Syntactic Developments in Sranan

Creolization as a Gradual Process

[p. 91]

Chapter Four: Clefting

4.1. Introduction

The position of this chapter at the end of the descriptive part of this diachronic study to some extent reflects its content, i.e. in the sense that the construction under discussion (predicate and non-predicate clefting) does not exhibit that much historical change to begin with¹. Although as such this result is not exactly what one is looking for in an investigation of syntactic change, it is nevertheless important in its negative outcome, especially when it is seen in its relation to the quite significant changes in the copula and comparative system that have been described in the preceding chapters. On the other hand, although what changes can be observed may be less striking in most cases or even marginal in some, this does not mean that they are less interesting from a purely linguistic point of view.

The fact remains, however, that the syntactic mechanism of clefting as it appears in the modern language was by and large established by the middle of the 18th century and shows no dramatic development in the two centuries after that. It will be clear that this does not fit in very nicely with the gradual, transgenerational view of creolization that was put forward earlier in this study on the basis of the rather slow stabilization of the other two construction types. Of course, there is a possibility - at least theoretically - that clefting has gone through important changes in the pre-1750, i.e. undocumented period, but even if this would be the case (which is unlikely in view of the fact that pre-1750 Sranan was more of a pidgin than a creole language), the rate of stabilization would still be much faster than it is for the other constructions.

This forces us to amend our ‘model’ of creolization to the extent that creolization is not only not an instantaneous, discrete process - as opposed to a gradual, continuous one - , but also that it is not uniform and constant in the rate with which it operates, but variable and differential instead. An important question for future research will be to decide what the factors are that make some structures ‘settle’ faster than others in the creolization process.

As far as clefting is concerned, I will venture the hypothesis that it is the absence of conflict between source and target languages which is the determinant factor. As far as predicate clefting is concerned, this is quite obvious, since it does not occur in English (nor in any other Germanic or Romance language). As to non-predicate clefting, this

1 The data for this chapter was drawn from the same sources that were used for the examination of the copula. The overall sample numbers a total of 298 cleft sentences.

[p. 92]

syntactic procedure is largely similar in sub-and superstrate languages, although there are some differences with respect to the position of the ‘cleft marker’ and the presence of a ‘complementizer’ between the ‘focus’ and the ‘non-focus’ part of the sentence². (In section 4.2.1. I will go into the question why Sranan may have chosen to place its cleft marker sentence-initially and to leave its complementizer unexpressed.)

It is important to note, in my opinion, that the adaptation of the view of creolization referred to above is in accordance with the idea that creolization is not a monocausal, but rather a multi-causal process in which no single factor can be held responsible to the exclusion of all the others. Where many factors influence a process it is only natural for this to be a composite and multiform instead of a monolithic and uniform one.

The structure of this chapter is as follows: first, in section 4.2., some formal and functional aspects of clefting will be described, according to whether they occur in all clefts (section 4.2.1.) or only in predicate (section 4.2.2.) or non-predicate clefts (section 4.2.3.). Then, in section 4.3., the developmental change in this construction will be summarized and discussed. Finally, in section 4.4., we will go into the consequences of these findings for the concept of creolization.

2 The terminology employed in this chapter will be as follows: a cleft sentence is a sentence in which an element is intensified or emphasized by moving it to the head of the sentence and accompanying it with a cleft marker. In Sranan this is the copula da/na or the negative copula a (no); in other languages this may be a determiner, as in Ewe, or a ø, as in Igbo with direct objects. If the fronted element is a predicate (i.e. a verb or an adjective), this is copied in the original position. A cleft sentence may be divided into two parts, the focus (=clefted element + cleft marker) and the non-focus (=the remaining part of the sentence); the two parts may have an element in between them, usually a relativizer or a conjunction, which we will refer to as the complementizer. Thus in the sentence da duysi tongo joe leri? (Van Dyk c.1770: 29), lit. ‘is Dutch language you learn?’, i.e. ‘are you learning Dutch?’ the following elements can be discerned: da = the cleft marker; duysi tongo = the clefted element; da duysi tongo = the focus; joe leri = the non-focus; the complementizer, as usual, is ø.

4.2. Form and function of cleft sentences

4.2.1. General aspects of clefting

This section will deal with aspects that are relevant to both types of clefting, i.e. the cleft marker, the complementizer

[p. 93]

and negation.

The cleft marker. Two aspects of the cleft marker will be discussed: its form and its position. As to the first, it can safely be assumed that the cleft marker da/na, which is obligatorily present in Sranan (but see below), is identical to the equative copula and not to the homophonous determiner. This is so because the phonological change from da to na coincides chronologically with that in the equative copula and not with that in the determiner. Na as a cleft marker is first attested in Cesaari (1836-37), in King (1864-1870) and King (1891-1894), and in Albitrouw (1894), exactly the same sources where equational na appears for the first time; the form of the determiner in all these sources is still da. It is important to note this, since determiners do sometimes occur as cleft markers (for instance in Ewe; see Westermann 1907: 104). The fact that the cleft marker in Sranan is identical to the equative copula is in accordance with the claim that all cleft sentences, universally speaking, are derived from underlying equationals (e.g. Harries-Delisle 1978). In a later source (Herskovits 1936) the cleft marker is sometimes reduced to 'a or a; the first of these reduced forms can also be found in equative sentences in this source. Some examples will serve to illustrate this development:

(1)	da	da disi joe wanni? (Focke 1855: 149)
		is this you want

‘do you want this one?’

(2)	na	na Noah djaroesoe jagi poeroe den na ondro
		is Noah jealousy chase pull them LOC under

na kroisi foe joe (King 1891-94: 52)

the cross of you

‘it's Noah's jealousy that has chased them away from your cross’

(3)	'a	'a so a pikin, 'a so a sa tan
		is so he small, is so he TNS stay

‘he will remain as little as he is’

(Herskovits 1936: 158)

(4)	a	a so gronman fom Keskesi te...a kiri hem
		is so farmer beat Monkey till he killed him

‘that's how the farmer beat Monkey to death’

(Herskovits 1936: 166)

The negative cleft marker has the form da no in all 18th century sources, with the exception of Schumann (1783) where besides da no also no is used, and a no in all post-1800 texts. (See section 2.2 for an explanation of the ordering (d)a no instead of no da/na in the negative copula; the fact that the negative cleft marker has this form is further evidence that it is a copula and not a determiner.) Compare the following examples:

[p. 94]

(5)	da no	da no boi gudu da somma habi
		is not small wealth that person have

‘that person is very wealthy’

(Schumann 1783: 65)

(6)	a no	a no brede wawan de holi soema liebi
		is not bread only ASP keep person alive

‘man does not live by bread alone’

(Luke 1829: 26)

(7)	no	no so mi ben taki? (Schumann 1783: 207)
		not so I TNS say

‘didn't I say just that?’

The absence of a in negative clefts such as (7) may well be related to its absence in negative equations in general in 18th century texts. In these sources about 35% of all equative copulas are ø (see Fig.1), a large part of these occurring in negative sentences such as the following:

(8)	mi no negeri fo joe (Van Dyk c.1770: 53)
	I not negro for/of you

‘I'm not your slave’

Despite the absence of a cleft marker the classification of sentences like (7) as clefts is fully warranted because of the marked position of the negative element: if (7) were not a cleft, it would occupy its canonical position, which is preverbal in Sranan, as in creoles in general. In other words, sentence (7) would read:

(9)	mi no ben taki so?
	I not TNS say so

‘didn't I say so?’

The absence of a cleft marker is not at all exceptional from a typological-universal point of view, especially in languages with a ø equative copula like early Sranan (Harries-Delisle 1978: 425). Thus, in Igbo for instance the copula can be omitted when direct objects are clefted:

Igbo

(10)	ulo ka m'huru (Spencer 1924: 110)
	ø a-house COMP I-saw

‘I saw a house’

As to the second aspect of the cleft marker, its position, it can be observed that, with very few exceptions, this is always clause- or sentence-initial, and, as in most other creoles (but see Holm 1980: 374, note 4), placed directly before the fronted element. This is rather surprising since this is a universally marked position: its unmarked position is between the fronted constituent and the remaining sentence (Harries-Delisle 1978: 430). It gets even more surprising when we know that also in most of the substrate languages (Ewe, Twi and Yoruba) the cleft marker is placed behind the fronted element; only in Igbo is its position sentence-initial:

[p. 95]

Yoruba

(11)	aso ni mo rà (Rowlands 1969: 25)
	cloth is I bought

‘I bought cloth’

Igbo

(12)	o wu na ya ka aga iwusa ya
	it is on him COMP he will-pour it

‘he'll poor it on him’

(Green & Igwe 1963: 46)

Boretzky (1983: 228) thinks that this marked position results from ‘general syntactic rules’, but superstrate influence may have been at work as well, in my view. The fact that da/na emerged in Sranan as a sentence-initial, introductory copula may have been an additional factor. But the most important reason probably is, that in a strict SVO language like Sranan sequences of a clefted NP followed by a copula would be interpreted as a subject followed by its predicate.

There is one example where the cleft marker, although it does precede the focussed element, does not appear at the beginning of the sentence:

(13)	fa mi sa kis' a so pikinso fo na pis'
	how I MOD get is so little-bit of the urine

fo keskesi? (Herskovits 1936: 184)

of Monkey

‘how shall I get only that little of Monkey's urine?’

It appears to me that the fossilization of na so into an unanalyzed, single morpheme entity (which will be discussed more fully below), which is not interpreted as a cleft construction consisting of a copula plus a focussed element, allows it to be placed in sentence-medial position. On the other hand, sentence-medial clefting is not an unknown option from a typological-universal point of view (Harries-Delisle 1978: 432).

The complementizer. Apart from a few doubtful cases, which will be discussed shortly, cleft sentences in Sranan do not contain a complementizer (which may be a conjunction, a relativizer or a particle) between the focus and the nonfocus. It is also absent in most of the relevant substrate languages (Ewe, Twi, Yoruba), but it does appear in Igbo (compare sentences (10) and (12) above). It is also present in English, although optionally when a direct object is involved; in colloquial speech the relative pronoun can even be omitted with subjects:

(14)	it is John (whom) I saw

(15)	it is John (who) is responsible here

As in Sranan, the complementizer is absent in most, but not in all creoles. However, when it is present, the cleft marker is usually absent, e.g. in Principense and Guinea-Buissau Creole Portuguese:

Principense

(16)	éli ki sa kási aré (Boretzky 1983: 222)
	that REL house king

‘that is the king's house’

[p. 96]

Guinea-Buissau Creole Portuguese

(17)	báka ki si pápe (Holm 1988: 181)
	cow REL is superior

‘the cow is superior’

According to Boretzky (1983: 223) the same type of structure is also possible in Haitian, although the example he gives does contain a cleft marker:

Haitian

(18)	se la pli ki tobe pada lanwit
	it's the rain REL fall during the-night

‘it was raining during the night’

(Boretzky 1983: 223)

Perhaps (18) is a mixture of a cleft sentence with only a cleft marker, which is the normal Haitian construction, and one where only a relativizer is present; alternatively, it could also be explained as a product of decreolization.

The structure exemplified in (16) and (17) suggests that in these languages the relative pronoun performs the function of a cleft marker, placed behind the fronted element, which would be in accordance with West African practice. To this might be added that, according to Balmer & Grant (1929) (quoted in Harries-Delisle 1978: 433), the Twi cleft marker na is a merger of the copula ne and the relativizer a. Besides this a post-focus position for cleft markers is also in agreement with universal tendencies.

The absence of a complementizer in the form of a relative pronoun in cleft sentences in creoles generally could be related to the fact that relative clauses in creoles also frequently do not contain relative pronouns³, especially in the early stages (Romaine 1988: 242). This is important in view of the hypothesized underlying relationship between clefting and relativization (Schachter 1973). Thus, for instance, in early Sranan sometimes sentences like the following can be found:

(19)	wan libisomma membre, takki, hem kan helpi hem
	a person think say he can help him

srefi, a kori hem srefi (Schumann 1783: 130)

self he deceive him self

‘somebody who thinks that he can help himself, is deceiving himself’

According to Boretzky (1983: 209) sentences like these still occur in the modern language.

Now we shall examine the few cases where there is something present which at first sight might look like a complementizer, but which in most cases will prove to be something else. First, there are some cases like the following, where disi is probably not a relative pronoun but

3 This opinion is, as far as Sranan is concerned, shared by Boretzky (1983: 211), who states that, since the relative pronoun di(si) is relatively young (derived from Du. ‘die’) and no older relative is found, one can assume that in the older stages relativization was without relativizer.

[p. 97]

a demonstrative, postposed to the pronoun and serving to emphasize it, a procedure quite common in Sranan:

(20)	a no mi disi ben gi disi libi baka, ma da wi
	is not I DEM TNS give this life back, but is our

boen-ati lobi Helpiman ben gi en libi baka

good-heart beloved Saviour TNS give his life back

‘I didn't bring him back to life, but our good, beloved Saviour Jesus did’ (Kraag 1894-96: 35)

The fact that the other cleft construction in this sentence does not contain a relative pronoun either strengthens the analysis chosen above.

Then there are two sentences in a dialogue, which are presented by the author (Weygandt 1798) as alternative and semantically identical answers to the question ‘Who's knocking at the door?’

(21)	da soema diesie dee kon loekoe joe
	is somebody REL ASP come see you

‘it's somebody who's coming to see you’ (Weygandt 1798: 124)

(22)	da viesietie diesie joe dee kiesie (ibid.)
	is visitor REL you ASP receive

‘it's a visitor whom you'll be receiving’

Although the original Dutch gloss (‘Het is gezelschap dat gij krijgt’, lit. ‘It is company that you get’) might suggest a cleft interpretation, a non-cleft reading is equally well possible. In the latter case da has the function of introductory copula, which is a normal option in early Sranan, especially in response to WH-questions. (Compare e.g. sentence (6a) in Ch. 2, which is also a response to a WH-question.) The remainder of the sentence would then have to be viewed as a normal relative clause introduced by a relative pronoun, which is indeed the interpretation given in the glosses above. The non-cleft reading of (21) is supported by the fact that pragmatically it is not a good candidate for clefting in this conversational context, i.e. it would be odd to emphasize ‘somebody’ in ‘Somebody is coming to see you’ as an answer to ‘Who's knocking at the door?’.

There are two other cases of what look like cleft sentences containing a complementizer (in the form of a conjunction this time), which at closer inspection prove to be cases of right dislocation. The complementizer-like element is datti, which is infrequently used as a conjunction instead of taki (Grammatik 1854: 65). However, in the source from which both sentences are taken (Focke 1855) the entry under datti only gives its pronominal function (demonstrative ‘that’). Since both sentences have exactly the same structure only one of them is given below:

(23)	da wan sani datti, a kom méki gi mi
	it-is a thing that he come make give me

‘he kicked up a hell of row’

(Focke 1855: 113)

The position of the comma, suggesting a pause after datti, provides further evidence that it is not a conjunction here. Rather it should be seen as a demonstrative pronoun in right-

[p. 98]

dislocated position, derived from datti wan sani. The part after the comma is a relative clause without relative pronoun with sani as its antecedent. This non-cleft analysis is strengthened by the fact that right dislocations of this type are common in Sranan, e.g.:

(24)	da wan areen datti (Schumann 1783: 77)
	it-is a rain that

‘that was a terrible shower!’

A final argument is provided by the fact that if (23) were a cleft sentence, its subject datti would most probably have to be followed by a dummy predicate (de)⁴, since subjects have to be followed by their verbs in Sranan.

There is one type of clefting where a complementizer, although disguised as a verb, does indeed appear optionally, and that is where the clefted constituent designates a reason or cause. Whereas in the first of the following examples meki still has to be construed as a verb, with the clefted element as its subject-NP, in the other two it is clearly a conjunction:

(25)	da joe póesoe mi, méki mi fádon (Focke 1855: 105)
	is you push me make me fall-down

‘I fell because you pushed me’

(26)	a no foe hángri-tem héde, méki mi sa káli
	is not for hunger-time reason COMP I MOD call

Tája ‘Tata’ (Focke 1855: 130)

‘taja’ (a vegetable) ‘Father’

‘I won't call “taja” “Father” just because I'm hungry’

(27)	na di a kan frei hei, mek' a tak' so
	is because he can fly high COMP he say so

‘he suggested this because he can fly high’

(Herskovits 1936: 192)

That meki is not a verb in the latter two sentences is indicated by the fact that it does not have an NP as its subject. Its analysis as a conjunction is supported by the fact that in (26) it is the verb kali and not meki which takes the TMA marker sa, and by the fact that in (27) the subject of tak' has the nominative form a instead of the

4 This is supported by the fact that if predicate noun clefting is possible in modern Sranan (which is not entirely clear), the subject of the non-focus will certainly have to be followed by a dummy verb de. This appears from the fact that as an equivalent of ‘my brother is a soldier’ only (a) (although marginally), but not (b) is possible (Hein Eersel, p.c.):

(a)	? na srudati mi brada de
(b)	* na srudati mi brada

This constraint seems to hold for other creoles as well: compare e.g. Haitian se ti mun li ye, lit. ‘is little person he is’, i.e. ‘he's a child’ (Boretzky 1983: 223).

[p. 99]

oblique form en. Incidentally, the interpretation of meki as a conjunction in cases like these is also fully supported by contemporary linguistic documents, such as Focke (1855: 80), Grammatik (1854: 65) and Wullschlägel (1856: 55), who all demonstrate its use as a conjunction besides that as a verb. Category switches like the one reported here are not at all unusual in creole languages (cf. e.g. Voorhoeve 1980 for Sranan). Compare the following example from Krio:

Krio

(28)	na we I lek yu mek a tok to yu
	is because I like you COMP I talk to you

‘I talk to you because I like you’

(Boretzky 1983:221)

Although meki is still in use as a conjunction in modern Sranan (Donicie 1959: 106), it does not seem to be used in cleft constructions like the ones above anymore (Hein Eersel, p.c.). Rather, it is used as a verb, as in (25) above, suggesting a category switch back to its original status. Thus, the modern equivalent of (26) is the following:

(29)	a no hangri-tem sa meki mi kali Taja ‘Tata’ is not hunger-time MOD make me call taja ‘Father’

‘I won't call “taja” “Father” just because I'm hungry’ (Hein Eersel, p.c.)

Finally, it should be noted that meki in sentences like (26) and (27) is not an obligatory element in the sources I examined:

(30)	a no fu dati ede yu mu ferakti yu
	is not for that reason you should despise your

eygi tongo (Koenders 1946-49: 140)

own language

‘that's no reason to despise your own language’

(31)	a no di Tjotjò-fowloe pikien, joe sa
	is not because Tjotjo-bird small you MOD

swári hem nánga wiwiri (Focke 1855: 138)

swallow it with feathers

‘the fact that the Tjotjo-bird is small is no reason for you to eat it with its feathers’

The fact that a complementizer turns up only in sentences where either a PP or a full clause has been clefted, suggests that the ‘heaviness’ of the clefted constituent may have something to do with it.

Negation. Although negative clefting occurs with the same kinds of constituents as clefting without negation, in some cases it appears to have the special function of restricting the scope of negation to the one element in the clefted constituent immediately following it. This element may be a gradable adjective (pikin or boi, ‘small’), a quantifier (ala, ‘all’) or a numeral. Viewed that way, negative clefting is simply a way of moving the negative element right before its scope, a procedure which is otherwise impossible, due to its fixed preverbal position. In the case of adjectives, this procedure (traditionally called ‘litotes’) leads to an intensified positive reading as in:

[p. 100]

(32)	da no boi-areen fadomm tideh
	is not small-rain fall-down today

‘there was a terrible shower today’

(Schumann 1783: 65)

This is also possible with predicate clefting:

(33)	da no boi fadomm a fadomm (ibid.)
	is not small fall it fall

‘there was a terrible shower’

(34)	a no alla tafra habi drai foetoe
	is not all tables have turned legs

‘not all tables have turned legs’

(Helstone 1903: 11)

(35)	a no tien soema ben kom krin (Luke 1829: 106)
	is not ten people TNS come clean

‘weren't there ten people purified?’

Clefting, however, is not the only means available for restricting the scope of the negation:

(36)	ala soema, di lidom na bedi, a no slibi
	all people who lie LOC bed is not sleep

‘not everybody who is lying in bed is asleep’

(Wullschlägel 1856: 28)

Anyhow, it seems clear that the function of clefting in (32) - (35) is to disambiguate between the two readings that are possible in a non-cleft sentence. Thus, for instance (34), in a non-cleft structure has the following two readings:

(37)	alla tafra no habi drai foetoe
	all tables not have turned legs

‘not all tables have turned legs’ or: ‘no table has turned legs’

Finally, sometimes (a) no appears at the beginning of a sentence in what looks like a cleft construction, but is not, since the whole sentence is under the scope of the negation:

(38)	no wan deh habi tin na tu juru?
	not a day have ten and two hours

‘are there not twelve hours in a day?’

(Schumann 1783: 208)

The difference between sentences (32) - (35) on the one hand and (38) on the other is that in the latter the negation is placed in a higher sentence, as in English ‘It is not the case that...’, whereas in the former it is intrasentential. The result of bringing the whole sentence under the scope of the negation is to produce a rhetoric question.

4.2.2. Predicate clefting.

The first question that should be answered about predicate clefting is what kinds of predicates can be clefted. In my sample, which contains no more than 29 cases of predicate clefting, examples can be found of clefting of verbs, adjectives and auxiliaries. Of the latter two there is only

[p. 101]

one example for each category (if we accept Seuren's (1987) analysis of siki, ‘ill’, as a verb, not an adjective):

Adjective

(39)	da boi no stoutoe, ma da morsoe a de morsoe
	the boy not naughty but is dirty he is dirty

toemoesi (Grammatik 1854: 47)

too-much

‘the boy isn't naughty, but he is a mucky pup’

Auxiliary

(40)	da moese joe sa moese (ibid.)
	is must you TNS must

‘you'll have to’

According to Jansen et al. (1978: 139) auxiliary clefting is not possible for all speakers nor for all auxiliaries in modern Sranan. It is possible for moesoe, also when it is followed by a main verb in the matrix sentence:

(41)	na moesoe mi moesoe waka go na wowojo
	is must I must walk go LOC market

‘I really must go to the market’

(Jansen et al. 1978: 139)

This is a viable option in other creoles as well:

Haitian

(42)	se vle yo pa vle luvri
	is want they not want open

‘they don't want to open up’

(Boretzky 1983: 223)

As far as clefting of verbs is concerned, between 80 and 90% of these are intransitives: in addition to the two transitives quoted below, there are three pseudo-transitives, of which (45) is an example:

Transitive verb

(43)	da tési joe de tési mi mófo (Focke 1855: 135)
	is taste you ASP taste my mouth

‘you're really trying my secrecy’

(44)	da koksi ju koksi mi (Schumann 1783: 154)
	is coax you coax me

‘you're really taunting me’

Pseudo-transitive verb

(45)	mi aksi Masra, efi da lesi wi moe lesi
	I ask Master if is read we must read

‘I'm asking you whether we should read (or do something else)’ (Grammatik 1854: 46)

(The other two cleft pseudo-transitives in the sample are jeri, ‘hear’, and furfur, ‘steal’.)

The majority of clefted predicates, however, is intransitive, as in the following examples:

Intransitive verb

(46)	da komm mi de komm (Grammatik 1854: 6)
	is come I ASP come

‘I'm coming right away’

[p. 102]

(47)	no piki siki ju ben siki (Schumann 1783: 246)
	not little ill you TNS ill

‘you were very ill indeed’

The question as to why it is mainly intransitive verbs that are prone to clefting remains largely obscure to me, although there are one or two things that may be said in order to clarify that issue somewhat. First, there appears to be a relationship with a construction type that I will label ‘predicate relativization’⁵ for the moment, which is also used with, - although by no means restricted to - , intransitive verbs. It is illustrated by the following examples. (See Grammatik (1854: 47) for further illustration):

Predicate relativization

(47)	da siki, dissi mi siki, da no Gadosiki
	the ill(ness) that I ill is not God-ill(ness)

‘the illness from which I suffer is not a natural one’ (Schumann 1783: 246)

(48)	di ben hangri vo si da verloesoe, disi da
	who TNS hungry for see the redeem that the

Messias sa kom verloesoe Israel (Luke 1829: 18)

Messiah TNS come redeem Israel

‘who were eager to see how the Messiah would redeem Israel’

Unfortunately, none of the cases of predicate relativization in Sranan that I have encountered receives an intensive interpretation, which would obviously strengthen the link with predicate clefting, but it does sometimes in another creole, related to Sranan, i.e. Krio:

Krio

(49)	yu bin fo si di it we i ben de it
	you TNS for see the eat that he TNS ASP eat

‘you should have seen how much (or: the way) he was eating’ (Williams 1976: 122)

In Sranan, however, the main function of predicate relativization appears to be the encoding of what in many other languages is expressed by derived nominals:

(50)	da poeloe, di dem poeloe mi (Grammatik 1854: 8)
	the pull that they pull me

‘my exclusion’

In Yoruba the intensified meaning is caused by the clefting of the entire relativized predicate, including the relative clause:

5 This term was suggested to me by Pieter Seuren (p.c.). Williams (1976: 122) uses the term ‘relativized verbs’, while Sebba (1987: 66) alternatively speaks about ‘topic relativization’ and ‘relative topicalization’ to refer to the same phenomenon.

[p. 103]

Yoruba

(51)	jijáde t'ó jáde l'ó rí mi
	come-out that-he come-out it-is-he saw me

‘as soon as I came out he saw me’

(Rowlands 1969: 189)

Although I have not encountered this kind of ‘relativized predicate clefting’ in Sranan, it is known to occur in other creoles, e.g. Haitian:

Haitian

(52)	we m te we-l, m kopra (Boretzky 1983: 225)
	see I TNS see-him I understand

‘as soon as I saw him, I understood’

The other factor which might cast some light on the problem of why especially intransitive verbs are being clefted is related to the phenomenon of predicate relativization, i.e. the existence of what are called ‘inner’ or ‘cognate’ objects in languages that can relativize predicates. Thus, for instance in Sranan we have:

(53)	a fadomm wan ogri fadomm (Luke 1829: 42)
	it fall an ugly fall

‘it fell completely apart’

An almost exact parallel can be found in Igbo:

Igbo

(54)	o dhàrà àdhà (Green & Igwe 1963: 31)
	it fall fall

‘it fell completely apart’

The hypothesized relationship between predicate clefting, predicate relativization and the phenomenon of cognate objects is nicely illustrated by the following examples.

Predicate clefting

(55)	da no boi fadomm a fadomm (Schumann 1783: 65)
	is not small fall it fall

‘there was a terrible shower’

Predicate relativization

(56)	da fadom, di a fadom (Grammatik 1854: 7)
	the fall that he fall

‘the fall that he had’

Cognate object

(57)	a fadomm wan ogri fadomm (Luke 1829: 42)
	it fall an ugly fall

‘it fell completely apart’

That the latter two are related seems pretty clear, but this is not so obvious for the former of the three. Deriving a sentence like (55) from a sentence like (57) means that the clefted predicate is interpreted as the inner object of the identical verb in the matrix sentence. One of the problems in such a derivation would be the disappearance of the article wan, ‘a’. In favour of the object-like status of the clefted predicate, however, speaks the fact that this predicate has been nominalized. That this is so appears from the fact that it cannot take TMA-markers (Jansen et al. 1978: 138), that it can be modified (as in (55) above) and that it takes the equative copula da/na, which is the copula that is preferred with nominal predicates (as was established in Ch. 2; see

[p. 104]

also Bynoe-Andriolo & Yillah 1975: 237).

A final feature of predicate clefting that has to be noted is the fact that its function is much more often intensifying (in a wide sense), than (contrastive-)emphatic, a fact which is also noted by Bynoe-Andriolo & Yillah (1975: 234), who say that these structures

‘(...)underline the fact that the subject is actually performing the action mentioned, or is performing it to a marked degree. They are frequently used as exclamations of wonder or astonishment. They are not primarily intended to focus on the action of the verb or to differentiate it from that of some other possible verb (...)’
(Bynoe-Andriolo & Yillah 1975: 234-235)

This is not to say that the contrastive-emphatic function is completely absent in Sranan, as appears from (58) below, but the intensifying function is much more frequent. (Compare sentences (59) and (60) below)

Contrastive-emphatic

(58)	Masra, a no de kre, ma da lafoe a de lafoe
	Master, he not ASP cry, but is laugh he ASP laugh

‘he's not crying, Sir, he's laughing’

(Grammatik 1854: 6)

Intensifying

(59)	da bassia a de bassia (Schumann 1783: 57)
	is bend he ASP bend

‘he's bending down real low’

(Schumann's translation: ‘er bückt sich recht tief’)

(60)	da furfur a furfur (Schumann 1783: 104)
	is steal he steal

‘he's a real thief’

(Schumann's translation: ‘er ist ein rechter Dieb’)

The intensifying function, besides the contrastive one, is also found in the substrate languages that do have predicate clefting⁶:

Twi

(61)	oyáw nko ara nà wote hó yàw
	quarrel EMPH.PART EMPH. PART CONJ they quarrel

‘they live in continual quarreling’

(Christaller 1875: 147)

6 Apparently no clefting occurs at all in Kikongo and Kimbundu; predicate clefting appears to be absent in Ewe and Igbo.

[p. 105]

Yoruba

(62)	nínà ni ng ó nà á (Rowlands 1969: 158)
	beat it-is I will beat him

‘I'll give him a good beating’

Finally, the intensifying function of predicate clefting may lend further support to the idea, expressed earlier, that there is a relationship between this construction, predicate relativization and cognate object constructions, since the latter also, at least sometimes, have an intensifying meaning.

4.2.3. Non-predicate clefting

The discussion of non-predicate clefting will be divided into separate subsections according to the nature of the clefted element. The following elements appear in clefted position in the sample: WH-elements, NP's, PP's, Adverbial Phrases and clauses. Each of these will be dealt with in turn.

WH-elements. Although clefting of question words occurs in several creoles (Saramaccan, Jamaican, Papiamentu and Negerhollands; see Holm 1988: 180), there are only two examples in my sample. In both cases the question word is embedded in a larger constituent: in (63) oe, ‘which’, is part of the NP oe Masra, in (64) oe soema, ‘who’, is part of the PP foe oe soema. Strictly speaking this would put them in the category of NP and PP clefting:

(63)	da oe Masra datie? (Weygandt 1798: 91)
	is what master that

‘what kind of gentleman is it?’

(64)	da foe oe soema pletie diesja
	is of what person plate this-here

‘whose plate is this?’

(Weygandt 1798: 94)

The fact that WH-clefting does not occur in modern Sranan and the fact that in other creoles the clefted WH-element never seems to be used attributively may make an alternative analysis more plausible, i.e. the interpretation of (63) and (64) as cases of right dislocation. This is supported by the fact that e.g. (63) has a structure which is parallel to that of (24) above, for which also a right dislocation analysis was suggested.

In the Kwa languages, where WH-clefting is well known, it also seems to be restricted to question words that have constituent status:

Yoruba

(65)	tani ó rà á? (Rowlands 1969: 26)
	who-is he bought it

‘who bought it?’

Twi

(66)	daben na yehun wo? (Balmer & Grant 1929: 69)
	when is we-saw you

‘when did we see you?’

This also seems to hold for the creoles mentioned above, e.g.

[p. 106]

Jamaican:

Jamaican

(67)	a wa Anti sen fi mi? (Holm 1988: 180)
	is what Aunti send for me

‘what has Auntie sent for me?’

Unfortunately, I have no explanation for the fact why Sranan, unlike some other English- and Dutch-based creoles, does not have clefting of question words. Decreolization offers no explanation because of the absence of WH-clefting in early Sranan and because some other creoles that do have this type of clefting, such as Jamaican, have been decreolized to a much greater extent.

Noun Phrases. NP's can be clefted when performing one of the following grammatical functions: subjects, direct objects, indirect objects and predicate nominals, although evidence on the latter two is open to other interpretations as well. The only case⁷ where the clefted NP may be interpreted as an indirect object is the following:

(68)	no hem mi aksi (Schumann 1783: 208)
	not him I ask

‘I didn't ask him’

Note that the absence of the copula in negative clefts is a regular feature of this source. The original German translation of (68) (‘ich habe nicht ihn gefragt’) allows for both a direct object and an indirect object interpretation of hem. The latter is supported by the fact that clefting of indirect objects is a normal option in modern Sranan (Hein Eersel, p.c.):

(69)	na Kofi mi gi wan buku (Jansen et al. 1978: 137)
	is Kofi I give a book

‘I gave Kofi a book’

Subject NP's are clefted more often than direct objects, and direct objects more often than indirect objects, with Prepositional Phrases and predicate nominals in between the latter two. This would seem to fit in nicely with the Noun Phrase Accessibility Hierarchy that Keenan and Comrie (1977) found for relativization. An important difference, of course, is that this hierarchy reflects an implicational universal, based on a cross-linguistic investigation of the possibility of relativization, whereas the Sranan data only reflects a frequency distribution, which, by the way, might be skewed by the overall frequency of different NP types. Although it would be an interesting question to examine whether a similar hierarchy exists for clefting, this could only be answered in a separate study, so I will not go any further into the

7 There is one more possible case of indirect object clefting in the sample:

(c)	da Gado joe moe begi (Grammatik 1854: 6)
	is God you must beg

‘you should ask God’
Of course, in this sentence a direct object interpretation is just as well possible as in (68).

[p. 107]

matter here.

I will now give some examples to illustrate clefting of the different NP types mentioned above:

Subject NP

(70)	da mie papa kiesie em na wan foe en kompee
	is my daddy get it LOC one of his friends

na fransie kondree (Weygandt 1798: 123)

LOC French country

‘my daddy got it from one of his friends in France’

Object NP

(71)	we, na den soema dati den kari leti obiaman
	well, is the people DEM they call true obiaman

‘well, those people are called true “obia” men (i.e. witches)’ (King 1864-70: 18)

Predicate Noun

(72)	da wan dungru datti (Schumann 1783: 77)
	is a darkness that

‘it is pitch-dark’

As to the last example, it should be remarked that, despite the intensifying meaning in this and other examples of the same type, a non-cleft interpretation may be more plausible, as has been argued on p.98 above. These arguments hold for all similar cases, since they all, basically, have the same structure, i.e. COPULA - PREDICATE NOUN - datti - (COMPLEMENT).

All NP types that can be clefted may have the form of a pronoun, but, except for subject pronouns, their occurrence is not very frequent. An example of a clefted object pronoun:

(73)	da disi joe wanni? (Focke 1855: 149)
	is this you want

‘do you want this one?’

The frequency distribution of pronoun clefting reflects that of NP clefting in general, and may similarly be related to the overall frequency of different pronoun functions⁸.

Prepositional Phrases. All of the eleven cases of PP clefting in the sample involve the preposition fu, ‘for’. Eight of these are part of the complex preposition fu...ede, ‘because of’, as in:

8 One other type of pronoun clefting that should be mentioned is the idiomatic, and now obsolete, expression da hem, ‘therefore’ as in:

(d)

joe hiti da sani na gron: da hem mi figi hem (Grammatik 1854: 61)
you hit the thing LOC ground: is it I sweep it ‘you threw it on the ground, that's why I swept it away’

Since, according to Focke (1855: 45), da hem is short for da hem hede, this can be analyzed as a cleft, lit. ‘is this (the) reason’, i.e. ‘this is the reason’.

[p. 108]

(74)	dafoe datie hédé joe no kan werie den soesoe
	is-for that reason you not can wear these shoes

‘that's why these shoes don't fit’

(Weygandt 1798: 115)

As was explained earlier, sentences like (74) often contain a complementizer meki between the focus and the non-focus part of the sentence.

Of the three non-causal PP's in clefted position one is final (75), one is possessive (or ablative) (76), and one is benefactive (77):

(75)	a no vo dati mi kom, ma mi wani haksi Masra
	is not for that I come but I want ask Master

wansani (Grammatik 1854: 61)

something

‘I didn't come for that, but I want to ask you something’

(76)	da doopoe vo Johannes, da vo hemel a ben
	the baptism of John is of/from heaven it TNS

de, ofoe vo liebisoema? (Luke 1829: 118)

be or of/from man

‘did John's baptism belong to heaven or to man?’

or: ‘was John's baptism from heaven or from man?’

(77)	we, da vo oenoe fosi Gado ben wiekie hem
	well, is for you first God TNS awake his

Pikien Jezus (Acts 1829: 162)

son Jesus

‘well, it's primarily for you that God has awaken his son Jesus’

There does not seem to be any a priori reason why fu would be the only preposition to appear in clefted PP's, although the high proportion of causal PP's might point to a pragmatic factor: perhaps the emphatic expression of reasons and causes is pragmatically so salient that it leads to this type of focussing. There might also be a bias in the sample, since in modern Sranan other prepositions besides fu can occur in clefted PP's as well:

(78)	na nanga wan gon Jan kiri a sneki
	is with a gun Jan kill the snake

‘Jan killed the snake with a gun’

(Jansen et al. 1978: 138)

Adverbial Phrases. The bulk of clefted Adverbial Phrases (80%) indicate manner, and in 90% of these the fronted constituent is the single adverb so, ‘thus’. Of the rest, the majority encodes either cause, which was treated above under PP clefting, place or time. Each of these will be illustrated below:

Manner

(79)	da so da wortoe kon leti dja toe
	is thus the word come right here too

‘thus the word came right here too’

(King 1894-96: 35)

The very high proportion of so in clefted manner

[p. 109]

adverbials calls for an explanation. There are several indications that da/na so is not so much a cleft construction, but rather a fossilization whose original cleft structure is no longer perceived. First, there is the fact that historically speaking so is clearly the first element to occur in clefted position: in the oldest source that has any clefting at all (Van Dyk c. 1770) fourteen out of seventeen cleft constructions involve so. Second, in some sources (e.g. Albitrouw 1894) so is used with da, whereas other clefted constituents take na, the form which is used for equation in these sources. Third, na so sometimes occurs in sentencemedial position, which is otherwise impossible in Sranan (compare sentence (13) above). And, finally, there are several sources that state explicitly that da/na so is a fixed expression meaning ‘this way’, ‘like this’ (Helstone 1903: 56; Wullschlägel 1856: 210)⁹.

An example of a clefted manner adverbial not involving so is the following:

Manner

(80)	ma na kibri fasi den de doe dati
	but is secret fashion they ASP do that

‘but they do it secretly’ (King 1864-70: 17)

(Note that na is interpreted as a copula here, not as a preposition: various sources indicate that the expression is either kibri fasi, ‘secretly’ (Schumann 1783: 151; Grammatik 1854: 51) or na wan kibri fasi. id. (Wullschlägel 1856: 120-21).)

Finally, some examples are given of clefted adverbials indicating time, place and frequency:

Time

(81)	a no tra-de'-tra-de' nómo mi pai joe
	is not other-day-other-day only I pay you

‘didn't I pay you just the other day?’

(Focke 1855: 22)

Place

(82)	da daplesie mie ben sie em (Weygandt 1798: 97)
	is that-place I TNS see him

‘I saw him there’

Frequency

(83)	da toe tron kabà mi báli joe (Focke 1855: 143)
	is two times already I call you

‘I already called you twice’

As far as place adverbials are concerned, a problem is constituted by cases where na dape occurs at the beginning of the sentence, since in principle na could be a preposition here:

9 Counterevidence against this hypothesis is provided by the fact that na so has been calqued in Suriname Dutch as is so, which speaks in favour of its analytic structure (Pieter Seuren, p.c.).

[p. 110]

(84)	na dape joe sa si taki heiden abi
is/LOC there you TNS see say heathens have

fansortoe sani foe... (King 1864-70: 18)

all-kinds things to...

‘there you'll see that heathens have all kinds of things to...’

In several sources (Schumann 1783: 218; Focke 1855: 99; Wullschlägel 1856: 58) na dape is equated with dape, in which case na probably has to be analyzed as a preposition, which is not implausible in view of the historical derivation of dape from da plesie, ‘that place’ (see section 2.3.1.). The same sources are contradictory, however, in that some (e.g. Wullschlägel) have na dape, whereas others (e.g. Focke) have da dape, while both have da as a copula. The most probable solution is, I think, that na dape may be a cleft construction, but if it is, it is probably a fossilized one like na so. Its fossilization may be responsible for its equation with dape, referred to above.

Clauses. Of the five cases where an entire clause seems to be clefted three involve a typical construction occurring only in odo's (proverbs). The other two have been given above as (25) and (27) in the discussion of the complementizer-status of meki. They are repeated here for convenience as (85) and (86):

(85)	na di a ka' frei hei, mek' a tak' so
	is because he can fly high COMP he say so

‘he suggested this because he can fly high’

(Herskovits 1936: 192)

(86)	da joe póesoe mi, méki mi fádon (Focke 1855: 105)
	is you push me make me fall-down

‘I fell because you pushed me’

As far as other creoles are concerned, the only example of clause clefting known to me is the Krio sentence comparable to (85), which is cited as (28) above. Some West African languages (Ewe, Yoruba) appear to be able to ‘cleft’ whole sentences, although the term becomes rather vacuous here:

Yoruba

(87)	mo n-foso ni (Boretzky 1983: 225)
	I am-washing it-is

‘I'm washing clothes’

The three cases of clefting in odo's are the following:

(88)	a no so léki arén brákka, a no so a de
	is not so like rain black, is not so it ASP

fadón (Focke 1855: 5)

fall-down

‘the rain won't fall down as hard as it seems’

(89)	a no so léki Páttoe biggi, a no so njanjám
	is not so like pot big, is not so food

sa fóeloe hem (Focke 1855: 98)

MOD fill it

‘a pot does not necessarily contain as much as it can’

[p. 111]

(90)	a no tid'ía wiwíri fadón na wátra, a no
	is not today leaf fall LOC water is not

tid'ía a póri^10 (Focke 1855: 135)

today it spoil

‘a leaf doesn't rot on the same day it falls into the water’

What is remarkable about (88) - (90) is that the clefting occurs both in the first and in the second clause, the latter of which seems to have the function of resuming the whole clefted first clause. This ‘resumptive cleft’ may be necessitated by the heaviness of the clefted element, i.e. the first clause. It is significant that in all cases of clefting of whole clauses there is an element (either meki or a resumptive cleft) intervening between the focus and the non-focus part of the sentence. This suggests that the resumptive cleft has a function comparable to that of a complementizer.

Finally, two cases have to be mentioned where a whole clause is preceded by na, but still has remained in its original position. The function seems to be the expression of wonder or confirmation, which is not unusual for clefting:

(91)	no, mi no ben taigi taki na Dia morsu?
	now I not TNS tell say is Deer dirty

‘now, didn't I tell you that Deer is dirty?’

(Herskovits 1936: 162)

(92)	di a waka so na yu ston süt' so
	since it walk so is your testicles sweet so

‘since it happens that your testicles are so sweet’

(Herskovits 1936: 182)

This sentence is reminiscent of the Yoruba sentence quoted as (87) above, where a whole sentence is ‘clefted’. Indeed, the cleft marker nl in Yoruba is also used at the end of subordinate clauses placed before the main clause (Rowlands 1969: 158). Viewed that way, na may also be a marker of subordination, especially in (92) where no other marker is present.

10 The first clause in this sentence is probably best analyzed as a conditional, expressed through parataxis, a procedure which is quite common in Sranan.

4.3. Historical development

In this section the historical development of clefting in Sranan will be briefly summarized. As was already noted in the introduction to this chapter, the period under investigation shows no dramatic developments since the system of predicate and non-predicate clefting was by and large established around the middle of the 18th century. However, what changes have been observed during the investigation will now be recapitulated according to whether a specific feature

[p. 112]

has changed its form, has disappeared or has been introduced into the language during this period.

First, there is a formal change in the cleft marker from da to na, which is in accordance with the overall development of the equative copula. Second, the following features or elements have disappeared from the language: the θ copula in negative clefts, which is common in Schumann's (1783) dictionary; the use of meki as a complementizer (although it is still used as a conjunction in other sentence types; see Donicie 1959: 106); the use of disi as a complementizer, although its analysis as such is doubtful to begin with; the clefting of WH-elements, although embedded in NP's (as in (63) and (64) above); and, finally, the idiomatic da hem, ‘therefore’ construction. Third, the following features have been introduced during or after the period under study: clefting of indirect objects has become a normal option; clefting of PP's is not restricted to PP's containing fu as a preposition; and, finally, (although examples are very scarce) clefting of clauses seems to have become possible only by the middle of the 18th century.

4.4. Conclusion

Clefting (but not pseudo-clefting) is a universally rare syntactic operation (Harries-Delisle 1978: 436). This excludes a universalist (e.g. Bickertonian) explanation for its occurrence in Sranan and other creoles. Since it does, however, occur in Kwa languages as well as in English, a subor superstrate origin seems inescapable; most probably the two have reinforced each other. As to the particular structure that was chosen in Sranan to encode clefting (i.e. COPULA - FOCUS - NON-FOCUS), this is completely in agreement with universal tendencies for SVO languages (Harries-Delisle 1978: 446). It also agrees with the English structure, except for the absence of a complementizer, which is optional in English depending on the grammatical function of the clefted constituent, and the absence of a dummy subject ‘it’.

An explanation for the latter is that dummy subjects are only necessary in languages where a sentence-initial copula may signal a question (o.c.: 446), which is not the case in Sranan. As to the former, no explanation is available from typological-universal research: the only information given by Harries-Delisle is that the nature of the relative clause marker (i.e. an invariant marker or a pronoun) does not influence its deletability (o.c.: 443). An explanation may be found in the fact that relative clauses in Sranan are sometimes unmarked. Another factor may be the fact that no complementizer is used in cleft constructions in most Kwa languages, except Igbo, which is also exceptional in that it has a pre-focus copula while the others have it behind the clefted element.

The ordering which is found in Ewe, Twi and Yoruba (i.e. FOCUS - COPULA - NON-FOCUS) is also compatible with universal

[p. 113]

tendencies, although in Harries-Delisle's derivation from an underlying pseudo-cleft (o.c.: 449) this requires an extra operation of the extraction-rule: first it operates on the copula and then on the Non-Focus. The latter fact, i.e. the dual as opposed to the single operation of a rule, may tentatively be held responsible for the structure that was chosen in Sranan, since it is less complex derivationally. Summarizing, it seems that the formal marking of clefting in Sranan is, at least partly, universally determined, whereas its occurrence per se is largely a substrate phenomenon. It seems worthwhile noting that a similar conclusion was reached above with respect to the structure of the copula system.

As to the diachronic side of the matter, it seems to me that the relative absence of change points towards an early stabilization of cleft constructions in Sranan. In my opinion two related points emerge from this finding: first, creolization is not a uniform process which affects all areas of syntax at the same rate. In other words, some syntactic constructions stabilize more quickly than others. The other point concerns the question why this should be so, i.e. what is it that makes some constructions ‘settle’ almost immediately, whereas in others change and variation keep going on, sometimes even to the present day. It will be clear that a lot more research has to be done on other areas of syntax and on other creoles, before the hypothesis of ‘differential creolization’ can be confirmed. But if so, it will have some important consequences for the concept of creolization.

As far as an explanation of the early stabilization of clefting is concerned, this may have something to do with the fact that, contrary to the comparative and copula systems, there is relatively little conflict between the source and target languages in this particular construction. This is immediately obvious for predicate clefting, but it also goes for non-predicate clefting, at least to some extent. Thus, they agree in the fact that the equative copula is chosen as a cleft marker, that they may omit a complementizer and that the focus is put at the beginning of the sentence, and not at the end, which is just as well possible, universally speaking. In terms of second language acquisition: there is little interference between source and target language.

This line of thought, however, can only be followed when one starts from the premise that second language speakers have played a major role in the formation of Sranan. In my view, it is they who are largely responsible for the continuous input of new impulses into the creolizing language, leading to continuous restructuring over long periods of time. This view is supported by what is known about the, rather exceptional, demographic situation in Suriname during the first century of its existence. This evidence, in addition to some linguistic material revealing the pidgin character of early 18th century Sranan, will be presented in the following chapter.