LINGUIST List 11.1511

Tue Jul 11 2000

Review: Bloom: How Children Learn the Meaning of Words

Editor for this issue: Andrew Carnie <>

What follows is another discussion note contributed to our Book Discussion Forum. We expect these discussions to be informal and interactive; and the author of the book discussed is cordially invited to join in. If you are interested in leading a book discussion, look for books announced on LINGUIST as "available for discussion." (This means that the publisher has sent us a review copy.) Then contact Andrew Carnie at


  1. Bart Geurts, Review Bloom 2000

Message 1: Review Bloom 2000

Date: Mon, 10 Jul 2000 12:43:13 +0200
From: Bart Geurts <>
Subject: Review Bloom 2000

Paul Bloom: How Children Learn the Meanings of Words. MIT Press,
Cambridge, Mass., 2000. Pp. 300 + xii. $39.95 (cloth).

Reviewed by:

Bart Geurts, University of Nijmegen & Humboldt University, Berlin

The title of this book is slightly misleading: "How Children Learn the
Meanings of Nominal Expressions Denoting Middle-Sized Observables"
would have been more accurate. But it would have been less catchy, I
suppose, and it wouldn't have been quite correct, either, since Bloom
does not confine his discussion to nominals, although that is where
the main emphasis lies.

Bloom's theoretical stance may be characterized as eclectic, as long
as this is taken in an entirely positive sense. If his essay has a
central message, it is that there is no "module" for acquiring words.
Bloom's view is that lexical learning involves several aspects of
cognition, none of which is custom-built for this purpose, ranging
from the ability to understand other people's doings in intentional
terms, on the one hand, to a proclivity for seeing the world in terms
of objects and events, on the other.

One of the main prerequisites for lexical learning, according to
Bloom, is what has come to be known as a "theory of mind". It is a
familiar observation that adults view each other as "intentional
systems", to use Dennett's (1987) term, that is to say as rational
beings who plan their actions so as to achieve certain goals. We
normally adopt this intentional stance when we want to explain or
predict each other's behaviour, and it is commonly assumed that this
holds especially for linguistic behaviour. One wouldn't perhaps expect
the intentional stance to be involved in such a lowly task as learning
words, but Bloom maintains that it is, and he has some telling
experimental results to back up his position. For example, in an
experiment by Baldwin (1993), babies were given an object to play with
while the experimenter examined the contents of a bucket in front of
him and labeled it with a new word, such as "It's a modi."
Interestingly, subsequent testing revealed that 18-month-olds did not
link this word to the object they were currently focused on, but
instead followed the experimenter's gaze so as to determine the
intended referent. Such results not only are an embarrassment to naive
associationist theories of word learning, as Bloom points out, but
they also suggest that already at a very tender age children are
sensitive to other people's intentions.

The little language learner's social cognition is employed in other
ways, as well, for example in the acquisition of personal pronouns. In
particular, indexicals like "I" and "you" are problematic because the
child is always addressed with the latter and never with the former;
so how does it figure out that "I" refers to the speaker and "you" to
the hearer? Following Oshima-Takane (1999), Bloom proposes that
children learn the meanings of personal pronouns by attending to other
people's conversations, and observing how referential terms alternate
as speakers take turns. An account along these lines is supported by
the fact that autistic children, whose social cognition tends to be
less developed, often have problems with personal pronouns. This may
seem like an obvious idea, but as Bloom notes it is in fact "a radical
proposal, as it flies in the face of the assumption that, at least for
children in Western societies, word meanings are learned from
child-directed speech." (p. 125) Furthermore, if this is how pronouns
are learned, it confirms Bloom's general point that children's
learning strategies are varied and, in many respects, quite

Although Bloom stresses the relevance of the theory of mind, he also
makes it clear that it is not the only factor in word learning.
Another factor of equal importance is the child's predisposition to
parse the world into discrete objects and events. The force of this
predisposition was demonstrated by an experiment in which children
were shown two red apples and three yellow bananas, and asked to count
the different colours or kinds of fruit. Preschoolers, who are focused
on objects, tended to answer "five" (Shipley and Shepperson 1990).
Other experiments suggest that even babies are aware that unobservable
properties of an object may be more important than its superficial
characteristics (another blow for naive associationism), and Bloom
argues that this "essentialism", as he calls it, holds not only for
natural-kind categories but for artifacts, too.

In general, Bloom tends to play down the relevance of syntactic clues
to the learning of words. His position is that syntax is only one
source of information, whose relative importance varies from case to
case. Some words may be learned almost entirely without syntactic
support, whereas others are difficult to learn in the absence of
syntactic information. The former class contains concrete content
words ("horse", "red", "eat"); the latter, abstract words and
numerals, for example.

Bloom devotes an entire chapter to number words, arguing that in this
particular case syntactic information is needed for a child to grasp
the notion of number term, and that, furthermore, the counting system
provided by the language (if any: some languages don't support
counting) is essential to the child's understanding of numbers. This
account raises the spectre of a Whorfian stance on the relationship
between language and thought, but Bloom argues at some length that, on
the whole, language is not indispensable even for "structured and
abstract thought" (p. 259)

One of the achievements of Bloom's book is that it conveys a very
strong impression of how difficult a task word learning really is, and
how much it takes to grasp even the sense of a simple content word.
But although I fully agree with Bloom's assessment of the intricacy of
lexical learning, I sometimes had the feeling that he might be
overrating the cognitive skills of infants. A case in point is his
discussion of lexical contrast. In view of the fact that there are
very few lexical synonyms, it is a good learning heuristic to assume
that new words will refer to concepts that are not already named.
Various experimental findings suggest that children use a rather
strong version of this heuristic. For example, if a preschooler is
shown a banana and a whisk, and then asked to pick out the "fendle",
he is more likely to choose the whisk. Moreover, children often find
it difficult to learn superordinate terms, or to accept that two
people may bear the same name.

How can such facts be accounted for? One explanation would be to
assume that children apply a heuristic that works well enough most of
the time, but needs some fine-tuning for dealing with special cases,
such as superordinate terms and non-unique names. And although in
general I am wary of nativist explanations, I don't think it would be
far-fetched to claim that such a default rule might be innate, in some
sense or other. Bloom favours a different explanation, however, in
which the theory of mind plays a central role:

 A child might reason as follows (implicitly, of course):
 1. I know that a banana is called "banana".
 2. If the speaker meant to refer to the banana, she would
 have asked me to show her the banana.
 3. But she didn't; she used a strange word, "fendle".
 4. So she must intend to refer to something other than the banana.
 5. A plausible candidate is the whisk.
 6. "Fendle" must refer to the whisk. (p. 68)

It is one thing to claim that children exploit clues about other
people's intentions. It is quite another thing to maintain that even
before they reach their second birthday, infants routinely employ the
neo-Gricean reasoning exhibited in this passage (if only implicitly).
I find this hard to accept, though not because I doubt that small
children can adopt the intentional stance. It is just that this train
of thought is so complex and hinges on a counterfactual premise, to
boot. Furthermore, other data discussed by Bloom suggest that small
children have a rather insecure grasp of intentional concepts, as
witnessed by the fact that they find it difficult to attribute false
beliefs to others, which makes it even more doubtful that they are
capable of reasoning in terms of counterfactual intentions.

Adopting the intentional stance is a quite sophisticated thing to do,
but even sophisticated things can be done in quite unsophisticated
ways. One can adopt the intentional stance and use information about
other people's beliefs and desires _without_ any deep understanding of
what it means to believe or want something. It is possible that young
children have a bag of tricks for exploiting intentional information,
but that their understanding of intentionality takes years to mature,
and I think it is rather likely that this is how intentional reasoning
develops. (This is why I prefer to avoid the term "theory of mind" and
have used Dennett's terminology instead.)

In the opening pages of his essay on Tolstoy, Isaiah Berlin (1953)
draws a contrast between the intellectual styles of "foxes" and
"hedgehogs". Hedgehogs are people "who relate everything to a single
central vision", whereas foxes are "those who pursue many ends, often
unrelated and even contradictory, connected, if at all, only in some
de facto way". Famous hedgehogs, according to Berlin, are Plato,
Pascal, Hegel, Dostoevsky, and Nietzsche; whereas Aristotle, Goethe,
and Joyce are foxes. In the concluding chapter of his book, Bloom
extends the dichotomy to psychological processes: "Hedgehog processes
are well captured by algorithms and involve a relatively small set of
capacities; fox processes are harder to make explicit and involve a
relatively large and diverse set of capacities. [S^] Looking at the
extremes, a reflex is very hedgehog; making up a joke is exceedingly
fox." (p. 263) It will be clear from the foregoing discussion that, in
his views on lexical learning, Bloom sides with the foxes: learning
words crucially involves pooling information from disparate sources in
ways that vary from one word to the next. Furthermore, although
children are often quick at grasping the meaning of a word, it
sometimes takes them months to learn what a word means, as in the case
of the number words. Therefore, to the extent that lexical learning
may be seen as a psychological process at all, it is more like
inventing a joke.

"How Children Learn the Meanings of Words" is a book that I am sure
will be valued not only by researchers in the area of language
acquisition, but by non-experts, as well. Drawing upon an impressive
stock of data and arguing in an eloquent and admirably transparent
way, Bloom makes it abundantly clear that lexical learning is not
just an interesting topic in its own right, but also a fruitful way of
approaching central topics in cognitive science.

As usual, the publisher has invited renowned experts to comment upon
the book, and their praise is lavish, as one may expect: "It's a
great read. Anybody concerned with language, cognition, or development
will find much of interest here." (Susan Carey) "The story [Bloom]
tells is clear, engaging, and well-documented, with a pleasant absence
of contention and polemic." (Eve V. Clark) "This is a tremendously
important book." (Susan A. Gelman) It seldom happens that I am tempted
to agree with such accolades, but this is one of those rare occasions.


Baldwin, D.A. 1993: Infants' ability to consult the speaker for clues
to word reference. Journal of Child Language 20: 395-418.

Berlin, I. 1953: The Hedgehog and the Fox. Weidenfeld and Nicholson,

Dennett, D.C. 1987: The Intentional Stance. MIT Press, Cambridge,

Oshima-Takane, Y. 1999: The learning of first and second person
pronouns in English. In: R. Jackendoff, P. Bloom, and K. Wynn (eds.),
Language, Logic, and Concepts: Essays in Honor of John Macnamara. MIT
Press, Cambridge, Mass.

Shipley, E.F. and B. Shepperson 1990: Countable entities:
Developmental changes. Cognition 34: 109-136.

About the reviewer: Bart Geurts has worked on various topics in
semantics and pragmatics, including presupposition, quantification,
and negation. He published a book on presupposition and anaphora in
1999. For further information, see
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue