Syntactic Developments in Sranan

Creolization as a Gradual Process

[p. 25]

Chapter two: The copula system

2.1. Introduction

In this chapter I present a general picture of the historical development of the copula¹ system of Sranan, but I will also, on the basis of the historical evidence, try to clarify some problems of synchronic syntax in this area. I will devote special attention to one particular aspect, viz. equation, since it is in this category that the most dramatic changes can be observed. The implications of these changes for the concept of creolization and the explanatory power of universals and substrata for this case will be discussed. At the end of the chapter I will summarize the history of the Sranan copula in a tentative, reconstructed scenario, which should be viewed as a summary rather than a completely worked out hypothesis. Before all this I will begin my exposition with an attempt to try to confirm an explicit hypothesis about the origin of one of the copula functions, i.e. the development of the equative copula from an anaphoric pronoun.

The copula was chosen as one of the objects of this study, among other reasons, because it appears to be connected with a central aspect of Sranan syntax, the tense/aspect system (de is one of the copula forms as well as preverbal aspect marker). As will be shown later, it is also related to the pronominal system (da (later na) is one of the copula forms as well as pronoun (it/that) and determiner (the/that)). Finally, the fact that Sranan, like all other Afro-American creoles (Alleyne 1980: 88), is not at all prone to copula deletion (as is often the case with pidgins and other simplified varieties of English (Ferguson 1971)), but, on the contrary, is rather fine grained in this respect provides an extra reason for an in-depth study.

1 The term ‘copula’ as used in this study covers the equative (or equational) as well as the existential, locative and possessive functions of BE, the use of BE in adjectival predicates and its use as a predicator of adverbial and infinitival complements. The term ‘equative’ covers both the attributive and the identifying copula. The copula in its function as a focussing element in cleft sentences will be discussed in Ch. 4.

2.2. Origin

In this section I will trace the pronominal origin of one segment of the copula system, the intimately connected forms da, na and 'a². First, the existence of a separate sentence

2 na and 'a are later developments of da; in what follows da will be used as a cover term for all three forms.

[p. 26]

introducing copula da will be posited, and then its pronominal origin and character will be discussed, especially with respect to the light this will shed on some problems of synchronic syntax.

In almost all copula functions where, in non-sentenceinitial position, de is the preferred form, we find instances of da (instead of a de) sentence-initially (cf. (3) and (4) in Ch. 1). This da functions in a way quite similar to French c'est... or English it's... , i.e. as a sentence introducing element, containing a semantically empty pronominal subject. Compare the following sample sentences, where the (a)-sentences have da, whereas the (b)-sentences have de:

possessive:

(1a)	da foe mi (Weygandt 1798: 94)
	it(-is) of/for me

‘it's mine’

(1b)	datti de vo dem (Schumann 1783: 81)
	that is of/for them

‘that belongs to them’

adverbial:

(2a)	da so (Stedman 1796: 362)
	it(-is) so

‘that's the way it is’

(2b)	so a de (Wullschlägel 1856: 8)
	so it is

‘quite so’

infinitival:

(3a)	da vo taki (Makzien 1902: 1)
	it(-is) for say

‘That means...’

(3b)	a de vo wiki froekoe vo go loekoe wrokko
	it is for wake early for go see work

‘you have to wake up early to supervise the work’

(Helmig van der Vegt 1844: 48)

adjective:

(4a)	na troe san Bijblie takki (Cesaari 1836: 280)
	it(-is) true what Bible say

‘what the Bible says is true’

(4b)	son de tranga kaba
	sun is strong already

‘it's very hot already’

(Helmig van der Vegt 1844: 37)

[p. 27]

equative (attr.):

(5a)	da wan pleh, dissi de noja tranga na modo
	it(-is) a game, which is now strong in fashion

‘it's a game that is very popular now’

(Helmig van der Vegt 1844: 38)

(5b)	a de foetoeboi gi massra (Focke 1855: 14)
	he is servant give master

‘he's a servant to his master’

equative (ident.):

(6a)	da mi, Filida (Van Dyk c.1770: 73)
	it(-is) me, Filida

‘it's me, Filida’

(6b)	joe de da Santawan vo Gado (Luke 1829: 30)
	you are the holy-one of God

‘you (i.e. Christ) are God's holy son’

As regards (4b) it should be noted that the preferred copula with adjectival predicates is ø, not de, but that does not diminish the value of (4a) as an example of sentence-initial da (in its later development na in this case). Further, it is clear that this use of da is not to be expected with existentials, since there the subject, by definition, cannot be semantically empty. As far as locatives are concerned, there is only one - doubtful - example (the main function of initial da here seems to be that of a clefting element):

locative:

(7a)	da deeja da da winklie foe Masra G.?
	it(-is) here is the shop of Mr. G.?

‘is this where Mr. G.'s shop is?’

(Weygandt 1798: 105)

(7b)	wan masra dee na doro (Weygandt 1798: 91)
	a gentleman is at door

‘there's a gentleman at the door’

Both the fact that da is virtually obligatory in this position and the fact that it is generalized over almost all de functions, indicate that it should be regarded as a separate and perhaps primary category³, from which its use as non-sentence-initial copula was derived. This view is corroborated by evidence pertaining to the deictic origin of da, which we will discuss now.

There is evidence that besides sentence-initial da, lacking semantic content, another form, the demonstrative pronoun datti ‘that’, was used with deictic reference, as in:

3 Holm (1980: 368-369) distinguishes a separate category of introducing copula in Miskito Coast Creole, Jamaican Creole and, for that matter, in Black English.

[p. 28]

(8)	datti da^4 somma, dissi ju de suku
	that the person that you ASP look-for

‘that's the person you're looking for’

(Schumann 1783: 78)

Although da's modern variant a usually only has definite and not deictic meaning, in older Sranan prenomial da was also used as a demonstrative pronoun (Voorhoeve 1953: 53; 73; see also Schumann 1783, s.v. da, which is glossed not only as a definite article, but also as a demonstrative pronoun (dieser, diese, dieses)). It was only later that da was restricted to the role of definite article and that demonstratives were expressed by ‘circum-nominal’ constructions like a man dati, ‘that man’ (lit. ‘the man that’). Also in Saramaccan da is not, as Schumann would have it, a definite article (which is di), but, in Schuchardt's words, a ‘substantivisches Demonstrativ’, equivalent to dat(t)i (Schuchardt 1914a: 58). A relationship between determiner and demonstrative in Sranan would not be very surprising anyhow, at least from a comparative point of view, since equivalents of that are the most common source of definite articles in the languages of the world (Givón 1979: 316). So both the fact that sentence-initial datti is used in a way similar to da and the fact that the determiner da used to have demonstrative meaning lead to the conclusion that both da's were derived from datti and are thus pronominal in origin.

If sentence-initial da is indeed pronominal, this, in combination with its separate status shown above, could very well strengthen the case for it being functionally, as well as historically, a primary category. If this is correct, then its use as an equative copula, placed between subject and predicate, can be derived historically from this primary function through topicalizations like (9), which abound in Schumann's dictionary, and thus seem to have been very frequent in early Sranan:

(9)	‘adjabre’, da Djutongo (Schumann 1783: 46)
	‘adjabre’, that(-is) Djutongo

‘(the word) “adjabre” is Djutongo⁵’

Here the carefully placed comma designates the pause that is so typical of this kind of construction. This development of the copula is completed when later this resumptive pronoun/copula is reanalyzed as a regular copula in non-topicalized sentences like:

(10)	da somma da wan boen somma (Grammatik 1854: 6)
	that person is a good person

‘that's a good person’

4 Since da somma is clearly definite, da cannot be a copula here.

5 Djutongo, according to Smith (1987: 125-127), is not to be equated with Saramaccan, but refers to the mixed Portuguese-English creole, which was its predecessor.

[p. 29]

The development of a copula from a resumptive pronoun is by no means a unique phenomenon in the world's languages. Li and Thompson (1976) report it for such diverse languages as Mandarin Chinese, Palestinian Arabic, Hebrew, Wappo (Yukan), and Zway (Semitic). Pronominal copulas have also been reported for the French-based creoles by Saint-Jacques-Fauquenoy (1972), for Kilba (a Chadic language) by Schuh (1983), and for Kikongo by Seidel & Struyf (1910: 38). An example from Guyanais:

Guyanais

(11)	mó fré sa maso
	my brother this bricklayer

‘my brother's a bricklayer’

(Saint-Jacques-Fauquenoy 1972: 228)

If this deictic, pronominal origin of da is correct⁶, a number of synchronic problems connected with its use as a copula disappear. For instance, the fact that da cannot be preceded by no (which we would expect it could, if it were a true verb), is explained by its pronominal status. If we take a sentence like (6a) above (which is repeated here as (12)), and analyze da as a pronoun, then this is followed by a zero copula:

(12)	da ø mi, Filida (Van Dyk c.1770: 73)
	it ø me, Filida

‘it's me, Filida’

Now, the rule in Sranan is for the negator to precede the verb and that is precisely what it does in a sentence like (13), where it is followed by the empty copula site. Consequently, we are faced with a da no instead of a no da sequence, and this is the rule for all negative copula sentences:

(13)	da no ø bon zanti? (Van Dyk c.1770: 66)
	it not ø good thing?

‘isn't it good?’

It is interesting to note that da no never occurs between subject and predicate, but in that position is always replaced by a a no sequence, as in (14):

(14)	takiman a no doeman (Focke 1855: 24)
	talkman is-not do-man

‘talking is not the same as acting’

Fairly early on this same negative copula is also found at the beginning of sentences, such as the following:

(15)	a no Gado? (Wennekers 1822: 33)
	it-is not God?

‘is it not God?’

In a few, rare cases the sentence-medial copula under negation is replaced by the locative/existential verb de(e), but this is probably due to heavy Europeanization of the

6 It should be noted here that the deictic character of na has already been observed by Voorhoeve (1953: 76-77) and Donicie (1959: 39), although these authors were not able to document this origin historically.

[p. 30]

source in which these sentences are found (Helstone 1903). More important, although da is later replaced by na, the sequence na no is (and always has been) ungrammatical, whether sentence-initially or medially, and always has to be a no instead. The fact that a no makes its appearance long before the change of da into na (which has a as a later variant), excludes the possibility that both a's are one and the same element⁷. It is unclear what the source of negative a is, and this agrees with the observation that a no is perceived as a single morpheme without any internal structure by native speakers of the language (Pieter Seuren, p.c.). In other words, Sranan is a language with a special form for the negated copula, and in this respect it is by no means unique among the languages of the world: this phenomenon also occurs, for instance, in a language like Amharic (Leon Stassen, p.c.).

Another aspect of copular da that can be explained by its pronominal character, is that it cannot be preceded by TMA particles: TENSE + da always becomes ben de; MOOD + da always becomes sa de. Also, it cannot occur in infinitival position:

tense:

(16)	da wendje nem ben de Maria (Luke 1829: 8)
	the girl name TNS be Mary

‘the girl's name was Mary’

mood:

(17)	a sa de troe sani (Weygandt 1798: 101)
	it MOOD be true thing

‘I suppose it's true’

infinitival:

(18)	a moe de wan boen presi wanten na bigin foe en
	it must be a good place at-once at start of it

‘it should be a good location right-away’

(Albitrouw c.1896: 17)

It is, of course, true that, especially in the later stages (after 1800), the functions performed by de in (17) and (18), viz. nominal predication, become increasingly expressed by de anyhow, irrespective of whether it is preceded by a TMA particle, but this does not diminish the value of these examples. It is rather the negative fact that da occurs never in these positions that counts here.

The same factor that was adduced above in connection with negation, is called upon here. Since da was originally a pronoun and as such was followed by a zero copula, it was this zero element which was to take TMA markers and auxiliaries such as moe. But with da being reanalyzed as a

7 This agrees with the observation (due to Renata de Bies) that in spoken language the two forms are reduced differently: e.g., mi na becomes mia, whereas mi a becomes ma.

[p. 31]

true copula, the result of this would be a situation where we would wind up with postverbal markers, instead of preverbal ones, which are the rule for all other Sranan verbs. The solution to this problem was to replace the equative copula da by the locative/existential verb de in these positions. This procedure eliminated all traces of a zero copula, so that all markers and auxiliaries in a quite natural way found their proper places, i.e. in preverbal position. The reason why it was precisely the locative/existential verb that was selected for this purpose is discussed below.

There are two cases in the sample where there appear to be postverbal markers, but in both cases an alternative analysis is preferred:

(19)	da ben toe horro, dissi mi go slibi
	it TNS/was two hours, that I go sleep

‘it was two o'clock when I went to bed’

(Helmig van der Vegt 1844: 42)

(20)	da ben wan mooi sabathem
	it TNS/was a beautiful evening

‘it was a beautiful evening’

(Weygandt 1798: 100)

Since in both of these cases da is in sentence-initial position, it is reasonable to assume that here it has not been reanalyzed as a true copula, but still functions as a pronoun (as in (12) and (13) above). The consequence of this is that ben in these cases has to be viewed as a preverbal marker preceding a zero copula. The assumption of a zero copula here is entirely warranted in view of the discussion above.

Another feature of the copula that might be explained by its deictic origin is the fact that the sequences a da (old) and a na (modern) ‘he/she/it is’ are always ungrammatical and replaced by hem da ‘he/she is’ or simply da ‘it is’ (in early Sranan), and by en na ‘he/she is’ or na ‘it is’ (in modern Sranan) respectively. Compare the following examples from 19th and 20th century Sranan:

(21)	bikasi tidei da Helpiman gebore gi oen na ini
	because today the Saviour born give you LOC in

‘because today the Saviour was born for you in the

da foto vo David, hem da Kristus, da Masra

the city of David, he be Christ, be Lord

city of David, he is Christ, the Lord’

(Luke 1829: 12)

(22)	en na datra (Seuren 1981: 1055)
	he/she is doctor

‘he/she is a doctor’

(23)	da mi, Filida (Van Dyk c.1770: 73)
	it-is me, Filida

‘it's me, Filida’

[p. 32]

(24)	na datra (Seuren 1981: 1055)
	it-is doctor

‘it's a doctor’

(Note: hem and en are the oblique and emphatic forms of a, but clearly no emphatic reading is intended in (21) or (22).) The fact that the neuter subject ‘it’ is not expressed at all, as in (23) and (24), is only to be expected in the light of the postulation of a special sentence-introducing, pronominal copula above. The fact that non-neuter subjects cannot be expressed by a in these cases might be explained by assuming (as is indeed done by Voorhoeve (1953: 73)) that the third person singular pronoun a was historically derived from the demonstrative pronoun da/na: in that case both forms would be (at least diachronically) identical, and, as is well known, languages tend to avoid the direct succession of identical forms. Hence, emphatic en, being phonetically very dissimilar to a, was called upon to perform the function of the personal subject.

A problem that remains to be discussed is the phonological change of da (as a copula and as a demonstrative pronoun) into na. Related to the question of the particular choice of na is the question of why this change originated with copular as opposed to deictic da. Copular na is attested from the very first native source at our disposal (Cesaari 1836-37: 280) onwards, through other native speaker texts such as King 1864-70, King 1891-94⁸ and Albitrouw 1894, in all of which the pronoun is still expressed as da. Herskovits's tales (1936) are the earliest source in our sample to have na in both functions. If both morphemes do indeed have a common source, as suggested above, why would they not undergo the same phonological change at the same time?

The fact that da could be an artefact of European spelling will not account for this, since in that case one would expect the correct spontaneous spelling na by native speakers in both functions, unless, of course, they had adopted European spelling conventions at the time. (Incidentally, it is known that at least one of them, Johannes King, taught himself how to read and write (King 1973: 3-4), which makes a European spelling implausible, at least for this case.) A possible explanation is that syncretism of the deictic function with English that (and of the related determiner with the) caused this form to retain the older spelling, whereas the change to na for the copula could tentatively be viewed to be influenced by the general preposition⁹. The latter hypothesis may seem far-fetched, but it is interesting to note that Welmers (1973: 312) suggests a relationship for Igbo between prepositional nà and the

8 The manuscript has da, which was changed into na by the editor (King 1973: 11).

9 See p. 38 for further possible evidence.

[p. 33]

present action particle nà, which in turn ‘may well be associated with “being at”’. This observation is especially relevant since the morpheme in question in both languages is, at least partially, identical in both form and function. We are left, of course, with the fact that the phonetic difference between both functions is later neutralized.

The latter problem is not accounted for either by Alleyne's hypothesis (which is quite fruitful in other respects) (1980: 89), that in the da/na alternation in various creole languages one can see the modern reflexes of one common historical root *nda, which yielded na and da as alternants. This hypothesized origin becomes all the more likely as it is known that prenasalized stops were often incorrectly spelt by Europeans (cf. Donicie & Voorhoeve (1963: vii) who report Schumann's (1778) misspellings nju and dju for Saramaccan ndju). Voorhoeve's suggestion that pronominal na could be derived from a Bantu noun class prefix (Voorhoeve 1953: 72-72) in conjunction with Daeleman's observation that the initial ‘nasal compound’ (i.e. prenasalized stop) is proportionally very common in the Kongo languages as a noun class feature lends further support to Alleyne's hypothesis. Both explanations are not mutually exclusive, of course, since phonological and functional causes may very well have reinforced each other.

2.3. Historical development

In this section the historical development of the entire copula system, with the exception of clefting (to which a separate chapter is devoted; see Ch. 4), will be described. The copula system as a whole can roughly be divided into four subsystems: location (in the wider sense, i.e. covering existence and possession), adjectival predication, nominal predication and ‘complementation’ - a wastebasket category of sentences where the copula is followed by a, mostly adverbial or infinitival, complement. Since in three of these areas the modern system, by and large, was already established by the end of the 18th century, these three (location, ‘complementation’ and adjectival predication) will be discussed relatively briefly and with an emphasis on their relevance for synchronic problems. In the fourth area (nominal predication or equation), however, a rather dramatic change can be observed, which deserves a more elaborate treatment.

2.3.1. Location, existence, possession

Location/existence. The distinction between these two types of sentences is a notorious problem in traditional grammar. But since this is not the place to try to solve it, I follow one of the authorities in this field and take as a distinguishing criterion for ‘existentiality’ the possibility of deleting the locative adverbial without affecting the

[p. 34]

central meaning of the sentence (Lyons 1968: 389-90). The consequence of this strategy is that, for instance, sentence (25) is taken as an existential, whereas (26) is seen as a locative:

(25)	na mindri foe nengre foeroe wisiman de
	LOC middle of negroes many witches are

‘there are many witches among the blacks’

(King 1864-1870: 17)

(26)	mastra, wini no de na battra (Van Dyk c.1770: 61)
	master, wine not is LOC bottle

‘master, there's no wine in the bottle’

The problem in Sranan, however, is a little more complicated in that the existential/locative verb itself (de) is presumably, through a process of zero-derivation, derived from the locative adverb de, ‘there’. That there was a locative adverb de (besides dape) in earlier stages of the language, as there still is today, is indicated by the relevant entries in the dictionaries by Focke and Wullschlägel. The former has two entries for de, one for the locative/existential verb and aspect particle, and one for the locative adverb ‘there’ as in Tan dè, ‘stay there’. The latter gives de and (na) dapee as alternatives under the entry ‘dort, dorthin’ (‘there’ - locative/directional). Although de is not included in its adverbial function in Schumann's (1783) dictionary, I take it to have been present in the 18th century. Its reanalysis as a verb could be explained by positing a pidgin-like stage with zero copulas (as is in fact done in Chapter 5), where the locative adverb would immediately follow the subject of the sentence and would thus be a primary candidate for a category switch to verbal status. Sentences like the following abound, especially in early sources:

(27)	mastra, mi de (Van Dyk c.1770: 47)
	master, I ø there

‘master, here I am’

(28)	mastra den bassia de (Van Dyk c.1770: 66)
	master the overseers ø there

‘master, the overseers are here’

The fact that de in sentences like these does not have a clear locative meaning, but rather indicates the notion of ‘being present’, is in line with the fact that the true demonstrative locative adverb was (and still is) expressed by dape. This demonstrative function of dape is in agreement with its etymological derivation, i.e. from ‘that place’. This etymology is indicated by etymological spellings like daplesie (Weygandt 1798: 97), da plesie (idem: 135), da presie (idem: 134), da pré (idem: 74), from which an evolution, through metathesis, to drape (the still existing variant of dape) is easily explained.

Returning to our discussion of a criterion for existentiality (i.e. the deletion of the locative adverbial), it will be evident that this criterion is not watertight for

[p. 35]

sentences like (27) and (28) above, since both are clearly locatives, although there is no locative adverbial present (since this is ‘part of’ the verb). In spite of this I have not collapsed the categories of existence and location in my analysis, because there is an important difference between what are intuitively true existentials on the one hand and true locatives on the other. This is, as could be expected on semantic grounds, that true existentials, as opposed to locatives, can never have a zero copula, since it is precisely the ‘being’ itself (the existence) which is being predicated, as in:

(29)	fansortoe wisiman de (King 1864-70: 18)
	all-kinds witches are/exist

‘there are all kinds of witches’

This condition, of course, still holds for modern Sranan:

(30)	jorka no de (Donicie 1959: 36)
	ghosts not are/exist

‘ghosts don't exist’

Locatives, on the other hand, can have a zero-copula, although this is largely restricted to pre-dia (‘here’) position. This seeming idiosyncrasy, which has evolved into a categorial rule for pre-dia and into a variable rule for pre-dape position in modern Sranan (Seuren 1981: 1067), will be explained below.

First we will have to discuss the apparent deletion of locative BE before prepositional na, which is attested in only two sources. Both attestations are quoted below, with relevant contextual information:

(People are looking for two men who have disappeared at a funeral ceremony. They go by boat to Kambaloewa, but the men aren't there.)

(31)	Slee na bilosé èn Kambaloewa na oposè.
	Slee LOC downstream and Kambaloewa LOC upstream.

‘Slee (which was located) downstream and K.

Den toe kampoe dati ben de moro krosbé na

The two villages DEM TNS be more close-by LOC

upstream, these two villages were the closest in

birti èn den ben de na srefi sé foe da liba

area and they TNS be LOC same side of the river

the area and they were at the same side of the river’

(Albitrouw 1894: 30)

There are, however, several reasons for thinking that (31) is not a case of copula deletion. First, since the context is in past tense (as appears from the second sentence in (31)), a present tense for the zero-copula runs counter to what one would expect. Second, the configuration of the two sentences in (31) and their relation to each other point to the possibility of a left-dislocated structure, so that both sentences are really only one sentence, with the first being the subject of the sentence as a whole, which (probably because of its ‘heaviness’) is repeated by the ‘anaphoric NP’

[p. 36]

den toe kampoe dati.

The second attestation is from Herskovits: Anansi has put a tub of water under the tree. Hunter comes to collect the money Anansi owes him, but his feet are dirty so he has to wash them. Anansi then says:

(32)	luku, wan tobo watra na ondro na bom
	look a tub water LOC under the tree

‘look, there is a tub of water under the tree’

(Herskovits's transl.) (Herskovits 1936: 152)

As appears from Herskovits's translation, he indeed interpreted (32) as containing a zero locative copula in front of na, and the resulting interpretation is consistent with the context given. There is another possibility, where no zero copula is posited, but which is less plausible within this context, and that is to take luku wan tobo watra as a fixed expression, meaning ‘to prepare a tub of water’. Compare the relevant entry in Schumann:

(33)	lukku wan tobo watra gi mi (Schumann 1783: 182)
	look a tub water give me

‘see to it that I get a tub of water’

This second reading, however, is less plausible since it can be inferred from the context that the tub was already present under the tree. Also, the latter reading would demand a different intonation pattern with no pause, and thus no comma after luku. These difficulties can be avoided by resorting to yet another interpretation, somewhere between the two discussed so far. This third interpretation would read something like: ‘look around for a tub of water under the tree’. This reading is in keeping with the idiomatic sense given above, as well as with the contextual information. The comma, however, would have to be scratched, since there can be no pause after luku. On this interpretation we will not have to postulate a - otherwise unique - zero locative copula before na.

Still, if Herskovits's interpretation is maintained, it may be useful to remind the reader here that a relationship between the preposition ‘at’ and the locative verb ‘being at’ is not as far-fetched as it might seem at first sight. Compare in this context Welmers's observation, quoted on p.32, of precisely this relationship in Igbo, which is all the more relevant since it concerns a phonologically identical form (nà). And although Welmers is careful to note that in Igbo ‘the /nà/ of locative phrases cannot be analyzed as any kind of verbal construction in the present structure of Igbo’ (Welmers 1973: 312), a historical relationship is an obvious possibility. This would mean that in (32) we would not have a zero locative copula, but rather an old form of the preposition na, still containing the verbal meaning of ‘being at’. I have not pursued the possibility of such a historical relationship for Sranan any further.

After this short excursion we can now turn our attention to those cases where there undoubtedly is a zero locative copula, i.e. in pre-dia and pre-dape positions. As to the latter of these, the variable rule of de-deletion before dape

[p. 37]

is a relatively recent innovation, its first occurrence in the sample being from 1896:

(34)	en wan ipi soema ben dape, omtrenti wan 20
	and a heap people TNS ø there, about a 20

‘and there were a lot of people there, about

soema (Albitrouw 1896: 16)

people

twenty’

In fact, this is the only occurrence of the operation of this rule in the sample, but it has become much more frequent in the modern language, judging by Seuren's statement that ‘BE may likewise be ø when followed by at least some adverbs of place such as drape (“there”) (an alternative form of dape, JA), as in (...) a (de) drape’ (Seuren 1981: 1067). I take it that pre-dape copula deletion was introduced on analogy with other similar expressions such as NP - de (as in (27) and (28) above) and NP - dia (see below). Both are present in fairly early stages of the language, i.e. the second part of the 18th century. Compare for instance:

(35)	a dea (Schumann 1783: 78)
	it ø here

‘it's here’

Although pre-dia copula deletion was a variable rule in the early days (see below), it has become categorical fairly early on, at least by the time Focke compiled his dictionary (1855):

(36)	a d'íä (Focke 1855: 47)
	it ø here

‘it's here’

In spite of the fact that so far I have used the term deletion in connection with these expressions, it should be made clear that, historically speaking, they do not involve deletion at all. As I will demonstrate below, what has happened to de in these contexts is not so much deletion as fusion, i.e. with ia ‘here’, which is the older form of dia and which can be found (although disguised by incorrect spelling) in older stages as in:

(37)	gran tanki fo myn heer a komi ja fo loeke
	great thanks for my Sir he come here for look

‘thank you very much Sir for coming here to visit

da pranasi wan trom (Van Dyk c.1770: 93)

the plantation one time

the plantation’

If we correct the obvious misprint komi ja into kom ija (komi has never been an alternative form for kom/kon), the word reads ija, which is a perfectly regular reflex of English ‘here’ (compare dia from ‘deer’)¹⁰. Even as late as 1855 (Focke 1855: 47) ia is found alongside dia, in a spelling

10 The stress shift from ija to ia is comparable to that in yesi (< ‘ears’).

[p. 38]

(íä: acute accent indicates word stress; diaeresis supposedly indicates syllabification) that supports the derivation from an older form ija, that was attested above. In parentheses Focke provides an etymology for dia, but he incorrectly links it to the relative pronoun/temporal conjunction di (‘who’, ‘when’), with which it has nothing to do. The historical evidence clearly indicates that dia is a merger of ia en de, the locative verb ‘be’. That this is so appears from older spellings like dea, deeja and deja, while forms like *dija are never found (which we would expect if Focke's hypothesis were correct):

(38)	mie deeja heelie boen (Weygandt 1798: 94)
	I am-here very well

‘this is the right place for me’

(39)	wan briefie foe em deja foe joe
	a letter of/from him is-here for you

‘there's a letter from him for you here’

(Weygandt 1798: 94)

Also occurring, but rarely, are hybrid forms with doubly expressed de as in:

(40)	a de dejaso (Helmig 1844: 43)
	he is (is-)here

‘he's here’

(dejaso is an alternative form of deja, perhaps modelled on Dutch hierzo, an alternative, substandard form of hier, ‘here’.) The presence of an overt locative verb in (40) could possibly be explained by European influence, i.e. as representing bakra tongo, ‘white man's Sranan’.

The absorption of de in ia to yield a (superficial) zero copula may have happened on analogy with the very frequent NP - de formula used in locative and existential sentences without a locative adverbial phrase as (27) and (28) above. Because of its frequency and because of its identical semantic domain (location) this structure may have acted as a model for sentences with de ia, leading them to ‘delete’ their copula by incorporating it into the adverb.

One environment where zero locative verbs occur and which has not been mentioned thus far, is that of direct and indirect questions. Although the normal procedure here is to have an overt copula, there are a few instances in the sample where such a copula is not present, e.g.:

(41)	oe plee da klosie diesie joe go tekie?
	which place the cloth ø that you go take

‘where's the cloth that you were going to pick up?’

(Weygandt 1798: 109)

(42)	mek' 'a moni tan, taki mi pe Kakaforu!
	make the money be, tell me where Kakaforu ø!

‘Let the money wait, tell me where Cock is!’

(Herskovits 1936: 154)

That a zero copula is rather exceptional here appears from the fact that all three sources where it occurs (the two mentioned in (41) and (42) plus Van Dyk c.1770) do have a

[p. 39]

surface copula in all other instances of the same pattern, as in:

(43)	mara oe plee da klosie foe mie kamizolo dee?
	but which place the cloth for my shirt is

‘but where's the cloth for my shirt?’

(Weygandt 1798: 108)

I have no explanation for the occurrence of zero copulas in these cases, except that it could be related to the ‘prototypical’ copula-less cases of location, as exemplified in (27) and (28). That is to say, the absence of a copula in (41) and (42) could be a relic of a purely adverbial status of de in declarative sentences; the matrix sentence from which the interrogative sentence is derived would thus not contain a copula.

Finally, two cases of locative sentences have to be discussed where not a zero copula is used but da, the equative copula which is normally used for nominal predication, especially identification. (The development of this category is discussed below.) The two sentences are from the same source and are of a similar structure, although the second is a cleft, while the first is not:

(44)	déjà da wan boen plesie (Weygandt 1798: 94)
	here is a good place

‘here's a good place for you’

(45)	da deeja da da winklie foe Masra G?
	is here is the shop of Mister G.

‘is this where Mr. G.'s shop is?’

(Weygandt 1798: 105)

(In (45) it is the second occurrence of da with which we are concerned here: the first is the clefting copula, the third is the determiner.) It is clear that the locative verb da in both these cases can have nothing to do with the general preposition na (compare (31) and (32) above where a locative na-copula cannot be excluded completely), simply because this preposition has never had the form da. Rather, it would seem that the selection of da in these cases has something to do with the fronted position of dia. In normal position dia is not accompanied by a locative verb, as was illustrated above (compare (35), (36), (38) and (39)). Now, it could be that as a result of its fronted position the adverb acquires something of a nominal character, since it is in a position which is normally occupied by a subject NP. As a result of this it would be quite logical to have it followed by the equative copula, the copula whose function it is to link two NP's. On the other hand, as these cases are so few in number and their occurrence is restricted to one source, it would not be unwise to allow for an idiosyncrasy of the author here, which could or could not be related to the specific lect (i.e. bakra-tongo) he was using.

Possession. Although the number of possessive sentences in the sample is very small (some 20 cases), we can confidently say that up to about 1900 possessive predication is expressed

[p. 40]

by de fu, ‘be of/for’, ‘belong to’, when the possessed item is definite (compare (46)), while indefinite possessees appear as objects of the verb abi, ‘have’ (compare (47)). Donicie's grammar, however, informs us (Donicie 1959: 39) that in the modern language not de but a (a later development of the copula na) is used with definite possessees, as in (48):

(46)	datti de vo dem (Schumann 1783: 81)
	that is of/for them

‘that belongs to them’

(47)	dem no habi so-hudu dea (Schumann 1783: 116)
	they not have such-wood here

‘they don't have that kind of timber here’

(48)	a sani a fu en (Donicie 1959: 39)
	the thing is of/for him

‘that belongs to him’

In connection with the latter Donicie notes that a, when preceded by TMA particles, has to change into de (as is also the case with other uses of copular na, see pp.34-35 above). The use of a as a possessive verb in modern Sranan does not come as a complete surprise from a historical point of view, since there is one case in the sample where da is used as the verb of possession:

(49)	da foe mi (Weygandt 1798: 94)
	it-is of/for me

‘it's mine’

In section 2.2. this use of da was explained as a case of the sentence-introducing copula, which was later replaced by a de as in:

(50)	a de vo mi (Wullschlägel 1856: 12)
	it is of/for me

‘it's mine’

It thus seems that there is something like a cyclical movement in the language, as far as the expression of possession is concerned: from da via de to a. Unfortunately, I have no explanation for this development. (It should be noted, in parentheses, that there is one case in the sample where a zero copula is used to express possession, but in view of the source where it is found (Wennekers 1822: 24, see Appendix B) I do not think this should be taken too seriously.)

There is one interesting exception to the rule, stated above, that with indefinite possessees the verb abi appears instead of de. In ascribing mental or physical (especially unpleasant) states to animate (especially human) beings, such as ‘being hungry’, ‘being terrified’, but also ‘being pregnant’, the expression of these states, although in the form of an indefinite or unspecific NP, never appears as the object of abi. Instead, such NP's are either the subject of abi (or a related word like kisi, ‘catch’ or holi, ‘hold’ or even kili, ‘kill’), or they are the prepositional complement in a comitative ‘be with...’ construction. Thus we have for

[p. 41]

instance:

(51)	dem poti soema di de nanga hangri
	the poor people who are with hunger

‘the poor people who are hungry’

(Luke 1829: 12)

(52)	joe sisa Elisabet de nanga bele toe na hem
	your sister Elizabeth is with belly too LOC her

‘even your sister Elizabeth is pregnant in her

ouroe jari srefi (Luke 1829: 10)

old age even

old age’

The words that appear in this kind of construction in the sample, besides the ones mentioned above, are: broko hatti, ‘broken hearted’, djompo hatti, ‘restless’ and skreki, ‘terror’. Besides these, there is one case where the roles of subject and ‘state-NP’ are reversed:

(53)	Tigri, wan pina de nanga mi
	Tiger, a suffering is with me

‘Tiger, I'm having a hard time’

(Herskovits 1936: 150)

This reversal of the roles of ‘patient’ (in the original sense of ‘one who is suffering’) and ‘state’ can also be observed in sentences of the other type mentioned above, i.e. those with abi or its equivalent, as in (54) en (55) below:

(54)	hekki kissi^11 mi (Schumann 1783: 128)
	hiccup catch me

‘I've got the hiccup’

(55)	mi kissi hekki (ibid.)
	I catch hiccup

‘I've got the hiccup’

Note that, to express the concept of ‘being hungry’, the Sranan speaker may choose from three alternatives: first, the ‘comitative’ structure exemplified in (51), second, a ‘stateprominent’ construction as in (56), and finally, a ‘patientprominent’ one as in (57); the difference between the latter two is one of intensity, as the translation shows:

(56)	hangri de kili mi (Schumann 1783: 120)
	hunger ASP kill me

‘I'm starving’

(57)	mi habi hangri (Schumann 1783: 120)
	I have hunger

‘I'm hungry’

It is interesting to note that both the ‘comitative’ (patient-prominent) type and the ‘catch’ (state-prominent)

11 Compare also korsu e freyri mi, ‘I'm going to have a fever’ (lit. ‘fever is flirting with me’) and korsu kisi mi, ‘I'm having a fever’ (lit. ‘fever has caught me’) (Renata de Bies, pers. comm.).

[p. 42]

type occur in West African languages, although maybe not with the same semantic restriction, i.e. that of expressing a, usually unpleasant, mental or physical state:

Tshiluba

(58)	ndi ne muana (Burssens 1946)
	I-am with child

‘I have a child’

Twi

(59)	mewo sika (Balmer & Grant 1929: 58)
	I-am-with money

‘I have money’

Seidel & Struyf (1910: 43) note for Kikongo that ‘le verbe “avoir” s'exprime par kala (...) = être, avec la préposition ye (avec)’ and add that this holds only for alienable possession; unfortunately, they do not provide any examples. In passing, it may be worthwhile to remark that this type of possessive seems to be typically African: in a typological study of 59 languages (Arends 1982: 49) it turned out to occur in only four languages, three of them being African.

As to the ‘catch’ type, an example from Twi is cited by Alleyne (1980: 119, referring to Schuchardt 1979), but unfortunately no gloss or translation is given. The context, though, indicates that the meaning must be something like ‘hunger is catching me’ (cf. (56)):

Twi

(60)	okom de m' (Alleyne 1980: 119)
	hunger is catching me (?)

‘I'm starving’ (?)

More information is available on Yoruba, of which Rowlands (1969: 127) states:

‘The Yoruba way of expressing emotions, sensations and ailments differs from English in that the person involved is usually the grammatical object, not the subject of the verb of the sentence, e.g. the Yoruba for “I feel cold” is òtútù' mí mi, lit. “cold has caught me”.’

Compare in this connection sentence (61) from Sranan, which is an almost exact calque of the Yoruba expression just cited:

(61)	koure morro mi (Schumann 1783: 160)
	cold overpower me

‘I'm very cold’

Rowlands (ibid.) goes on to note that ‘in some cases a reverse order is possible’, implying a semantic difference as to the active participation of the ‘patient’ in the process. The same is true of Sranan (compare (54) and (55)), although the semantic difference may be less salient.

2.3.2. Complementation

This (ill-named) category is intended to cover all instances of copula sentences that do not belong to any of the other categories distinguished within this study. Since what is

[p. 43]

predicated is, in most cases, an adverbial or infinitival complement, the above label was chosen. Although generally de is the preferred copula for both types of complement, there are a few cases of a zero element, but these are largely restricted to ‘quasi-adverbs’ such as klarie, ‘ready’, which behave like adjectives, and even one instance of the use of na:

(62)	dati na so, ma Philip na libisuma, net so leki
	that is so, but Philip is human, just so as

‘that may be true, but Philip is a human being

ibri trawan (Bruma c. 1958: 142)

every other

just like everybody else’

The use of na in (62) is surely exceptional from a historical point of view, and cannot be easily explained. Within the context of the overall development of the copula system it is less surprising, in view of the fact that also in possessive sentences de has become replaced by na (see p. 39).

As to the use of a zero copula, there are four cases which can all, more or less easily, be explained. There is one instance from Herlein (1718), whose use of a zero element is not surprising since he does not use any overt copula in the entire fragment (see Ch. 5). In Nepveu's corrections of the fragment (1765) no copula is added, so I take it that in this period (say, up to about 1750) Sranan still had enough of a pidgin character to use zero copulas in these contexts. (This point will be discussed more extensively in Ch. 5.) Then, there is one source (Wennekers 1822) where the copula is omitted in a comitative construction (with a non-literal meaning), but this source can be shown to be unreliable on independent grounds (see sections 2.3.3. and 2.3.4. below). The same source also uses a zero in the expression of possession one time, as was noted above. Finally, there is a case from Herskovits (1936) which will be quoted here in full:

(63)	'A so a pikin, 'a so a sa tan. Na so hem mama
	is so he ø small, is so he TNS stay. Is so his mama

‘Little as he is, so he will remain. His mother

na hem papa (Herskovits 1936: 158)

and his papa ø

and his father are like that’

Because of the clefting this case is different from what would constitute a clear case of a pre-adverbial zero copula. But still one would expect a surface copula de to remain after the fronting of the adverb so. Unless, of course, there has not been any de present in the non-clefted sentence, but rather na, as in (62) above. In that case a possible explanation could be that the original copula has fused with the clefting copula, with which it is homophonous.

[p. 44]

As to the use of a zero copula before ‘quasi-adverbs’: apparently these behave like adjectival predicates, for which omission of the copula is the unmarked option (see section 2.3.3.). The adverbs concerned (none of which is ever attested with a surface copula in the entire sample) are: klari, ‘ready’, lati, ‘late’, noefe, ‘enough’, toemoesi, ‘too much’ (the latter two as non-modifying adverbs, i.e. not accompanying an adjectival predicate), fara, ‘far’ and vanoodoe, ‘necessary, needful’. Some examples are:

(64)	tanki mastra a noefe (Van Dyk c.1770: 59)
	thanks master it ø enough

‘please, master, it (i.e. the beating) is enough’

(65)	a farra toemoesi (Helmig van der Vegt 1844: 37)
	it ø far too-much

‘it's too far’

The adjectival character is expressed even more clearly with one adverb, klari, ‘ready’, which has a zero copula in finite as well as in infinitival position. The latter is a characteristic feature of true adjectival predicates, as we will see below (section 2.3.3.). Examples include:

(66)	n'jam n'jam klarie? (Weygandt 1798: 93)
	dinner ø ready?

‘dinner's ready?’

(67)	da n'am n'jam sa klarie na ienie wan affoe
	the dinner TNS ø ready LOC inside a half

‘dinner will be ready in half an hour’

joeroe (Weygandt 1798: 94)

hour

Although none of the other adverbs mentioned here ever appears in infinitival position in the entire sample, we may safely assume that they would behave just like klari, i.e. as adjectival predicates. This only goes to show once more how thin category boundaries are in a creole language like Sranan (cf. Voorhoeve 1981). The same point will be made below with respect to predicate adjectives, when they will be shown to behave like (stative) verbs.

It may be interesting to note in this connection that several of the relevant West African languages do not have a separate category of adverbs. E.g. Rowlands says ‘that Yoruba has no special class of words, like English words ending in “-ly”, which we can obviously label “adverbs”’ (Rowlands 1969: 145). With respect to Twi (Fante) it is said that ‘there are not many words which are exclusively adverbs’ (Balmer & Grant 1929: 139). And, finally, for Igbo the category of adverbs does not appear in a summing-up of the parts of speech in Green and Igwe's grammar (1963: 13). An obvious hypothesis is that the behaviour of adverbs in Sranan may have something to do with substrate influence, but I have not investigated this point any further.

The discussion of de as a complement-introducing copula is best divided into two parts, according to the nature of

[p. 45]

the complement. First, adverbial and then infinitival complements will be discussed. As far as adverbials are concerned, it is clear that, with the exception of the quasi-adverbs mentioned above, de is the obligatory copula in these contexts from the second half of the 18th century onwards. In the earliest document (Herlein 1718), however, the one case with an adverbial complement has a zero copula:

(68)	oe fasse nam vor joe Mastre?
	what fashion name of/for your master ø?

‘what's the name of your master?’

(Herlein 1718: 281)

The fact that this is an interrogative sentence may be relevant here. Compare the situation with locatives, where sometimes the copula is omitted in direct and indirect questions (see p. 38). In Chapter 5 the absence of a copula in this sentence and in the correction by Nepveu (1765) is related to the supposedly pidginoid character of early Sranan. The normal situation, i.e. with an overt copula is illustrated by the following sentences:

(69)	a no dee so (Weygandt 1798: 94)
	it not is so

‘it's not like that’

(70)	nono Masra, diesie dé noja na modo
	no-no master, this is now LOC fashion

‘no Sir, this one is fashionable now’

(Weygandt 1798: 105)

(71)	da zonde dee leki wan wisi (Wennekers 1822: 31)
	the sin is like a wisi (i.e. witchcraft)

‘sin is like witchcraft’

(72)	a de vo datti hedde dissi va mi moesoe tan?
	it is for that reason this of/for me must wait

‘is that the reason why mine has to wait?’

(Helmig van der Vegt 1844: 39)

(73)	a de na wroko (Wullschlägel 1856: 18)
	he is LOC work

‘he's at work’

To summarize, what appears from these examples (and many more in the Appendix) is that, with the one exception of so, which in modern Sranan can have na as a copula (see p. 43), de is and almost always was the obligatory copula with adverbial complements, regardless of their semantic function (temporal, modal, etc.). The fact that it is the locative verb which is selected to perform this function, could very well be related to the fact that what is expressed in the adverbial complement is unstable in terms of the ‘time-stability-parameter’ (Givón 1979: 320-23) and thus unsuitable to be expressed by the ‘time-independent’ equative copula da/na. (See section 2.3.4. for a more elaborate discussion of this point in connection with adjectival and nominal predication; the same argument could be made with respect to the

[p. 46]

expression of possession and existence by de: both involve a temporal situation, except, perhaps, where a theological concept of eternity is involved, as in Gado de, ‘God exists’.)

Turning our attention now to infinitival complements, we will first give a few examples:

(74)	...da hattibron, disi de vo kom (Luke 1829: 22)
	the anger that is for come

‘...the anger which is about to come’

(75)	a de vo wiki froekoe vo go loekoe wrokko
	it is for wake early for go look work

‘you have to wake up early to supervise the work’

(Helmig van der Vegt 1844: 48)

(76)	a no de vo verandre (Wullschlägel 1856: 7)
	it not is for change

‘it can't be changed’

(77)	den no de foe feni den (Albitrouw 1894: 30)
	they not are for find them

‘they couldn't find them’

(78)	disi wanwan soema de toe vo finni, disi de
	these few people are also for find who ASP

‘those few people can also be found who keep

kibri dem srefi vo dem fasi dati

keep them selves from the manners those

themselves away from those customs’

(Makzien 1902: 4)

(79)	wantem trawan de vo hindre hem nanga dem
	sometimes others are for annoy him with the

‘sometimes others may annoy him with their empty

soso kondre-singi (Makzien 1902: 4)

empty country-songs

folk-songs’

From these examples it appears that de fu either gets a future tense interpretation (as in (74)) or a modal interpretation, which involves an obligation (as in (75)) or a possibility ((76) - (79)). Within the latter category there is a difference between (76) and (78) on the one hand, and (77) and (79) on the other: in the former the subject of the sentence performs the semantic function of patient, whereas in the latter the grammatical subject is the agent of the action expressed by the verb. The ‘subject=patient-type’ may be modelled on Dutch, where exactly the same structure is possible:

Dutch

(80)	het is niet te veranderen
	it is not to change

‘it can't be changed’

[p. 47]

(81)	er zijn enkele mensen te vinden die...
	there are some people to find who...

‘some people can be found who...’

The future tense type (as in (74)) might be modelled on an English structure such as:

(82)	the time which is to come
	‘the time that will come’

Finally, the modal ‘subject=agent-type’ may have been influenced by a related construction with fu which also involves a modal interpretation (although one of obligation, not possibility) and where fu is preceded by abi, not de. It seems clear, by the way, that this abi fu, ‘have to’ construction was taken from English. Some examples are:

(83)	joe no ha' foe go moro (Focke 1855: 41)
	you not have for go more

‘you don't have to go anymore’

(84)	so a fo tan dape te den ben kom
	so he-had for remain there until they TNS come

‘so he had to remain there until they came and

feni hem (Herskovits 1936: 166)

find him

found him’

(85)	en so Konum a fo gi Anansi wan leti-mindri
	and so King had for give Anansi one precise-half

‘and so the King had to give Anansi half of his

kondre fo hem (idem: 172)

land of him

Kingdom’

Note that in glossing (84) I proceed from the assumption that a is a contraction of the pronoun a, ‘he’ and the verb a, ‘have’, a reduced form of abi. Contractions of this type are very frequent in Sranan, and it may very well be that this is what caused all the fuss about fu being a modal auxiliary in Sranan. This idea was promoted by Byrne (1984: 100) and Bickerton (1984: 181), both, I suppose, led by the mistaken idea¹² that fu is a modal auxiliary in Saramaccan:

Saramaccan

(86)	i fu naki di mii (Bickerton 1984: 180)
	you for hit the child

‘you should hit the child’

Sentence (85), where the subject is Konum and not a (as in (84)), provides clear evidence that fo is a complementizer preceded by the verb a, and not an auxiliary itself. Apart from that, I have never seen any evidence that fu is a verb in Sranan: it cannot be tensed, infinitized or clefted, as

12 That fu is not a modal in Saramaccan either is supported by Wijnen & Alleyne (1987: 43), who also derive a fu from a a fu, ‘he has to’.

[p. 48]

true modals can, such as musu:

(87)	a ben moesoe foe fadon (Focke 1855: 82)
	he TNS must for fall-down

‘he had to fall down’

(88)	da moesoe joe sa moesoe (Grammatik 1854: 47)
	is must you TNS must

‘you'll have to’

It is ironic that Bickerton rests his claim for the modal status of fu entirely on sentence (84), which he explicitly states (o.c.: 181) he has taken from Washabaugh (1975: 129). The irony is that in the latter article (84) is immediately followed by (85) which, as shown above, provides clear evidence to the contrary with regard to the modal status of fo. This calls for an extension of Bickerton's ‘First Law of Creole Studies’ (Bickerton 1981: 83): not only can every creolist's analysis be directly contradicted by that creolist's own texts and citations, but also by the texts and citations he omits.

2.3.3. Adjectival predication

The question of the categorial status of predicative adjectives in Sranan - are they ‘plain’ adjectives or are they (stative) verbs? - has received quite some attention lately. Seuren (1981) sparked off the discussion by claiming that predicative adjectives are adjectives, preceded by an underlying copula which is realized as a zero unless otherwise specified. The reason for him to do so was that, although in ‘prototypical’ cases (i.e. simple declarative sentences) predicative adjectives behave exactly like stative verbs, under certain conditions an overt copula de appears. This happens in contexts where the subject is not immediately followed by the predicate. Until Seuren's article appeared, the generally accepted analysis had been the one by Voorhoeve (1957) who, on the basis of their combinatory possibilities with TMA-particles, chose to classify these adjectives as verbs. In an unpublished paper (Arends 1985), which was a preliminary version of the present chapter, I supported this view, with the restriction that indeed, as Seuren claims, these ‘adjectival verbs’ sometimes behave differently from ‘verbal verbs’ (like sabi, ‘know’). In a recent article Sebba (1986) reacted to Seuren's analysis by adducing evidence for the verbal status of adjectival predicates. Seuren (1986), in his reaction to Sebba, defended his earlier view, but later (1987) accepted the exceptional character of one of these (siki, ‘ill’), which he showed has to be classified as a verb. Although I do not pretend to be able to solve this controversy, I hope my historical data may contribute to a clarification of the problem. In order to get a clear picture of the relevant facts I will first present a survey of the behaviour of predicative adjectives in a number of different contexts in modern Sranan. After that I will present my historical data in the same order, so as to

[p. 49]

facilitate comparison.

Predicative Adjectives in Modern Sranan

(a) simple declaratives

(89)	a liba bradi
	the river wide

‘the river is wide’

(90)	*a liba de bradi
	the river is wide

(b) infinitives

(91)	a liba musu bradi
	the river must wide

‘the river must be wide’

(92)	a liba musu de bradi
	the river must be wide

‘the river must be wide’

(c) wh-questions

(93)	o bradi a liba de?
	what^13 wide the river is?

13 It may even be that in a sentence like (93) bradi is not an adjective but a noun (‘width’). Note that I have glossed o as ‘what’, not as ‘how’. The correct gloss of the sentence would have to be: ‘what width the river is?’. In this connection it may be interesting to know that the same type of adjectival predication occurs in a language like Kikongo, where a sentence like ‘they are happy’ is constructed as ‘they are happiness’ (Bentley 1887: 1027). In relation to this it is also important that in my opinion the correct gloss of o is ‘what/which’ (ultimately derived from ‘who’), not ‘how’ as most authors (Echteld 1961: 96; Focke 1855: 45) would have it. The derivation of o from ‘who’ is supported by older spellings like hu, as in hupeh, ‘where’ and hutem, ‘when’ (Schumann 1783: 133-34) and oe as in oe plesse, ‘where’ and oe tem, ‘when’ (Herlein 1718: 280-81). The evidence is strengthened by the fact that the regular reflex of the English diphthong in ‘how’ in Sranan would be ow (as in kow < ‘cow’), whereas /u:/ is regularly reflected as /u/ (cf. tu < ‘two’) (Smith 1987: 436; 462-63). It is also significant that other creoles that have analytical question words (Saramaccan, Haitian, Krio, Negerhollands) always choose the equivalent of ‘what/which’, not ‘how’ (Boretzky 1983: 154). Finally, every Kwa language investigated for this study (Ewe, Igbo, Yoruba and Twi) constructs its question words with an equivalent of ‘what/which’, not ‘how’ (Boretzky o.c.: 155-56; Rowlands 1969: 141, 160, 174; Balmer & Grant 1929: 139). Pieter Seuren (p.c.) suggests that the derivation of o from ‘who’ does not constitute evidence for the nominal status of predicates like bradi, because of cases like omeni, ‘how much’ which clearly cannot be analyzed as ‘what + Noun’. I feel, however, that omeni may have been built on analogy with other question words like oten, ‘when’ and (o)pe, ‘where’; this could imply that omeni is a single, unanalyzable unit, in which o is not present as a separate morpheme.

[p. 50]

‘how wide is the river?’

(94)	o bradi a liba bradi?
	what wide the river wide?

‘how wide is the river?’

(95)	fa a liba bradi?
	how the river wide?

‘how wide is the river?’

(96)	*o bradi a liba?
	what wide the river?

(d) preposed modifiers (so, ‘so’, tumusi, ‘very’, nofo, ‘enough’; but not moro, ‘more’ (see under (f))

(97)	a liba de so bradi
	the river is so wide

‘the river is that wide’

(98)	*a liba so bradi
	the river so wide

(99)	a liba bradi so
	the river wide so

‘the river is that wide’

(100)	*a liba de bradi so
	the river is wide so

(e) clefts

(101)	na bradi a liba bradi
	is wide the river wide

‘the river is really wide’

(102)	*na bradi a liba (de)
	is wide the river (is)

(f) comparatives

(103)	a liba disi moro bradi leki a trawan
	the river this more wide than the other-one

‘this river is wider than that one’

(104)	a liba disi bradi moro leki a trawan
	the river this wide more than the other-one

‘this river is wider than that one’

(105)

a liba disi bradi moro a trawan

[p. 51]

the river this wide exceed the other-one

‘this river is wider than the other one’

(106)	*a liba disi de moro bradi leki a trawan
	the river this is more wide than the other-one

(g) unrestructured Dutch loans

(107)	mi de enthoesiast (= Du. ‘enthousiast’)
	I am enthusiastic

‘I am enthusiastic’

(108)	* a de moy (< Du. ‘mooi’)
	she is beautiful

A few remarks will have to be made concerning this survey. First, although (90) is generally considered to be ungrammatical (Sebba 1986: 110), there are some situations, language varieties and groups of speakers for which it is not, at least when the predicate contains a word like tru, ‘true’. Donicie (1959: 38) notes that both a tru and a de tru are possible, although the latter is not for older, monolingual speakers, unless it has a clear emphatic function. It is more frequent in the younger generation, perhaps through interference from Dutch; it also occurs in non-emphatic contexts, and in Europeanized literary and pulpit language (Voorhoeve 1971). Second, although Seuren (1981: 1055) presents (92) as a normal and (91) as a special sociolectal variant, the picture that emerges from my historical sample is somewhat different. This is especially surprising in view of the characteristics attributed to the speakers of the latter variety, which are ‘young’, ‘male’ and ‘urban’. If synchronic variation reflects historical change, one would expect (92) to be the alternative form found more frequently in older sources. Third, as I have been informed (Renata de Bies, p.c.), (94) is indeed grammatical, but it is used very infrequently. This corresponds to my observation that it does not occur in my sample. Fourth, Renata de Bies (p.c.) also informs me that (95) is found especially with older speakers, which is in agreement with the fact that it is found in older sources within the sample. Donicie (o.c.: 98), however, gives (95) and (93) as equivalents, with no additional information. Incidentally, he does not mention (94) at all as a third variant. Fifth, Seuren (1986: 130) suggests that bradi in both (97) and (99) has to be interpreted in a neutral, non-oriented way (i.e. not implying that the river is indeed wide). The non-neutral sentence would involve a double occurrence of bradi, as in:

(109)	a liba bradi so bradi
	the river wide so wide

‘the river is só (terribly) wide’

And, finally, although (106) is generally assumed to be ungrammatical in modern Sranan, it is found as late as 1936, even though most of the occurrences in the sample date from before 1850. (See Ch. 3 for a detailed discussion.):

(110)

bika a ben de moro hei leki Falek

[p. 52]

because he TNS be more high than Falcon

‘because he was higher than Falcon’

(Herskovits 1936: 194)

With this synchronic picture in mind we will now proceed to a presentation of the historical data. In doing so we will follow the same order as in the above survey.

(a) simple declaratives. The normal case for predicative adjectives here is to have no overt copula. Only in a minority of cases (approximately 10%) does a copula de appear; most of these cases fall under category (g) or represent instances of the semantic differentiation of adjectives in a stative and a non-stative reading. (Both will be discussed below.) A few examples of sentences which do not belong to either of these categories are:

(111)	a de fulu tumusi, a de go passa abra
	it is full too-much, it ASP go pass over

‘it's too full, it's going to overflow’

(Schumann 1783: 59)

(112)	son dee tranga kaba (Weygandt 1798: 98)
	sun is strong already

‘it's very hot already’

(113)	mi bri' ji de lau (Focke 1855: 17)
	I believe you are crazy

‘are you crazy?’

(114)	Gado wroko de bigi vo troe (Makzien 1913: 51)
	God work is big for true

‘God's doings are really great’

As far as a rationale for these few cases is concerned, it might be remarked that in (113) and (114) an emphatic intention cannot be excluded: cf. Donicie's remark on the preceding page. As to (111) and (112), a possibility here is that, instead of functioning as stative verbs (as predicative adjectives with a zero copula seem to be doing), these two are used here as non-stative (more specifically inchoative) verbs, i.e. ‘to begin to be full’, ‘to begin to be strong’. In that case de would have to be glossed not as a copula, but rather as an aspect marker de, which is obligatory for this category of verbs in the present tense. That what appear to be adjectives may indeed be used as inchoative verbs, becomes clear from examples like the following, where bigi, ‘big’ has the meaning of ‘to become big’, i.e. ‘to begin to be big’, not of ‘to be big’ (compare also Seuren 1981: 1048):

(115)	Tyotyoforu n'e go bigi moro. Na so no
	Tyotyoforu not-ASP go get-big more. Is so not

‘T. (a small bird) is not going to get bigger.

mo a de bigi (Herskovits 1936: 156)

more he ASP get-big

He only gets this big’

(116)

A no habi fo bigi moro. ‘A so a pikin,

[p. 53]

He not has to get-big more. Is so he small,

‘He (i.e. Toad) doesn't have to get any

‘a so a sa tan (idem: 158)

is so he TNS remain

bigger. He will remain as little as he is’

Of course, apart from these tentative explanations, all four of (111) - (114) may reflect synchronic variation as a result of incomplete decreolization.

To show that in the majority of cases a zero copula is used, I will give four examples with the same ‘adjective’ as in (111) - (114). Sometimes a semantic difference can be observed, as between tranga in (112) and in (118):

(117)	mi belle fulu (Schumann 1783: 59)
	my belly full

‘I am satisfied’

(118)	mi belle tranga (idem: 46)
	my belly strong

‘I have constipation’

(119)	dem kom tan leki dem lau (Luke 1829: 38)
	they come be like they crazy

‘they lost their minds’

(120)	mi no bigi yete (Herskovits 1936: 156)
	I not big yet

‘I'm not big yet’

(For many more examples of this type the reader is referred to the relevant section in Appendix B.)

(b) infinitives. Here the situation is similar to that described under (a), i.e. there are a few exceptional cases with a surface copula de (approximately 10%), but in the majority of cases there is no such element. The former are all from the same source (Wennekers 1822), which has been shown to be unreliable on independent grounds (see sections 2.3.1. and 2.3.2. above; see also section 2.3.4. below). One example is given here:

(121)	datti no sa dee takroe? (Wennekers 1822: 5)
	that not MOD be evil

‘isn't that evil?’

Examples with no overt copula include:

(122)	a sa boen na mie foetroe (Weygandt 1798: 92)
	it TNS good LOC me for-true

‘I will be very pleased’

(123)	aránja moe groen, bifósi a lépi (Focke 1855: 11)
	orange must green before it ripe

‘an orange must be green, before it gets ripe’

Within this category many examples can be found to support the view, expressed above, that some predicative adjectives are in fact inchoative verbs. For instance, apart from (124), there are eight more examples where dede means ‘die’, i.e. ‘begin to be dead’ and not ‘be dead’:

[p. 54]

(124)	da somma no sa dedde, a dorro kaba
	the person not TNS die, he get-through already

‘the man's not going to die, he's through it’

(Schumann 1783: 88)

(125)	mi fom zomma morre den za ziki kwetti
	I beat people more they TNS get-ill quite

‘If I'll beat these people more, they'll get really ill’

(Van Dyk c.1770: 57)

(126)	djompo-hátti foe sabakóe
	jump-heart of/for sabakoe (a kind of bird)

‘the fearful nature of the sabakoe prevents it

méki a no kan fáttoe (Focke 1855: 24)

makes it not can get-fat

from getting fat’

If this view of predicative adjectives as inchoative (i.e. non-stative) verbs could be substantiated by further research, it would open the possibility of a new analysis, somewhat analogous to Alleyne's proposal (1980: 98; 1986) to interpret them as perfective passives of causative verbs. That is, mi siki would have to be interpreted as either ‘I have become ill’ or as ‘I have been made ill’. Note that this view runs counter to earlier proposals, made also by myself and discussed at the beginning of this section, to view predicative adjectives as stative verbs. Within the context of the present, whose aim is mainly historical, I have not investigated this matter any further.

(c) wh-questions. There is only one example in the sample with o:

(127)	o brada (sic) a dee (Weygandt 1798: 107)
	what wide it is

‘how wide is it?’

(Note that bradi is misspelt here as brada.)

Further, there are a few examples with (hoe) fa, ‘what fashion’, ‘how’, but they are either constructed differently (as (128)) or they have a different meaning (as (129))¹⁴. To show that this construction type did indeed exist in older Sranan in the same form as it still does now, I present one example from outside the sample (sentence (130)).

(128)	we, jere now fa den heiden kondre
	well, listen now how the heathen villages

‘well, listen now how heavy life is in the

libi de so ebi (King 1864-70: 16)

life is so heavy

heathen villages’

14 (128) has de so before the adjective; (129) cannot refer to ‘being dead’ since it concerns the manner in which somebody is actually committing suicide.

[p. 55]

(129)	hoe fa a didde (Van Dyk c.1770: 75)
	what fashion he die

‘how is he dying?’ (i.e. committing suicide)

(130)	Azáu sâbi fa hem lási brâdi, a
	Elephant knows how its bottom wide, it

‘the elephant knows how broad its bottom is,

swâri kokronoto (Focke 1855: 128)

swallows coconuts

it swallows coconuts’

The fact that (130) is a proverb, representing an older stage of the language, together with the fact that the only other example, (128), is constructed in a deviant way, may very well point to the possibility that the fa-type is an older variant, which is now largely restricted to older speakers (see p.52). In view of the fact that there is only one attestation of the o-type with de, (127), the complete absence of this type with copying of the adjective (cf. (94)) may not be all that significant. In other words: it may very well be due to the (relatively) small size of the sample.

(d) preposed modifiers. Of the nine cases in the sample seven are constructed with so, while the remaining two have pikien(so), ‘a little’ as a modifier. Further there are some cases which do not properly belong to this category since there is no preposed modifier involved. Still, they have been classified here, because they correspond to the other cases in that the subject of the sentence is not immediately followed by the predicative adjective. In all cases there is an overt copula present:

(131)	ju no de so poti (Schumann 1783: 53)
	you not are so poor

‘you're not that poor’

(132)	mie dee piekienso swakki jetee (Weygandt 1798: 97)
	I am a-little weak still

‘I still feel a little weak’

(133)	moro switi joe de, moro mi kan lobbi joe
	more sweet you are, more I can love you

‘the sweeter you are, the more I can love you’

(Grammatik 1854: 13)

(134)	ma foe den boesi kondre moro ebi libi
	but for/of the bush villages more heavy life

‘but bushland life is the most difficult’

de (King 1864-70: 16)

is

[p. 56]

(135)	syen^15 a de te wi frigiti wi afo
	shame it is when we forget our forefathers

‘but it is shameful when we forget our forefathers’

(Koenders 1946-49: 138)

Note that in (134) as well as (135) contrastive emphasis is involved, which is clearly borne out by the contextual information not repeated here. Summarizing, it would seem that there is a general principle involved, dictating that a surface copula be present when the subject is not immediately followed by the adjectival predicate, unless the intervening element (such as moro or an auxiliary like musu) is conceptually incorporated in that predicate. This accounts not only for the cases under (131)-(132) but also for the other cases presented here, as well as for the behaviour of wh-questions. Seuren (1986b) puts forward a possible explanation for this principle in that its function is to signal non-canonical, i.e. non-S V (0), word order.

(e) clefts. All cases of predicate clefting of an adjectival predicate in the sample involve copying of that predicate. These cases (including some marginal cases which not everybody will agree are clefts: see Ch. 4 for details) are illustrated here by (136) and (137):

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

‘the boy's not naughty, but he's incredibly

morsoe toemoesi (Grammatik 1854: 47)

dirty too-much

dirty’

(137)	na siki Tigri siki, a dede
	is ill Tiger ill, he dead

‘Tiger is very ill, he's dead’

(Herskovits 1936: 152)

(f) comparatives. All three types current in modern Sranan occur in the sample, even in some of the oldest sources.

Besides these copula-less variants we find another one, with an overt copula preceding moro, which is ungrammatical in the modern language. Still, in the sample it has been attested as late as 1936 (Herskovits). All four of these types will be illustrated by a sample sentence. For further information on the comparative the reader is referred to Ch. 3.

(138)	bakra moro koni leki ningre
	white-man more smart than negro

‘whites are smarter than blacks’

(Grammatik 1854: 13)

15 syen is a noun originally, but it clearly functions as an adjective here, illustrating the phenomenon of multifunctionality, so well known in creole languages like Sranan (Voorhoeve 1981).

[p. 57]

(139)	ju betre morro leki mi? (Schumann 1783: 62)
	you better more than me

‘are you better than me?’

(140)	a biggi morro mi (idem: 62)
	he big surpass me

‘he's taller than me’

(141)	hem de moro switi leki joe (Grammatik 1854: 12)
	he is more sweet than you

‘he's sweeter than you’

Of the nineteen examples in the sample with preposed moro twelve have a zero copula and seven have de. Of the latter, five are from pre-1850 sources. This clearly indicates that this is the older variant. The transition from the overt-copula-type exemplified in (141) to the zero-copula-type illustrated in (138) can be explained by postulating a process of grammaticalization: in the former moro is still taken as a lexical adverb ‘more’, which, since it is preposed to the adjective, necessitates an overt copula; in the latter it is beginning to be grammaticalized as a comparative inflection of the adjective, comparable to the English ‘-er’ suffix, and thus, with no modifier intervening between subject and predicate, leads to the selection of a zero copula. (This issue will be discussed more fully in Ch. 3.)

(g) unrestructured Dutch loans. In most of the cases where an adjective is involved which phonologically still clearly is a Dutch loan (such as getrouw, ‘faithful’, zeker, ‘certain’, voorzigtig (sic), ‘careful’ and zelfstandig, ‘independent’), a copula appears. But also in this area there seems to have been some variation, as appears from the comparison of (143) and (144), in both of which the adjective has been rephonologized to the same extent (ferkoutoe, ‘having a cold’ < Du. ‘verkouden’ and benauwtoe, ‘anxious’ < Du. ‘benauwd’), whereas it triggers a surface copula in only one:

(142)	ala sani disi de onmogelijk na libisoema
	all things that are impossible LOC people

‘everything which is impossible for man’

(Kraag 1894-96: 35)

(143)	mi de ferkoutoe pikienso (Helmig 1844: 44)
	I am having-a-cold a-little

‘I'm having a bit of a cold’

(144)	mi hatti benauwtoe (Wullschlägel 1856: 7)
	my heart anxious

‘my heart is anxious’

Finally, a few remarks must be made concerning an issue which does not fall within the scope of this survey properly, since I cannot relate it to the situation in the modern language. The reason is simply that I have never seen a discussion of the issue from a synchronic point of view. The point I want to discuss concerns the fact that for some

[p. 58]

adjectives (e.g. bun, ‘good’) the presence or absence of a copula seems to be related to their semantic differentiation. This differentiation is always related to the category of stativity, i.e. for bun there is a differentiation between two meanings, one stative (‘O.K.’) and one non-stative (‘mentally/physically well’). Relevant examples include:

(145)	a bon (Van Dyk c.1770: 58)
	it good

‘it's O.K.’

(146)	mi de bun (Schumann 1783: 74)
	I am good

‘I'm fine’

A similar differentiation can be observed in the opposition dede, ‘be dead’ vs. de dede, ‘take pains’:

(147)	yu mama dede na Sotwatrasei
	your mama dead LOC Saltwaterside

‘your mother is dead (has died) at Saltwaterside’

(Herskovits 1936: 158)

(148)	a de déde de tappoe hem (Focke 1855: 21)
	he ASP take-pains ASP stop him

‘he's doing his best to stop him’

In connection with bun perhaps, instead of stativity, the more general notion of ‘time-stability’ (Givón 1979: 320-23) should be called upon here, because this concept will prove to play a major role in the differentiation of the equative copula as well (see section 2.3.4.). If this analysis is accepted, the time-stability-parameter appears to be a central category in the syntax of Sranan: it operates in the verbal as well as in the adjectival and nominal systems. In the former its function is to assign the aspect particle only to non-stative verbs, i.e. mi e waka, ‘I walk/am walking’ vs. mi sabi, ‘I know’; this will not be discussed any further here. Indeed, the difference between one's ‘being well’ and the time-less concept of something ‘being O.K.’ is not so much a difference of action/process vs. state, but rather one of being time-dependent or not. Thus, on a time-stability-continuum the former would be more on the unstable, the latter more on the stable pole. A similar parameter seems to operate in languages like Ewe and Yoruba, where there is a difference in the predication of characteristics which are permanent and those which are temporary. Thus in Yoruba two different copulas are used, one (jé) to express permanent, and one (se) to express temporary states:

Yoruba

(149)	ó jé okùnrin (Rolands 1969: 153)
	he is man

‘he is male’

(150)	ó se okùnrin (ibidem)
	he is man

‘he is manly/brave’

It may be significant that in Ewe, where most predicative

[p. 59]

adjectives are really verbs, some characteristics such as ‘being well/healthy’ are expressed with the aid of the locative verb BE:

Ewe

(151)	ele afi (Westermann 1907: 73)
	he-is here

‘he's here’

(152)	ele nyuie (ibid.)
	he-is fine

‘he's fine’

The parallel to Sranan mi de bun, where the copula selected is also locative/existential BE, should be obvious. The question of why it is locative BE that is selected to express adjectival (as well as nominal) predication in a number of languages will be discussed in the next section.

Summarizing: compared with the situation of today three major differences can be observed in the expression of adjectival predication. First, in simple declaratives there is sometimes an overt copula. Second, predicative adjectives in infinitive position hardly ever have a copula. And third, in comparatives with preposed moro a surface copula sometimes appears, especially in older texts. The first and third of these cases were tentatively explained where they were presented in the text. As to the second, this could tentatively be explained by resorting to the same kind of explanation that was suggested in connection with the absence of a copula in comparatives with pre-adjectival moro: auxiliaries like moesoe, ‘must’, kan, ‘can’ and sa, ‘will’ are incorporated conceptually into the verb phrase and thus do not separate the subject of the sentence from the predicate, which prevents the copula from appearing between them.

2.3.4. Nominal predication.

The historical development of the equative copula is the only area within this study which lends itself to a fully quantitative analysis. Of the 1176 copula sentences in the sample 222 represent nominal predicates (either of the identifying or the attributive type). If these sentences are analyzed according to the selection of one out of three copula forms (da, de or ø) the following picture emerges (from Arends 1986: 111):

[p. 60]

Figure 1: Historical development of the equative copula in Sranan

The most important conclusion to be drawn from this figure is that, somewhere between 1800 and 1850, de takes over the prominent position of da as the equative copula and that this change is later consolidated. Also, the zero copula shows a clear decline in its usage after 1800. The relatively high frequency of the zero copula in stage I (around 35%) is in agreement with the fact that it is the only copula form to be found in the one pre-1750 source available (Herlein 1718). In other words, the decrease in usage of the zero copula from 100% in 1718 to less than 10% in 1800 supports the idea of the pidgin character of early Sranan and the concomitant hypothesis of gradual creolization which will be discussed below and elaborated more fully in Ch. 5. It will be seen that zero copulas are largely restricted to special contexts, such as negation, which will be dealt with below.

One proviso with respect to Fig.1 should be made. The peak in the proportion of de in the 1800-1850 period is due to one source that is prone to overgeneralization of this form in all functions (Wennekers 1822; see also sections 2.3.1., 2.3.2. and 2.3.3.). If corrected for this overrepresentation, the development would be much more gradual than Fig.1 makes it appear to be.

In an attempt to find an explanation for this rather dramatic change the equative category was split up into two subcategories, that of attribution (description, qualification) and that of identification. The first type is exemplified by a sentence like ‘That man is a teacher’, the second by a sentence like ‘That man is my neighbour’.

According to Lyons (1977: 471-472) the syntactic difference between the two depends principally upon two facts: 1) the nominal predicates in the two types do not belong to the same ‘expression class’, i.e. they cannot be freely interchanged; and 2) subject and complement of identifying, but not of attributive, sentences are freely permutable. If we divide all equational sentences according to these criteria into

[p. 61]

these two subcategories, we can clearly see that the rise of de over da is related to the emergence of these separate categories from the - previously uniform - class of equatives. Compare Figs. 2 and 3 (from Arends 1986: 112):

Figure 2: Historical development of the attributive copula in Sranan

Figure 3: Historical development of the identifying copula in Sranan

(Note that the same remark that was made with respect to the peak in the de-curve in Fig.1 should be made in connection with Figs. 2 and 3.)

Figures 2 and 3 show that, after an initial stage (1750-1800) where da was the preferred copula for attribution as well as identification, the attributive category develops a preference for de (which was already marginally present), while the expression of identity is largely delegated to da. In both cases the other copula is possible as a secondary alternative, whereas the zero copula shows a decline in both functions. As to the origin of the split of the formerly uniform category of equation, I would say that it is most probably the emergence of attribution as a distinct category which has been the driving force in this process. This is based on the idea that the process under discussion is not just one which has to do only with equation, but one which has wider implications in the sense that it is an example of the introduction of the concept of ‘temporal instability’ into the language (compare section 2.3.3. where this idea is discussed in connection with adjectival predicates). The idea is strengthened by the fact that it is de, the indicator of non-stativity with verbs and certain adjectival predicates, which is selected to mark the category of attribution, whereas da, which was the preferred copula for the undifferentiated equative category in stage I, is reserved for the expression of identity. In other words, the language has changed from a stage where it had only one equative category which was undifferentiated with respect to time (expressed by da), to a stage where it does differentiate its equative category with respect to time by expressing the time-independent category of identification by its ‘timeless’

[p. 62]

copula da (which cannot be tensed) and by marking the time-dependent category of attribution by its non-stative, i.e. time-dependent copula de.

In order to illustrate this development relevant examples will be quoted from stages I and II, in each case representing the situation which is typical for the period concerned:

Stage I (attribution)

da

(154)	adjossi, da Bakkratongo
	(the word) ‘adjossi’ is Bakratongo

‘ldadjossi” is Bakratongo’ (Europeans' Sranan)

(Schumann 1783: 46)

ø

(155)	djinja wan bune dressi vo tranga
	ginger ø a good medicine for strong

‘ginger is a good medicine against head-aches’

hedde-hatti (Schumann 1783: 84)

head-ache

(identification)

da

(156)	droifi watra, da wini (Schumann 1783: 92)
	grape water is wine

‘wine is juice of grapes’

de no examples

ø

(157)	who^16 somma datty? (Stedman 1796: 227)
	what person ø that?

‘who's that?’

Stage II (attribution)

de

(158)	mi wefi de wan bejari soema toe (Luke 1829: 8)
	my wife is an aged person too

‘my wife is also very old’

(identification)

de

(159)	efi joe de Gado pikien (Luke 1829: 26)
	if you are God child

‘if you are the son of God’

16 Strangely enough this form is spelled qui somma in the 1799 Dutch translation of Stedman (Pieter Seuren, pers. comm.). I have no explanation for this other than the printer's fancy to insert a French interrogative pronoun here.

[p. 63]

ø

(160)	joe pikien ben Gado, bifo joe meki Hem
	your child TNS ø God before you made him

‘your son was God before you made him’

(Luke 1829: 19)

Sentence (160) serves to illustrate the fact that zero copulas are not always restricted to special contexts, such as negation. Sentence (160) may be compared to sentences (19) and (20) (see section 2.2.), where also a preverbal marker precedes a zero copula. Or should we rather say that ben is verbal in these cases instead of preverbal, carrying the original meaning of its supposed etymon been, ‘was’?

The idea that it is not just the opposition between attribution and identification which is marked by the selection of da versus de, but rather the ordering on a scale of time-stability is supported by the following two examples, which are taken from one source:

(161)	Philip na libisuma (Koenders 1946-1949: 142)
	Philip is living-person

‘Philip is a human being’

(162)	Philip de wan takru suma (Koenders 1946-49: 142)
	Philip is a bad person

‘Philip is a mean person’

The comparison of these two sentences is especially striking since it is the same noun (suma, ‘person’) which makes up the head of the nominal predicate. Apparently, the difference between the two lies in the fact that in one case a time-stable condition (‘being a human being’) and in the other a time-unstable condition (‘being a mean person’) is being predicated. One cannot help being reminded of sentences (149) and (150) in Yoruba where a similar opposition (‘being a man’ versus ‘being manly’) leads to the selection of two different copula forms.

As stated earlier, zero copulas can occur in attributive as well as in identifying sentences, but their use is largely restricted to contexts where some other element (such as a negator, a demonstrative pronoun or a clefting copula) is present to take over the function of the copula:

negative

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

‘I'm not your slave’

demonstrative

(164)	soortoe maniri dati? (Wullschlägel 1856: 20)
	what manner ø that?

‘what kind of behaviour is that?’

cleft

(165)	da hem da somma, disi joe soekoe
	is he ø the person that you look-for

‘he's the one you're looking for’

(Grammatik 1854: 19)

[p. 64]

As to (163), it will be remembered that the a no sequence, which is normal for negative copula sentences, is a single morpheme for native speakers, so that it may be not quite correct to speak of a ‘zero copula’ in cases like this, it not being at all clear that the a is a copula in the first place. As will be shown in Ch. 4, no instead of a no is also quite frequent in negative cleft sentences. Regarding (164) it should be remarked that the Sranan copula da was derived from the pronoun of the same form, but ultimately from its predecessor dati (compare sentence (8) above), which in (164) apparently fulfills both functions, i.e. that of pronoun and of copula. Finally, in (165) the presence of the clefting copula may be the cause of the absence of the identifying copula, in which case it performs a dual function.

As far as an explanation for the above-mentioned differentiation is concerned, I think it is best seen as a grammatical innovation, aimed at enhancing semantic transparency and resulting in a different form for each semantically distinguishable category (see Seuren & Wekker 1986). Viewed this way, the problem of the structure and differentiation of the equational system has two sides to it - the distribution of forms over different semantic categories, and the semantic distinctions themselves.

As to the first (formal) side, it must be remembered that Givón's time-stability concept was adduced to explain the differentiation of the equative copula. Within this framework the selection of de, the morpheme used as preverbal aspect marker with non-stative verbs and thus already functioning as a marker of time-instability, is a very natural thing to happen, since it is precisely the factor of time-instability which sets attribution apart from identification. But perhaps one might go one step further back into the etymological history of de and see the original, locative function as the main factor in this development.

Location can be seen as a primary category, from which other categories are derived by linguistic metaphor (Anderson 1971). The relation between location, existence and possession is a classic example in this respect (Lyons 1967). For instance, in Niger-Congo the relationship between location and possession is very common (see section 2.3.1.; see also Welmers 1973: 308-309), but, more interestingly for our case, languages of the Kwa branch within this family (such as Igbo, Yoruba and Twi) relate these two categories to the marking of present action (Welmers 1973: 315). This is important because in Sranan preverbal de also serves primarily to indicate present action (traditionally called ‘aspect’ in the literature on the subject): unmarked verbs designate a state (i.e. time-stability), as in mi wroko, ‘I (have) worked’ or mi sabi, ‘I know’, whereas marked verbs indicate present action (i.e. time-instability), as in mi (d)e wroko, ‘I work/am working’ or mi (d)e sabi, ‘I'm beginning to know’.

The impression one gets from this is that in Sranan as

[p. 65]

well as in Kwa time dependency (‘being in time’) is expressed by a spatial metaphor (‘being in space’¹⁷). This explains not only why it is locative de that is selected to mark present action, but also why it is generalized to mark the nonstative members of semantically differentiated pairs of adjectives like bun (see section 2.3.3.) and, additionally, to differentiate time-dependent attribution from time-independent (or maybe more accurate ‘time-less’) identification. It is significant in this respect that languages that distinguish locative and equative copulas in affirmative present tense contexts, but not in negative or past tense sentences, always choose the locative in the latter case (Leon Stassen, pers. comm.). Although the use of locative (or locative/existential) BE as an equative copula can be demonstrated for Kwa, it is by no means restricted to this group. For instance, Munro (1976) has revealed the existential origin of the equative copula in the Yuman languages. Further investigation will have to show whether the phenomenon has universal aspects or not. For the moment a few examples from Igbo will suffice to show the relationship:

Igbo (location)

(166)	o di n'ulo (Spencer 1924: 74)
	he is LOC house

‘he's in the house’

(equation)

(167)	ayi di madu ato (ibid.)
	we are three persons

‘we are three persons’

Within this line of reasoning it is only to be expected that non-verbal, tenseless da would be used for the expression of the time-independent relation of identity. That it is also the preferred copula in odo's (proverbs containing timeless truths) (Favery et al. 1976: 89) is not surprising either:

(168)	soema bére da liba (Focke 1855: 10)
	person belly is river

‘the human heart is unfathomable’

As to the second part of the problem mentioned above - that of the underlying semantic differentiation of the copula system -, there are strong indications of substrate influence, especially from Kwa. All four Kwa languages that were investigated have a very fine-grained copula system with an abundance of different copula forms: Ewe, Twi and Yoruba have between six and eight different options, through which all kinds of subtle semantic distinctions can be formally

17 This suggestion was made by Leon Stassen (pers. comm.), who also referred me to the articles by Munro and Li & Thompson. In this connection also compare Welmers' observation that ‘location and present action appear to show an underlying relationship in a variety of languages’ (Welmers 1973: 312).

[p. 66]

marked. For instance, Ewe and Twi have a separate copula to express functions like that of ‘minister’ as opposed to other kinds of professions like being a ‘teacher’:

Twi

(169)	odzi sofo (Balmer & Grant 1929: 39)
	he-is minister

‘he's a minister’

(170)	oye kyerekyerefo (ibid.)
	he-is teacher

‘he's a teacher’

(For Ewe see Westermann (1907: 73 ff.; for Yoruba see (149) and (150) above, where two examples are given to demonstrate the fact that this language has one copula to express temporary, and another to express permanent states.) Although the distinction between identity and attribution is sometimes formally marked in Kwa (as in Igbo (Welmers 1973: 311) and Twi (Balmer & Grant 1929: 39 ff.)), this does not seem to be the crucial difference. Fixed class membership (like ‘being a man’ as opposed to ‘being a linguist’), for instance, is grouped together with identity in Ewe (Westermann l.c.). All this put together points in the direction of time-stability as the crucial factor in the differentiation of the equative category in these substrate languages.

Although, as far as I know, no systematic studies are available concerning universal aspects of copula systems, the ‘pre-Greenbergian’ literature on the expression of BE in the languages of the world (Benveniste 1966; Locker 1954) does not indicate that the differentiation of the equational category is a universal phenomenon. In addition, this differentiation is not accommodated in Bickerton's bioprogram, as he lists the equative as an undifferentiated category (Bickerton 1981: 68). Within the field of creoles it is thus an intriguing question whether this distinction is made in other creoles as well (but has gone unnoticed thus far), or whether Sranan is an exception to the rule. Whether the distinction is universal or not, the locus in time at which the differentiation took place in Sranan (one and a half centuries after the language's inception) makes a bioprogrammatic explanation impossible, since the bioprogram is by definition fully operative only in the first generation of creole speakers.

The general - and rather surprising - conclusion to be drawn from the foregoing is that the underlying semantic differentiation of the equative category seems largely substrate driven, whereas the formal marking of this differentiation appears to be something of a universal phenomenon.

2.4. Summary and conclusion

In order to give an overall picture of the development of the Sranan copula I have reconstructed the historical scenario presented in Fig.4. In this scenario the intermediate

[p. 67]

character of the equative within the entire copula system is represented by the fact that it derives the morphemes to express and differentiate its function from the two extremes of the copula spectrum, i.e. the truly verbal locative BE (de) on the time-unstable and the clearly non-verbal pronoun (da) on the stable pole of the time-stability continuum:

Figure 4: Reconstructed scenario for the development of the Sranan copula

The most important conclusion that can be drawn from the development of the Sranan copula, especially that of the equative, is that creoles do not always become stable systems within one generation. Long after the transition from the hypothesized preceding pidgin stage they are still amenable to large scale restructuring. It is of course true that all languages change, but both the rate at which the particular change, described above, took place, as well as the central character of the domain in which it happened, indicate that it is of a different, perhaps specifically ‘creole’ nature. Or, to put it in the form of a question: could it not be the case that the observed change is somehow still part of the creolization process, in the sense that the language - as a not yet fully stabilized and relatively young system - is still in search of its stable form?

Viewed this way, creolization is not a discrete, single-generation process as some would have it (cf. Bickerton 1984), but rather one that is gradual, extending over a number of generations. It seems worthwhile to contemplate this hypothesis since the idea of ‘instant’ creolization is a basic assumption for the language bioprogram hypothesis, which makes far-reaching claims of a highly abstract and at the same time highly specific nature about the origin of creoles. One would expect any such theory to be based on detailed examination of the known historical facts. As it is, however, this specific theory appears highly vulnerable when held against what historical evidence is available. The

[p. 68]

gradual character of creolization does not exclude, of course, universals as a factor in creole genesis. The important question, however, is to determine the relative extent to which these factors are at work and the extent to which they are supplemented by other factors, such as substrata.