Article

Perceiving words during reading: Lack of facilitation from prior peripheral exposure

Abstract and Figures

As their eye movements were being monitored, college students read short texts displayed on a cathode-ray tube. As they read, the contents of certain word locations changed from fixation to fixation, alternating between two words differing in two letters. This manipulation had no effect on reading unless the subjects happened to regress to or reread the word later. The results indicated that these words, which were low in contextual constraint, were read only when directly fixated, and that there was no facilitation from prior peripherally obtained information about the words.
Content may be subject to copyright.
Perception &Psychophysics
1982,32
(3),271-281
Perceiving words during reading: Lack
of
facilitation from prior peripheral exposure
GEORGE W. McCONKIE, DAVID ZOLA, HARRY E. BLANCHARD,
and GARY S. WOLVERTON
University
of
Illinois at Urbana-Champaign, Champaign, Illinois
As their eye movements were being monitored, college students read short texts displayed
on a cathode-ray tube. As they read, the contents of certain word locations changed from
fixation to fixation, alternating between two words differing in two letters. This manipula-
tion had no effect on reading unless the subjects happened to regress to or reread the word
later. The results indicated
that
these words, which were low in contextual constraint, were
read only when directly fixated, and that there was no facilitation from prior peripherally
obtained information about the words.
271
Present evidence suggests that visual information
acquired from peripheral visual areas on one fix-
ation during reading facilitates the identification
of words available foveally on the next fixation
(McConkie &Rayner, 1976a; Rayner, 1975, 1978).
While there are many possible mechanisms by which
such facilitation might be achieved, for present pur-
poses these can be divided into two broad cate-
gories. First, it may be that words are often per-
ceived in subword units (letters, letter groups, syl-
lables, etc.), with one or more subunits of a word
being perceived on one fixation and the remainder
on the next (McConkie, 1979). Thus, words may be
identified from subparts which are frequently ob-
tained on different fixations. This position will be
referred to as the subword-unit hypothesis.
It
would
suggest that parts of a word perceived on one fixa-
tion are not reprocessed on the next, and that the
individual fixation span (Underwood &McConkie,
Note 1), or the region perceived during a fixation,
does not necessarily extend to word boundaries.
A· second way in which peripherally obtained in-
formation might facilitate processing
of
the foveal
stimulus on the next fixation will be referred to as
the information-accrual hypothesis. This assumes
that the stimulus pattern, while lying too far into
the periphery to support word identification, has still
been partially processed, and the information gained
facilitates word or meaning identification on the next
fixation. For instance, it may be that certain features
of a word, such as its general shape or length, or
features of some letters in the word, have been ob-
This research was conducted under Grants MH 32884 and
MH 33408 from the National Institute of Mental Health to the
first author and National Institute of Education Contract HEW-
NIE-C-400-76-0l16 to the Center for the Study of Reading. Copies
of this paper can be obtained by writing to George W. McConkie,
Center for the Study of Reading, 51 Gerty Drive, Champaign,
Illinois 61820.
Copyright 1982 Psychonomic Society, Inc.
tained, or that some sort of preprocessing has per-
mitted priming of the lexical entries of a word set
which includes the correct word, on the basis of
semantic or phonological information, for instance.
In general, this hypothesis assumes that the informa-
tion obtained is not the identification of subparts
of the word (letters or orthographic units), but is
more general information that places constraints on
what the word might be. Admittedly, the distinction
between these two hypotheses becomes difficult if
one thinks of features
of
words or letters as being
subparts of those units, but this will not cause dif-
ficulty in the context of the present study.
The other possibility, of course, is that information
obtained peripherally from a word on one fixation
does not facilitate its processing when it comes into
the foveal region for the next fixation. This could oc-
cur either because such visual analysis of peripheral
words is not attempted during fixations in reading
or because, when identification of a word fails, all
information accumulated about it is discarded and
processing begins anew on the next fixation. This
would require that words typically be identified only
from the visual pattern present during a fixation,
without the use of information obtained during prior
fixations. There would certainly have to be excep-
tions to this, as when words are split between two
lines or two pages, or when a word is so long as to
require more than one. fixation to perceive it. These
conditions would necessitate the additional ability to
perceive and use subparts, such as syllables or the
parts of compound words. However, it would as-
sume that this was not normally the way perception
proceeds in skilled reading. This position will be re-
ferred to as the word-unit hypothesis, since it as-
sumes that entire words are typically identified dur-
in a single fixation, rather than being perceived in
subparts or accruing information about them to be
used later.
0031-5117/82/090271-11 $01.35/0
272
McCONKIE, ZOLA, BLANCHARD, AND WOLVERTON
Present evidence favors the peripheral acquisition
of
information,
and
thus stands against the word-
unit hypothesis, although it is unclear whether this
facilitation occurs in an information-accrual or sub-
word-unit manner. Having erroneous letters in the
periphery during one fixation can inflate the duration
of
the next fixation, when that region is brought into
foveal vision, even though the errors have now been
replaced by normal text (O'Regan, 1980; Rayner,
1975; Underwood &McConkie, Note 1). In addi-
tion, as Rayner (1975) found, if a word is changed.
during the saccade taking the eyes to
that
location,
some disruption results, even when both words are
appropriate to the context. Similarly, in a word-
naming task, having a word in the periphery during
one fixation reduces the time required to name it
when it is in the fovea on the next fixation (Rayner,
1978; Rayner, McConkie, &Ehrlich, 1978), although
this facilitation may depend on subjects' familiarity
with the word set used or on the degree
of
contextual
constraint operating (McClelland &O'Regan, 1981;
Paap
&Newsome, 1981). These results have been
taken as support for the notion that information ob-
tained peripherally on one fixation is brought to
bear in perceiving those words on the next fixation.
The information carried across is apparently
not
strictly visual in nature (McConkie &Zola, 1979),
and
there is some evidence that it may
not
be se-
mantic or phonetic (Rayner, McConkie, &Zola,
1980). However, these studies have failed to indicate
whether the information obtained peripherally was
of
subword units (McConkie, 1979) or more general
information accrual. One study suggests
that
it may
be subword units that are acquired, although the task
used was one
of
naming words rather
than
one
of
normal reading (Rayner, McConkie, &Zola, 1980).
The present study attempted to investigate, in a
more controlled manner, whether peripheral facilita-
tion was taking place during reading, and, more
importantly, to do this in a way which would allow
a choice between the subword-unit hypothesis
and
the information-accrual hypothesis as an explanation
for any observed facilitation. In order to do this,
sets of four words which differed in only two letters
(for example, pears,
bears,
peaks, and
beaks)
were
identified. Sentences which contained one word loca-
tion in which any
of
the four words could appropri-
ately fit were then written. Subjects read these sen-
tences, displayed on a cathode-ray tube (CRT) under
computer control, as their eye movements were
being monitored. During saccades in the region
around the critical word location, the display was
changed from one
of
two
of
the words
that
differed
in both letters (e.g.,
bears
and peaks) to the other.
Thus, the word in
that
location alternated between
two
of
the possible words on successive fixations.
If
the subword-unit hypothesis is an accurate de-
scription
of
perception during reading, there should
be times when subjects would acquire the first letter
of
the word during one fixation
and
the fourth
letter during the next, thus perceiving a word which
was, in fact, never present on the CRT
(beaks
or
pears
in the above example). This could well occur
without any evidence of disruption in the eye-
movement pattern, since the reader would likely not
know the word had changed. Any combination
of
letters perceived as the word changed would result in
an appropriate reading of the sentence. On the other
hand, if perception occurs by an information-accrual
process, then the change
of
the pattern in that word
position from one fixation to the next should pro-
duce disruption in reading, as found in prior studies,
but it would be unlikely that the reader would re-
port seeing words that were never actually on the
CRT. Finally, if subjects showed no evidence of dis-
ruption from the changing letters,
and
also failed to
perceive words
not
present on the CRT (but con-
structed from parts of words that were present),
this would be taken as evidence for the word-unit
hypothesis.
METHOD
Subjeds
The subjects were 16 University
of
Illinois undergraduates
who had normal uncorrected vision, were native speakers of
English, and were paid for their participation in the experiment.
All the subjects had previously participated for 4 to 6 h in another
experiment involving the same general type of task.
Materials
Twenty-five quadruplets of five-letter words were chosen such
that each word in a set differed from one
of
the other words in
only the first letter, from a second word in only the fourth,
and from the remaining word in both the
rust
and fourth letters.
For each quadruplet, a short text of one to three sentences was
constructed which made sense when any of the four words oc-
cupied a particular word position, to be referred to as the cri-
tical word location. These texts are presented in the Appendix.
In addition to these texts, eight others of similar length and style
were used as warm-up and filler sentences.
Apparatus
The text was displayed one line at a time on a Digital Equip-
ment Corporation Model
VT-ll
CRT. The CRT has a P-31
phosphor that decays to
1070
of the original intensity in 500
lJSec
and a hardware character generator capable
of
producing upper-
and lowercase letters. The display was refreshed every 3 rnsec.
Thus, display changes could be made within 3 rnsec without inter-
rupting the refresh cycle. The CRT was 68 em away from the
subject, which made 1 deg of visual angle equivalent to four
character positions. The subject was provided with a button which
called the next line
of
text onto the CRT.
Eye movements were monitored with an SRI Dual Purkinje
Eyetracker. The procedures currently used maintain an accuracy
within
Y4
deg
of
visual angle. Only the horizontal channel of
the eyetracker was monitored. The CRT and the eyetracker were
interfaced with a Digital Equipment Corporation
PDP-ll/40
computer, which was programmed to sample eye position every
millisecond
and
to make display changes at certain times when
the eyes were identified as being in a saccade.
Procedure
The subjects were fitted with a bite bar and headrest in order
to minimize head movement. The 33 texts were split into two
groups with a short rest in between. Before and after reading
each group, the subjects were engaged in a calibration task, in
which they successively fixated on five dots placed at equidistant
points on the CRT on the line on which text was to be displayed.
As each dot was fixated, the subject pressed a button, which
caused the computer to sample the voltage level of the eyetracker
for that position. These values were used to translate eyetracker
voltage levels to eye-position locations on the CRT while the sub-
ject was reading. Values obtained before and after reading each
group
of
texts were compared to judge the accuracy of the data.
While the subjects read each text, display changes occurred
during saccades made in the region of the critical word. The dis-
play change consisted of substituting one line of text for another
while the eyes were in a saccade. In the experimental condition,
the substituted line
of
text was the same as the original except
that the first and fourth letters of the critical word had been
changed, replacing the word with its alternative. So, for successive
fixations in the vicinity of the critical word, two words differing
in two letters alternated from one fixation to another. For ex-
ample, if on one fixation the word blame were present, on the
second, it would bef/are, on the third, blame, on the fourthf/are,
etc. (with the actual switching occurring during the saccades be-
tween the fixations). In the control condition, the substituted
line of text was identical to the original line, so that on each fixa-
tion the same alternative was present.
The region within which this switching occurred was defined
in the following manner. Three boundaries were set on each line
containing a critical word location. The first, the enabling bound-
ary, was always 11 character positions to the right
of
the begin-
ning of the line. No display changes were permitted until the eyes
had fixated at least once to the left of that boundary. The second,
the initiating boundary, was 18 character positions to the left of
the critical word location, or one position to the right of the en-
abling boundary if it was 18 or fewer character positions to the
left of the critical word location. The first display change occurred
during the saccade following the first flxation to the right of the
initiating boundary, given that the eyes had previously fixated
left of the enabling boundary, and the word changed during each
following saccade until terminated by one of two possible events.
The third boundary, the terminating boundary, was 11 character
positions to the right of the first letter of the critical word lo-
cation. Display changes were permanently disabled for a given
line following the first fixation to the right of the terminating
boundary, or following the first regressive saccade, after such
changes had been initiated. Thus, display changes occurred only
during saccades following fixations that lay between the initiating
and terminating boundaries. Within this region, they occurred
only if (1) the enabling boundary had previously been crossed,
(2) no previous regressions had been made since the region was
entered, and (3) this region had not been read previously (that is,
the initiating and terminating boundaries had not previously been
crossed, in that order). With this algorithm, the decision as to
whether a display change should occur during a saccade was made
during the prior fixation, making it possible to insure that all
changes took place early during the saccade and that none oc-
curred at or after the time the eyes were coming into a fixation.
After reading each text, the subject was presented with each
of the words in the quadruplet relevant to that text, one at a time.
The order
of
these test words was randomized for each text.
The subject was instructed to indicate whether each test word
had been present in the text. The subject was supplied with two
buttons, one to indicate
"yes"
and one to indicate
"no."
Note
that in each test set for texts in the experimental condition,
two words had actually been in the text and the other two had not.
In the control condition, only one of the four words had been
present during reading.
PERCEIVING WORDS DURING READING 273
Each subject received the texts and test items in the same
order, but the order of assignment of conditions to texts was
counterbalanced across subjects. The subjects were divided into
eight groups. Groups 1 to 4 had 13 experimental and 12 control
texts. For Groups 5 to 8, the conditions to which the texts were
assigned were reversed, resulting in 12 experimental and 13 con-
trol texts. Groups I to 4 each had a different word in the critical
word location when the text initially appeared, as did Groups 5 to
8. The initially appearing word was the only word present in the
control texts; in the experimental texts, the contents of the critical
word location alternated between the initially appearing word and
the word created by changing both the first and fourth letters.
Thus, the conditions, texts, and word alternatives were all counter-
balanced with respect to each other.
RESULTS
The results will be reported as providing answers
to four questions. First, how accurate were the sub-
jects' responses on the test items? Second, did the
subjects report
seeing
words that were never present as
they read under the experimental conditions? Third,
within what region were the critical letters perceived?
Fourth, did the letter changes that were taking place
cause interference (or reduce normal facilitation)
during reading?
Accuracy
of
the Subjects' Responses
The different patterns
of
responses to the test items
and the frequency with which each occurred are
shown in Table 1. An examination of the response
patterns for the texts read under the control con-
dition, in which the same word always occupied
the critical word location, indicates that
850/0
of the
time the subjects reported seeing only the word ac-
tually present in the text. On another
5.50/0
of the
instances, they chose the correct word plus another
word; on 3.5%, they selected no word; and on the
remaining 6%, they chose a wrong word, one not
in the sentence.
Table 1
Responses Given to Test Word
Frequency
Response Type CE
Correct Response
Total 170 161
Single Correct Response 170 156
Two Correct Responses *5
Errors
Total 30 39
No Response 78
Single Erroneous Response 12 17
Multiple Words, Including Correct Response 11 13
Multiple Words, Not Including Correct Response 01
Overall
Total 200 200
Note-C
=control texts; E =experimental texts.
*Not applicable because there was only one correct response.
274
McCONKIE,
ZOLA,
BLANCHARD,
AND
WOLVERTON
O+--r---r---.,.--,4!!o,..,l~~---r--r---r----r~..->t!lT-
-16 -14 -12 -10
-8 -6
-4
-2 0 2 4 6 8 10
~
he
~
CI
ass
i
fie
s
~{ble
a{~}s
_
fro
m~
LOCATION OF FIXATIONS
Figure 1. Number of fixations at each letter position with re-
spect to the critical word location, and the number of these on
which the word selected as having been seen was present durfnR
that fixation. Only data from instances in which the subject did
not regress into this region, and in which the subject responded
correctly on the test, are Included.
cation of the critical word with minima to either
side, reflecting the tendency
of
subjects to fixate
the centers of words (O'Regan, 1981; Rayner, 1979a;
Zola, Note 2). This maximum does not necessarily
reflect any tendency to fixate the critical word more
than other words; it simply reflects the fact that the
data are grouped with respect to the position of that
word. No such consistent correspondence exists for
other words across the passages when the data are
grouped in this manner.
The data
of
interest are the frequencies with which
the word present during fixations at different loca-
tions was the word reported as being seen. Each
such fixation is a candidate for being the fixation on
which the word was identified (though,
of
course,
this does not mean that the word was identified
on all these fixations). The frequencies
of
these fix-
ations are shown by triangles in Figure 1. As can be
seen, in every instance but one in which a fixation
fell directly on the critical word, the word present
on that fixation was reported as having been seen.
Of
11
fixations on the space prior to and following
the word, the word reported as being seen was pres-
ent on 6. And during fixations on the five locations
before or after the word (including spaces before
and after), the word reported was present on only
5 of the 88 fixations. In each
of
these latter cases,
there was another fixation directly on the critical
word or on the space before or after, on which the
reported word was present. Thus, it appears that
the critical word was being identified only on fix-
ations directly on that word, or sometimes on the
space before or after. Apparently, whether the
first or fourth letter of the critical word location
.....
TOTAL NUMBER FIXATIONS
AT THAT LOCATION
t::.--l:>
NUMBER FIXATIONS ON WHICH
SELECTED WORD WAS PRESENT
2
4
16
20
18
III
Z
o14
~
11
12
~
10
a:
1Il
8
::li
~
6
Tbe
Region Within Whicb tbe Letters
Were Being Identified
In order to obtain a general indication
of
the re-
gion within which the eyes were centered when the
critical word was being read, experimental condition
instances were selected in which the subject (1) made
a single correct response and (2) showed no regres-
sions and no refixations following a regression in the
region around the critical word (from 10 letter posi-
tions prior to it to 5 letter positions following it)
and in which (3) the letter was switching during sac-
cades and (4) the equipment was tracking the read-
er's eyes properly. This yielded a total
of
73 un-
ambiguous cases. All fixations on these sentences
between 15 letter positions to the left of the critical
word and 6 to the right were then classified accord-
ing to (1) their location with respect to the critical
word and (2) whether or not the word present during
that fixation was the one reported as having been
seen. These data are plotted in Figure 1. The circles
indicate the total number of fixations centered at
each letter position. There is a maximum at the 10-
Of the 23 instances in which an erroneous word
was selected, in 22 the selected word shared one
letter with the original. Only once did a subject
choose a word differing in both letters. Thus, even
when a wrong word was reported, it tended to main-
tain information from the original word. In the 12 in-
stances in which a single erroneous word was se-
lected, it shared the first letter with the original word
four times, the fourth letter seven times, and neither
letter only once. Thus, there was no tendency for the
first letter to be maintained more frequently than
the fourth, as might have been expected. Finally,
it should be noted that in every case in which an
error was made (including selecting none
of
the test
words), the subjects' eyes had been centered on the
critical word on at least one fixation. The errors
did not arise from physically skipping over the word.
In general, then, subjects were quite accurate in se-
lecting the word that was in the text in the control
condition, although some errors were made.
Frequency of Selecting Nonpresented
Words
in tbe Experimental Condition
Table 1 gives the frequencies with which subjects
reported having seen words that were never present
on the CRT as they read. While this occurred more
frequently under the experimental condition than
under the control, this difference was small (31 vs.
23) and not statistically significant. Thus, it appears
that these five-letter words were seldom, if ever,
being perceived from subunits acquired on two suc-
cessive fixations. In the experimental condition, as
in the control, most
of
the time the subjects reported
having seen one of the words that was present in the
text as they read.
was being employed in word perception during a
fixation depended less on the location
of
that let-
ter with respect to the fixation location (that is, on
how many letter positions it lay to left or right
of
the center
of
vision), and more on whether the word
containing it was directly fixated. The
data
include
instances in which the fixation was centered on the
fifth letter
of
the word, and even on the space fol-
lowing it, thus placing the first letter
of
the word
four or five letter positions to the left
of
the fixa-
tion point, on the fixation on which the word was
identified.
That
initial letter of the word could not
have been acquired on the prior fixation, because a
different letter had occupied the location on that
fixation. Likewise, there are other instances in which
the eyes were fixated just one or two letter positions
to the left
of
the initial letter
of
the word, and that
letter was apparently not picked up, in spite of prior
evidence suggesting that the perceptual span tends
to be asymmetric to the right (McConkie &Rayner,
1976b; Rayner, Well, &Pollatsek, 1980; Underwood
&McConkie, Note 1).
It
appears, then, that when these readers were
making a rightward series
of
saccades along a line
of text, the region influencing word perception dur-
ing a fixation was defined in terms
of
word units,
rather than in terms of a certain number
of
letters
to right and left
of
the fixation point.
If
a word
was fixated only once (as was typically the case in
these data), that fixation was typically the one that
provided the visual information for the perception
of
that word.
The data from those instances on which a sub-
ject regressed to the critical word or reread the re-
gion containing the critical word after having re-
gressed to an earlier region were also examined.
In these instances, during most of the fixations on
the
word,
the word reported as having been seen
was present. This was typically because
of
the letter-
switching algorithm used. When a subject regressed,
the word was changed, but further changing was dis-
continued. Thus, if the subject made a fixation on the
word, went one fixation beyond, and then returned to
the word for several fixations, the word present on the
fixation following the regression and all further fixa-
tions would be the same as the word
that
was pres-
ent on the initial fixation. In spite
of
this aspect
of the study, there were 53 instances in the experi-
mental condition in which subjects
had
fixations
centered directly on the critical word or on the space
before or after, during which each
of
the two words
was present. An examination of the responses made
following these sentences indicated that a single cor-
rect response was given 36 times, or
68010.
Both
words were reported three times, a correct plus one
or two incorrect words six times, only an incorrect
word three times, and no response six times. Thus,
most
of
the time when a word was refixated after
PERCEIVING WORDS DURING READING 275
it
had
been changed, only one of the words was
reported as having been seen, and the frequency
of
reporting both was less
than
that
of
reporting a cor-
rect word and an erroneous one. While further re-
search is needed to investigate the possibility that
subjects were not always reporting all the words
they actually saw, it appears possible that, when re-
gressions or rereadings are involved, it is not neces-
sarily the case that aword is being identified each
time it is fixated. Such a finding would raise ques-
tions about the function
of
such fixations, but the
present
data
provide no additional information on
this issue.
Effect of Changing Letters on Reading
Several aspects of the data indicated that changing
letters from one fixation to another was producing
an effect on reading. The total number
of
fixations
on the critical word was 387 in the experimental con-
dition and 314 in the control. The mean duration
of
fixations falling in the region from five letters prior
to, to five letters following, the critical word was
266 msec, compared with 254 msec for the control
condition.
In order to provide unambiguous evidence con-
cerning the effect
of
changing letters, the data were
searched for those instances on which the first fix-
ation on the critical word was preceded by a saccade
during which the letters in the critical word were
changed.
Data
from the control condition were se-
lected in an identical manner, since the same al-
gorithm had been used in controlling the display
in that condition. This produced adata set
of
157 in-
stances in the experimental condition and 151 in the
control condition. The first fixation on the word was
labeled fixation
FO,
the saccade following it, sac-
cade
Sl,
and
the next fixation, fixation
Fl.
For
each
subject, the mean
FO
and
Fl
fixation durations
and
the mean
Sl
saccade length were calculated for those
experimental and control sentences which met the
above criteria. Means for the experimental and con-
trol conditions were then calculated by averaging
the subject means. These were compared by t test
for correlated means. The results are presented in
Table 2 and indicate that no detectable effect was
observed on either fixation FO or saccade Sl ,
but
that amarginal effect may have been present on
fixation
Fl.
Further analysis indicated that this latter
effect disappeared after all F1 fixations that fol-
lowed regressive Sl saccades, or those that were cen-
tered on the critical word location (these being the
second fixation on the word,
and
typically being pre-
ceded by a regression), were removed. Finally, as
Table 2 shows, when the F1 fixations were sub-
divided into those that fell to the right
of
the critical
word and those that fell on or to the left
of
it, a
significant effect was found only for the latter.
It
appears, then, that the
Fl
fixation durations only
276 McCONKIE, ZOLA, BLANCHARD, AND WOLVERTON
Table 2
Eye-Movement Data Following Display Change
Prior
to Fixation FO
Condition
Data Experimental Control df
Number of Instances
FOfixation duration
SI forward saccade length
Number of regressive S 1 saccades
Fl
fixation duration
F1 fixation duration when not on critical word
F1 fixation duration when S1 is not regressive
F1 fixation duration when
Fl
is right of the word
F1 fixation duration when F1 is left of or on word
157
283
7.16
26
275
267
261
268
281
151
273
6.96
20
254
264
261
274
214
.79
1.07
1.43
<1
<1
<1
2.88
15
15
<.09
<.00
Note-Fixation
durations are given in milliseconds. Saccade lengths are given in number
of
character positions; 4character positions =
1 deg
of
visual angle.
showed an effect when regressions
and/or
refixations
of the critical word were involved.
It
should be noted
that these were cases in which the word changed again
during saccade
SI,
so the word was different on fix-
ation
Fl
from what it had been on
FO.
It
seems
quite likely that this effect on fixation
Fl
was ac-
tually due to this second change rather than the first.
Thus, there is no evidence that having a different
word present during the fixation prior to that on
which the word was directly fixated had any effect
on processing that word.
It
appears that the change
in the word has an influence only in some instances
when a second fixation is made on the word or
when there is a regression on the following saccade.
Finally, it seemed possible that the display change
influenced reading only when the fixation prior to
the fixation on the critical word location was close
enough to it for visual detail to be resolved. To test
this, the data were split into those instances in which
the fixation prior to fixation
FO
was fewer than five
character positions to the left of the critical word loca-
tion and those in which it was five or more character
positions to the left. Mean
FO
fixation durations for
experimental and control conditions in the first
of
these cases were 287 and 276 msec; in the second
case, they were 278 and 264 msec. Neither
of
these
differences was significant (t <1 and t =1.39, p =.09,
successively), and there was no evidence for the pre-
,dieted interaction.
As reported earlier, in the total data set there were
more fixations in the region of the critical word
in the experimental condition than in the control
(from five character positions before to five after
the word), and these fixations show slightly longer
durations in the experimental condition. Once again,
when these data are partitioned into first-pass fixa-
tions (those taking the eyes further along the line
than they had previously been) and those involving
regressions and rereadings, the difference is found
only in the latter set. Mean fixation durations for
first-pass data are 258 vs. 257 msec for experimental
and control conditions, respectively; corresponding
values for the other set are 283 and 245 msec.
It
seems evident, then, that the effects of changing
words occurred only when the subjects regressed or
reread the text. The remaining question is whether
the display changes themselves induced these regres-
sions and rereadings. Table 2 indicates that in the
data which were selected on having a display change
during the saccade prior to fixation
FO,
there was
very little difference betweenthe frequency of regres-
sions in the experimental and control conditions on
saccade S1. Furthermore, frequency distributions of
the number
of
times that the critical word was fIX-
ated in the experimental and control conditions
showed very little difference in the frequency of
being fixated only once (104 vs. 111, respectively).
The primary difference in these distributions was in
the frequency with which the critical word received
3 or more fixations, which was 44 for the experi-
mental condition (of which 13 instances showed 5-10
fixations) and 21 for the control (of which only 4
showed 5 or 6 fixations, none with more). Thus, it
appears that the changing of the words did not in-
duce the regressions and rereadings, but that, when
a regression occurred, the fact that the word was
sometimes now different tended to induce additional
fixations in the experimental condition. These fixa-
tions tended to be longer in duration than fixations
involved in regressions and rereading in the control
condition.
It
is concluded, then, that there is no evidence
that the changing of the critical word from one fIX-
ation to another was having any detectable effect,
either in producing longer fixations, shorter sac-
cades, or a greater likelihood of regressing. When,
for some reason other than the display change, there
was a regression or the critical word was refixated,
the discrepancy produced by the display change did
sometimes have an effect. Thus, there is no evidence
that information about the first or fourth letters of
the word (or word shape involving these letters) was
acquired from the periphery on one fixation
and
then influenced the processing
of
the
critical word
on the next fixation, when it was brought into
foveal vision.
DISCUSSION
With respect to the hypotheses posed earlier, it
seems clear
that
the perception
of
the words studied
is best described by the word-unit hypothesis. There
was no evidence that subword units were being acquired
on successive fixations
and
integrated into a single
perception, or
that
the visual information being
manipulated in this study was being obtained periph-
erally from aword on one fixation,
and
then facil-
itating its perception on the next. Rather, the critical
words were being perceived during only one fixation,
that on which
the
word was directly fixated,
and
the
changing
of
the letters
had
an effect only if the
reader regressed or refixated the word for some reason.
These results are quite different from those ob-
tained by Rayner (1975)
and
strongly challenge the
notion
that
word perception during one fixation is
facilitated by information obtained peripherally dur-
ing a prior fixation (McConkie &Rayner, 1976a;
Patberg &Yonas, 1978; Rayner, 1978, 1979b; Rayner,
Inhoff, Morrison, Slowiaczek, &Bertera, 1981;
Rayner, McConkie, &Ehrlich, 1978; Rayner,
McConkie, &Zola, 1980; Smith, 1971).
If
informa-
tion
about
the critical words was being acquired
on prior fixations,
that
information must
not
include
letter features, even for the initial letter
of
the word,
or even word-shape information, since word shape
was changed in a number
of
the instances used in
the study.
If
such information was being acquired,
the present study found no evidence
that
it was
being used to facilitate reading.
It
is necessary, then, to consider why it was
that
the results
of
this study seem so
contrary
to prior
theory
and
research. One possibility is
that
there was
something peculiar
about
this study
that
inhibited
the
normal
integrative processes during reading.
This could be due either to the nature
of
the ma-
terials used or to
the
nature
of
the task. With re-
gard to the materials, it should be noted
that
the
critical word location was a relatively unconstrained
word position. This was necessary in order for it
to be capable
of
containing any
of
four different
words, selected only on the basis
of
their having
certain letter similarities.
It
may be
that
the acquisi-
tion
and
use
of
peripheral visual information to facil-
itate later word recognition occurs only when there
exists a higher level
of
contextual constraint (Haber,
1978;
McClelland &O'Regan, 1981;
Paap
&Newsome,
1981). This is a possibility
that
requires further in-
vestigation. On the other hand, the task could be
suspect as well, especially in the present study. Sub-
jects were asked to read short texts, consisting
of
PERCEIVING WORDS DURING READING 277
one to three sentences,
and
then
to select from
among
several visually similar word alternatives.
It
is pos-
sible
that
this could induce a
sort
of
word-by-word
consideration
of
the text
that
is different from
the
processing
that
takes place in more normal reading.
Since all the subjects in this study
had
previously
participated in a similar
but
much larger study in
which they read more
than
300 such texts, they may
have developed apeculiar reading strategy.
To
test
this, four naive subjects were tested in a somewhat
different manner.
After
reading each passage,
the
subject came
off
the bite
bar
and
answered an
oral
question which was designed to require more
than
a
one-word answer
and
would reveal which
of
the
words was perceived during reading. The
data
from
these subjects were very much like those from
the
subjects in the main experiment. In particular, from
texts read in the experimental condition there was
only one instance
(of
a
total
of
44) in which a sub-
ject reported having read aword
that
was never
presented. The control condition produced four
such instances. Once again, there was no evidence
for the subword-unit hypothesis,
and
there was no
other evidence for greater confusion or difficulty
in responding to the words in the experimental condi-
tion. Thus, it seems unlikely
that
the results were
due
to the task employing a test
that
required subjects
to select from among visually similar words.
Another
possible reason for
the
discrepant results
is
that
earlier conclusions may have been in error.
All studies involving eye-movement-contingent dis-
play control which have been taken as evidence for
the existence
of
facilitation
from
prior peripheral in-
formation have involved changing the display in
some
manner
from one fixation to the next. In most,
some stimulus pattern (erroneous letters or a grating)
is present in the visual periphery on one fixation,
but
is removed or relocated on the next fixation, so
that
the
part
of
the text now in the fovea
and
near-fovea
is veridical. When this results in a detectable change
in reading (increasing reading time or causing specific
changes in the eye-movement pattern), it has been
assumed
that
this was due, at least partially, to inter-
ference from peripheral visual information or to its
not providing the normal facilitation
of
later foveal
processing (McConkie, 1979; Rayner, 1979b).
However, it should be
noted
that
these studies
actually provide three separate possible sources
of
difficulty for reading.
The
first is the existence
of
inappropriate stimulus patterns in the periphery dur-
ing a fixation. Underwood
and
McConkie (Note 1)
have specifically explored the effects
of
having er-
roneous letters at different retinal locations, for in-
stance.
The
second is the fact
that
the stimulus pat-
tern on one fixation is somehow different from
what
it was on the prior fixation.
The
third is the fact
that
changes are occurring on the
CRT
which, by them-
selves,
can
have a disrupting effect on reading. While
278
McCONKIE, ZOLA, BLANCHARD, AND WOLVERTON
it has been demonstrated that briefly replacing text
with other text or letter strings during a saccade has
no effect on reading, it is also true that changes oc-
curring during the first 30 msec of a fixation produce
disruptive effects on reading (Wolverton, Note 3).
Thus, it is extremely critical for this type of research
that any display changes occur at a time when the
stimulus movement associated with the change is not
perceived. No study has yet been done which shows
how late in the saccade, or how early in the fixation,
these
changes
can be made without the simple existence
of movement on the CRTs having an effect on read-
ing, and no study has employed an appropriate con-
trol condition for such an influence. Of course, the
claim that peripheral visual information is being ac-
quired and used to facilitate later foveal perceptual
processing is based on the assumption that some or
all of the observed effects of display changes in the
relevant studies have arisen from the second influence,
nonidentical stimulus patterns on
successive
fixations,
rather than from either of the others.
All but the studies which have involved reading
of continuous text have also used text-inappropriate
peripheral visual
patterns-words
spelled backwards
(O'Regan, 1980), other letters substituted for text
original letters (McConkie &Rayner, 1975, 1976b;
Rayner, 1975; Underwood &McConkie, Note 1), or
gratings (Rayner &Bertera,
1979;
Rayner, Inhoff,
Morrison, Slowiaczek, &Bertera, 1981; Rayner &
Pollatsek,
1981).
Thus, it is possible that in all such
studies the changes in reading which were observed
resulted from the presence of inappropriate periph-
eral visual patterns themselves, rather than from the
existence of change in the pattern from one fixation
to the next.
The two studies in which this was not the case,
other than the present one, were by Rayner (1975)
and McConkie (Note 4). In the Rayner study, there
was one condition (condition W-SL) in which the
content of one word location was changed from one
word to a second during the saccade which took the
eyes to that word. Both words were appropriate in
the context, and they had the same initial and final
letters and the same general word shape. The present
study would seem to have produced a potentially
more noticeable change, since the'initial letter was
changed and in many of the instances the word shape
was changed as well. Still, Rayner found an effect
of his manipulation, an increase of about 20 msec
in the immediately following fixation duration,
whereas the present study found no effect. It is pos-
sible that the effect in the Rayner study was actually
due to the third type of influence, detection of move-
ment in the display associated with the making
of
the change. Several analyses were conducted to con-
sider this possibility at the time the study was done
(Rayner, 1974), and all but one yielded nonsignifi-
cant results. However, a reconsideration of the data
presented indicates that, while the differences were
not large, most were in the direction suggested by
the hypothesis that changes were indeed being seen.
For instance, when fixations were divided into those
which occurred on the letter following the boundary
that triggered the display change (and thus had the
greatest possibility for the change's occurring after
the saccade was completed) vs. fixations further from
the boundary (which hence provided greater saccade
time after a change was called for), only one of five
tests was significant, but four of the five showed
longer fixations in the first case than in the second.
The average difference between the means of these
five comparisons was greater than the 20-msec dif-
ference found for the experimental manipulation in
the study. In contrast to this study, in which display
changes could be triggered late in the saccade and
thus might not be complete until the eyes were actually
in fixation, both the present study and a previous
one (McConkie, Note 4) made changes only early
in the saccade. McConkie (Note 4) caused a single
letter to change during each saccade, alternating
the
contents of a critical location between two words,
both of which were contextually appropriate. In the
present study, two letters were changed. In both
studies, the change occurred about 8 msec after the
onset of the saccade, and since it required only 3 msec,
it was completed long before the end
of
the saccade,
which lasted at least 20 msec. Neither of these studies
provided any evidence that changing a word from
one fixation to the next had any effect on reading
unless the reader later reread the word. Thus, it may
be that when effects of changing words or letters
from one fixation to the next have been observed in
the reading
of
continuous text, these were due to the
presence
of
erroneous or inappropriate peripheral
patterns during fixations, or to perceived movement
of the text when changes occurred late in saccades
or early in fixations, rather than to the mismatch of
visual patterns from one fixation to the next.
The present results seem particularly damaging to
two proposals that have been made concerning the
use
of
peripheral information. Bouma
(1978)
sug-
gested that information from the periphery was
transmitted to the brain much more slowly than
foveal information, and that in a series
of
fixations
during reading the peripherally obtained information
from a word and the later foveally obtained informa-
tion from the same word arrived at appropriate brain
centers at about the same time, thus supporting one
another and facilitating the perception
of
that word.
Smith (1971) proposed that the region within which
words are read was sufficiently large for the same
word to be actually read on several fixations, thus
facilitating processing and providing redundancy
which would inhibit misreadings. It seems clear that
neither of these proposals describes perception of the
words studied in the present experiment. Rather, the
evidenceindicates that the words were read with in-
formation obtained during only one fixation.
Even if peripheral information is not being used to
facilitate later foveal perception, this is not to say
that peripheral information is not useful during read-
ing. Other studies indicate that sometimes words not
directly fixated are still being read (Kliegl, Olson,
&Davidson, in press; Hogaboam, Note 5) and that
lengths of words in the periphery, quite apart from
their other characteristics, can influence where the
eyes are sent on a saccade (O'Regan, 1980). How-
ever, these are characteristics of the stimulus that are
available peripherally from the present fixation, and
do not require integration across fixations.
If
some
type of information is being carried across fixations
in reading, its nature is not presently apparent. The
carry-over of strictly visual aspects of letters and
word shapes does not seem to occur (McConkie &
Zola, 1979), and the present study seems to eliminate
the carry-over of specific letters or semantic informa-
tion. The fact that changing words has no effect
eliminates the possibility of semantic priming based
on peripheral patterns (Inhoff &Rayner, 1980). In
summary, then, these results argue that reading is
based on available retinal information rather than on
patterns perceived during prior fixations. Whether
good readers are more adept at using available pe-
ripheral information (Fisher, 1976; Patberg &Yonas,
1978) or not (Underwood, Note 6) is still a matter
requiring investigation.
The results
of
this study also argue that the region
within which visual information is used for word
identification during a fixation is defined in terms of
word units, not a specific number of letters to left
and right
of
the fixation (McConkie, in press). While
there has been prior evidencethat this is so to the left
of the fixation point (Rayner, Well, &Pollatsek,
1980), the same is not true for the region to the right
(Underwood &McConkie, Note 1). For instance,
these last investigators found that when letters be-
yond the third to the right of the fixation point were
replaced by other letters, the disruptive effect was
just as great when this manipulation did not cause a
change in the fixated word as when it did. The same
was true when all letters beyond the fifth to the right
were replaced. Thus, there was no evidence that
errors were not being perceived when they lay outside
the fixated word, or that this occurred less frequently
or had less of an effect. Combining these results with
the results of the present study raises an interesting
question. Is it possible that during a fixation the
reader responds to orthographic irregularities in a
word that is not identified on the fixation, and yet,
at the same time, does not use information perceived
from such a word when it is brought into the fovea
on the next fixation? Since the present data are
insufficient to answer this question, it must remain
a topic for future investigation.
If
the answer is
PERCEIVING
WORDS
DURING
READING
279
positive, this will be strong evidence that ortho-
graphic structure is perceived directly in reading,
rather than becoming important only in lexicalaccess
(Underwood &McConkie, Note 1).
Finally, the results from the present study bear
directly on the issue of how best to derive a measure
of processing time in reading from eye-movement
data. One basic question has been whether percep-
tion during a fixation should be considered to be in
letter or letter-group units (McConkie, Hogaboam,
Wolverton, Zola, &Lucas, 1979) or word units
(Just &Carpenter, 1980; Hogaboam &McConkie,
Note 7). The present results clearly favor the latter
and are in harmony with Just and Carpenter's "eye-
mind assumption," although other
evidence
indicates
that words other than the word fixated are at times
read during a fixation (Kliegl, Olson, &Davidson,
1982; Hogaboam &McConkie, Note 7). Appar-
ently, at least when a person is reading carefully,
five-letter words which are relatively unconstrained
are read only when directly fixated.
REFERENCE NOTES
1. Underwood, N. R., & McConkie, G. W. The effect
of
en-
countering errors at different retinal locations during reading.
Unpublished manuscript, Center for the Study of Reading, Uni-
versity of Illinois, 1982.
2. Zola, D. The effect
of
redundancy on the perception
of
words in reading (Tech. Rep. 216). Urbana: Center for the Study
of Reading, University of Illinois, 1981.
3. Wolverton, G. S. The acquisition
of
visual information
during fixations and saccades in reading. Paper presented at the
annual meeting of the American Educational Research Associa-
tion, San Francisco, April 1979.
4. McConkie, G. W. Where do we read? Paper presented at the
annual meeting of the Psychonomic Society, San Antonio, Tex.,
November 1978.
5. Hogaboam, T. W. Time on text: The use
of
eye movement
records in assigning processing time to text. Paper presented at the
annual meeting of the American Educational Research Associa-
tion, San Francisco, April 1979.
6. Underwood, N. R. The span
of
letter recognition
of
good and
poor readers (Tech. Rep. 251). Urbana: Center for the Study of
Reading, University of Illinois, 1982.
7. Hogaboam, T. W., &McConkie, G. W. The rocky road from
eye fixations to comprehension (Tech. Rep. 207). Urbana: Center
for the Study of Reading, University of Illinois, 1981.
REFERENCES
BOUMA,
H. Visual search and reading: Eye movements and func-
tional visual field: A tutorial review. In J. Requin (Ed.), Attention
and performance VII. Hillsdale, N.J: Erlbaum, 1978.
FISHER,
D. F. Spatial factors in reading and search: The case
for space. In R. A. Monty &J. W. Senders (Eds.), Eye move-
ments and psychological processes. Hillsdale, N.J: Erlbaum,
1976.
HABER,
R. N. Visual perception. Annual Review
of
Psychology,
1978,19,31·59.
IN
HOFF,
A. W., &
RAYNER,
K. Parafoveal word perception: A
case against semantic preprocessing. Perception
.I
Psychophysics,
1980,17,457-464.
JUST,
M. A., &
CARPENTER,
P. A. A theory of reading: From eye
280
McCONKIE, ZOLA, BLANCHARD, AND WOLVERTON
fixations to comprehension. Psychological Review, 1980, 87,
329-354.
KLIEGL, R.,
OLSON,
R. K., &
DAVIDSON,
B. J. Regression anal-
ysis as a tool for studying reading processes:
Comment
on
Just
and Carpenter's eye fixation theory. Memory &Cognition,
1982,10,287-296.
KLIEGL, R.,
OLSON,
R. K., &
DAVIDSON,
B. J. On problems of
unconfounding perceptual
and
language processes. In K. Rayner
(Bd.), Eye movements in reading: Perceptual and language pro-
cesses.New York: Academic Press, in press.
McCLELLAND,
J. L., &O'REGAN, J. K. Expectations increase the
benefit derived from parafoveal visual information in reading
words aloud. Journal
of
Experimental Psychology: Human
Perception and Performance, 1981,7,634-644.
MCCONKIE,
G. W. On the role
and
control of eye movements in
reading. In P. A. Kolers, M. E. Wrolstad, &H.
Bouma
(Eds.),
Processing
of
visible language. New York: Plenum Press, 1979.
McCONKIE,
G. W. Eye movements and perception during reading.
In K. Rayner (Ed.), Eye movements in reading: Perceptual
and language processes. New York: Academic Press, in press.
MCCONKIE,
G. W.,
HOGABOAM,
T. W.,
WOLVERTON,
G. S.,
ZOLA,
D., &
LUCAS,
P. A. Toward the use of eye movements
in the study of language processing. Discourse Processes, 1979,
2,1S7-177.
McCONKIE, G. W., &
RAYNER,
K. The span of the effective
stimulus during afixation in reading. Perception &Psycho-
physics, 1975, 17, 578-586.
MCCONKIE,
G. W., &
RAYNER,
K. Identifying the
span
of
the
effective stimulus in reading: Literature review
and
theories of
reading. In H. Singer & R. B. Ruddell (Eds.), Theoretical models
and processes
of
reading. Newark, Del: International Reading
Association, 1976. (a)
McCONKIE, G.
W.,
&
RAYNER,
K. Asymmetry of
the
perceptual
span
in reading. Bulletin
of
the Psychonomic Society, 1976,
8, 365-368. (b)
McCONKIE, G. W., &
ZOLA,
D. Is visual information integrated
across successive fixations in reading? Perception &Psycho-
physics, .1979,25,221-224.
O'REGAN, K. The control of saccade size and fixation
duration
in reading: The limits of linguistic control. Perception &Psy-
chophysics, 1980,28,112-117.
O'REGAN, K. The
"convenient
viewing position" hypothesis. In
D. F. Fisher, R. A. Monty, &J. W. Senders (Eds.), Eye move-
ments: Cognition and visualperception. Hillsdale,
N.J:
Erlbaum,
1981.
PAAP, K. R., &
NEWSOME,
S. L. Parafoveal
information
is
not.
sufficient to produce semantic or visual priming. Perception &
Psychophysics, 1981,29,457-466.
PATBERG,
J. P., &
YONAS,
A. The effects of the reader's skill
and
the difficulty
of
the text on the perceptual
span
in reading.
Journal
of
Experimental Psychology: Human Perception and
Performance, 1978,4,545-552.
RAYNER,
K. The perceptual span and peripheral cues in reading.
Unpublished dissertation, Cornell University, 1974.
RAYNER,
K. The perceptual
span
and
peripheral cues in reading.
Cognitive Psychology, 1975,7,65-81.
RAYNER,
K. Foveal
and
parafoveal cues in reading. In J. Requin
(Ed.), Attention and performance VII. Hillsdale,
N.J:
Erlbaum,
1978.
RAYNER,
K. Eye guidance in reading: Fixation locations within
words. Perception,
1979,8,21·30.
(a)
RAYNER,
K. Eye movements in reading: Eye guidance
and
integra-
tion. In P. A. Kolers, M. E. Wrolstad, &H.
Bouma
(Eds.),
Processing
of
visible language. New York:
Plenum
Press,
1979. (b)
RAYNER,
K., &
BERTERA,
J. H. Reading without a fovea. Science,
1979,206,468-469.
RAYNER,
K.,
INHOFF,
A. W.,
MORRISON,
R.
E.,
SLOWIACZEK,
M. L., &
BERTERA,
J. H. Masking of foveal
and
parafoveal
vision during eye fixations in reading. Journal
of
Experimental
Psychology: Human Perception and Performance, 1981, 7,
167-179.
RAYNER,
K.,
MCCONKIE,
G. W., &
EHRLICH,
S. Eye movements
and
integrating information across fixations. Journal
of
Experi-
mental Psychology: Human Perception and Performance, 1978,
4,529-544.
RAYNER,
K.,
MCCONKIE,
G. W., &
ZOLA,
D. Integrating in-
formation across eye movements. Cognitive Psychology, 1980,
12,206-226.
RAYNER,
K., &
POLLATSEK,
A. Eye movement control during
reading: Evidence for direct control. Quarterly Journal
of
Ex-
perimental Psychology, 1981,33,351-373.
RAYNER,
K., WELL, A. D., &
POLLATSEK,
A. Asymmetry of the
effective visual field in reading. Perception &Psychophysics,
1980,27,537-544.
SMITH,
F. Understanding reading. New York:
Holt,
Rinehart &
Winston, 1971.
APPENDIX
1. As Kevin approached his grandfather's bungalo
down the long country lane, he was taken aback by its
/weedy/weepy/seedy/seepy/ appearance and its need for a
coat of paint.
2. Ruth's great aunt is definitely the most /mushy/musty/
gushy/gusty/ person she has ever met. .
3. Dr. Koppof was able to demonstrate that the
/blare/
blame/flare/flame/ which had been so disconcerting to the
natives of the region originated from a neighboring tribe.
4. With considerable hesitation, Phil asked Sue to /trade/
trace/grade/grace/ his table, not knowing what the out-
come might be.
5. English teachers typically believe that formal writing
has
the ability to clean out the /musty/musky/dusty/dusky/
reachesof the mind and that it contributes to clear thinking.
6. The second grader was doing very well in the spelling
bee until he was given the word
'/couch/cough/touch/
tough/', which he missed by a single letter.
7. Mr. Gilmore has written a book in which he classifies
/peaks/pears/beaks/bears/ from around the world. How-
ever, it will never be a best seller.
8. They have interesting /meals/meats/seals/seats/ at
the Emporium. You really should make a visit there the
next time you are in the city.
9. While the old herder shivered in front of his pot-
bellied stove, he was led to contemplate the results of his
actions. While having his /goats/goals/coats/coals/ high
seemed like a good idea at the time, he now realized the
disadvantages.
10. Mr. Blackwell was, in order, an industrialist, a run-
ner, a collector,
and
a hypochondriac. The many /pills/
piles/mills/miles/ he had accumulated over the years were
often the topic of conversation among his friends.
11. The Recreation Housing Committee reported that
five /teams/tears/beams/bears/ were recently discovered
that we previously did not know about. Boyd has been
assigned to seek more infbrmation and to report back to
the Committee. .
12. The high point of the Great Zaslow's act is when he
/Ieads/leaps/heads/heaps/ all twenty tigers toward the gate
of the cage at once.
13.
It
has been whispered that deaf Talu /beats/bears/
heats/hears/ his chickens in the large bamboo cage.
14. In this particular form of treatment, the physician
/seats/seals/heats/heals/ the injured bone directly in the
socket.
15. Rosslyn decided that her husband's music and his
/feats/fears/beats/bears/ were no longer tolerable and it
was finally time for her to leave.
16. The children rang the school bell as the old teacher
shopped at the fruit market next door. "Although I love
the /dears/deals/pears/peals/," she said mournfully,
"I
can't take any more of them for a few days."
17. Everyone knows about the /raids/rails/maids/mails/
in my country, but the government is unable to do anything
about them.
18.
It
has been said that a highly trained dog /heels/
heeds/feels/feeds/ well only at the call of its master.
19. The old peddler had so many /warts/wares/carts/
cares/ that it was hard for him to exert enough energy to
make his way along the streets of the city.
PERCEIVING WORDS DURING READING 281
20. After a careful examination of the Frenchman's
/works/words/corks/cords/ the judges declared him to
be the winner.
21. The officer inspected the /belts/bells/welts/wells/
before having his supper.
22. After hours of searching for them, the old
/deans/
dears/beans/bears/ were found in the park.
23. As the moon slid behind the clouds, the
/yards/
yarns/bards/barns/ had a strange effect on him.
24. The anthropologist, Dr. Barter, claimed that, unlike
the compassionate and helpful Samoans, an Eskimo fisher-
man never /baits/bails/waits/wails/ for anyone, even his
closest friends.
25. Yesterday afternoon Mr. Johns showed his
/warts/
wares/harts/hares/ to his friends.
(Manuscript received February 10, 1982;
accepted for publication
June
11, 1982.)
Article
• Current models of fluent reading often assume that fast and automatic word recognition involves the use of supraletter feature corresponding to the envelope or shape of the word when it is printed in lowercase. The advantages of mixed case over pure case and of pure lowercase over pure uppercase have often been taken as evidence favoring the word-shape hypothesis. Alternative explanations for these phenomena are offered. Exp I, with 400 undergraduates, showed that previous demonstrations of word-shape effects during proofreading are better described as individual letter effects. Exps II–IV, with 144 Ss, explored the possibility that word shape facilitates lexical access through uncertainty reduction. In all 3 experiments, performance on words with rare shapes was compared to those with common shapes. There were no effects of shape frequency in either tachistoscopic recognition or lexical-decision tasks. This was true regardless of the degree to which the visual shape cue was supplemented by the nonvisual factors of familiarity and expectancy. Possible reasons why fluent readers ignore word shape are discussed within the framework of a model that assumes that automatic word recognition is mediated by the activation of abstract letter identities. (48 ref) (PsycINFO Database Record (c) 2012 APA, all rights reserved) • Current models of fluent reading often assume that fast and automatic word recognition involves the use of supraletter feature corresponding to the envelope or shape of the word when it is printed in lowercase. The advantages of mixed case over pure case and of pure lowercase over pure uppercase have often been taken as evidence favoring the word-shape hypothesis. Alternative explanations for these phenomena are offered. Exp I, with 400 undergraduates, showed that previous demonstrations of word-shape effects during proofreading are better described as individual letter effects. Exps II–IV, with 144 Ss, explored the possibility that word shape facilitates lexical access through uncertainty reduction. In all 3 experiments, performance on words with rare shapes was compared to those with common shapes. There were no effects of shape frequency in either tachistoscopic recognition or lexical-decision tasks. This was true regardless of the degree to which the visual shape cue was supplemented by the nonvisual factors of familiarity and expectancy. Possible reasons why fluent readers ignore word shape are discussed within the framework of a model that assumes that automatic word recognition is mediated by the activation of abstract letter identities. (48 ref) (PsycINFO Database Record (c) 2012 APA, all rights reserved)
Article
This chapter discusses the role of eye movements in reading. It describes some limits of the eye-mind assumption. The eye-attention assumption holds that, by recording the movements of the eye, an external observer may have access to the current contents of conscious processing. With stationary eyes, observers can attend a few degrees away from fixation, to make use of advance information about the location of target events. With saccadic eye-movements permitted, attention can be shown to move ahead of the eyes. These demonstrations of eye-movements and independent attention-movements across a two-dimensional space are further qualified when the dimension of depth is added to the display. Under very specific conditions attention can be switched from one scene to another without any ocular indication of a movement. For the eye-mind assumption to hold, the evidence would need to demonstrate that all processing of an object is completed during its inspection. In addition, the eye-mind assumption accounts for a large proportion of the variance in fixation durations.
Article
Full-text available
This chapter reviews historical evidence for change blindness emphasizing both the historical precursors and the innovations of more recent approaches. The recent focus on change blindness comes against a backdrop of decades of empirical research on change detection, priming, and saccadic integration. Change blindness is the finding that observers often fail to notice large changes to objects or scenes when the change coincides with a visual disruption. Regardless of the true explanation for change blindness (CBB), the phenomenon shows that people misunderstand important aspects of vision and visual memory, and that these misunderstandings can lead them astray when predicting their own performance and the performance of others. Mismatches between intuitions and performance are often precisely those for which psychological science is most warranted and relevant. Within psychology, such mismatches serve as an informal guide to help decide what represents an interesting problem for research. Relying too much on the assumption that changes draw attention, however, might lead drivers to focus on other activities such as talking on a phone. In addition to the early empirical literature on change detection, several other fields have considered failures of change detection. Perhaps the clear example comes from the development and analysis of the art of filmmaking.
Article
The abstract for this document is available on CSA Illumina.To view the Abstract, click the Abstract button above the document title.
Article
Skilled readers read passages that were displayed on a Cathode Ray Tube controlled by a computer. The readers' eye movements were monitored and certain critical words were changed by the computer as the eye was in motion. The experimental technique utilized in the study provided data on how wide the area is from which a reader acquires information during a fixation in silent reading. The results also delineate different types of visual information that are acquired from various areas within the perceptual span. It was found that a reader was able to make a semantic interpretation of a word that began 1–6 character spaces from his fixation point. When he fixated 7–12 character spaces prior to a word, he was able to pick up such gross visual characteristics as word shape and initial and final letters. It was concluded that the skilled reader is able to take advantage of information in the periphery. However, the size of the area from which he does is rather small.
Article
Three problems in the use of eye movement data for the study of language processing are discussed in this report: the perceptual span problem, the data summary problem, and the eye-mind lag problem. Recent research on perception during reading that bears on these problems is also described. Finally, a general approach to the use of eye movement data for studying language processing, based on present knowledge of perceptual processing and eye movement control during reading, is presented. (Author/FL)
Article
As six subjects read a history text, their eye movements/fixations were monitored and recorded. A processing profile was created for each subject, and the correlations between pairs of profiles were computed. The correlations were all positive, but quite small. Although the differences between correlations were not large enough to indicate a definite rank order, one generalization emerged: the higher correlations were associated with processing spans shorter than or equal to the saccade length. This finding tends to support the idea that the perceptual span extends from one fixation point rightword to the point of the next fixation. In the production of a single profile that represented the processing load at various points in the text, it was observed that less attention/fixation was given to the definite article "the," and that increased processing times occurred for dates and numbers in the text. Since many issues concerning the reading process are critically dependent upon notions of processing load, the construction and analysis of processing profiles from eye movement records can be quite useful to reading researchers and theorists. (RL)
Article
To identify when visual information is acquired during reading, an experiment was designed in which rapid display changes were made contingent upon the state of the eye. This was accomplished by programing a computer to detect within a few milliseconds whether the eye is in a saccade or fixation. By then replacing a line of text with some other line for a 30 millisecond period during the saccade or at different times during the fixation it was hoped that it could be determined when the reader was sensitive to visual input from the text. The subjects were college undergraduates. The results showed that (1) little, if any, visual information is acquired during the saccade; (2) saccadic threshold elevation does not substantially reduce the acquisition of visual information during the early portion of the fixation; (3) visual information is being acquired throughout the fixation; (4) the kind of visual information being acquired may change over the course of the fixation; and (5) eye movements are under rather immediate control, responding to stimulus manipulations within the fixation where the manipulation occurs. (MKM)
Article
Used an on-line computer technique to determine whether 3 skilled readers (high school students) acquired visual information equally far to the left and right of central vision during fixations in reading. None of the Ss appeared to use visual information more than 4 character positions to the left of the fixation point (smaller distances were not tested), though all of them acquired visual information substantially further than that to the right. Thus, the region of useful visual information in reading is asymmetric around the fixation point. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
Article
Conducted 2 experiments in which 18 undergraduates named visually presented words preceded by a "preview" stimulus presented in parafoveal vision. When there was no constraint on which word would be shown, the preview had little effect on the time required to name the foveal presentation of the target word. However, when the target was chosen from a constrained set of 8 words seen repeatedly by S, the preview produced a large naming facilitation. Likewise, when constraints were generated by preceding context, the preview again produced a large naming facilitation. A weak context, which by itself was insufficient to facilitate performance, nevertheless was sufficient to increase the size of the preview-benefit effect. Results suggest that Ss were able to take 2 sources of information, each of which by itself is insufficient to facilitate performance, and combine them to derive a substantial benefit. (11 ref) (PsycINFO Database Record (c) 2012 APA, all rights reserved)