Newport, E. 1990. Maturational Constraints on Language Learning. Cognitive Science 14:11--28.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....in the reading time and comprehension of certain classes of relative clauses. Moreover, it has been hypothesized that language learning in humans is successful precisely because limits on information processing capacities allow children to ignore much of the linguistic data they receive (Newport, 1990). Some computational language learning systems (e.g. Elman (1990) actu ally build a short term memory directly into the architecture of the system. Our baseline case representation does not necessarily make use of this restricted memory bias, however. Each case is described in terms of the ....

E. Newport. 1990. Maturational Constraints on Language Learning. Cognitive Science, 14:11-28.

....simplify the world allowing for the building of simple neural representations that act as building blocks or stepping stones for the later development of more complex neural representations. Such bootstrapping strategies have been hypothesized in children s language acquisition. Newport s[4] Less is More hypothesis states that children s limited memorial and attentional abilities enable children to segment language into its smaller componential parts, and that adults diculty in learning new language stems in part from a poor ability to do this segmentation [5] Elman [6] built a ....

E. L. Newport, \Maturational constraints on language learning," Cognitive Science, vol. 14, pp. 11-28, 1990.

....window of the simple recurrent network was initially restricted and increased in five steps during training. When the network was trained without using either of the two incremental learning techniques it was unable to learn the task. This observation seemed to be in accordance with the work of [29]. However, recently the results and conclusions of Elman [8] were questioned by [42] They reported experiments using similar language tasks where data incremental learning is not necessary. The networks learned the task when trained directly on the most complex data sets. As explanation of these ....

....concerned . 23] They note that the elements of timing and evolution are missing from Elman s models [8] Kirby Hurford [23] further claim that . any innate pre programming for input sensitive growth must have evolved . and they propose a method to incorporate the less is more [29] and starting small [8] ideas into an evolutionary framework to model the evolution of incremental learning. In our own case studies on data incremental learning we trained first order recurrent networks on a onestep look ahead prediction task. The input data consisted of symbol sequences ....

E. Newport. Maturational constraints on language learning. Cognitive Science, 14:11--21, 1990.

....fraction of the set of theoretically possible languages. That language has evolved in close relation to the development of the human language mechanism is a phylogenetic point, but it has ontogenetic plausibility, too. Based on evidence from studies of both first and second language learners, Newport (1990) has proposed a Less is More hypothesis which suggests paradoxically, that the more limited abilities of children may provide an advantage for tasks (like language learning) which involve componential analysis (p. 24) Maturationally imposed limitations in perception and memory forces children ....

.... notion of a crucial period in language learning) This means that because of age differences in perceptual and memorial abilities, young children and adults exposed to similar linguistic environments may nevertheless have very different internal data bases on which to perform linguistic analysis (Newport, 1990: p. 26) In relation to morphology, Newport discusses whether a learner necessarily needs a priori knowledge akin to UG in order to segment language into the right units corresponding to morphemes. She finds that such a segmentation is, indeed, possible, even without advance knowledge of the ....

[Article contains additional citation context not shown here]

Newport, E. (1990). Maturational Constraints on Language Learning. Cognitive Science, 14, p. 11-28.

....this limitation might be a prerequisite for the integration of different sensory modalities. These ideas were backed by experiments with rat pups whose eyes were prematurely opened directly after birth [8] these rats developed an abnormal homing behavior. Similarly, the Less is More hypothesis [13] argues that late learners of language do not reach native speaker competence because they are too advanced in their cognitive abilities and store linguistic units as a whole. By contrast, early learners, due to their limited processing capabilities, break down the units into their constituents ....

Newport, E. L. (1990). Maturational constraints on language learning. Cognitive Science, 14, 11--28.

....di#erences in the reading time and comprehension of certain classes of relative clauses. Moreover, it has been hypothesized that language learning in humans is successful precisely because limits on information processing capacities allow children to ignore much of the linguistic data they receive (Newport, 1990). Some computational language learning systems (e.g. Elman (1990) actually build a short term memory directly into the architecture of the system. It should be clear that the baseline instance representation for the relative pronoun task does not make use of short term memory limitations: the ....

Newport, E. (1990). Maturational constraints on language learning. Cognitive Science, 14, 11--28.

....function, and do model learning cost in these simulations learning is eventually entirely displaced, given a constant environment, as expected. 9 of acquatic species. He suggests, in particular, that languages have evolved to be easily learnable by a procedure which starts small , following Newport (1990) and Elman (1993) with a limited working memory only capable of seeing local grammatical dependencies. Furthermore, Deacon (1997:328f) argues that the surface grammatical organization of languages changes with such speed relative to genetic evolution that there could not have been consistent ....

....a sentence type can be used to determine whether it can function as a trigger at a particular stage in learning, thus filtering random presentation of triggers and ensuring that triggers are presented in (partial) order of decreasing parsability. This corresponds to incorporating an analogue of Newport s (1990) starting small hypothesis into the model (see section 1.3) The algorithm is summarized in Figure 12. A valid category assignment to a trigger (VCA(t i ) is defined as a pairing of a lexical syntactic category with each word in the SF of t i , w 1 : c 1 ; w 2 : c 2 ; w n : c n such ....

Newport, Elissa (1990) `Maturational constraints on language learning', Cognitive Science, vol.14, 11--28.

....Since much of syntax is encoded in grammatical elements realized as affixes, difficulty in analyzing such affixes could seriously hamper language development. What factors internal to the organism might be responsible for the difference between the treatment of triggers in L1 and L2 acquisition Newport (1990) suggests that there may be a neurobiological factor relevant for the critical period which results in bound morphemes being processed by second language learners. We suspect, however, that ultimately the distinction between bound and free morphemes as triggers may be derived from phonology ....

....Since much of syntax is encoded in grammatical elements realized as affixes, difficulty in analyzing such affixes could seriously hamper language development. What factors internal to the organism might be responsible for the difference between the treatment of triggers in L1 and L2 acquisition Newport (1990) suggests, based on her findings on native, early and late ASL first language acquisition that the processing of complex morphology undergoes a major qualitative shift around the age of puberty (and perhaps also a minor shift well before puberty, sometime after the age of four. Thus, there may be ....

Newport, E. (1990) 'Maturational constraints on language learning.' Cognitive Science 14:11-28.

....contrast, when the network was given fully accurate memory and trained on the full complex grammar from the outset, it failed to learn the task. Elman suggested that the limited cognitive resources of the child may, paradoxically, be necessary for effective language acquisition, in accordance with Newport s (1990) less is more proposal. This paper reports on attempts to replicate of some of Elman s findings using similar networks but more sophisticated languages. In contrast with his results, it was found that networks were able to learn quite readily even when confronted with the full complexity of ....

Newport, E. L. (1990). Maturational constraints on language learning.

....The work in language acquisition most closely associated with the advantages of starting small, or otherwise incomplete by comparison with adults, is Elissa Newport s. In a series of studies, she and colleagues have shown the plausible learning advantages of limited resources in children (Newport, 1990, 1988; Goldowsky and Newport, 1993) Whatever it is in the infant that starts small and then grows, its growth facilitating language acquisition, the rate of this growth seems likely to be controlled in large part by genetic factors. And the evolution of the genotype that codes for this growing ....

Newport, Elissa L. 1990. "Maturational constraints on language learning". Cognitive Science, 14: 11-28.

.... appear; Pinker, 1989) 7 (v) Children assume that each meanings and strings, grammatical morphemes in this case, map one to one, an assumption also known as the Uniqueness postulate (Wexler and Culicover, 1980; Clark, 1987; Pinker, 1984) Less is More hypothesis (Newport, 1984; Newport, 1988; Newport, 1990; Carey and Gelman, 1991) Goldowsky and Newport, 1992) cf. Elman, 1993) Specifically, since tense and aspect are confounded, we assume that the one meaning is aspect, demonstrably closer to the verb than tense (see Slobin s introduction to Newport (1988, p. 17) Having set out the framework ....

Newport, Elissa L. 1990. Maturational constraints on language learning. Cognitive Science, 14:11--28.

....complex domains. Surprisingly, limitations can play a crucial role in providing appropriate sequences, meaning that a period of highly limited immature functioning in early life can be necessary to achieving competence in an adult. The importance of limitations has been called the Less is More (Newport, 1990) and the Starting Small (Elman, 1993) principle. The validity of this principle has been demonstrated computationally in the context of language learning (Newport, 1990; Goldowsky and Newport, 1993; Elman, 1993) The trick is that, in some situations, limitations can simplify a domain while ....

....life can be necessary to achieving competence in an adult. The importance of limitations has been called the Less is More (Newport, 1990) and the Starting Small (Elman, 1993) principle. The validity of this principle has been demonstrated computationally in the context of language learning (Newport, 1990; Goldowsky and Newport, 1993; Elman, 1993) The trick is that, in some situations, limitations can simplify a domain while still preserving some useful elements of the full complex domain essentially performing task decomposition. This decomposition can make a complex task tractable because ....

Newport, E. L. (1990). Maturational constraints on language learning. Cognitive Science, 14(1):11-- 28.

....in the reading time and comprehension of certain classes of relative clauses. Moreover, it has been hypothesized that language learning in humans is successful precisely because limits on information processing capacities allow children to ignore much of the linguistic data they receive [Newport, 1990] . Some computational language learning systems (e.g. Elman [1990] actually build a short term memory directly into the architecture of the system. WHirlpool s baseline case representation does not necessarily make use of this restrict memory bias, however. Each case is described in terms of the ....

Newport, E. Maturational Constraints on Language Learning. Cognitive Science, 14:11--28, 1990.

....it is difficult to maintain both recovery of a full LF as the success criterion and the restriction to updating a single parameter per trigger. Some of the consequences of such variants within the parameter setting framework are explored experimentally in section 4. Elman (1993) argues, following Newport (1990), that language learning starts small , restricted by working memory limitations which block the learner from seeing complex triggering data until later in the learning process. It is known that working memory capacity increases through childhood (e.g. Baddeley, 1976, 1992) and that this ....

Newport, E. (1990) `Maturational constraints on language learning', Cognitive Science, vol.14, 11--28.

....in the reading time and comprehension of certain classes of relative clauses. Moreover, it has been hypothesized that language learning in humans is successful precisely because limits on information processing capacities allow children to ignore much of the linguistic data they receive (Newport, 1990). Some computational language learning systems (e.g. Elman (1990) actually build a short term memory directly into the architecture of the system. Our baseline case representation does not necessarily make use of this restricted memory bias, however. Each case is described in terms of the ....

E. Newport. 1990. Maturational Constraints on Language Learning. Cognitive Science, 14:11--28.

....like to consider here is the question of what it is about the so called critical period that might facilitate learning language. Interestingly, much of the work in learnability theory neglects the fact that learning and development co occur. An especially relevant exception, as we shall see, is Newport s (1988, 1990) less is more proposal, 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 ....

....interaction between maturational limitations and language learning Elissa Newport has suggested that indeed, early resource limitations might explain the apparent critical period during which languages can be learned with native like proficiency. Newport calls this the less is more hypothesis (Newport, 1988, 1990). It is well known that late learners of a language (either first or second) exhibit poorer performance, relative to early or native learner. Newport suggests that examination of the performance of early (or native) learners when it is at a comparable level to that of the late learners (i.e. ....

Newport, E.L. (1990). Maturational constraints on language learning. Cognitive Science, 14:1128.

....presented in Elman (1991b, 1993) and in Goldowsky Newport (1993) suggesting that the initial memory limitations of children may, at least in part, help with the bootstrapping of linguistic structure. Data from first and second language learning further supports this less is more hypothesis (Newport, 1990 for further discussion of maturational constraints on learning, see chapter 5, section 5.3) 4.2 Results In order to provide an independent basis for the assessment of general network performance on the two grammars, the simple statistical prediction method developed in connection with the ....

....134 135) In this section, I will present an processing based account of language learning in which a reappraisal of the poverty of the stimulus argument will pave the way for a solution to the language learning paradox. Based on evidence from studies of both first and second language learners, Newport (1990) has proposed a Less is More hypothesis which suggests paradoxically, that the more limited abilities of children may provide an advantage for tasks (like language learning) which involve componential analysis (p. 24) Maturationally imposed limitations in perception and memory forces children ....

[Article contains additional citation context not shown here]

Newport, E. (1990) Maturational Constraints on Language Learning. Cognitive Science, 14, 11--28.

....By contrast, when the network was given fully accurate memory and trained on the complex grammar from the outset, it failed to learn the task. Elman suggested that the limited cognitive resources of the child may, paradoxically, be necessary for effective language acquisition, in accordance with Newport s (1990) less is more proposal. This paper reports on attempts to replicate of some of Elman s findings using similar networks but more sophisticated languages. In contrast with his results, it was found that networks were able to learn quite readily even when confronted with the full complexity of the ....

Newport, E. L. (1990). Maturational constraints on language learning.

....involving function word manipulations show that limitations on represen i tation, of the type we might expect for young children, does not interfere with obtain ng an approximately correct linguistic categorization, and in some cases can actually e p improve upon it. This could be an example of Newport s (1988, 1990) Less is Mor henomenon, where constraints on young learners information processing abilities may lead to more successful learning. We would like to re emphasize that we are not proposing that our analytic proa cedures are a model of form class acquisition. Rather, we have used these analyses to ....

Newport, E. L. (1990). Maturational constraints on language learning. Cognitive Science, 14, 11-28.

No context found.

Newport, E. 1990. Maturational Constraints on Language Learning. Cognitive Science 14:11--28.

No context found.

Newport, E.L. (1990). Maturational Constraints on Language Learning. Cognitive Science, 14, p11-28.

No context found.

Newport, E. (1990). Maturational constraints on language learning. Cognitive Science, 14 (pp. 11-28).

No context found.

CA. E. Newport. 1990. Maturational constraints on language learning. Cognitive Science, 14:11--28.

No context found.

Newport, E. (1990). Maturational constraints on language learning. Cognitive Science, 14 (pp. 11-28).

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