Adult language learners under cognitive load do not over-regularize like children
Adult language learners under cognitive load do not over-regularize like children
复制标题
认知负荷下的成人语言学习者不会像儿童那样过度规范化
DOI：
--
复制
发表时间：
2010
期刊：
影响因子：
--
通讯作者：
N. Burns
中科院分区：
文献类型：
--
作者：
A. Perfors;N. Burns
Adult language learners under cognitive load do not over-regularize like children Amy Perfors (amy.perfors@adelaide.edu.au) School of Psychology, University of Adelaide Nicholas Burns (nicholas.burns@adelaide.edu.au) School of Psychology, University of Adelaide Abstract The “less is more” hypothesis suggests that one reason adults and children differ in their language acquisition abilities is that they also differ in other cognitive capacities: for instance, the relatively poor memory and/or processing abilities of children may make them more likely to over-regularize inconsistent in- put (Singleton & Newport, 2004; Hudson Kam & Newport, 2005). We investigate this hypothesis by placing adults under a high cognitive load using a standard task. Does their ten- dency to over-regularize in a simultaneous language-learning task increase? Results indicate that although the cognitive load is high enough to impair overall learning, neither the pres- ence of load nor poor working memory predicts greater over- regularization. This suggests that if the “less is more” hypoth- esis explains over-regularization in children, the relevant cog- nitive capacity is not one that was impaired by our load task. Keywords: language acquisition; over-regularization; statisti- cal learning; memory; processing; development Introduction Children and adults differ both qualitatively and quantita- tively in their ability to acquire a new language. Adults have difficulty with many aspects of language acquisition, from phonetic perception (Werker & Tees, 1984; Werker & Lalonde, 1988; Kuhl, 2004), to language processing (Clahsen & Felser, 2006), to certain aspects of syntax (e.g., Johnson & Newport, 1989; Birdsong, 2006). Scientists have proposed many theories to account for the difference between children and adults; these theories differ in both the degree and type of contribution made by pre-existing language-specific biases. Although nearly everyone agrees that (due to the inherent log- ical problem of induction posed by language learning) some bias must be necessary to explain successful language acqui- sition, explanations about the nature of the bias – and the dif- ference between children and adults – vary considerably. Some argue that there is a fundamental difference between first- and second-language acquisition. They posit that acqui- sition in children is guided by an innate Universal Grammar and by language-specific acquisition procedures, whereas adult acquisition is directed by more domain-general learning mechanisms (e.g., Bley-Vroman, 1990). However, there are many other possibilities, since children and adults also dif- fer profoundly in their cognitive capabilities, knowledge, as- sumptions, and typical linguistic input. For one thing, learn- ing a second language is made more difficult by interference from the first language; the evidence that experience with a first language influences acquisition of a second language is extensive (e.g., Mayberry, 1993; Iverson et al., 2003; Tan, 2003; Weber & Cutler, 2003; Hernandez, Li, & MacWhinney, 2005). This observation overlaps considerably with the re- lated point that adult brains are less malleable than the brains of children (Elman et al., 1996; MacWhinney, 2005). Adults and children also differ in their style of learning (Ullman, 2004) and in the nature of the social support (Snow, 1999) and linguistic input (Fernald & Simon, 1984) they receive. The observation that children perform more poorly than adults across most domains of cognitive ability, including memory and processing speed, has led to another hypothe- sis, often called “less is more.” It suggests that the relative cognitive deficits in children may actually help with language acquisition by enabling them to isolate and analyze the sepa- rate components of a linguistic stimulus (Newport, 1988), or by leading them to over-regularize inconsistent input (Hudson Kam & Newport, 2005; Singleton & Newport, 2004). Indeed, it is apparent that children over-regularize while adults often do not. Deaf children exposed to the inconsistent sign lan- guage of hearing parents will over-regularize that language and produce regular grammatical forms (Singleton & New- port, 2004), as will children exposed to inconsistent input in an artificial language (Hudson Kam & Newport, 2005; Goldowsky, 1995). By contrast, adult language learners are known to produce highly variable, inconsistent utterances, even after years of experience with the language and after their grammars have stabilized (Wolfram, 1985; Johnson, Shenkman, Newport, & Medin, 1996). The difference between children and adults has also been found in non-linguistic domains. If adults must predict some phenomenon (e.g., a light flashing or a certain card being drawn from a deck), they will tend to probability match: if the phenomenon occurs 70% of the time, they will expect it 70% of the time they are asked (see, e.g., Myers, 1976; Shanks, Tunney, & McCarthy, 2002, for an overview). Children are more likely to predict that the phenomenon will occur closer to 100% of the time (e.g., Weir, 1964; Derks & Paclisanu, 1967). A similar pattern has been found in causal reasoning: children over-regularize by assuming that causes are deter- ministic, while adults do not (Schulz & Sommerville, 2006). Although the tendency toward over-regularization is well- established, the reason for the difference between adults and children is far from clear. As previously mentioned, the “less is more” hypothesis suggests that over-regularization may be due to some aspect of children’s cognitive capacities, such as their poorer memory or slower processing speed (Newport, 1988). Adults do tend to over-regularize more when the in- put is complex, when the probabilities involved are small (Gardner, 1957; Weir, 1964; Gluck & Bower, 1988; Hud- son Kam & Newport, 2009), or when lexical retrieval is more difficult (Hudson Kam & Chang, 2009). This may be because
