Pbio 1001934

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/265057343

How Could Language Have Evolved?

Article in PLoS Biology · August 2014


DOI: 10.1371/journal.pbio.1001934 · Source: PubMed

CITATIONS READS

231 5,260

4 authors, including:

Ian Tattersall Robert Berwick


American Museum of Natural History Massachusetts Institute of Technology
314 PUBLICATIONS 8,124 CITATIONS 131 PUBLICATIONS 5,441 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Cognitive Linguistics: Theories of Formal Semantics and Machine Knowledge Learning View project

All content following this page was uploaded by Ian Tattersall on 12 April 2016.

The user has requested enhancement of the downloaded file.


Essay

How Could Language Have Evolved?


Johan J. Bolhuis1,2*, Ian Tattersall3, Noam Chomsky4, Robert C. Berwick5
1 Cognitive Neurobiology and Helmholtz Institute, Departments of Psychology and Biology, Utrecht University, Utrecht, The Netherlands, 2 Department of Zoology and
Sidney Sussex College, University of Cambridge, Cambridge, United Kingdom, 3 Division of Anthropology, American Museum of Natural History, New York, New York,
United States of America, 4 Department of Linguistics and Philosophy, MIT, Cambridge, Massachusetts, United States of America, 5 Department of Electrical Engineering &
Computer Science and Brain and Cognitive Sciences, MIT, Cambridge, Massachusetts, United States of America

Abstract: The evolution of the what it can tell us about the mechanisms evolution of language (Box 1) [1,4].
faculty of language largely remains of language. Here we argue that the basic However, speech and speech perception,
an enigma. In this essay, we ask principle that underlies language’s hierar- while functioning as possible external
why. Language’s evolutionary chical syntactic structure is consistent interfaces for the language system, are
analysis is complicated because it with a relatively recent evolutionary not identical to it. An alternative external-
has no equivalent in any nonhu- emergence. ization of language is in the visual domain,
man species. There is also no as sign language [1]; even haptic external-
consensus regarding the essential Conceptualizations of ization by touch seems possible in deaf and
nature of the language ‘‘pheno- Language blind individuals [5]. Thus, while the
type.’’ According to the ‘‘Strong evolution of auditory-vocal learning may
Minimalist Thesis,’’ the key distin- The language faculty is often equated be relevant for the evolution of speech, it is
guishing feature of language (and with ‘‘communication’’—a trait that is not for the language faculty per se. We
what evolutionary theory must shared by all animal species and possibly maintain that language is a computational
explain) is hierarchical syntactic also by plants. In our view, for the cognitive mechanism that has hierarchical
structure. The faculty of language purposes of scientific understanding, lan- syntactic structure at its core [1], as
is likely to have emerged quite guage should be understood as a particular outlined in the next section.
recently in evolutionary terms, computational cognitive system, imple-
some 70,000–100,000 years ago, mented neurally, that cannot be equated
and does not seem to have under- The Faculty of Language
with an excessively expansive notion of According to the ‘‘Strong
gone modification since then,
‘‘language as communication’’ [1]. Exter-
though individual languages do of Minimalist Thesis’’
course change over time, operating nalized language may be used for com-
within this basic framework. The munication, but that particular function is In the last few years, certain linguistic
recent emergence of language and largely irrelevant in this context. Thus, the theories have arrived at a much more
its stability are both consistent with origin of the language faculty does not narrowly defined and precise phenotype
the Strong Minimalist Thesis, which generally seem to be informed by consid- characterizing human language syntax. In
has at its core a single repeatable erations of the evolution of communica- place of a complex rule system or accounts
operation that takes exactly two tion. This viewpoint does not preclude the grounded on general notions of ‘‘culture’’
syntactic elements a and b and possibility that communicative consider- or ‘‘communication,’’ it appears that
assembles them to form the set {a, b}. ations can play a role in accounting for the human language syntax can be defined
maintenance of language once it has in an extremely simple way that makes
appeared or for the historical language conventional evolutionary explanations
change that has clearly occurred within much simpler. In this view, human
It is uncontroversial that language has the human species, with all individuals language syntax can be characterized via
evolved, just like any other trait of living sharing a common language faculty, as a single operation that takes exactly two
organisms. That is, once—not so long ago some mathematical models indicate [1–3]. (syntactic) elements a and b and puts them
in evolutionary terms—there was no A similar misconception is that language is together to form the set {a, b}. We call this
language at all, and now there is, at least coextensive with speech and that the basic operation ‘‘merge’’ [1]. The ‘‘Strong
in Homo sapiens. There is considerably evolution of vocalization or auditory-vocal Minimalist Thesis’’ (SMT) [6] holds that
less agreement as to how language learning can therefore inform us about the merge along with a general cognitive
evolved. There are a number of reasons
for this lack of agreement. First, ‘‘lan-
Citation: Bolhuis JJ, Tattersall I, Chomsky N, Berwick RC (2014) How Could Language Have Evolved? PLoS
guage’’ is not always clearly defined, and Biol 12(8): e1001934. doi:10.1371/journal.pbio.1001934
this lack of clarity regarding the language
Published August 26, 2014
phenotype leads to a corresponding lack of
clarity regarding its evolutionary origins. Copyright: ß 2014 Bolhuis et al. This is an open-access article distributed under the terms of the Creative
Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium,
Second, there is often confusion as to the provided the original author and source are credited.
nature of the evolutionary process and
Funding: JJB is funded by Utrecht University and by Netherlands Organization for Scientific Research (NWO)
grants (ALW Open Competition and NWO Gravity and Horizon Programmes) (http://www.nwo.nl/). The funders
had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Essays articulate a specific perspective on a topic of
broad interest to scientists. Competing Interests: The authors have declared that no competing interests exist.
* Email: j.j.bolhuis@uu.nl

PLOS Biology | www.plosbiology.org 1 August 2014 | Volume 12 | Issue 8 | e1001934


Box 1. Comparative Linguistics: Not Much to Compare has been argued [15], motor sequences
resemble more the ‘‘sequence of letters in
A major stumbling block for the comparative analysis of language evolution is the alphabet than the sequences of words
that, so far, there is no evidence for human-like language syntax in any in a sentence’’ ([15], p. 221). (For expos-
nonhuman species [4,41,42]. There is no a priori reason why a version of such a itory purposes, we omit here several
combinatorial computational system could not have evolved in nonhuman technical linguistic details about the label-
animals, either through common descent (e.g., apes) or convergent evolution ling of these words; see [16].) Along with
(e.g., songbirds) [1,18]. Although the auditory-vocal domain is just one possible the conceptual atoms of the lexicon, the
external interface for language (with signing being another), it could be argued SMT holds that merge, plus the internal
that the strongest animal candidates for human-like syntax are songbirds and interface mappings to the conceptual
parrots [1,41,42]. Not only do they have a similar brain organization underlying system, yields what has been called the
auditory-vocal behavior [4,43,44], they also exhibit vocal imitation learning that ‘‘language of thought’’ [17].
proceeds in a very similar way to speech acquisition in human infants [4,41,42].
More narrowly, the SMT also suffices to
This ability is absent in our closest relatives, the great apes [1,4]. In addition, like
human spoken language, birdsong involves patterned vocalizations that can be automatically derive some of the most
quite complex, with a set of rules that govern variable song element sequences central properties of human language
known as ‘‘phonological syntax’’ [1,4,41,42,45]. Contrary to recent suggestions syntax. For example, one of the most
[46,47], to date there is no evidence to suggest that birdsong patterns exhibit the distinctive properties of human language
hierarchical syntactic structure that characterizes human language [41,48,49] or syntax is that of ‘‘displacement,’’ along
any mapping to a level forming a language of thought as in humans. Avian vocal- with what is sometimes called ‘‘duality of
learning species such as parrots are able to synchronize their behavior to variable semantic patterning.’’ For example, in the
rhythmic patterns [50]. Such rhythmic abilities may be involved in human sentence ‘‘(Guess) what boys eat,’’ ‘‘what’’
prosodic processing, which is known to be an important factor in language takes on a dual role and is interpreted in
acquisition [51]. two places: first, as a question ‘‘operator’’
at the front of the sentence, where it is
pronounced; and second, as a variable that
requirement for computationally minimal Thus, before the appearance of merge, serves as the argument of the verb eat, the
or efficient search suffices to account for there was no faculty of language as such, thing eaten, where it is not pronounced
much of human language syntax. The because this requires merge along with the (Figure 1). (There are marginal exceptions
SMT also requires two mappings: one to conceptual atoms of the lexicon. Absent to the nonpronunciation of the second
an internal conceptual interface for this, there is no way to arrive at the ‘‘what’’ that, when analyzed carefully,
thought and a second to a sensory-motor essentially infinite number of syntactic support the picture outlined here.) Given
interface that externalizes language as language structures, e.g., ‘‘the brown the free application of merge, we expect
speech, sign, or other modality [1]. The cow,’’ ‘‘a black cat behind the mat’’ [9– human languages to exhibit this phenom-
basic operation itself is simple. Given 11], etc. This view leaves room for the enon of displacement without any further
merge, two items such as the and apples possibility that some conceptual atoms stipulation. This is simply because operat-
are assembled as the set {the, apples}. were present antecedent to merge itself, ing freely, without any further constraints,
Crucially, merge can apply to the results of though at present this remains entirely merge derives this possibility. In our
its own output so that a further application speculative. Even if true, there seems to be example ‘‘(Guess) what boys eat,’’ we
of merge to ate and {the, apples} yields the no evidence for an antecedent combina- assume that successive applications of
set {ate, {the, apples}}, in this way torial and hierarchical syntax. Further- merge as in our earlier example will first
deriving the full range of characteristic more, merge itself is uniform in the derive {boys, {eat, what}}—analogous to
hierarchical structure that distinguishes contemporary human population as well {boys, {eat, apples}}. Now we note that
human language from all other known as in the historical record, in contrast to one can simply apply merge to the two
nonhuman cognitive systems. human group differences such as the adult syntactic objects {boys,{eat, what}} and
As the text below and Figure 1 shows, ability to digest lactose or skin pigmenta- {what}, in which {what} is a subcompo-
merge also accounts for the characteristic tion [12]. There is no doubt that a normal nent (a subset) of the first syntactic object
appearance of displacement in human child from England raised in northern rather than some external set. This yields
language—the apparent ‘‘movement’’ of Alaska would readily learn Eskimo-Aleut, something like {what, {boys, {eat,
phrases from one position to another. or vice versa; there have been no con- what}}}, in this way marking out the two
Displacement is not found in artificially firmed group differences in the ability of required operator and variable positions
constructed languages like computer pro- children to learn their first language, for what.
gramming languages and raises difficulties despite one or two marginal, indirect,
for parsing as well as communication. On and as yet unsubstantiated correlative The Nature of Evolution
the SMT account, however, displacement indications [13]. This uniformity and
arises naturally and is to be expected, stability points to the absence of major Evolutionary analysis might be brought
rather than exceptional, as seems true in evolutionary change since the emergence to bear on language in two different ways.
every human language that has been of the language faculty. Taken together, First, evolutionary considerations could be
examined carefully. Furthermore, hierar- these facts provide good evidence that used to explain the mechanisms of human
chical language structure is demonstrably merge was indeed the key evolutionary language. For instance, principles derived
present in humans, as shown, for instance, innovation for the language faculty. from studying the evolution of communi-
by online brain imaging experiments [7], It is sometimes suggested that external cation might be used to predict, or even
but absent in nonhuman species, e.g., motor sequences are ‘‘hierarchical’’ in this explain, the structural organization of
chimpanzees taught sign language demon- sense and so provide an antecedent language. This approach is fraught with
strably lack this combinatorial ability [8]. platform for language [14]. However, as difficulties. Questions of evolution or

PLOS Biology | www.plosbiology.org 2 August 2014 | Volume 12 | Issue 8 | e1001934


Paleoanthropology
Language leaves no direct imprint in
the fossil record, and the signals imparted
by putative morphological proxies are
highly mixed. Most of these involve speech
production and detection, neither of which
by itself is sufficient for inferring language
(see Box 2). After all, while the anatomical
potential to produce the frequencies used
in modern speech may be necessary for
the expression of language, it provides no
proof that language itself was actually
Figure 1. The binary operation of merge (X,Y) when Y is a subset of X leads to the employed. What is more, it is not even
ubiquitous phenomenon of ‘‘displacement’’ in human language, as in Guess what boys
eat. Left: The circled structure Y, corresponding to what, the object of the verb eat, is a subset of
necessary for language, as the visual and
the circled structure X, corresponding to boys eat what. Right: The free application of merge to X, haptic externalization routes make clear.
Y in this case automatically leads to what occupying two syntactic positions, as required for Moreover, even granting that speech is a
proper semantic interpretation. The original what remains as the object of the verb so that it can requirement for language, it has been
serve as an argument to this predicate, and a copy of what, ‘‘displaced,’’ is now in the position of argued convincingly [25,26] that equal
a quantificational operator so that the form can be interpreted as ‘‘for what x, boys eat x.’’ proportions of the horizontal and vertical
Typically, only the higher what is actually pronounced, as indicated by the line drawn through the
lower what. portions of the vocal tract are necessary for
doi:10.1371/journal.pbio.1001934.g001 producing speech. This conformation is
uniquely seen in our own species Homo
sapiens. In a similar vein, the aural ability
function are fundamentally different from after the appearance on the planet of of nonhuman primates like chimpanzees
those relating to mechanism, so evolution anatomically distinctive Homo sapiens or extinct hominid species such as H.
can never ‘‘explain’’ mechanisms [18]. For around 200,000 years ago [20,21], neanderthalensis to perceive the sound
a start, the evolution of a particular trait The second, more traditional way of frequencies associated with speech
may have proceeded in different ways, applying evolutionary analysis to lan- [26,27] says nothing about the ability of
such as via common descent, convergence, guage is to attempt to reconstruct its these relatives to understand or produce
or exaptation, and it is not easy to establish evolutionary history. Here, too, we are language. Finally, neither the absolute size
which of these possibilities (or combination confronted with major explanatory obsta- of the brain nor its external morphology as
of them) is relevant [18,19]. More impor- cles. For starters, language appears to be seen in endocasts has been shown to be
tantly, evolution by natural selection is not unique to the species H. sapiens. That relevant to the possession of language in
a causal factor of either cognitive or neural eliminates one of the cornerstones of an extinct hominid (Figure 2) [28]. Recent
mechanisms [18]. Natural selection can be evolutionary analysis, the comparative research has determined that Neander-
seen as one causal factor for the historical method, which generally relies on features thals possessed the modern version of the
process of evolutionary change, but that is that are shared by virtue of common FOXP2 gene [29], malfunctions in which
merely stating the essence of the theory of descent (Box 1) [1,4,18]. Alternatively, produce speech deficits in modern people
evolution. As we have argued, communi- analysis can appeal to convergent evolu- [4,30]. However, FOXP2 cannot be
cation cannot be equated with language, tion, in which similar features, such as regarded as ‘‘the’’ gene ‘‘for’’ language,
so its evolution cannot inform the mech- birds’ wings and bats’ wings, arise inde- since it is only one of many that have to be
anisms of language syntax. However, pendently to ‘‘solve’’ functionally analo- functioning properly to permit its normal
evolutionary considerations—in particu- gous problems. Both situations help expression.
lar, reconstructing the evolutionary history constrain and guide evolutionary expla- In terms of historically calibrated
of relevant traits—might provide clues or nation. Lacking both, as in the case of records, this leaves us only with archae-
hypotheses as to mechanisms, even though language, makes the explanatory search ology, the archive of ancient human
such hypotheses have frequently been more difficult. In addition, evolutionary behaviors—although we have once again
shown to be false or misleading [18]. analysis of language is often plagued by to seek indirect proxies for language. To
One such evolutionary clue is that, popular, naı̈ve, or antiquated conceptions the extent that language is interdependent
contrary to received wisdom, recent anal- of how evolution proceeds [19,22]. That with symbolic thought [20], the best
yses suggest that significant genetic change is, evolution is often seen as necessarily a proxies in this domain are objects that
may occur in human populations over the slow, incremental process that unfolds are explicitly symbolic in nature. Opin-
course of a few hundred years [19]. Such gradually over the eons. Such a view of ions have varied greatly as to what
rapid change could also have occurred in evolutionary change is not consistent with constitutes a symbolic object, but if one
the case of language, as we will argue current evidence and our current under- excludes stone and other Paleolithic
below. In addition, as detailed in the next standing, in which evolutionary change implements from this category on the
section, paleoanthropological evidence can be swift, operating within just a few fairly firm grounds that they are prag-
suggests that the appearance of symbolic generations, whether it be in relation to matic and that the techniques for making
thought, our most accurate proxy for finches’ beaks on the Galapagos, insect them can be passed along strictly by
language, was a recent evolutionary event. resistance to pesticides following WWII, imitation [31], we are left with objects
For instance, the first evidence of puta- or human development of lactose toler- from the African Middle Stone Age
tively symbolic artifacts dates back to only ance within dairy culture societies, to (MSA) such as pierced shell beads from
around 100,000 years ago, significantly name a few cases out of many [19,22–24]. various ,100,000-year-old sites (e.g.,

PLOS Biology | www.plosbiology.org 3 August 2014 | Volume 12 | Issue 8 | e1001934


Box 2. The Infamous Hyoid Bone stimulus was most plausibly the appear-
ance of language in members of a species
A putative relationship between basicranial flexion, laryngeal descent, and the that demonstrably already possessed the
ability to produce sounds essential to speech was suggested [52] before any fossil peripheral vocal apparatus required to
hyoid bones, the sole hard-tissue components of the laryngeal apparatus, were externalize it [20,22]. Then, within a
known. It was speculated that fossil hyoids would indicate when speech, and by remarkably short space of time, art was
extension language, originated. A Neanderthal hyoid from Kebara in Israel invented, cities were born, and people had
eventually proved very similar to its H. sapiens homologue, prompting the reached the moon. By this reckoning, the
declaration that speech capacity was fully developed in adult H. neanderthalensis language faculty is an extremely recent
[53]. This was soon contested on the grounds that the morphology of the hyoid is acquisition in our lineage, and it was
both subsidiary [25] and unrelated [26] to its still-controversial [36] position in the acquired not in the context of slow,
neck. A recent study [54] focuses on the biomechanics, internal architecture, and
gradual modification of preexisting sys-
function of the Kebara fossil. The authors conclude that their results ‘‘add support
for the proposition that the Kebara 2 Neanderthal engaged in speech’’ ([54], p. 6). tems under natural selection but in a
However, they wisely add that the issue of Neanderthal language will be fully single, rapid, emergent event that built
resolved only on the basis of fuller comparative material. While the peripheral upon those prior systems but was not
ability to produce speech is undoubtedly a necessary condition for the expression predicted by them. It may be relevant to
of vocally externalized language, it is not a sufficient one, and hyoid morphology, note that the anatomical ability to express
like most other lines of evidence, is evidently no silver bullet for determining language through speech was acquired at
when human language originated. a considerable cost, namely the not-
insignificant risk of adults choking to death
[25,36], as simultaneous breathing and
[32]) and the ,80,000-year-old geomet- Nevertheless, the conclusion lines up with swallowing became impossible with the
rically engraved plaques from South what is known from genomics. descent of the larynx. However, since this
Africa’s Blombos Cave [33] as the earliest Our species was born in a technologi- conformation was already in place before
undisputed symbolic objects. Such objects cally archaic context [35], and significant- language had demonstrably been acquired
began to be made only substantially after ly, the tempo of change only began picking (see Box 2), the ability to express language
the appearance, around 200,000 years up after the point at which symbolic cannot by itself have been the counter-
ago, of anatomically recognizable H. objects appeared. Evidently, a new poten- vailing advantage. Finally, there has been
sapiens, also in Africa [34]. To be sure, tial for symbolic thought was born with no detectable evolution of the language
this inference from the symbolic record, our anatomically distinctive species, but it faculty since it emerged, with no known
like much else in paleontology, rests on was only expressed after a necessary group differences. This is another signa-
evidence that is necessarily quite indirect. cultural stimulus had exerted itself. This ture of relatively recent and rapid origin.

Figure 2. A crude plot of average hominid brain sizes over time. Although after an initial flatlining this plot appears to show consistent
enlargement of hominid brains over the last 2 million years, it is essential to note that these brain volumes are averaged across a number of
independent lineages within the genus Homo and likely represent the preferential success of larger-brained species. From [20]. Image credit: Gisselle
Garcia, artist (brain images).
doi:10.1371/journal.pbio.1001934.g002

PLOS Biology | www.plosbiology.org 4 August 2014 | Volume 12 | Issue 8 | e1001934


For reasons like these, the relatively have identified with words. Without compatible with the known empirical
sudden origin of language poses difficulties merge, there would be no way to assemble evolutionary evidence. Such an account
that may be called ‘‘Darwin’s problem.’’ the arbitrarily large, hierarchically struc- also aligns with what we currently know
tured objects with their specific interpre- about the relatively few genomic differ-
The Minimalist Account of tations in the language of thought that ences between our species and other
Language—Progress towards distinguish human language from other ancestral Homo species—e.g., only about
Resolving ‘‘Darwin’s Problem’’ animal cognitive systems—just as Darwin 100 coding gene differences between
insisted: ‘‘A complex train of thought can Homo sapiens and H. neanderthalensis,
The Strong Minimalist Thesis (SMT) be no more carried out without the use of the majority of them in nonlanguage areas
[6], as discussed above, greatly eases the words, whether spoken or silent, than a such as the olfactory and immune systems
explanatory burden for evolutionary anal- long calculation without the use of figures [40]. Furthermore, as far as we can tell
ysis, since virtually all of the antecedent or algebra’’ ([39], p. 88). With merge, from direct historical evidence, the capac-
‘‘machinery’’ for language is presumed to however, the basic properties of human ity that emerged, namely the ability of any
have been present long before the human language emerge. Evolutionary analysis child to learn any human language, has
species appeared. For instance, it appears can thus be focused on this quite narrowly remained frozen for 10,000 years or more.
that the ability to perceive ‘‘distinctive defined phenotypic property, merge itself, To be sure, such observations must be
features’’ such as the difference between as the chief bridge between the ancestral interpreted with great care and can
the sound b, as in bat, as opposed to p, as and modern states for language. Since this remain only suggestive as long as we lack
in pat, might be present in the mammalian change is relatively minor, it accords with the knowledge to even crudely connect
lineage generally [37,38]. The same holds what we know about the apparent rapidity genomic changes to the relevant pheno-
for audition. Both comprise part of the of language’s emergence. types. Even given these caveats, it appears
externalization system for language. Fur- that there has simply not been enough
thermore, the general constraint of effi- Conclusions time for large-scale evolutionary changes,
cient computation would also seem plau- as indicated by the SMT. Clearly, such a
sibly antecedent in the cognitive The Strong Minimalist Thesis that we novel computational system could have
computation of ancestral species. The only have sketched here is consistent with a led to a large competitive advantage
thing lacking for language would be merge, recent and rapid evolutionary emergence among the early H. sapiens who
some specific way to externalize the of language. Although this thesis is far possessed it, particularly when linked to
internal computations and, importantly, from being established and contains many possibly preexisting perceptual and motor
the ‘‘atomic conceptual elements’’ that we open questions, it offers an account that is mechanisms.

References
1. Berwick RC, Friederici AD, Chomsky N, Bolhuis 12. Bersaglieri T, Sabeti PC, Patterson N, Vander- 25. Lieberman P (2007) The evolution of human
JJ (2013) Evolution, brain, and the nature of ploeg T, Schaffner SF, et al. (2004) Genetic speech: Its anatomical and neural bases. Curr
language. Trends Cogn Sci 17: 89–98. signatures of strong recent positive selection at the Anthropol 48: 39–66.
2. Nowak MA, Komarova NL, Niyogi P (2001) lactase gene. Am J Hum Genet 74: 1111–1120. 26. Fitch TW (2000) The evolution of speech: A
Evolution of universal grammar. Science 291: 13. Ladd DR, Dediu D (2013) Genes and linguistic comparative review. Biol Philos 20: 193–230.
114–118. tone. In: Pashler H, editor. Encyclopedia of the 27. Martinez I, Rosa M, Quam R, Jarabo P, Lorenzo
3. Niyogi P, Berwick RC (2009) The proper mind. London: Sage Publications. pp. 372–373. C, et al. (2012) Communicative capacities in
treatment of language acquisition and change in 14. Pulvermüller F (2014) The syntax of action. Middle Pleistocene humans from the Sierra de
a population setting. Proc Natl Acad Sci USA Trends Cogn Sci 18: 219–220. Atapuerca in Spain. Quat Int 295: 94–101.
106: 10124–10129. 15. Moro A (2014) Response to Pulvermüller: the 28. DeSalle R, Tattersall I (2012) The brain: Big
4. Bolhuis JJ, Okanoya K, Scharff C (2010) Twitter syntax of actions and other metaphors. Trends bangs, behaviors, and beliefs. New Haven (Con-
evolution: Converging mechanisms in birdsong Cogn Sci 18: 221. necticut): Yale University Press.
and human speech. Nat Rev Neurosci 11: 747– 16. Chomsky N (2013) Problems of projection. 29. Krause L, Lalueza-Fox C, Orlando L, Enard W,
759. Lingua 130: 33–49. Green RE, et al. (2007) The derived FOXP2
5. Chomsky C (1986) Analytic study of the Tadoma 17. Berwick R, Chomsky N (2011) The biolinguistic variant of modern humans was shared with
method: language abilities of three deaf-blind program: The current state of its development. Neandertals. Curr Biol 17: 1908–1912.
individuals. J Speech & Hearing Res 29: 332– In: Di Sciullo AM, Boeckx C, editors. The 30. Varga-Khadem F, Gadian DG, Copp A, Mishkin
347. Reprinted in: Piattelli-Palmarini M, Berwick biolinguistic enterprise. Oxford: Oxford Univer- M (2005) FOXP2 and the neuroanatomy of speech
RC, editors (2013) Rich Languages from Poor sity Press. pp. 19–41. and language. Nat Rev Neurosci 6: 131–138.
Inputs. Oxford: Oxford University Press. pp. 18. Bolhuis JJ, Wynne CDL (2009) Can evolution 31. Ohnuma K, Aoki K, Akazawa T (1997) Trans-
241–270. explain how minds work? Nature 458: 832–833. mission of tool-making through verbal and non-
6. Chomsky N (2000) Minimalist inquiries: The 19. Bolhuis JJ, Brown GR, Richardson RC, Laland verbal communication: Preliminary experiments
framework. In: Martin R, Michaels D, Uriager- KN (2011) Darwin in mind: New opportunities in Levallois flake production. Anthropol Sci 105:
eka J, editors. Step by step: Essays on minimalist for evolutionary psychology. PLoS Biol 9: 159–168.
syntax in honor of Howard Lasnik. Cambridge e1001109. 32. Vanhaeren M, D’Errico F, Stringer C, James SL,
(Massachusetts): MIT Press. pp. 89–155. 20. Tattersall I (2008) An evolutionary framework for Todd JA (2006) Middle Paleolithic shell beads in
7. Pallier C, Devauchelle A-D, Dehaene S (2011) the acquisition of symbolic cognition by Homo Israel and Algeria. Science 312: 1785–1788.
Cortical representation of the constituent struc- sapiens. Comp Cogn Behav Revs 3: 99–114. 33. Henshilwood C, d’Errico F, Yates R, Jacobs Z,
ture of sentences. Proc Natl Acad Sci USA 108: 21. Dubreuil B, Henshilwood CS (2013) Material Tribolo C, et al. (2002) Emergence of modern
2522–2527. culture and language. In: Lefebvre C, Comrie B, human behavior: Middle Stone Age engravings
8. Yang C (2013) The ontogeny and phylogeny of Cohen H, editors. New perspectives on the origin from South Africa. Science 295: 1278–1280.
language. Proc Natl Acad Sci U S A 110: 6324– of language. Amsterdam: John Benjamins. pp. 34. McDougall I, Brown FH, Fleagle JG (2005)
6327. 147–170. Stratigraphic placement and age of modern
9. Hurford J (2012) Language in the light of 22. Tattersall I (2012) Masters of the planet. The humans from Kibish, Ethiopia. Nature 433:
evolution. Oxford: Oxford University Press. search for our human origins. New York: 733–736.
10. Jackendoff R (2003) Foundations of language: Palgrave Macmillan. 35. Clark JD, Beyene Y, WoldeGabriel G, Hart WK,
Brain, meaning, grammar, evolution. Oxford: 23. Thompson JH (2013) Relentless evolution. Chi- Renne PR, et al. (2003) Stratigraphic, chronolog-
Oxford University Press. cago: University of Chicago Press. ical and behavioural contexts of Pleistocene Homo
11. Bickerton D (1981) Roots of language. Ann Arbor 24. Nei M (2013) Mutation-driven evolution. Oxford: sapiens from Middle Awash, Ethiopia. Nature
(Michigan): Karoma Publishers. Oxford University Press. 423: 747–752.

PLOS Biology | www.plosbiology.org 5 August 2014 | Volume 12 | Issue 8 | e1001934


36. Lieberman DE (2011) The evolution of the 43. Gobes SMH, Bolhuis JJ (2007) Bird song Everaert M, editors. Birdsong, speech, and
human head. Cambridge, MA: Harvard Univer- memory: A neural dissociation between song language. Exploring the evolution of mind and
sity Press. recognition and production. Curr Biol 17: 789– brain. Cambridge (Massachusetts): MIT Press.
37. Kuhl PK, Miller JD (1975) Speech perception by 793. pp. 1–26.
the chinchilla: Voiced-voiceless distinction in 44. Moorman S, Gobes SMH, Kuijpers M, Kerkhofs 50. Patel AD (2014) The evolutionary biology of
alveolar plosibve consonants. Science 190: 69–72. A, Zandbergen MA, et al. (2012) Human-like musical rhythm: Was Darwin wrong? PLoS Biol
38. Hauser M, Fitch T (2003) What are the uniquely brain hemispheric dominance in birdsong learn- 12: e1001821.
human components of the language faculty? In: ing. Proc Natl Acad Sci USA 109: 12782–12787. 51. Shukla M, White KS, Aslin R (2011) Prosody
Christiansen MH, Kirby S, editors. Language 45. Marler P (1970) A comparative approach to vocal guides the rapid mapping of auditory forms onto
Evolution. Oxford: Oxford University Press. pp. learning: song development in white-crowned visual objects in 6-mo-old-infants. Proc Natl Acad
158–181. sparrows. J Comp Physiol Psychol 71: 1–25. Sci USA 108: 6038–6043.
39. Darwin C (1871) The Descent of man, and 46. Gentner TQ, Fenn KM, Margoliash D, Nuss- 52. Laitman JT, Heimbuch RC, Crelin ES (1979)
selection in relation to sex. London: John Murray. baum HC (2006) Recursive syntactic pattern The basicranium of fossil hominids as an
40. Geschwind DH, Rakic P (2013) Cortical evolu- learning by songbirds. Nature 440: 1204–1207. indicator of their upper respiratory systems.
tion: Judge the brain by its cover. Neuron 80: 47. Abe K, Watanabe D (2011) Songbirds possess the Am J Phys Anthropol 51: 15–34.
633–647. spontaneous ability to discriminate syntactic rules. 53. Arensburg B, Tillier A-M, Vandermeersch B,
41. Berwick RC, Okanoya K, Beckers GJL, Bolhuis Nat Neurosci 14: 1067–1074. Duday H, Schepartz LA, et al. (1989) A Middle
JJ (2011) Songs to syntax: The linguistics of 48. Beckers GJL, Bolhuis JJ, Okanoya K, Berwick Palaeolithic human hyoid bone. Nature 338: 758–
birdsong. Trends Cogn Sci 15: 113–121. RC (2012) Birdsong neurolinguistics: Songbird 760.
42. Bolhuis JJ, Everaert M, editors (2013) Birdsong, context-free grammar claim is premature. Neu- 54. D’Anastasio R, Wroe S, Tuniz C, Mancini L,
speech, and language. Exploring the evolution of roReport 23: 139–145. Cesana DT, et al. (2013) Micro-biomechanics of
mind and brain. Cambridge (Massachusetts): 49. Everaert M, Huybregts R (2013) The design the Kebara 2 hyoid and its implications for
MIT Press. principles of natural language. In: Bolhuis JJ, speech in Neanderthals. PLoS ONE 8: e82261.

PLOS Biology | www.plosbiology.org 6 August 2014 | Volume 12 | Issue 8 | e1001934

View publication stats

You might also like