Professional Documents
Culture Documents
Case Closed Testing The Feature Pool Hyp PDF
Case Closed Testing The Feature Pool Hyp PDF
Case closed?
Testing the feature pool hypothesis
John McWhorter
Columbia University
1. Introduction
One of the prime attractions of scientific inquiry is the discovery of the counter-
intuitive. As such, the idea that creole genesis entails simply a mixture between
languages, with simplification playing no significant role, is intriguing. The impli-
cation is that the traditional sense of what a ‘creole’ is has been an illusion.
Thus it is not surprising that the ‘feature pool’ analysis of creole genesis, most
often supported by Mufwene (2001 and other works), adopted of late by Enoch
Aboh (e.g. 2009) and Umberto Ansaldo (e.g. 2009), and regularly cited by most
other creolists, is so readily accepted. It is an elegant idea, and even more attractive
in paralleling biological reproduction.
However, the ‘feature pool’ hypothesis does not, in the end, account for the
body of data that it addresses. An opposing idea is that creole genesis involves not
only mixture, but also considerable simplification of grammars, to a degree far
beyond what grammars undergo naturally, under contact or not.
Conventional wisdom in creole studies would appear to be that this hypoth-
esis, which I unhesitantly term Creole Exceptionalism, has been refuted. I would
like to explain why I do not believe this to be the case.
The ‘feature pool’ hypothesis (henceforth FP) fares beautifully as an idea, but
much less so as a mechanism. Creole studies would benefit from a careful evalua-
tion of this paradigm before judging it correct.
It is certainly correct when creolists cite the FP to show that creoles acquired most
of their features from their source languages, along the lines of, for example, Siegel
(2008). However, the basic hybridity of creoles is, in itself, obvious to all. Spelling
Journal of Pidgin and Creole Languages 27:1 (2012), 171–182. doi 10.1075/jpcl.27.1.07mcw
issn – / e-issn – © John Benjamins Publishing Company
172 John McWhorter
→ ←
→ ←
The FP certainly does not stipulate that the mixture occurs randomly. The idea is
that a competition between features is decided on the basis of prominence, trans-
parency, frequency, salience, and typological similarity. However, these features
determine language contact in general, as conclusively taxonomized in Thoma-
son & Kaufman’s (1988) foundational thesis, and further elucidated in summary
language contact sources such as Winford (2002) and Matras (2009). This means
that FP adherents are arguing that qualitatively, nothing significant distinguishes
the birth of Sranan Creole from that of Romanian. In both cases, languages came
together, tout court.
Counterintuitive and contradicting several decades of research on language
change and language contact, this is a radical proposition. As such, it requires
rigorous demonstration. Has this happened?
3. Argumentation
Aboh & Ansaldo characterize the absence of a plural affix in Surinam creoles to
plurality being ‘semantically vacuous’, an analysis that would surprise most lin-
guists, including those who, as Plag notes, classify the plural as inherent morphol-
ogy, carrying meaning in contrast to contextual morphology (e.g. concord) which
does not.
Plag notes these kinds of problems throughout the article — as well as inac-
curacies in data — and the conclusion is inescapable that Aboh & Ansaldo (2007)
does not make a successful case for the FP. A search for more successful argu-
mentation will certainly lead to the work of FP’s founder, Mufwene. Here, how-
ever, while we have eminently useful statement of the FP’s assumptions, there is no
more actual demonstration with data than in the previously cited sources.
For example, Mufwene (2009: 375) treats the usage of for as both a comple-
mentizer and modal in many creoles as what he terms a ‘showpiece’ of FP argu-
mentation. That argument, typical in form of Mufwene’s FP work in general (e.g.
Mufwene 2001), is enlightening in itself — but in the post-Bickertonian sense I
alluded to above. As an argument for FP, this kind of argument is fundamentally
incomplete. All agree that mixture plays a major role in creole genesis. Our ques-
tion is whether simplification does not. Demonstrations of the mixed origins of a
few items cannot answer that question — nor even could dozens of such demon-
strations, any more than dry patches of land between puddles could demonstrate
that it had not rained.
4. Applicability
What happens, for example, when we try to apply the FP to even a creole grammar
as a whole? Palenquero Creole Spanish is a useful case. It was born from a feature
pool provided by just two languages, Spanish and Kikongo (cf. Schwegler 1996).
Kikongo is not analytic as the Kwa language contributors to so many creoles are.
Rather, Kikongo and Spanish have quite a bit in common, both richly inflected
and SVO:
Kikongo (Bentley 1887: 526) (C8P = noun class 8 plural):
(1) O ma-tadi ma-ma ma-mpembe ma-mpwena
AUG C8P-stone C8P-DEM C8P-white C8P-big
i ma-u ma-ma tw-a-mw-ene.
COP C8P-that C8P-DEM we-them-see-PERF.
Spanish:
(2) Est-a-s piedr-a-s grand-e-s y blanc-a-s
DEM-FEM-PL stone-FEM-PL big-FEM-PL and white-FEM-PL
if not as affixes, then as free morphemes along the lines of noun classifiers in so
many languages worldwide?
Crucial: Palenquero does retain one Kikongo noun class marker, ma-, as its
plural-marking clitic. But this only leaves more questions that the FP is not de-
signed to answer. First, why the restriction to the definite and the animate? I pro-
pose that this traces to simplification: diachronically, animate definite is typically
where plural marking begins in a language that did not have it before (McWhorter
2006: 171). Why the plural but not gender? Plag, as well as Kihm (2003) and myself
(2011), would refer to creolization preserving inherent inflection (e.g. number)
but eliminating the contextual (e.g.concord). As of now, FP adherents can refer
only to their sense of what is ‘semantically vacuous’ — which under their analysis
would include rather than exclude the plural marker.
Moreover, even if they were to add the contextual/inherent distinction as a
software ‘update’, it would be of no help in explaining why Palenquero has no defi-
nite determiner or perfect tense.
Nativization? FP adherents might object that the framework does account for di-
vergences from feature pool sources, referring to arguments by Veenstra (2006)
and Kouwenberg (2009) that such divergences are due to later nativization by chil-
dren. These arguments, solid as genesis arguments in general, cannot help us here.
No account of nativization encompasses the elimination of concord, articles and
Differential Object Marking, the narrowing of the scope of number marking, and
eliminating grades of past tense distinction. Nor, obviously, do these losses submit
to analysis as subsequent grammaticalizations, along the lines of Mufwene (1996).
Palenquero’s divergences from Spanish and Kikongo simply do not fit any concep-
tion of the FP.
Other languages in the pool? The only thing that could possibly help an FP account
is that Palenquero was likely seeded by a form of Portuguese pidgin or creole,
relexified by Spanish (Schwegler 1998: 229–30); cf. a famous comment by Father
Alonso Sandoval that at least some of the first slaves spoke a ‘broken’ variety called
‘the language of Sao Tomé’. Sao Tomense Creole Portuguese is quite similar to
Palenquero in lacking the five features above, and thus one might suppose that
Palenquero incorporated ‘features’ from some form of this Sao Tomé variety as
well as from Spanish and Kikongo.
It must be clear that the Portuguese element in Palenquero consists of a small
collection of morphemes; it is unlikely that the Portuguese pidgin/creole was a
robust presence in the genesis context. However, even if it was, and the Palen-
quero feature pool was thus composed of Spanish, Kikongo, and a Sao Tomé
Portuguese pidgin/creole, the FP account again founders. For one, if this ‘broken’
Sao Tomé variety was a pidgin, then given Mufwene’s frequent dissociation of
pidgin-level varieties from creole genesis, under FP this variety was irrelevant in
any case.
However, even supposing it was not, this Sao Tomé pidgin/creole would have
been just one among three contributors to the feature pool, and the other two were
more typologically similar to one another than to it. Why, then, would features
from Sao Tomense be processed as more prominent, transparent, frequent, and
salient — or more typologically compatible with the creators’ native Kikongo? A
conception of a competition between a Romance language, a Bantu one, and a
Sao Tomense Portuguese pidgin or creole in which the result is Palenquero could
never be judged predictive or systematic. Why was the Sao Tomé variety a kind of
‘strange attractor’ if creole genesis is simply a mixture of features?
The implications must be clear. It is starkly obvious that any empirical account
of Palenquero’s birth from a pool of features must pay as much attention to what
Palenquero did not incorporate from its feature pool as to what it did. This means,
quite simply, a creole genesis model with an ‘exceptionalist’ component. That is,
my argument is not that the FP is mistaken, but that it is incomplete.
Significantly, Palenquero is not a fluke but a norm, paralleled by Guinea-Bis-
sau Creole Portuguese’s analyticity despite its emergence in a feature pool of Por-
tuguese and well-inflected West Atlantic languages, and Nubi Creole Arabic’s only
faint grammatical resemblance to either Arabic or Nilo-Saharan languages, and, in
fact, most if not all creole languages. If it is true that as Aboh has put it ‘the claim
that creoles are simplified versions of their sources is a fallacy’, FP adherents have
yet to demonstrate it.
Something else renders the FP a highly preliminary proposal. Namely, its adher-
ents have yet to engage substantially with counter-arguments.
An example comes from Saramaccan, whose determining substrate language
was Fongbe. In McWhorter (2004), I noted nine Fongbe features that surpass
Saramaccan equivalents in complexity as defined in the Introduction, even when
the Saramaccan one is modelled on the Fongbe one, and that it seems impossible
to compose a like list of Saramaccan features more complex than Fongbe equiva-
lents. McWhorter (2001) notes several features that creoles regularly fail to incor-
porate from their feature pools, most of which would not fall under the contextual
inflection ‘update’ mentioned above, such as Central Philippines languages’ focus
marking in the Philippines Creole Spanishes (expressed in part with a free mor-
pheme), and evidential markers in creolized Chinook Jargon, whose source lan-
guages were rich with them. Short of an FP account, we must conclude that Creole
Exceptionalism explains these facts better.
Supporting that verdict is that FP argumentation tends to evade the indica-
tions of simplification in creole genesis. ‘We must wonder’, Mufwene (2009: 393–4)
briefly notes in a footnote followed by a brief passage, why creoles tend to eliminate
redundancy, proposing that the loss is due to unspecified ‘cross-systemic or intra-
systemic interactional pressures’. But this vastly minimizes the challenge: there is
a great deal more than redundancy that creoles do not choose from their feature
pool. Similarly, Ansaldo & Nordhoff (2009: 354) mention — again, in a footnote
— the absence of inflections in Chinook Jargon (which, as noted above, creolized)
as requiring ‘alternative explanation’. However, the Chinook Jargon problem is but
an isolated symptom, of a profound incompatibility between creole language data
as a whole and the FP paradigm.
Adherents of a framework with promise eagerly take on challenging data in
detail. As long as this is anything less than standard procedure among adherents
of FP, the hypothesis can qualify only as tentative. (Kouwenberg 2010 on Parkvall
2008, although I do not support its conclusions, is an exception — that proves the
rule.) Surpassing that tentaive status will require detailed response not only to
arguments here and elsewhere, but to new ones bolstering Parkvall (2008), such as
Bakker, Daval-Markussen, Parkvall, & Plag (2011).
7. Unintended consequences
Finally, I suggest that creolists consider two implications of the FP that are not
deliberate, but demand attention nonetheless.
First, the FP suggests a watchcry: The proper goal of the study of creole lan-
guages is to demonstrate that there is nothing especially interesting about them. I
submit that this is a distinctly peculiar and uninspiring mission. Under the FP,
upon what linguistic grounds do we treat these languages we have called ‘creoles’
as the subject of a journal, conferences, and anthology volumes? Is creole studies,
as Parkvall (2001: 150) asked, ‘cutting off the branch it is sitting on?’
Of course, the fact that a hypothesis is less ‘dramatic’ does not render it false.
Chomskyan syntacticians, presented with the radically streamlined machinery of
Peter Culicover and Ray Jackendoff ’s ‘Simpler Syntax’ paradigm, have objected
that ‘if this approach is right, then syntax isn’t interesting’ (Culicover & Jackendoff
2005: 540). Yet all indications are that Simpler Syntax is closer to truth than the
ingrown, semiempirical mechanisms of Minimalism. In the same way, the fact that
FP eliminates creoles as a legitimate subject of inquiry cannot, in itself, prejudice
us against it. However, given that the FP is a call to eliminate a linguistic subfield,
we will presumably accept it only upon rigorous assessment of its logical validity.
Second, the FP entails, inherently, an insult to many creolist scholars.
The insult is, ironically, the product of what is considered an enlightened re-
sponse. In examining solid descriptions of older languages such as Manambu of
Papua New Guinea (Aikhenvald’s 2008), Jarawara of the Amazon (Dixon 2004),
or even analytic ones like Sino-Tibetan Lahu (Matisoff 1973) one cannot help but
perceive much more overspecification, structural elaboration, and irregularity
than has been described in any creole language. Aboh & Smith (2009) and Muf-
wene’s (2009: 387–90) warnings about the pitfalls of assessing languages according
to ‘bit complexity’ are well taken. However, they do not belie the stark contrast to
which I refer, once referred to by Bickerton (1988: 318) as ‘elusive’ but obvious.
This contrast cannot help but strike anyone, regardless of their views about FP
or Creole Exceptionalism. The question is not whether that difference exists, but
what we attribute it to.
FP encourages a choice here that creates the insult in question. Namely, we are
to suppose that creolists have yet to describe creoles well enough to show that they
are as complex as older languages (e.g. Arends 2001). However, this is a trenchant
and dismissive judgment of the many creolists who have granted us detailed gram-
mars of creole languages and articles about aspects of their structural makeup.
Short of embracing this uncivility, we must allow that these scholars’ descriptions
have been accurate.
This, however, requires acknowledging that creole genesis is not only a matter
of selections from a feature pool, but also a robustly simplificatory process, result-
ing in languages less accreted with needless elaboration than older ones.
8. Conclusion
FP, in beginning with something as obvious as that creoles are born amidst mix-
ture, can seem so reasonable in itself that the slippage between it and the facts can
be counterintuitive. However, that slippage is vast. A seemingly unexceptionable
statement such as this one (Mufwene 2009: 386):
‘The extent of morphological complexity (in terms of range of distinctions) re-
tained by a ‘contact language’ largely reflects the morphological structures of the
target language and the particular languages that it came in contact with’.
References
Aboh, Enoch O. 2009. Competition and selection: that’s all! In Enoch O. Aboh & Norval Smith,
Complex processes in new languages, 317–344. Amsterdam: John Benjamins.
Aboh, Enoch O. & Umberto Ansaldo. 2007. The role of typology in language creation. In Um-
berto Ansaldo, Stephen Matthews, & Lisa Lim (eds.). Deconstructing creole, 39–66. Amster-
dam: John Benjamins.
Aboh, Enoch O. & Norval Smith. 2009. Complex processes in new languages. Amsterdam: John
Benjamins.
Aboh, Enoch O. & Norval Smith. 2009. Simplicity, simplification, complexity and complexi-
fication. In Enoch O. Aboh & Norval Smith, Complex processes in new languages, 1–25.
Amsterdam: John Benjamins.
Aikhenvald, Alexandra Y. 2008. The Manambu language of East Sepik, Papua New Guinea. Ox-
ford: OUP.
Ansaldo, Umberto & Sebastian Nordhoff. 2009. Complexity and the age of languages. In Enoch
O. Aboh & Norval Smith (eds.), Complex processes in new languages, Creole Language Li-
brary. Amsterdam: John Benjamins.
Arends, Jacques. 2001. Simple grammars, complex languages. Linguistic Typology 5(2/3).
180–182.
Bakker, Peter, Aymeric Daval-Markussen, Mikael Parkvall, & Ingo Plag. 2011. Creoles are
typologically distinct from non-creoles. Journal of Pidgin and Creole Languages 26(1). 5–42.
Bentley, W. Holman. 1887. Dictionary and grammar of the Kikongo language. London: Trübner
& Co.
Bickerton, Derek. 1981. Roots of language. Ann Arbor: Karoma.
Bickerton, Derek. 1988. Creole languages and the bioprogram. In Frederick Newmeyer (ed.),
Linguistics: the cambridge survey (Vol. II), 268–284. Cambridge: CUP.
Culicover, Peter W. & Ray Jackendoff. 2005. Simpler syntax. Oxford: OUP.
Dixon, R.W.M. 2004. The Jarawara language of Southern Amazonia. Oxford: OUP.
Kihm, Alain. 2003. Inflectional categories in creole languages. In Ingo Plag (ed.), Phonology and
morphology of creole languages, 333–363. Tübingen: Niemeyer.
Kouwenberg, Silvia. The invisible hand in creole genesis: reanalysis in the formation of Berbice
Dutch. In Enoch O. Aboh & Norval Smith, Complex processes in new languages, 115–158.
Amsterdam: John Benjamins.
Kouwenberg, Silvia. 2010. Creole studies and linguistic typology: Part 1. Journal of Pidgin and
Creole Languages 25. 173–86.
Matisoff, James A. 1973. The grammar of Lahu. Berkeley: University of California Publications.
Matras, Yaron. 2009. Language contact. Cambridge: CUP.
McWhorter, John H. 2001. The rest of the story: restoring pidginization to creole genesis theory.
Journal of Pidgin and Creole Languages 17:1–48.
McWhorter, John H. 2004. Saramaccan and Haitian as young grammars: the pitfalls of syntac-
tocentrismin creole genesis research. Journal of Pidgin and Creole Languages 19. 77–138.
McWhorter, John H. 2006. What the creolist learns from Cantonese and Kabardian (Review
article of Phonology and Morphology in Creole Languages, ed. by Ingo Plag). Diachronica
23: 143–184.
McWhorter, John H. 2007. Language interrupted: signs of non-native acquisition in standard lan-
guage grammars. Oxford: OUP.
McWhorter, John H. 2011. Tying Up Loose Ends: The creole prototype After All. Diachronica
28. 82–117.
Mufwene, Salikoko S. 1996. Creolization and grammaticization: what creolistics could contrib-
ute to research on grammaticization. In Philip Baker & Anand Syea (eds.), Changing mean-
ing, changing functions, 5–28. London: University of Westminster.
Mufwene, Salikoko S. 2001. The ecology of language evolution. Cambridge: CUP.
Mufwene, Salikoko S. 2009. Restructuring, hybridization, and complexity in language evolution.
In Enoch O. Aboh & Norval Smith, Complex processes in new languages, 367–400. Amster-
dam: John Benjamins.
Parkvall, Mikael. 2001. Creolistics and the quest for creoleness: a reply to Claire Lefebvre. Jour-
nal of Pidgin and Creole Languages 16. 147–151.
Parkvall, Mikael. 2008. The simplicity of creoles in a cross-linguistic perspective. In Matti Mi-
estamo, Kaius Sinnemäki, & Fred Karlsson (eds.), Language complexity: typology, contact,
change, 265–285. Amsterdam: John Benjamins.
Plag, Ingo. 2008a. Creoles as interlanguages: Inflectional morphology. Journal of Pidgin and Cre-
ole Languages 23. 114–135.
Plag, Ingo. 2008b. Creoles as interlanguages: Syntactic structures. Journal of Pidgin and Creole
Languages 23. 307–328
Plag, Ingo. 2011. Creolization and admixture: Typology, feature pools, and second language
acquisition. Journal of Pidgin and Creole Languages. 89–110.
Schwegler, Armin. 1996. ‘Chi ma nkongo’: lengua y rito ancestrales en El Palenque de San Basilio
(Colombia). Frankfurt: Vervuert.
Schwegler, Armin. 1998. El palenquero. In Matthias Perl & Armin Schwegler (eds.), América
negra: panorámica actual de los estudios lingüísticos sobre variedades criollas y afrohispanas,
219–291. Frankfurt: Vervuert.
Schwegler, Armin & Kate Green. 2007. Palenquero (Creole Spanish). In John Holm & Peter
Patrick (eds), Comparative creole syntax, 273–306. London: Battlebridge.
Siegel, Jeff. 2008. The emergence of pidgin and creole languages. Oxford: OUP.
Thomason, Sarah J. & Terence Kaufman. 1988. Language contact, creolization, and genetic lin-
guistics. Berkeley: University of California Press.
Veenstra, Tonjes. 2006. Modeling creole genesis: headedness in morphology. In Ana Deumert
& Stephanie Durrleman (eds.), Structure and variation in language contact, 61–83. Amster-
dam: John Benjamins.
Winford, Donald. 2002. An introduction to contact linguistics. Malden MA: Blackwell.
Author’s address
John McWhorter
Columbia University
Slavic Department
708 Hamilton Hall
New York, NY 10027
jhmcw5@yahoo.com