Cultural Evolution of Language 315 Cultural Evolution of Language K Smith, University of Edinburgh, Edinburgh, UK Cultural Transmission and ß 2006 Elsevier Ltd. All rights reserved. Cultural Evolution Children learn language from examples of language Introduction use, often called the Primary Linguistic Data (PLD). The PLD must be the linguistic behavior of other Language is a culturally transmitted system – children language users, as children acquire the language of learn the language of their speech community on the their linguistic community. This linguistic behavior is basis of the linguistic behavior of that community. a consequence of the internalized linguistic knowl- This cultural transmission can lead to the cultural edge of these other individuals, and the PLD therefore evolution of the linguistic system, whereby language reflects the linguistic knowledge of the individuals in changes over time as a consequence of pressures act- a child’s speech community. These individuals have ing on it during its cultural transmission. grammars that define their knowledge of language Cultural evolution potentially offers an explana- and that knowledge of language guides their linguistic tion for the origins of a linguistic system with the behavior. The child is therefore attempting to acquire design features and functionality of human language, a system of linguistic knowledge, a grammar, based on as well as an explanation for the subsequent change data that are the consequence of the linguistic knowl- of such systems – the processes that explain the ways edge, or grammars, of other individuals. Andersen in which languages change on a historical timescale summarizes this relationship between linguistic data can also explain how languages themselves emerged. and mental grammar as follows: ‘‘the verbal output of Furthermore, the fact that languages themselves any speaker is determined by the grammar he has can evolve has implications for linguistic theories internalized . . . any speaker’s internalized grammar more generally. One goal for linguistic theory is to is determined by the verbal output from which it has achieve explanatory adequacy (Chomsky, 1965) – been inferred’’ (1973: 767). an explanation of how language is acquired on the Language therefore matches the sorts of general basis of linguistic data. This goal for linguistic theory definition of culturally transmitted systems given by explicitly links features of language with components anthropologists or evolutionary theorists concerned of an innate capacity for language. However, the with culture (e.g., Boyd and Richerson, 1985). Lan- evolutionary dynamics arising from cultural trans- guage, like other cultural systems, is transmitted from mission mean that we cannot infer the characteristics generation to generation via a process of learning of the language faculty directly from a single episode from the behavior of others. of language acquisition, but must consider how any What are the basic steps in this transmission innate language faculty will interact with cultural of language? Obviously, internalized knowledge of transmission to shape languages over cultural time. language (a mental grammar) is not directly physical- I begin by explaining exactly what is meant when ly transferred from individual to individual. Rather, we say that language is culturally transmitted and language is transmitted via a more indirect route – that this cultural transmission leads to cultural evolu- following Andersen (1973), Hurford (1990), and tion and then I discuss what kinds of linguistic phe- Kirby (2002), Hurford explains that ‘‘a language per- nomena can be explained in these terms. I then go on sists historically through successive instantiations in to consider three theories of the cultural evolution of two quite different media: (1) mental grammars of language, each of which proposes rather different individuals, and (2) public behavior in the form of mechanisms for cultural evolution. Note that this is utterances’’ (Hurford, 2002: 302). Hurford dubs this not an exhaustive list of all possible mechanisms – the ‘expression/induction’ cycle. Language is trans- rather, these three theories offer a contrasting range mitted via expression (application of internal knowl- of accounts for the evolution of linguistic structure. edge to produce observable behavior) and induction Finally, I consider the ways in which cultural evolu- (acquisition/learning of a mental grammar based on tion has significance for the wider linguistic commu- some observed set of behavior). nity. Although such evolutionary processes have The cultural transmission of language, based on its obvious implications for our understanding of lan- repeated transformation from grammar to data to guage origins and language change, their impact po- grammar and so on, leads to the possibility of cultural tentially runs deeper, to goals that lie at the heart of evolution. Cultural evolution, in the broadest poss- classic synchronic linguistics. ible terms, is change in a culturally transmitted system 316 Cultural Evolution of Language over time. The cultural evolution of language, then, is Cultural Evolution as a Consequence of Cultural change in a linguistic system over time as a conse- Transmission quence of its cultural transmission. The very fact of cultural transmission can drive the What determines the way in which a culturally cultural evolution of language. This theory as outlined transmitted system, such as language, will evolve? here is based on ideas presented and developed by I briefly outline below some mechanisms that drive Simon Kirby, Jim Hurford, Morten Christiansen, and cultural evolution. However, before doing so, it Terry Deacon (see, e.g., Christiansen and Devlin, 1997; is worth considering how far such an account can Christiansen and Ellefson, 2002; Deacon, 1997; take us. Hurford, 2002; Kirby, 2001, 2002). For convenience, One obvious goal would be to explain language I refer to these theories as transmission theories. change on a historical timescale – language change In order for a language to survive over repeated is manifestly a consequence of the cultural transmis- episodes of cultural transmission, it must be possible sion of language. One theory to be discussed below for language learners to learn that language from the (Croft, 2000) takes this as a primary goal. But how PLD, and so languages must be learnable. The key far back in time must we go before cultural transmis- insight of transmission theories is that this learnabil- sion and evolution stop being relevant? 100 years? ity constraint introduces an evolutionary pressure 1000 years? 100 000 years? As long as language that results in languages themselves changing over has been culturally transmitted, cultural processes time so as to become more learnable. Transmission and cultural evolution will have been at work. theories therefore emphasize the fact that language Consequently, theories of the cultural evolution of itself adapts to its medium of transmission. language are increasingly regarded as useful tools I summarize two ways in which the learnability for explaining the origins of aspects of linguistic of linguistic systems can vary: due to their gener- structure. alizability from a finite set of data (bottlenecked The appeal of taking the process of cultural evolu- transmission) and due to learnability considerations tion to its logical explanatory limits is that it poten- arising from the biases of language learners (biased tially offers a single, unified mechanism or set of transmission). mechanisms that can explain the origins of linguistic systems and also their subsequent change. In other Language Evolution as a Consequence of the Data words, cultural transmission potentially offers a Bottleneck If a linguistic system is to persist over uniform mechanism that explains both the genesis cultural time, it must pass through the cycle from data of language (a qualitative shift from a nonlinguistic to grammar to data multiple times. One consequence system to a linguistic system) and language change of this passage through the medium of linguistic data (subsequent quantitative shift), at whatever temporal is that language must be learnable from a finite set of granularity is required. Of course, this is not to deny data produced under real-world conditions. that biological evolution plays a role in explaining Languages are capable of expressing an infinite the origins of language – even if we subscribe to the range of concepts, and any member of this infinite strongest cultural account, we must still explain the array of expressions is interpretable in turn. Acquir- emergence of the capacity for a culturally transmitted ing a language therefore entails the acquisition of a system of communication. Indeed, there may have system for producing and understanding such an infi- been significant interactions between biological nite set of meaningful utterances. However, the sys- and cultural evolution during this process (see, e.g., tem for generating this infinite set of utterances must Deacon, 1997). be acquired from a finite set of data – it is necessarily true that language learners do not see all the sentences Mechanisms for Cultural Evolution of a language during the language learning process, because this would take an infinite amount of time. What causes a linguistic system to change over time? This data bottleneck is one aspect of the poverty of There are several possible mechanisms for the cultur- the stimulus problem, which is typically advanced as al evolution of language. I outline three prominent an argument suggesting that linguistic structure must theories that tackle this question in rather different be largely prespecified in language learners. ways and offer a range of possible answers. Briefly, Simon Kirby and colleagues (see, e.g., Kirby et al., the three mechanisms proposed to drive the cultural 2004 for references) have argued that this data evolution of language are as follows: (1) cultural bottleneck introduces a pressure for recursively com- transmission itself; (2) language use; and (3) ultimate positional linguistic structure. In other words, two function for reproduction. significant design features of language (recursion Cultural Evolution of Language 317 and compositionality) can be explained as a con- systems that are not generalizable. Languages them- sequence of the cultural transmission of language selves will evolve over cultural time to be more and and its subsequent cultural evolution. Bottlenecked more structured and more and more generalizable, as transmission theories therefore provide a causal link a consequence of the pressure introduced by the data between a feature of language’s cultural transmis- bottleneck. Regular morphology and recursive compo- sion (the data bottleneck) and universal properties sitionality therefore represent adaptations by language of linguistic systems. in response to its cultural transmission. Kirby argues that the data bottleneck results in This theory provides a cultural explanation for pressure for certain types of linguistic structure: these basic design features of language. However, ‘‘This learning bottleneck leads inevitably to the instances of irregularity and noncompositionality are emergence of a language in which structure is not uncommon in language. Morphological para- preserved in the mapping between semantics and digms exhibit some irregularity. Idioms are an obvi- strings’’ (Kirby, 2002: 199). Structure-preserving ous example of noncompositionality. More generally, mappings, ubiquitous in language, are manifested it has been argued that certain social functions of in compositional syntax and regular morphology. language are encoded and expressed in a noncompo- These systems are structure-preserving in that simi- sitional way, as holistic associations between complex lar signals (either syntactically or morphologically meanings and unanalyzed signals (Wray, 2002). Bot- structured) convey similar (once again, structured) tlenecked transmission theories also offer an expla- meanings – structure in semantic space is reflected in nation for such deviations from the general properties the structure of signals, as a consequence of the regu- of regularity and compositionality. lar and compositional process of forming complex It was assumed above that all meanings that lan- signals. guage users wish to express are equally frequent. This The distinctive feature of regular and compositional is not the case – some concepts presumably need to be morphosyntactic systems is that they are generalizable: expressed more frequently than others. The signals Individual structured meaning–signal associations in a associated with frequently expressed concepts will compositional system can be generalized from other reliably occur in the data that language learners are meaning–signal associations generated by that system, exposed to, even in the presence of a data bottleneck, by virtue of identifying the relevant generalizations. due to their frequency. Consequently, these associa- Consequently, a regular and compositional system tions are under little pressure to be part of a regular can be stable over multiple generations of cultural or compositional system. In contrast, infrequently transmission even if language learners do not observe expressed meanings are unlikely to make it into the full set of possible meaning–signal associations in the PLD and are therefore under increased pressure the data they learn from. This is, of course, the situation to be expressed compositionally – the only way such that language learners face – they are attempting to meaning–signal associations can survive repeated epi- acquire a system that is capable of producing an infinite sodes of cultural transmission is if they are capable of set of meaning–signal associations on the basis of a being generalized. finite set of examples. This frequency–irregularity prediction inherent in In contrast, noncompositional or irregular systems bottlenecked transmission theories matches the lin- are not generalizable. By definition, the signal asso- guistic data. To take a specific example, the 10 most ciated with a particular meaning in such a system is frequently used verbs in English have irregular past- arbitrary in terms of structure preservation, and there tense forms. In morphological systems more general- is no compositional or regular relationship between ly, frequent forms tend to be irregular. Larger holistic elements of meaning and components of signal. Lan- units (e.g., idioms) also seem to occur frequently guage learners must observe these idiosyncratic, un- (Wray, 2002). This relationship between frequency structured associations in order for these associations and irregularity naturally falls out of the bottlenecked to survive. Given the data bottleneck, this cannot be transmission account. the case for all meaning–signal associations produced by a noncompositional system. Consequently, non- Language Evolution as a Consequence of Learning compositional systems are unstable over cultural time. Bias Bottlenecked transmission theories link learn- The data bottleneck implicit in cultural transmission ability to properties of the data available to language therefore introduces a pressure for generalizability. learners. Biased transmission theories focus on anoth- Linguistic systems (or subparts of linguistic systems) er aspect of the learnability problem: fit to the biases that are highly generalizable, such as recursively of language learners. Language learners must have compositional syntax or regular morphology, will be some strategy for learning a language. If all aspects more likely to survive cultural transmission intact than of linguistic structure are not equal under this 318 Cultural Evolution of Language learning strategy – if some particular grammatical morphosyntactic systems with the ease or difficulty construction, or morphological paradigm, or config- with which children will acquire such systems. uration of lexical items, or phonological system, is Kenny Smith (see, e.g., Smith, 2004) has shown more difficult to acquire than an alternative – then we that the cultural consequence of these biases will should expect to see such constructions disappear be the evolution of linguistic systems that exhibit from the linguistic system over time. As Deacon puts maximal transparency – systems where the lexicon it, ‘‘[l]anguage operations that can be learned quickly exhibits a perfect one-to-one correspondence be- and easily by children will tend to get passed on to the tween meanings and words, and the morphosyntactic next generation more effectively and intact than those component exhibits a perfectly compositional, regu- that are difficult to learn. So, languages should lar system for expressing complex meanings, with change through history in ways that conform to a single unambiguous token in the surface form for children’s expectations’’ (Deacon, 1997: 110). each element of meaning. This process of cultural Morten Christiansen and colleagues (see, e.g., evolution can potentially explain certain diachronic Christiansen and Ellefson, 2002 for references) phenomena, such as various kinds of leveling and explain certain typological distributions in terms of simplification of morphological paradigms. These cultural evolution driven by the biases of language same mechanisms can also explain the emergence learners. To give one example, languages with a con- of such transparent linguistic systems from unorga- sistent ordering of heads across phrasal categories nized, nonfunctional systems of meaning–signal (e.g., either consistently head-initial or consistently mappings. Transparent linguistic systems are obvi- head-final) are more common than languages that are ously optimal for communication, in that they allow inconsistent with respect to head order. Christiansen every possible meaning to be expressed unambigu- and Devlin (1997) explain this crosslinguistic tendency ously. However, it is important to emphasize that as a cultural consequence of the biases of a general- such systems can evolve culturally without any purpose sequential learning device (modeled using specific pressure for communication. Transparent an artificial neural network). They found that such systems are simply the most learnable kind, assuming learning devices make more errors when attempting that language learners have the types of biases sug- to learn languages with inconsistent head ordering gested by the developmental literature outlined and, furthermore, the patterns of errors made by their above. general-purpose sequence learning device mirrored the To summarize, the basic evolutionary mechanism distribution of languages in the world. Crucially, cul- provided by transmission theories is adaptation of tural evolution driven by learner bias explains this link language itself to its medium of transmission. This – languages that are hard to learn (e.g., because of leads to adaptations that are functional from the inconsistent head ordering) are rare, because they perspective of language. This is not to say that these do not conform to the biases of language learners and adaptations cannot also be of use to language users, are therefore less likely to persist through repeated but this is a side effect of the cultural evolution of episodes of cultural transmission than easy-to-learn language, rather than a driving force for linguistic languages. evolution. Other work on biased transmission has focused on Cultural Evolution as a Consequence of how learning biases with respect to transparency of Language Use the meaning–signal mapping might impact language over cultural time. The developmental linguistics lit- The second theory of cultural evolution that I review erature suggests that child language learners have here is that presented in Croft (2000), which repre- difficulty with synonymy and homonymy – children sents an interesting synthesis of both functional and expect that a single object will not have several possi- sociohistorical accounts of language change. Croft’s ble names (Markman, 1989) and that a single word main aim is to provide a unified set of mechanisms for will not have several possible meanings (Mazzocco, explaining a variety of processes of language change, 1997). These amount to child biases in favor of trans- and the level of linguistic detail supporting his argu- parency in the lexical system, where each distinct ment is impressive. Croft proposes two primary meaning maps to a single unique surface form. mechanisms for cultural evolution. First, language At the morphosyntactic level, Slobin has claimed users will introduce innovations into the linguistic that child language learners ‘‘strive to maintain a system during the process of communication. Second, one-to-one mapping between underlying semantic social factors determine the differential cultural structures and surface forms’’ (Slobin, 1977: 186). transmission of the competing linguistic variants Slobin explicitly linked the degree of transparency in that these innovations introduce. Cultural Evolution of Language 319

Croft proposes several mechanisms for functional particular form is reinterpreted by a hearer as being innovation – the introduction of new linguistic forms provided by context (physical or linguistic). In other or structures into a language. The two main mechan- words, the speaker uses some form F to express a isms are form–function reanalysis,where‘function’ is particular meaning M, but the hearer reinterprets roughly synonymous with ‘meaning,’ and intraference. the form–meaning mapping in such a way that M is According to Croft’s theory, form–function reanaly- provided somewhere else – say, in another part of the sis occurs during language use. Language use involves utterance. Consequently, the form F becomes seman- attempting to communicate a particular meaning be- tically bleached – for the hearer, form F does not tween a speaker and a hearer using a set of linguistic convey the meaning M. conventions that relate form and meaning. How- In addition to outlining several other subtypes of ever, the meaning behind an utterance is necessarily form–function analysis, Croft suggests that processes fuzzy, as is the system linking meaning and form – such as grammaticalization (which is often described interlocutors have no way of perfectly identifying as depending on context-induced reinterpretation) the communicative intentions of their partners, or of can be subsumed under these general processes of gaining direct access to the meaning–form mappings form–function reanalysis. employed. Instead, both meaning and the system of What consequences will hypoanalysis and hyper- meaning–form mappings must be inferred during use. analysis have for a language? Croft suggests that This error-prone process of inference allows innova- the repeated application of these two types of reanal- tions to emerge in the linguistic system – an utterance ysis will result in the emergence of syntagmatic that embodies a particular structured mapping be- isomorphism – a one-to-one correspondence between tween meaning and form for one interlocutor may elements of meaning and elements of form within be interpreted by another as encoding a somewhat each particular utterance. Forms that are associated different meaning and/or a different form–meaning with little semantic content will become increasingly mapping. Croft listed several types of form–function bleached (via hyperanalysis) and will be eliminated reanalysis that can occur due to this process of infer- or will acquire a specific semantic function (via hy- ence of meaning during language use. I focus here on poanalysis), resulting in semantic ‘load’ being spread hypoanalysis and hyperanalysis. out among the elements of an utterance. Biased trans- Hypoanalysis occurs when a particular element of mission and form–function reanalysis therefore offer meaning that is conventionally given by context – alternative explanations for linguistic transparency. either the context in which the form is used or the In principle, were hyperanalysis and hypoanalysis linguistic context provided by other parts of an utter- the only forces at play, we could imagine a linguistic ance – is reinterpreted by the speaker as being asso- system with a transparent but idiosyncratic meaning– ciated with a particular linguistic form. For example, form mapping for every distinct meaning–form the speaker may use form F in a context that provides pair. Croft introduces a further mechanism, intrafer- meaning M. The hearer instead interprets M as being ence, which operates paradigmatically, across differ- conveyed by form F and, for the hearer, form F ent individual structured form–meaning mappings. acquires additional semantic content. Intraference is driven by overlap in meaning or This process of hypoanalysis can be illustrated with function – different linguistic forms may convey the respect to the umlaut change in Germanic languages. same meaning, and language users may identify this In Pre-Old English, plurality could be expressed using semantic overlap. This allows overlapping forms to an -[i] suffix (e.g., singular [go:s] corresponding to be used in innovative ways, as alternative ways Modern English ‘goose’, plural [go:si]). Anticipatory of conveying the common meaning. For example, assimilation subsequently resulted in fronting of the German underwent a similar umlaut change to stem vowel in the plural as a result of the -[i] suffix English. Though the fronted-vowel means of expres- (e.g., singular [go:s], plural [gø:si]). In early Old sing plurality in English never achieved high produc- English, the -[i] suffix was lost by a further phonolog- tivity, in Early New High German this means of ical change, leaving, e.g., singular [go:s], plural [gø:s]. expressing plurality was extended throughout large Under these circumstances, the alteration between parts of the nominal system, via intraference (Croft, back [o:] and fronted [ø:] acquires semantic content 2000: 128). via hypoanalysis – the fronted version of the vowel Intraference is essentially a mechanism for general- now conveys plurality, as is the case in Modern ization – identification of the common meaning of English ‘geese.’ two or more linguistic forms and subsequent use The opposite of hypoanalysis is hyperanalysis. of either form to convey that meaning. Croft sug- Hyperanalysis occurs when a particular element of gested that intraference is a mechanism for what is meaning that is conventionally associated with a often called analogical extension, resulting in, for 320 Cultural Evolution of Language example, regularization of morphological paradigms. form–meaning mappings during the continual nego- Croft, in common with bottlenecked transmission tiation for meaning inherent in communication, theories, includes a caveat regarding frequency in his Nowak and colleagues see a much more direct link explanation of this regularization process. High- between language’s function for communication and frequency meaning–form associations are less prone its cultural evolution. to intraference from other form–meaning mappings The argument runs as follows: individuals in a due to being ‘‘independently represented units in the population have different grammars, and different mind. Frequent forms thus develop and maintain grammars vary in the communicative payoff that irregularities’’ (Croft, 2000: 149). they offer an individual using that grammar (depen- The mechanisms outlined above are Croft’s dent, of course, on the proportion of the different mechanisms for innovation – introduction of new grammars in the population). Successful communica- linguistic devices for fulfilling a particular communi- tors have disproportionate access to reproduction, cative function. Croft also provides a mechanism for and children acquire the grammar of their biological competition between alternative form–function map- parent(s). Under these assumptions, those grammars pings, which is essentially social. There is an estab- that offer the highest communicative payoff increase lished tradition in the sociolinguistic literature that in frequency in a population – grammars that maxi- accounts for the linguistic behavior of individuals mize communicative (and therefore reproductive) in terms of prestige and covert prestige. In these success will be selected and will proliferate. This is terms, the choice of a particular linguistic form in essentially a process where cultural evolution closely preference to alternative forms constitutes an act of mirrors biological evolution under natural selection – identity on the part of the speaker. In Croft’s model, cultural evolution favors traits that lead to biological such acts of identification determine the cultural reproduction, because biological reproduction is the spread of particular (functionally motivated) systems only route to cultural persistence. of form–meaning mapping. Nowak et al. have used this general framework Can Croft’s theory achieve the maximal possible to look at the function-driven cultural evolution of explanatory reach – can it, in principle, explain both vocabulary systems and to quantify how accurate language change (its intended explicandum) and lan- language acquisition must be if a population is to guage origins? Croft’s mechanism for the cultural converge on a shared grammar. In terms of broader transmission of linguistic variants is motivated by application, their model of cultural evolution dove- inherently arbitrary social considerations. However, tails nicely with models of biological evolution to Croft envisions a constant input of new, functional provide evolutionary explanations for the capacity variants into the arbitrary, socially motivated process for phonemic coding and compositionality. As such, of propagation. This potentially provides a mecha- it constitutes a useful and flexible tool for inves- nism for the emergence and subsequent evolution tigating a wide range of evolutionary questions of communicatively functional, grammaticalized about language, using established techniques from linguistic systems. evolutionary biology. The value of this theory is more questionable Cultural Evolution as a Consequence of when we move beyond questions of language origins Reproductive Utility to consider the subsequent change in linguistic sys- A third mechanism for the cultural evolution of lan- tems (although it should be emphasized this was not guage is provided by Martin Nowak and colleagues the primary explanatory goal of this theory). It is (see Nowak and Komarova, 2001, for review). This highly unlikely that the majority of attested lan- is an instance of a more general theory of cultural guage changes are driven by differential reproduction. evolution, sometimes called the ‘‘natural selection of The main problem with this biologically motivated cultural variations’’ (Boyd and Richerson, 1985), and theory of cultural evolution, then, is that it cannot is included here primarily as an example of a theory (or should not) provide a uniform mechanism for of cultural change that is heavily influenced by thinking about the cultural origins and subsequent approaches adopted from evolutionary biology. evolution of linguistic systems. This has been ac- Nowak and colleagues, like Croft, see language knowledged by the proponents of this theory, who as a means for communication. Languages, or ele- state that ‘‘Neutral [i.e., not reproductively driven] ments of the linguistic system, are useful inasmuch language dynamics provide an appropriate descrip- as they further this goal. Whereas Croft envisages tion for many language changes studied in historical language changing as a consequence of reanalysis of linguistics’’ (Komarova and Nowak, 2003: 457). Cultural Evolution of Language 321

Cultural Evolution and the languages are culturally stable and therefore, ulti- Linguistic Enterprise mately, possible from the viewpoint of cultural evolu- tion. To put it in cruder terms, we might be tempted Theories of the cultural evolution of language inter- to look at the linguistic systems of the world, recog- face at various points with other areas of linguistics. nize that all such linguistic systems are recursively For example, such theories may be motivated by ex- compositional, and conclude that this recursive com- perimental evidence relating to children’s language- positionality constitutes part of the learner’s innate learning biases, or by the rich range of data available prespecification of a possible human language. The on language change, or they might explain a particu- cultural account of this linguistic universal shows that lar typological distribution, or a characteristic design this is not necessarily a valid conclusion to draw – the feature of language. These are all worthwhile links cultural transmission of language, and its concomi- between theories dealing with the cultural evolution tant evolution, offers another possible explanatory of language and other subfields of linguistic research mechanism for this constraint on the possible forms and bode well for continuing integration of cultural of language. In more general terms, any cultural ways of thinking into these fields. dynamic is likely to obscure the relationship between As mentioned in the Introduction to this article, attested features of language and innate constraints, there is one further, more fundamental way in which due to the fact that cultural evolution also acts to an understanding of cultural evolution impacts on constrain the range of possible human languages. As linguistic theory. Noam Chomsky, in his 1965 book such, the cultural transmission of language potential- Aspects of the theory of syntax and subsequent ly plays a key role in the central goal of developing works, gives a set of criteria for evaluating the explanatorily adequate linguistic theories. adequacy of a linguistic theory, the ultimate crite- Second, taking cultural evolution seriously may rion being that linguistic theories should achieve also impact on the second foundation of Chomsky’s explanatory adequacy. A linguistic theory achieves explanatorily adequate theory, which requires that explanatory adequacy if it can account for how a children be provided with a strategy for selecting a language learner acquires a descriptively adequate grammar of the appropriate form that is compatible grammar on the basis of exposure to linguistic data, with the linguistic data. The biased transmission theo- where those linguistic data are of the sort that a ries of cultural evolution turn this issue on its head. language learner can expect to be exposed to. Rather than the onus being on the child to identify a Chomsky argues that such an explanatorily ade- grammar consistent with the data, the onus is on quate theory must provide an account of children’s language itself to ensure that it is identified by chil- innate linguistic knowledge, characterized informally dren from the data it produces. If a particular aspect as a ‘language instinct’ or more formally as ‘‘first, a of linguistic structure is routinely misidentified by linguistic theory that specifies the form of a possible language learners, it will not persist. Conversely, if human language, and second, a strategy for selecting language learners find it easy to identify a particular a grammar of the appropriate form that is compatible aspect of structure from data, they will themselves with the primary linguistic data’’ (Chomsky, 1965: produce further examples of this structure, which will 25). As previously argued by, e.g., Hurford (1990), lead to the continued survival of that linguistic struc- Kirby (2002), Kirby et al. (2004), and Brighton et al. ture. Cultural evolution provides a mechanism by (2005), the nature of these capacities cannot be con- which languages can adapt so as to make themselves sidered in isolation of the dynamics of cultural trans- readily identifiable, rather than placing the burden mission and cultural evolution. of explanation on the child’s innate knowledge of First, cultural evolution resulting from the cultural language. transmission of language can act as an alternative A theory of language that does not reckon with mechanism that constrains the range of possible these types of phenomena, arising from the cultural human languages. For example, the bottlenecked transmission of language, risks overestimating the transmission account predicts that only highly regular burden placed on language learners and is con- and highly compositional systems will be possible sequently in danger of overemphasizing the role human languages when viewed on any significant that innate linguistic knowledge must play in an cultural timescale. Language learners may be capable explanatorily adequate linguistic theory. of learning and representing systems of varying degrees of compositionality and as such these may constitute ‘possible human languages’ from the See also: Design Features of Language; Formal Models point of view of individual language learners. How- and Language Acquisition; Formulaic Language; Levels ever, only highly regular, recursively compositional of Adequacy, Observational, Descriptive, Explanatory; 322 Cultural Evolution of Language

Linguistic Universals, Chomskyan; Origin and Evolution regularity and irregularity.’ IEEE Transactions on Evolu- of Language. tionary Computation 5, 102–110. Kirby S (2002). ‘Learning, bottlenecks and the evolution of Bibliography recursive syntax.’ In Briscoe E (ed.). 173–203. Kirby S, Smith K & Brighton H (2004). ‘From UG to uni- Andersen H (1973). ‘Abductive and deductive change.’ versals: Linguistic adaptation through iterated learning.’ Language 40, 765–793. Studies in Language 28, 587–607. Boyd R & Richerson P J (1985). Culture and the evolution- Komarova N L & Nowak M A (2003). ‘Language dynam- ary process. Chicago, IL: University of Chicago Press. ics in finite populations.’ Journal of Theoretical Biology Brighton H, Kirby S & Smith K (2005). ‘Cultural selection 221, 445–457. for learnability: Three hypotheses underlying the view Macnamara J (ed.) (1977). Language learning and thought. that language adapts to be learnable.’ In Tallerman London: Academic Press. M (ed.). 291–309. Markman E M (1989). Categorization and naming in Briscoe E (ed.) (2002). Linguistic evolution through lan- children: Problems of induction. Cambridge, MA: MIT guage acquisition: Formal and computational models. Press. Cambridge: Cambridge University Press. Mazzocco M M (1997). ‘Children’s interpretations of Chomsky N (1965). Aspects of the theory of syntax. homonyms: A developmental study.’ Journal of Child Cambridge, MA: MIT Press. Language 24, 441–467. Christiansen M & Devlin J (1997). ‘Recursive inconsisten- Nowak M A & Komarova N L (2001). ‘Towards an evolu- cies are hard to learn: A connectionist perspective on tionary theory of language.’ Trends in Cognitive Sciences universal word order correlations.’ In Shafto M & 5, 288–295. Langley P (eds.). 113–118. Roca I M (ed.) (1990). Logical issues in language acquisi- Christiansen M & Ellefson M (2002). ‘Linguistic adapta- tion. Dordrecht: Foris. tion without linguistic constraints: The role of sequen- Shafto M & Langley P (eds.) (1997). Proceedings of tial learning in language evolution.’ In Wray A (ed.). the 19th Annual Cognitive Science Society Conference. 335–358. Mahwah, NJ: Lawrence Erlbaum Associates. Croft W (2000). Explaining language change: An evolu- Slobin D I (1977). ‘Language change in childhood and tionary approach. London: Longman. history.’ In Macnamara J (ed.). 185–221. Deacon T (1997). The symbolic species. London: Penguin. Smith K (2004). ‘The evolution of vocabulary.’ Journal of Hurford J R (1990). ‘Nativist and functional explanations Theoretical Biology 228, 127–142. in language acquisition.’ In Roca I M (ed.). 85–136. Tallerman M (ed.) (2005). Language origins: Perspectives Hurford J R (2002). ‘Expression/induction models of lan- on evolution. Oxford: Oxford University Press. guage evolution: Dimensions and issues.’ In Briscoe E Wray A (ed.) (2002). The transition to language. Oxford: (ed.). 301–344. Oxford University Press. Kirby S (2001). ‘Spontaneous evolution of linguistic struc- Wray A (2002). Formulaic language and the lexicon. ture: An iterated learning model of the emergence of Cambridge: Cambridge University Press.

Culture in Language Teaching C Kramsch, University of California, which a national language was spoken by a homoge- Berkeley, CA, USA neous national citizenry. In the last 10 years, this ß 2006 Elsevier Ltd. All rights reserved. unitary conception of one language equaling one national culture has become problematic. National standard languages have come to be seen as arbitrary Culture has always been an integral component of constructions of the 19th-century nation states as language teaching. Until World War II, culture used much as the social and political institutions that to be seen as the literate or humanities component constitute national cultures. At a time of growing of language study. After the war and following the economic and political globalization, when cultural communicative turn in language pedagogy, it became encounters are increasingly mediated by information synonymous with the way of life and everyday technologies, whose and what culture(s) should we behaviors of members of speech communities, teach: national, regional, or global culture? Urban or bound together by common experiences, memories, rural culture? High brow or popular culture? Oral, and aspirations. These communities were seen as written, or cyberculture? Gay culture? Marketing grounded in the nation – the national context in culture? And what disciplinary discourse should we