LINGUIST List 10.928

Wed Jun 16 1999

Review: MacWhinney: The Emergence of Language

Editor for this issue: Andrew Carnie <carnielinguistlist.org>


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 carnielinguistlist.org

Directory

  1. Dina Belyayeva, RE: Available for Review: Historical, Lang Acquisition

Message 1: RE: Available for Review: Historical, Lang Acquisition

Date: Mon, 07 Jun 1999 04:38:40 -0400
From: Dina Belyayeva <belyayevphys.ufl.edu>
Subject: RE: Available for Review: Historical, Lang Acquisition

MacWhinney, B. (Ed.). (1999). The Emergence of Language. Lawrence 
Erlbaum Associates, Publishers. Mahwah, NJ. 500 pages.


Reviewed by Dina Belyayeva

Synopsis:

This book is the first comprehensive collection of papers that promote an 
emergentist approach to language acquisition. Emergintism is a conceptual 
framework that does not explicitly reject either nativist or empiricist position 
but rather takes a step further from the current plateau of the nature-nurture 
debate by showing how language emerges from interactions between biological 
and environmental processes. The 16 chapters of the book were initially 
presented at the 28th Carnegie Mellon Symposium on Cognition, and offer 
contributions from a broad range of perspectives, such a connectionist, lexicalist, cognitive, functional and social-pragmatic. 

Overview and Critical Evaluation:

The book is exceptionally well written which makes some highly specialized 
topics accessible to a much wider audience of linguists and psychologists 
interested in language acquisition issues. 

The first chapter (The Emergence of Language: A Conspiracy Theory -- 
Jeffrey L. Elman) offers an outline of a connectionist perspective on language 
developed by Elman and his colleagues in their 1996 book RETHINKING 
INNATENESS. Unlike nativism that defines innateness in terms of specific 
wiring at the level of representations, emergentism redefines innateness by 
offering a taxonomy of levels: representational, architectural and chronotopic 
(responsible for timing). An innate behavior, therefore, is a result of interactions between processes that modify environmental input at each
of these levels. Elman uses computer simulations to demonstrate
how non-domain-specific processing constraints result in domain-specific
behavior. Although he purposefully draws all the attention to a single
level (chronotopic), more elaboration on how levels interact (particularly 
Table 1.1.) is needed for readers to appreciate the proposed taxonomy.

Elizabeth Bates and Judith C. Goodman (Chapter 2 - On the Emergence of 
Grammar from the Lexicon) take on the debate about domain-specificity of 
language by offering a unified lexicalist approach to grammar acquisition and 
processing. In this chapter they pursue two objectives: to demonstrate (1) that 
grammar and lexicon are acquired by the same mental-neural mechanisms, 
and (2) that these mechanisms are not unique to language. To support their 
position they offer a summary of longitudinal and cross-sectional data from 
normal children that reveals strong correlation between vocabulary size and 
grammar development. Their reevaluation of existing experimental and 
longitudinal studies dealing with atypical populations helps to rid of some 
stereotypes that were often used as evidence for grammar-specific neural 
wiring in the brain. 

A more moderate emergentist view on the role of the lexicon in grammar 
acquisition is offered by Adele E. Goldberg (Chapter 7 -- The Emergence 
of the Semantics of Argument Structure Constructions). According to this 
view, argument structure acquisition is an emergent property that derives 
from the processes of categorization and generalization of the meaningful 
input. Within this approach a particular role is assigned to light verbs that 
are characterized by more general semantics and greater accessibility in 
a wider variety of contexts. 

An interesting diachronic perspective on emergence of grammar is proposed 
in Chapter 3 (Generativity and Variation: The Notion 'Rule of Grammar' Revisited
 -- T. Givon). As a premise for the discussion Givon offers a review of 
philosophical antecedents to nativist and empiricist positions. A Cognitive-
Adaptive Perspective that he later details takes an intermediate position that 
is more in line with the emergentist trend. A range of typological and 
variational data is presented to support for this middle-ground position.
 Some acronyms (pp 96-97) may be not familiar to a wider audience of readers
and need to be explained.

Chapters 4, 5, and 6 offer processing accounts that take into consideration 
statistical aspects of linguistic input. 

Chapter 4 (The Emergence of Grammaticality in Connectionist Networks - 
Joseph Allen & Mark S. Seidenberg) begins with a discussion of problems 
created by the competence-performance distinction adopted by the generative 
approach. According to this approach statistical aspects of linguistic input
 (e.g. impoverished input, negative evidence) are excluded as important factors 
in language learning. To demonstrate the contrary the authors chose the 
concept of grammaticality. A&S used a variant of a simple recurrent network 
to demonstrate that grammaticality judgments do not necessarily require 
access to principles of grammar, but may be based on statistical regularities 
of linguistic input. 

In Chapter 5 (Disambiguation and Grammar as Emergent Soft Constraints - 
Risto Miikkulainen & Marshall R. Mayberry, III) M&M present computer 
simulations to demonstrate how particular kinds of linguistics constraints 
(soft constraints) emerge from statistical regularities of word co-occurrence. 
Although their models successfully apply soft constraints by correlating 
new input with past contexts, it's not clear whether they will be able to 
resolve structural ambiguities that trigger garden-path processing. 

Maryellen C. MacDonald (Chapter 6 - Distributional Information in Language Comprehension, Production, and Acquisition: Three Puzzles and a Moral) 
uses analyses of verb modification ambiguity and heavy-NP shift to 
demonstrate how processing constraints emerge from distributional information 
that was made available to a speaker in prior comprehension and production 
events. The same system of emergent constraints is proposed to govern 
language acquisition. In conclusion she offers a review of studies that can 
develop this branch of language acquisition research. It is surprising though 
that she does not consider the Optimality Theory accounts as potential 
contributors to the field. 

Brian MacWhinney (Chapter 8 -The Emergence of Language from 
Embodiment) presents a unified theoretical framework of language organization 
and processing by using a cognitive ability of perspective-taking as a starting 
point for emergence of embodied meaning. Four perspectival systems 
(affordances; spatio-temporal reference frames; causal action chains; and 
social roles) are used as central organizing principles that provide common 
cognitive ground for many distinct language phenomena. MacWhinney outlines neurophysiological implications of the proposed hypothesis and clearly defines 
its limitations.

Catherine E. Snow (Chapter 9 - Social Perspectives on the Emergence of 
Language) promotes a view according to which children's linguistics abilities 
emerge from interactions between children's social capacities and social-
pragmatic conditions of their immediate environment. She introduces a new 
form of bootstrapping -- pragmatic precocity -- as the primary force that helps 
to take the language learning process off the ground. 

Chapters 10, 11, and 12 address issues of lexical acquisition and processing. 
The first chapter in the group (Children's Noun Learning: How General Learning 
Processes Make Specialized Learning Mechanisms - Linda B. Smith) presents 
results from a series of longitudinal studies that demonstrate how general 
mechanisms of associative and attentional learning create shape bias in learning 
of count noun terms. The conclusion that word learning emerges from general l
eaning processes not special to language follows the "party line" of the 
emergentisit program outlined by Elman. The author, however, is not sufficiently 
explicit on the criteria that distinguishes between language-specific and 
general learning processes. The remark that attentional learning can be
also achieved by non-linguistic perceptible cues, such as hand gestures,
inadvertently entails that signing is a non-linguistic behavior. 

Chapter 11 (Emerging Cues for Early Word Learning - Roberta Michnick 
Golinkoff, Kathy Hirsh-Pasek & George Hollich) proposes the developmental 
lexical framework that captures mechanisms of early word learning. The 
authors adopt a hybrid emergentist position that attempts to offer a common 
ground for various approached to word learning, such as the constraints-based, 
social-pragmatic and domain-general. The principles that were largely borrowed 
from the constraints-based literature are said to evolve from basic into more 
complex as children become more sophisticated in their abilities to weight a 
variety of cues. A new method (The Interactive Intermodal Preferential Looking 
Paradigm) is offered to illustrate how children learn to incorporate these
principles. The method does not require cumbersome eye tracking equipment
to measure duration of infants' attentional states. Yet, it is not clear
how this paradigm can be implemented to register whether a child actually
follows experimenter's gaze. As a model, the proposed hierarchy of principles
lacks internal structure that could be used to define mechanisms responsible for
transitions between and within tiers. 

William Merriman (Chapter 12 - Competition, Attention, and Young Children's 
Lexical Processing) presents a model of children's lexical processing (CALLED) 
that ties many loose ends existing in the literature. The model's centerpiece is 
a device that uses associations between various dimensions (features, functions, 
exemplars) and contextual cues (objects, scenes, events) to acquire and access 
words. Words' retrieval can be affected by learned attentional responses to
words. Factors affecting attention include recency and frequency effects,
distinctiveness of features, and social, pragmatic and linguistic cues. What
makes this model particularly attractive is that its competition-attention
component can be used to construe lexical principles (e.g. Shape Bias and
Mutual Exclusivity) as emergent properties. It also offers a rule that can
predict which principles are most prominent at different points of lexical
development. 

The remaining chapters address issues of phonological development.

In Chapter 13 (Statistical Learning in Linguistic and Nonlinguistic Domains - 
Richard N. Aslin, Jenny R. Saffran & Elissa L. Newport) preferential listening 
technique was used to demonstrate infants' sensitivity to phonotactic 
regularities of artificial languages. The obvious discord between the
experimental findings suggesting that general learning mechanisms are
employed in language learning and concluding remarks that take a rather
sharp turn towards nativism creates a rather awkward impression. If the 
purpose of the paper were to demonstrate that "unconstrained learning
mechanisms will not, by themselves, 
correctly learn just those things that every human baby learns" , than the study 
should have been designed in a way that could generate supporting evidence. 

Chapter 14 (The Emergence of Phonology from the Interplay of Speech 
Comprehension and Production: A Distributed Connectionist Approach -David C. 
Plaut and Christopher T. Kello) and Chapter 15 (The Emergence of Faithfulness - 
Joseph Paul Stemberger & Barbara Handford Bernhardt) address the issue of 
phonological development from two different perspectives - connectionist and 
Optimality Theory (OT). P&K present a computer stimulation of the framework 
that captures computational aspects of phonological processing as the basis 
or emergent phonological representations. The stimulation provides an impressive 
research tool by being able to demonstrate how a model can learn from variable 
input and use the extracted knowledge to gradually achieve the target production 
level. S&B present their variant of OT, according to which constraints that
guide children's phonological development are not innate domain-specific
processes, but rather emergent principles sensitive to communicative and
Bilingualism are the major areas of my research interests. Other areas of 
interest include semantic memory disorders and models of language production 
and comprehension.
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue