Professional Documents
Culture Documents
Automaticity and Inference Generation During Reading Comprehension
Automaticity and Inference Generation During Reading Comprehension
net/publication/247498589
CITATIONS READS
47 115
2 authors:
Some of the authors of this publication are also working on these related projects:
Individual differences in reading to summarize texts: Evidence from eye movements reading and from an automatic system to assessment using LSA View project
All content following this page was uploaded by Paul van den Broek on 20 December 2016.
A reader must simultaneously carry out many cognitive processes while moving
through a text. Each process could demand all the reader's attention and pre-
vent any other process from occurring. However, as a reader's skill develops,
some of these processes begin to demand less and less attention, thus freeing
attentional resources for other, simultaneous processes. The development of
automaticity is a hallmark of skilled reading, yet there are questions regarding
which processes truly become automatic with practice. We intuitively know that
we make a wide variety of inferences as we read that both clarify and enrich a
text. However, research provides evidence that only a minimum of these infer-
ences occur automatically. Because educators should make use of this research,
recommendations are given here for fostering automaticity in inference gener-
ation.
Try to not think about an elephant. Not very easy, is it? The very
mention of the word makes you think about it. You can't stop yourself;
it's automatic. Now read the following passage:
It is quite apparent that the lack of sound means Toby has no money
to buy a present. Yet at no point does the passage mention the words
"money" or "buy." We don't struggle to understand why Toby would
shake the bank or even to know that the "it" that he shook was indeed
the bank. In fact, it would be difficult for most skilled readers to stop
themselves from filling in the missing pieces. As with the "elephant"
example, there is an automatic nature to the process, a sense that
comprehension occurs outside of our control.
The fact that there are cognitive processes beyond our control
We thank Jay Samuels for numerous helpful comments on earlier versions of this
paper.
Address correspondence to Richard Thurlow, Center for Education, Widener Univer-
sity, One University Place, Chester, PA 19013-5792, USA.
Reading & Writing Quarterly: Overcoming Learning Difficulties, 13:165-181, 1997 165
Copyright © 1997 Taylor & Francis
1057-3569/97 $12.00 + .00
166 R. THURLOW AND P. VAN DEN BROEK
1984; Posner & Snyder, 1975; Schneider, Dumais, & Shiffrin, 1984). To
understand this debate, note that in the discussion above, various
characteristics of "automaticity" are implied: Automatic processes (a)
are initiated and run their course outside of conscious control, (b) de-
Downloaded By: [University of Minnesota] At: 20:11 7 December 2007
mand few attentional resources, (c) are effortless, (d) produce little
interference with other tasks, and (e) are fast.
The first two characteristics—initiation/execution without con-
scious control and low demands on attention—describe the mechanism
of automatic processes. They are, in essence, the defining characteris-
tics of automaticity and thus are agreed upon by most researchers.
However, they are theoretical constructs that cannot be observed di-
rectly, but only through their effects on overt behaviors. This is where
the other three characteristics—effortless, fast, and producing little
interference—enter the scene. They describe possible behavioral con-
sequences of a process being automatic. A particular task will be per-
formed more quickly, with less effort, and with less interference, when
it occurs in an automatic fashion than when it occurs via a conscious
process. Because of this, automaticity should be viewed as a relative
concept.
Trouble starts when one proposes that automatic processes are fast
(or effortless or create little interference) in an absolute sense. Such
proposals lead to the adoption of absolute criteria for automaticity,
such as: for a process to be automatic it needs to be completed within
a certain amount of time (usually within hundredths or tenths of a
second), to take no effort whatsoever, or to show no interference with
other cognitive tasks. Tempting though these criteria may be, they are
inaccurate. Take, for example, the contraction reflex of the iris when a
bright light is shone into the eye. This reflex is a stereotypical example
of an automatic process, yet it would be disqualified by at least two of
the above criteria: the muscle contraction may take several seconds,
and it will consume measurable amounts of energy. The interference
criterion is also limited: although "automatic" processes such as driv-
ing may not interfere with performance on relatively simple secondary
tasks, they frequently do interfere with performance on more complex
tasks such as navigating an unfamiliar city.
The use of absolute criteria for automaticity leads to disagreements,
because for any criterion one can find exceptions. Thus, a particular
process may be relatively automatic by virtue of being much faster
than its conscious counterpart, and may still be slow by an absolute
standard. The same applies to the other properties of automaticity
mentioned. Although many automatic processes are very fast, require
little effort, and show little interference with the execution of other
tasks in an absolute sense, it is possible for a process to be automatic
168 R. THURLOW AND P. VAN DEN BROEK
Present
wants to get for
-buy
Downloaded By: [University of Minnesota] At: 20:11 7 December 2007
Toby Chris
Piggy Bank
result
No Sound
FIGURE 1 A minimally coherent representation of the ideas and relations in the Toby
story.
, Present
wants to get s* N. for
Toby Chris
shakes
goes to
Piggy Bank
No Sound
FIGURE 2 A less coherent representation, which is missing some of the relations be-
tween ideas.
AUTOMATICITY AND INFERENCE GENERATION 171
Present
wants to get for
-buy
Downloaded By: [University of Minnesota] At: 20:11 7 December 2007
can't buy
Toby needs Chris
Money
goes to shakes
has none
to get
Piggy Bank
usually has Coins
result
not there
No Sound
FIGURE 3 A very coherent representation of both explicit and implicit text information.
attention at one time to examine how each sentence relates to all the
others. Instead, we must integrate each new sentence only with adja-
cent information, unless we retrieve earlier, distant information from
memory.
A fourth constraint on a reader's ability to maintain coherence is the
reader's own standard for coherence (van den Broek, Risden, & Huse-
bye-Hartmann, 1995). Readers differ in the degree of coherence that
they pursue. Some readers may be satisfied when they understand
each individual fact or event. Others may aim for grasping the rela-
tions among various pieces of information. Still others may not be
satisfied until they know the theme or moral of the text. These differ-
ences are reflected in the amount of effort that readers invest in han-
dling difficult inferences, missing information, ambiguity, or simply a
lack of clarity. The differences may have their origin in a variety of
sources. For instance, some readers will not worry about complete
comprehension and will ignore breaks in coherence, while other read-
ers may simply be unaware that a coherence break has occurred, or
may not have the ability to remedy it. Even within a reader the stan-
dards for coherence vary, depending on the situation and the reader's
goals. For example, we typically are more relaxed about the amount of
coherence in a text when we read it for entertainment than when we
read it to study for an examination.
172 R. THURLOW AND P. VAN DEN BROEK
others, likely and enriching, are inferences that elaborate on the mean-
ing of the text. Even though there is agreement that readers are ca-
pable of making all of these inferences, a debate exists regarding
whether or not the inferences are made automatically during the read-
Downloaded By: [University of Minnesota] At: 20:11 7 December 2007
Golding, & Graesser, 1992; Thurlow, 1991). So, causal relations are also
inferred quickly as readers move through a text. However, some stud-
ies that have examined individual differences show a weaker pattern
of results for less skilled readers, implying that they may not make
Downloaded By: [University of Minnesota] At: 20:11 7 December 2007
compels the reader to make new inferences that are much more coher-
ent. The "AHA!" feeling is so automatic that it is just as difficult to
suppress as thinking about that elephant.
We have now discussed several influences on whether or not an
Downloaded By: [University of Minnesota] At: 20:11 7 December 2007
EDUCATIONAL IMPLICATIONS
Recall our statement that skill in inference generation entails both
recognition of when to initiate inference making, and generation of the
actual inference. Certain characteristics of skilled readers have impli-
cations for both of these aspects.
First, it is apparent that skilled readers do have a goal of construct-
ing a coherent memory representation. They make, at least, a mini-
mum number of automatic inferences that help integrate new material
with the previous text. If integration is difficult, they will take time to
work out the difficulty with elaborations. A recommendation for read-
ing instruction is to make sure developing readers are aware that the
pieces of a text are meant to go together and that they, the readers,
have a task to find out how the pieces fit. One instructional technique
would be to inform students that the author also has a task. The
author's task is to make the pieces easy to put together. Class discus-
sions can examine the students' beliefs about the author's overall mes-
sage and how well different sections of the text contribute to that
message. By evaluating how well authors have succeeded at their task,
the students will explore the difficulties in their own task.
Second, skilled readers have strategies and criteria for repairing
coherence breaks. They re-read difficult passages; they look back in the
text for more information; they even make up their own explanations
that make sense. These are all teachable skills, and instruction in such
strategies is important (Irwin; 1986). However, readers should also be
taught to be conservative in using these strategies. For one thing, not
all relevant information is required for coherence. Often, all a reader
178 R. THURLOW AND P. VAN DEN BROEK
elaborative inferences that will add to the richness of the reading ex-
perience.
Downloaded By: [University of Minnesota] At: 20:11 7 December 2007
REFERENCES
Bloom, C. P., Fletcher, C. R., van den Broek, P. W., Reitz, L., & Shapiro, B. P.
(1990). An on-line assessment of causal reasoning during text comprehen-
sion. Memory & Cognition, 18, 65-71.
Bransford, J. D., Stein, B. S., Vye, N. J., Franks, J. J., Auble, P. M., Mezynski,
K. J., & Perfetto, B. A. (1982). Differences in approaches to learning: An
overview. Journal of Experimental Psychology: General, 111, 390-398.
Cirilo, R. K. (1981). Referential coherence and text structure in story compre-
hension. Journal of Verbal Learning and Verbal Behavior, 20, 358-367.
Collins, A. M., Brown, J. S., & Larkin, K. (1978). Inference in text understand-
ing. In R. J. Spiro, B. C. Bruce, & W. F. Brewer (Eds.), Theoretical issues in
reading comprehension (pp. 385—407). Hillsdale, NJ: Lawrence Erlbaum
Associates.
Dell, G. S., McKoon, G., & Ratcliff, R. (1983). The activation of antecedent
information during the processing of anaphoric references in reading. Jour-
nal of Verbal Learning and Verbal Behavior, 22, 121-132.
Duffy, S. A. (1986). Role of expectations in sentence integration. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 12, 208-219.
Fletcher, C. R., & Bloom, C. P., (1988). Causal reasoning in the comprehension
of simple narrative texts. Journal of Memory and Language, 27, 235-244.
Fletcher, C. R., van den Broek, P. W., & Arthur, E. J. (1996). A model of
narrative comprehension and recall. In B. K. Britton and A. C. Graesser
(Eds.), Models of understanding text (pp. 141-163). Hillsdale, NJ: Lawrence
Erlbaum Associates.
Graesser, A. C , & Clark, L. F., (1985). The structures and procedures of im-
plicit knowledge. Norwood, NJ: Ablex.
Halliday, M., & Hassan, R. (1976). Cohesion in English. London: Longman.
Haviland, S. E., & Clark, H. H. (1974). What's new? Acquiring new information
as a process in comprehension. Journal of Verbal Learning and Verbal
Behavior, 13, 512-521.
Husebye-Hartmann, E. (1992). Causal elaborative inferences in text compre-
hension: Implications for psychology and education. Unpublished doctoral
dissertation, University of Minnesota.
Irwin, J. W. (1986). Teaching reading comprehension processes. Englewood
Cliffs, NJ: Prentice-Hall.
Just, M. A., & Carpenter, P. A. (1987). The psychology of reading and language
comprehension. Newton, MA: Allyn and Bacon.
Kahneman, D., & Treisman, A. (1984). Changing views of attention and auto-
maticity. In R. Parasuraman and D. R. Davies (Eds.), Varieties of attention
(pp. 29-61). New York: Academic Press.
Kintsch, W. A., & van Dijk, T. A. (1978). Toward a model of text comprehension
and production. Psychological Review, 85, 363-394.
180 R. THURLOW AND P. VAN DEN BROEK