________________________________
Thinking and Speech and Protocol Analysis
Peter Smagorinsky
University of Georgia
Abstract
Some form of verbal report=8Bthat is, a research=20
participant=B9s concurrent or retrospective verbal account of=20
thought processes during problem-solving activities=8Bhas been=20
used throughout this century as the data base from which=20
psychologists have developed theories of human mentation. =20
Newell and Simon (1972) and Ericsson and Simon (1980, 1993)=20
have provided extensive justification for using one such=20
method, protocol analysis, as a tool for investigating cognition=20
from an information processing (IP) perspective. Their=20
arguments have characterized protocol analysis as a=20
methodology capable of providing evidence of the ways in=20
which people attend to information stored in short term=20
memory (STM) in order to solve problems, with the evidence=20
providing them with the grounds from which to generate=20
models of human cognitive processes. A different view of=20
protocol analysis is available from the perspective of cultural-
historical activity theory (CHAT) based on the work of=20
Vygotsky (1987), Leont=B9ev (1981), and others, and its concern=20
with the mediation of human development by culturally- and=20
historically-grounded signs and tools. Because of its emphasis=20
on culturally channeled development, a CHAT perspective=20
views speech, including the speech that serves as evidence for=20
cognition in psychological research, as a tool that potentially=20
enables changes in consciousness. In this article I outline a=20
CHAT perspective that accounts for protocol analysis along=20
three key dimensions: (1) the relationship between thinking=20
and speech from a representational standpoint, (2) the social=20
role of speech in research methodology, and (3) the influence of=20
speech on thinking during data collection. The purpose of this=20
discussion is not to refute the IP perspective on protocol=20
analysis but to illustrate how this method can be viewed=20
through a CHAT lens and to identify alternative assumptions=20
that must be made in order to use it from a CHAT perspective.
------------------------------------
How to Study Thinking in Everyday Life:=20
Contrasting Think-Aloud Protocols With Descriptions and=20
Explanations of Thinking
K. Anders Ericsson Herbert A. Simon
The Florida State University Carnegie-Mellon University
Tallahassee, Florida Pittsburgh, Pennsylvania
Submitted commentary on Peter Smagorinsky's paper=20
"Thinking and Speech and Protocol Analysis" in Mind, Culture,=20
& Activity
In his paper Smagorinsky (this volume) describes how protocol=20
analysis (Ericsson & Simon, 1980, 1984, 1993) can be applied=20
to study thinking within the perspective of cultural-historical=20
activity theory (CHAT). Smagorinsky=B9s main claim is that=20
verbalization of thinking as speech is a =B3process through which=20
thinking reaches a new level of articulation=B2 (p. 29) and he=20
explicitly questions the empirical evidence reviewed by us that=20
it is possible under some circumstances to have subjects think=20
aloud without altering the course of their thinking. A primary=20
reason for his misunderstanding appears to be that=20
Smagorinsky focused almost exclusively on our earlier=20
publications that described our original theoretical proposals=20
and thereby missed the numerous subsequent experiments=20
explicitly testing them. In fact, Ericsson and Simon (1993)=20
discusses over thirty additional studies that provide results=20
consistent with our theoretical framework. In this commentary=20
we briefly review the evidence that supports different types=20
of verbalization activities and describe how our distinctions=20
match the classic distinction between thinking as inner speech=20
and as social speech and give explicit quotes by Vygotsky=20
(1962) where he argues for very similar differences. Within=20
our framework we can incorporate the circumstances where=20
verbalization of thinking (thinking aloud) can be made without=20
reactive effects and other circumstances where verbal=20
descriptions and explanations of thinking serve =B3as a tool that=20
potentially enables changes in consciousness=B2 (Smagorinsky,=20
this volume, p. 2).
------------------------------------
Learning From Distributed Theories of Intelligence
Paul Cobb
Vanderbilt University
Abstract
The analysis reported in this paper is grounded in the=20
practice of classroom-based developmental or=20
transformational research, and focuses on the distributed=20
views of intelligence developed by Pea (1993) and by=20
Hutchins (1995). The general areas of agreement with this=20
theoretical perspective include both the non-dualist=20
orientation and the critical role attributed to tool use. Against=20
this background, I focus on two aspects of the distributed=20
view that I and my colleagues have found necessary to=20
modify for our purposes. The first concerns the legitimacy of=20
taking the individual as the unit of analysis, and here I argue=20
that the distributed view implicitly accepts key tenets of=20
mainstream American psychology=B9s characterization of the=20
individual even as it explicitly rejects it. The second=20
modification concerns distributed intelligence=B9s=20
characterization of tool use. Drawing on a distinction made by=20
Dewey, I argue that it is more useful for the purposes of=20
instructional design to focus on activity that involves the tool=20
as an instrument, rather than focusing on the tool itself.