gooniyandi and Ungarinjin).

There are a very small number of languages with the compound type and without the morphological type in the sample. In all areas where these languages are located, however, they are too few and far between to suggest any distinct geographical patterns.

As pointed out in section 2, there are languages without either the morphological type or the compound type, for example, Modern Greek, Kilivila, Krongo and Maybrat. Given the very small number of languages with the compound type and without the morphological type, it comes as no surprise that the geographical patterns suggested above for languages without the morphological type also more or less hold true for languages without the morphological type and the compound type. Thus there are two potential clusters of languages lacking the nonperiphrastic causative: Southeast Asia (Hmong Njua, Mulao, Thai and Yay) and northwestern Australia (Djaru, Gooniyandi and Ungarinjin).

Finally, languages with both the morphological type and the compound type are not numerous, but there are at least two areas where such languages seem to cluster together: Southeast Asia (Bawm, Burmese, Khmu’, Lahu and Lalo) and the southeastern part of North America (Biloxi, Koasati and Tunica). However, more data will be required to substantiate these clusters.

4. Theoretical issues

For the past three decades or so, the causative construction has been a recurrent research topic in linguistics. Most research has focused on the morphological type (and also the compound type as a variation on the morphological type) with respect to the grammatical relation of the causee noun phrase, and on the causative as a valency or grammatical relation changing operation (Comrie 1976b and Dixon 2000; also see Song 1996: ch. 6). It is claimed that the grammatical relation of the causee noun phrase in the morphological type of causative — which is assumed to consist of a matrix clause and an embedded clause at some abstract level — can be predicted by appealing to a hierarchy of grammatical relations, although some linguists have argued that the morphological type of causative is modeled conceptually and grammatically on simple clauses. Formal theories also have proposed different approaches to representation of causativization as a grammatical relation changing operation. These approaches can be characterized broadly as lexical or syntactic.