LINGUIST List 12.1779

Mon Jul 9 2001

Review: Hollich et al., Breaking the Language Barrier

Editor for this issue: Terence Langendoen <terrylinguistlist.org>

Directory

1. David Golumbia, Review of Hollich et al. (2000) Origins of Word Learning

Hollich, George J., Kathy Hirsh-Pasek and Roberta Michnick Golinkoff (2000)
Breaking the Language Barrier: An Emergentist Coalition Model for the
Origins of Word Learning. In collaboration with Rebecca J. Brand, Ellie
Brown, He Len Chung, Elizabeth Hennon, and Camille Rocroi and with
commentary by Lois Bloom. Monographs of the Society for Research in Child
Development, Serial No. 262, Vol. 65, No. 3. Blackwell Publishers, hardback
ISBN: 0-631-22154-9, 141pp, $32.95.

Reviewed by David Golumbia, New York, NY

OVERVIEW

The book describes a sequence of experiments carried out in the Interactive
Intermodal Preferential Looking Paradigm (Interactive IPLP) designed by the
authors, a program that adds the critical variable of social interaction
into the already-important Intermodal Preferential Looking Paradigm (IPLP)
described and utilized in Hirsh-Pasek and Golinkoff (1996). The earlier
program follows the visual attention of infants to a choice of stimuli based
on verbal cues (a simple example is a voice saying "look at the ball" while
pictures of a ball and a bat appear on two different screens). The
Interactive IPLP locates the verbal stimuli in the pronouncements of an
examiner in the room with the infant and parent. By directing the
experimenter's gaze toward objects in both his or her and the infant's
fields of vision, the Interactive IPLP introduces a defined variable for
social influence into the experimental environment. It is characteristic of
the detailed approach of these investigators to both experiment design and
to theoretical underpinnings that a fundamentally multivariate understanding
of language is explored throughout the Monograph in precise and
context-sensitive detail.

SUMMARY

Chapter 1. What Does It Take to Learn a Word?; Chapter 2. The Emergentist
Coalition Model.

The authors begin with an empirical problem and a methodological one. The
empirical problem is that there seems to be a transition in infant
development occurring approximately between 12 and 19 months of age, where
the child leaps from acquiring words at a rate of around two per week to
acquiring them as rapidly as nine per day. The methodological problem is to
reconcile the various, strong, and seemingly conflicting accounts of
language development that the authors group into three camps they call
constraints/principles (intending to cover standard generativist and
optimality-theoretic approaches), associationist, and social/pragmatic.

The book's first two chapters propose that the Interactive IPLP is
compatible with an "emergentist coalition model" that "posits that
children's lexical development is the product of intricate, epigenetic
interactions between multiple factors" (17). The model is tested via
evaluations of three interconnected hypotheses: 1) "Children are sensitive
to multiple cues, attentional, social, and linguistic, in word learning"; 2)
"Children differentially weigh certain cues over others in the course of
word learning"; 3) "Principles of word learning are emergent as each
principle changes from an immature to a mature state" (18). An sequence of
twelve experiments is used to test these three hypotheses.

Chapter 3. The Interactive Intermodal Preferential Looking Paradigm
(Rocroi); Chapter 4. Learning Novel Nouns: Children Use Multiple Cues
(Brand).

The first two experimental chapters, attributed individually to students but
sharing the same clear and conversational voice employed in the rest of the
book, provide overviews of the experimental method used (Chapter 3) and the
first, in many ways grounding experiments (Chapter 4). In the Interactive
IPLP, a child sits in his or her parent's lap, presented with a scene that
is more like play than it is like a test or examination: the naturalness of
the test setting, while not stressed in this volume, is nevertheless part of
what makes the program itself compelling.

The test is based around two toys or objects that are attached with velcro
to a small table between the child and (in some cases) an examiner. "After
playing with two toys, the experimenter labels one of the toys (e.g., 'this
is a glorp'). In a subsequent test phase, the child sees the two toys
presented side-by-side on a display board. The now hidden experimenter
requests the target object. . . . it is hypothesized that children who learn
the label will look longer at the target object than at the non-target
object" (32). The experiments target their three core hypotheses by varying
the visual salience of the target object (color, appeal) as well as social
variables and more directly linguistic ones.

The first three of twelve experiments in the sequence set levels for what is
to come (Chapter 4). Unlike the subsequent nine experiments; the first three
include children of different ages, and establish that "at the ages of 12
and 24 months, children weight perceptual and social cues very differently.
. . . taken together, the first three experiments suggest that 12-month-olds
rely less on the cue of social eye gaze and relatively more on perceptual
salience, whereas 24-month-olds rely more on social cues" (61).
Interestingly, "[24-month-olds] take the perspective of the speaker into
account in linking a name to a referent; [12-month-olds], while sensitive to
conflicting cues, only seem to learn words that correspond to their own
perspective."

Chapter 5. What Does It Take for 12-Month-Olds to Learn a Word? (Hennon,
Chung, Brown); Chapter 6. Is 12-Month Old Word Learning Domain-General,
Socially Determined, or Emergent?

In the main set of nine experiments, labeled experiments 4 through 12, the
experimental subjects shift to 12-month-olds exclusively. Experiments 4
through 9 (Chapter 5) question how much stimulus is necessary for a
12-month-old to conclusively respond to an object label. The answer seems to
be, a great deal; while the naming performance of the experimenter does not
seem to impress an infant easily or initially, repeated insistence does
eventually result in name recognition. The eventual use of heavy repetition
(the name pronounced ten times in a short timeframe) suggests to the
experimenters that it is precisely a coalition of factors that result in
naming, and that the augmented strength of any one of several factors may
result in a learning event.

Experiments 10 through 12 (Chapter 6) start from this observation and
attempt to use it to pinpoint the domain-specificity of word learning. "If
early word learning relies primarily on associative processes, then babies
should link, for example, musical notes or noises to objects just as they
link words to objects" (86). Whistles, clicks, and then digital sounds are
added to the environment, but fail to attract the infant's attention. In
Experiment 12 perceptual salience is added back into the mix, and here the
coalition of cues does result in a naming event, suggesting that "the
emergentist hypothesis is correct: young word learners will accept object
noises as 'labels' if they are given enough attentional and social support"
(96). The contrast with the earlier experiments shows that "apparently, by
12 months of age, babies already know that sounds that do not emanate from
the mouth are a bad bet to serve as labels" (98).

Chapter 7. General Discussion.

In their discussion, the authors return to their three primary hypotheses
and review briefly how the experimental sequence confirms them. The immature
principle of reference, accessible to early infants but associated primarily
with the subjective point of view, undergirds a mature principle reference
that starts to emerge by 24 months of age and is able to respond to and
incorporate third-party points of view and so social facts. Some suggestions
for future research are indicated.

Commentary. Pushing the Limits on Theories of Word Learning. (Bloom)

Bloom's brief but dense commentary may be the most bracing section of this
book. An important researcher into child language development from the
empirical (observational) side (see Bloom 1973, 1991, 1993), Bloom seems to
see into theoretical questions only glanced at by the main authors. To her,
too, collaborative and integrative approaches, to which the authors'
emergentist approach claims membership, might come "closer . . . to the
truth" than more exclusive approaches (126). Nevertheless, some specific
aspects of the experimental sequence seem narrow: "development from a
first-person perspective in expressing the child's own intentionality to a
concern for the intentional states of other persons is a more general
developmental phenomenon and is not limited to a principle of reference for
word learning in the 2nd year" (127). This leads her to raise three areas in
which experimental research programs in general seem to her to be
problematic: "the persistently myopic focus on object words in word learning
research; the phantom child in the model; and the missing affect in theories
and research on word learning." Both in general and, as the authors admit,
in this Monograph, an untenably direct mapping of word to object is assumed,
since "words do not map directly to objects and events in the real world"
(128). So is the principle of reference the authors locate "also the
'heuristic' the child uses to test alternative hypotheses about what a word
might mean, and therefore explain developmental change? Or, more likely, is
the principle a convenient and accurate description, after the fact, of the
developmental changes that take place in the process of word learning?"
(129). Thus the authors have relied on a relatively circular theoretical
mechanism that helps to obscure the more central fact that "the changes in
word learning that take place in the 2nd year are epiphenomenal--a
by-product of developments in perception, cognition, and social sensitivity,
rather than attributable to development of the specifically lexical
principle of reference offered to explain it" (130). As the authors
acknowledge, the focus on nouns seems not to comport well with the way
infants actually do learn language. Whatever word learning mechanisms they
use apply with equal efficacy to words across classes: "even if verbs and
other words that are not object names represent less than half, or a
quarter, or only a tenth of a child's first 50 words, object specific
principles cannot explain early word learning unless the principles
themselves are tinkered with by building in ad-hoc procedures for
'overriding' them" (131).

Briefly, Bloom's "phantom child" problem refers to the "missing authority of
the child in the acquisition process," "the active, on-line thinking that
goes on in the moments of word learning"? (132). "Missing affect" refers to
the fact that the experimental sequence uses object salience as a proxy for
interest, therefore bypassing the critical place of the "child's affective
investment in the task," in a sense ruling out critical social knowledge
even in the course of attempting to reflect it.

DISCUSSION

Readers familiar with Hirsh-Pasek and Golinkoff (1996) will note a shift in
focus from determining the origins of syntactic principles toward locating
what they call the "origin" of word learning but seems more directly to be
its rapid acceleration around 19 months. Experimentally, this turns out to
mean a focus on just what features of word learning there are that 12-month
old infants lack and 19-month old infants possess. Theoretically, as the
authors and commentator point out, this means proceeding with assumptions
about the equivalence between words and nouns, and between learning nouns
and fixing reference in the physical world, that may weight the experiment
itself in some ways. It is also fair to say that the authors allow the
notion of "emergence" to blend somewhat amorphously with an idea of
"integration," and that to some extent the bare development of linguistic
skills is a kind of "emergence" that is not meant to be under direct
examination. Indeed, more interaction with the literature on emergence would
have helped to situate the experiment more closely (see Bates and Goodman
1999, Elman 1999, Golumbia 1998, Hopper 1987, 1988, and Weber 1997; an early
presentation of some of the material in this Monograph as Golinkoff,
Hirsh-Pasek and Hollich 1999 attempts to make these connections somewhat
more explicit). Despite the narrowing of focus on a certain aspect of word
learning, the sequence seems less specific in its theoretical entailments
than does the sequence in Hirsh-Pasek and Golinkoff (1996). That study
helped to show how a rigid system of syntactic organization seemed
incompatible with an account sensitive to the way language develops in the
child. The focus on syntax and syntactic phrases, while arguably a larger
topic than is word learning, somehow seems more precise and less problematic
than the principle of reference. In other words, the limited experimental
paradigm of the IPLP seems to suggest facts about syntactic development in
general while the Interactive IPLP, at least in this example, seems more
contained in what it reveals about word learning.

The authors seem aware of these problems, invoking Quine briefly (though
less helpfully than one might have hoped) to try to probe at whether it is
words or labels that they are learning about in the experimental sequence.
In addition to these, one wonders about the use of objects that might have
been, if even briefly, encountered by infants in their environments, and the
use of novel words (e.g. "glorp") that the infant will not have encountered
in their familiar language environment and therefore, unlike much of the
rest of the language they are acquiring, has been created for a specific
situation. One wonders if the sound patterns in these novel terms comport
well with the infants' native tongues. (The focus is narrow enough that the
authors must point out, post-hoc as it were, that the clicks and pssts used
in Experiment 10 as apparently non-linguistic mouth sounds are of course
close if not identical to sounds found in the inventory of African and Greek
languages; see 91).

Despite these theoretical conundra, the clarity of experimental goals and
descriptions and their relation to an alternative theoretical paradigm like
emergentism make this an important volume. The experimental sequence shows
clearly that multiple cues from a range of social stimuli are required
during word learning. The design and operation of the experiment and no less
its documentation as evidenced in this Monograph can serve as models for
participatory research, in terms of both the collaborators and subjects. As
the authors suggest, further refinements to either narrow the scope of the
experiment toward object labeling in a theoretical sense, or to widen the
scope to incorporate more kinds of words as they are acquired, may provide
even more fruitful results.

In short, this is an admirably designed and executed experiment, concisely
and admirably documented, that makes important points about the multivalent
process that is child word learning; perhaps it is mark of the experiment's
strong design that it seems to raise as many questions as it answers, as
does language learning itself.


REFERENCES

Bates, E. and Goodman, J. C. (1999). On the Emergence of Grammar from the
Lexicon. In MacWhinney (1999), 29-80.

Bloom, L. (1973). One Word at a Time: The Use of Single-Word Utterances
Before Syntax. The Hague: Mouton.

Bloom, L. (1991). Language Development from Two to Three. Cambridge:
Cambridge University Press.

Bloom, L. (1993). The Transition from Infancy to Language: Acquiring the
Power of Expression. Cambridge: Cambridge University Press.

Elman, J. (1999). The Emergence of Language: A Conspiracy Theory. In
MacWhinney (1999), 1-28.

Golinkoff, R. M., Hirsh-Pasek, K. and Hollich, G. (1999). Emergent Cues for
Early Word Learning. In MacWhinney (1999), 305-330.

Golumbia, D. (1998). "Emergence." In M. A. Gernsbacher and S. J. Derry
(eds.), Proceedings of the 18th Annual Conference of the Cognitive Science
Society. Mahwah, NJ: Lawrence Erlbaum, 412-417.

Hirsh-Pasek, K. and Golinkoff, R. M. (1996) The Origins of Grammar: Evidence
from Early Language Comprehension. Cambridge, MA: The MIT Press.

Hopper, P. (1987) "Emergent Grammar." In Berkeley Linguistics Society.
Proceedings of the 13th Annual Meeting. Berkeley, CA: Berkeley Linguistics
Society, 139-157.

Hopper, P. (1988). "Emergent Grammar and the A Priori Grammar Postulate." In
Tannen, D. (ed.) Linguistics in Context: Connecting Observation and
Understanding. Advances in Discourse Processes, Volume 29. Norwood, NJ:
Ablex Publishing, 117-134.

MacWhinney, B. (ed.) (1999) The Emergence of Language. Hillsdale, NJ:
Lawrence Erlbaum.

Weber, T. (1997). "The Emergence of Linguistic Structure: Paul Hopper's
Emergent Grammar Thesis Revisited." Language Sciences 19(2), 177-196.


The reviewer is an independent scholar who works on cultural studies of
linguistics, philosophy and computation.