Hare, M., and Elman, J.L., "Learning and morphological change," Cognition, vol.56, pp.61-98, 1995.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....complex models just to account for that phenomenon. For example the irregularity in the acquisition of English past tense forms 4 was once considered as evidence for rule based mechanisms in the mind [52] but experiments have shown that such a curve can be obtained with simple neural models [53, 31]. The third approach is characterized by one of my favourite definitions of artificial intelligence. It does not impose restrictions on the methodology, and sets a practical goal. Implicit in this definition is Marr s dissociation of theory from representation and implementation [46] If we ....

Hare, M., and Elman, J.L., "Learning and morphological change," Cognition, vol.56, pp.61-98, 1995.

....modularity. Given the discussion of Dunn Kirsner [6] this should not be too much of a shock, but it does complicate our modelling endeavors. We are left with the question: Can this happen in real life, and if so, what does it mean In reality, word frequencies are not just random. Hare Elman [7] have shown how language evolution naturally results in a correlation between irregularity and frequency and so a balancing of the effects of frequency and regularity does really happen. It seems that real language does have an inbuilt confound. If we are not careful, we will be equally justified ....

Hare, M. & Elman, J.L. (1995). Learning and Morphological Change. Cognition, 56, 61-98.

.... word forms (e.g. exhibit phonological exchange errors) Ultimately, we believe that connectionist models of the acquisition of the skills of speaking (and comprehending) will contribute to explanations of the nature of language why it is the way it is (see, e.g. Christiansen Devlin, 1997; Hare Elman, 1995; Gupta Dell, 1999) Moreover, we believe the PDP approach offers the best chance to explain 539 LANGUAGE PRODUCTION production as a skill, as something that one learns to do over years of experience. Perhaps most importantly, a PDP approach to language production expresses its commonalties ....

Hare, M., & Elman, J. L. (1995). Learning and morphological change. Cognition, 56, 61--98.

....themselves are arti#cial, and it will be crucial to test the network on less regular stimuli from real languages. While this has not been attempted, the large literature on the learning of regular and irregular morphology by neural networks #Cottrell Plunkett, 1991; Daugherty Seidenberg, 1992; Hare Elman, 1995; MacWhinney Leinbach, 1991; Plunkett Marchman, 1991; Rumelhart McClelland, 1986# is relevant. The indication is that networks somewhat similar to the one discussed in this paper are not thrown o# by the combination of regular and irregular morphology. The treatment of #semantics in the ....

Hare, M. & Elman, J. L. #1995#. Learning and morphological change. Cognition, 56, 61#98.

....of current research The field of research into language evolution by means of computational simulations as outlined here is still quite young, thus there is only a small number of research results available so far, of which I will briefly present two. 2.6. 1 Learning and morphological change In [Hare Elman 95] two experiments are described which make use of some of the basic ideas set out above, though the authors obviously had more specific goals. The aim of their work is to simulate the historical change of the morphology of the verb system of Old English into that of the language as known today, ....

....claimed to be necessary for language to evolve. 2. 7 Conclusion For this thesis I have concentrated on simulations of language evolution from no initial language by replicating work originally reported in [Batali 98] This experiment, which I will describe in detail in chapter 3, is similar to [Hare Elman 95] in modelling continued transmission of language using connectionist methods. It is, however, fundamentally different in directly implementing the generation of surface forms as sequences of characters. The initial agent is not given any knowledge and the simulation does not rely on frequency ....

[Article contains additional citation context not shown here]

Mary Hare and Jeffrey L. Elman. Learning and morphological change. Cognition, 56:61-- 98, May 1995.

.... of grammatical structures, for example from an introductory linguistics text (Lawrence et al. 1998) Furthermore, they re ect human performance on a number of language tasks (Weckerly and Elman, 1992; Christiansen and Chater, 1998) and can account for historical descriptions of language change (Hare and Elman, 1995). As well as processing constraints, RNNs have learning constraints. That is, they are constrained not only in what they can represent, but in what they can learn. Returning to our original question, we consider how the emergence of a language between communicating RNNs will be a ected by their ....

Hare, M. and Elman, J. (1995). Learning and morphological change. Cognition, 56:61-98.

....years by the proposal that rather than language savvy infants , languages themselves adapt to human learners, and the ones that survive are infant friendly languages [2 4] To date, relatively few empirical studies have explored how such adaptation of language works in practice. Hare and Elman [5] demonstrated that classes of past tense forms could evolve over simulated generations in response to changes in the frequency of verbs, using neural networks. Kirby [6] showed, using a symbolic system, how compositional languages are more likely to emerge when learning is constrained to a ....

M. Hare and J. Elman. Learning and morphological change. Cognition, 56:61--98, 1995.

....generally) have historically played important roles in several core areas of linguistic research, including phonological theory (e.g. Halle Mohanan 1985) language acquisition (e.g. Berko 1958; Bybee Slobin 1982) adult psycholinguistics (e. g, Bybee Moder 1983) diachronic change (e.g. Hare Elman 1995), and the genetic (Rice 1996) and neurobiological (Ullman et al. 1997) bases of language. These studies illustrate the value of exploring a specific aspect of language in considerable detail. However, it is also true that if the connectionist approach were proved to be incapable of explaining ....

Hare, Mary, & Jeffrey L. Elman. 1995. Learning and morphological change. Cognition 56.6198.

....change in relatively mature and complex languages, although the methodology of these simulations, extended and refined, would be very suitable for models in historical linguistics. Examples of such applications to historical language change, from quite contrasting theoretical backgrounds, are Hare and Elman (1995) and Niyogi and Berwick (1997) No biological evolution: In these models, there are no differences between individuals at the point when they are introduced into the population. They all have identical capacities for responding to their environment, either in the production of utterances or in ....

Hare, M., and Elman, Jeff (1995) "Learning and morphological change", Cognition, 56:61-98.

....German (see Clahsen, Rothweiler, Woest Marcus, 1993; Marcus, Brinkmann, Clahsen, Wiese, Woest Pinker, 1995) If true, such lack of cross linguistic validity would pose serious problems for connectionist models of morphology. However, recent connectionist work has addressed minority defaults. Hare, Elman and Daugherty (1995) trained a multi layer feedforward network (with additional cleanup units see Plaut, this issue, for an explanation) to map between phonological representations of stems and past tenses for a set of verbs representative of very early Old English. The training set consisted of five classes of ....

Hare, M. & Elman, J.L. (1995). Learning and morphological change. Cognition, 56, 61--98.

.... of their major appeal is the ability to incorporate syntax and semantics into a single encompassing model [3] They have also demonstrated competence in learning a wide range of grammatical structures [7] and often reflect real world data on natural language tasks [2, 10] and language change [5]. Although these results have been affected by it, the issue of constraints of RNNs has not been explicitly examined. It seems important then, to investigate the constraints of recurrent networks and the way that they influence the properties and emergence of language. This paper is motivated by ....

M. Hare and J. Elman. Learning and morphological change. Cognition, 56:61--98, 1995.

....appearance of generally stronger restrictions on relative clause markers and the disappearance of inflections first in nouns and, subsequently, in verbs 18 . The proposed effects of human learning and processing constraints on linguistic change is further substantiated by recent simulations by Hare Elman (1994), investigating the changes in English verb inflection over the last 1,100 years. The morphological system of Old English (ca. 870) was quite complex involving at least 10 different classes of verb inflection (with a minimum of six of these being strong ) The simulations involved several ....

.... results show that verbs which either belong to small classes, lack consistent defining characteristics, or are low in frequency should change most rapidly; change in other verbs will depend on the precise extent to which they possess the characteristics which make them resistant to assimilation (Hare Elman, 1994: p. 31) The change taking place in the verb inflection system being transferred between generations closely resembles, in considerable detail, the historical change in English verb inflection leading from the highly complex past tense system of Old English to the modern English incorporating one ....

[Article contains additional citation context not shown here]

Hare, M. & Elman, J.L. (1994) Learning and Morphological Change. Unpublished Ms. Centre for Research in Language, University of California, San Diego.

....are artificial, and it will be crucial to test the network on less regular stimuli from real languages. While this has not been attempted, the large literature on the learning of regular and irregular morphology by neural networks (Cottrell Plunkett, 1991; Daugherty Seidenberg, 1992; Hare Elman, 1995; MacWhinney Leinbach, 1991; Plunkett Marchman, 1991; Rumelhart McClelland, 1986) is relevant. The indication is that networks somewhat similar to the one discussed in this paper are not thrown off by the combination of regular and irregular morphology. The treatment of semantics in the ....

Hare, M. & Elman, J. L. (1995). Learning and morphological change. Cognition, 56, 61--98.

....in the cases we analyze the restructuring of noun classes can be accounted for on the basis of the morphophonemic similarities of items and their type token frequency. This is exactly the type of information that connectionist simulations are known to be sensitive to (Plunkett Marchman 1991; Hare Elman 1995). The case we will examine is the restructuring of the gender system in Romance. In particular, we will show how a sophisticated connectionist network was applied to model the change from the three gender system found in Latin to the two gender system of Old French. The importance of this project ....

....word expressing the same similar concept in Gaulish, but the evidence for this could be drawn only on the basis of possibilities (a) and (b) 2.5. From language history to the modeling design The connectionist approach to the restructuring of Romance gender was inspired largely by the work by Hare and Elman (1995), who used sophisticated connectionist simulations to model changes in the past tense system of Old English. Among other historical developments, Hare Elman were able to model the assimilation of weak verb class Ic (nerjan type verbs) into weak verb class II (lufian type verbs) This development ....

[Article contains additional citation context not shown here]

Hare, M. and Elman, J. 1995. Learning and morphological change. Cognition 56: 61-98.

No context found.

Hare, M., & Elman, J. L. (1995). Learning and Morphological Change, Cognition, 56(1).

No context found.

M. Hare and J. Elman. Learning and morphological change. Cognition, 56(1):61-- 98, 1995.

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