Volume 47, Issue 2

The article reports on a series of studies designed to test for similarities and differences between the implicit and explicit learning of categories and rules in children and adults. First, the distinction between implicit, nonconscious and automatic learning processes on the one hand, and explicit hypotheses testing and problem solving on the other hand is briefly introduced. While explicit learning processes have been shown to develop rather late, preschool children appear as capable as adults when implicitly inducing the formal morphophonological and syntactic regularities of complexly structured input. This is empirically shown by using miniature languages within an implicit learning context. Over and above these studies show that not only explicit but also implicit learning is highly dependent on input structure. Specifically, children and adults succeeded in learning more complex morphophonological regularities and formal word categories if and only if the input contained additional cues to structure that covaried with word categories and rules. In addition, this result is extended to the learning of visual-conceptual categories and rules. When explicit and implicit modes of learning were compared, implicit, but not explicit learning turned out to be highly “modality specific”, i.e., correlated phonological cues (but not visual-conceptual ones) facilitated the implicit learning of morphophonological rules and categories, while visual-conceptual cues promoted the learning of visual-conceptual categories and rules. Contrary to implicit learning, explicit learning is comparatively more focused on visual-conceptual cues and regularities. The results of the miniature language studies are discussed and related to natural bidirectional interactions between cognitive-conceptual development on the one hand, and lexical and formal-language development on the other.

All proponents of generative approaches to language learning argue that the syntactic knowledge which language learners acquire is underdetermined by the input. Therefore, they assume an innate language acquisition device which constrains the hypothesis space of children when they acquire their native language. However, it is still a matter of debate how general or domain-specific this acquisition mechanism is and whether it is fully available from the onset of language acquisition. This article provides an overview of the different answers that have been provided for these questions within generative linguistics. Moreover, it shows how the generative concept of “learning” has been applied to the acquisition of syntax, morphology, phonology and vocabulary, language processing, L2-acquisition, nontypical language development, creoles and language change. Finally, current developments, merits and problems of the generative approach to learning are discussed. The focus of this discussion will be on efforts to reduce assumptions about domain-specific innate predispositions for language learning.

In optimality theory (OT) the essence of both language learning in general (learnability) and language acquisition (the actual development children go through) entails the ranking of constraints from an initial state of the grammar to the language-specific ranking of the target grammar. This is the common denominator in all OT studies on language acquisition and learning. There are many unsettled issues, however. Are the constraints innate or do they emerge during acquisition (nature-nurture)? And if they emerge, where do they come from? What is the initial state? Does the (re)ranking of constraints only involve the demotion of markedness constraints, the promotion of faithfulness constraints, or can it be achieved by both the demotion and the promotion of constraints? Another issue is whether comprehension and production are mediated by the same grammar or whether there is one grammar for comprehension and another for production. This article reviews the current state of affairs in language acquisition studies in OT and ends with some critical remarks and speculations on how the field is likely to develop.

In the field of language acquisition the term bootstrapping stands for the assumption that the child is genetically equipped with a specific program to get the process of language acquisition started. Originally set within the principles and parameters framework bootstrapping mechanism are considered as a linkage between properties of the specific language the child is exposed to and pre-existing linguistic knowledge provided by universal grammar . In a different view — primarily developed within the so-called prosodic bootstrapping account — bootstrapping mechanism direct the child's processing of the input thereby constraining the child's learning in a linguistically relevant way. Thus, the attendance to specific input cues provides the child with information to segment the input in linguistically relevant units which constitute restricted domains for more general learning mechanism like e.g., distributional learning. The paper will present a review of empirical findings that show that children are in fact equipped with highly sensitive and efficient mechanism to process their speech input which initially seem to be biased to prosodic information. It will be argued that further research within this framework has to deal with the reliability of the proposed relevant input cues despite crosslinguistic variation and with children's ability to overcome an initial reliance on single cues in favor of an integration of different sources of information.

It was long considered to be impossible to learn grammar based on linguistic experience alone. In the past decade, however, advances in usage-based linguistic theory, computational linguistics, and developmental psychology changed the view on this matter. So-called usage-based and emergentist approaches to language acquisition state that language can be learned from language use itself, by means of social skills like joint attention, and by means of powerful generalization mechanisms. This paper first summarizes the assumptions regarding the nature of linguistic representations and processing. Usage-based theories are nonmodular and nonreductionist, i.e., they emphasize the form-function relationships, and deal with all of language, not just selected levels of representations. Furthermore, storage and processing is considered to be analytic as well as holistic, such that there is a continuum between children's unanalyzed chunks and abstract units found in adult language. In the second part, the empirical evidence is reviewed. Children's linguistic competence is shown to be limited initially, and it is demonstrated how children can generalize knowledge based on direct and indirect positive evidence. It is argued that with these general learning mechanisms, the usage-based paradigm can be extended to multilingual language situations and to language acquisition under special circumstances.

In the past twenty years the connectionist approach to language development and learning has emerged as an alternative to traditional linguistic theories. This article introduces the connectionist paradigm by describing basic operating principles of neural network models as well as different network architectures. The application of neural network models to explanations for linguistic problems is illustrated by reviewing a number of models for different aspects of language development, from speech sound acquisition to the development of syntax. Two main benefits of the connectionist approach are highlighted: implemented models offer a high degree of specificity for a particular theory, and the explicit integration of a learning process into theory building allows for detailed investigation of the effect of the linguistic environment on a child. Issues regarding learnability or the need to assume innate and domain specific knowledge thus become an empirical question that can be answered by evaluating a model's performance.

Children learn how to learn language, and they get better as they go along. This article presents an overview of research inspired by a dynamic systems view of language learning that shows it to be a self-organizing process in which children create the units they need from the regularities present in the environment in which they are situated.

This article outlines a nonlinear dynamic systems approach to language learning on the basis of developmental cognitive neuroscience. Language learning, on this view, is a process of experience-dependent shaping and selection of broadly defined domain-general and domain-specific genetic predispositions. The central concept of development is (neuro)cognitive growth in terms of self-organization. Linguistic structure-building is synergetic and emergent insofar as the acquisition of a critical mass of elements on a local level (e.g., words) results in the emergence of novel qualities and units on a macroscopic level (e.g., syntax). We argue that language development does not take a linear path but comes in phases of intermittent turbulence, fluctuation, and stability, along a “chaotic itinerary”. We review qualitative, quantitative and computational applications of this concept in the lexical, morphological, and syntactic domain. We identify as the most significant property of the dynamic approach the temporal nature of language learning. As a medium-term forecast we anticipate a further diversification of the dynamic approach, an increase in more formal approaches, and a stronger interest in issues of embodiment and embeddedness.