DOI: 10.1126/science.298.5598.1569 , 1569 (2002); 298 Science et al. Marc D. Hauser, How Did It Evolve? The Faculty of Language: What Is It, Who Has It, and www.sciencemag.org (this information is current as of August 23, 2007 ): The following resources related to this article are available online at http://www.sciencemag.org/cgi/content/full/298/5598/1569 version of this article at: including high-resolution figures, can be found in the online Updated information and services, found at: can be related to this article A list of selected additional articles on the Science Web sites http://www.sciencemag.org/cgi/content/full/298/5598/1569#related-content http://www.sciencemag.org/cgi/content/full/298/5598/1569#otherarticles , 10 of which can be accessed for free: cites 72 articles This article 262 article(s) on the ISI Web of Science. cited by This article has been http://www.sciencemag.org/cgi/content/full/298/5598/1569#otherarticles 26 articles hosted by HighWire Press; see: cited by This article has been http://www.sciencemag.org/cgi/collection/neuroscience Neuroscience : subject collections This article appears in the following http://www.sciencemag.org/about/permissions.dtl in whole or in part can be found at: this article permission to reproduce of this article or about obtaining reprints Information about obtaining registered trademark of AAAS. c 2002 by the American Association for the Advancement of Science; all rights reserved. The title SCIENCE is a Copyright American Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. Science (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by the on August 23, 2007 www.sciencemag.org Downloaded from
: subject collectionsThis article appears in the following
http://www.sciencemag.org/about/permissions.dtl in whole or in part can be found at: this article
permission to reproduce of this article or about obtaining reprintsInformation about obtaining
registered trademark of AAAS. c 2002 by the American Association for the Advancement of Science; all rights reserved. The title SCIENCE is a
CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. Science (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by the
The Faculty of Language: What Is It, Who HasIt, 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 substantialinterdisciplinary cooperation. We suggest how current developments in linguistics canbe profitably wedded to work in evolutionary biology, anthropology, psychology, andneuroscience. We submit that a distinction should be made between the faculty oflanguage in the broad sense (FLB) and in the narrow sense (FLN). FLB includes asensory-motor system, a conceptual-intentional system, and the computationalmechanisms for recursion, providing the capacity to generate an infinite range ofexpressions from a finite set of elements. We hypothesize that FLN only includesrecursion and is the only uniquely human component of the faculty of language. Wefurther argue that FLN may have evolved for reasons other than language, hencecomparative studies might look for evidence of such computations outside of thedomain of communication (for example, number, navigation, and social relations).
If a martian graced our planet, it would bestruck by one remarkable similarity amongEarth’s living creatures and a key difference.
Concerning similarity, it would note that allliving things are de-signed on the basis ofhighly conserved de-velopmental systemsthat read an (almost)universal language en-coded in DNA basepairs. As such, life isarranged hierarchical-ly with a foundationof discrete, unblend-able units (codons, and,for the most part,genes) capable of com-bining to create increas-ingly complex and vir-tually limitless varietiesof both species and in-dividual organisms. Incontrast, it would noticethe absence of a univer-sal code of communi-cation (Fig. 1).
If our martian nat-uralist were meticu-lous, it might notethat the faculty medi-ating human communication appears remark-ably different from that of other living crea-
tures; it might further note that the humanfaculty of language appears to be organizedlike the genetic code—hierarchical, genera-tive, recursive, and virtually limitless with
respect to its scope of expression. With thesepieces in hand, this martian might begin towonder how the genetic code changed in sucha way as to generate a vast number of mutu-ally incomprehensible communication sys-tems across species while maintaining clarityof comprehension within a given species. Themartian would have stumbled onto some ofthe essential problems surrounding the
question of language evolution, and of howhumans acquired the faculty of language.
In exploring the problem of language evo-lution, it is important to distinguish betweenquestions concerning language as a commu-nicative system and questions concerning thecomputations underlying this system, such asthose underlying recursion. As we argue be-low, many acrimonious debates in this fieldhave been launched by a failure to distinguishbetween these problems. According to oneview (1), questions concerning abstract com-putational mechanisms are distinct fromthose concerning communication, the lattertargeted at problems at the interface betweenabstract computation and both sensory-motorand conceptual-intentional interfaces. Thisview should not, of course, be taken as aclaim against a relationship between compu-
tation and communication. It is possible, aswe discuss below, that key computationalcapacities evolved for reasons other thancommunication but, after they proved to haveutility in communication, were altered be-cause of constraints imposed at both the pe-riphery (e.g., what we can hear and say or seeand sign, the rapidity with which the auditorycortex can process rapid temporal and spec-
1Department of Psychology, Harvard University,Cambridge, MA 02138, USA. 2Department of Linguis-tics and Philosophy, Massachusetts Institute of Tech-nology, Cambridge, MA 02138, USA.*To whom correspondence should be addressed. E-mail: [email protected]
Fig. 1. The animal kingdom has been designed on the basis of highly conserved developmental systems that read an almostuniversal language coded in DNA base pairs. This system is shown on the left in terms of a phylogenetic tree. In contrast, animalslack 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, pragmat-ics, memory limitations).
At least three theoretical issues cross-cutthe debate on language evolution. One of theoldest problems among theorists is the“shared versus unique” distinction. Most cur-rent commentators agree that, although beesdance, birds sing, and chimpanzees grunt,these systems of communication differ qual-itatively from human language. In particular,animal communication systems lack the richexpressive and open-ended power of humanlanguage (based on humans’ capacity for re-cursion). The evolutionary puzzle, therefore,lies in working out how we got from there tohere, given this apparent discontinuity. A sec-ond issue revolves around whether the evo-lution of language was gradual versus salta-tional; this differs from the first issue becausea qualitative discontinuity between extantspecies could have evolved gradually, involv-ing no discontinuities during human evolu-tion. Finally, the “continuity versus exapta-tion” issue revolves around the problem ofwhether human language evolved by gradualextension of preexisting communication sys-tems, or whether important aspects of lan-guage have been exapted away from theirprevious adaptive function (e.g., spatial ornumerical reasoning, Machiavellian socialscheming, tool-making).
Researchers have adopted extreme or in-termediate positions regarding these basically
independent questions, leading to a widevariety of divergent viewpoints on the evo-lution of language in the current literature.There is, however, an emerging consensusthat, although humans and animals share adiversity of important computational andperceptual resources, there has been sub-stantial evolutionary remodeling since wediverged from a common ancestor some 6million years ago. The empirical challengeis to determine what was inherited un-changed from this common ancestor, whathas been subjected to minor modifications,and what (if anything) is qualitatively new.The additional evolutionary challenge is todetermine what selectional pressures led toadaptive changes over time and to under-stand the various constraints that channeledthis evolutionary process. Answering thesequestions requires a collaborative effortamong linguists, biologists, psychologists,and anthropologists.
One aim of this essay is to promote astronger connection between biology andlinguistics by identifying points of contactand agreement between the fields. Al-though this interdisciplinary marriage wasinaugurated more than 50 years ago, it hasnot yet been fully consummated. We hopeto further this goal by, first, helping toclarify the biolinguistic perspective on lan-guage and its evolution (2–7). We thenreview some promising empirical ap-proaches to the evolution of the language
faculty, with a special focus oncomparative work with non-human animals, and concludewith a discussion of how in-quiry might profitably advance,highlighting some outstandingproblems.
We make no attempt to becomprehensive in our coverage ofrelevant or interesting topics andproblems. Nor is it our goal toreview the history of the field.Rather, we focus on topics thatmake important contact betweenempirical data and theoretical po-sitions about the nature of the lan-guage faculty. We believe that ifexplorations into the problem oflanguage evolution are to progress,we need a clear explication of thecomputational requirements forlanguage, the role of evolutionarytheory in testing hypotheses ofcharacter evolution, and a researchprogram that will enable a produc-tive interchange between linguistsand biologists.
Defining the Target: TwoSenses of the Faculty ofLanguage
The word “language” has highly divergentmeanings in different contexts and disci-plines. In informal usage, a language is un-derstood as a culturally specific communica-tion system (English, Navajo, etc.). In thevarieties of modern linguistics that concernus here, the term “language” is used quitedifferently to refer to an internal componentof the mind/brain (sometimes called “internallanguage” or “I-language”). We assume thatthis is the primary object of interest for thestudy of the evolution and function of thelanguage faculty. However, this biologicallyand individually grounded usage still leavesmuch open to interpretation (and misunder-standing). For example, a neuroscientistmight ask: What components of the humannervous system are recruited in the use oflanguage in its broadest sense? Because anyaspect of cognition appears to be, at least inprinciple, accessible to language, the broadestanswer to this question is, probably, “most ofit.” Even aspects of emotion or cognition notreadily verbalized may be influenced by lin-guistically based thought processes. Thus,this conception is too broad to be of muchuse. We therefore delineate two more restrict-ed conceptions of the faculty of language, onebroader and more inclusive, the other morerestricted and narrow (Fig. 2).
Faculty of language— broad sense(FLB). FLB includes an internal computa-tional system (FLN, below) combined withat least two other organism-internal sys-
Fig. 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); FLNincludes the core grammatical computations that we suggest are limited to recursion. See text for morecomplete 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 onthe precise nature of these systems, andabout whether they are substantially sharedwith other vertebrates or uniquely adaptedto the exigencies of language, we take asuncontroversial the existence of some bio-logical capacity of humans that allows us(and not, for example, chimpanzees) toreadily master any human language withoutexplicit instruction. FLB includes this ca-pacity, but excludes other organism-internal systems that are necessary but notsufficient for language (e.g., memory, res-piration, digestion, circulation, etc.).
Faculty of language—narrow sense(FLN). FLN is the abstract linguistic compu-tational system alone, independent of the oth-er systems with which it interacts and inter-faces. FLN is a component of FLB, and themechanisms underlying it are some subset ofthose underlying FLB.
Others have agreed on the need for arestricted sense of “language” but have sug-gested different delineations. For example,Liberman and his associates (8) have arguedthat the sensory-motor systems were specifi-cally adapted for language, and hence shouldbe considered part of FLN. There is also along tradition holding that the conceptual-intentional systems are an intrinsic part oflanguage in a narrow sense. In this article, weleave these questions open, restricting atten-tion to FLN as just defined but leaving thepossibility of a more inclusive definitionopen to further empirical research.
The internal architecture of FLN, so con-ceived, is a topic of much current researchand debate (4 ). Without prejudging the is-sues, we will, for concreteness, adopt a par-ticular conception of this architecture. Weassume, putting aside the precise mecha-nisms, that a key component of FLN is acomputational system (narrow syntax) thatgenerates internal representations and mapsthem into the sensory-motor interface by thephonological system, and into the conceptu-al-intentional interface by the (formal) se-mantic system; adopting alternatives thathave been proposed would not materiallymodify the ensuing discussion. All approach-es agree that a core property of FLN is recur-sion, attributed to narrow syntax in the con-ception just outlined. FLN takes a finite set ofelements and yields a potentially infinite ar-ray of discrete expressions. This capacity ofFLN yields discrete infinity (a property thatalso characterizes the natural numbers). Eachof these discrete expressions is then passed tothe sensory-motor and conceptual-intentionalsystems, which process and elaborate thisinformation in the use of language. Eachexpression is, in this sense, a pairing of soundand meaning. It has been recognized for thou-sands of years that language is, fundamental-
ly, a system of sound-meaning connections;the potential infiniteness of this system hasbeen explicitly recognized by Galileo, Des-cartes, and the 17th-century “philosophicalgrammarians” and their successors, notablyvon Humboldt. One goal of the study of FLNand, more broadly, FLB is to discover justhow the faculty of language satisfies thesebasic and essential conditions.
The core property of discrete infinity isintuitively familiar to every language user.Sentences are built up of discrete units: Thereare 6-word sentences and 7-word sentences,but no 6.5-word sentences. There is no long-est sentence (any candidate sentence can betrumped by, for example, embedding it in“Mary thinks that . . .”), and there is no non-
arbitrary upper bound to sentence length. Inthese respects, language is directly analogousto the natural numbers (see below).
At a minimum, then, FLN includes the ca-pacity of recursion. There are many organism-internal factors, outside FLN or FLB, that im-pose practical limits on the usage of the system.For example, lung capacity imposes limits onthe length of actual spoken sentences, whereasworking memory imposes limits on the com-plexity of sentences if they are to be under-standable. Other limitations—for example, onconcept formation or motor output speed—represent aspects of FLB, which have their ownevolutionary histories and may have played arole in the evolution of the capacities of FLN.Nonetheless, one can profitably inquire into the
evolution of FLN without animmediate concern for theselimiting aspects of FLB. Thisis made clear by the observa-tion that, although manyaspects of FLB are sharedwith other vertebrates, thecore recursive aspect of FLNcurrently appears to lack anyanalog in animal communi-cation and possibly other do-mains as well. This point,therefore, represents thedeepest challenge for a com-parative evolutionary ap-proach to language. We be-lieve that investigations ofthis capacity should includedomains other than commu-nication (e.g., number, socialrelationships, navigation).
Given the distinctionsbetween FLB and FLN andthe theoretical distinctionsraised above, we can definea research space as sketchedin Fig. 3. This researchspace identifies, as viable,problems concerning theevolution of sensory-motorsystems, of conceptual-in-tentional systems, and ofFLN. The comparative ap-proach, to which we turnnext, provides a frameworkfor addressing questionsabout each of these com-ponents of the faculty oflanguage.
The ComparativeApproach to LanguageEvolutionThe empirical study of theevolution of language is be-set with difficulties. Lin-guistic behavior does notfossilize, and a long tradi-
Fig. 3. Investigations into the evolution of the faculty of languageare confronted with a three-dimensional research space thatincludes three comparative-evolutionary problems cross-cut bythe core components of the faculty of language. Thus, for eachproblem, researchers can investigate details of the sensory-motorsystem, the conceptual-intentional system, FLN, and the interfac-es 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 andcranial endocasts has led to little consensusabout the evolution of language (7, 9). Amore tractable and, we think, powerful ap-proach to problems of language evolution isprovided by the comparative method, whichuses empirical data from living species todraw detailed inferences about extinct ances-tors (3, 10–12). The comparative method wasthe primary tool used by Darwin (13, 14 ) toanalyze evolutionary phenomena and contin-ues to play a central role throughout modernevolutionary biology. Although scholars in-terested in language evolution have often ig-nored comparative data altogether or focusednarrowly on data from nonhuman primates,current thinking in neuroscience, molecularbiology, and developmental biology indicatesthat many aspects of neural and developmen-tal function are highly conserved, encourag-ing the extension of the comparative methodto all vertebrates (and perhaps beyond). Forseveral reasons, detailed below, we believethat the comparative method should play amore central role in future discussions oflanguage evolution.
An overarching concern in studies of lan-guage evolution is with whether particularcomponents of the faculty of languageevolved specifically for human language and,therefore (by extension), are unique to hu-mans. Logically, the human uniqueness claimmust be based on data indicating an absenceof the trait in nonhuman animals and, to betaken seriously, requires a substantial body ofrelevant comparative data. More concretely,if the language evolution researcher wishes tomake the claim that a trait evolved uniquelyin humans for the function of language pro-cessing, data indicating that no other animalhas this particular trait are required.
Although this line of reasoning may ap-pear obvious, it is surprisingly common for atrait to be held up as uniquely human beforeany appropriate comparative data are avail-able. A famous example is categorical per-ception, which when discovered seemed sofinely tuned to the details of human speech asto constitute a unique human adaptation (15,16 ). It was some time before the same under-lying perceptual discontinuities were discov-ered in chinchillas and macaques (17, 18),and even birds (19), leading to the oppositeconclusion that the perceptual basis for cate-gorical perception is a primitive vertebratecharacteristic that evolved for general audito-ry processing, as opposed to specific speechprocessing. Thus, a basic and logically in-eliminable role for comparative research onlanguage evolution is this simple and essen-tially negative one: A trait present in nonhu-man animals did not evolve specifically forhuman language, although it may be part ofthe language faculty and play an intimate rolein language processing. It is possible, of
course, that a trait evolved in nonhuman an-imals and humans independently, as analogsrather than homologs. This would preservethe possibility that the trait evolved for lan-guage in humans but evolved for some otherreason in the comparative animal group. Incases where the comparative group is a non-human primate, and perhaps especially chim-panzees, the plausibility of this evolutionaryscenario is weaker. In any case, comparativedata are critical to this judgment.
Despite the crucial role of homology incomparative biology, homologous traits are notthe only relevant source of evolutionary data.The convergent evolution of similar charactersin two independent clades, termed “analogies”or “homoplasies,” can be equally revealing(20). The remarkably similar (but nonhomolo-gous) structures of human and octopus eyesreveal the stringent constraints placed by thelaws of optics and the contingencies of devel-opment on an organ capable of focusing a sharpimage onto a sheet of receptors. Detailed anal-ogies between the parts of the vertebrate andcephalopod eye also provide independent evi-dence that each component is an adaptation forimage formation, shaped by natural selection.Furthermore, the discovery that remarkablyconservative genetic cascades underlie the de-velopment of such analogous structures pro-vides important insights into the ways inwhich developmental mechanisms canchannel evolution (21). Thus, although po-tentially misleading for taxonomists, anal-ogies provide critical data about adaptationunder physical and developmental con-straints. Casting the comparative net morebroadly, therefore, will most likely reveallarger regularities in evolution, helping toaddress the role of such constraints in theevolution of language.
An analogy recognized as particularly rele-vant to language is the acquisition of song bybirds (12). In contrast to nonhuman primates,where the production of species-typical vocal-izations is largely innate (22), most songbirdslearn their species-specific song by listening toconspecifics, and they develop highly aberrantsong if deprived of such experience. Currentinvestigation of birdsong reveals detailed andintriguing parallels with speech (11, 23, 24).For instance, many songbirds pass through acritical period in development beyond whichthey produce defective songs that no amount ofacoustic input can remedy, reminiscent of thedifficulty adult humans have in fully masteringnew languages. Further, and in parallel with thebabbling phase of vocalizing or signing humaninfants (25), young birds pass through a phaseof song development in which they spontane-ously produce amorphous versions of adultsong, termed “subsong” or “babbling.” Al-though the mechanisms underlying the acquisi-tion of birdsong and human language are clear-ly analogs and not homologs, their core com-
ponents share a deeply conserved neural anddevelopmental foundation: Most aspects ofneurophysiology and development—includingregulatory and structural genes, as well as neu-ron types and neurotransmitters—are sharedamong vertebrates. That such close parallelshave evolved suggests the existence of impor-tant constraints on how vertebrate brains canacquire large vocabularies of complex, learnedsounds. Such constraints may essentially forcenatural selection to come up with the samesolution repeatedly when confronted with sim-ilar problems.
Testing Hypotheses About theEvolution of the Faculty of LanguageGiven the definitions of the faculty of lan-guage, together with the comparative frame-work, we can distinguish several plausiblehypotheses about the evolution of its variouscomponents. Here, we suggest two hypothe-ses that span the diversity of opinion amongcurrent scholars, plus a third of our own.
Hypothesis 1: FLB is strictly homologousto animal communication. This hypothesisholds that homologs of FLB, including FLN,exist (perhaps in less developed or otherwisemodified form) in nonhuman animals (3, 10,26). This has historically been a popular hy-pothesis outside of linguistics and closelyallied fields, and has been defended by somein the speech sciences. According to thishypothesis, human FLB is composed of thesame functional components that underliecommunication in other species.
Hypothesis 2: FLB is a derived, uniquelyhuman adaptation for language. Accordingto this hypothesis, FLB is a highly complexadaptation for language, on a par with thevertebrate eye, and many of its core compo-nents can be viewed as individual traits thathave been subjected to selection and perfect-ed in recent human evolutionary history. Thisappears to represent the null hypothesis formany scholars who take the complexity oflanguage seriously (27, 28). The argumentstarts with the assumption that FLB, as awhole, is highly complex, serves the functionof communication with admirable effective-ness, and has an ineliminable genetic compo-nent. Because natural selection is the onlyknown biological mechanism capable of gen-erating such functional complexes [the argu-ment from design (29)], proponents of thisview conclude that natural selection hasplayed a powerful role in shaping many as-pects of FLB, including FLN, and, further,that many of these are without parallel innonhuman animals. Although homologousmechanisms may exist in other animals, thehuman versions have been modified by nat-ural selection to the extent that they can bereasonably 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 hy-pothesize that most, if not all, of FLB is basedon mechanisms shared with nonhuman animals(as held by hypothesis 1). In contrast, we sug-gest that FLN—the computational mechanismof recursion—is recently evolved and unique toour species (33, 34). According to this hypoth-esis, much of the complexity manifested inlanguage derives from complexity in the pe-ripheral components of FLB, especially thoseunderlying the sensory-motor (speech or sign)and conceptual-intentional interfaces, com-bined with sociocultural and communicativecontingencies. FLB as a whole thus has anancient evolutionary history, long predating theemergence of language, and a comparativeanalysis is necessary to understand this com-plex system. By contrast, according to recentlinguistic theory, the computations underlyingFLN may be quite limited. In fact, we proposein this hypothesis that FLN comprises only thecore computational mechanisms of recursion as
they appear in narrow syntax and the mappingsto the interfaces. If FLN is indeed this restrict-ed, this hypothesis has the interesting effect ofnullifying the argument from design, and thusrendering the status of FLN as an adaptationopen to question. Proponents of the idea thatFLN is an adaptation would thus need to supplyadditional data or arguments to support thisviewpoint.
The available comparative data on animalcommunication systems suggest that the facultyof language as a whole relies on some uniquelyhuman capacities that have evolved recently inthe approximately 6 million years since ourdivergence from a chimpanzee-like commonancestor (35). Hypothesis 3, in its strongestform, suggests that only FLN falls into thiscategory (34). By this hypothesis, FLB containsa wide variety of cognitive and perceptualmechanisms shared with other species, but onlythose mechanisms underlying FLN—particu-larly its capacity for discrete infinity—areuniquely human. This hypothesis suggests that
all peripheral components of FLB are sharedwith other animals, in more or less the sameform as they exist in humans, with differencesof quantity rather than kind (9, 34). What isunique to our species is quite specific to FLN,and includes its internal operations as well as itsinterface with the other organism-internal sys-tems of FLB.
Each of these hypotheses is plausible tosome degree. Ultimately, they can be distin-guished only by empirical data, much of whichis currently unavailable. Before reviewing someof the relevant data, we briefly consider somekey distinctions between them. From a compar-ative evolutionary viewpoint, an importantquestion is whether linguistic precursors wereinvolved in communication or in somethingelse. Proponents of both hypotheses 1 and 2posit a direct correspondence, by descent withmodification, between some trait involved inFLB in humans and a similar trait in anotherspecies; these hypotheses differ in whetherthe 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 systemVocal 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 ofaction-perception systems
Studies assessing whether mirror neurons, which provide a core substrate for theaction-perception system, may subserve gestural and (possibly) vocal imitation
(67, 68, 71)
Discriminating the sound patternsof language
Operant conditioning studies of the prototype magnet effect in macaques and starlings (52, 120)
Constraints imposed by vocal tractanatomy
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 productionand perception
Cross-modal perception and sign language in humans versus unimodal communication inanimals
(3, 25, 123)
FLB—conceptual-intentional systemTheory of mind, attribution ofmental states
Studies of the seeing/knowing distinction in chimpanzees (84, 86–89)
Capacity to acquire nonlinguisticconceptual 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 socialrelationships
(3, 78, 90, 91, 93,94, 97)
Imitation as a rational, intentionalsystem
Comparative studies of chimpanzees and human infants suggesting that only the latterread intentionality into action, and thus extract unobserved rational intent
(125–127)
Voluntary control over signalproduction as evidence ofintentional communication
Comparative studies that explore the relationship between signal production and thecomposition of a social audience
(3, 10, 92, 128)
FLN—recursionSpontaneous and training methodsdesigned to uncover constraintson rule learning
Studies of serial order learning and finite-state grammars in tamarins and macaques (114, 116, 117,129)
Sign or artificial language intrained apes and dolphins
Studies exploring symbol sequencing and open-ended combinatorial manipulation (130, 131)
Models of the faculty of languagethat attempt to uncover thenecessary and sufficientmechanisms
Game theory models of language acquisition, reference, and universal grammar (72–74)
Experiments with animals thatexplore the nature and contentof number representation
Operant conditioning studies to determine whether nonhuman primates can representnumber, including properties such as ordinality and cardinality, using suchrepresentations in conjunction with mathematical operands (e.g., add, divide)
Although many aspects of FLB very likelyarose in this manner, the important issue forthese hypotheses is whether a series of gradualmodifications could lead eventually to the ca-pacity of language for infinite generativity. De-spite the inarguable existence of a broadlyshared base of homologous mechanisms in-volved in FLB, minor modifications to thisfoundational system alone seem inadequate togenerate the fundamental difference—discreteinfinity—between language and all knownforms of animal communication. This claim isone of several reasons why we suspect thathypothesis 3 may be a productive way to char-acterize the problem of language evolution.
A primary issue separating hypotheses 2and 3 is whether the uniquely human capac-ities of FLN constitute an adaptation. Theviewpoint stated in hypothesis 2, especiallythe notion that FLN in particular is a highlyevolved adaptation, has generated much en-thusiasm recently [e.g., (36 )], especiallyamong evolutionary psychologists (37, 38).At present, however, we see little reason tobelieve either that FLN can be anatomizedinto many independent but interacting traits,each with its own independent evolutionaryhistory, or that each of these traits could havebeen strongly shaped by natural selection,given their tenuous connection to communi-cative efficacy (the surface or phenotypicfunction upon which selection presumablyacted).
We consider the possibility that certain spe-cific aspects of the faculty of language are“spandrels”—by-products of preexisting con-straints rather than end products of a history ofnatural selection (39). This possibility, whichopens the door to other empirical lines of inqui-ry, is perfectly compatible with our firm supportof the adaptationist program. Indeed, it followsdirectly from the foundational notion that adap-tation is an “onerous concept” to be invokedonly when alternative explanations fail (40).The question is not whether FLN in toto isadaptive. By allowing us to communicate anendless variety of thoughts, recursion is clearlyan adaptive computation. The question iswhether particular components of the function-ing of FLN are adaptations for language, spe-cifically acted upon by natural selection—or,even more broadly, whether FLN evolved forreasons other than communication.
An analogy may make this distinctionclear. The trunk and branches of trees arenear-optimal solutions for providing an indi-vidual tree’s leaves with access to sunlight.For shrubs and small trees, a wide variety offorms (spreading, spherical, multistalked,etc.) provide good solutions to this problem.For a towering rainforest canopy tree, how-ever, most of these forms are rendered im-possible by the various constraints imposedby the properties of cellulose and the prob-lems of sucking water and nutrients up to the
leaves high in the air. Some aspects of suchtrees are clearly adaptations channeled bythese constraints; others (e.g., the popping ofxylem tubes on hot days, the propensity to betoppled in hurricanes) are presumably un-avoidable by-products of such constraints.
Recent work on FLN (4, 41–43) suggeststhe possibility that at least the narrow-syntacticcomponent satisfies conditions of highly effi-cient computation to an extent previously unsus-pected. Thus, FLN may approximate a kind of“optimal solution” to the problem of linkingthe sensory-motor and conceptual-intentionalsystems. In other words, the generative process-es of the language system may provide anear-optimal solution that satisfies the interfaceconditions to FLB. Many of the details of lan-guage that are the traditional focus of linguisticstudy [e.g., subjacency, Wh- movement, theexistence of garden-path sentences (4, 44)] mayrepresent by-products of this solution, gener-ated automatically by neural/computationalconstraints and the structure of FLB—components that lie outside of FLN. Evennovel capacities such as recursion are imple-mented in the same type of neural tissue as therest of the brain and are thus constrained bybiophysical, developmental, and computation-al factors shared with other vertebrates. Hy-pothesis 3 raises the possibility that structuraldetails of FLN may result from such preexistingconstraints, rather than from direct shaping bynatural selection targeted specifically at com-munication. Insofar as this proves to be true,such structural details are not, strictly speaking,adaptations at all. This hypothesis and thealternative selectionist account are both viableand can eventually be tested with comparativedata.
Comparative Evidence for the Facultyof LanguageStudy of the evolution of language has accel-erated in the past decade (45, 46 ). Here, weoffer a highly selective review of some ofthese studies, emphasizing animal work thatseems particularly relevant to the hypothesesadvanced above; many omissions were nec-essary for reasons of space, and we firmlybelieve that a broad diversity of methods andperspectives will ultimately provide the rich-est answers to the problem of language evo-lution. For this reason, we present a broadersampler of the field’s offerings in Table 1.
How “special” is speech? Comparativestudy of the sensory-motor system. Starting withearly work on speech perception, there has beena tradition of considering speech “special,” andthus based on uniquely human mechanismsadapted for speech perception and/or produc-tion [e.g., (7, 8, 47, 48)]. This perspective hasstimulated a vigorous research program study-ing animal speech perception and, more recent-ly, speech production. Surprisingly, this re-search has turned up little evidence for uniquely
human mechanisms special to speech, despite apersistent tendency to assume uniqueness evenin the absence of relevant animal data.
On the side of perception, for example,many species show an impressive ability toboth discriminate between and generalizeover human speech sounds, using formants asthe critical discriminative cue (17–19, 49–51). These data provide evidence not only ofcategorical perception, but also of the abilityto discriminate among prototypical exem-plars of different phonemes (52). Further, inthe absence of training, nonhuman primatescan discriminate sentences from two differentlanguages on the basis of rhythmic differenc-es between them (53).
On the side of production, birds and non-human primates naturally produce and per-ceive formants in their own species-typicalvocalizations (54–59). The results also shedlight on discussions of the uniquely humanstructure of the vocal tract and the unusualdescended larynx of our species (7, 48, 60),because new evidence shows that several oth-er mammalian species also have a descendedlarynx (61). Because these nonhuman specieslack speech, a descended larynx clearly hasnonphonetic functions; one possibility is ex-aggerating apparent size. Although this par-ticular anatomical modification undoubtedlyplays an important role in speech productionin modern humans, it need not have firstevolved for this function. The descended lar-ynx may thus be an example of classic Dar-winian preadaptation.
Many phenomena in human speech percep-tion have not yet been investigated in animals[e.g., the McGurk effect, an illusion in whichthe syllable perceived from a talking head rep-resents the interaction between an articulatorygesture seen and a different syllable heard; see(62)]. However, the available data suggest amuch stronger continuity between animals andhumans with respect to speech than previouslybelieved. We argue that the continuity hypoth-esis thus deserves the status of a null hypothe-sis, which must be rejected by comparativework before any claims of uniqueness can bevalidated. For now, this null hypothesis of notruly novel traits in the speech domain appearsto stand.
There is, however, a striking ability tied tospeech that has received insufficient atten-tion: the human capacity for vocal imitation(63, 64). Imitation is obviously a necessarycomponent of the human capacity to acquirea shared and arbitrary lexicon, which is itselfcentral to the language capacity. Thus, thecapacity to imitate was a crucial prerequisiteof FLB as a communicative system. Vocalimitation and learning are not uniquely hu-man. Rich multimodal imitative capacitiesare seen in other mammals (dolphins) andsome birds (parrots), with most songbirdsexhibiting 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 mon-keys show almost no evidence of visuallymediated imitation, with chimpanzees show-ing only slightly better capacities (66 ). Evenmore striking is the virtual absence of evi-dence for vocal imitation in either monkeysor apes (3). For example, intensively trainedchimpanzees are incapable of acquiring any-thing but a few poorly articulated spokenwords, whereas parrots can readily acquire alarge vocal repertoire. With respect to theirown vocalizations, there are few convincingstudies of vocal dialects in primates, therebysuggesting that they lack a vocal imitativecapacity (3, 65). Evidence for spontaneousvisuomanual imitation in chimpanzees is notmuch stronger, although with persistent train-ing they can learn several hundred handsigns. Further, even in cases wherenonhuman animals are capable of im-itating in one modality (e.g., songcopying in songbirds), only dolphinsand humans appear capable of imita-tion in multiple modalities. The de-tachment from modality-specific in-puts may represent a substantialchange in neural organization, onethat affects not only imitation butalso communication; only humanscan lose one modality (e.g., hear-ing) and make up for this deficit bycommunicating with complete com-petence in a different modality (i.e.,signing).
Our discussion of limitations isnot meant to diminish the impressiveachievements of monkeys and apes,but to highlight how different themechanisms underlying the produc-tion of human and nonhuman primategestures, either vocally expressed orsigned, must be. After all, the aver-age high school graduate knows up to60,000 words, a vocabulary achievedwith little effort, especially whencontrasted with the herculean effortsdevoted to training animals. In sum,the impressive ability of any normalhuman child for vocal imitation mayrepresent a novel capacity thatevolved in our recent evolutionaryhistory, some time after the diver-gence from our chimpanzee-like an-cestors. The existence of analogs indistantly related species, such asbirds and cetaceans, suggests consid-erable potential for the detailed com-parative study of vocal imitation. There are,however, potential traps that must be avoid-ed, especially with respect to explorations ofthe neurobiological substrates of imitation.For example, although macaque monkeysand humans are equipped with so-called“mirror neurons” in the premotor cortex thatrespond both when an individual acts in a
particular way and when the same individualsees someone else act in this same way (67,68), these neurons are not sufficient for imi-tation in macaques, as many have presumed:As mentioned, there is no convincing evi-dence of vocal or visual imitation in mon-keys. Consequently, as neuroimaging studiescontinue to explore the neural basis of imita-tion in humans (69–71), it will be importantto distinguish between the necessary and suf-ficient neural correlates of imitation. This isespecially important, given that some recentattempts to model the evolution of languagebegin with a hypothetical organism that isequipped with the capacity for imitation andintentionality, as opposed to working out howthese mechanisms evolved in the first place[see below; (72–74 )]. If a deeper evolution-
ary exploration is desired, one dating back toa chimpanzee-like ancestor, then we need toexplain how and why such capacitiesemerged from an ancestral node that lackedsuch abilities (75) (Fig. 4).
The conceptual-intentional systems of non-linguistic animals. A wide variety of studiesindicate that nonhuman mammals and birds
have rich conceptual representations (76, 77).Surprisingly, however, there is a mismatch be-tween the conceptual capacities of animals andthe communicative content of their vocal andvisual signals (78, 79). For example, although awide variety of nonhuman primates have accessto rich knowledge of who is related to whom, aswell as who is dominant and who is subordi-nate, their vocalizations only coarsely expresssuch complexities.
Studies using classical training approach-es as well as methods that tap spontaneousabilities reveal that animals acquire and use awide range of abstract concepts, includingtool, color, geometric relationships, food, andnumber (66, 76–82). More controversially,but of considerable relevance to intentionalaspects of language and conditions of felici-
tous use, some studies claim that animalshave a theory of mind (83–85), including asense of self and the ability to represent thebeliefs and desires of other group members.On the side of positive support, recent studiesof chimpanzees suggest that they recognizethe perceptual act of seeing as a proxy for themental 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 andmonkeys, 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, butperhaps no other nonhuman animals, have arudimentary theory of mind. On the side ofnegative support, other studies suggest thateven chimpanzees lack a theory of mind,failing, for example, to differentiate betweenignorant and knowledgeable individuals withrespect to intentional communication (88,89). Because these experiments make use ofdifferent methods and are based on smallsample sizes, it is not possible at present toderive any firm conclusions about the pres-ence or absence of mental state attribution inanimals. Independently of how this contro-versy is resolved, however, the best evidenceof referential communication in animalscomes not from chimpanzees but from a va-
riety of monkeys and birds, species for whichthere is no convincing evidence for a theoryof mind.
The classic studies of vervet monkeyalarm calls (90) have now been joined byseveral others, each using comparable meth-ods, with extensions to different species (ma-caques, Diana monkeys, meerkats, prairiedogs, chickens) and different communicativecontexts (social relationships, food, inter-group aggression) (91–97 ). From these stud-ies we can derive five key points relevant toour analysis of the faculty of language. First,individuals produce acoustically distinctivecalls in response to functionally importantcontexts, including the detection of predatorsand the discovery of food. Second, the acous-tic morphology of the signal, although arbi-
trary in terms of its association with a partic-ular context, is sufficient to enable listeners torespond appropriately without requiring anyother contextual information. Third, the num-ber of such signals in the repertoire is small,restricted to objects and events experiencedin the present, with no evidence of creativeproduction of new sounds for new situations.Fourth, the acoustic morphology of the callsis fixed, appearing early in development, withexperience only playing a role in refining therange of objects or events that elicit suchcalls. Fifth, there is no evidence that calling isintentional in the sense of taking into accountwhat other individuals believe or want.
Early interpretations of this work suggest-ed that when animals vocalize, they are func-
tionally referring to the objects and eventsthat they have encountered. As such, vervetalarm calls and rhesus monkey food calls, totake two examples, were interpreted as word-like, with callers referring to different kindsof predators or different kinds of food. Morerecent discussions have considerably weak-ened this interpretation, suggesting that if thesignal is referential at all, it is in the mind ofthe listener who can extract informationabout the signaler’s current context from theacoustic structure of the call alone (78, 95).Despite this evidence that animals can extractinformation from the signal, there are severalreasons why additional evidence is requiredbefore such signals can be considered as pre-cursors for, or homologs of, human words.
Roughly speaking, we can think of a partic-
ular human language as consisting of words andcomputational procedures (“rules”) for con-structing expressions from them. The computa-tional system has the recursive property brieflyoutlined earlier, which may be a distinct humanproperty. However, key aspects of words mayalso be distinctively human. There are, first ofall, qualitative differences in scale and mode ofacquisition, which suggest that quite differentmechanisms are involved; as pointed out above,there is no evidence for vocal imitation in non-human primates, and although human childrenmay use domain-general mechanisms to ac-quire and recall words (98, 99), the rate atwhich children build the lexicon is so massivelydifferent from nonhuman primates that onemust entertain the possibility of an indepen-
dently evolved mechanism. Further-more, unlike the best animal exam-ples of putatively referential signals,most of the words of human lan-guage are not associated with specif-ic functions (e.g., warning cries, foodannouncements) but can be linked tovirtually any concept that humanscan entertain. Such usages are oftenhighly intricate and detached fromthe here and now. Even for the sim-plest words, there is typically nostraightforward word-thing relation-ship, if “thing” is to be understood inmind-independent terms. Withoutpursuing the matter here, it appearsthat many of the elementary proper-ties of words—including those thatenter into referentiality—have onlyweak analogs or homologs in naturalanimal communication systems, withonly slightly better evidence from thetraining studies with apes and dol-phins. Future research must thereforeprovide stronger support for the pre-cursor position, or it must insteadabandon this hypothesis, arguing thatthis component of FLB (conceptual-intentional) is also uniquely human.
Discrete infinity and constraintson learning. The data summarized thus far,although far from complete, provide overallsupport for the position of continuity betweenhumans and other animals in terms of FLB.However, we have not yet addressed oneissue that many regard as lying at the heart oflanguage: its capacity for limitless expressivepower, captured by the notion of discreteinfinity. It seems relatively clear, after nearlya century of intensive research on animalcommunication, that no species other thanhumans has a comparable capacity to recom-bine meaningful units into an unlimited vari-ety of larger structures, each differing sys-tematically in meaning. However, littleprogress has been made in identifying thespecific capabilities that are lacking in otheranimals.
Fig. 5. Human and nonhuman animals exhibit the capacity to compute numerosities, including small precisenumber quantification and large approximate number estimation. Humans may be unique, however, in the abilityto show open-ended, precise quantificational skills with large numbers, including the integer count list. In parallelwith 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 anynatural language user can produce and under-stand has an important implication for lan-guage acquisition, long a core issue in devel-opmental psychology. A child is exposed toonly a small proportion of the possible sen-tences in its language, thus limiting its data-base for constructing a more general versionof that language in its own mind/brain. Thispoint has logical implications for any systemthat attempts to acquire a natural language onthe basis of limited data. It is immediatelyobvious that given a finite array of data, thereare infinitely many theories consistent with itbut inconsistent with one another. In thepresent case, there are in principle infinitelymany target systems (potential I-languages)consistent with the data of experience, andunless the search space and acquisition mech-anisms are constrained, selection amongthem is impossible. A version of the problemhas been formalized by Gold (100) and morerecently and rigorously explored by Nowakand colleagues (72–75). No known “generallearning mechanism” can acquire a naturallanguage solely on the basis of positive ornegative evidence, and the prospects for find-ing any such domain-independent deviceseem rather dim. The difficulty of this prob-lem leads to the hypothesis that whateversystem is responsible must be biased or con-strained in certain ways. Such constraintshave historically been termed “innate dispo-sitions,” with those underlying language re-ferred to as “universal grammar.” Althoughthese particular terms have been forcibly re-jected by many researchers, and the nature ofthe particular constraints on human (or ani-mal) learning mechanisms is currently unre-solved, the existence of some such con-straints cannot be seriously doubted. On theother hand, other constraints in animals musthave been overcome at some point in humanevolution to account for our ability to acquirethe unlimited class of generative systems thatincludes all natural languages. The nature ofthese latter constraints has recently becomethe target of empirical work. We focus hereon the nature of number representation andrule learning in nonhuman animals and hu-man infants, both of which can be investigat-ed independently of communication and pro-vide hints as to the nature of the constraintson FLN.
More than 50 years of research using clas-sical training studies demonstrates that ani-mals can represent number, with careful con-trols for various important confounds (80). Inthe typical experiment, a rat or pigeon istrained to press a lever x number of times toobtain a food reward. Results show that ani-mals can hit the target number to within aclosely matched mean, with a standard devi-ation that increases with magnitude: As thetarget number increases, so does variation
around the mean. These results have led tothe idea that animals, including human in-fants and adults, can represent number ap-proximately as a magnitude with scalar vari-ability (101, 102). Number discrimination islimited in this system by Weber’s law, withgreater discriminability among small num-bers than among large numbers (keeping dis-tances between pairs constant) and betweennumbers that are farther apart (e.g., 7 versus8 is harder than 7 versus 12). The approxi-mate number sense is accompanied by a sec-ond precise mechanism that is limited to val-ues less than 4 but accurately distinguishes 1from 2, 2 from 3, and 3 from 4; this secondsystem appears to be recruited in the contextof object tracking and is limited by workingmemory constraints (103). Of direct rele-vance to the current discussion, animals canbe trained to understand the meaning of num-ber words or Arabic numeral symbols. How-ever, these studies reveal striking differencesin how animals and human children acquirethe integer list, and provide further evidencethat animals lack the capacity to create open-ended generative systems.
Boysen and Matsuzawa have trainedchimpanzees to map the number of objectsonto a single Arabic numeral, to correctlyorder such numerals in either an ascending ordescending list, and to indicate the sums oftwo numerals (104–106 ). For example, Boy-sen shows that a chimpanzee seeing two or-anges placed in one box, and another twooranges placed in a second box, will pick thecorrect sum of four out of a lineup of threecards, each with a different Arabic numeral.The chimpanzees’ performance might sug-gest that their representation of number is likeours. Closer inspection of how these chim-panzees acquired such competences, howev-er, indicates that the format and content oftheir number representations differ funda-mentally from those of human children. Inparticular, these chimpanzees required thou-sands of training trials, and often years, toacquire the integer list up to nine, with noevidence of the kind of “aha” experience thatall human children of approximately 3.5years acquire (107 ). A human child who hasacquired the numbers 1, 2, and 3 (and some-times 4) goes on to acquire all the others; heor she grasps the idea that the integer list isconstructed on the basis of the successorfunction. For the chimpanzees, in contrast,each number on the integer list required thesame amount of time to learn. In essence,although the chimpanzees’ understanding ofArabic numerals is impressive, it parallelstheir understanding of other symbols andtheir referential properties: The system appar-ently never takes on the open-ended genera-tive property of human language. This limi-tation may, however, reveal an interestingquirk of the child’s learning environment and
a difference from the training regime of ani-mals: Children typically first learn an arbi-trary ordered list of symbols (“1, 2, 3, 4 . . . ”)and later learn the precise meaning of suchwords; apes and parrots, in contrast, weretaught the meanings one by one withoutlearning the list. As Carey (103) has argued,this may represent a fundamental differencein experience, a hypothesis that could betested by first training animals with an arbi-trary ordered list.
A second possible limitation on the classof learnable structures concerns the kinds ofstatistical inferences that animals can com-pute. Early work in computational linguistics(108–110) suggested that we can profitablythink about language as a system of rulesplaced within a hierarchy of increasing com-plexity. At the lowest level of the hierarchyare rule systems that are limited to localdependencies, a subcategory of so-called“finite-state grammars.” Despite their attrac-tive simplicity, such rule systems are inade-quate to capture any human language. Natu-ral languages go beyond purely local struc-ture by including a capacity for recursiveembedding of phrases within phrases, whichcan lead to statistical regularities that areseparated by an arbitrary number of words orphrases. Such long-distance, hierarchical re-lationships are found in all natural languagesfor which, at a minimum, a “phrase-structuregrammar” is necessary. It is a foundationalobservation of modern generative linguisticsthat, to capture a natural language, a grammarmust include such capabilities (Fig. 5).
Recent studies suggest that the capacity tocompute transitional probabilities—an exam-ple of a rule at the lowest level of the hierar-chy—might be available to human infantsand provide a mechanism for segmentingwords from a continuous acoustic stream(111–113). Specifically, after familiarizationto 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 proba-bility of appearing within the corpus, infantsare readily able to discriminate these tri-grams from others that are uncommon.Although this ability may provide a mech-anism for word segmentation, it is appar-ently not a mechanism that evolved unique-ly in humans or for language: The samecomputation is spontaneously available tohuman infants for visual sequences andtonal melodies (113), as well as to nonhu-man primates (cotton-top tamarins) testedwith the same methods and stimuli (114 ).Similarly, in the same way that humaninfants appear capable of computing alge-braic rules that operate over particular CVsequences (115), so too can cotton-toptamarins (116 ), again demonstrating thatthe 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, andalmost certainly did not evolve specificallyfor language.
Fitch and Hauser (117 ) recently complet-ed a study comparing finite-state and phrase-structure grammar acquisition in humanadults and tamarins, using the same subjectsand methods as the studies above. Thephrase-structure rule tested was AnBn, whereA and B were each represented by one of aset of eight different CVs. The rule thereforespecified both a set of consistent strings (nA’s must precede n B’s) and a set of incon-sistent strings; the latter consisted of viola-tions of order (B tokens precede A tokens) orof patterning (alternations of A’s and B’ssuch as ABAB). Results showed that humanadults rapidly learned this rule implicitly,distinguishing consistent and inconsistentstrings. Tamarins, in contrast, failed in threeseparate experiments testing their ability toacquire this grammar, but they readilymastered a finite-state variant (ABn) imple-mented with the same stimuli and testingconditions. This suggests that tamarins have alimited capacity to learn the type of long-distance hierarchical dependencies necessaryto achieve the class of phrase-structure gram-mars. If true, this limitation would place se-vere restrictions on their capacity to learn anynatural human language. It is currently un-clear whether this limitation generalizes toother animals, and whether it is similarlyimposed on humans at different stages ofdevelopment. Nonetheless, such experimentsprovide an empirical approach to exploringkey differences between humans and animalsrelevant to FLN.
Our review has stressed the usefulnessof animal data for theories about humans,but this exchange need not be one-way. Asthe research program we have sketchedprogresses, more general principles aboutcognitive evolution may emerge. For exam-ple, suppose we adopt the conception ofhypothesis 3, oversimplifying radically,that the interface systems—sensory-motorand conceptual-intentional—are given, andthe innovation that yielded the faculty oflanguage was the evolution of the compu-tational system that links them. The com-putational system must (i) construct an in-finite array of internal expressions from thefinite resources of the conceptual-intention-al system, and (ii) provide the means toexternalize and interpret them at the senso-ry-motor end. We may now ask to whatextent the computational system is optimal,meeting natural conditions of efficientcomputation such as minimal search and nobacktracking. To the extent that this can beestablished, we will be able to go beyondthe (extremely difficult, and still distant)accomplishment of finding the principles ofthe faculty of language, to an understanding
of why the faculty follows these particularprinciples and not others. We would thenunderstand why languages of a certain classare attainable, whereas other imaginablelanguages are impossible to learn and sus-tain. Such progress would not only open thedoor to a greatly simplified and empiricallymore tractable evolutionary approach to thefaculty of language, but might also be moregenerally applicable to domains beyondlanguage in a wide range of species—per-haps especially in the domain of spatialnavigation and foraging, where problems ofoptimal search are relevant. For example,elegant studies of insects, birds, and pri-mates reveal that individuals often searchfor food by an optimal strategy, one involv-ing minimal distances, recall of locationssearched, and kinds of objects retrieved(77, 118, 119). Only after a concerted, mul-tidisciplinary attack on the problems oflanguage evolution, paralleling 40 years ofoptimal foraging research, will we learnwhether such similarities are more thansuperficial.
ConclusionsWe conclude by making three points. First, apractical matter: Linguists and biologists,along with researchers in the relevant branch-es of psychology and anthropology, canmove beyond unproductive theoretical debateto a more collaborative, empirically focusedand comparative research program aimed atuncovering both shared (homologous or anal-ogous) and unique components of the facultyof language. Second, although we have ar-gued that most if not all of FLB is shared withother species, whereas FLN may be unique tohumans, this represents a tentative, testablehypothesis in need of further empirical inves-tigation. Finally, we believe that a compara-tive approach is most likely to lead to newinsights about both shared and derived fea-tures, thereby generating new hypothesesconcerning the evolutionary forces that led tothe design of the faculty of language. Specif-ically, although we have said relatively littleabout the role of natural selection in shapingthe design features of FLN, we suggest thatby considering the possibility that FLNevolved for reasons other than language, thecomparative door has been opened in a newand (we think) exciting way.
Comparative work has generally focusedon animal communication or the capacity toacquire a human-created language. If, how-ever, one entertains the hypothesis that recur-sion evolved to solve other computationalproblems such as navigation, number quanti-fication, or social relationships, then it ispossible that other animals have such abili-ties, but our research efforts have been tar-geted at an overly narrow search space (Fig.3). If we find evidence for recursion in ani-
mals, but in a noncommunicative domain,then we are more likely to pinpoint the mech-anisms underlying this ability and the selec-tive pressures that led to it. This discovery, inturn, would open the door to another suite ofpuzzles: Why did humans, but no other ani-mal, take the power of recursion to create anopen-ended and limitless system of commu-nication? Why does our system of recursionoperate over a broader range of elements orinputs (e.g., numbers, words) than other ani-mals? One possibility, consistent with currentthinking in the cognitive sciences, is thatrecursion in animals represents a modularsystem designed for a particular function(e.g., navigation) and impenetrable with re-spect to other systems. During evolution, themodular and highly domain-specific systemof recursion may have become penetrable anddomain-general. This opened the way for hu-mans, perhaps uniquely, to apply the powerof recursion to other problems. This changefrom domain-specific to domain-general mayhave been guided by particular selective pres-sures, unique to our evolutionary past, or as aconsequence (by-product) of other kinds ofneural reorganization. Either way, these aretestable hypotheses, a refrain that highlightsthe importance of comparative approaches tothe faculty of language.
References and Notes1. N. Chomsky, Aspects of the Theory of Syntax (MITPress, Cambridge, MA, 1965).
2. """" , Reflections on Language (Pantheon, NewYork, 1975).
3. M. D. Hauser, The Evolution of Communication (MITPress, Cambridge, MA, 1996).
4. R. Jackendoff, Foundations of Language (OxfordUniv. 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 theWorld: Inside the Mind of Another Species (Univ. ofChicago 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 Rela-tion 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: Persis-tent Problems, R. B. Masterton, W. Hodos, H. Jerison,Eds. (Wiley, New York, 1976), pp. 175–179.
21. W. J. Gehring, Master Control Genes in Developmentand 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. (MITPress, Cambridge, MA, 1999), pp. 391–418.
23. P. Marler, J. Neurobiol. 33, 1 (1997).24. F. Nottebohm, in The Design of Animal Communi-cation, 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 toSpeech: 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, NewYork, 1986).
30. D. Bickerton, Species and Language (Univ. of Chica-go Press, Chicago, 1990).
31. R. Dunbar, Grooming, Gossip and the Evolution ofLanguage (Harvard Univ. Press, Cambridge, MA,1996).
32. D. Kimura, Neuromotor Mechanisms in Human Com-munication (Oxford Univ. Press, Oxford, 1993).
33. N. Chomsky, Rules and Representations (ColumbiaUniv. Press, New York, 1980).
34. M. D. Hauser, in Language, Brain, and CognitiveDevelopment: 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 Transi-tions of Evolution (Freeman, Oxford, 1995).
37. L. Barrett, R. Dunbar, J. Lycett, Human EvolutionaryPsychology (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 & BindingTheory (Blackwell, Oxford, 1991).
45. J. R. Hurford, M. Studdert-Kennedy, C. Knight, Eds.,Approaches to the Evolution of Language: Social andCognitive Bases (Cambridge Univ. Press, Cambridge,1998).
46. A. Wray, Ed., The Transition to Language (OxfordUniv. 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 Physi-ology 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 Lan-guage: 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 (OxfordUniv. 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, Neuroimage15, 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, Science291, 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 Lectureson Human Values, G. Peterson, Ed. (Univ. of UtahPress, 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 (MITPress, 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 theAnimal Language Controversy (MIT Press, Cam-bridge, 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. ChildDev. 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 toCognitive Science, H. Roitblatt, Ed. (MIT Press, Cam-bridge, 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 ofWords (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, Cog-
nition 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 Evolu-
tion 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 Ani-
mals: 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 CognitiveEthology: 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, Cognition16, 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 ofTonal Music (MIT Press, Cambridge, MA, 1983).
134. N. Wallin, B. Merker, S. D. Brown, The Origins ofMusic (MIT Press, Cambridge, MA, 2000).
135. R. Zatorre, I. Peretz, The Biological Foundations ofMusic (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
