Pabiku and Ga Ti Ga : Two Mechanisms Infants Use to Learn About the World

Some of the things that adults learn about language, and about the world, are very specific, whereas others are more abstract or rulelike. This article reviews evidence showing that infants, too, can very rapidly acquire both specific and abstract information, and considers the mechanisms that infants might use in doing so. Some of the things we know about language are very specific. We might know, for instance, that the word inextricably is often followed by the word bound but rarely by the word attached, that the verb raced is more often used as a past-tense form (the Preakness winner raced past the other horses) than as a past participle (the horse that was raced past the other horses suddenly fell down), or that the letters st are sometimes followed by r but never by x, and so forth. Other aspects of what we know about language are more abstract. We know, for example , that we can form a sentence by combining any plural noun (say, cats) with any verb phrase with plural agreement (say, chase mice, yielding the sentence cats chase mice), and that we can form the progressive of any verb stem by adding –ing to it (e.g., fax-faxing). We see the same contrast in our knowledge of the world. For example , we know the specific fact that horses that mate with donkeys give birth to mules, but also the more general fact that when an animal of a certain species has babies, the babies will ordinarily be of the same species. If you had never before heard of a gerenuk (a giraffe-necked gazelle that lives in Africa), you would automatically assume (correctly) that gerenuks give birth to baby gerenuks. (This is no tau-tology—one can easily imagine a world in which cats give birth to dogs, gerenuks give birth to wilde-beests, and fennec foxes give birth to dandelions.) Both of these kinds of knowl-edge—facts about particular items and abstractions that hold more broadly—are in some sense statistically reliable reflections of the world, but the abilities to learn and represent these kinds of knowledge appear to rely on different mechanisms, one of which pertains to relations between specific elements , the other of which pertains to what one might call " algebraic rules " —relationships between variables, or open-ended schemas in which we can freely substitute any instance. Human infants appear to be able to represent and acquire …