How children learn to communicate discriminatively

Abstract

How do children learn to communicate, and what do they learn? Traditionally, most theories have taken an associative, compositional approach to these questions, supposing children acquire an inventory of form-meaning associations, and procedures for composing / decomposing them; into / from messages in production and comprehension. This paper presents an alternative account of human communication and its acquisition based on the systematic, discriminative approach embodied in psychological and computational models of learning, and formally described by communication theory. It describes how discriminative learning theory offers an alternative perspective on the way that systems of semantic cues are conditioned onto communicative codes, while information theory provides a very different view of the nature of the codes themselves. It shows how the distributional properties of languages satisfy the communicative requirements described in information theory, enabling language learners to align their expectations despite the vastly different levels of experience among language users, and to master communication systems far more abstract than linguistic intuitions traditionally assume. Topics reviewed include morphological development, the acquisition of verb argument structures, and the functions of linguistic systems that have proven to be stumbling blocks for compositional theories: grammatical gender and personal names.

Book Language and Its Implications for Children’s Language, Literacy, and Development

“…Although language is complex, it contains structure via a multitude of probabilistic relationships between multiple levels of correlated features (e.g., Seidenberg & MacDonald, 2018). Several theoretical accounts treat language development as the product of general learning mechanisms operating across this rich input (e.g., Ramscar, 2021). In this view, the transition to adultlike language comprehension and production is a gradual one, reflecting the accumulation of experience with sounds, words, sentences, and their contexts over a number of years.…”

Understanding the Phonetic Characteristics of Speech Under Uncertainty—Implications of the Representation of Linguistic Knowledge in Learning and Processing

“…Theoretically, the network simulations reported in our study were rooted in discriminative learning (Ramscar and Yarlett, 2007 ; Ramscar et al, 2011 , 2013a , b ; Ramscar, 2019 , 2021b ). This framework conceptualizes learning—during perception and production—as a process that serves to discriminate informative relationships between a set of cues and a set of outcomes in a cognitive system.…”

“…On one hand, it limits the ability of these models to discover abstract structures—such as inflectional functions—that may be present in a set of training data. On the other hand, simply because of their simplicity, they constrain modelers to utilizing input and output structures that explicitly code for the cues and outcomes that they believe to be important to the process being modeled (see Ramscar, 2021b , for a more detailed discussion of this point).…”

“…That is, in order for two speakers to have a conversation, they must share the same “source code” (Ramscar, 2019 , 2021b ) that underlies the language they are using. A listener uses what they have learned about the shared code to predict the messages intended by speakers.…”

“…Accordingly, we shall begin by discussing how previous computational models of speech production have addressed these issues, and how they were used to make predictions about the phonetic characteristics of speech. Then, since both Tucker et al ( 2019 ) and Tomaschek et al ( 2019 ) are rooted in the theory of discriminative learning , a cognitive theory of how language (and actually any kind of behavior) is learned (Ramscar and Yarlett, 2007 ; Ramscar et al, 2010 , 2013b ; Ramscar, 2019 , 2021b ), we shall examine the constraints that this theory imposes on the way the input-output structure of models is configured.…”