The Saram ramaccan imp implos losive ives: too tools for for ling linguist istic arch rchaeolog logy?
Norval Smith and Vinije Haabo University of Amsterdam & Wageningen University 1. I ntroduction ntroduction
Comparatively recently it was established (Haabo 2000) that Saramaccan, a mixed English/Portuguese-lexifier creole language of Surinam, possessed two types of voiced stop. Next to the plain voiced series of /b, d, dj, g, gb/ it turned out out also to have two implosive voiced stops /∫, Î/. This This article icle represents a prelimin limina ary attempt to answer two question ions. Firs Firstly, ly, the question ion of the historical source of these two voiced stop series in Saramaccan. Why do we have two types of labial and coronal stops in Saramaccan? Secondly, can we utilize this difference to reveal any kind of stratification in the Saramaccan lexicon? We will postpone the first question to the last, and deal with the second question first. There are two reasons for this approach. First of all, there is a great deal more evidence bearing on the second question, and secondly, by adopting this approach approach we progress from more recent history history back back into into thepast. past. In addi additition, on, the first questi question on is the more fundamental. Now, to start with, we give one possible take on the consonant system of Saramaccan. . (1) The Saramaccan Saramaccan consonant system Stops: voice voicelless voice voiced d implosive Fricatives: voice voicelless voice voiced d Nasals: asals: Liquids: Gli Glides:
L abial
Alveolar veolar Palata Palatall
Velar
L abial bial-Ve -V elar
p b
t d
k g
kw kp1 gw gb
∫
Î
f v m
s z n l
tj dj
hw (?) (?)2
j
Glot Glotta tall
h
w
1
Some dialects of Saramaccan appear to distinguish /kw/ and /kp/, others do not (Smith and Haabo, 2004)
2
We wil not discuss the question of the possibility of a phoneme /hw/ here.
2. The The Saram Saramacca accan n le lexicon xicon
The The Saramaccan lex lexico icon contains ins words fro from a great variet iety of sources. Ho How wever we will restrict ict ourselves here to five important linguistic sources: (2)
Engli nglish: Portugue ortuguese: se: Fon: K ikongo: Dutch:
the dominant nant superstrate lexical xical donor the second second most domi dominant lexical exical donor the main African ri can substrate substrate langua anguage ge anothe anotherr important portant Africa rican n substra substrate te langua nguage ge as the coloni colonial al langua anguage ge,, and and a continua continuall source of lexical xical item tems
We ignore ignore here here the various various Am Amerindi erindian an languages that have have contri contributed to the Saramaccan lexicon, including Carib, Wayana, and Tupi. It is not clear to us that this would alter the picture that we hope will emerge from this study. The The dist istribu ibution ion of the two types of voice iced stops is at firs first sigh ight difficu ifficult to relat late to any other feature. As far as the linguistic source of particular lexical items is concerned, this does not necessarily provide us with a clear guide as to which type of stop to expect in any given word. Because of significant differences in the distributions of plain voiced and implosive stops in terms of their place of articulation, we will deal separately with labial and coronal stops. 3. L abial stops stops
L et us look first at some some example ples of the lab labiial stops corre correspondi sponding to /b/ in in thevarious various source source languages. In English-derived words we find both types. At first sight there is no seeming logic as to why one set of words has implosives, and another plain voiced stops. Tab Ta ble 1.
Imp Implos losive ives and plain lain voice iced lab labial ial stops in Eng English lish-de -derive ived words in Saramaccan
Engli nglish
Saramaccan accan
gloss gloss
bottle
∫áta
'bottle'
black
∫aáka
'black'
bastard
bása
'bastard'
beg
bégi
'beg, pray'
The same applies to words of Portuguese origin, except that voiced /b/ is much scarcer. Table 2.
Implosives and plain voiced labial stops in Portuguese-derived words in Saramaccan
Portuguese
Saramaccan
gloss
butar
∫utá
'place, lay'
boca
∫úka
'mouth'
baixhar
baziá
'sink, descend'
but: (de)baixho
∫ásu
'low, below'
When we turn to Kikongo words we find only words with an implosive. The Kikongo forms used here are taken from Swartenbroekx (1973) unless otherwise stated. Note that the the automatic lengthening of vowels before nasals in Kikongo is not marked here. It is also relevant to remark at this point that stops in nasal clusters in Saramaccan arenever implosives. So with the exception condition that implosives never occur after nasals in Saramaccan, we can say that all voiced labial stops in words of Kikongo origin are implosives: Table 3.
Implosive labial stops in Kikongo-derived words in Saramaccan
Kikongo
Saramaccan
gloss
ba-kúlu
∫akúlu
'supernatural apparition'
mbámbi
∫ámbi
'sp. lizard'
mbaansya (Daeleman 1973)
∫andja
'side, beside, wall'
bi-lóngo
∫iO⁄NgO
'ritual ingradients'
Fon or Gbe3 words, finally, reveal the same pattern.
3
The Gbe-derived lexical items in Saramaccan appear to be either fromFon or closely-related Gbe lects.
Table 4.
Implosive labial stops in Fon/Gbe-derived words in Saramaccan
Fon
Saramaccan
gloss
bèsé
∫ése
'frog, toad'
bod ZrE⁄(-tó)
∫ódjee
'hypocrisy'
baŸ
∫a
'draw water'
vEŸ
∫E
'red'
Although we have restricted our examination here to initial labial stops, it can be stated that intervocalic labials display a similar picture. What can we concludefrom these observations? If we were certain that the African strands in the Saramaccan vocabulary were old, i.e. date from the 17th century, we could take the occurrence of implosives as a sign of age in terms of adoption into the language. Now, this is something we can be fairly confident about, since Fon and Kikongo are not only the most important African lexical contributors to the Saramaccan vocabulary, the dominance of these two areas in the slave trade is early, and in the case of Fon stops early in the eighteenth century (Arends 1995) for reasons connected with Slave Coast internal politics (Aboh and Smith to appear). We can also date the formation of Saramaccan as such quite well. There are strong arguments for dating this to approximately 1680 (cf. Smith 2002; Smith and Cardoso 2004; McWhorter 1997). The Portuguese lexical items display virtually the same picture of implosive stops, with a single exception. And there is good reason again for locating the Portuguese influence on Saramaccan to the period between 1665 and approximately 1710. Greater precision on this last point is possible, but not necessary here. The English words display a different picture, but as it is obvious that English influence on Saramaccan was early – it is the main lexifier language after all, we need an explanation for this fact. The most obvious explanation is that there are two lexical layers or strata chronologically speaking, as there is no apparent phonological conditioning. It seems clear that Saramaccan has been influenced by its sister creole language Sranan, the language of the capital, Paramaribo, and the former plantation area on the coast. One layer then might represent the original English element in Saramaccan, and the other later accretions from Sranan Following on from the idea that implosives appear in the older African and Portuguese items in the language, the simplest hypothesis would then be that the words with implosives represent the original English-derived stratum of vocabulary in Saramaccan, and those with voiced stops the later loanwords from Sranan. It might seem that it would be difficult to decide what the cut-off point between these two groups would be but we do have certain external evidence to take into account. Firstly we know that all /b, d/ in Ndyuka are implosives. Ndyuka is considered to descend from Plantation Sranan (Price 1983; Migge 1998) so that at the time of formation of this creole, Plantation Sranan must be assumed to have had implosives as well. This should not surprise us since the three languages all ultimately derive (at least as far as their English element is concerned) from the same source4. 4
A historical phonological comparison reveals that only a small number out of hundreds of English-derived
Ndyuka was formed in the period following 1715 (Price 1976). Now present-day Sranan does not have implosives, so we must assume that it lost them at some period after the formation of Ndyuka, presumably due to Dutch influence.Alternatively, Plantation Sranan and Paramaribo Sranan parted may well company earlier on in the eighteenth century in this respect, the plantation dialects retaining their implosives longer, and the town dialect losing them under Dutch influence. We know that there were different varieties of Sranan spoken in Paramaribo and in the plantation area (Van der Berg to appear). Words like /bása/ and /bégi/ were already in Saramaccan by 1778, the year of Schumann's first word-list of this language, although Schumann does not distinguish implosives and voiced stops, so we might conceive of the possibility that there was a period of increased Sranan influence on Saramaccan following the peace treaty with the Dutch in 1762. The Moravian Church started its missionary activities among the Saramaccan tribe shortly afterwards. This hypothesis is of course only precisely that - a hypothesis. What about Dutch-derived words? What pattern do we find here? We find both types of labial stop in this case too. Table 5.
Implosives and plain voiced labial stops in Dutch-derived words in Saramaccan
Dutch
Saramaccan
gloss5
blaas
∫aási
'bladder, balloon'
balk
∫áiki
'cross-beam'
brak(en)
baláki
'bring up'
blommetje
bolómítji
'flower'
The form /baláki/ is specifically marked "(NE)"6 in Donicie & Voorhoeve (1963) indicating that their informant regarded it as a (recent) loan from Sranan. Note that all four words have an internal liquid in English, but not consistently in Saramaccan. The only liquid in Saramaccan is the lateral /l/, and in virtually all English-derived and Portuguese-derived lexical items internal liquids have been lost. In Dutch words we see a dichotomous pattern - in older words, like those already present in Schumann (1778) they are lost, in more modern words, like the abovementioned /baláki/ they are preserved. So, in the above table we can see a relationship between preservation of the liquid, and the occurrence of the plain voiced stop - which is also the phonetic value of /b/ in Dutch, indicating the later adoption of this type of labial, at least in these examples. How about obviously modern Dutch words? Do these tell us anything? Probably not so easily, as we cannot immediately tell by inspection whether they have come into the language diirectly from Dutch, or through the intermediary of Sranan. But here is a list fairly obvious
words in the three Surinam creoles go back to phonologically different etyma in Early Modern English (Smith 1987). 5 6
The gloss referred to here is the Saramaccan gloss. NE =Negerengels, an older name for Sranan.
recent importations. We will compare the Saramaccan form with its Sranan equivalent if available, to allow us to see any signs of independent development. Table 6.
Modern Dutch words with labial stops in Sranan and Saramaccan Dutch
Saramaccan
Sranan
gloss
voetbal
ba⁄al⁄ i
futubál
'football'
brom
bOlO⁄m
brom(báysigri)
'moped'
box
bO⁄ku⁄su
bóksu
'(loud)speaker'
ambulance
abila⁄nsi
anbyláns
'ambulance'
fabriek
fabili⁄ki
fabríki
'factory'
bankroet
bangulu⁄tu
bankrútu
'bankrupt'
gebit
ka∫E⁄ti
(<) tífi, k ´bét(i)
'false teeth'
batterij
∫atulE⁄i
batréy
'torch'
It seems clear that the majority of the most obviously modern Dutch items have plain stops in Saramaccan, which could be explained by them being borrowed through the intermediary of Sranan. Let us turn now to a consideration of the situation with coronal stops 4. Coronal stops
a. initial In initial position we find both the implosive and plain stop types represented in the Englishderived vocabulary Table 7.
Initial implosives and plain voiced coronal stops in English-derived words
English
Saramaccan
gloss
down +go
ÎO⁄ngO
'go down'
dead
ÎE⁄ÎE
'dead'
drink
⁄ngi dii
'drink' (n.)
dig
díki
'dig'
But although we find both types of stop here, there is a major difference between coronal English-derived words and their labial counterparts. By far the majority of initial coronal stops are plain voiced stops. We will introduce some (simple) statistics to illustrate this later on. Interestingly enough, words where the stop has developed from an initial voiced dental fricative seem all to have an implosive in Saramaccan. This is a small group of pronominal words in English, and presumably belongs to the oldest stratum of English words in Saramaccan. But it is also a small group. Table 8.
Initial implosives in English-derived pronominal words in Saramaccan
English
Saramaccan
gloss
there
ÎE
'there'
this
Îísi
'this'
them
Îe
'they'
that
Îa
'be'
Words, from whatever source, with successive labial and coronal stops, have implosives in both cases if the labial precedes, but if the coronal comes first this is realised as a plain voiced stop and the labial as an implosive. This reinforces the idea of a major difference in the status of labial and coronal voiced stops in Saramaccan. Table 9.
Implosives and plain voiced stops in words containing both labial and coronals
source
Saramaccan
gloss
biddag (Dutch)
∫E⁄Îa⁄ki
'Christmas'
bed (Eng.)
∫e⁄Îi
'bed'
devil (Eng.)
didi⁄∫i
'devil'
dobrar (Port.)
do∫a⁄
'fold (shut)'
Note that stems with two labials, or two coronals always have stops of the same type (except for the above-mentioned restriction on stop-type after nasals). Both are implosives. First some labial examples:
Table 10. Implosives in words containing both two voiced labial stops source
Saramaccan
gloss
Bantu language
∫aku⁄∫a
'sp. banana'
unknown source
∫a∫a⁄u
'stupid'
baboon (Eng.)
∫a∫u⁄nu
'baboon'
bebir (Port.)
∫e∫e⁄
'drink'
and then coronal oness: Table 11. Implosives in words containing both two voiced coronal stops source
Saramaccan
gloss
dead (Eng.)
ÎE⁄ÎE
'dead'
unknown source
Îe⁄eÎ ⁄ e
'palmworm'
lie down (Eng.)
ÎiÎo⁄n
'go into hospital'
de dia (Port.)
ÎiÎi⁄a
'in the daytime'
When we turn to Portugese words, we find a similar pattern with initial coronal stops to the English one - a few words with implosives, and the large majority with plain voiced stops. Table 12. Initial implosives and plain voiced coronal stops in Portuguese-derived words Portuguese
Saramaccan
gloss
deixhar
Îisa⁄
'leave, let go'
de dia
ÎiÎi⁄a
'in the daytime'
dentro
déndu
'inside'
dormir
duumi
'sleep'
Kikongo words do not exhibit any cases with an initial voiced coronal stop, so we ignore them here, while we will postpone the discussion of Fon words until later.
b. medial In the case of the medial coronals we find a similar situation to the initial and medial labial case, but radically different from the case of initial coronals. So English-derived words mostly have medial implosives. Table 13. Medial implosive and plain voiced coronal stops in English-derived words English
Saramaccan
gloss
yesterday
e⁄siÎe
'yesterday'
Friday
fee⁄Îa
'Friday'
god
ga⁄Îu
'god'
head
he⁄Îi
'head'
paddle
pa⁄da
‘oar’
as do Portuguese-derived ones. Table 14. Medial implosives and plain voiced coronal stops in Portuguese-derived words Portuguese
Saramaccan
gloss
mascarado
masikaa⁄ ⁄Îu
'mask'
pedir
pi⁄ ⁄Îi⁄
'ask, beg'
sardinha
saaÎi⁄an
'sp. fish'
escada
sika⁄ ⁄Îa
'ladder'
cedro
se⁄du
'cedar'
Once again Kikongo words do not exhibit any case with an intervocalic medial coronal. So far then, we have identified a strange dualism in the treatment of coronal. Initially, plain voiced stops are most frequent, and medially implosives (except of course post-nasally). 5. Statistics
Let us now look at theactual numbers of examples of the various types present in the two conditions in the words from the different languages. First, looking at the initial labials, we find the following cases:
Table 15. Frequency of initial labial plain voiced and implosive stops in Saramaccan labial initial
plain voiced
implosive
English
5
43
Portuguese
1
26
Kikongo
0
9
Dutch
15
29
Here the preponderance of implosives is striking. The only significant source of voiced stops initially is found in the lexical items adopted from Dutch, and even here the implosives are in a substantial majority. The larger proportion of plain stops we could interpret as being due to more recent borrowings from Sranan, as we have indicated above. Next we examine medial labials, ignoring words with labial or coronal initials, as we have seen that these have an influence on the type of medial stop. Table 16. Frequency of medial labial plain voiced and implosive stops in Saramaccan labial medial
plain voiced
implosive
English
0
17
Portuguese
0
20
Kikongo
0
2
Dutch
3
11
The figures do not significantly differ from those for initials. In both cases the implosives are in the large majority. If we sumthe two tables we get 24 voiced - 157 implosive. The suspicion - as yet unconfirmed - is that the voiced cases could well be loans from Sranan, this suspicion is given some basis because the figures for Dutch are deviant: initial Dutch plain voiced stops 15:29 [34%], other sources 6:78 [7%]; and medial Dutch plain voiced 3:11 [21%], other sources 0:39 [0%]. Let us now turn our attention to the coronals, first the initial cases.
Table 17. Frequency of initial coronal plain voiced and implosive stops in Saramaccan coronal initial
plain voiced
implosive
English
15
7
Portuguese
4
3
Kikongo
-
-
Dutch
22
3
Here we see a completely different picture, as we have suggested. The voiced initials are more numerous than the implosives. Dutch is again different in that it has an even greater preponderance of plain voiced stops. Now we examine the medial cases, once again ignoring words with labial or coronal initials. Table 18. Frequency of medial coronal plain voiced and implosive stops in Saramaccan coronal medial
plain voiced
implosive
English
2
7
Portuguese
1
5
Kikongo
-
-
Dutch
3
13
Here we observe that the initial and medial cases differ significantly. The difference between Dutch and the other source languages disappears, although in this case the number of examples is too small to be of much significance. The proportions found are: initial plain Dutch voiced 22:3 [88%], other sources 19:10 [66%]; and medial Dutch plain voiced 3:13 [19%], other sources 3:12 [20%]. The initial implosive figures for English may be inflated, either because pronominal words have undergone some special development due to their non-lexical status, or more likely because because at least three of the cases occur only in enclitic words, and are as such maybe better regarded as medial cases rather than initial ones. Now we sum up the statistics7.
7
Here we ignore the possible medial status of enclitic words, and treat them as initial cases.
Table 19. Overall frequencies of plain voiced and implosive stops in Saramaccan Excluding Dutch Plain voiced Implosive
Plain voiced
Implosive
Initial Labial
6
78
21
107
Medial Labial
0
39
3
50
Total Labial
6
117
24
157
Initial Coronal
19
10
41
13
Medial Coronal
3
12
6
25
Total Coronal
22
22
47
38
Total
28
139
71
195
Ignoring the Dutch-derived lexical items, which we assume gives us a better idea of the original situation, we can conclude that the striking result emerges that originally we probably had a situation with only implosive labials, while there were apparently two kinds of coronals, skewed in distribution as between initial and medial positions, with the plain voiced stops dominating in initial position and the implosive ones in medial position. At this point we might think of relating the distribution of coronal stops to the differential allophony of English lax stops - lax voiceless initially versus plain voiced intervocalically. This might seem at first sight not to be the whole explanation however, for two reasons. Firstly we would not expect the difference we observe between the distributions of labials and coronals, and secondly we find a preponderance of voiced coronals initially, while in English voicing is precisely characteristic of medial allophones. Although as far as the last point is concerned it might be argued that implosives sound even more “voiced” than plain voiced stops. Let us now deal with the first potential problem as this might seem to be serious, but is, as we will show, extremely interesting, and turns out to be a very important piece of evidence. If we indeed assume that there were two coronals stops available, but only one labial (in view of the small number of plain voiced labials in seemingly older words) in early Saramaccan, this would allow us to utilize the above explanation in respect of the differences in type between initial and medial coronal stops in words of English origin. We would have both implosive and plain voiced coronal stops available, and these are differentially distributed in parallel fashion to the two allophones of lax coronal stops in English. But if we have only one labial stop available – the implosive - this would perforce appear both in initial and medial contexts. In the next section we will go on to ascribe the labial-coronal asymmetry to Fon influence. 6. Fon influence?
Now it so happens that the Gbe languages in general, and Fon in particular, have precisely one voiced labial stop and two voiced coronal stops.
(3)
/b/ /∂ / /d/
voiced labial stop post-alveolar voiced stop dental voiced stop
These three phonemes, while phonetically all plain voiced stops, are not all the same type of phonological "animal". Fon has two types of voiced stop, paired and unpaired (Capo 1991). Paired stops are stops participating in a voicing contrast. So, in terms of the table below, /d/ is paired with /t/, /g/ with /k/, /gb/ with /kp/. /b/ and /∂ / are however unpaired. They do not have a voiceless counterpart8. The full consonant system of Fon is set out in the following table, which takes account of certain features of the phonological behaviour of /b, ∂ /. We will return to this point below. Table 20. The Fon consonant system dental/ alveolar
palatal
velar
labiovelar
voiceless stop
t
c
k
kp
voiced stop
d
j
g
gb
labial
uvular
labiouvular
voiceless fricative
f
s
X
XW
voiced fricative
v
z
Â
ÂW
"sonorants"
b
m
sonorants continued
~
∂ ~
n
y~ y)
w~ w)
l~ l)
/b, ∂ / show a common, and striking, phonological behaviour. We have termed them here "sonorants", which should strike most readers as odd (and is supposed to). They are in complementary distribution with the corresponding nasals, which are clearly sonorants. Before oral vowels we find the allophones [b] and [∂ ]; before nasal vowels we find the allophones [m] and [n]. In this feature they resemble the sonorant consonants /l, y, w/, which also have nasal allophones when they precede nasalized vowels.
8
This is no longer true of the modern Gbe languages, as they have borrowed numerous lexical items from European languages. This is irrelevant for the period we are talking about however.
(4)
preceding a non-nasalized vowel (examples from Fon, Agbome dialect (Gbéto 2002))9 [lo] /lo/ 'crocodile' [abO] /abO/ 'fore-arm' [a∂u] /a∂u/ 'tooth' preceding a nasalized vowel [l‡a‡] /la‡/ 'meat' [mO‡] /bO‡/ 'road' [anO‡] /a∂O‡/ ‘breasts’
What kind of phonemes are /b/ and /∂ /, then? Note that they differ in a number of ways from other "voiced stops". Firstly, they lack a corresponding voiceless partner, as we have noted above. Capo (1991) has referred to these as [unpaired] stops, but as Gbéto (2002) notes, this feature [unpaired] is an arbitrary and non-phonological one. Secondly, we have the special nasal allophones of /b, ∂ / which we have just noted. Thirdly, we have a variety of tonal effects. We will follow Gbéto (2002) here in the assumption that the lowering tonal effect displayed by the paired voiced stops, was not present with sonorants, nor significantly with /b, ∂ / in Proto-Gbe. According to Gbéto, the southernmost Gbe languages still possess this feature. And fourthly, the two coronal stops differ in phonetic place of articulation. The unpaired stop is post-alveolar in the Gbe languages, while the unpaired one is articulated further forward. Note that all four of these features are typical of "implosive" stops. Greenberg (1970) provides an extensive discussion of the general features of these sounds. Firstly, implosives are nearly always voiced. Some Mayan languages have voiceless implosive allophones in word- or syllable-final position – frequent contexts for the devoicing of stops. And occasionally, we find (non-contrastive) voiceless implosive sounds in other contexts. Secondly, the phononological relation between implosives and plain nasals is briefly touched on by Greenberg (1970: 137-138). Thirdly, that implosives tend to cluster with voiceless obstruents rather than voiced ones in their tonal effects is a widespread phenomenon (cf. Greenberg 1970: 132-133). And fourthly, the retraction of coronal implosives is reported frequently from languages around the world (Greenberg 1970: 129) . However, /b, ∂ / in Gbe are not implosives. What are they then? Before we study this question in more detail, let us take a look at the reflexes of Fon /d, b, ∂ / in Saramaccan. We will list all the lexical items identified so far with reflexes of these three sounds.
9
I have omitted tones in these examples.
Table 21. /d/ - the paired Fon (post-)dental Fon
Saramaccan
gloss
degOn
adingO
'shrimp'
lEgEdE
lEgEdE
'lie'
vodu⁄n
voÎu⁄
'sacred snake'
agida
agida
'big drum'
dangbe⁄
dagbe⁄
'large snake, snake god'
denkpe
dEkpE
'dagger'
de‹n
dE⁄n
'coxcomb'
We see that out of seven Saramaccan words identified as having Fon paired /d/, six display a plain voiced /d/ in Saramaccan. One word only shows an implosive /Î/, indicated by gray shading in the table. Table 22.
/b/ - the unpaired Fon labial
Fon
Saramaccan
gloss
ba
∫a
'gather, draw (water)'
be⁄se
∫e⁄se
'toad, frog'
bojrE⁄-to⁄
∫o⁄djee
'hypocrisy'
bu‹
∫u⁄uu⁄ ⁄
'completely' (ideophone)
bosa⁄klE
a∫osi⁄kele
'cricket'
Here we see that the unpaired Fon labial corresponds to an implosive in the five cases identified in Saramaccan. Table 23. /∂ / - the unpaired Fon post-alveolar Fon ka⁄∂a⁄
Saramaccan ka⁄Îa
gloss 'green tree-snake'
So far only one case has been identified. /∂ / is however a fairly rare sound in Fon. This single case displays Saramaccan implosive /Î/. We summarize the results in the following table.
Table 24. The Saramaccan reflexes of Fon labial and coronal stops Fon paired Fon obstruent stop Fon “sonorant stop”
Saramaccan unpaired
d: 7
plain
implosive
d: 6
Î: 1
b: 5
∫: 5
∂ : 1
Î: 1
Note the near-parallellism between the last two (pairs of) columns. Not only do the Saramaccan forms go back to Fon originals, they mirror closely the paired-unpaired distinction of the Gbe languages. While these facts might suggest to us us that Saramaccan /∫, Î/ and Gbe /b, ∂ / are birds of a feather, this does help us very much in identifying the nature of the birds concerned. We would like to have a more reasoned phonological characterization of Saramaccan /∫, Î/ and Gbe /b, ∂ /. Without getting too technical in this article, we would refer the interested reader to the thesis on nasals by Botma (2004). There he argues that phonetic nasals are phononological sonorant stops, in the context of his theory of Element-based Dependency Phonology, thereby giving recognition to both features of nasals - their sonorancy, and the fact that they have complete closure (stopping) in the oral passage. But the two Fon nasals [m, n] are only allophones of their nonnasal counterparts /b, ∂ /. Botma gets round this conundrum by arguing that while nasals are the normal realisation of sonorant stops, this does not mean that sonorant stops are always realised as nasals. They do behave like sonorants, and this gives us the answer to the problem of their allophony in Fon. When they precede non-nasalized vowels they are are plain sonorant stops (not realized as nasals); when they precede nasalized vowels, the sonorant stops are contextually harmonized, undergoing nasalization. The Saramaccan implosives no longer reflect this allophony, except for a few irregular cases. Due to the overwhelming amount of lexical material imported from other sources a phonemic split has taken place, giving us four distinctive phonemes /m, ∫, n, Î/ rather than two. To what extent could we regard the Saramaccan /∫, Î/ as sonorant stops? We will not attempt to answer this question technically here, although examples of implosives with this phonological status are discussed in Botma's work. Thinking back to our four parallels between implosives and the Gbe sonorant stops we can only positively identify one aspect of the Saramaccan implosives at this time that is reminiscent of “sonorant stopness", i.e. their unpairedness. Since Saramaccan has both plain voiceless and voiced phonemes, the voiced implosives are unpaired. Additionally, a restriction to labial and coronal places of articulation is a not uncommon feature of sonorant stops and their nasal instantiations. 7. Further back in history
Historical phonologists of African languages consider Proto-Volta-Congo - the precursor of both Kwa and Bantu language groups - to havepossessed a series of so-called voiced lenes (lax consonants). This is the name assigned by Stewart (1994) to the group of reconstructed
implosives and sonorants. Gbe linguists regard the implosives to have been lost prior to the common ancestral stage of the Gbe languages - Proto-Gbe. We are now faced with fact that the Saramaccan cognates of Fon lexical items have implosives corresponding to the unpaired /b, ∂ / of Fon. Are we to assume that we have following sequence: (5)
The possible historical stages of development of the Saramaccanimplosives i. Î Proto-Volta-Congo 'b 'd = ∫ implosives10 b ∂ Proto-Gbe "unpaired" voiced stops b ∂ Fon "unpaired" voiced stops ∫ Î Saramaccan implosives
Under traditional assumptions this would be unexpected - returning to a phonological starting point after departing from it. Under the assumption that Pre-Gbe and Proto-Gbe situations are instantiations of sonorant stops it becomes much less odd, however. (6)
The possible historical stages of development of the Saramaccan implosives ii. Proto-Volta-Congo sonorant stops [realised as implosives] Proto-Gbe sonorant stops [realised as voiced stops] Fon sonorant stops [realised as voiced stops] Saramaccan sonorant stops (?) [realised as implosives]
Phonologically then, we do have birds of a feather here – somorant stops. Whether these are realized as implosives or as plain voiced stops is reduced to a matter of phonetic realization. This does beg the question, however, whether we should not make a further assumption that therealisation of sonorant stops as implosives has been continuous from Proto-Volta-Congo up to Saramaccan. We illustrate this in (7): (7)
The possible historical stages of development of the Saramaccan implosives iii. Proto-Volta-Congo sonorant stops [realised as implosives] Proto-Gbe sonorant stops [realised as implosives] early Fon: sonorant stops [realised as implosives] Fon sonorant stops [realised as voiced stops] Saramaccan sonorant stops (?) [realised as implosives]
In other words, a hypothesis whereby the realisation of sonorant stops in Fon is as implosives up till at least 1700 would seem to be favoured by the Saramaccan evidence. Note that we are not suggesting that all the Gbe languages would necessarily have retained their implosives for the same length of time.
10
At the present there are languages like Ébrié and Avikam which preserve the alternation between implosives and nasals (Gbéto 2002).
8. Conclusions
In conclusion we would like to claim that we have at least partially achieved our aims. It is possible to make use of the distinction between plain and implosive voiced stops to distinguish strata in the Saramaccan lexicon. It is also fairly clear that Saramaccan developed from an initial state in which it had three stops in the labial/coronal region rather than four. The plain voiced labials presumably only entered Saramaccan later as the result of Sranan loans. The variable representation of English coronal stops can be explained as an allophonic response to the availability of two types of coronal stop in Fon; the lack of anything parallel in the labials can be explained as due to the availability of only one type of labial stop in Fon. The fact that these three labial and coronal stops were available is yet another indication of Gbe/Fon substrate influence in Saramaccan. And finally the fact that Saramaccan has implosives, while the corresponding sounds in all modern forms of Gbe are plain stops can be explained in two ways (9, 10). On the face of it, however, we would suggest that it seems likely that Fon had implosives, and retained them to a fairly late period. This would be the null hypothesis. 9. Consequences for future research
A final remark of some significance requires to be made as a consequence of the stratification of the English-derived lexicon that we have demonstrated exists in Saramaccan. We have identified two strata. These reveal themselves most clearly in respect of initial voiced labial stops. Firstly we have a stratum of words with initial implosive labial stops, which we have identified as the original English-derived element, and secondly we have a stratum of words with initial plain voiced stops, which we have identified as English-derived words borrowed through the intermediary of Sranan. As a consequence the proportions of original English and Portuguese lexical items in Saramaccan are changed by an as yet indeterminate amount. Indeterminate, because other English-derived vocabulary does not so handily provide us with clues of stratification as this group of words does. However, the relative roles in the formation of Saramaccan of English and Portuguese clearly require to be re-evaluated. In fact Smith and Cardoso (2004) have already anticipated this result to a certain extent in terms of their discussion of the true relative proportions of Englishderived and Portuguese-derived verbs in Saramaccan, while Aboh (to appear) discusses the possibility of Portuguese syntactic influence in Saramaccan. Saramaccan appears to be getting more and more mixed the more it is subjected to detailed scrutiny.
Bibliography
Aboh, Enoch, to appear. The left periphery in the Surinamese creoles and Gbe: On the modulayty of substrate transfer. In: Pieter Muysken and Norval Smith (eds.), The trans-atlantic sprachbund. Aboh, Enoch and Norval Smith, in preparation. Migrations and Geo-linguistics in the Eastern Aja-Tado cultural area. In: Pieter Muysken and Norval Smith (eds.), The trans-atlantic sprachbund. Arends, Jacques, 1995. Demographic Factors in the Formation of Sranan. J acques Arends (ed.), The Early Stages of Creolization, pp. 233-85. Amsterdam: John Benjamins Publishing Company. Van der Berg, Margot, to appear. Creole at birth. Utrecht: LOT Netherlands Graduate School of Linguistics. Botma, Bert, 2004. Phonological aspects of nasality: an element-based dependency approach. Utrecht: LOT Netherlands Graduate School of Linguistics. Capo, Hounkpatin, 1991. A comparative phonology of Gbe. Berlin/Garome: Foris Publications/Labo Gbe (Int.). Daeleman, Jan, 1973. Kongo elements in Saramacca Tongo. Journal of African Languages 11: 144. Donicie, Anton and Jan Voorhoeve, 1963. De Saramakaanse woordenschat. Amsterdam: Bureau voor Taalonderzoek in Suriname van de Universiteit van Amsterdam. Gbéto, Flavien, 2002. Les consonnes voisées “implosives” et leurs rôles dans la tonogenese et la nasalité des consonnes dans quelques languages volta-congo. Études Gbe 1: 7-31. Greenberg, J oseph, 1970. Some generalizations concerning glottalic consonants, especially implosives. International Journal of American Linguistics 36: 123-145. Haabo, Vinije, 2000. Fonologie van het Saramaccaans. [Unpublished Ms. University of Leiden.] Migge, Bettina, 1998. Substrate Influence in the Formation of the Surinamese Plantation Creole: A Consideration of Sociohistorical Data and Linguistic Data from Ndyuka and Gbe. [Ph.D. dissertation, Ohio State University.] Price, Richard, 1975. Kikoongo and Saramaccan: a reappraisal. Bijdragen tot de Taal-, Land- en Volkenkunde 131:461-478. Price, Richard, 1976. TheGuiana Maroons: A historical and bibliographical introduction. Baltimore: The Johns Hopkins University Press. Price, Richard, 1983. First-Time: The Historical Vision of an Afro-American People. Baltimore: Johns Hopkins University Press. McWhorter, John, 1997. Towards a new model of creole genesis. New Y ork: Peter Lang. Schumann, C. 1778. Saramaccan Deutsches Wörter-Buch. [Ms., Moravian Brethren, Bambey, Surinam.] Smith, Norval, 1987. The genesis of the creole languages of Surinam. [Unpublished D.Litt. thesis, University of Amsterdam.] Smith, Norval, 2002. The history of the Surinamese creoles II: Origin and differentiation, in Jacques Arends and Eithne Carlin (eds.), Language atlas of Suriname. Leiden: KITLV Press, 131-151.
Smith Norval and Hugo Cardoso, 2004. A new look at thePortuguese element in Saramaccan. Journal of Portuguese Linguistics 3: 115-147. [Special issue on Portuguese creoles edited by Norval Smith.] Smith, Norval, and Vinije Haabo, 2004. Suriname creoles: Phonology. In: Bernd Kortmann and Edgar W. Schneider (eds.), A Handbook of Varieties of English, Vol 2. Berlin: Mouton de Gruyter, 525-566. Stewart, J ohn, 1994. Article review: The comparative phonology of Gbe and its significance for that of Kwa and Volta-Congo. Journal of Africa Languages and Linguistics 15: 175-193. Swartenbroekx, Pierre, 1973. Dictionnaire Kikongo et kituba/français. Bandunu: CEEBA.
