Bowerman, Melissa, and Stephen Levinson, ed. (2001) Language
Acquisition and Conceptual Development. Cambridge University Press.
Reviewed by Laura Wagner, NYU
This book contains 19 papers (plus an introductory overview of
them by Levinson) considering the connection between language acquisition
and conceptual development. For the majority of the papers, the
perspective taken is neo-Whorfian: the process of language acquisition
helps to structure the concepts being developed. There are too many
papers in this volume to review them all separately, so instead I will
consider them grouped by the sections in which they appear in the book.
I want to apologize at the outset, therefore, because this approach leads
me to focus on the common themes across the papers which has often forced
me to omit many interesting particular arguments the papers have to offer.
Contents
Part 1: Foundational Issues
The first three papers consider the impact of having language
itself on one's conceptual abilities. The paper by Langer compares
primate and human conceptual development. This paper serves as a useful
anchor for the collection by reminding the reader both of our evolutionary
ancestry (we share many concepts with primates) and of the special role
that language plays in defining us as a species. Gopnik's paper reviews
her "Theory Theory" view of conceptual development and the role that
emerging language plays in helping children shape their theories of how
the world works. The final paper in this section, by Spelke & Tsivkin,
considers the role that language might play in connecting our conceptual
abilities. They review their research showing that children with limited
language, like rats (and also adults engaged in a verbal shadowing task),
can't use landmarks to help them orient in a room although their ability
to use geometric information is preserved. They argue that many complex
concepts (including ones like "to the left of the blue wall") are composed
of simpler concepts which can be combined only by means of language. The
Spelke & Tsivkin paper takes what is perhaps the most radical stance in
the book: they claim that there are some concepts -- including some
apparently simple spatial relations -- that just can't be thought without
language.
Part 2: Constraints on Word Learning?
This section consists of three papers aimed at showing that no
language specific constraints are necessary to explain word learning.
Smith presents data showing that, for example, the shape bias found in
early word learning in fact develops over time and has self-organizing
properties of the sort one might expect from a connectionist architecture.
Tomasello reviews his social-pragmatic theory of word learning and Bloom
reviews his kitchen-sink theory of word learning. None of these papers
takes an explicitly Whorfian stance, though the focus on domain general
abilities is certainly compatible with one. The resounding lesson from
this section is that the input to children is rich and complex in a
variety of ways, and that children can exploit far more information --
about emergent patterns in the language they're learning, about the
intentions of their interlocuters -- than most simple constraints
approaches give them credit for.
Part 3: Entities, Individuation, and Quantification
The first three papers of this section are linked by a common
concern with the mass/count distinction. Carey reviews her experiments
showing pre-linguistic children are sensitive to the mass/count
distinction: they track numerical identity for objects (e.g. a toy truck)
but not for substances (e.g. sand). Gentner & Boroditsky review data
showing that Japanese and American children agree about how to extend the
reference for complex objects (on the basis of shape) and substances (on
the basis of substance), but differ with respect to simple objects with no
clear function or interconnected parts: American children extend reference
based on shape and Japanese children on the basis of substance. Lucy &
Gaskins compare English speaking and Yucatec speaking children in a
non-linguistic classification task and find that given an ambiguous
object, the English speakers classify more on the basis of function while
the Yucatec speakers focus more on substance. Yucatec and Japanese both
differ from English in being classifier languages in which mass nouns are
basic syntactically and count interpretations are derived via classifiers.
Taken together, these papers argue for a basic universal mass/count
distinction which can be refined and influenced by exposure to particular
languages.
The paper by Deutsch et al. was one of my favorites for its
creative investigation of the role of differential input on pronoun use.
They looked at the use of pronouns in production and comprehension in only
children, children with siblings, and twins. They find that having
siblings accelerates pronoun acquisition in general, although twins show
some interesting delays in referring to their other twin. This paper
emphasizes that differences in input come not just from different
languages, but also from different acquisition situations.
The final two papers in this section, by Brooks et al. and Drozd,
concern children's early understanding of quantifiers. The Brooks et al.
paper examines early use of "all" and "each" cross-linguistically and
arrives at a universalist conclusion: the acquisition of these follows
similar patterns in English, Portuguese and Mandarin. The Drozd paper is
a thorough review of the debate on the acquisition of "every" but doesn't
have much to say about the relationship between this debate and conceptual
development.
Part 4: Relational Concepts in Form-Function Mapping
The first six papers in this section all consider children's early
language production and the extent to which is reflects language
particular patterns. Clark reviews data showing that in many cases,
children's production shows evidence of emergent semantic categories.
These categories are not supported by the input data but represent the
universal conceptual organization the child imposes on the linguistic
data. Contra Clark, the remaining papers all aim to show that children's
production is in line with their target language from the beginning.
Slobin provides a general introduction to this position and contrasts it
with the universalist position. Behrens documents early competence with
the German tense/aspect system; Bowerman & Choi, Brown, and DeLeon review
their data on early language specific competence with spatial terms in
Korean, Tzeltal and Tzotzil respectively. These papers aim to show that
children acquire spatial (and temporal) language without error, despite
large differences in spatial (and temporal) organization
cross-linguistically. They suggest, in general, that such error-free
learning is possible only if the linguistic input itself is organizing
children's concepts.
The final paper in this section is by Levinson and reviews his
research showing that adults' solution to a spatial ordering task reflects
the spatial system of their language: Dutch uses relative spatial terms
(e.g. "right" "left") and Dutch speakers use a relative spatial strategy
in solving a non-linguistic ordering task; Tzeltal uses absolute spatial
terms (e.g. "north" "south") and Tzeltal speakers use an absolute spatial
strategy on the same task.
Evaluation
Overall, this is a wonderful book -- broad in scope, very up to
date in the positions presented, and in general, very readable. For those
previously unfamiliar with the topic, it provides an excellent
introduction to how people approach the question of the relationship
between language acquisition and conceptual development. It is also an
excellent review of the state of the art for those who want to keep up on
this area of research.
My primary disappointment with the book was how few of the papers
actually investigated the relationship between language and conceptual
development. In their paper, Lucy & Gaskins lay out clear and appropriate
standards for what constitutes evidence that language has influenced
concepts. One of their critical standards is that the test of concepts
must be non-linguistic; otherwise, all that is being tested is a subject's
skill with their native language. To say that it is language that is
shaping concepts we must have an independent measure of what those
concepts are. Some of the papers in this volume have taken this point to
heart (Spelke & Tsivkin, Carey, Gentner & Boroditsky, Lucy & Gaskins,
Levinson) but, particularly with the papers on spatial language, this
point has been lost entirely.
Moreover, the data presented in the papers that really tested
non-linguistic conceptual differences between speakers of different
languages is actually at odds with the claims made by many of those
studying language acquisition exclusively. Both Gentner & Boroditsky and
Lucy & Gaskins report that the language-influenced conceptual differences
increase with age: teenagers and adults show the greatest conceptual
differences and very young children (age 2;6) barely show any at all.
The conclusion to be drawn from these papers seems to be that the effect
of language on thought is a cumulative one; the longer you speak a
language, the more influence it has over your concepts. This position
contrasts sharply with the claims of, for example, Bowerman & Choi, who
argue that the error free production of spatial terms (from children as
young as 2;0) is because the language specific patterns are shaping the
concepts from the very beginning. However, since Bowerman & Choi (along
with the rest of the acquisition papers in part 4) never test children's
spatial concepts aside from their production of spatial terms, there is no
real way to evaluate their claim. Children's knowledge of the
determiner/classifier systems of English and Yucatec (cf. Lucy & Gaskins)
was presumably in place well before age 9, the earliest age at which those
speakers show non-linguistic conceptual differences. Early competence
with some bit of language tells us nothing about the effects of that
language on thought.
Indeed, judging from the descriptions of the semantic domains
coded by language, there is little surprise that children learn all the
different systems so easily; the differences between the systems isn't all
that great. In laying out the conceptual domain to be lexified, Gentner &
Borodisky (for nouns) and Bowerman & Choi (for spatial terms) each present
what looks like a universal conceptual space and the types of possible
language-specific parameterizations. To be fair, not all the papers go
this route (e.g. De Leon's paper on Tzotzil spatial terms) but the idea
recurs often enough to suggest that the neo-Whorfian approach is really
not all that far from the universalist Chomskyan position after all. In
fact, one of the great surprises overall in this book is how little is
actually being claimed by most authors. For example, Gopnik and Deutsch
et al. argue that specific differences in the input can influence the rate
of acquisition of some concepts or words but neither argue that the
differences influence whether or not these concepts or words are acquired
at all. Similarly, the effects reported in Gentner & Boroditsky, Lucy &
Gaskins, and Levinson all rely on ambiguous situations that can be
reasonably interpreted in one of two ways; only in these cases does it
appear that language can be the deciding factor of how to construe a
situation. Ironically, the authors imputing the strongest role to
language in conceptual development are Spelke & Tsivkin and Carey, the
authors least concerned with cross-linguistic comparison and most
concerned with conceptual development per se.
In the end, as was pointed out in some of the papers here, the
truth of the Whorfian hypothesis -- that language influences thought -- is
an empirical question. Several of the authors here noted the fact that
for many years, linguists presumed the universalist position across the
board and people interested in investigating the Whorfian question
received little support or attention. This volume shows how much things
have changed; if anything, some of the papers in this book go too far in
presuming the a priori truth of the Whorfian hypothesis. It is far from
settled at this point how and to what extent language might influence
conceptual development but this volume makes it clear that these are
questions well worth asking.
This book was reviewed by Laura Wagner, a post-doctoral fellow at NYU.
My research centers on children's acquisition of tense and aspect, and
their non-linguistic representation of event concepts.
|
