R E V I E W : N E U R O S C I E N C E
The Faculty of Language: What Is It, Who Has
It, and How Did It Evolve?
Marc D. Hauser,1
* Noam Chomsky,2
W. Tecumseh Fitch1
We argue that an understanding of the faculty of language requires substantial
interdisciplinary cooperation. We suggest how current developments in linguistics can
be profitably wedded to work in evolutionary biology, anthropology, psychology, and
neuroscience. We submit that a distinction should be made between the faculty of
language in the broad sense (FLB) and in the narrow sense (FLN). FLB includes a
sensory-motor system, a conceptual-intentional system, and the computational
mechanisms for recursion, providing the capacity to generate an infinite range of
expressions from a finite set of elements. We hypothesize that FLN only includes
recursion and is the only uniquely human component of the faculty of language. We
further argue that FLN may have evolved for reasons other than language, hence
comparative studies might look for evidence of such computations outside of the
domain of communication (for example, number, navigation, and social relations).
I
f a martian graced our planet, it would be
struck by one remarkable similarity among
Earth’s living creatures and a key difference.
Concerning similarity, it would note that all
living things are designed
on the basis of
highly conserved developmental
systems
that read an (almost)
universal language encoded
in DNA base
pairs. As such, life is
arranged hierarchically
with a foundation
of discrete, unblendable
units (codons, and,
for the most part,
genes) capable of combining
to create increasingly
complex and virtually
limitless varieties
of both species and individual
organisms. In
contrast, it would notice
the absence of a universal
code of communication
(Fig. 1).
If our martian naturalist
were meticulous,
it might note
that the faculty mediating
human communication appears remarkably
different from that of other living creatures;
it might further note that the human
faculty of language appears to be organized
like the genetic code—hierarchical, generative,
recursive, and virtually limitless with
respect to its scope of expression. With these
pieces in hand, this martian might begin to
wonder how the genetic code changed in such
a way as to generate a vast number of mutually
incomprehensible communication systems
across species while maintaining clarity
of comprehension within a given species. The
martian would have stumbled onto some of
the essential problems surrounding the
question of language evolution, and of how
humans acquired the faculty of language.
In exploring the problem of language evolution,
it is important to distinguish between
questions concerning language as a communicative
system and questions concerning the
computations underlying this system, such as
those underlying recursion. As we argue below,
many acrimonious debates in this field
have been launched by a failure to distinguish
between these problems. According to one
view (1), questions concerning abstract computational
mechanisms are distinct from
those concerning communication, the latter
targeted at problems at the interface between
abstract computation and both sensory-motor
and conceptual-intentional interfaces. This
view should not, of course, be taken as a
claim against a relationship between computation
and communication. It is possible, as
we discuss below, that key computational
capacities evolved for reasons other than
communication but, after they proved to have
utility in communication, were altered because
of constraints imposed at both the periphery
(e.g., what we can hear and say or see
and sign, the rapidity with which the auditory
cortex can process rapid temporal and spec-
1
Department of Psychology, Harvard University,
Cambridge, MA 02138, USA. 2
Department of Linguistics
and Philosophy, Massachusetts Institute of Technology,
Cambridge, MA 02138, USA.
*To whom correspondence should be addressed. Email:
mdhauser@wjh.harvard.edu
Fig. 1. The animal kingdom has been designed on the basis of highly conserved developmental systems that read an almost
universal language coded in DNA base pairs. This system is shown on the left in terms of a phylogenetic tree. In contrast, animals
lack a common universal code of communication, indicated on the right by unconnected animal groups. [Illustration: John Yanson]
S C I E N C E ’ S C O M P A S S ● R E V I E W
www.sciencemag.org SCIENCE VOL 298 22 NOVEMBER 2002 1569
tral changes) and more central levels (e.g.,
conceptual and cognitive structures, pragmatics,
memory limitations).
At least three theoretical issues cross-cut
the debate on language evolution. One of the
oldest problems among theorists is the
“shared versus unique” distinction. Most current
commentators agree that, although bees
dance, birds sing, and chimpanzees grunt,
these systems of communication differ qualitatively
from human language. In particular,
animal communication systems lack the rich
expressive and open-ended power of human
language (based on humans’ capacity for recursion).
The evolutionary puzzle, therefore,
lies in working out how we got from there to
here, given this apparent discontinuity. A second
issue revolves around whether the evolution
of language was gradual versus saltational;
this differs from the first issue because
a qualitative discontinuity between extant
species could have evolved gradually, involving
no discontinuities during human evolution.
Finally, the “continuity versus exaptation”
issue revolves around the problem of
whether human language evolved by gradual
extension of preexisting communication systems,
or whether important aspects of language
have been exapted away from their
previous adaptive function (e.g., spatial or
numerical reasoning, Machiavellian social
scheming, tool-making).
Researchers have adopted extreme or intermediate
positions regarding these basically
independent questions, leading to a wide
variety of divergent viewpoints on the evolution
of language in the current literature.
There is, however, an emerging consensus
that, although humans and animals share a
diversity of important computational and
perceptual resources, there has been substantial
evolutionary remodeling since we
diverged from a common ancestor some 6
million years ago. The empirical challenge
is to determine what was inherited unchanged
from this common ancestor, what
has been subjected to minor modifications,
and what (if anything) is qualitatively new.
The additional evolutionary challenge is to
determine what selectional pressures led to
adaptive changes over time and to understand
the various constraints that channeled
this evolutionary process. Answering these
questions requires a collaborative effort
among linguists, biologists, psychologists,
and anthropologists.
One aim of this essay is to promote a
stronger connection between biology and
linguistics by identifying points of contact
and agreement between the fields. Although
this interdisciplinary marriage was
inaugurated more than 50 years ago, it has
not yet been fully consummated. We hope
to further this goal by, first, helping to
clarify the biolinguistic perspective on language
and its evolution (2–7). We then
review some promising empirical approaches
to the evolution of the language
faculty, with a special focus on
comparative work with nonhuman
animals, and conclude
with a discussion of how inquiry
might profitably advance,
highlighting some outstanding
problems.
We make no attempt to be
comprehensive in our coverage of
relevant or interesting topics and
problems. Nor is it our goal to
review the history of the field.
Rather, we focus on topics that
make important contact between
empirical data and theoretical positions
about the nature of the language
faculty. We believe that if
explorations into the problem of
language evolution are to progress,
we need a clear explication of the
computational requirements for
language, the role of evolutionary
theory in testing hypotheses of
character evolution, and a research
program that will enable a productive
interchange between linguists
and biologists.
Defining the Target: Two
Senses of the Faculty of
Language
The word “language” has highly divergent
meanings in different contexts and disciplines.
In informal usage, a language is understood
as a culturally specific communication
system (English, Navajo, etc.). In the
varieties of modern linguistics that concern
us here, the term “language” is used quite
differently to refer to an internal component
of the mind/brain (sometimes called “internal
language” or “I-language”). We assume that
this is the primary object of interest for the
study of the evolution and function of the
language faculty. However, this biologically
and individually grounded usage still leaves
much open to interpretation (and misunderstanding).
For example, a neuroscientist
might ask: What components of the human
nervous system are recruited in the use of
language in its broadest sense? Because any
aspect of cognition appears to be, at least in
principle, accessible to language, the broadest
answer to this question is, probably, “most of
it.” Even aspects of emotion or cognition not
readily verbalized may be influenced by linguistically
based thought processes. Thus,
this conception is too broad to be of much
use. We therefore delineate two more restricted
conceptions of the faculty of language, one
broader and more inclusive, the other more
restricted and narrow (Fig. 2).
Faculty of language— broad sense
(FLB). FLB includes an internal computational
system (FLN, below) combined with
at least two other organism-internal sysFig.
2. A schematic representation of organism-external and -internal factors related to the faculty of language.
FLB includes sensory-motor, conceptual-intentional, and other possible systems (which we leave open); FLN
includes the core grammatical computations that we suggest are limited to recursion. See text for more
complete discussion.
S C I E N C E ’ S C O M P A S S
22 NOVEMBER 2002 VOL 298 SCIENCE www.sciencemag.org1570
tems, which we call “sensory-motor” and
“conceptual-intentional.” Despite debate on
the precise nature of these systems, and
about whether they are substantially shared
with other vertebrates or uniquely adapted
to the exigencies of language, we take as
uncontroversial the existence of some biological
capacity of humans that allows us
(and not, for example, chimpanzees) to
readily master any human language without
explicit instruction. FLB includes this capacity,
but excludes other organisminternal
systems that are necessary but not
sufficient for language (e.g., memory, respiration,
digestion, circulation, etc.).
Faculty of language—narrow sense
(FLN). FLN is the abstract linguistic computational
system alone, independent of the other
systems with which it interacts and interfaces.
FLN is a component of FLB, and the
mechanisms underlying it are some subset of
those underlying FLB.
Others have agreed on the need for a
restricted sense of “language” but have suggested
different delineations. For example,
Liberman and his associates (8) have argued
that the sensory-motor systems were specifically
adapted for language, and hence should
be considered part of FLN. There is also a
long tradition holding that the conceptualintentional
systems are an intrinsic part of
language in a narrow sense. In this article, we
leave these questions open, restricting attention
to FLN as just defined but leaving the
possibility of a more inclusive definition
open to further empirical research.
The internal architecture of FLN, so conceived,
is a topic of much current research
and debate (4). Without prejudging the issues,
we will, for concreteness, adopt a particular
conception of this architecture. We
assume, putting aside the precise mechanisms,
that a key component of FLN is a
computational system (narrow syntax) that
generates internal representations and maps
them into the sensory-motor interface by the
phonological system, and into the conceptual-intentional
interface by the (formal) semantic
system; adopting alternatives that
have been proposed would not materially
modify the ensuing discussion. All approaches
agree that a core property of FLN is recursion,
attributed to narrow syntax in the conception
just outlined. FLN takes a finite set of
elements and yields a potentially infinite array
of discrete expressions. This capacity of
FLN yields discrete infinity (a property that
also characterizes the natural numbers). Each
of these discrete expressions is then passed to
the sensory-motor and conceptual-intentional
systems, which process and elaborate this
information in the use of language. Each
expression is, in this sense, a pairing of sound
and meaning. It has been recognized for thousands
of years that language is, fundamentally,
a system of sound-meaning connections;
the potential infiniteness of this system has
been explicitly recognized by Galileo, Descartes,
and the 17th-century “philosophical
grammarians” and their successors, notably
von Humboldt. One goal of the study of FLN
and, more broadly, FLB is to discover just
how the faculty of language satisfies these
basic and essential conditions.
The core property of discrete infinity is
intuitively familiar to every language user.
Sentences are built up of discrete units: There
are 6-word sentences and 7-word sentences,
but no 6.5-word sentences. There is no longest
sentence (any candidate sentence can be
trumped by, for example, embedding it in
“Mary thinks that . . .”), and there is no nonarbitrary
upper bound to sentence length. In
these respects, language is directly analogous
to the natural numbers (see below).
At a minimum, then, FLN includes the capacity
of recursion. There are many organisminternal
factors, outside FLN or FLB, that impose
practical limits on the usage of the system.
For example, lung capacity imposes limits on
the length of actual spoken sentences, whereas
working memory imposes limits on the complexity
of sentences if they are to be understandable.
Other limitations—for example, on
concept formation or motor output speed—
represent aspects of FLB, which have their own
evolutionary histories and may have played a
role in the evolution of the capacities of FLN.
Nonetheless, one can profitably inquire into the
evolution of FLN without an
immediate concern for these
limiting aspects of FLB. This
is made clear by the observation
that, although many
aspects of FLB are shared
with other vertebrates, the
core recursive aspect of FLN
currently appears to lack any
analog in animal communication
and possibly other domains
as well. This point,
therefore, represents the
deepest challenge for a comparative
evolutionary approach
to language. We believe
that investigations of
this capacity should include
domains other than communication
(e.g., number, social
relationships, navigation).
Given the distinctions
between FLB and FLN and
the theoretical distinctions
raised above, we can define
a research space as sketched
in Fig. 3. This research
space identifies, as viable,
problems concerning the
evolution of sensory-motor
systems, of conceptual-intentional
systems, and of
FLN. The comparative approach,
to which we turn
next, provides a framework
for addressing questions
about each of these components
of the faculty of
language.
The Comparative
Approach to Language
Evolution
The empirical study of the
evolution of language is beset
with difficulties. Linguistic
behavior does not
fossilize, and a long tradiFig.
3. Investigations into the evolution of the faculty of language
are confronted with a three-dimensional research space that
includes three comparative-evolutionary problems cross-cut by
the core components of the faculty of language. Thus, for each
problem, researchers can investigate details of the sensory-motor
system, the conceptual-intentional system, FLN, and the interfaces
among these systems.
S C I E N C E ’ S C O M P A S S
www.sciencemag.org SCIENCE VOL 298 22 NOVEMBER 2002 1571
tion of analysis of fossil skull shape and
cranial endocasts has led to little consensus
about the evolution of language (7, 9). A
more tractable and, we think, powerful approach
to problems of language evolution is
provided by the comparative method, which
uses empirical data from living species to
draw detailed inferences about extinct ancestors
(3, 10–12). The comparative method was
the primary tool used by Darwin (13, 14) to
analyze evolutionary phenomena and continues
to play a central role throughout modern
evolutionary biology. Although scholars interested
in language evolution have often ignored
comparative data altogether or focused
narrowly on data from nonhuman primates,
current thinking in neuroscience, molecular
biology, and developmental biology indicates
that many aspects of neural and developmental
function are highly conserved, encouraging
the extension of the comparative method
to all vertebrates (and perhaps beyond). For
several reasons, detailed below, we believe
that the comparative method should play a
more central role in future discussions of
language evolution.
An overarching concern in studies of language
evolution is with whether particular
components of the faculty of language
evolved specifically for human language and,
therefore (by extension), are unique to humans.
Logically, the human uniqueness claim
must be based on data indicating an absence
of the trait in nonhuman animals and, to be
taken seriously, requires a substantial body of
relevant comparative data. More concretely,
if the language evolution researcher wishes to
make the claim that a trait evolved uniquely
in humans for the function of language processing,
data indicating that no other animal
has this particular trait are required.
Although this line of reasoning may appear
obvious, it is surprisingly common for a
trait to be held up as uniquely human before
any appropriate comparative data are available.
A famous example is categorical perception,
which when discovered seemed so
finely tuned to the details of human speech as
to constitute a unique human adaptation (15,
16). It was some time before the same underlying
perceptual discontinuities were discovered
in chinchillas and macaques (17, 18),
and even birds (19), leading to the opposite
conclusion that the perceptual basis for categorical
perception is a primitive vertebrate
characteristic that evolved for general auditory
processing, as opposed to specific speech
processing. Thus, a basic and logically ineliminable
role for comparative research on
language evolution is this simple and essentially
negative one: A trait present in nonhuman
animals did not evolve specifically for
human language, although it may be part of
the language faculty and play an intimate role
in language processing. It is possible, of
course, that a trait evolved in nonhuman animals
and humans independently, as analogs
rather than homologs. This would preserve
the possibility that the trait evolved for language
in humans but evolved for some other
reason in the comparative animal group. In
cases where the comparative group is a nonhuman
primate, and perhaps especially chimpanzees,
the plausibility of this evolutionary
scenario is weaker. In any case, comparative
data are critical to this judgment.
Despite the crucial role of homology in
comparative biology, homologous traits are not
the only relevant source of evolutionary data.
The convergent evolution of similar characters
in two independent clades, termed “analogies”
or “homoplasies,” can be equally revealing
(20). The remarkably similar (but nonhomologous)
structures of human and octopus eyes
reveal the stringent constraints placed by the
laws of optics and the contingencies of development
on an organ capable of focusing a sharp
image onto a sheet of receptors. Detailed analogies
between the parts of the vertebrate and
cephalopod eye also provide independent evidence
that each component is an adaptation for
image formation, shaped by natural selection.
Furthermore, the discovery that remarkably
conservative genetic cascades underlie the development
of such analogous structures provides
important insights into the ways in
which developmental mechanisms can
channel evolution (21). Thus, although potentially
misleading for taxonomists, analogies
provide critical data about adaptation
under physical and developmental constraints.
Casting the comparative net more
broadly, therefore, will most likely reveal
larger regularities in evolution, helping to
address the role of such constraints in the
evolution of language.
An analogy recognized as particularly relevant
to language is the acquisition of song by
birds (12). In contrast to nonhuman primates,
where the production of species-typical vocalizations
is largely innate (22), most songbirds
learn their species-specific song by listening to
conspecifics, and they develop highly aberrant
song if deprived of such experience. Current
investigation of birdsong reveals detailed and
intriguing parallels with speech (11, 23, 24).
For instance, many songbirds pass through a
critical period in development beyond which
they produce defective songs that no amount of
acoustic input can remedy, reminiscent of the
difficulty adult humans have in fully mastering
new languages. Further, and in parallel with the
babbling phase of vocalizing or signing human
infants (25), young birds pass through a phase
of song development in which they spontaneously
produce amorphous versions of adult
song, termed “subsong” or “babbling.” Although
the mechanisms underlying the acquisition
of birdsong and human language are clearly
analogs and not homologs, their core components
share a deeply conserved neural and
developmental foundation: Most aspects of
neurophysiology and development—including
regulatory and structural genes, as well as neuron
types and neurotransmitters—are shared
among vertebrates. That such close parallels
have evolved suggests the existence of important
constraints on how vertebrate brains can
acquire large vocabularies of complex, learned
sounds. Such constraints may essentially force
natural selection to come up with the same
solution repeatedly when confronted with similar
problems.
Testing Hypotheses About the
Evolution of the Faculty of Language
Given the definitions of the faculty of language,
together with the comparative framework,
we can distinguish several plausible
hypotheses about the evolution of its various
components. Here, we suggest two hypotheses
that span the diversity of opinion among
current scholars, plus a third of our own.
Hypothesis 1: FLB is strictly homologous
to animal communication. This hypothesis
holds that homologs of FLB, including FLN,
exist (perhaps in less developed or otherwise
modified form) in nonhuman animals (3, 10,
26). This has historically been a popular hypothesis
outside of linguistics and closely
allied fields, and has been defended by some
in the speech sciences. According to this
hypothesis, human FLB is composed of the
same functional components that underlie
communication in other species.
Hypothesis 2: FLB is a derived, uniquely
human adaptation for language. According
to this hypothesis, FLB is a highly complex
adaptation for language, on a par with the
vertebrate eye, and many of its core components
can be viewed as individual traits that
have been subjected to selection and perfected
in recent human evolutionary history. This
appears to represent the null hypothesis for
many scholars who take the complexity of
language seriously (27, 28). The argument
starts with the assumption that FLB, as a
whole, is highly complex, serves the function
of communication with admirable effectiveness,
and has an ineliminable genetic component.
Because natural selection is the only
known biological mechanism capable of generating
such functional complexes [the argument
from design (29)], proponents of this
view conclude that natural selection has
played a powerful role in shaping many aspects
of FLB, including FLN, and, further,
that many of these are without parallel in
nonhuman animals. Although homologous
mechanisms may exist in other animals, the
human versions have been modified by natural
selection to the extent that they can be
reasonably seen as constituting novel traits,
perhaps exapted from other contexts [e.g.,
social intelligence, tool-making (7, 30–32)].
S C I E N C E ’ S C O M P A S S
22 NOVEMBER 2002 VOL 298 SCIENCE www.sciencemag.org1572
Hypothesis 3: Only FLN is uniquely human.
On the basis of data reviewed below, we hypothesize
that most, if not all, of FLB is based
on mechanisms shared with nonhuman animals
(as held by hypothesis 1). In contrast, we suggest
that FLN—the computational mechanism
of recursion—is recently evolved and unique to
our species (33, 34). According to this hypothesis,
much of the complexity manifested in
language derives from complexity in the peripheral
components of FLB, especially those
underlying the sensory-motor (speech or sign)
and conceptual-intentional interfaces, combined
with sociocultural and communicative
contingencies. FLB as a whole thus has an
ancient evolutionary history, long predating the
emergence of language, and a comparative
analysis is necessary to understand this complex
system. By contrast, according to recent
linguistic theory, the computations underlying
FLN may be quite limited. In fact, we propose
in this hypothesis that FLN comprises only the
core computational mechanisms of recursion as
they appear in narrow syntax and the mappings
to the interfaces. If FLN is indeed this restricted,
this hypothesis has the interesting effect of
nullifying the argument from design, and thus
rendering the status of FLN as an adaptation
open to question. Proponents of the idea that
FLN is an adaptation would thus need to supply
additional data or arguments to support this
viewpoint.
The available comparative data on animal
communication systems suggest that the faculty
of language as a whole relies on some uniquely
human capacities that have evolved recently in
the approximately 6 million years since our
divergence from a chimpanzee-like common
ancestor (35). Hypothesis 3, in its strongest
form, suggests that only FLN falls into this
category (34). By this hypothesis, FLB contains
a wide variety of cognitive and perceptual
mechanisms shared with other species, but only
those mechanisms underlying FLN—particularly
its capacity for discrete infinity—are
uniquely human. This hypothesis suggests that
all peripheral components of FLB are shared
with other animals, in more or less the same
form as they exist in humans, with differences
of quantity rather than kind (9, 34). What is
unique to our species is quite specific to FLN,
and includes its internal operations as well as its
interface with the other organism-internal systems
of FLB.
Each of these hypotheses is plausible to
some degree. Ultimately, they can be distinguished
only by empirical data, much of which
is currently unavailable. Before reviewing some
of the relevant data, we briefly consider some
key distinctions between them. From a comparative
evolutionary viewpoint, an important
question is whether linguistic precursors were
involved in communication or in something
else. Proponents of both hypotheses 1 and 2
posit a direct correspondence, by descent with
modification, between some trait involved in
FLB in humans and a similar trait in another
species; these hypotheses differ in whether
the precursors functioned in communication.
Table 1. A sampler of empirical approaches to understanding the evolution of the faculty of language, including both broad (FLB) and narrow (FLN)
components.
Empirical problem Examples References
FLB—sensory-motor system
Vocal imitation and invention Tutoring studies of songbirds, analyses of vocal dialects in whales, spontaneous imitation
of artificially created sounds in dolphins
(11, 12, 24, 65)
Neurophysiology of
action-perception systems
Studies assessing whether mirror neurons, which provide a core substrate for the
action-perception system, may subserve gestural and (possibly) vocal imitation
(67, 68, 71)
Discriminating the sound patterns
of language
Operant conditioning studies of the prototype magnet effect in macaques and starlings (52, 120)
Constraints imposed by vocal tract
anatomy
Studies of vocal tract length and formant dispersion in birds and primates (54–61)
Biomechanics of sound production Studies of primate vocal production, including the role of mandibular oscillations (121, 122)
Modalities of language production
and perception
Cross-modal perception and sign language in humans versus unimodal communication in
animals
(3, 25, 123)
FLB—conceptual-intentional system
Theory of mind, attribution of
mental states
Studies of the seeing/knowing distinction in chimpanzees (84, 86–89)
Capacity to acquire nonlinguistic
conceptual representations
Studies of rhesus monkeys and the object/kind concept (10, 76, 77, 124)
Referential vocal signals Studies of primate vocalizations used to designate predators, food, and social
relationships
(3, 78, 90, 91, 93,
94, 97)
Imitation as a rational, intentional
system
Comparative studies of chimpanzees and human infants suggesting that only the latter
read intentionality into action, and thus extract unobserved rational intent
(125–127)
Voluntary control over signal
production as evidence of
intentional communication
Comparative studies that explore the relationship between signal production and the
composition of a social audience
(3, 10, 92, 128)
FLN—recursion
Spontaneous and training methods
designed to uncover constraints
on rule learning
Studies of serial order learning and finite-state grammars in tamarins and macaques (114, 116, 117,
129)
Sign or artificial language in
trained apes and dolphins
Studies exploring symbol sequencing and open-ended combinatorial manipulation (130, 131)
Models of the faculty of language
that attempt to uncover the
necessary and sufficient
mechanisms
Game theory models of language acquisition, reference, and universal grammar (72–74)
Experiments with animals that
explore the nature and content
of number representation
Operant conditioning studies to determine whether nonhuman primates can represent
number, including properties such as ordinality and cardinality, using such
representations in conjunction with mathematical operands (e.g., add, divide)
(102–106, 132)
Shared mechanisms across
different cognitive domains
Evolution of musical processing and structure, including analyses of brain function and
comparative studies of music perception
(133–135)
S C I E N C E ’ S C O M P A S S
www.sciencemag.org SCIENCE VOL 298 22 NOVEMBER 2002 1573
Although many aspects of FLB very likely
arose in this manner, the important issue for
these hypotheses is whether a series of gradual
modifications could lead eventually to the capacity
of language for infinite generativity. Despite
the inarguable existence of a broadly
shared base of homologous mechanisms involved
in FLB, minor modifications to this
foundational system alone seem inadequate to
generate the fundamental difference—discrete
infinity—between language and all known
forms of animal communication. This claim is
one of several reasons why we suspect that
hypothesis 3 may be a productive way to characterize
the problem of language evolution.
A primary issue separating hypotheses 2
and 3 is whether the uniquely human capacities
of FLN constitute an adaptation. The
viewpoint stated in hypothesis 2, especially
the notion that FLN in particular is a highly
evolved adaptation, has generated much enthusiasm
recently [e.g., (36)], especially
among evolutionary psychologists (37, 38).
At present, however, we see little reason to
believe either that FLN can be anatomized
into many independent but interacting traits,
each with its own independent evolutionary
history, or that each of these traits could have
been strongly shaped by natural selection,
given their tenuous connection to communicative
efficacy (the surface or phenotypic
function upon which selection presumably
acted).
We consider the possibility that certain specific
aspects of the faculty of language are
“spandrels”—by-products of preexisting constraints
rather than end products of a history of
natural selection (39). This possibility, which
opens the door to other empirical lines of inquiry,
is perfectly compatible with our firm support
of the adaptationist program. Indeed, it follows
directly from the foundational notion that adaptation
is an “onerous concept” to be invoked
only when alternative explanations fail (40).
The question is not whether FLN in toto is
adaptive. By allowing us to communicate an
endless variety of thoughts, recursion is clearly
an adaptive computation. The question is
whether particular components of the functioning
of FLN are adaptations for language, specifically
acted upon by natural selection—or,
even more broadly, whether FLN evolved for
reasons other than communication.
An analogy may make this distinction
clear. The trunk and branches of trees are
near-optimal solutions for providing an individual
tree’s leaves with access to sunlight.
For shrubs and small trees, a wide variety of
forms (spreading, spherical, multistalked,
etc.) provide good solutions to this problem.
For a towering rainforest canopy tree, however,
most of these forms are rendered impossible
by the various constraints imposed
by the properties of cellulose and the problems
of sucking water and nutrients up to the
leaves high in the air. Some aspects of such
trees are clearly adaptations channeled by
these constraints; others (e.g., the popping of
xylem tubes on hot days, the propensity to be
toppled in hurricanes) are presumably unavoidable
by-products of such constraints.
Recent work on FLN (4, 41–43) suggests
the possibility that at least the narrow-syntactic
component satisfies conditions of highly efficient
computation to an extent previously unsuspected.
Thus, FLN may approximate a kind of
“optimal solution” to the problem of linking
the sensory-motor and conceptual-intentional
systems. In other words, the generative processes
of the language system may provide a
near-optimal solution that satisfies the interface
conditions to FLB. Many of the details of language
that are the traditional focus of linguistic
study [e.g., subjacency, Wh- movement, the
existence of garden-path sentences (4, 44)] may
represent by-products of this solution, generated
automatically by neural/computational
constraints and the structure of FLB—
components that lie outside of FLN. Even
novel capacities such as recursion are implemented
in the same type of neural tissue as the
rest of the brain and are thus constrained by
biophysical, developmental, and computational
factors shared with other vertebrates. Hypothesis
3 raises the possibility that structural
details of FLN may result from such preexisting
constraints, rather than from direct shaping by
natural selection targeted specifically at communication.
Insofar as this proves to be true,
such structural details are not, strictly speaking,
adaptations at all. This hypothesis and the
alternative selectionist account are both viable
and can eventually be tested with comparative
data.
Comparative Evidence for the Faculty
of Language
Study of the evolution of language has accelerated
in the past decade (45, 46). Here, we
offer a highly selective review of some of
these studies, emphasizing animal work that
seems particularly relevant to the hypotheses
advanced above; many omissions were necessary
for reasons of space, and we firmly
believe that a broad diversity of methods and
perspectives will ultimately provide the richest
answers to the problem of language evolution.
For this reason, we present a broader
sampler of the field’s offerings in Table 1.
How “special” is speech? Comparative
study of the sensory-motor system. Starting with
early work on speech perception, there has been
a tradition of considering speech “special,” and
thus based on uniquely human mechanisms
adapted for speech perception and/or production
[e.g., (7, 8, 47, 48)]. This perspective has
stimulated a vigorous research program studying
animal speech perception and, more recently,
speech production. Surprisingly, this research
has turned up little evidence for uniquely
human mechanisms special to speech, despite a
persistent tendency to assume uniqueness even
in the absence of relevant animal data.
On the side of perception, for example,
many species show an impressive ability to
both discriminate between and generalize
over human speech sounds, using formants as
the critical discriminative cue (17–19, 49–
51). These data provide evidence not only of
categorical perception, but also of the ability
to discriminate among prototypical exemplars
of different phonemes (52). Further, in
the absence of training, nonhuman primates
can discriminate sentences from two different
languages on the basis of rhythmic differences
between them (53).
On the side of production, birds and nonhuman
primates naturally produce and perceive
formants in their own species-typical
vocalizations (54–59). The results also shed
light on discussions of the uniquely human
structure of the vocal tract and the unusual
descended larynx of our species (7, 48, 60),
because new evidence shows that several other
mammalian species also have a descended
larynx (61). Because these nonhuman species
lack speech, a descended larynx clearly has
nonphonetic functions; one possibility is exaggerating
apparent size. Although this particular
anatomical modification undoubtedly
plays an important role in speech production
in modern humans, it need not have first
evolved for this function. The descended larynx
may thus be an example of classic Darwinian
preadaptation.
Many phenomena in human speech perception
have not yet been investigated in animals
[e.g., the McGurk effect, an illusion in which
the syllable perceived from a talking head represents
the interaction between an articulatory
gesture seen and a different syllable heard; see
(62)]. However, the available data suggest a
much stronger continuity between animals and
humans with respect to speech than previously
believed. We argue that the continuity hypothesis
thus deserves the status of a null hypothesis,
which must be rejected by comparative
work before any claims of uniqueness can be
validated. For now, this null hypothesis of no
truly novel traits in the speech domain appears
to stand.
There is, however, a striking ability tied to
speech that has received insufficient attention:
the human capacity for vocal imitation
(63, 64). Imitation is obviously a necessary
component of the human capacity to acquire
a shared and arbitrary lexicon, which is itself
central to the language capacity. Thus, the
capacity to imitate was a crucial prerequisite
of FLB as a communicative system. Vocal
imitation and learning are not uniquely human.
Rich multimodal imitative capacities
are seen in other mammals (dolphins) and
some birds (parrots), with most songbirds
exhibiting a well-developed vocal imitative
S C I E N C E ’ S C O M P A S S
22 NOVEMBER 2002 VOL 298 SCIENCE www.sciencemag.org1574
capacity (65). What is surprising is that monkeys
show almost no evidence of visually
mediated imitation, with chimpanzees showing
only slightly better capacities (66). Even
more striking is the virtual absence of evidence
for vocal imitation in either monkeys
or apes (3). For example, intensively trained
chimpanzees are incapable of acquiring anything
but a few poorly articulated spoken
words, whereas parrots can readily acquire a
large vocal repertoire. With respect to their
own vocalizations, there are few convincing
studies of vocal dialects in primates, thereby
suggesting that they lack a vocal imitative
capacity (3, 65). Evidence for spontaneous
visuomanual imitation in chimpanzees is not
much stronger, although with persistent training
they can learn several hundred hand
signs. Further, even in cases where
nonhuman animals are capable of imitating
in one modality (e.g., song
copying in songbirds), only dolphins
and humans appear capable of imitation
in multiple modalities. The detachment
from modality-specific inputs
may represent a substantial
change in neural organization, one
that affects not only imitation but
also communication; only humans
can lose one modality (e.g., hearing)
and make up for this deficit by
communicating with complete competence
in a different modality (i.e.,
signing).
Our discussion of limitations is
not meant to diminish the impressive
achievements of monkeys and apes,
but to highlight how different the
mechanisms underlying the production
of human and nonhuman primate
gestures, either vocally expressed or
signed, must be. After all, the average
high school graduate knows up to
60,000 words, a vocabulary achieved
with little effort, especially when
contrasted with the herculean efforts
devoted to training animals. In sum,
the impressive ability of any normal
human child for vocal imitation may
represent a novel capacity that
evolved in our recent evolutionary
history, some time after the divergence
from our chimpanzee-like ancestors.
The existence of analogs in
distantly related species, such as
birds and cetaceans, suggests considerable
potential for the detailed comparative
study of vocal imitation. There are,
however, potential traps that must be avoided,
especially with respect to explorations of
the neurobiological substrates of imitation.
For example, although macaque monkeys
and humans are equipped with so-called
“mirror neurons” in the premotor cortex that
respond both when an individual acts in a
particular way and when the same individual
sees someone else act in this same way (67,
68), these neurons are not sufficient for imitation
in macaques, as many have presumed:
As mentioned, there is no convincing evidence
of vocal or visual imitation in monkeys.
Consequently, as neuroimaging studies
continue to explore the neural basis of imitation
in humans (69–71), it will be important
to distinguish between the necessary and sufficient
neural correlates of imitation. This is
especially important, given that some recent
attempts to model the evolution of language
begin with a hypothetical organism that is
equipped with the capacity for imitation and
intentionality, as opposed to working out how
these mechanisms evolved in the first place
[see below; (72–74)]. If a deeper evolutionary
exploration is desired, one dating back to
a chimpanzee-like ancestor, then we need to
explain how and why such capacities
emerged from an ancestral node that lacked
such abilities (75) (Fig. 4).
The conceptual-intentional systems of nonlinguistic
animals. A wide variety of studies
indicate that nonhuman mammals and birds
have rich conceptual representations (76, 77).
Surprisingly, however, there is a mismatch between
the conceptual capacities of animals and
the communicative content of their vocal and
visual signals (78, 79). For example, although a
wide variety of nonhuman primates have access
to rich knowledge of who is related to whom, as
well as who is dominant and who is subordinate,
their vocalizations only coarsely express
such complexities.
Studies using classical training approaches
as well as methods that tap spontaneous
abilities reveal that animals acquire and use a
wide range of abstract concepts, including
tool, color, geometric relationships, food, and
number (66, 76–82). More controversially,
but of considerable relevance to intentional
aspects of language and conditions of felicitous
use, some studies claim that animals
have a theory of mind (83–85), including a
sense of self and the ability to represent the
beliefs and desires of other group members.
On the side of positive support, recent studies
of chimpanzees suggest that they recognize
the perceptual act of seeing as a proxy for the
mental state of knowing (84, 86, 87). These
Fig. 4. The distribution of imitation in the animal kingdom is patchy. Some animals such as songbirds,
dolphins, and humans have evolved exceptional abilities to imitate; other animals, such as apes and
monkeys, either lack such abilities or have them in a relatively impoverished form. [Illustration: John Yanson]
S C I E N C E ’ S C O M P A S S
www.sciencemag.org SCIENCE VOL 298 22 NOVEMBER 2002 1575
studies suggest that at least chimpanzees, but
perhaps no other nonhuman animals, have a
rudimentary theory of mind. On the side of
negative support, other studies suggest that
even chimpanzees lack a theory of mind,
failing, for example, to differentiate between
ignorant and knowledgeable individuals with
respect to intentional communication (88,
89). Because these experiments make use of
different methods and are based on small
sample sizes, it is not possible at present to
derive any firm conclusions about the presence
or absence of mental state attribution in
animals. Independently of how this controversy
is resolved, however, the best evidence
of referential communication in animals
comes not from chimpanzees but from a variety
of monkeys and birds, species for which
there is no convincing evidence for a theory
of mind.
The classic studies of vervet monkey
alarm calls (90) have now been joined by
several others, each using comparable methods,
with extensions to different species (macaques,
Diana monkeys, meerkats, prairie
dogs, chickens) and different communicative
contexts (social relationships, food, intergroup
aggression) (91–97). From these studies
we can derive five key points relevant to
our analysis of the faculty of language. First,
individuals produce acoustically distinctive
calls in response to functionally important
contexts, including the detection of predators
and the discovery of food. Second, the acoustic
morphology of the signal, although arbitrary
in terms of its association with a particular
context, is sufficient to enable listeners to
respond appropriately without requiring any
other contextual information. Third, the number
of such signals in the repertoire is small,
restricted to objects and events experienced
in the present, with no evidence of creative
production of new sounds for new situations.
Fourth, the acoustic morphology of the calls
is fixed, appearing early in development, with
experience only playing a role in refining the
range of objects or events that elicit such
calls. Fifth, there is no evidence that calling is
intentional in the sense of taking into account
what other individuals believe or want.
Early interpretations of this work suggested
that when animals vocalize, they are functionally
referring to the objects and events
that they have encountered. As such, vervet
alarm calls and rhesus monkey food calls, to
take two examples, were interpreted as wordlike,
with callers referring to different kinds
of predators or different kinds of food. More
recent discussions have considerably weakened
this interpretation, suggesting that if the
signal is referential at all, it is in the mind of
the listener who can extract information
about the signaler’s current context from the
acoustic structure of the call alone (78, 95).
Despite this evidence that animals can extract
information from the signal, there are several
reasons why additional evidence is required
before such signals can be considered as precursors
for, or homologs of, human words.
Roughly speaking, we can think of a particular
human language as consisting of words and
computational procedures (“rules”) for constructing
expressions from them. The computational
system has the recursive property briefly
outlined earlier, which may be a distinct human
property. However, key aspects of words may
also be distinctively human. There are, first of
all, qualitative differences in scale and mode of
acquisition, which suggest that quite different
mechanisms are involved; as pointed out above,
there is no evidence for vocal imitation in nonhuman
primates, and although human children
may use domain-general mechanisms to acquire
and recall words (98, 99), the rate at
which children build the lexicon is so massively
different from nonhuman primates that one
must entertain the possibility of an independently
evolved mechanism. Furthermore,
unlike the best animal examples
of putatively referential signals,
most of the words of human language
are not associated with specific
functions (e.g., warning cries, food
announcements) but can be linked to
virtually any concept that humans
can entertain. Such usages are often
highly intricate and detached from
the here and now. Even for the simplest
words, there is typically no
straightforward word-thing relationship,
if “thing” is to be understood in
mind-independent terms. Without
pursuing the matter here, it appears
that many of the elementary properties
of words—including those that
enter into referentiality—have only
weak analogs or homologs in natural
animal communication systems, with
only slightly better evidence from the
training studies with apes and dolphins.
Future research must therefore
provide stronger support for the precursor
position, or it must instead
abandon this hypothesis, arguing that
this component of FLB (conceptualintentional)
is also uniquely human.
Discrete infinity and constraints
on learning. The data summarized thus far,
although far from complete, provide overall
support for the position of continuity between
humans and other animals in terms of FLB.
However, we have not yet addressed one
issue that many regard as lying at the heart of
language: its capacity for limitless expressive
power, captured by the notion of discrete
infinity. It seems relatively clear, after nearly
a century of intensive research on animal
communication, that no species other than
humans has a comparable capacity to recombine
meaningful units into an unlimited variety
of larger structures, each differing systematically
in meaning. However, little
progress has been made in identifying the
specific capabilities that are lacking in other
animals.
Fig. 5. Human and nonhuman animals exhibit the capacity to compute numerosities, including small precise
number quantification and large approximate number estimation. Humans may be unique, however, in the ability
to show open-ended, precise quantificational skills with large numbers, including the integer count list. In parallel
with the faculty of language, our capacity for number relies on a recursive computation. [Illustration: John Yanson]
S C I E N C E ’ S C O M P A S S
22 NOVEMBER 2002 VOL 298 SCIENCE www.sciencemag.org1576
The astronomical variety of sentences any
natural language user can produce and understand
has an important implication for language
acquisition, long a core issue in developmental
psychology. A child is exposed to
only a small proportion of the possible sentences
in its language, thus limiting its database
for constructing a more general version
of that language in its own mind/brain. This
point has logical implications for any system
that attempts to acquire a natural language on
the basis of limited data. It is immediately
obvious that given a finite array of data, there
are infinitely many theories consistent with it
but inconsistent with one another. In the
present case, there are in principle infinitely
many target systems (potential I-languages)
consistent with the data of experience, and
unless the search space and acquisition mechanisms
are constrained, selection among
them is impossible. A version of the problem
has been formalized by Gold (100) and more
recently and rigorously explored by Nowak
and colleagues (72–75). No known “general
learning mechanism” can acquire a natural
language solely on the basis of positive or
negative evidence, and the prospects for finding
any such domain-independent device
seem rather dim. The difficulty of this problem
leads to the hypothesis that whatever
system is responsible must be biased or constrained
in certain ways. Such constraints
have historically been termed “innate dispositions,”
with those underlying language referred
to as “universal grammar.” Although
these particular terms have been forcibly rejected
by many researchers, and the nature of
the particular constraints on human (or animal)
learning mechanisms is currently unresolved,
the existence of some such constraints
cannot be seriously doubted. On the
other hand, other constraints in animals must
have been overcome at some point in human
evolution to account for our ability to acquire
the unlimited class of generative systems that
includes all natural languages. The nature of
these latter constraints has recently become
the target of empirical work. We focus here
on the nature of number representation and
rule learning in nonhuman animals and human
infants, both of which can be investigated
independently of communication and provide
hints as to the nature of the constraints
on FLN.
More than 50 years of research using classical
training studies demonstrates that animals
can represent number, with careful controls
for various important confounds (80). In
the typical experiment, a rat or pigeon is
trained to press a lever x number of times to
obtain a food reward. Results show that animals
can hit the target number to within a
closely matched mean, with a standard deviation
that increases with magnitude: As the
target number increases, so does variation
around the mean. These results have led to
the idea that animals, including human infants
and adults, can represent number approximately
as a magnitude with scalar variability
(101, 102). Number discrimination is
limited in this system by Weber’s law, with
greater discriminability among small numbers
than among large numbers (keeping distances
between pairs constant) and between
numbers that are farther apart (e.g., 7 versus
8 is harder than 7 versus 12). The approximate
number sense is accompanied by a second
precise mechanism that is limited to values
less than 4 but accurately distinguishes 1
from 2, 2 from 3, and 3 from 4; this second
system appears to be recruited in the context
of object tracking and is limited by working
memory constraints (103). Of direct relevance
to the current discussion, animals can
be trained to understand the meaning of number
words or Arabic numeral symbols. However,
these studies reveal striking differences
in how animals and human children acquire
the integer list, and provide further evidence
that animals lack the capacity to create openended
generative systems.
Boysen and Matsuzawa have trained
chimpanzees to map the number of objects
onto a single Arabic numeral, to correctly
order such numerals in either an ascending or
descending list, and to indicate the sums of
two numerals (104–106). For example, Boysen
shows that a chimpanzee seeing two oranges
placed in one box, and another two
oranges placed in a second box, will pick the
correct sum of four out of a lineup of three
cards, each with a different Arabic numeral.
The chimpanzees’ performance might suggest
that their representation of number is like
ours. Closer inspection of how these chimpanzees
acquired such competences, however,
indicates that the format and content of
their number representations differ fundamentally
from those of human children. In
particular, these chimpanzees required thousands
of training trials, and often years, to
acquire the integer list up to nine, with no
evidence of the kind of “aha” experience that
all human children of approximately 3.5
years acquire (107). A human child who has
acquired the numbers 1, 2, and 3 (and sometimes
4) goes on to acquire all the others; he
or she grasps the idea that the integer list is
constructed on the basis of the successor
function. For the chimpanzees, in contrast,
each number on the integer list required the
same amount of time to learn. In essence,
although the chimpanzees’ understanding of
Arabic numerals is impressive, it parallels
their understanding of other symbols and
their referential properties: The system apparently
never takes on the open-ended generative
property of human language. This limitation
may, however, reveal an interesting
quirk of the child’s learning environment and
a difference from the training regime of animals:
Children typically first learn an arbitrary
ordered list of symbols (“1, 2, 3, 4 . . . ”)
and later learn the precise meaning of such
words; apes and parrots, in contrast, were
taught the meanings one by one without
learning the list. As Carey (103) has argued,
this may represent a fundamental difference
in experience, a hypothesis that could be
tested by first training animals with an arbitrary
ordered list.
A second possible limitation on the class
of learnable structures concerns the kinds of
statistical inferences that animals can compute.
Early work in computational linguistics
(108–110) suggested that we can profitably
think about language as a system of rules
placed within a hierarchy of increasing complexity.
At the lowest level of the hierarchy
are rule systems that are limited to local
dependencies, a subcategory of so-called
“finite-state grammars.” Despite their attractive
simplicity, such rule systems are inadequate
to capture any human language. Natural
languages go beyond purely local structure
by including a capacity for recursive
embedding of phrases within phrases, which
can lead to statistical regularities that are
separated by an arbitrary number of words or
phrases. Such long-distance, hierarchical relationships
are found in all natural languages
for which, at a minimum, a “phrase-structure
grammar” is necessary. It is a foundational
observation of modern generative linguistics
that, to capture a natural language, a grammar
must include such capabilities (Fig. 5).
Recent studies suggest that the capacity to
compute transitional probabilities—an example
of a rule at the lowest level of the hierarchy—might
be available to human infants
and provide a mechanism for segmenting
words from a continuous acoustic stream
(111–113). Specifically, after familiarization
to a continuous sequence of consonant-vowel
(CV) syllables, where particular trigrams
(three CVs in sequence, considered to be
“words” in this context) have a high probability
of appearing within the corpus, infants
are readily able to discriminate these trigrams
from others that are uncommon.
Although this ability may provide a mechanism
for word segmentation, it is apparently
not a mechanism that evolved uniquely
in humans or for language: The same
computation is spontaneously available to
human infants for visual sequences and
tonal melodies (113), as well as to nonhuman
primates (cotton-top tamarins) tested
with the same methods and stimuli (114).
Similarly, in the same way that human
infants appear capable of computing algebraic
rules that operate over particular CV
sequences (115), so too can cotton-top
tamarins (116), again demonstrating that
the capacity to discover abstract rules at a
S C I E N C E ’ S C O M P A S S
www.sciencemag.org SCIENCE VOL 298 22 NOVEMBER 2002 1577
local level is not unique to humans, and
almost certainly did not evolve specifically
for language.
Fitch and Hauser (117) recently completed
a study comparing finite-state and phrasestructure
grammar acquisition in human
adults and tamarins, using the same subjects
and methods as the studies above. The
phrase-structure rule tested was AnBn, where
A and B were each represented by one of a
set of eight different CVs. The rule therefore
specified both a set of consistent strings (n
A’s must precede n B’s) and a set of inconsistent
strings; the latter consisted of violations
of order (B tokens precede A tokens) or
of patterning (alternations of A’s and B’s
such as ABAB). Results showed that human
adults rapidly learned this rule implicitly,
distinguishing consistent and inconsistent
strings. Tamarins, in contrast, failed in three
separate experiments testing their ability to
acquire this grammar, but they readily
mastered a finite-state variant (ABn) implemented
with the same stimuli and testing
conditions. This suggests that tamarins have a
limited capacity to learn the type of longdistance
hierarchical dependencies necessary
to achieve the class of phrase-structure grammars.
If true, this limitation would place severe
restrictions on their capacity to learn any
natural human language. It is currently unclear
whether this limitation generalizes to
other animals, and whether it is similarly
imposed on humans at different stages of
development. Nonetheless, such experiments
provide an empirical approach to exploring
key differences between humans and animals
relevant to FLN.
Our review has stressed the usefulness
of animal data for theories about humans,
but this exchange need not be one-way. As
the research program we have sketched
progresses, more general principles about
cognitive evolution may emerge. For example,
suppose we adopt the conception of
hypothesis 3, oversimplifying radically,
that the interface systems—sensory-motor
and conceptual-intentional—are given, and
the innovation that yielded the faculty of
language was the evolution of the computational
system that links them. The computational
system must (i) construct an infinite
array of internal expressions from the
finite resources of the conceptual-intentional
system, and (ii) provide the means to
externalize and interpret them at the sensory-motor
end. We may now ask to what
extent the computational system is optimal,
meeting natural conditions of efficient
computation such as minimal search and no
backtracking. To the extent that this can be
established, we will be able to go beyond
the (extremely difficult, and still distant)
accomplishment of finding the principles of
the faculty of language, to an understanding
of why the faculty follows these particular
principles and not others. We would then
understand why languages of a certain class
are attainable, whereas other imaginable
languages are impossible to learn and sustain.
Such progress would not only open the
door to a greatly simplified and empirically
more tractable evolutionary approach to the
faculty of language, but might also be more
generally applicable to domains beyond
language in a wide range of species—perhaps
especially in the domain of spatial
navigation and foraging, where problems of
optimal search are relevant. For example,
elegant studies of insects, birds, and primates
reveal that individuals often search
for food by an optimal strategy, one involving
minimal distances, recall of locations
searched, and kinds of objects retrieved
(77, 118, 119). Only after a concerted, multidisciplinary
attack on the problems of
language evolution, paralleling 40 years of
optimal foraging research, will we learn
whether such similarities are more than
superficial.
Conclusions
We conclude by making three points. First, a
practical matter: Linguists and biologists,
along with researchers in the relevant branches
of psychology and anthropology, can
move beyond unproductive theoretical debate
to a more collaborative, empirically focused
and comparative research program aimed at
uncovering both shared (homologous or analogous)
and unique components of the faculty
of language. Second, although we have argued
that most if not all of FLB is shared with
other species, whereas FLN may be unique to
humans, this represents a tentative, testable
hypothesis in need of further empirical investigation.
Finally, we believe that a comparative
approach is most likely to lead to new
insights about both shared and derived features,
thereby generating new hypotheses
concerning the evolutionary forces that led to
the design of the faculty of language. Specifically,
although we have said relatively little
about the role of natural selection in shaping
the design features of FLN, we suggest that
by considering the possibility that FLN
evolved for reasons other than language, the
comparative door has been opened in a new
and (we think) exciting way.
Comparative work has generally focused
on animal communication or the capacity to
acquire a human-created language. If, however,
one entertains the hypothesis that recursion
evolved to solve other computational
problems such as navigation, number quantification,
or social relationships, then it is
possible that other animals have such abilities,
but our research efforts have been targeted
at an overly narrow search space (Fig.
3). If we find evidence for recursion in animals,
but in a noncommunicative domain,
then we are more likely to pinpoint the mechanisms
underlying this ability and the selective
pressures that led to it. This discovery, in
turn, would open the door to another suite of
puzzles: Why did humans, but no other animal,
take the power of recursion to create an
open-ended and limitless system of communication?
Why does our system of recursion
operate over a broader range of elements or
inputs (e.g., numbers, words) than other animals?
One possibility, consistent with current
thinking in the cognitive sciences, is that
recursion in animals represents a modular
system designed for a particular function
(e.g., navigation) and impenetrable with respect
to other systems. During evolution, the
modular and highly domain-specific system
of recursion may have become penetrable and
domain-general. This opened the way for humans,
perhaps uniquely, to apply the power
of recursion to other problems. This change
from domain-specific to domain-general may
have been guided by particular selective pressures,
unique to our evolutionary past, or as a
consequence (by-product) of other kinds of
neural reorganization. Either way, these are
testable hypotheses, a refrain that highlights
the importance of comparative approaches to
the faculty of language.
References and Notes
1. N. Chomsky, Aspects of the Theory of Syntax (MIT
Press, Cambridge, MA, 1965).
2. ࿜࿜࿜࿜, Reflections on Language (Pantheon, New
York, 1975).
3. M. D. Hauser, The Evolution of Communication (MIT
Press, Cambridge, MA, 1996).
4. R. Jackendoff, Foundations of Language (Oxford
Univ. Press, New York, 2002).
5. L. Jenkins, Biolinguistics (Cambridge Univ. Press,
Cambridge, 2000).
6. E. H. Lenneberg, Biological Foundations of Language
(Wiley, New York, 1967).
7. P. Lieberman, The Biology and Evolution of Language
(Harvard Univ. Press, Cambridge, MA, 1984).
8. A. Liberman, Speech: A Special Code (MIT Press,
Cambridge, MA, 1996).
9. W. T. Fitch, Trends Cognit. Sci. 4, 258 (2000).
10. D. L. Cheney, R. M. Seyfarth, How Monkeys See the
World: Inside the Mind of Another Species (Univ. of
Chicago Press, Chicago, 1990).
11. A. Doupe, P. Kuhl, Annu. Rev. Neurosci. 22, 567
(1999).
12. P. Marler, Am. Sci. 58, 669 (1970).
13. C. Darwin, On the Origin of Species (John Murray,
London, 1859).
14. ࿜࿜࿜࿜, The Descent of Man and Selection in Relation
to Sex (John Murray, London, 1871).
15. A. M. Liberman, K. S. Harris, H. S. Hoffman, B. C.
Griffith, J. Exp. Psychol. 54, 358 (1957).
16. A. M. Liberman, F. S. Cooper, D. P. Shankweiler, M.
Studdert-Kennedy, Psychol. Rev. 74, 431 (1967).
17. P. K. Kuhl, J. D. Miller, Science 190, 69 (1975).
18. P. K. Kuhl, D. M. Padden, Percept. Psychophys. 32,
542 (1982).
19. K. R. Kluender, R. Diehl, P. R. Killeen, Science 237,
1195 (1987).
20. S. J. Gould, in Evolution, Brain and Behavior: Persistent
Problems, R. B. Masterton, W. Hodos, H. Jerison,
Eds. (Wiley, New York, 1976), pp. 175–179.
21. W. J. Gehring, Master Control Genes in Development
and Evolution: The Homeobox Story (Yale Univ.
Press, New Haven, CT, 1998).
22. R. M. Seyfarth, D. L. Cheney, in The Design of Animal
S C I E N C E ’ S C O M P A S S
22 NOVEMBER 2002 VOL 298 SCIENCE www.sciencemag.org1578
Communication, M. D. Hauser, M. Konishi, Eds. (MIT
Press, Cambridge, MA, 1999), pp. 391–418.
23. P. Marler, J. Neurobiol. 33, 1 (1997).
24. F. Nottebohm, in The Design of Animal Communication,
M. D. Hauser, M. Konishi, Eds. (MIT Press,
Cambridge, MA, 1999), pp. 63–110.
25. L. A. Petitto, P. Marentette, Science 251, 1483
(1991).
26. K. R. Kluender, A. J. Lotto, L. L. Holt, in Listening to
Speech: An Auditory Perspective, S. Greenberg, W.
Ainsworth, Eds. (Erlbaum, Mahwah, NJ, in press).
27. R. Jackendoff, Trends Cognit. Sci. 3, 272 (1999).
28. S. Pinker, P. Bloom, Behav. Brain Sci. 13, 707 (1990).
29. R. Dawkins, The Blind Watchmaker (Norton, New
York, 1986).
30. D. Bickerton, Species and Language (Univ. of Chicago
Press, Chicago, 1990).
31. R. Dunbar, Grooming, Gossip and the Evolution of
Language (Harvard Univ. Press, Cambridge, MA,
1996).
32. D. Kimura, Neuromotor Mechanisms in Human Communication
(Oxford Univ. Press, Oxford, 1993).
33. N. Chomsky, Rules and Representations (Columbia
Univ. Press, New York, 1980).
34. M. D. Hauser, in Language, Brain, and Cognitive
Development: Essays in Honor of Jacques Mehler, E.
Dupoux, Ed. (MIT Press, Cambridge, MA, 2001), pp.
417–434.
35. W. Enard et al., Nature 418, 869 (2002).
36. J. Maynard Smith, E. Szathmary, The Major Transitions
of Evolution (Freeman, Oxford, 1995).
37. L. Barrett, R. Dunbar, J. Lycett, Human Evolutionary
Psychology (Princeton Univ. Press, Princeton, NJ,
2002).
38. D. Buss, Evolutionary Psychology (Allyn & Bacon,
London, 1999).
39. S. J. Gould, R. C. Lewontin, Proc. R. Soc. London 205,
281 (1979).
40. G. C. Williams, Adaptation and Natural Selection
(Princeton Univ. Press, Princeton, NJ, 1966).
41. N. Chomsky, The Minimalist Program (MIT Press,
Cambridge, MA, 1995).
42. C. Collins, Local Economy (MIT Press, Cambridge,
MA, 1997).
43. S. D. Epstein, N. Hornstein, Working Minimalism
(MIT Press, Cambridge, MA, 1999).
44. L. Haegeman, Introduction to Government & Binding
Theory (Blackwell, Oxford, 1991).
45. J. R. Hurford, M. Studdert-Kennedy, C. Knight, Eds.,
Approaches to the Evolution of Language: Social and
Cognitive Bases (Cambridge Univ. Press, Cambridge,
1998).
46. A. Wray, Ed., The Transition to Language (Oxford
Univ. Press, Oxford, 2002).
47. A. Liberman, D. H. Whalen, Trends Cognit. Sci. 4,
187 (2000).
48. P. Lieberman, Uniquely Human (Harvard Univ. Press,
Cambridge, MA, 1991).
49. R. J. Dooling, C. T. Best, S. D. Brown, J. Acoust. Soc.
Am. 97, 1839 (1995).
50. J. M. Sinnott, C. H. Brown, J. Acoust. Soc. Am. 102,
588 (1997).
51. M. S. Sommers, D. B. Moody, C. A. Prosen, W. C.
Stebbins, J. Acoust. Soc. Am. 91, 3499 (1992).
52. K. R. Kluender, A. J. Lotto, L. L. Holt, S. L. Bloedel, J.
Acoust. Soc. Am. 104, 3568 (1998).
53. F. Ramus, M. D. Hauser, C. T. Miller, D. Morris, J.
Mehler, Science 288, 349 (2000).
54. W. T. Fitch, J. Acoust. Soc. Am. 102, 1213 (1997).
55. ࿜࿜࿜࿜, J. P. Kelley, Ethology 106, 559 (2000).
56. M. D. Hauser, C. S. Evans, P. Marler, Anim. Behav. 45,
423 (1993).
57. M. J. Owren, R. Bernacki, J. Acoust. Soc. Am. 83,
1927 (1988).
58. M. J. Owren, J. Comp. Psychol. 104, 20 (1990).
59. D. Rendall, M. J. Owren, P. S. Rodman, J. Acoust. Soc.
Am. 103, 602 (1998).
60. V. E. Negus, The Comparative Anatomy and Physiology
of the Larynx (Hafner, New York, 1949).
61. W. T. Fitch, D. Reby, Proc. R. Soc. London Ser. B 268,
1669 (2001).
62. J. D. Trout, Psychol. Rev. 108, 523 (2000).
63. M. Donald, in Approaches to the Evolution of Language:
Social and Cognitive Bases, J. R. Hurford, M.
Studdert-Kennedy, C. Knight, Eds. (Cambridge Univ.
Press, Cambridge, 1998), pp. 44–67.
64. M. Studdert-Kennedy, Hum. Neurobiol. 2, 191
(1983).
65. V. M. Janik, P. J. B. Slater, Anim. Behav. 60, 1 (2000).
66. M. Tomasello, J. Call, Primate Cognition (Oxford
Univ. Press, Oxford, 1997).
67. G. Rizzolatti, M. A. Arbib, Trends Cognit. Sci. 2, 188
(1998).
68. G. Rizzolatti, L. Fadiga, L. Fogassi, V. Gallese, Arch.
Ital. Biol. 137, 169 (1999).
69. T. Chaminade, A. N. Meltzoff, J. Decety, Neuroimage
15, 318 (2002).
70. J. Decety, T. Chaminade, J. Grezes, A. N. Meltzoff,
Neuroimage 15, 265 (2002).
71. M. Iacoboni et al., Science 286, 2526 (1999).
72. M. A. Nowak, N. L. Komarova, P. Niyogi, Science
291, 114 (2001).
73. M. A. Nowak, N. L. Komarova, Trends Cognit. Sci. 5,
288 (2001).
74. M. A. Nowak, J. B. Plotkin, V. A. Jansen, Nature 404,
495 (2000).
75. M. A. Nowak, N. L. Komarova, P. Niyogi, Nature 417,
611 (2002).
76. C. M. Heyes, F. Huber, The Evolution of Cognition
(MIT Press, Cambridge, MA, 2000).
77. S. Shettleworth, Cognition, Evolution and Behavior
(Oxford Univ. Press, New York, 1998).
78. D. L. Cheney, R. M. Seyfarth, in The Tanner Lectures
on Human Values, G. Peterson, Ed. (Univ. of Utah
Press, Salt Lake City, UT, 1998), pp. 173–210.
79. M. D. Hauser, Wild Minds: What Animals Really Think
(Holt, New York, 2000).
80. C. R. Gallistel, The Organization of Learning (MIT
Press, Cambridge, MA, 1990).
81. I. M. Pepperberg, The Alex Studies (Harvard Univ.
Press, Cambridge, MA, 2000).
82. D. Premack, Gavagai! or the Future History of the
Animal Language Controversy (MIT Press, Cambridge,
MA, 1986).
83. ࿜࿜࿜࿜, G. Woodruff, Behav. Brain Sci. 4, 515
(1978).
84. D. Premack, A. Premack, Original Intelligence
(McGraw-Hill, New York, 2002).
85. D. C. Dennett, Behav. Brain Sci. 6, 343 (1983).
86. B. Hare, J. Call, B. Agnetta, M. Tomasello, Anim.
Behav. 59, 771 (2000).
87. B. Hare, J. Call, M. Tomasello, Anim. Behav. 61, 139
(2001).
88. C. M. Heyes, Behav. Brain Sci. 21, 101 (1998).
89. D. J. Povinelli, T. J. Eddy, Monogr. Soc. Res. Child
Dev. 247 (1996).
90. R. M. Seyfarth, D. L. Cheney, P. Marler, Science 210,
801 (1980).
91. W. P. G. Dittus, Anim. Behav. 32, 470 (1984).
92. C. S. Evans, P. Marler, in Comparative Approaches to
Cognitive Science, H. Roitblatt, Ed. (MIT Press, Cambridge,
MA, 1995), pp. 241–282.
93. J. Fischer, Anim. Behav. 55, 799 (1998).
94. S. Gouzoules, H. Gouzoules, P. Marler, Anim. Behav.
32, 182 (1984).
95. M. D. Hauser, Anim. Behav. 55, 1647 (1998).
96. C. N. Slobodchikoff, J. Kiriazis, C. Fischer, E. Creef,
Anim. Behav. 42, 713 (1991).
97. K. Zuberbuhler, D. L. Cheney, R. M. Seyfarth,
J. Comp. Psychol. 113, 33 (1999).
98. P. Bloom, L. Markson, Trends Cognit. Sci. 2, 67
(1998).
99. P. Bloom, How Children Learn the Meanings of
Words (MIT Press, Cambridge, MA, 2000).
100. E. M. Gold, Inform. Control 10, 447 (1967).
101. S. Dehaene, The Number Sense (Oxford Univ. Press,
Oxford, 1997).
102. C. R. Gallistel, R. Gelman, Trends Cognit. Sci. 4, 59
(2000).
103. S. Carey, Mind Lang. 16, 37 (2001).
104. S. T. Boysen, G. G. Bernston, J. Comp. Psychol. 103,
23 (1989).
105. N. Kawai, T. Matsuzawa, Nature 403, 39 (2000).
106. T. Matsuzawa, Nature 315, 57 (1985).
107. K. Wynn, Cognit. Psychol. 24, 220 (1992).
108. N. Chomsky, Logical Structure of Linguistic Theory/
Excerpted Manuscript (Plenum, New York, 1975).
109. ࿜࿜࿜࿜, IRE Trans. Inform. Theory 2 (no. 2), 113
(1956).
110. ࿜࿜࿜࿜, G. Miller, Inform. Control 1, 91 (1958).
111. Z. S. Harris, Language 31, 190 (1955).
112. J. R. Saffran, R. N. Aslin, E. L. Newport, Science 274,
1926 (1996).
113. J. Saffran, E. Johnson, R. N. Aslin, E. Newport, Cognition
70, 27 (1999).
114. M. D. Hauser, E. L. Newport, R. N. Aslin, Cognition
78, B53 (2001).
115. G. Marcus, S. Vijayan, S. Bandi Rao, P. M. Vishton,
Science 283, 77 (1999).
116. M. D. Hauser, D. Weiss, G. Marcus, Cognition 86,
B15 (2002).
117. W. T. Fitch, M. D. Hauser, in preparation.
118. N. S. Clayton, A. Dickinson, Nature 395, 272 (1998).
119. C. R. Gallistel, A. E. Cramer, J. Exp. Biol. 199, 211
(1996).
120. P. Kuhl, Percept. Psychophys. 50, 93 (1991).
121. P. F. MacNeilage, Behav. Brain Sci. 21, 499(1998).
122. M. Studdert-Kennedy, in Approaches to the Evolution
of Language: Social and Cognitive Bases, J. R.
Hurford, M. Studdert-Kennedy, C. Knight, Eds.
(Cambridge Univ. Press, Cambridge, 1998), pp. 202–
221.
123. H. McGurk, J. MacDonald, Nature 264, 746 (1976).
124. L. R. Santos, G. M. Sulkowski, G. M. Spaepen, M. D.
Hauser, Cognition 83, 241 (2002).
125. G. Gergerly, H. Bekkering, I. Kiraly, Nature 415, 755
(2002).
126. A. N. Meltzoff, M. K. Moore, Infant Behav. Dev. 17,
83 (1994).
127. A. Whiten, D. Custance, in Social Learning in Animals:
The Roots of Culture, C. M. Heyes, J. B. G.
Galef, Eds. (Academic Press, San Diego, CA, 1996),
pp. 291–318.
128. P. Marler, S. Karakashian, M. Gyger, in Cognitive
Ethology: The Minds of Other Animals, C. Ristau, Ed.
(Erlbaum, Hillsdale, NJ, 1991), pp. 135–186.
129. H. S. Terrace, L. K. Son, E. M. Brannon, Psychol. Sci.,
in press.
130. L. M. Herman, D. G. Richards, J. P. Wolz, Cognition
16, 129 (1984).
131. E. S. Savage-Rumbaugh et al., Monogr. Soc. Res.
Child Dev. 58 (1993).
132. E. M. Brannon, H. S. Terrace, Science 282, 746
(1998).
133. F. Lerdahl, R. Jackendoff, A Generative Theory of
Tonal Music (MIT Press, Cambridge, MA, 1983).
134. N. Wallin, B. Merker, S. D. Brown, The Origins of
Music (MIT Press, Cambridge, MA, 2000).
135. R. Zatorre, I. Peretz, The Biological Foundations of
Music (National Academy Press, New York,
2000).
136. For comments on an earlier draft of the manuscript,
we thank D. Cheney, R. Jackendoff, L. Jenkins, M.
Nowak, M. Piatelli-Palmerini, S. Pinker, and R.
Seyfarth.
S C I E N C E ’ S C O M P A S S
www.sciencemag.org SCIENCE VOL 298 22 NOVEMBER 2002 1579