Evolution and Language: Overview

William Croft, University of New Mexico, Albuquerque, NM, USA

Abstract

The close relationship between biological evolution and language was noted by Darwin himself in an oft-quoted passage
from The Descent of Man: “the formation of different languages and of different species, and the proofs that both have been
developed through a gradual process, are curiously parallel” (Darwin, 1882). In fact, the development of evolutionary theory
in biology was inspired in part by the advances in historical linguistics in the early nineteenth century. In the twentieth
century, evolutionary theory did not have much influence on linguistics, in part due to the advent of structuralism and the
focus on synchronic linguistic analysis. In the past two decades, there has been a considerable increase in interest in the
relationship between language and evolution. The interest has been manifested in three areas of recent research. The first is
the evolutionary origin of the human language capacity, a topic that was actively avoided in linguistics for a century. The
second is the employment of techniques of phylogeny reconstruction from biology in the analysis of genetic families of
languages. Finally, there is the application of theories of evolutionary processes to language change, based on the hypothesis
that, as Darwin wrote, the two are strikingly parallel.

The Evolutionary Origin of Language us more about the evolutionary origin of the human language
capacity.
The evolutionary origin of language is not technically an The taboo against studying the evolutionary origin of the
example of parallelism between biological evolution and language capacity has now fallen. A number of different
language. It simply is an instance of biological evolution, approaches have been taken to tackling the question. All of
specifically the evolution of a behavioral capacity in one them are necessarily indirect. Two of the commonest
particular species. This topic was avoided in linguistics approaches are computer-based models simulating a society of
for over a century because of the absence of empirical interacting agents, to explore under what circumstances
evidence about any stage of language prior to modern human language structures might emerge (e.g., Kirby, 1999); and
language. comparative ethology, comparing the communicative,
All modern human languages are general-purpose symbolic, and interactional capacities of our nearest relatives,
communication systems of a similar order of complexity; the primates, and other species that possess vocal
there is no reason to believe that any contemporary human communicative systems, such as various species of birds and
languages represent an earlier stage in the evolutionary devel- cetaceans (e.g., Tomasello, 2008). Child language acquisition
opment of the language capacity. Written records of language is also used as a reference point for language origins
are very recent – no more than 5000 years old – and they make (Tomasello, 2008). Comparison to child language acquisition
clear that languages of that age are no different from contem- must be treated with caution because children learn language
porary human languages (see Writing, Evolution of). Languages by being exposed to a modern human language, which will
otherwise do not leave material evidence, so the archaeological influence their course of development. Also, children largely
record leaves no direct record of the evolution of the language already have the cognitive and social capacities of modern
capacity. humans, so they do not represent the sort of hominins in
Evidence from hominin skeletal remains suggests that there which the modern human language capacity evolved. More
have been anatomical changes that indicate the increased recently, experiments using artificial language learning have
ability of our ancestors to produce a wide range of vocaliza- come to be used in the search for language origins (see
tions, and to control our breathing in such a way to produce Evolution and Language: Cultural Transmission). Again, the
linguistic sounds. It is generally assumed that the ability to subjects of artificial language learning experiments already
produce a rich variety of vocalizations implies the existence of have the cognitive and social capacities of modern human
language at the time of these anatomical changes. However, it beings, and they also have knowledge of a modern human
could be that such vocalizations had other functions than language at the point that they engage in the experiments.
a general-purpose communication system, such as music and This latter point also applies to proposals about the origins
ceremony, or that the sort of language that such vocalizations of language based on more naturalistic examples of language
were used for was quite different in its structure and commu- ‘creation,’ such as pidgins, creoles, and certain sign languages.
nicative range than modern human language. Likewise, the Many earlier studies focused on the structure of the
great expansion of brain capacity in the development of linguistic signal. In comparative ethology, the question was
humankind is an indicator of the emergence of the cognitive whether or not certain structures of human language utterances
capacity for modern human language. But lacking more specific were unique, or could be found in the communication systems
details of neural structure in hominins, this change cannot tell of other species. For example, it was claimed that vervet

364 International Encyclopedia of the Social & Behavioral Sciences, 2nd edition, Volume 8 http://dx.doi.org/10.1016/B978-0-08-097086-8.81033-8
 Evolution and Language: Overview 365

monkey alarm calls were symbolic, in that they distinguish consisted of only one symbolic unit (a one-word stage). Some
three different kinds of predators, and that bee dances and argue that a multiword stage developed by combining these
some bird songs involve a combinatoric syntax not unlike that single units in a single utterance, while others argue that
of human languages. In the modeling approaches, the agents a multiword stage developed by reanalyzing the single
would produce strings of units and through simple rules of symbolic unit (a complex vocalization) into meaningful parts.
interpretation and feedback from such interactions, stable Other proposals are based on the functions of the linguistic
meanings of the units (i.e., symbols) and regular syntactic units. An earlier stage may involve gestures that are indexical,
combinations were shown to emerge over several generations establishing joint attention in order to coordinate joint actions;
of interactions. names for individuals and symbolic gestures that denote things
More recent studies have taken a broader perspective on the or actions emerge in a later stage. Or an earlier stage may
evolution of the human language capacity. Language structure involve more direct interactions such as commands or ques-
does not evolve for its own sake, but only as part of a larger tions; utterances whose sole purpose is to share information,
system of social interaction among human beings. Even the especially about displaced experience, develop in a later stage.
function of communication, often taken to be the raison d’être Such intermediate stages of pre-language only make sense in
for language, actually serves the ultimate goal of achieving the context of plausible intermediate stages of social interaction
some joint action between persons. These joint actions form between what is observed in other primates (or other species)
the basis of human culture and society. In other words, the and what occurs between human beings in contemporary
evolution of language is part of the evolution of human social societies.
interaction and human culture (Tomasello, 2008).
This broader perspective implies, for example, that the
computational modeling of the emergence of language will The Use of Biological Methods for Language
have limited success with disembodied agents simply reacting Classification (Phylogenies)
to whether a string of units refers to some object or not. In one
modeling project from this perspective, the agents are robots The most obvious respect in which biological evolution and
that have perceptual and motor abilities, and the modeling language are parallel is that languages form lineages in much
includes visual processing and the establishment of joint the same way as species do (see Evolution and Language:
attention on the part of the robots toward a potential referent Phylogenetic Analyses). Languages have parents and daughters,
for any utterance that is produced (Steels, 2011). Of course, this and form trees of ancestors and descendants; these trees are
type of modeling is very ambitious, and requires solving called phylogenies in biology (see Comparative Method in
problems in visual and auditory processing, motor activity, Evolutionary Studies). In both domains, the ancestors are no
and social interaction as well as language structure processing longer present, because they have either become extinct or the
and cognitive categorization and combination. In other ancestral species or language has split up into the contempo-
words, one must model human beings in their full rary species or languages. In only a few cases is there a material
perceptual, motor, cognitive, and social abilities in order to record of ancestral forms – fossil individuals of ancient species
model the emergence of language. and written records of ancient languages. As a consequence,
In the case of comparative ethological approaches, the evidence for the phylogeny of both species and languages is
broader perspective implies looking at the social behavioral indirect, namely, the traits of the currently existing species or
capacities of primates and other species, including joint languages. Comparative work among languages uncovers
attention, reasoning about other individual’s beliefs (‘theory of genetic trees, just as comparative work among species uncovers
mind’), coordination of actions, and the ability to engage in biological taxa. In some cases, historical linguists have arrived
cooperative activity, as well as cognitive abilities of categori- at conclusions about the nature of comparative historical
zation, causal reasoning, and representing knowledge of dis- analysis before biologists have. For example, an important
placed experience (i.e., not the ‘here and now’) that have been principle of comparative historical linguistics is that a trait that
traditionally associated with the evolution of the language is an innovation among daughter languages is more important
capacity. All of these social behaviors are preconditions for the evidence that the daughter languages form a subgroup than
evolution of modern human languages. Comparative studies retention of a shared trait from an ancestral language. This
suggest that many other species have the simple cognitive principle was later independently adopted in the cladistic
capacities listed above, but few if any have the social cognitive approach to constructing phylogenies (Hull, 1988).
capacities required for language, and not to the degree that Thus, both the basic fact – branching lineages – and the
modern humans do. basic method – comparison of present-day members – are the
Language embedded in its social interactional context is same for biology and language. For this reason, recent work has
part of a very complex suite of human behavioral capacities. It increased in the application of phylogeny reconstruction
is highly unlikely that this complex suite evolved suddenly. methods from biology to language data (see Evolution and
Thus, the transition to modern human language in the Language: Phylogenetic Analyses). For the most part, it has
hominid lineage was probably a gradual one, with a number of been biologists using these methods on language data, rather
intermediate stages of ‘pre-language.’ Unfortunately, there is than linguists adopting these methods for historical linguistics.
still a great gap between animal communication systems The basic reason for this fact is that the biological methods are
known to us and modern human language, so just what those essentially quantitative (and increasingly complex), while the
intermediate stages are is largely a matter of speculation. One methods of traditional historical linguists are essentially
common proposal is that there was a stage in which utterances qualitative.
366 Evolution and Language: Overview

In comparative historical linguistics, the traits used to clas- phylogeny reconstruction algorithms is in how to explore the
sify languages into genetic families (i.e., phylogenies) are search space of possible trees.) One consequence of this fact is
words, that is, combinations of form and meaning. Words are that a phylogeny reconstruction algorithm does not produce
particularly good traits to use for phylogenies. Words generally a single ‘best’ family tree; it produces a probability distribution
evolve independently of each other, so a large number of words of greater or less likely trees for the data.
represents a large set of independent pieces of evidence for Against this concern is the fact that even crude methods can
a phylogeny. (In some cases, words are derivationally related, find the major subgroups for the language families to which
such as long and length, and so are not entirely independent.) they are applied, such as the major subgroups of Indo-
The pairing of form and meaning is largely arbitrary, so other European. That is, the phylogenetic signal is very strong. But
factors are unlikely to lead to convergence across languages. that signal is strong enough that visual inspection of cognate
(There are cases of onomatopoeia, such as meow, but they are lists reveals those aspects of the phylogeny, and employment of
few and are avoided by historical linguists.) For this reason, quantitative methods is unnecessary. When it comes to
historical linguists avoid schematic grammatical traits, such as resolving phylogenetic questions which are still a matter of
subject-verb-object word order and the presence of grammat- debate among historical linguists, then the ability to find the
ical gender. These traits are limited in their possibilities (much most likely trees becomes an important issue in applying
more so than the possible pairings of forms and meanings), quantitative methods to historical linguistic problems.
and their presence is often due to universals of grammatical Any list of cognates for a set of languages is going to give
structure, of the sort studied in linguistic typology (see conflicting information about the phylogeny of those
Linguistic Typology), rather than common ancestry of the languages. Languages lose cognate forms; the original cognate
languages that contain them. In biological terms, words are changes meaning, or is lost entirely, and its form is replaced by
weakly filtered by selection but schematic grammatical patterns another, unrelated form. Hence a missing cognate is not
are not. Individual words form lineages themselves, but these necessarily evidence that the language does not belong to the
word lineages bundle together, especially in basic vocabulary. family in question. Deciding whether a set of cognate forms is
This is what allows us to say that a language is the daughter of a common innovation, justifying placing those languages into
another language: most of its linguistic material comes from a subgroup, or a common retention, whose survival or loss is
a single source. Of course, some words come from a language an accident, is not an automatic process. It requires balancing
other than its parent; these are borrowings. Borrowings are evidence from other cognate distributions to decide which
more common in nonbasic vocabulary, although they occur in cognates define a subgroup and which ones do not. The
basic vocabulary as well. To the extent that borrowings can be phylogeny reconstruction algorithms encode a particular way
identified and excluded from analysis, then the phylogeny of to make these decisions, based on their utility in biological
languages is tree like: a single parent and multiple daughters. evolution. The algorithm may or may not have the same utility
Most applications of biological methods in phylogeny in language change (this factor has not been discussed in the
reconstruction to language data begin with a set of words that relevant literature).
are assumed to be cognate, that is, inherited from a common The application of biological phylogeny reconstruction
ancestor language (called a protolanguage). That is, for a set of algorithms to language classification is actually a hybrid
languages, the data consists of whether or not some or all of method. The data to which the algorithm is applied is already
the languages share a cognate for that meaning, for all the processed by a linguist: a linguist has decided which words are
meanings in the word list. The phylogeny reconstruction cognates and which ones are not. One of the chief problems in
algorithms use computational statistical methods to construct determining cognates is whether words that are similar in form
a phylogeny for the languages in question, given the distri- and meaning across languages are borrowings or not. Some
bution of cognate forms across the languages. These methods putative cognates may turn out to be borrowings. Some
have increased in sophistication in recent years, aided by phylogeny reconstruction programs produce networks rather
increases in computational power (e.g., Bouckaert et al., 2012; than trees. In most if not all cases in language history, ancestry
see Comparative Method in Evolutionary Studies; Evolution vs contact are clearly distinct; for example, English is
and Language: Phylogenetic Analyses). a Germanic language, albeit with many words borrowed via
The phylogeny reconstruction methods provide a more contact with French. Historical linguists look for patterns in the
precisely quantifiable judgment weighing the evidence from distribution of similar forms in order to differentiate common
the complex distribution of a large number of cognates. They ancestry from borrowing. For example, similar forms across the
have been applied to universally accepted, chronologically bulk of basic vocabulary is indicative of common ancestry,
relatively shallow families such as Austronesian (in Oceania) whereas similar forms concentrated in particular semantic
and Bantu (in Africa) in order to clarify the subgroupings of domains of nonbasic vocabulary (e.g., animal husbandry,
those language families and provide support for hypotheses religious ceremony) is indicative of contact; and a high degree
about the prehistoric migrations of the speakers of those of similarity of forms to the forms in just one language suggest
languages. Such methods have their limits, however. For even that the latter language is the source of borrowing. Judgments
a relatively small number of languages, the number of possible based on differentiating types of vocabulary, or patterns of
trees that can be constructed is far larger than the number that distribution across sets of languages, are not yet incorporated
can be tested using current computing power. Hence, the into phylogeny reconstruction algorithms.
phylogeny reconstruction algorithm can only test a small What is of most interest to linguists (and nonlinguists) is
subset of the possible trees against the data provided by the list the establishment of chronologically deeper language families
of cognate forms. (Much of the recent effort in refining than those widely accepted among linguists today. The
 Evolution and Language: Overview 367

comparative historical linguistic method starts from generally called replication. Indeed, evolution can be thought
languages that are accepted to form a group; phylogeny of as a model of change by replication, as opposed to
reconstruction algorithms do the same, by starting with a model of inherent change. Replication also occurs in
cognates for accepted families. Deeper language families have cultural transmission, including language change (Mesoudi
been proposed, but are controversial among linguists. In et al., 2004; Pagel, 2009). In both biological and cultural
principle, it is possible to create algorithms that would iden- evolution, there is debate over what is the replicator, that is,
tify form–meaning similarities among a complete set of words what it is that replicates. In neo-Darwinian evolutionary
across a large set of languages not accepted to be related; look theory, the paradigm case of the replicator is the gene, and we
for correspondences between the sounds occurring in equiv- will use this example for initial illustration.
alent positions in the words, which is strong evidence for The fundamental properties of the replication process, most
cognacy; use distributional patterns of form–meaning of which are explicated by Darwin, can be grouped into two
similarities to differentiate likely cognates from likely categories (Dawkins, 1976). The first category has to do with
borrowings; and use all of this information to postulate likely how replicators form lineages. A replicated unit can itself be
family trees. It is only very recently that the first efforts to replicated; the replication process can proceed indefinitely.
devise such algorithms are being made. And some replicators must survive long enough to be
Biologists interested in prehistoric human population replicated themselves (these properties are also called
history have compared phylogenies reconstructed from bio- longevity and fecundity). The second category has to do with
logical data to language family trees, and have observed inheritance, namely what makes the replication process
parallels between biological phylogenies of humans (and the nonrandom. Replicated units possess much of the structure
migration patterns they imply) and language family trees. The of the original unit. The replication process is mostly faithful.
combination of biological and linguistic phylogenies has been But replication can produce heritable variation: that is, the
used to clarify the prehistoric movements of Austronesian and replication process introduces variation, and that variation is
Bantu peoples, where the language families are widely cumulative over generations of replicators.
accepted; and the migration of Native Americans into the New Among theorists of cultural evolution, three sorts of entities
World, where the language families that resemble the biolog- have been proposed as replicators: artifacts, behaviors, and
ical phylogenies remain controversial (Reich et al., 2012). The concepts, that is, mental entities. An issue with all of these
parallels between biological and language phylogenies do not replicator types is that their replication must be indirect. Arti-
necessarily match in all details, because human populations facts and behaviors are reproduced by human agents, who use
may give up their heritage language and shift to another their cultural knowledge (i.e., mental conceptual structures) to
language. Nevertheless, considering both biological and replicate them. For concepts to be replicated from one person
language families together gives us a more refined picture of to another (‘learning’), they must be publicly displayed in
human prehistory. terms of human actions (behavior) or products (artifacts). In
addition, concepts are not directly observable and there is
evidence that the neural structures and patterns of activation
The Parallel between Biological for the same concept may differ from person to person.
and Linguistic Evolution In language change, the debate has been traditionally
framed in terms of whether language change takes place
The fact that both languages and species form lineages, and through imperfect learning by children, or via language use. In
numerous other parallels between biological evolution and evolutionary terms, this is a debate over whether the replicator
language change, has led many linguists to discuss analogies is a grammar (that is, a speaker’s knowledge about his or her
between the two domains. Although observing analogies may language) or linguistic structures in utterances (such structures
provide inspiration for theories of language change in are replicated every time we speak). Both theories have been
linguistics, they do not provide a principled basis for con- debated since the emergence of historical linguistics in the
structing an evolutionary theory of language change. In nineteenth century, and the debate continues.
particular, when one finds a disanalogy between biological There is considerable empirical evidence that imperfect
evolution and language change, it is unclear whether the dis- learning by children cannot be the source of language change
analogy is due to simply a difference between the two empirical (Croft, 2000). Children’s errors in acquisition (that is, their
domains, or the disanalogy points to profound differences linguistic innovations) do not match the processes observed
between biological evolution and language change. In fact, in language change. Children are very good at learning the
domain-general theories of evolutionary units and processes language spoken around them even though they are not
have been proposed. These theories are intended to subsume provided with direct negative evidence about the structure of
both biological evolution and some form of cultural evolution their language (specialists in child language acquisition call
(see Cultural Evolution: Overview). There is major debate over this the ‘no negative evidence problem’). Finally, children’s
the appropriateness of these theories; the following sections language is not emulated by other social groups in language
will briefly summarize the major issues. change; instead, children expend their effort in emulating
Most biologists consider the fundamental principles of other social groups, in particular their older peers when they
Darwin’s evolutionary theory (Darwin, 1859) to continue to become adolescents.
hold (see Darwinism). These principles are descent with By contrast, the sort of variation that is found in the
modification and selection. Descent with modification leads phonetic realization of sounds, and in the verbalization of
to lineages, and occurs through a process that is now experience into words and constructions, does match observed
368 Evolution and Language: Overview

language changes (Croft, 2010). An important element of the The General Analysis of Selection provides a framework for
usage-based model of language change is that knowledge of an evolutionary model of the processes that have been observed
language includes knowledge of language variation, and that in sociolinguistics. Sociolinguistics observes variation in
language variation is pervasive in language. Hence language language use in a speech community. (Sociolinguists generally
change is actually a process by which change occurs in the do not consider the origins of that variation; it has been argued
distribution of linguistic variants across phonetic space to result from language production, as discussed earlier.) In
(for sounds) or conceptual space (for words or constructions). particular, some of the variation in language use is socially
While adults are not as flexible as children in changing their structured: that is, it correlates with social parameters such
patterns of language use, there is some flexibility, which is as gender, age, and socioeconomic status. The association of
greater for some linguistic units (e.g., new words) and less for variants with particular social parameters appears to drive the
others (accent); and adults vary in their receptiveness to novel propagation of certain variants in the speech community, that
linguistic variants. is, selection of those variants by the speakers in the community.
Cultural replicators have been called memes (Dawkins, Another element of a generalized evolutionary theory that
1976). However, memetics has combined the basic properties plays a central role in cultural evolution and language change is
of replicators and replication with a ‘selfish gene/selfish population thinking. The term ‘population thinking’ refers to
meme’ theory of selection. Selection represents the other the definition of species that emerged by the neo-Darwinian
fundamental principle of Darwinian evolutionary theory. synthesis in the early twentieth century. The population defi-
Selection is a process by which some replicators are nition replaces the essentialist definition of species that
successfully propagated in a population while other preceded evolutionary theory. In the essentialist approach to
replicators go extinct (in the extreme case). Selection is species, a species is defined as an ahistorical type with certain
a separate process from replication. Most evolutionary essential properties. The essentialist approach is problematic
biologists recognize that evolution is a two-step process: the for actual biological species for several reasons. There is a high
generation of variation via replication and selection of some degree of variation among individuals in a single species;
variants over others (also called differential replication). ‘essential’ traits may be lacking in many individuals.
In the ‘selfish gene/meme’ approach, the replicator (gene or Conversely, two species may overlap substantially in their
meme) functions as both the locus of replication and of traits, so ‘essential’ traits would not distinguish species (and in
selection. In biology, the organism is merely a vehicle to fact, there are many disputes about how many species there are
facilitate the gene’s successful replication, and by analogy, in in particular genera). Finally, the ‘essential’ traits of a species
cultural change, a human agent is merely a means for concepts change over time as species evolve.
to differentially replicate themselves. This approach has been The population approach avoids all of these problems. In the
controversial in both biology and theories of cultural evolution population approach, a species is defined not by an essential trait
including language change (but see Ritt, 2004). of individuals, but an interactional trait among the members,
Most biologists consider the selfish gene theory to miss the namely interbreeding (in sexual organisms) and its converse,
important roles that entities other than the gene play in bio- reproductive isolation between populations. This interactional
logical evolution. Analogously, the selfish meme theory misses trait is also the way in which populations form lineages:
the important roles that entities other than the meme, in interbreeding leads to reproduction and lack of interbreeding
particular human agents, play in cultural evolution. Another eventually splits a population. Recently, different definitions
generalized theory of evolution, the General Analysis of of species have been proposed. However, these definitions
Selection, provides a richer, though still highly abstract, theory appear to be the consequence of the fact that populations diverge
of evolutionary processes (Hull, 1988). The General Analysis of slowly when interbreeding stops between the populations, and
Selection includes the replication process, but analyzes the recent species definitions differ as to what point diverging
selection as a process of environmental interaction that populations are said to constitute distinct species.
causes differential replication to take place. Environmental Languages suffer from the same problems with essentialist
interaction defines two more roles, the environment and the definitions as biological species. Languages are internally highly
interactor. In the neo-Darwinian theory of biological evolu- variable, so that one cannot provide an ‘essential’ set of words
tion, the canonical interactor is the organism: an organism and/or grammatical rules that would define a language. Two
interacts with its environment, and natural selection brings languages may share large amounts of lexicon and grammar,
about the differential replication of its genes depending on the and there are disputes in many cases as to whether there are two
survival and reproductive success or failure of the organism. languages or two dialects of a single language. Finally, languages
The General Analysis of Selection can also be applied to change, so ‘essential’ features of languages may change.
language (Croft, 2000). The interactor is the human agent, The general solution is to offer a social definition of
namely the speaker. The speaker’s environment includes the a language, since a structural (i.e., essentialist) definition is
experience being communicated, the speakers with whom he unsatisfactory. More precisely, a population definition of
or she converses, and the social structures that bind those a speech community will avoid the problems of defining
speakers together in a speech community. In that context of a language, since a language is produced by members of
language use, speakers make choices as to which linguistic the speech community (Croft, 2000; Mufwene, 2001).
variants to replicate. The variants that get replicated more A population definition of a speech community is a set of
frequently are propagated through a speech community, and speakers that converse with each other. Since what the
in the end a historical linguist observes that a language members of the speech community produce in conversation
change has taken place. is the language of that community, it accommodates
 Evolution and Language: Overview 369

variation in the language. When a speech community splits, The level of impact of evolutionary thinking on linguistic
the languages they speak gradually diverge as the speakers theorizing remains to be seen.
of the two communities stop interacting. At first, the
languages are structurally quite similar, but over time they See also: Comparative Method in Evolutionary Studies; Cultural
will come to differ; linguists may disagree about the Evolution: Overview; Darwinism; Evolution and Language:
point at which they would say there are now two distinct Cultural Transmission; Evolution and Language: Phylogenetic
languages. Analyses.
Language populations, that is, speech communities, are
never completely communicatively isolated from their neigh-
bors. The result is language contact. Although this is sometimes
Bibliography
held up as a disanalogy between language change and bio-
logical evolution, there is in fact much hybridization in the
Bouckaert, R., Lemey, P., Dunn, M., Greenhill, S.J., Alekseyenko, A.V., Drummond, A.J.,
biological world outside of animals, and many instances of Gray, R.D., Suchard, M.A., Atkinson, Q.D., 2012. Mapping the origins and expansion
introgression that leads to the transfer of genes from one of the Indo-European language family. Science 337, 957–960.
population to another through incomplete reproductive Bybee, J.L., 2010. Language. Usage and Cognitive. Cambridge University Press,
isolation. Language contact leads to sometimes large-scale Cambridge.
Croft, W., 2000. Explaining Language Change: An Evolutionary Approach. Longman,
transfer of linguistic structures or features from one speech
Harlow, Essex.
community to another, particularly in the creation of new Croft, W., 2010. The origins of grammaticalization in the verbalization of experience.
speech communities through the immigration of multiple Linguistics 48, 1–48.
populations in colonial situations (new dialect formation and Darwin, C.R., 1859. On the Origin of Species, first ed. John Murray, London. Available
creole formation; Trudgill, 2004; Mufwene, 2001, 2008). As in at: http://darwin-online.org.uk/contents.html (accessed 30.07.12).
Darwin, C.R., 1882. The Desent of Man, and Selection in Relation to Sex, second ed.
the case of the selection of certain linguistic variants in a single John Murray, London. Available at: http://darwin-online.org.uk/contents.html
speech community, selection of linguistic variants from the (accessed 30.07.12).
members of different speech communities that come together Dawkins, R., 1976. The Selfish Gene. Oxford University Press, New York.
in a colonial situation is influenced by social factors that Ellis, N.C., Larsen-Freeman, D. (Eds.), 2009. Language as a Complex Adaptive
System. (Special Supplement to Language Learning 59 (4).)
operate in the circumstances of the colonization (the
Hull, D.L., 1988. Science as a Process: An Evolutionary Account of the Social and
dominance of some groups and enslavement of others, the Conceptual Development of Science. University of Chicago Press, Chicago, IL.
demographic proportions of speakers, and so on). Kirby, S., 1999. Function, Selection and Innateness: The Emergence of Language
The employment of an evolutionary framework for under- Universals. Oxford University Press, Oxford.
standing language change allows for the development of Mesoudi, A., Whiten, A., Laland, K.N., 2004. Is human cultural evolution Darwinian?
Evidence reviewed from the perspective of the origin of species. Evolution 58, 1–11.
mathematical models of mechanisms and processes of Mufwene, S., 2001. The Ecology of Language Evolution. Cambridge University Press,
language change. Such models are agent-based models, in Cambridge.
which entities that represent speakers (as human agents) Mufwene, S., 2008. Language Evolution: Contact, Competition and Change.
interact with each other and produce linguistic structures, Continuum, London.
Pagel, M., 2009. Human language as a culturally transmitted replicator. Nature
constrained by a range of competing factors. The process of
Reviews Genetics 10, 405–415.
replicating these linguistic structures feeds back into future Reich, D., et al., (63 authors), 2012. Reconstructing native American population
linguistic behavior of the agents in a cumulative fashion, history. Nature 488, 370–374.
leading to evolutionary change in the language produced by Ritt, N., 2004. Selfish Sounds and Linguistic Evolution: A Darwinian Approach to
the agents/speakers. In other words, these mathematical Language Change. Cambridge University Press, Cambridge.
Steels, L., 2011. Modeling the cultural evolution of language. Physics of Life Reviews
models represent language as a complex adaptive system 8, 339–356.
(Ellis and Larsen-Freeman, 2009). Tomasello, M., 2008. Origins of Human Communication. MIT Press, Cambridge, MA.
The relationship between evolution and language is suffi- Trudgill, P., 2004. New-dialect Formation: The Inveitability of Colonial Englishes.
ciently parallel that both can be subsumed under a generalized Edinburgh University Press, Edinburgh.
model of evolutionary processes, and specific mathematical
models can be applied to both. Although many linguists, Relevant Websites
particularly historical linguists, have observed the parallels
between evolution and language, a reassessment of the nature http://www.evolutionarylinguistics.org – Evolutionary Linguistics.
of theories and models of language based on an evolutionary http://www.eva.mpg.de – Max Planck Institute for Evolutionary Anthropology.
perspective is a relatively recent phenomenon in linguistics. http://www.evolang.org – Evolution of Language International Conferences.