Plunkett, K., & Marchman, V. (1993). From rote learning to system building. Cognition, 49, 21-69.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....a statistical learning mechanism is sufficient to account for complex rule like behavior, or whether additional rule learning mechanisms are needed. In the past this debate has primarily taken place within specific areas of language acquisition, such as inflectional morphology (e.g. Pinker, 1991; Plunkett Marchman, 1993) and visual word recognition (e.g. Coltheart, Curtis, Atkins Haller, 1993; Seidenberg McClelland, 1989) More recently, Marcus et al. 1999) have presented results from experiments with 7 month olds, apparently showing that infants acquire abstract algebraic rules after two minutes of ....

Plunkett, K., & Marchman, V. (1993). From rote learning to system building. Cognition, 48, 21-69.

....on what structures develop before a concerted attempt at explaining transition becomes feasible. Recent developments in machine learning have made it possible to think in detailed, mechanistic terms about transitions in cognitive development. Both connectionist (Chauvin, 1989; McClelland, 1988; Plunkett Marchman, 1990) and symbolic level (Newell, 1990; Simon, Newell, Klahr, in press) models of cognitive development are now starting to appear. This paper examines connectionist models of human cognitive development with regard to how well they cope with the issue of stages. There is no question about whether ....

....in learning the past tense of English verbs. Developmental psycholinguistic research has suggested that correct performance on irregular verbs is followed by incorrect over regularization of irregular verbs (for example, saying runned for ran) and then by correct performance on the irregulars. Plunkett and Marchman (1990) have constructed back propagation networks that simulate these psychological regularities. Plunkett and Marchman found that they could simulate the inverted U effect by manipulating the type and token frequencies of regular and irregular verb forms. To obtain over regularization errors, the input ....

[Article contains additional citation context not shown here]

Plunkett, K., & Marchman, V. (1990). From rote learning to system building. Technical Report #9020, Center for Research in Language, University of California, San Diego, CA.

....catastrophes are associated with a kind of self organization that can only take place in non linear systems [ p.415, italics added. Furthermore, as neural networks show similar stages in their development as training proceeds (e.g. Rumelhart and McClelland (1986b, chapter 18) and Plunkett and Marchman (1990)) it seems that catastrophe theory can be valuable in explaining the behavior of artificial networks. 3.3.3 Dynamic systems and cognitive growth Van Geert (1991) describes a dynamic systems model that takes cognitive growth to be a system of supportive and competitive interactions between ....

....application of these types of theory that may shed new light on the analysis of network behavior. The explicit notation of mathematical definitions like those in Van Geert or Van der Maas and Molenaar might go some way to help understand what goes on in the stages that for example Elman (1993) and Plunkett and Marchman (1990) have noted in their experiments with networks on the subject of incremental learning. In chapter 5 these experiments will be described in greater detail. 13 It seems, however, that Molenaar is getting research like this underway (Olthof, personal communication) 20 So, in summary, although it ....

[Article contains additional citation context not shown here]

Plunkett, K. and Marchman, V. (1990). From rote learning to system building.

....Rather, all the hidden units move to reduce the current largest error (Fahlman Lebiere 1990) Eventually hidden unit responsibilities are sorted out, but in the intervening time stages can be observed. An example of this occurs in the learning of past tenses (Rumelhart McClelland 1986a; Plunkett Marchman 1990; Marchman 1992) 2. Over generalization. Hidden unit herding is one form of over generalization, but a network without hidden units can also over generalize. An initial period of over generalization could be seen as one stage, with later stages occurring as the network learned the fine ....

Plunkett, K. & Marchman, V. A. (1990). From rote learning to system building. Technical report 9020, Center for Research in Language, University of California, San Diego, La Jolla, CA.

....no generalisation, i.e. if overfitting occurs, learned forms may vary considerably. We cannot say anything about the harmony of these forms, but convergence must result in this case from the reliability of data. In the case of language acquisition, this constraint has often been acknowledged (e.g. (Plunkett Marchman, 1990)) or presented as problematic by some authors, as we will see. 5.2 Learning Inharmonious Forms Some aspects of visual perception are claimed to be indifferent. Stratton s wellknown experiments show that if you see the world upside down through special glasses, then after a week without removing ....

Plunkett, K. & Marchman, V. (1990). From Rote Learning to System Building. San Diego: CRL Technical Report 9020, Univ. of California.

....which is very consistent with the results obtained here. The typical assumption is that both the learning device and training input are static. One might wonder what the consequences are of having either the learning device (network or child) or the input data not be constant during learning. Plunkett and Marchman (1990) have shown that while the basic influences of type token frequency and phonological predictability are similar to the condition of non incremental learning, better overall learning is achieved when the training corpus for a connectionist model is allowed to slowly grow in size. We might also ask ....

Plunkett, K., & Marchman, V. (1990). From rote learning to system building. Center for Research in Language, TR 9020. University of California, San Diego.

....net from the previous generation. The first net was 17 This development is somewhat similar to the shift from rote learning to system building evidenced in the acquisition of vocabulary items in early childhood and, importantly, also observed in a connectionist model of this acquisition process (Plunkett Marchman, 1993). 18 Here it should be noted that as pointed out by Kiparsky, 1976 there appears to be cases where a language has changed towards a more complex morphology (presumably with no significant change in syntax) Such changes often occurs in relatively isolated cultures where the influx of new ....

....order to explain how the current regular irregular system developed (Hare Elman, 1994: p. 34 35) The model furthermore suffers from an ontogenetic version of this problem which has to do with the explanation of how the rule component develops in early language acquisition. On the other hand, Plunkett Marchman (1993) have presented a connectionist account in which a single mechanism explains the qualitative shift from rote learning to system building in the early acquisition of vocabulary items. On this developmental note, I will now turn to the ontogenetic development of language as reflected by the above ....

Plunkett, K. & Marchman, V. (1993) From Rote Learning to System Building.

No context found.

Plunkett, K., & Marchman, V. (1993). From rote learning to system building. Cognition, 49, 21-69.

No context found.

K. Plunkett. From rote learning to system building. acquiring verb morphology in children and connectionist nets. Cognition, 48(1):21--69, 1993.

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