hardwired and independent of theenvironment and how much theyare dependent on incominginformation. Infants aresophisticated learners and itseems that, in language andmusic, their perceptual abilitiesare driven by innate mechanismsand learning by experience [8]. Inboth domains their learning isdriven by the ability to extractstatistical regularity from stimuli.Their responses to strings ofwords and sounds depend on theprobability that one element of astring will follow another based onprevious experience. In otherwords, it seems that the infantsare making statistical inferencesregarding the external world.What we now need to know is theextent to which the algorithms forextracting these regularities andbuilding predictions are similar oreven shared at some stages ofdevelopment across domainssuch as music and language.

The issue is a deep one: Wemake predictions about visualobjects, about others’ emotionalresponses, about speech, musicand the movement of objects inthe world. The ground has shifted,then, from thinking abouthardwired knowledge to thinkingabout hardwired ways of

acquiring knowledge. Whetheradults can recapture the earlypower of these learningmechanisms or whether moredeveloped mechanisms can beadapted to learn things in differentways is also a question opened bythis line of work.

What of the adults in Hannonand Trehub’s experiment [7]?Would they have learned like the12 month olds if they just hadmore time, or had they missed acritical time window after whichthey could not use the samelearning mechanism? Again ahalfway house might be the rightplace to stop: it may not be thatthe window is slammed shut, butthat the adults continue to usepredictive learning mechanismsonly on a different, usually lessrich and more fixed body of neuralrepresentations than are availableto children. Adults can learn, ofcourse, but they have toovercome the limitations set bytheir early experience andsubsequent neural sensitivities.Balkan dance classes are nowopen for enrolment.

References1. Hubel, D.H., and Wiesel, T.N. (1963).

Receptive fields of cells in striate cortexof the very young, visually inexperiencedkittens. J. Neurophysiol. 26,994–1002.

2. Kuhl, P.K. (1994). Learning andrepresentation in speech and language.Curr. Opin. Neurobiol. 5, 812–822.

3. Werker, J.F., and Tees, R.C. (1984).Cross-language speech perception:Evidence for perceptual reorganisationduring the first year of life. Infant Behav.Dev. 7, 49–63.

4. Werker, J.F., and Lalonde, C.E. (1988).Cross-language speech perception:initial capabilities and developmentalchange. Dev. Psychol. 24, 672–683.

5. Lynch, M.P., Eilers, R.E., Oller, D.K., andUrbano, R.C. (1990). Innateness,experience, and music perception.Psychol. Sci. 1, 272–276.

6. Hannon, E.E., and Trehub, S.E. (2005).Metrical categories in infancy andadulthood. Psychol. Sci. 16, 48–55.

7. Hannon, E.E., and Trehub, S.E. (2005).Tuning in to musical rhythms: Infantslearn more readily than adults. Proc.Natl. Acad. Sci. USA 102, 12639–12643.

8. Streeter, L.A. (1976). Languageperception of two month old infantsshows effects of both innatemechanisms and experience. Nature259, 39–41.

9. Hirsch, H.V., and Spinelli, D.N. (1970).Visual experience modifies distribution ofhorizontally and vertically orientedreceptive fields in cats. Science 168,869–871.

10. Saffran, J.A., Aslin, R.N., and Newport,E.L. (1996). Statistical learning by eightmonth old infants. Science 274,1926–1928.

11. Saffran, J.A., Johnson, E.K., Aslin, R.N.,and Newport, E.L. (1999). Statisticallearning of tone sequences by humaninfants and adults. Cognition 70, 27–52.

1University of Newcastle and WellcomeDepartment of Imaging Neuroscience,University College London, London, UK.2Institute of Cognitive Neuroscienceand Department of Psychology,University College London, London, UK.

DOI: 10.1016/j.cub.2005.10.019

Current Biology Vol 15 No 21R884

Adam Clark Arcadi

Human language exhibits manyunique features compared withthe communication systems ofother animals. The most obviousof these is its expressive power— the grammatical structure oflanguage permits an infinitenumber of meaningful utterances[1]. Language behavior dependson the ability to model the mentalstates of conspecifics [2]. And theability to produce and processspeech involves specialized oro-

facial, respiratory and perceptualabilities [3]. But these potentiallyunique features are alsosupported by capacities thatshow continuities with otherspecies. For example, membersof some species partitioncontinuous acoustic variationcategorically, exhibit lateralizationin perceptual processing, requireauditory feedback to learnspecies-specific vocalizations,engage in timed vocalinteractions, vary call productiondepending on their audience, and

encode information aboutexternal events in their calls [4].

Tracing the evolution oflanguage requires clarifying thenature of continuities betweenhuman and nonhuman cognitivestructures and communicationsystems in order to specify likelypathways by which language’sunique features could haveemerged [5]. One intriguingresearch area, explored bySlocombe and Zuberbühler [6] in arecent issue of Current Biology,concerns the possibility that someanimal signals refer to objects orevents external to the signaler, andmay therefore be similar to words.The first evidence of suchreferential potential came from theobservation that wild vervetmonkeys (Cercopithecus aethiops)produce acoustically distinct alarmcalls in response to their threemost important predators —

Language Evolution: What DoChimpanzees Have to Say?

Although unique in important ways, language shares some propertieswith other animal communication systems. Comparative analyses ofnonhuman primate vocalizations can shed light on the evolution oflanguage’s special features.

eagles, leopards and snakes —and that listeners responddifferently to each [7].Broadcasting recordings of thesecalls in the absence of actualpredators elicits the samebehavioral responses fromlisteners as the predatorsthemselves, suggesting that thevocalizations in some way ‘standfor’ the predators [8]. Similarobservations have been made forthe alarm calls of ring-tailed lemurs(Lemur catta) [9], Diana andCampbell’s monkeys (C. diana, C.campbelli) [10], and domesticchickens (Gallus gallus) [11].

There are, however, importantways in which such calls differfrom human words. Human wordsare built according tophonological rules from a finitenumber of sound contrastscharacteristic of a given languagecommunity [12]. The consequenceof this structural property,whereby meaningless units arecombined into meaningful ones, isthat languages contain a vastnumber of words. By contrast,vocal repertoires in animals arelimited to a few dozen calls, andevidence for phonological rules isscarce [13]. In addition, unlike thealarm calls of some primatespecies, a large percentage ofwords refer to concepts ratherthan physical entities, or have noreferents at all, functioninginstead as grammatical items(articles, prepositions and soforth). Finally, there aresubstantial philosophical andempirical obstacles to specifyingwhat animal vocalizations mightmean to signalers and receivers,and thus how similar they couldbe cognitively to human words.For example, does hearing aleopard alarm call produce amental picture of a leopard in thevervet listener, or simply anescape routine [14]?

In recognition of the fact thatanimal signals cannot be thesame as words, the term‘functionally referential’ wascoined to characterize those callsthat appear ‘to encode sufficientinformation about referentcharacteristics to allowconspecific receivers to respondappropriately’ [15]. Two empiricalrequirements for demonstrating

functional reference arerecognized: high productionspecificity (a call is alwaysassociated with the samestimulus), and high responsespecificity in the absence of othercues (listener responses toplaybacks match responses toreferents) [16]. The empiricaldemonstration of functionalreference is most compellingwhen the vocalizations underconsideration are from differentcall types, rather than beingvariants of the same vocalization,and when listener responses differin kind rather than in degree.When these conditions are met, itis more likely that the motivationalcomponent of calls, which canoperate as a contextual cue tosupplement the referentialcomponent of signals, can beexperimentally controlled [15].

As they reported in CurrentBiology last month, Slocombe andZuberbühler [6] conductedexperimental playbacks ofchimpanzee food calls to acaptive chimpanzee in order toexamine the referential potentialof these calls (Figure 1). Manyanimals and birds are known toproduce specific vocalizationsupon discovery of food, andexperimental studies havesuggested that some of thesesignals meet the criteria offunctional reference [17]. Wild

chimpanzees produce one type ofcall, the ‘rough grunt,’ upondiscovering food and in no othercontexts, thus fulfilling therequirement of productionspecificity [18]. Experimentalstudies have shown that the rateof rough grunting depends on theamount and divisibility of foodpresent [19]. Field workers havenoted that rough grunts areacoustically variable, especially inbeing relatively high or lowpitched, but field observationshave not revealed differences infood quality or quantityassociated with this acousticvariability. Instead, acousticdifferences have been associatedwith social context [20]. Thus,suggestions of functionalreference for this call have beenrestricted to food discovery ingeneral.

Based on acoustic analyses of82 calls delivered in 19 bouts,Slocombe and Zuberbühler [6]were able to distinguish two roughgrunt variants produced by threecaptive chimpanzees: a higherpitched one given to a preferredfood type, and a lower pitchednoisier one given to a less prizedfood. The chimpanzees werehabituated to two feedingstations, a tree baited with thepreferred food and a tree baitedwith the less desirable food.Experimental playbacks (17 tests

Dispatch R885

Figure 1. Liberius, the juvenile male chimpanzee tested with food grunt playbacks [6].Photograph courtesy of Katie Slocombe.

and 10 controls) of the two roughgrunt variants elicited differentialsearching behavior by a juvenilemale. In four of the six initial trials,this male first searched the treecontaining the foodcorresponding to the playbackstimulus. In the 11 remainingtrials, the male searched the treewith the preferred food firstirrespective of playback stimulus,but searched more intensivelyafter hearing grunts given to thepreferred food. The authorsconclude that the two roughgrunts meet the criteria offunctional reference, containingsufficient information about thenature of food sources to guidetheir subject’s searching behavior.This conclusion goes beyondprevious suggestions of functionalreference for chimpanzee roughgrunts in claiming not just that therough grunt is functionallyreferential in general, but thatdifferent rough grunt variants referto different foods or foodpreferences.

Although the demonstration offunctional reference in achimpanzee vocalization wouldbe an important contribution tolanguage origins research,Slocombe and Zuberbühler’s [6]findings should be viewed withconsiderable caution for severalreasons. First, responses to theplaybacks were not sharplydifferentiated (low responsespecificity, unlike monkeyresponses to alarm calls), aproblem also encountered in theanalysis of rhesus monkey foodcalls [17]. Second, the authorswere unable to control forindividual, age, and sexdifferences in call structure. Fieldanalyses have indicated thatyounger individuals give higherpitched rough grunts [20]. In thisstudy, a subadult male and adultfemale contributed 6 of 9 bouts tothe preferred food, but only 5 of10 bouts to the less preferredfood. Moreover, it is unclear whateffect social context had on callproduction. Again, field analyseshave indicated that social

excitement affects call structure[20]. Since the hypothesizedreferential signals are variants ofthe same call type, it is thereforedifficult to evaluate thecontribution of motivational stateas a contextual cue embedded inthe vocal playback [15]. Takentogether, these ambiguities meanthat production specificity, likeresponse specificity, has notbeen clearly demonstrated.Larger sample sizes of bothcallers and test subjects areneeded to disentangle thepossible sources of variation inrough grunt acoustic structureand to isolate their effects onlisteners.

Despite these shortcomings,this first attempt to test forfunctional reference in achimpanzee vocalization shouldstimulate further research. Effortsin captivity can profit from theopportunity to manipulate keyaspects of social context andmotivational state in bothsenders and receivers. Renewedattention to food calling in thewild can help specify itsfunctional significance. Tracingthe evolution of language willalways be hampered by the factthat early hominid species thatlikely possessed proto-linguisticabilities are extinct. Nevertheless,it is a fair hypothesis that at leastsome of the substrate on whichlanguage is founded can beexplored through comparativeanalyses of the vocal behavior ofour closest phylogeneticrelatives.

References 1. Chomsky, N. (1980). Rules and

Representations. (New York: ColumbiaUniversity Press).

2. Seyfarth, R.M., and Cheney, D.L. (2003).Signalers and receivers in animalcommunication. Annu. Rev. Psychol. 54,145–173.

3. Lieberman, P. 1991. Uniquely Human:The Evolution of Speech, Thought, andSelfless Behavior. (Cambridge,Massachusetts: Harvard UniversityPress).

4. Evans, C.S., and Marler, P. (1995).Language and animal communication:Parallels and contrasts. In ComparativeApproaches to Cognitive Science. H.L.Roitblat and J-A. Meyer, eds.(Cambridge, Massachusetts: MIT Press),pp. 341–382.

5. Hauser, M.D., Chomsky, N., and Fitch,W.T. (2002). The faculty of language:What is it, who has it, and how did itevolve? Science 298, 1569–1579.

6. Slocombe, S., and Zuberbühler, K.(2005). Functionally referentialcommunication in a chimpanzee. Curr.Biol. 15, 1779–1784.

7. Struhsaker, T.T. (1968). Auditorycommunication among vervet monkeys(Cercopithecus aethiops). In SocialCommunication among Primates. S. A.Altmann, ed. (Chicago: University ofChicago Press), pp. 281–324.

8. Seyfarth, R.M., Cheney, D.L., and Marler,P. (1980). Monkey responses to threedifferent alarm calls: Evidence ofpredator classification and semanticcommunication. Science 210, 801–803.

9. Macedonia, J. (1990). What iscommunicated in the antipredator callsof lemurs: Evidence from playbackexperiments with ring-tailed lemurs andruffed lemurs. Ethology 86, 177–190.

10. Zuberbühler, K. (2003). Referentialsignaling in non-human primates:Cognitive precursors and limitations forthe evolution of language. Adv. Stud.Behav. 33, 265–307.

11. Evans, C.S., Evans, L., and Marler, P.(1993). On the meaning of alarm calls:Functional reference in an avian vocalsystem. Anim. Behav. 46, 23–38.

12. Jakobson, R., and Halle, M. (1980).Fundamentals of Language, 4th edn.(The Hague: Mouton).

13. Mitani, J.C., and Marler, P. (1989). Aphonological analysis of male gibbonsinging behavior. Behaviour 109, 20–45.

14. Owren, M.J., and Rendall, D. (2001).Sound on the rebound: Bringing formand function back to the forefront inunderstanding nonhuman primate vocalsignaling. Evol. Anthrop. 10, 58–71.

15. Marler, P., Evans, C.S., and Hauser, M.D.(1992). Animal signals: Motivational,referential, or both? In Nonverbal VocalCommunication. H. Papousek, U.Jürgens, and M. Papousek, eds.(Cambridge: Cambridge UniversityPress), pp. 66–86.

16. Macedonia, J.M., and Evans, C.S. (1993).Variation among mammalian callsystems and the problem of meaning inanimal signals. Ethology 93, 177–197.

17. Hauser, M.D. (1998). Functional referentsand acoustic similarity: Field playbackexperiments with rhesus monkeys. Anim.Behav. 55, 1647–1658.

18. Goodall, J. (1986). The Chimpanzees ofGombe: Patterns of Behavior.(Cambridge, Massachusetts andLondon: Belknap Press of HarvardUniversity Press).

19. Hauser, M., Teixidor, P., Field, L., andFlaherty, R. (1993). Food-elicited calls inchimpanzees: Effects of food quantityand divisibility. Anim. Behav. 45,817–819.

20. Marler, P., and Tenaza, R. (1977).Signaling behavior of apes with specialreference to vocalization. In HowAnimals Communicate. T. A. Sebeok, ed.(Bloomington: Indiana University Press),pp. 965–1032.

Department of Anthropology, 264McGraw Hall, Cornell University, Ithaca,New York 14853, USA.

DOI: 10.1016/j.cub.2005.10.020

Current Biology Vol 15 No 21R886