Upload
yuvharpaz
View
213
Download
0
Embed Size (px)
Citation preview
8/14/2019 kahana faust 2002
1/10
Neuropsychologia 40 (2002) 892901
Priming summation in the cerebral hemispheres: evidence fromsemantically convergent and semantically divergent primes
Miriam Faust, Allon KahanaDepartment of Psychology, Bar-Ilan University, Ramat Gan 52900, Israel
Received 31 January 2000; received in revised form 19 January 2001; accepted 31 July 2001
Abstract
Theability to activate andto maintain a large andrelatively undifferentiated semantic field has been thought to be an important component
of lexical semantic processing by the right hemisphere (RH). An implication of this unique propensity of the RH was examined in the
present study that included two divided visual field priming experiments with SOAs of 800 and 2500ms. The experiments investigated the
ability of the RH and the left hemisphere (LH) to summate activation from multiple primes followed by a laterally presented ambiguous
target word. The priming words either converged onto the same semantic representation (i.e. all three words related to either the dominant or
to the subordinate meaning of the target) or diverged onto distinct semantic representations (i.e. two words related to the dominant and one
to the subordinate meaning of the target, or vice versa). Resultsindicated that for either an 800 or 2500 ms stimulus onset asynchrony (SOA)
the LH benefited most from three semantically convergent primes that converged onto the dominant meaning of the ambiguous target word.
There was no facilitation when three subordinate primes preceded the target. When the primes diverged onto different meanings, there
was significant facilitation for the 800 ms SOA only. In contrast, with an 800 ms SOA, the RH benefited only from semantically divergent
primes, that diverged onto alternate meanings of the ambiguous target word. With a 2500 ms SOA, the RH benefited from all combinations
of primes. The discussion focuses on the implications for language processing of the differences between the two hemispheres in the scope
and temporal pattern of the multiple prime effect. 2002 Elsevier Science Ltd. All rights reserved.
Keywords: Hemispheric asymmetry; Visual fields; Multiple priming; Lexical ambiguity
1. Introduction
Although the superiority of the left hemisphere (LH) for
language processing is indisputable, it is now clear that word
meaning access is bilateral. Data from normal brain and from
patients with injury to the RH imply that each hemisphere
processes the meaning of words, although not necessarily in
the same way. The differences between the two hemispheres
in word meaning activation may result in the right hemi-
sphere (RH) having a great deal of influence on language
processing (for reviews see [6,8,13,15,16]). According to the
RH coarse semantic coding theory [6,9], after encountering
a word, the LH engages in relatively fine semantic coding,
strongly focusing activation on a single interpretation of a
word and a few close or contextually appropriate associates,
whereas the RH engages in coarse semantic coding, weakly
and diffusely activating alternative meanings and more dis-
tant associates. The present study examined an important
implication of the coarse semantic coding by the RH: the
Corresponding author. Tel.: +972-3-5318547; fax: +972-3-5350267.
E-mail address: [email protected] (M. Faust).
ability to summate activation from multiple primes that di-
verge onto distinct, unrelated semantic representations (e.g.
money-river-vault as priming stimuli for the target word
bank).
Research on language processing in the two cerebral
hemispheres relies heavily on semantic priming methods,
that allow for the evaluation of RH language processing
independent of the overall superiority of the LH for word
recognition. Taken as a whole, the results of priming studies
reveal qualitatively distinct patterns of priming in the two
hemispheres, suggesting that different semantic information
is most accessible within each hemisphere at different mo-
ments during language processing. Two factors determine
hemispheric differences in semantic access and retrieval:
the nature of semantic relations between words and the time
course of meaning activation [16]. While the LH maintains
a relatively small semantic field, including only closely re-
lated meanings and a single interpretation, the RH weakly
activates a much broader range of related meanings, in-
cluding peripheral and unusual meanings (e.g. metaphoric
interpretations, multiple meanings of ambiguous words, see
[1,2,14]). Furthermore, the range of meaning activation in
0028-3932/02/$ see front matter 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 2 8 -3 9 3 2 (0 1 ) 0 0 1 7 4 - 9
8/14/2019 kahana faust 2002
2/10
M. Faust, A. Kahana/ Neuropsychologia 40 (2002) 892901 893
the RH is not just wider, but also less discriminant. While in
the LH, strongly related meanings are activated much more
than weakly related meanings, in the RH weakly related
meanings are also highly activated. Thus, the RH obtains
equal priming from high and low dominant category primes
[19], from a single strongly related associate as from a set
of three weak associates [9], from words related via twosemantic relations as from words related via only one [18],
from contextually relevant as from contextually nonrelevant
meanings of ambiguous words [23,25] and from the domi-
nant and subordinate meanings of ambiguous words [2,16].
The differences between the hemispheres in meaning
availability are usually more pronounced for long stim-
ulus onset asynchronies (SOAs), when strategic and/or
post-access prime processing and inhibition for unrelated
words may only occur within the LH. When relatively long
SOAs were used, larger priming effects in the RH were
found for subordinate meanings of ambiguous words at
750 ms SOA [14], for metaphoric meanings of words at
800 ms SOA [1], for contextually nonrelevant meanings ofambiguous words at 900 ms SOA [23] and 1000 ms SOA
[25], for several weakly related prime words at 700 ms SOA
[9] and for several prime words forming a compound with
the target word at 2000, 7000, 15 000 ms SOA [10].
These results suggest that while LH fine semantic coding
[9] has a clear advantage for most linguistic processes, RH
coarse semantic coding is critical for mediating those aspects
of comprehension which require the simultaneous consider-
ation of more than one plausible meaning, or the sustained
activation of a wide range of word meanings. This unique
ability of the intact RH could account for the difficulties that
RH damaged persons have in appreciating jokes, metaphors,connotations, idioms, sarcasm and indirect requests, which
may depend on the continued activation of multiple mean-
ings [1113]. Thus, these patients seem to behave as if their
intact LH is operating without the semantic support of the
damaged RH [15].
To sum up, the results of semantic priming studies as
well as data collected from brain damaged patients converge
on the idea that the RH contributes to language processing
mainly by subserving widespread activation of word mean-
ings, without subsequent selection. The undifferentiated ac-
tivation of alternative, and sometimes contradictory, inter-
pretations for some indefinite period [14,15,26] may have
significant implications, that have not been explicitly tested
in past research, for the ability of the RH to benefit from
semantically divergent information.
These differences between LH focusing on the most
salient or dominant meaning of a word versus RH main-
taining multiple meanings of words, are relevant to the
processing of both single words and words appearing in a
sentence context [14,23,25]. For every word, even if it is
seemingly unambiguous, there is a large body of divergent
semantic information associated with it. The components of
this semantic information, i.e. the semantic features of the
word, include the definitive properties as well as associa-
tions of the word [6]. Only a subset of all possible semantic
features of a word is relevant in a given context, or, most
salient when there is no context. This is particularly true for
ambiguous words, which may have meaning subsets that
are semantically incompatible with each other. Semantic
processing in the LH may consist of initially activating all
semantic information related to an ambiguous word, thenselecting the dominant or contextually relevant meaning
while actively inhibiting the less frequent or contextually
nonrelevant semantic features [1,14,23,25,31].
In contrast to single words, the selected meaning of
words appearing in sentences is aided by the syntax and
message-level semantics of the sentence. Consequently
only the contextually relevant meanings of the words are
maintained [30]. Earlier studies have shown that the syntac-
tic and message-level semantic components of a sentence
aid in facilitating either the dominant or the subordinate
meaning of target words presented to the RVF/LH even at
relatively longer SOAs [23,25]. For single word primes, at
longer SOAs, no facilitation was found for the subordinatemeaning of target words presented to the RVF/LH [14]. No
systematic studies have been performed to date, however,
on the effect of lists of several single word primes related to
the subordinate meaning of an ambiguous target presented
to the RVF/LH. Thus, one of the questions addressed by a
multiple prime paradigm is how will a list of multiple word
primes related to the subordinate meaning of the target af-
fect target recognition? Will a multiple word list facilitate
an RVF target even at longer SOAs in a manner similar to
that of words embedded in a sentence, or will the LH show
no facilitation for the subordinate meanings at the longer
SOAs in a manner similar to that of single word primesthat are not organized into sentences? For example, primes
consisting of lists of several single words weakly related to
an RVF target, show less facilitation than a strongly related
single word prime [9]. Multiple weak primes, however, do
show strong facilitation for RVF targets when presented in
a sentence structure [22,24]. Consequently, when the SOA
is relatively long, the LH should definitely benefit from
multiple primes that converge on the dominant meaning of
an ambiguous target word, and thus multiply activate the
same subset of dominant semantic features (for discussions
of the effects of multiple primes, see [14,20]). However,
when multiple primes converge on the subordinate meaning
of an ambiguous word, priming may or may not be found
in the LH for relatively long SOAs, depending on whether
the primes are treated as a list of single words or provide
sufficient context for the LH to narrow activation to the sub-
ordinate meaning in a manner similar to that of a sentence
prime.
With regard to lists of divergent primes, the LH may be at a
relative disadvantage. Because of the lack of featural similar-
ity between the different unrelated meanings of an ambigu-
ous word and of the active inhibition of alternative meanings,
the LH may be relatively unable to benefit from semanti-
cally divergent primes that require the sustained availability
8/14/2019 kahana faust 2002
3/10
894 M. Faust, A. Kahana/ Neuropsychologia 40 (2002) 892901
of discrete and separable subsets of semantic features, in-
cluding less frequent and incompatible features. Thus, while
the stronger activation of dominant word meanings or se-
mantic features is conducive to most language tasks, the LH
may be at a relative disadvantage when the recognition of an
ambiguous target word depends on the summation of acti-
vation from multiple primes that diverge onto two different,incompatible meanings of an ambiguous word.
In contrast, the RH diffusely activates and maintains large
semantic fields, containing many distantly related semantic
features [1,14,15,17]. The larger, and less discriminant, the
semantic field, the more likely it is to overlap with semantic
fields activated by other input words. Thus, in the RH, the
constellation of features instantiated by one meaning of the
ambiguous word could overlap, at least partly, with that of
the other meanings. Due to the sustained and undifferentiated
activation of alternate, and even incompatible [2], meanings
in this hemisphere, the RH might benefit almost to the same
degree from multiple primes that converge on either the
dominant or subordinate meanings of ambiguous words aswell as from multiple primes that diverge onto two different,
unrelated meanings of ambiguous words.
The current study was conducted to investigate what ef-
fect multiple primes that either converge on the dominant
or subordinate meaning or diverge on both dominant and
subordinate meanings may have on the activation of lexi-
cally ambiguous words presented to the right visual field
(RVF)/LH and the left visual field (LVF)/RH. An earlier
study [5] examined the combinatorial influence of two word
convergent and divergent primes on lexical decision to tar-
gets presented to the central visual field after a relatively
short SOA of 266 ms. The target words were either ambigu-ous or unambiguous for the divergent and convergent con-
ditions, respectively, and the pattern of priming reported for
both convergent and divergent primes was additive. How-
ever, the experimental conditions of that study, i.e. very short
SOAs and central visual field stimulation of primes and tar-
gets do not allow us to conclude which hemisphere and form
of processing was involved. In the present study, a lexical
decision-priming paradigm, which allowed examination of
the relative activation of different sets of meanings in the two
hemispheres, was used. We predicted that at relatively long
SOAs, RVF/LH ambiguous target words following multiple
primes related to the dominant meaning would be facilitated.
The question remains open as to whether multiple primes
that converge on the subordinate meaning will facilitate tar-
get words presented to the RVF. Due to the presence of one
or two primes related to their dominant meaning, ambiguous
RVF target words following semantically divergent multiple
primes should be facilitated, but, perhaps, to a lesser extent
than in the case of three dominant primes. In contrast, when
ambiguous target words are presented to the left visual field
(LVF)/RH, all types of priming word triads should facilitate
target word recognition, although according to Beemans
theory greater priming is still predicted in the two conver-
gent conditions than in the two divergent conditions [6,9].
2. Experiment 1
2.1. Methods
2.1.1. Participants
The participants were 20 Bar-Ilan University students,
native Hebrew speakers, aged 2025 years and strongly righthanded (laterality quotient of at least +90 on the Edinburgh
Inventory) [28]. All had normal or corrected to normal vision
and volunteered to participate in the experiment.
2.1.2. Materials
The stimulus pool consisted of priming words and target
word and nonword items, all in Hebrew. The target stimuli
were two-to-four letter words and nonwords. The 75 target
words were ambiguous, having at least two different, unre-
lated dictionary meanings (e.g. shniya in Hebrew means
secondboth as a unit of time and as an ordinal number). Ten
students, who did not participate in the experiment, were pre-
sented with 150 ambiguous words and asked to provide themeaning of each word when presented in isolation. The first
meaning given by each judge was designated as the domi-
nant meaning. Only words with a clear distinction between
their dominant (first choice of at least 75% of the judges)
and subordinate (first choice of no more than 25% of the
judges) meanings were chosen for the experiment. The 75
nonword targets were constructed by changing one letter of
each word target stimulus.
The priming stimuli were three-to-six letter words that
were related to either the dominant or subordinate meanings
of the ambiguous target words, or were completely unre-
lated. Related targets were generated in a word associationpretest in which 10 students were given each ambiguous
word and asked to write all related words that came to mind.
In a second pretest, the three related words most frequently
mentioned for each meaning of each ambiguous target word
were presented as a group to 20 different students and they
were asked to generate the corresponding ambiguous word
(the target). Only targets that were generated within 3 min by
at least 14 judges were selected for the experiments. Thus,
the final pool of priming stimuli consisted of 225 words
related to the dominant meanings of the ambiguous target
words and 225 words related to the subordinate meanings
of the ambiguous target words. The 225 unrelated priming
words were from the same list, randomly paired with unre-
lated target words.
Five kinds of word triads were constructed from these
priming words. Examples of priming word triads for the
target second (shniya in Hebrew) are provided following
the description of each type of prime:
1. A convergent (three dominant) prime: all three words
were related to the dominant meaning of the ambiguous
target word (minute, watch, time).
2. A divergent (two dominant + one subordinate) prime:
two words were related to the dominant meaning and
8/14/2019 kahana faust 2002
4/10
M. Faust, A. Kahana/ Neuropsychologia 40 (2002) 892901 895
one word was related to the subordinate meaning of the
ambiguous target word. The subordinate priming word
was always presented as the second stimulus, i.e. between
the two dominant priming words (minute, first, time).
3. A divergent (one dominant + two subordinate) prime:
two words were related to the subordinate meaning and
one word was related to the dominant meaning of theambiguous target word. The dominant priming word was
always presented as the second stimulus, i.e. between the
two subordinate priming words (first, watch, line).
4. A convergent (three subordinate) prime: all three words
were related to the subordinate meaning of the ambiguous
target word (first, last, line).
5. An unrelated prime: all three words were completely un-
related to the ambiguous target word (bomb, apple, joy).
As can be seen from the example given above, in the two
convergent priming conditions (1, 4) all three primes were
related both to the ambiguous target words and to each other,
while in the two divergent priming conditions (2, 3), oneprime was related only to the target word while the other two
primes were related both to the target word and to each other,
but not to the third prime [5]. However, the divergent primes
could also be related to each other via mediated priming,
with the target being the mediator [29]. In the unrelated
priming condition (5), all three primes were related neither
to the target word nor to each other.
There were 1500 experimental permutations [375 priming
stimuli (75 triads consisting of three unrelated words, 75 tri-
ads consisting of three dominant words, 75 triads of consist-
ing of two dominant words+one subordinate word, 75 triads
consisting of two subordinate words and one dominant word,75 triads consisting of three subordinate words) visual
field (left/right) lexicality (word/nonword)]. Five stimu-
lus lists, each containing 300 target stimuli (150 words, 150
nonwords) were needed to completely rotate items over ex-
perimental conditions in the full experimental design. Thus,
each participant viewed each target item twice, but in a
different experimental condition and visual field. Analysis
of the first versus second presentation of each target item
indicated that repetition did not affect the results, neither
as a main effect nor in interaction with any of the other
variables. Consequently, responses to both first and second
presentation of stimuli were included in all of the analy-
ses. Cell means were based on 15 trials per condition per
participant.
2.1.3. Apparatus and procedure
Each participant completed one experimental session, dur-
ing which he/she was exposed to a single list. Stimulus pre-
sentation and responses were controlled and recorded by a
Pentium PC 586 computer. The priming word triads were
presented simultaneously in normal horizontal orientation,
at, above and below the center of the screen, followed by the
target stimulus displayed 2 to the left or right of the cen-
trally presented +. Targets subtended, on average, 1.90
of horizontal visual angle (0.7 vertical) at a viewing dis-
tance of 50 cm.
The participant placed his/her right index finger on the
middle key of the computer mouse and waited for a focus-
ing signal (500 ms duration) which appeared on the cen-
ter of the screen and indicated the onset of a trial. Imme-
diately following the disappearance of the focusing signal,the priming word triad appeared for 500ms (pilot testing
indicated that the three words could be read and under-
stood within 500 ms). Participants were instructed to read
the words silently. Next, the focusing signal reappeared and
remained on the screen for 400 ms, until the end of the target
stimulus presentation, to ensure full fixation. The participant
was instructed at the beginning of the session to focus on
the central + and not to move his/her eyes while it was
present. Three hundred ms after the appearance of the fo-
cusing signal (total SOA800 ms), the target stimulus was
presented randomly to the RVF of LVF for 100 ms. Par-
ticipants were tested individually in a quiet, dim lit room.
They were instructed to indicate as rapidly and accuratelyas possible whether the target stimulus was a word or a non-
word by moving their finger from the middle mouse key
to the right or left mouse keys. Assignment of the keys to
word/nonword responses was counterbalanced over partici-
pants. The next trial began when a response was made or, in
the case of no response, after 3 s from the target onset. The
session began with a practice list, consisting of 30 priming
word triads and target stimuli not used in the experimental
lists.
2.2. Results
A 25 repeated measures analysis of variance with visual
field (left/right) and type of prime (dominant, dominant +subordinate, subordinate+dominant, subordinate, unrelated)
was performed on accuracy and correct RTs for target words.
Accuracy and RT for target nonwords presented to the RVF
and LVF were 68% and 1001 ms, and 63% and 999 ms,
respectively.
Means and S.D.s for response times and percent of correct
responses for ambiguous target words presented to the RVF
and LVF following the five types of primes are given in
Table 1. All correct RTs ranging from 300 to 1600 ms were
included in the analyses, 96% of all of the responses.
The main effect of visual field was significant for RT
and accuracy. Participants responded more quickly and ac-
curately to RVF (801 ms, 88%) than to LVF (845 ms, 81%)
target words, RT F (1, 19) = 13.9, P < 0.01; accuracy
F (1, 19) = 7.3, P < 0.01. In addition, a main effect of type
of prime was obtained for RT, F (4, 76) = 10.1, P < 0.001,
and accuracy, F (4, 76) = 15.9, P < 0.001. Post-hoc anal-
yses (Neuman Keuls, P < 0.05) revealed that this was due
to faster and more accurate responses to dominant (801 ms,
89%), dominant + subordinate (786 ms, 85%) and subordi-
nate + dominant (812 ms, 88%) primes than to either sub-
ordinate (847 ms, 84%) or unrelated (868 ms, 73%) primes.
8/14/2019 kahana faust 2002
5/10
896 M. Faust, A. Kahana/ Neuropsychologia 40 (2002) 892901
Table 1
Means and S.D.s for RT and percent correct for target words presented to the RVF and LVF following five different types of primes at 800ms SOA
Prime type RVF LVF
RT (ms) Percent correct RT (ms) Percent correct
Three dominant primes 743 (93) 92 (12) 853 (134) 85 (11)
Two dominant and one subordinate prime 759 (125) 87 (9) 813 (113) 83 (13)
One dominant and two subordinate primes 803 (101) 93 (6) 823 (112) 84 (11)Three subordinate primes 833 (130) 89 (13) 861 (102) 79 (13)
Unrelated primes 861 (107) 75 (13) 875 (141) 70 (21)
Responses to subordinate primes were more accurate than
to unrelated primes.
Of greatest importance, the predicted two-way interaction
between visual field and type of prime was obtained for RT,
F (4, 76) = 4.5, P < 0.01. The critical priming predictions
are represented by contrasting each type of prime to the
unrelated prime in the two visual fields.
Fig. 1 presents the overall priming in RT for target wordspresented to the RVF and the LVF following five types of
primes. As can be clearly seen in Fig. 1, for the RVF/LH
presented targets, differences between the unrelated con-
dition and the prime conditions were largest for dominant
[118 ms, F (1, 19) = 108.9, P < 0.0001] and dominant
+ subordinate [102 ms, F (1, 19) = 55.4, P < 0.0001]
primes. The facilitation from three and two dominant re-
lated primes was larger than from one dominant prime
(F (1, 19) = 3.5, P < 0.001). Thus, priming effects were
relatively smaller, although statistically reliable, for the
subordinate + dominant condition [58 ms, F (1, 19) = 17.1,
P < 0.001]. No priming was found for the subordinate
primes [28 ms, F (1, 19) = 2.2, P < 0.2]. However, forLVF/RH presented targets, differences between the unre-
lated condition and the prime condition were largest for the
dominant+subordinate [61 ms, F (1, 19) = 9.9, P < 0.005]
and the subordinate +dominant [52 ms, F (1, 19) = 4.3,
P < 0.05] priming conditions. No priming was found for
either the dominant [22 ms, F (1, 19) = 1.5, P < 0.3] or
subordinate primes [14 ms, F (1, 19) = 0.4, P < 0.6]. Re-
sponse latency was not correlated with response accuracy
Fig. 1. RT priming for target words presented to the RVF and LVF following four types of primes (800 ms SOA).
in either the RVF/LH, r(20) = 0.06 or the LVF/RH,
r(20) = 0.08. Thus, in accordance with previous research
[4] participants were not sacrificing accuracy for speed
when responding to either RVF/LH or LVF/RH targets
following the different prime combinations.
2.3. Discussion
The overall pattern of priming reported in experiment 1
indicates that the combinatorial influence of multiple re-
lated primes is different within each hemisphere. Results
for RVF/LH presented target suggest that at an SOA of
800 ms, the LH had no more access to subordinate meanings
of ambiguous words and could not benefit from the pres-
ence of primes that are related to the subordinate meaning
of the ambiguous target word. In the LH no facilitation was
uncovered for multiple primes that converged on a single
subordinate representation. Thus, the results for RVF tar-
get words presented after three subordinate primes are more
similar to that of single word primes than to that of sen-
tence primes [1,14,23]. Furthermore, facilitation from mul-tiple primes that diverged on both the dominant and sub-
ordinate representations seemed to be entirely dependent
on the presence of dominant primes. The facilitation from
two and three dominant related primes was larger than from
one dominant prime. Thus, priming in the LH tended to in-
crease with the number of dominant primes. In contrast, for
LVF/RH presented targets, only semantically divergent mul-
tiple primes facilitated target recognition significantly. Prim-
8/14/2019 kahana faust 2002
6/10
M. Faust, A. Kahana/ Neuropsychologia 40 (2002) 892901 897
ing from combinations of either two dominant related and
one subordinate related primes or two subordinate related
and one dominant related primes was significant whereas
three dominant or three subordinate primes did not have a
significant facilitation effect.
In experiment 2, the same primes and targets were again
presented to each hemisphere. The experiment was designedto determine how the availability of different combinations
of multiple primes changes over time in each hemisphere.
Therefore, in experiment 2 we used an even longer SOA, of
2500 ms, that is much longer than that used in most priming
studies of simple word recognition.
Previous research suggests that the differences between
the two hemispheres with respect to the availability of word
meanings may be more pronounced for long SOAs. Specifi-
cally, unique RH priming of more distant semantic relations
is found later in processing [17]. Furthermore, when very
long SOAs (several seconds) were used in the context of
insight problem solving, that requires the sustained avail-
ability of remote meanings, the RH tended to show not onlylarger priming effects than the LH, but even an advantage in
raw response times [7,10] and a tendency for an advantage
in response accuracy [26], an unusual result in normal, right
handed participants. Thus, results of experiment 2 were ex-
pected to replicate those observed in experiment 1 and to be
even more robust, showing greater priming for divergent and
for subordinate primes in the RH as compared to the LH.
3. Experiment 2
3.1. Methods
3.1.1. Participants
The 30 participants were Bar-Ilan University students,
native Hebrew speakers, aged 2025 years and strongly right
handed (laterality quotient of at least +90 on the Edinburgh
Inventory) [28]. All had normal or corrected to normal vision
and volunteered to participate in the experiment. None of
them had participated in experiment 1.
3.1.2. Materials
The stimulus pool consisted of priming words and target
word and nonword items, all in Hebrew, that were identicalto those used in experiment 1.
Table 2
Means and S.D.s for RT and percent correct for target words presented to the RVF and LVF following five different types of primes at 2500ms SOA
Prime type RVF LVF
RT (ms) Percent correct RT (ms) Percent correct
Three dominant primes 686 (152) 88 (9) 695 (130) 88 (9)
Two dominant and one subordinate prime 715 (157) 83 (12) 666 (111) 87 (14)
One dominant and two subordinate primes 707 (172) 82 (12) 673 (131) 86 (13)
Three subordinate primes 740 (184) 80 (14) 676 (127) 85 (10)
Unrelated primes 747 (179) 66 (16) 725 (152) 73 (17)
3.1.3. Apparatus and procedure
Apparatus and procedure were identical to those used in
experiment 1, except for the difference in SOAs. In experi-
ment 2, the priming words were presented for 2200 ms, so
that the total SOA was 2500 ms (2200 ms prime duration +300 ms focusing signal duration) as compared to 800 ms in
experiment 1 (500 ms prime duration+ 300 ms focusing sig-nal duration).
3.2. Results
A 25 repeated measures analysis of variance with visual
field (left/right) and type of prime (dominant, dominant +
subordinate, subordinate+dominant, subordinate, unrelated)
was performed on accuracy and correct RTs for target words.
Accuracy and RT for target nonwords presented to the RVF
and LVF were 72% and 803 ms, and 68% and 811 ms, re-
spectively.
Means and S.D.s for response times and percent of correct
responses for ambiguous target words presented to the RVF
and the LVF following the five types of primes are given in
Table 2. All correct RTs ranging from 300 to 1600 ms were
included in the analyses, 95% of all of the responses.
The main effect of visual field was significant for RT
and marginally significant for accuracy. However, the effect
was opposite to that expected. Participants responded more
quickly and accurately to LVF (687 ms, 84%) than to RVF
(720 ms, 80%) target words, RT F (1, 29) = 5.8, P < 0.05;
accuracy F (1, 29) = 2.4, P < 0.15. In addition, a main
effect of type of prime was obtained for RT, F (4, 116) =6.8, P < 0.001, and accuracy F (4, 116) = 30.4, P