Kirby, S.: Fitness and the selective adaptation of language. In Approaches to the Evolution of Language, J. R. Hurford, M. Studdert-Kennedy, and C. Knight (eds.), Cambridge University Press (1998) 359-383  Home/Search   Document Details and Download   Summary   Related Articles   Check

....hypothesis space. Such preferences can arise from general principles of learning such as Occam s Razor in the form of a simplicity metric over the representational framework employed (e.g. Mitchell, 1997) from functional considerations such as the relative parsability of di erent triggers (e.g. Kirby, 1998), or from prehistorical contingent properties of grammars genetically assimilated during the period of adaptation for the language faculty (Briscoe, 2000a) For example, in a parametric LA the learner might select between alternative grammars compatible with the triggering data (so far) by ....

Kirby, S. (1998) `Fitness and the selective adaptation of language' in Hurford, J., Studdert-Kennedy, M., and Knight, C. (ed.), Approaches to the Evolution of Language, Cambridge University Press, Cambridge, pp. 359-383.

....space. Such preferences can arise from general principles of learning such as Occam s Razor in the form of a simplicity metric over the representational framework employed (e.g. Mitchell, 1997) from functional considerations such as the relative parsability of different triggers (e.g. Kirby, 1998), or from prehistorical contingent properties of grammars genetically assimilated during the period of adaptation for the language faculty (Briscoe, 2000a) For example, a parametric LA might select between alternative grammars compatible with the triggering data (so far) by selecting the one ....

Kirby, S. (1998) `Fitness and the selective adaptation of language' in Hurford, J., Studdert-Kennedy, M., and Knight, C. (ed.), Approaches to the Evolution of Language, Cambridge University Press, Cambridge, pp. 359--383.

.... we also need to characterize the environment in which such a device could emerge and evolve (e.g. Pinker and Bloom, 1990) This environment must have included (proto)language(s) capable of creating (natural) selection pressure in favour of more efficient and reliable language acquisition (e.g. Kirby, 1998). In turn, the evolving language acquisition device created (linguistic) selection pressure in favour of more learnable languages (e.g. Kirby and Hurford, 1997a) Inductive bias in language acquisition emerges as a direct consequence of this coevolutionary process. Section 1.1 briefly surveys the ....

....by positing that the probability of successful interaction is partly a function of the WML of the sentence type chosen. An alternative assumption is that maturational working memory limitations will decrease the chances of less parsable sentence types functioning as effective triggers (see e.g. Kirby, 1997, 1998, 1999 for a similar position) To simulate this scenario, the same set of runs was done with random reproduction of LAgts but with memory limitations during learning. The results show a very similar pattern to those reported above, though the selection effect is weaker and it is only when the ....

[Article contains additional citation context not shown here]

Kirby, Simon (1998) `Fitness and the selective adaptation of language' in Jim Hurford, Michael Studdert-Kennedy and Chris Knight (ed.), Approaches to the Evolution of Language, Cambridge: Cambridge University Press, pp. 359--383.

....operations for copying bits of the database onto the blackboard. Rather, they will be given additional set of symbols to be used for communication, and a set of operations for learning symbols by guessing their meaning. The basic principles of such a system are clear from previous work, such as [17, 16, 22, 19, 2, 4, 12, 11, 13, 9]. However, a word of caution: the complexity of the additional speci cation and programming task is large, and the amount of CPU time that would be needed for agents to develop interesting languages from scratch may also be large. The negative character of the results in [5] may be indicative of ....

Simon Kirby, to appear. \Fitness and the selective adaptation of language", in J. Hurford, M. Studdart-Kennedy & Chris Knight (eds.) The Evolution of Human Language, CUP.

.... pressure favouring such a development, and its subsequent maintenance and refinement, is only coherent given a coevolutionary scenario in which a (proto)language supporting successful communication within a population had already itself evolved on a historical timescale (e.g. Hurford, 1987; Kirby, 1998; Steels, 1998) and continued to coevolve with the LAD (e.g. Briscoe, 1997, 1998, 2000a) The model of the LAD presented here builds on and extends previous work in the parameter setting framework (e.g. Chomsky, 1981; Clark, 1992; Gibson and Wexler, 1994; Niyogi and Berwick, 1996; Briscoe, 1997, ....

.... linguistic variants (e.g. Kroch, 1989) The extension of the theory of grammatical acquisition beyond parameter setting to one in which an infinite range of grammars can, in principle, be acquired would underpin a coevolutionary account of the development of human language and of the LAD (e.g. Kirby, 1998). The paper begins by summarizing (x2) the model of the LAD described in Briscoe (1997, 1998, 2000a) and experiments with (evolving) populations of language agents (LAgts) defined in terms of this model of the LAD (x3) It then (x4) describes the Bayesian extensions to this model designed to ....

[Article contains additional citation context not shown here]

Kirby, S. (1998) `Fitness and the selective adaptation of language' in Hurford, J., Studdert-Kennedy, M., and Knight, C. (ed.), Approaches to the Evolution of Language, Cambridge University Press, Cambridge, pp. 359--383.

....with internal preferences created by LP UG. In Briscoe (1998a, 1999, 2000) I address the question of how likely it is that the language faculty incorporates inductive bias. I argue that if the language faculty is the result of genetic assimilation (e.g. Waddington, 1942; Pinker and Bloom, 1990; Kirby, 1998), if there is correlation between LP UG and its genotypic specification (e.g. Mayley, 1996) and if the communicative success ratio (CSR) as defined in section 1.2, is interpreted as a measure of agent fitness, then it is highly likely that the language faculty incorporates inductive bias, ....

.... via a single macromutation or exaption (e.g. Berwick, 1998; Bickerton, 1998) is rendered irrelevant by the observation that organs which arise via exaptions of spandrels or by macromutations are still susceptible to subsequent modification by natural selection (e.g. Ridley, 1990; Lieberman, 1991, Kirby, 1998). Indeed in my simulations, language agents do begin with language faculties (either random or converged) and these are merely refined by subsequent evolution. If we accept that there is inductive bias in language acquisition, as I think we should, given both the evidence from the simulation ....

[Article contains additional citation context not shown here]

Kirby, S. (1998) `Fitness and the selective adaptation of language' in Hurford, J., Studdert-Kennedy, M., and Knight, C. (ed.), Approaches to the Evolution of Language, Cambridge University Press, Cambridge, pp. 359--383.

.... pressure favouring such an exaption, and its subsequent maintenance and refinement, is only coherent given a coevolutionary scenario in which a (proto)language supporting successful communication within a population had already itself evolved on a historical timescale (e.g. Hurford, 1987; Kirby, 1998; Steels, 1998) and continued to coevolve with the LAD (e.g. Briscoe, 1997, 1998a,b) The model of the LAD presented here builds on and extends previous work in the parameter setting framework (e.g. Chomsky, 1981; Clark, 1992; Gibson and Wexler, 1994; Niyogi and Berwick, 1996; Briscoe, 1997, ....

.... competing linguistic variants (e.g. Kroch, 1989) The extension of the theory of grammatical acquisition beyond parameter setting to one in which an infinite rage of grammars can, in principle, be acquired underpins a coevolutionary account of the development of human language and of the LAD (e.g. Kirby, 1998). The paper begins by summarizing (x1) the model of the LAD described in Briscoe (1997, 1998a,b) and experiments with (evolving) populations of language agents (LAgts) defined in terms of this model of the LAD (x2) It then (x3) describes the Bayesian and non parameter setting extensions to this ....

Kirby, S. (1998) `Fitness and the selective adaptation of language' in Hurford, J., Studdert-Kennedy, M., and Knight, C. (ed.), Approaches to the Evolution of Language, Cambridge University Press, Cambridge, pp. 359--383.

....human grammar and a set of biases (partially) ranking possible grammars by markedness. A variety of explanations have been offered for the emergence of an innate LAD with such properties based on saltation (Berwick, 1998; Bickerton, 1990, 1998) or genetic assimilation (Pinker and Bloom, 1990; Kirby, 1998). Formal models of parameter setting (e.g. Clark, 1992; Gibson and Wexler, 1994; Niyogi and Berwick, 1996; Brent, 1996) have demonstrated that development of a psychologically plausible and effective parameter setting algorithm, even for minimal fragments of UG, is not trivial. The account ....

.... pressure favouring genetic assimilation, and its subsequent maintenance and refinement, is only coherent given a coevolutionary scenario in which (proto)language(s) supporting successful communication within a population had already itself evolved on a historical timescale (e.g. Hurford, 1987; Kirby, 1998), probably with many of the constraints and biases subsequently assimilated already present in the (proto)language(s) as a consequence of linguistic selection, perhaps initially driven by quite general cognitive constraints such as working memory limitations. Here we report the results of a ....

Kirby, S. (1998) `Fitness and the selective adaptation of language' in Hurford, J., Studdert-Kennedy, M., and Knight, C. (ed.), Approaches to the Evolution of Language, Cambridge University Press, Cambridge, pp. 359--383.

....intersecting dependencies. The genetic assimilation of a language specific rule system (the UG component of the LAD) remains a theoretical possibility, even if the emergence of such abstract universals can be traced to non domain specific factors, such as working memory limitations (see also Kirby, 1998). Secondly, Deacon relies heavily on the starting small hypothesis and Elman s (1993) experiments training recurrent neural networks (RNN) to approximate recognition of context free languages. Whilst these experiments demonstrate a clear requirement for initially training on short sequences ....

Kirby, S. (1998) `Fitness and the selective adaptation of language' in Hurford, J., Studdert-Kennedy, M., and Knight, C. (ed.), Approaches to the Evolution of Language, Cambridge University Press, Cambridge, pp. 359--383.

....they can represent, but in what they can learn. The distinction between learning and representation is important when we consider how human languages have developed. For a natural language to be viable, it must not only be representable by its users, but also learnable by subsequent generations [6]. The learning and representational constraints of the human brain dictate the set of languages humans are able to understand and learn, and consequently the languages that have emerged. Recurrent neural networks (RNNs) have shown significant promise as computational models of various aspects of ....

S. Kirby. Fitness and the selective adaptation of language. In James Hurford, Chris Knight, and Michael Studdert-Kennedy, editors, Evolution of Language: Social and Cognitive Bases for the Emergence of Phonology and Syntax. in press, 1998.

.... pressure favouring such a development, and its subsequent maintenance and refinement, is only coherent given a coevolutionary scenario in which a (proto)language supporting successful communication within a population had already itself evolved on a historical timescale (e.g. Hurford, 1987; Kirby, 1998; Steels, 1998) and continued to coevolve with the LAD (e.g. Briscoe, 1997, 1998a,b) The model of the LAD presented here builds on and extends previous work in the parameter setting framework (e.g. Chomsky, 1981; Clark, 1992; Gibson and Wexler, 1994; Niyogi and Berwick, 1996; Briscoe, 1997, ....

.... linguistic variants (e.g. Kroch, 1989) The extension of the theory of grammatical acquisition beyond parameter setting to one in which an infinite range of grammars can, in principle, be acquired would underpin a coevolutionary account of the development of human language and of the LAD (e.g. Kirby, 1998). The paper begins by summarizing (x1) the model of the LAD described in Briscoe (1997, 1998a,b) and experiments with (evolving) populations of language agents (LAgts) defined in terms of this model of the LAD (x2) It then (x3) describes the Bayesian and non parameter setting extensions to this ....

[Article contains additional citation context not shown here]

Kirby, S. (1998) `Fitness and the selective adaptation of language' in Hurford, J., Studdert-Kennedy, M., and Knight, C. (ed.), Approaches to the Evolution of Language, Cambridge University Press, Cambridge, pp. 359--383.

....10 15 20 25 30 Final language Deprivation Figure 13: Final language attained against deprivation with 7 language stages, quality 0.5 and various ages for puberty. mating starts (i.e. puberty) It is likely that language independently evolved to be the right size to be learnt by puberty (Hurford Kirby 1997) 17 . Given this, if the critical period evolves to be timed with the end of normal language learning, then it will appear to be related to puberty. This, however, is an indirect causal connection. 10 Conclusions The main points raised by these results are: 1. The critical period is inevitable ....

....Whatever amount of language is available in the community in which the learner grows up will then affect the biological evolution of the learning strategies to efficiently acquire that language. This hypothesised cultural biological co evolution raises many issues that go beyond this paper (see Kirby 1997; Kirby Hurford 1997 for some discussion) 30 THE EVOLUTION OF INCREMENTAL LEARNING evolution of general purpose learning mechanisms put to the task of learning a culturally evolving system: language. The ways in which the particular task of learning language is nativised may not be ....

KIRBY, SIMON. 1997. Fitness and the selective adaptation of language. In Evolution of Language: Social and cognitive bases for the emergence of phonology and syntax, ed. by J. Hurford, C. Knight, & M. Studdert-Kennedy. To appear.

....linguistic competence of the adults because it is filtered through the arena of use [20] There are various pressures that operate during communication that will have a selective effect on the different linguistic variants that are being transmitted from generation to generation. In earlier work [23, 25, 22] Kirby has shown that the selective effect of the parser in the cycle of acquisition and use can give rise to language universals of the sort that typologists observe cross linguistically. It appears that languages adapt to aid their own survival over time. More correctly, proportions of competing ....

Simon Kirby. Fitness and the selective adaptation of language. In J. Hurford, C. Knight, and M. Studdert-Kennedy, editors, Evolution of Language: Social and cognitive bases for the emergence of phonology and syntax. 1997. To appear.

No context found.

Kirby, S.: Fitness and the selective adaptation of language. In Approaches to the Evolution of Language, J. R. Hurford, M. Studdert-Kennedy, and C. Knight (eds.), Cambridge University Press (1998) 359-383

No context found.

Kirby, S., (1998), Fitness and the selective adaptation of language, in Hurford, J., Knight, C., & Studdert-Kennedy, M., (eds.), Approaches to the Evolution of Language, pp. 359-383, Cambridge University Press.

No context found.

Simon Kirby. Fitness and the selective adaptation of language. In James R. Hurford, Chris Knight, and Michael Studdert-Kennedy, editors, Approaches to the evolution of language: social and cognitive bases, pages 359--383. Cambridge University Press, Cambridge, 1998. http://ling.ed.ac.uk/anonftp/pub/staff/ kirby/evolpaper.ps.

No context found.

Kirby, S. (1998a). Fitness and the selective adaptation of language. In Hurford, J., Knight, C., and Studdert-Kennedy, M., editors, Approaches to the Evolution of Language. Cambridge University Press.

No context found.

S. Kirby. Fitness and the selective adaptation of language. In J. Hurford, C. Knight, and M. Studdert-Kennedy, editors, Approaches to the Evolution of Language. Cambridge University Press, 1998.

No context found.

Kirby, S. (1997) Fitness and the Selective Adaptation of Language. In: Hurford, J., et.al. (ed.).

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC