The Evolution of Language: Hints from Creoles and Pidgins

2009. In Language Evolution and the Brain,
ed. by James Minett & William Wang,pp. 1-33.
City University of Hong Kong Press.
The Evolution of Language: Hints
from Creoles and Pidgins
Salikoko S. Mufwene
University of Chicago
1. Introduction
The subtitle of this essay should not be interpreted in the way
suggested by Bickerton (1990, 1995, 2002) or Givón (1998).
Creoles can inform our research on the evolution of language not
because there is anything empirically exceptional or unusual about
the way they emerged (see, e.g., DeGraff 2003, 2005) but because
the multitude of facts that we have learned over the past couple of
decades about their development has drawn our attention to the
kinds of ecological factors that should have informed any sound
genetic linguistics (Mufwene 2001, 2005, 2008). Much of recent
research has shown that these new language varieties are largely
a legacy of 17th and 18th-century European vernaculars spoken
in the colonies around the Atlantic and in the Indian Ocean. They
have reintegrated, on the model of gene recombination in biology
and under the influence of various African substrate languages (and
sometimes with elements imported from these), structural materials
from diverse varieties of their lexifiers1 (see, e.g., Chaudenson 2001,
2003 in the case of French creoles). They also emerged gradually,
illustrating the piecemeal way in which selection resolved competition
in language contact settings where the language of the economically
powerful (among other factors) had selective advantage.
Quite significantly from the point of view of evolution, the
literature on creole continua reminds us that languages as communal
phenomena are constructed from idiolects, that boundaries are
Note: The papers in this unedited edition should not be quoted or cited.

---

2
Language, Evolution and the Brain
artificially imposed between varieties identified as dialects, sociolects,
or languages,2 that they do not evolve uniformly, and that the
conventionalization and normalization of their structures are the
outcomes of the tacit negotiations that take place among their
speakers as they endeavor to communicate successfully. The literature
also informs us that languages have no lives independent of the
communicative activities of their speakers, that population structure
(see below) bears significantly on the differential evolution of a
language, as patterns of social interaction determine which particular
structural variants are more likely to be favored by which particular
speakers, depending on location, age, gender, social class, ethnicity/
race, frequency of interactions,3 and a host of other “ecological”
factors.
In this chapter, I capitalize on some of these factors that bear
on language evolution, especially structural variation and the nature
of the relevant feature pool, the ensuing competition, and factors
bearing on selection from among the variants, on normalization
(as emergence of norms) and on speciation (as emergence of new
language varieties from the same ancestor). I show how they can
inform us indirectly about the phylogenetic evolution of language
in mankind. I submit that nothing about the latter topic can be
explained without factoring in the ecology of the population of
interacting individuals, those aiming at communicating with each
other but not necessarily intending to develop a language.
I argue that human languages are not designed phenomena; the
term “design feature” used since Hockett (1960) to single out those
structural and functional characteristics that distinguish them from
animal means of communication is a convenient misnomer. I submit
that languages are emergent phenomena with their patterns arising
from repeated spontaneous attempts by individuals who share
spatial and social space, and must negotiate terms of coexistence, to
communicate with each other (see also Steels 2000, 2003). Although
invocations of adequate mental and physiological infrastructures
account for humans’ and their hominin ancestors’ ability to produce
and copy, or learn, some communicative strategies, these factors do
not explain how norms and communal languages emerge. Like Wang,
Ke & Minett (2004), I submit that social interaction is an important

---

The Evolution of Language: Hints from Creoles and Pidgins
3
ecological factor that had a major role to play in the evolution of
language, as much in the conventionalization of particular languages
as in their speciation from what Ruhlen (1994) calls “proto-
language” and others have identified as “proto-world.”4 However, it
will be necessary to explain briefly how creoles have led me to the
hypothesis I articulate below and why.
2. The normal evolution of creoles and pidgins
I subscribe to Alleyne’s (1971) position, later elaborated in
Chaudenson (1979, 2001, 2003) and Mufwene (2001, 2005), that
creoles have not evolved from erstwhile pidgins. A close examination
of colonial history and of the geographical distribution of creoles and
pidgins around the world should have disputed the traditional view
a long time ago. The fact that the morphosyntax of incipient pidgins
is simpler than that of creoles, the state of affairs on which linguists
have capitalized, is not sufficient evidence for assuming that creoles
evolved from antecedent pidgins. It simply reveals that pidgins have
evolved in a manner very similar to most Indo-European languages
over the past 3,000 years or so, from a more complex to a simpler
morphosyntax, with fewer and fewer inflections and more and more
periphrastic markers of especially TENSE, ASPECT, and MOOD.
Incipient pidgins represent extreme such cases in accordance with the
ecological conditions of sporadic contacts under which their lexifiers
were learned (see below).
It is not even certain that creoles emerged after the incipient
pidgins from which present-day expanded pidgins (e.g., Nigerian
and Cameroon Pidgin Englishes, and Tok Pisin) have evolved. The
evidence for the putative West African Pidgin English (WAPE)
invoked by Dillard (1972, 1985) to infer, misguidedly, the ultimate
pidgin origins of African American English dates only from the early
18th century (Mufwene 2000). So does the “Guinea Coast Creole
English” invoked by Hancock (1986) to support his Afrogenesis
hypothesis of Atlantic creoles.5 It is not even evident that there was
ever a generalized WAPE spoken in all English trade colonies up to
the 18th century. As pointed out by Huber (1999), some Portuguese

---

4
Language, Evolution and the Brain
variety (pidgin or other) had functioned in the region as a universal
trade language used, at least initially, by all European traders along
the West African coast.6 This is consistent with Ostler’s (2003)
observation that Portugal, which had in fact prevailed as a leading
maritime and trade power until the 17th century (i.e., before the
England, France, and Holland did), had succeeded in imposing its
language as a trade and diplomatic language on the African coast
and in Asia, all the way to Japan. Although this does not necessarily
entail that all later pidgins relexified from the Portuguese Pidgin
that was spoken on the West African coast (pace Thompson 1961
and Whinnom 1965, who also claimed that it must evolved by
relexification from the Mediterranean Lingua Franca), colonial
history supports the possibility that traders from other European
nations would have simply learned the prevailing trade language
of their business. This would have been similar to usage of “trade
Latin” in the Hanseatic League during the Middle Ages.
It is quite telling that the term pidgin did not emerge until either
the late 18th century (Bolton 2000) or the early 19th century (Baker
& Mühlhäusler 1990), soon after the English established a trade
colony in Canton, about a century after they had established trade
forts on the West coast of Africa in early 17th century. The term
emerged two centuries later than the first written attestations of the
term creole in Spanish and Portuguese, in the late 16th century, for
non-indigenous people born in the colonies. That was also a century
later than the first attestation, in the late 17th century, of the term for
a variety of Portuguese spoken in the present-day Casamance region
of Senegal (see Mufwene 1997 and the references cited therein).
Bolton (2002) attributes the emergence of Pidgin English to
the intensification of trade with China and the ensuing shortage
of interpreters, who had initially been trained by missionaries as
proselytes and had spoken closer approximations of native English.
As more and more people traded directly with the Europeans in
a language they were not familiar with, it was restructured to the
pidgin level, i.e., a contact-based language variety whose structures
where reduced to the minimum.
As explained in Mufwene (2005), this is an explanation that can
apply to Africa too, about which colonial history reveals a significant

---

The Evolution of Language: Hints from Creoles and Pidgins
5
role of interpreters during the early Euro-African contacts. During
the earliest explorations, the Portuguese had usually taken some
Africans (often from the royal family) with them back to Portugal,
leaving some of their crew behind in earnest. They would return
with them a few years later and these L2-Portuguese speakers would
later function as interpreters. During the later colonial ventures, the
Europeans would recruit Africans from the earlier colonies of the
northwestern coast as auxiliaries (and therefore interpreters, in some
naïve way of course) in the colonization of the southern parts.
In Hawaii, American missionaries first taught English to
members of the royal family, who would later become instrumental as
interpreters during the colonization and the economic exploitation of
the islands. In all these contact settings, pidgins would emerge when
contacts between the non-Europeans and Europeans (or Americans
in the case of Hawaii) increased, there were too few interpreters, and
more and more non-Europeans assumed they had learned enough
English to communicate in the European languages. As the trade
expanded, more and more interactions among non-Europeans in the
European languages would lead to their extensive divergence away
from the originals and the erosion of their morphosyntaxes.
Klein (2003) also reminds us that Europeans, especially the
Portuguese and Spaniards, had enslaved Africans as domestics
in Europe before they imported slaves to the New World. Some
of these were probably used as “interpreters” between masters
and bozal slaves in the New World, although the nature of initial
communication between the Europeans and their future interpreters
remains very mysterious. The question arises in fact about the first
homesteads in which the Europeans lived together with their African
captives. What is certain is that the contact ecologies were not those
of sporadic contacts and minimal interaction typically associated
with the development of pidgins. The very limited rate of slave
importations during the homestead phase, due to shortage of capital,
created a contact situation where the resident populations grew
more by birth than by immigration. As pointed out by Chaudenson
(2001, 2003), Creole children of the homestead phase most likely
did not speak creole varieties, which would develop during the
later, plantation society phase. White and Black Creole children of

---

6
Language, Evolution and the Brain
the homestead phase spoke native koiné varieties of the European
vernaculars (i.e., new varieties combining elements from the different
dialects in contact), variable as the latter were from one speaker to
another.
Overall, if creoles did not develop earlier than pidgins lexified
by European languages, they developed concurrently, but certainly
not later. They also developed in settings where pidgins did not, and
probably could not, emerge. In the Caribbean, pidgins identified
as baragouins are reported to have emerged in the contacts of
Europeans with Native Americans but not with Africans (Prudent
1980, Wylie 1995, Chaudenson 1992, 2001). Overall, both creoles
and pidgins evolved by basilectalization, from closer approximations
of the European colonial vernaculars to varieties more and more
divergent from the originals. As I observe in Mufwene (2001, 2005),
no language-restructuring mechanisms have been identified that
are peculiar to either creoles or pidgins. As also argued by many
creolists since Arends (1989), the evolution process seems to have
been gradual. The evidence of creole continua suggests in fact that
they have always been internally variable, reminding us that the term
idiolect, so central but still under-exploited in modern linguistics,
denotes individually variable varieties. What is attested in creole
speech communities is just a more evident case of the inter-idiolectal
variation that can be observed in any language community.
I articulate in Mufwene (2008) the most important reasons why
I do not share Bickerton’s (1990f) position that pidgins, from which
he assumes creoles have evolved, can especially inform us about his
putative “protolanguage,” that presumably critical transition from
pre-language to modern languages in human phylogeny. The first is
that unlike our hominin ancestors who developed that protolanguage
(whose empirical validity I question below) the modern humans
who developed pidgins, especially during trade contacts between
Europeans and non-Europeans from the 16th to the 19th century,
were speakers of modern languages. This is an important difference
from the late Homo erectus or the early Homo sapiens, who
presumably started from something quite different from a modern
language. The creators of pidgins were therefore able to draw on
fundamental properties of Universal Grammar (presumably the

---

The Evolution of Language: Hints from Creoles and Pidgins
7
biological endowment for language) and/or on particular resources
in their respective languages in modifying the relevant lexifier –
unintentionally, in their attempts to communicate – to the minimal
structures associated with pidgins. Note that pidgins still have lexical
and syntactic categories (such as NOUN, NOUN PHRASE, VERB,
AND VERB PHRASE) and, despite variation among them, individual
speakers systematically adopt SVO or SOV syntactic structures,
although they do not produce complex structures with especially
adverbial subordinate clauses.
Although Bickerton (1981f) argues that pidgins lack (complex)
syntax, what he has really shown is that the attested syntax is
inter-idiolectally variable. This simply means that the incipient
pidgins have not conventionalized their strucutures to the extent
that expanded pidgins have. In other words, the grammars used
individually by their speakers, which are otherwise internally
systematic, have not converged in ways observable in language
varieties of stable communities, where mutual accommodations have
made the idiolects more similar to each other.
With regard to variation, the difference between pidgins (or
creoles) and other languages is just a matter of degree of convergence
among speakers. Otherwise, reality shows that no two speakers
behave in identical ways; therefore, they probably do not use
identical grammars. Aside from the fact that speakers are no more
identical mentally than they are physiologically, the main reason for
interidiolectal variation is that each speaker has had a unique history
of linguistic interactions with members of their community and has
been exposed to a different subset of primary linguistic data from
which they could work out their respective individual grammars. It
is mutual accommodations through regular interactions with each
other that over time make their systems similar to each other or
perhaps just enable them to interpret each other’s outputs. This is
what I mean by convergence of idiolects, which Steels (2000, 2003)
calls system “coherence.” As he correctly observes, the convergence
is necessary in order for inter-individual communication to be
successful.
We must note that, contrary to Saussure’s assumption about the
conventional nature of language (also preserved in much of today’s

---

8
Language, Evolution and the Brain
formal linguistics), members of a speech community communicate
with each other not because they possess identical systems of their
language (which is only a construct) but because they are able to
interpret each other’s utterances successfully (Mufwene 1989).
Linguists should not brush under the rug the fact that members
of the same speech community sometimes misinterpret each other
even under conditions involving no noise or fatigue, a condition
that is corroborated by Wang et al’s (2004) modeling of language
evolution. It is like two computers communicating with each other
but sometimes experiencing translation problems.7
Regarding the simplicity of pidgin systems, one thing they reveal,
just like child language does for that matter, is that ontogenetically
human languages develop from simple to complex structures.
Phylogenetically, human language also appears to have evolved
from rudimentary to complex structures (see below), although the
histories of modern languages over the past 5,000 years or so also
shows evolutions from more complex to simpler morphosyntax. In
the case of children, the structural simplicity is apparently correlated
with cognitive maturation. The story for pidgins, however, is that of
initial attempts by individuals or groups already speaking mutually
unintelligible complex languages to establish communication among
themselves. It involves establishing communicative correspondences
between the languages the relevant individuals have been speaking
and what they are targeting jointly. It also involves knowing what
part of information packaging can be done away with and what
cannot.
What would be particularly informative in this case, in relation
to the phylogenetic evolution of language among humans, is
knowing how gradually a communal norm (some sort of social
consensus, so to speak) emerges out of the multitude communicative
strategies initiated individually by members of the interacting
community (see also Wang et al. 2005). Where the emergent
norm is intra-communally variable and where more than one
communal norm emerge, it must also be informative to know what
particular ecological factors bear on this variation. These include
POPULATION STRUCTURE, which determines who interacts with
whom and how frequently, who is most likely to accommodate

---

The Evolution of Language: Hints from Creoles and Pidgins
9
whom, the setting of the social contact and what motivates the
contact, the time period of the contact in the relevant populations’
history, the respective demographic strengths of the populations in
contact, and, where a significant demographic disproportion obtains,
which proportion or segment of the majority population interacts
with the minority and powerful population.
As shown in Chaudenson (1992, 2001, 2003) and Mufwene
(2001, 2005, 2008), these factors have borne as much on the
evolution of European languages into creoles and pidgins as in other
cases of language evolution. In this chapter, I focus on creoles and
pidgins not because their emergence has been exceptional or unusual
but simply because they have prompted us to do better genetic
linguistics, paying closer attention to ecological factors that influence
language birth and death, beyond the structural changes that have
traditionally interested historical linguists. We could thus also pay
attention to founder effects, generative entrenchment, periodization,
scaffolding, and patterns of population growth.
3. Hints from creoles about the evolution of
language
The development of creoles and pidgins reminds us of the fact that
languages are really not transmitted, at least not in the way that
genes are transmitted and inherited in a biological species, in which
offspring inherit passively what is transmitted from their parents.
As pointed out in Mufwene (2001), languages are learned by
unguided and ecology-specific inferences from the performances of
current speakers or signers. (See also Steels 2000, 2003 for a similar
position.) To make things more complicated, there are for every
learner typically multiple model speakers or signers, who influence
the learning process in different ways. Under ecological pressures
that vary from one learner and one setting to another, each learner
selects from various models variants that can cumulate to become
part of his/her productive idiolect. One must also bear in mind that
there is no perfect replication in the learning process, which is itself
subject to mental, perceptual and other physiological peculiarities

---

10
Language, Evolution and the Brain
of the learner. Like that of other languages, the emergence of creoles
and pidgins is largely the result of imperfect replication in settings
where contact with other languages increases the extent to which
the outcome of the appropriation of the target language is ultimately
more divergent at the communal level, particularly when the majority
of the learning population is adult.
Equally noteworthy about the emergence of creoles and pidgins
is the significance of periodization. Not every member of the relevant
population appropriates the target at the same time or from the same
speakers. The nature of the target changes from one generation to
the next. In the case of creoles and pidgins, the gradual restructuring
of the target away from the metropolitan norm led to situations
where later learners had as models speakers whose knowledge was
already divergent from that of the earlier generations of speakers.
Many of the models were no longer native speakers and spoke
varieties influenced by their own substrate languages. Repetition of
this process over generations made the emergent variety more and
more divergent, especially in settings where the learning process
was exclusively naturalistic, the proportion of native speakers of
the European variety kept dwindling, and either the non-European
population was growing more by importation of adults than by birth
in the case of creoles or the language kept spreading locally, among
non-Europeans needing it as a lingua franca. Reiteration of this
individual imperfect language appropriation process within a novel
population in a setting exogenous to the target language accounts for
the usually dramatic way in which creoles and pidgins diverge from
their lexifiers. Although many of their features are selections from
17th and 18th-century colonial European vernaculars, they have been
recombined into the new systems in novel ways, often influenced by
substrate languages.
However, we should focus now on how the norms of these new
varieties emerged. Although, we may not ever be able to answer this
question fully, we now know about how to conceptualize it and think
of elements that should help us answer it. As in any other population,
the individuals who collectively and cumulatively produced
creoles and pidgins never did this according to explicit “rules of
engagement” or a particular script that they all followed uniformly

---