- How children learn the meanings of words by Paul Bloom
How children learn the meaning of words is an exceptionally good book, one likely to be a foundational work on word meaning acquisition for some years to come. It is also an easy one for this reviewer to praise because it agrees with and strengthens the main thesis about word meaning (and many other psychological systems) that I and Gedeon Deak put forth in a 1995 paper which Bloom cites in his conclusions.
Basically, we proposed a metaphorical translation of Isaiah Berlin’s distinction between foxes and hedgehogs. Berlin classified intellectual theories as being either hedgehogs or foxes. Hedgehogs do one big thing well (quills), and hedgehog figures like Plato have one big idea (eternal ideas). Foxes succeed by doing many different things well, and fox-like intellectual systems, such as Aristotle’s or Machiavelli’s, do the same. We suggested that psychological systems can also be classified into hedgehog systems, which work off relatively few big principles and can be algorithmized, and fox systems, which are hard to describe explicitly and involve a great variety of processes. In language, Chomskyan core grammar is clearly a hedgehog. By contrast, people’s pragmatic use of language to achieve social goals stands squarely in the fox domain, consisting of a great variety of abilities and processes.
Our argument, with which B agrees, was that most theories of word meaning acquisition have characterized it as a hedgehog process, utilizing either ‘constraints’ (Markman 1992) or ‘dumb attentional mechanisms’ (Smith et al. 1996) (i.e. a very limited set of mechanisms or principles is supposed to cover everything). The fact that word meaning seems to be acquired very quickly and relatively accurately, even when the input is potentially ambiguous, provides the same basis for this hedgehog view that is supplied in Chomskyan argument about language acquisition. We thought that the acquisition of basic object nouns like chair and car might have some hedgehog structure, for reasons given below. But all the other kinds of words—nonbasic nouns, verbs, adjectives—would lack such a privileged conceptual level, and so no such small set of fastacting processes could be used to acquire them in a uniform way.
B makes what could be seen as a friendly amendment to our conclusion, claiming fox processes for all the basic object nouns, too. But, he solves the problem of quickness and accuracy somewhat differently than we did. Basically, B agrees with much current work in accepting that children do a good deal of ‘fast mapping’. That is, they figure out a good deal of the meaning of new words quickly. While he provides a few exceptions to this generalization, for the most part he stresses children’s ability to figure out a good deal accurately and quickly about the meaning of a new word. So how do they do this without invoking highly general ‘constraints’ on word learning?
First, there is children’s skill in using ‘theory of mind’ competence. Here B does not just mean the knowledge that others also have their own mind (i.e. perceptions and thoughts). This could be a ‘big idea’, a hedgehog idea, if this is all there was to children’s capacities. What engages B, quite rightly, is the accumulating evidence (reviewed extensively in Ch. 2) of children’s varied and flexible skills in actually reading other people’s intentions across a wide variety of situations. These ‘mind-reading’ skills do much of the work of various general ‘constraints’ and simple mechanisms that have been proposed (which would, in fact, only cover a small portion of actual vocabulary). Furthermore, evidence indicates children read other people’s nonlinguistic intentions in similar ways. This is not some special social insight or relevance reader for word meaning.
Another major move towards a fox’s world comes in B’s discussion of figuring out words from linguistic context alone, as in reading, or in situations where the linguistic context of the new word is otherwise central. B discusses at length the current...