Sie sind auf Seite 1von 18

Journal oj Psycholinguistic Research Vol.

12, No, 2, 1983

Semantic and Pragmatic Factors in the


Representation of "Near" and 'Far"

L u c i a C o l o m b o I a n d Philip H. K. S e y m o u r 2

Accepted August 3, 1982

The processing oJ two spatial terms which reJer to proximity and distance, near and far, was
investigated in a word-picture comparison task. Two alternative hypotheses were tested,
one based on the notion of linguisticmarkedness and supporting the positive status o f far,
and the other based on semantic-pragmatic factors and proposing that near is the positive
member oJ the pair. An analysis of the true-Jalse decision latencies indicated that near was
generally verifiedJaster than far in judgments in which the terms were used to describe both
ego-to-object and object-to-object distance. A third experiment addressed the question of
whether the advantage obtained in the verification oJ near may be accauntedJor in terms oJ
the priority oJ the interpersonal dimension oJ representation relative to the objective one.
The results are discussed with regard to the semantic representation oj the two words.

INTRODUCTION

This paper reports a series of experiments based on a semantic analysis of


the terms n e a r and Jar. These terms are used to express relative distance
on a spatial dimension which may be defined in terms of either (a) the
location of an object relative to the serf (the egocentric system), or (b) the
location of one object relative to another (the object-to-object system). As
is the case with other bipolar relations, the terms near and Jar are defined
relative to a norm which may vary from one context to another (Miller &

llstltuto dl Pslcologla, Umverslta dl Padova, Padova, Italy.


2Department of Psychology, University of Dundee, Dundee, Scotland.
75
0090-6905/83/0300-007550.300/00 Plenum Publishing Corporation
76 Colombo

J o h n s o n - L a i r d , 1976). To state that " X is near to Y" is to say that the


distance from X to Y is less than some previously established norm,
whereas to say that X is far from Y implies a distance which is greater than
the norm. This analysis defines Jar in terms of relatively large extent on
the dimension of distance, and near in terms of relatively small extent on
that dimension. However,jar is used both in a general or nominal sense to
indicate magnitude of distance (e.g., the question " H o w far is X ? " is open
to the answers " n e a r " or " f a r " ) , and in a contrastive sense (e.g., " X is
v e r y far from h e r e " ) , where the sense conveyed is that of great distance.
On the other hand, near always presupposes small distance, as ex-
emplified in the phrase " t h e house near the river," and shown by the fact
that expressions such as " t w i c e as f a r " are acceptable, whereas "twice as
n e a r " or " h a l f as n e a r " are somewhat strange and unusual. According to
this distinction, based on standard linguistic criteria (Bierwisch, 1967), Jar
m a y be considered the unmarked member of the pair, as in the case of
adjectives o f extent such as large, tall, and near, the marked member,
w h e r e the resemblance would be with small and short.
S t u d i e s carried out on the effect of markedness on a picture-word
matching task have shown that, in the case of locative expressions occur-
r i n g in the form of antonymous pairs the verification of the unmarked
m e m b e r is faster and more accurate than the verification of the marked
m e m b e r (Clark, 1974; Clark, Carpenter & Just, 1973; Seymour, 1975,
1979). If the conclusions from these experiments are extended to the pair
near-far, and if it is further assumed that the asymmetry in their usage
referred to above is related to lexical marking, one would expect a word-
picture comparison task in which the words near and far are matched
against the N E A R or FAR location of an object (relative to the self or to
a n o t h e r object), to reveal an advantage for far~FAR over near~NEAR
conditions when " t r u e " judgments are called for.
T h e r e are, however, further considerations regarding the semantics
o f near-far which should be taken into account. Metaphorical use of these
terms, especially in the areas of social and interpersonal relations, often
assigns a positive value to near and close and a negative value to far and
distant. Thus, in the "expression "friends who are near and dear to m e , "
near is used to refer to the intimacy of the relationships and carries
positive connotations. In the sentence " H i s work is far from being satis-
f a c t o r y , " on the other hand, Jar functions as the negation of a desirable
p r o p e r t y or quality. These metaphors may be based on an early stage in
the conceptualization of egocentric space, in which a distinction is made
b e t w e e n regions which are near to the self, containing visible accessible
objects, and regions which are far or remote, containing objects which are
Representation of "Near" and " F a r " 77

inaccessible, small, or vanishing. According to this distinction, perceptual


pragmatic properties of objects in egocentric space determine the se-
mantically positive quality of near with respect to Jar. One of these pro-
perties might be accessibility. Thus, in the statement "Xis near to F," the
preposition to carries a positive sense of approach (or ease of access),
whereas in the statement " X is f a r away from Y," the terms a w a y and
J r o m may be viewed as a negation of to, since the assertion "X moved to
Y" affirms that X is at Y, while the assertion " X moved away from Y"
states that X is not at Y (Gruber, 1976). Indeed, Lyons (1977) has argued
that terms such as near and j a r are implicitly dynamic in that underlying
them is the idea of movement, as in a journey with a specified starting
point and destination. If the notion of relative distance (greater or lesser
extent) is based on this dynamic conception of a metaphorical journey,
then a positive connotation should be conveyed by the term near (a near
object is more easily reached), but not by the term Jar (where the object is
not easily reached).
Another property of objects in egocentric space which bears on the
semantics of n e a r is that o f size; an object which is near to the self appears
relatively large, and this attribute might well be a cue in determining
distance (especially if other cues are lacking) as well as functioning in its
own right as a positive feature in the object representation. If it is further
assumed that relations in egocentric space have priority over those in
objective space (Clark, 1973; Miller & Johnson-Laird, 1976) then positive
connotations associated with n e a r may be extended from egocentric
space to objective space. Indeed, in analyses of the affective meaning
system carried out on various pairs of adjectives by Osgood, Sucy, and
Tannenbaum (1957), and Osgood, May, and Marion (1975), near was
found to be at the positive end of the scale and positively correlated with
adjectives having high loadings on the evaluation and activity factors. In
fact, the relation near-Jar itself shows in some of the analyses a not
insignificant loading on the evaluative factor.
According to the foregoing argument near possesses positive fea-
tures, some of which are derived from semantic-pragmatic properties and
others from attributes of egocentric space and these in turn may have
carried over into objective space. Consequently, near may be considered
the positive term of the pair. Such considerations lead to the prediction
that, on a word-picture comparison task, the matching of near would
show an advantage with respect to that of Jar, in terms of processing
accuracy or speed.
Thus, an analysis of the pair near-jar based on the notion of linguistic
markedness leads to different conclusions concerning which is the posi-
78 Colombo

tive member from one based on pragmatic/deictic considerations. It fol-


lows that depending on which analysis is correct, one would make differ-
ent predictions about the outcome of an experiment in which the terms
near and jar are matched against pictures incorporating the corresponding
relations. According to the linguistic markedness hypothesis and assum-
ing that the positive term is more easily processed, Jar would be verified
in a shorter time and with fewer errors than near. If, on the other hand,
the second hypothesis holds, then the opposite predictions would be ob-
tained.
In the following sections, three experiments will be described. In the
first, the terms near and Jar are judged in relation to pictures showing two
squares which are either large (NEAR) or small (FAR), the assumption
being that size functions as a cue in locating an object's distance from the
self. In the second experiment, the two terms are used to refer to inter-
object separation, and judgments are made about the distance between
two squares of the same size. An advantage of near~NEAR over Jar~FAR
judgments both in egocentric and in objective space would support the
hypothesis that near is the unmarked term. In the third experiment, the
two systems of representation (egocentric space and interobject space)
are simultaneously present and vary together, the aim being to establish
whether there is interference of one with the other. The occurrence of
interference would support the hypothesis that egocentric space has pri-
ority over objective space.

EXPERIMENT 1

The first experiment was conducted with the aim of determining


whether near judgments exhibit an advantage over Jar judgments in a
simple perceptual verification task. The task was designed to elicit near[
Jar judgments in relation to the egocentric dimension, with size being
employed as a cue for proximity or distance. In the main condition of the
experiment the subject was presented, on each trial, with the word near
or far printed below either two large squares (to be interpreted as NEAR)
or two small squares (to be interpreted as FAR). The task was to press a
" Y e s " key if the word matched the apparent distance of the shapes
(near[NEAR or Jar~FAR) and the " N o " key if it did not (near~FAR or
Jar[NEAR). It was expected, on the basis of previous research on the
prepositions above and below, that there would be an advantage for the
positive member of the pair on " Y e s " judgments, but no effect on " N o "
judgments.
Representation of " N e a r " and " F a r " 79

T w o supplementary conditions were also evaluated. The first of


these was intended to measure the strength of the markedness effect in a
relatively straightforward situation. The word large or small was pre-
sented and the subjects' task was to match it against the physical size of
the squares. Given that small is the marked (negative) member of this
pair, it was predicted that there would be an advantage for large/LARGE
o v e r small/SMALL judgments (see also Seymour, 1971). The second sup-
p l e m e n t a r y condition was designed to encourage the subject to represent
the squares in terms of greater or lesser distance from himself. The word
long or short was displayed, and the subject was told that it was to be
m a t c h e d against the apparent distance of the objects (two squares), that
is, the combinations short~NEAR and long~FAR were to be regarded as
true. It was hypothesized that long, as the expression of large extent on
the dimension (distance along the principal axis) would be the positive
m e m b e r o f the pair, and should therefore show a reaction time and ac-
c u r a c y advantage where " t r u e " judgments are made.

Method

Subjects. The subjects were 24 members of the University of Dundee.


T h e y were assigned at random to three groups of eight subjects each in
which the words used in the verification task were as follows: Group
1--near/far; Group 2--large~small; Group 3--short/long.
Apparatus. The experiments were conducted using an Apple II
m i c r o p r o c e s s o r . The stimuli were presented as white figures on a dark
g r o u n d on the 12 x 24 cms screen of a TV monitor linked to the video
o u t p u t o f the microprocessor. Words appeared in upper case characters
(standard Apple character set) at the lower center of the screen, 6 cms
beneath a pair of horizontally aligned identical squares. The size of the
squares on the visual display was 1 c m x 1 cm (small) or 3 cm x 3 cm
(large). The separation between the two squares was 4 cm in each case.
T h e subject observed the screen from a distance of 50-60 cms, and re-
s p o n d e d by pressing one of two keys (labeled YES and NO). The reaction
time was measured from the onset of the word-shape display to the instant
at which the key was pressed. At the beginning o f each trial, a row & f o u r
asterisks was displayed at the lower center of the screen for 1 sec. and
was then replaced, after an interval of .75 sec. by the word with the
squares above it. Following the response, the program displayed the re-
action time and an event code and indicated which key had been pressed,
and w h e t h e r the subject's response had been correct. If the response was
incorrect, a bleep sounded and the message " E R R O R " was displayed.
80 Colombo

The experimenter recorded errors via the keyboard, and the trials on
which these occurred were repeated at a later point in the sequence.
Procedure. Subjects were given preliminary practice in the use of the
apparatus by presenting the word yes or no on the screen and requiring
them to respond by pushing the appropriate key. The task was then ex-
plained as one which involved either: (1) the matching of the words near
or Jar against the apparent distance of the large (NEAR) or small (FAR)
squares (Group 1), (2) the matching of the words large or small against
the LARGE or SMALL squares (Group 2), or (3) the matching of the
words long or short against the apparent (NEAR or FAR) distance of the
shapes from the observer (Group 3). Following practice on the task, a
sequence of 64 trials was presented, involving 16 presentations of each of
the four word-shape combinations. Errors and other failures were re-
corded and the trials on which they occurred replaced. The trials were
randomized for each subject.

Results

The results obtained in the three experimental conditions were


analysed separately. The first one considered is the large-small condition,
which was included in order to verify the occurrence of the markedness
effects which have generally been found in studies of this kind. The re-
sults are summarized in Table I, and show the expected advantage for
large/LARGE over small~SMALL judgments on trials calling for " Y e s "
responses, both in terms of reaction time and error frequency. The RT

Table I. Mean "True" and "False" Reaction Time (in msecs) and Error Rates in
the Ego-to-Object Verification Task (Experiment 1)

Condition True False

large/LARGE small/SMALL large/SMALL small/LARGE


RT 639 826 797 796
1 Error per cent 0.6 6.2 1.3 2.9

near/NEAR far/FAR near/FAR far/NEAR


RT 861 1150 1154 1088
2 Error per cent 0.4 4.5 1.6 1.7

short/NEAR long/FAR short/FAR long/NEAR


RT 989 1056 1155 1115
3 Error per cent 2.6 3.3 4.3 1.9
IIII
Representation of "Near" and " F a r " 81

advantage was shown by all eight subjects (F = 47.00; df = 1,7; p <


0.001), and the accuracy advantage by seven of them. Negative responses
were slower than affirmative (F = 7.1, df = 1,7; p < 0.05), but were not
subject to effect of markedness. This outcome is qualitatively similar to
that o f previous studies, such as those on above and below carried out by
S e y m o u r (1969) and Chase and Clark (1971).
T h e results for the near-far condition are also shown in Table I.
T h e s e data are qualitatively similar to those obtained for large-small, and
provide clear evidence for there being an advantage of near over far
judgments. All eight subjects obtained lower RTs for near/NEAR judg-
ments than for Jar~FAR ones and seven of the eight show an error ad-
vantage as well. This effect occurred for affirmative responses (F = 59.8,
d j = 1,7; p < .001) but not for negative ones, which were, overall, also
slower than the former type (F = 20.4, dJ = 1,7; p < .01).
The third condition, using the words long and short, was designed to
e n c o u r a g e the subject to think in terms of long or small distance, the
assumption being that long (referring to large extension on this dimension)
would function as the unmarked member of the pair and produce a re-
versal o f the effect found in the previous two conditions (i.e., an ad-
vantage for long/FAR over short~NEAR). The results summarized in
Table I indicate that this reversal did not occur. The mean RT for short/
NEAR comparisons was 67 ms faster than that for long~FAR. This pattern
was shown by seven of the eight subjects, but the eighth showed a large
difference in the opposite direction, with the result that, overall, the effect
failed to reach a significant level (F = 4.69, df = 1,7; p < .05). As in the
previous two conditions, affirmative responses were in general faster than
negative ones ( F = 10.7, dJ = 1,7; p < .05).

Discussion

The reaction times obtained in the first conditions, in which the


words large and small were compared with figures of different size, pro-
vide a baseline measure of the time required to match a word with a figure
w h e r e the attribute in question is simply size. A comparison with the
near-jar condition, where the relation between the perceptual display and
the words was not so direct, shows a general increase of the order of 300
msec in the magnitude of the RT's, which might be explained by the
necessity for an intermediate " t r a n s f o r m a t i o n " to convert the initial en-
coding of the stimulus figure in terms of size into one which can be
m a t c h e d with the lexical code (i.e., L A R G E transforms to PROXIMAL
and S M A L L to N O N - P R O X I M A L ) . In the first condition, an increase (of
82 Colombo

the order of 200 msec) in the time required to verify "true" or "false"
items containing small and/or SMALL, by comparison with " t r u e " items
containing large, may be attributed to the presence of a negative com-
ponent in either the marked or perceptual representation of the former,
and suggests that the increase in the amount of processing time required
for the encoding of the marked term relative to the unmarked one is about
the same as for the negative transformation of the response code. A
similar trend is observed in the near-Jar condition, that is, an increase of
about 250-300 msec for both affirmative and negative responses to com-
binations with jar by comparison with affirmative responses to near/
NEAR. This seems to suggest that Jar behaves like the marked member of
the pair, and that the representation embodies the relation "non-
proximal" rather than "distant."
A more intriguing result is that found with the adjective pair long-
short; although the difference was not significant, there is a clear ten-
dency for combinations containing short to be verified faster than those
with long. This suggests that, although the two terms normally refer to
distance, an interpretation in terms of proximity is rendered appropriate
by the requirements of the task, and an additional stage of processing may
be carried out above or beyond that involved in matching combinations
with near and Jar, whereby size is first given in terms of proximity and the
latter code is then transformed into one for distance (long/short).
At this point we shall consider the hypothesis that the faster verifica-
tion of near and short combinations by comparison with Jar and long ones
occurs on account of the presence of a positive feature in the representa-
tion of the large version of the figure. This interpretation would require an
effect of size on both positive and negative responses, whereas no effect
was evident in the latter case; a finding which will be discussed at greater
length below. It is instead possible that the advantage found for near
judgments relative to Jar ones depends on which system of representation
is used in the task; when distance is measured with respect to the self, the
privileged relation may be "proximity," while for the interobject distance
the predominant relation may be "extent of separation" between the
objects. If the latter situation obtains, one would expect an advantage for
jar judgments over near ones. However, if the relation of "proximity" is,
whatever the situation, more central in the representation of near-jar than
that of "distance," then the advantage of near over Jar should also be
found in a task in which interobject judgments are made with respect to
distance.
Representation of "Near" and " F a r " 83

EXPERIMENT 2

An experiment was carried out in which the terms near and Jar were
matched against the distance between two objects. For those items in-
volving judgments of near the separation of the objects was either small (2
cms) or very small (I cm) while for items with Jar the separation was
either large (6 cms) or very large (12 cms). The different distances were
used in both NEAR and FAR conditions to check for a possible effect on
processing time of scanning the space between the objects. If such an
effect was present, the RTs for both near/NEAR and Jar]FAR judgments
should increase as a function of the distance between the squares, and an
advantage of near over Jar judgments should be most evident when
N E A R is represented by a very small separation, and FAR by a very large
separation. The alternative hypothesis would instead predict that the
smaller the separation of the objects, the easier they should be to judge in
relation to jar, on the assumption that the encoding of the picture would
be facilitated by their greater discliminability.

Method

Subjects. The subjects were 10 volunteers, students of the University


of Dundee.
Apparatus. The apparatus was the same as that used for the preced-
ing experiment.
Procedure. The experiment involved matching of the words near and
Jar, presented as previously at the bottom center of the screen, against the
object-to-object separation of two 2 cm • 2 cm squares. The distance
between squares in the N E A R condition was either 1 cm or 2 cm, and in
the F A R condition either 6 cm or 12 cm. Following a series of practise
trials the eight word-picture combinations were each presented 12 times
in a randomized sequence of 96 trials.

Results and Discussion

The results are summarized in Table II. An analysis of variance of the


affirmative RTs indicated that near judgments were significantly faster
than Jar judgments (F = 8.9, dJ = 1,9; p < .05), and that there was an
effect of separation (large versus small), which interacted with near
versus Jar (F = 25.4, dJ = 1,9; p < .001). The interaction was such that,
84 Colombo

Table II. Mean "True" and "False" Reaction Times (in ms) and Error Rates for
Object-to-Object Verification with Variation in Object Separation (Experiment 2)

Display
Separation near/NEAR far/FAR near/FAR far/NEAR

Close RT 839 1107 1050 984


Error per cent 0.3 4.9 2.0 1.2

Far RT 833 915 1006 1048


Error per cent 0.4 2.6 1,1 1.2

whereas separation had no effect on near judgments, it did have a strong


effect on Jar judgments (F = 32.5, dj = 1,9; p < .001). Overall negative
responses were again found to be slower than affirmative responses (F =
29.5, dj = 1,9; p < .001), but there were no significant effects of either
lexical item or object separation on the negative RTs.
The advantage on trials of near judgments with respect to jar ones
found in Experiment 1, was confirmed in this Experiment; near being
verified on average 175 msec faster than Jar when the results were col-
lapsed across variations in separation. The change in the system of repre-
sentation from the egocentric to the objective, did not produce a reversal
of the near/jar difference as would be predicted if the analysis based on
interpersonal distance was true for egocentric space and the linguistic
markedness hypothesis held in object-to-object space. It is interesting to
note that the time required to match near is about the same as in the first
experiment suggesting that approximately the same amount of processing
time might be involved in the two tasks and that the two pictorial attri-
butes (size and separation) are equivalent in this respect when interpreted
in terms of proximity. While the variation in separation did not affect the
verification of items containing near, it seems to have influenced the
verification of Jar/FAR displays. In fact, the interaction between lexical
item and separation on affirmative trials in Experiment 2 is such that in
the case of jar items the larger separation produced a decrease in the RTs,
a result which is contrary to the predictions of the scanning hypothesis,
since if the judgment that the figures are FAR apart involves a time-
consuming switching of attention from the one to the other, then such a
process should have combined additively with the response. Rather, the
effect is more likely to be explicable in terms of the greater discrimin-
ability of the more widely spaced pair. However, the pattern of results
does not appear to be explicable simply in terms of perceptual processes
Representation of "Near" and " F a r " 85

associated with the encoding of the figures since the effect of separation
on responses to FAR figures depends on which lexical item is involved in
the comparison. Reduction in the distance between the squares has a
marked effect when the subject verifies items containing Jar (i.e., on a
J a r / F A R trial) but not when he verifies ones with near (i.e., on a near/
F A R ) trial.
The results obtained in Experiment 2 confirm that near is easier to
process than jar, suggesting that even when the system in terms of which
the judgment is made does not take the self as reference point but instead
refers to interobject separation, the representation of the situation is in
terms of proximity, rather than distance, such that near and Jar take
positive and negative values, respectively, on the dimension concerned.
Had the magnitude of separation been interpreted in terms of small or
large distance in Experiment 2, then near would presumably have
functioned as the marked term, so that comparisons between it and the
figures would have taken more time than in Experiment 1, where, on both
perceptual and pragmatic grounds, it could be regarded as possessing
positive features. However, no such effect occurred (mean RTs for the
verification of near being 833 msec in Experiment 2 and 861 msec in
Experiment 1). On the other hand, although a negative feature in the
representation of small objects could account for the larger reaction time
for f a r [ S M A L L judgments (1150 msec in Experiment 1), it is not sufficient
to explain the overall advantage of near, since the prediction in the case of
the condition where near is matched with small separation would have
been of a reversal of the effect in favor of far.

EXPERIMENT 3

/It was argued in the introduction that the semantic representation of


near incorporates a positive feature as a consequence of the egocentric
dimension being encoded in terms of proximity rather than distance and
this property may account for the advantage in terms of matching speed of
near over far. It was also suggested that egocentric space may have
priority over objective space, and that as a result the positive feature
which is characteristic of near on the former dimension is assigned to the
same term on the latter. Experiment 3 was planned to test this hypothesis.
If the egocentric system is in some sense primary, a judgment regarding
the separation of two objects should be sensitive to variations in the
apparent proximity of the objects to the observer. Thus, if the subject is
required to judge whether two squares are near or far relative to one
86 Colombo

another, and if the shapes, as displayed on the screen, appear large or


small, this additional factor of size may influence the matching process in
such a way that large shapes facilitate NEAR judgments but inhibit FAR
judgments, and vice versa for small shapes. However, if the egocentric
dimension has priority, one would not expect to obtain the converse
effect of an influence of object-to-object separation on judgments of ego-
centric proximity.
In order to test these hypotheses a further experiment was designed in
which the words near and far were matched against pairs of squares
which varied in size (large or small) and separation (near or far). One
group of subjects was asked to match the words against the horizontal
separation of the objects, ignoring the variation in size. The second group
was asked to match the words against the apparent distance of the objects
from the self (with LARGE implying near, and SMALL implying far),
ignoring their separation.

Method

Subjects. The subjects were 30 volunteers from classes at the Uni-


versity of Dundee. Fifteen were assigned to the object-to-object task, and
15 to the ego-to-object task.
Apparatus. The apparatus was the same as for the preceding experi-
ment.
Displays. Eight types of display were prepared by printing the word
near or Jar at the lower center of the screen beneath two horizontally
aligned squares, either 1 c m x 1 cm or 3 cm x 3 cm, which were separated
by either 1 cm or 10 cm.
Procedure. Subjects were first given an explanation of the task, and
then a series of practice trials. Those in Group 1 were instructed to com-
pare the words against the horizontal separation of the squares, ignoring
the size variation. Those in Group 2 were requested to match the words
against the apparent proximity of the shapes, the expectation being that
the relatively large figures would be judged as near and the small ones as
remote, and that the variation in separation would be ignored. In the main
trial sequence each display was presented 16 times in a randomized order,
making a total of 128 trials.

Results

EJject oJ irrelevant size on object-to-object comparisons. Table III


gives a summary of the results obtained for Group 1 (object-to-object
Representation of "Near" and "Far" 87

Table III. Mean "True" and "False" Reaction Times (in ms) and Error Rates for
Object-to-Object Verification with Variation of Object Size (Experiment 3)

Display
Square size near/NEAR far/FAR near/FAR far/NEAR

Large RT 822 962 1041 998


Error per cent 0.9 1.4 1.1 0.9

Small RT 925 956 1019 1047


Error per cent 1.0 1.5 0.6 0.2

c o m p a r i s o n s ) . F o r affirmative responses, there was a 103 msec advantage


for large o v e r small figures in respect of near~NEAR judgments. This
effect was s h o w n by 11 of the 15 subjects. The converse effect predicted
f o r Jar/FAR j u d g m e n t s did not occur, the advantage of small o v e r large
figures being only 6 m s e c (shown by seven of the 15 subjects). An analysis
o f v a r i a n c e of the affirmative RTs indicated significant effects for size (F
= 9.9, dJ = 1,14; p < .01) and lexical item (near/jar) (F = 5.3, dJ = 1,13; p
< .05), and also a significant interaction between these factors (F = 9.8,
dJ = 1,13; p < .01). Further analyses of variance carried out on affirma-
tive r e s p o n s e s confirmed that the size effect was not significant for Jar~
F A R j u d g m e n t s , but that it was so for near[NEAR judgments (F -- 16.7, dJ
= 1,13; p < .01). As in the previous studies, negative responses were in
general slower than positive responses (F = 62.2, dJ = 1,13; p < .001).
T h e truth value interacted with both size and lexical item (F = 23.7, dJ =
1,13; p < .001), the near-Jar and size effects found for affirmative re-
s p o n s e s being absent for negative responses.
EJJect oJ irrelevant separation on ego-to-object comparisons. The
results for the second condition are summarized in Table IV. This experi-
m e n t c o n f i r m e d the findings of E x p e r i m e n t 1 insofar as it indicated a
strong effect in f a v o r of near judgments on affirmative trials (F = 21.3, dJ
= 1,13; p < 0.001). This effect was shown by 14 of the 15 subjects,
although one exceptional subject produced what was clearly a reversed
p a t t e r n , with errors and elevated RTs on near/NEAR judgments. Nega-
tive r e s p o n s e s were again consistently slower than affirmative responses
( F = 57.9, dJ = 1,13; p < .001), and the near-Jar effect was not significant
on negative trials. There was a small overall effect related to the separa-
tion of the shapes ( F = 7.5, dJ = 1,13; p < .05). This occurred because
r e s p o n s e s to the wider separation were 20 msec slower than responses to
the closer separation. H o w e v e r , this effect interacted with words and
88 Colombo

Table IV. Mean "True" and "False" Reaction Times (in ms) and Error Rates for
Ego-to-Object Verification with Variation in Object Separation (Experiment 3)

Display
Separation near[NEAR far/FAR near/FAR far/NEAR

Close RT 818 1014 1065 992


Error per cent 0.2 1.6 0.4 0.7

Far RT 851 1025 1050 1053


Error per cent 0.4 1.8 1.2 0.8

responses (F = 4.8, dJ = 1,13; p < .05). Follow-up tests suggested that the
only point at which separation exerted a consistent effect was on negative
responses to Jar/NEAR displays, where the RT was facilitated if the
shapes were close together.

Discussion

The aim, in this experiment, was to test the hypothesis that the
egocentric system is dominant in the representation of proximity. It was
anticipated that a variation in object size would intrude on judgments of
object-to-object separation, in such a way that large size would facilitate
near/NEAR judgments, and small size facilitate Jar/FAR judgments. At
the same time it was not expected that an irrelevant variation in object-to-
object separation would interfere with judgments of object-to-ego proxi-
mity.
The results obtained did not completely confirm these expectations.
Although presentation of large squares did facilitate near/NEAR judg-
ments, the converse effect of small size on JarlFAR judgments was not
observed. Moreover, variations in size had no effect on the negative
response reaction times. However, if the influence of size variation is due
to an effect occurring in the picture encoding stage, one would expect it to
be symmetrical, that is, for large object size to facilitate NEAR and inhibit
FAR judgments, and conversely for small size. Furthermore, an effect
related to picture encoding should have been evident on "false" trials as
well as on " t r u e " trials.
The results obtained suggest that the perceptual representation of an
object is not simply a function of the latter's physical characteristics, but
is to some extent constrained by the semantic features of the accompany-
ing verbal item. The concept activated when a word is presented may
Representation of "Near" and " F a r " 89

contain features which are more or less congruent with the characteristics
of the figure. If this match does not occur, or if the characteristics of the
object only partially match the concept to be verified, an increase in the
amount of processing time may be required. Interference may occur when
a feature which is a constituent of the concept to be verified but not
relevant to the task at hand, becomes activated. It may be that, in the
context of interobject separation, the semantic representation of prox-
imity is based on features related both to intrinsic properties of the objects
(e.g., size), and to properties of the space between them (e.g., amount of
separation), and that both become active on presentation of the verbal
item. In the representation of distance, on the other hand, only the sepa-
ration of the two objects is encoded. Thus, when near is to be compared,
both separation and size are encoded giving rise to the possibility of
interference of one with the other. A near~NEAR display containing large
squares will be assigned a positive value on the active dimensions, and no
conflict will occur, whereas a near~NEAR display containing small
squares will be assigned a negative value on the size dimension, so that
the presence of concurrently active features of opposite polarity may
require an increase in the amount of processing time. In the verification of
jar, on the other hand, the feature relating to the size of the objects will
not have been activated and will not, therefore, produce interference.
When size is the dimension relevant to the task (as in the second
condition of Experiment 3) and separation varies, interference would only
be expected if both dimensions of distance (ego-object and object-object)
had the same "weight," or in terms of the preceding argument, the same
degree of activation. As no interference was obtained in the aforemen-
tioned task, either in the verification of near or in the verification of Jar, it
can be argued that relations in egocentric space take precedence over
those in objective space. The only effect obtained in the matching of Jar
with objects of different size was an increase in the RTs, which was also
found in Experiment 1 and was there attributed to the additional process-
ing required to encode size in terms of distance.

GENERAL DISCUSSION

The aim of this study was to investigate the semantic structure of the
antonymous pair near-Jar. As in the case of other bipolar adjective pairs,
the opposition may be thought of in terms of a scale lying between two
poles or extremes, with the unmarked term lying at the positive end and
the marked term at the negative end. However, in the case of near~jar, it
90 Colombo

is not clear which term lies at the "positive" end, and it was the purpose
of the experiment reported to examine this issue both in relation to ego-
centric usage of the term, where reference is made to the proximity of an
object to the self, and in relation to object-to-object usage. According to
the conventional linguistic analysis, applied to either the ego-object or
object-object situation, f a r is the unmarked term and, in a picture-word
comparison task should be matched faster than the marked term near.
Alternatively, one could argue, on semantic-pragmatic grounds, that near
is endowed with a positive feature, through its association with closeness
on the interpersonal dimension, where the position of the self is (in
Clark's terms) that of "canonical" reference point. On the latter view, the
status of near and Jar may not be constant, but could depend on the
situation, so that the positive feature might shift from one term to the
other according to which system of representation is appropriate to the
task at hand, giving near a processing time advantage in the egocentric
system, and j a r in the interobject system.
The results of all three experiments suggest that the representation of
near is, in general, positively defined. The advantage found for near was
greatest in Experiment 1, where size was used as a cue for egocentric
distance, suggesting that its positive character derives from perceptual
properties of the referent. According to this interpretation properties such
as size play an important role in the representation of proximity with the
result that the interpretation of large figures as near requires less process-
ing than the interpretation of small figures as Jar, on the one hand, and
short, on the other, where the translation from size to magnitude of sepa-
ration is less direct. However, although the near-jar difference found in
the ego-to-object task may be related to the use of size as a cue for
proximity, this parameter cannot, in general, account for the advantage of
near over Jar. Firstly, the effect was also found in the tasks in which
interobject distance was matched (Experiments 2 and 3). Moreover, the
times required to match near show very little differences in going from
one system of representation to the other (861 msec (Experiment 1), 834
msec (Experiment 3) for the ego-to-object task; 836 msec (Experiment 2),
873 msec (Experiment 3) for the object-to-object task) suggesting that a
similar amount of processing is required to verify near, whether size or
extent of separation are employed as cues. In contrast, the verification of
j a r seems to be sensitive to the type of cue used in the different tasks
(ranging from 915 msec for the object-to-object task (Experiment 2) to
1150 msec for the ego-to-object task (Experiment 1)).
This pattern of results suggests that the linguistically positive
character of near relative to Jar owes something to semantic and prag-
Representation of "Near" and " F a r " 91

matic factors connected with the closeness of objects, as well as to per-


ceptually derived properties of the latter.
It was pointed out above that the self is the canonical reference point
on the dimension of distance, and that for this reason interpersonal dis-
tance may assume a priority over interobject distance. Experiment 3,
which was designed to test this hypothesis by means of an interference
paradigm, showed that near is subject to interference in one direction
only. That is, there is an effect of size on judgments of object-to-object
separation but no effect of separation on judgments of ego-object dis-
tance, supporting the argument that relations in egocentric space take
precedence over those in objective space. Judgments of jar, on the other
hand, are not subject to interference at all. The results of Experiment 3,
therefore, fail to support the strong version of the hypothesis that the
egocentric dimension is dominant with respect tO the interobject one, but
suggests that a degree of asymmetry is presentin the conceptual repre-
sentation of near and Jar. In the representation of near, the size of the
object has an important influence on judgments of both ego-to-object and
object-to-object separation, the most likely explanation being that the
positive feature of near originates in the former system and carries over to
the latter. By the same token, the representation of the term Jar when
used to refer to relatively great distance, on the egocentric dimension or
by extension on the object-to-object dimension acquires a negative con-
notation. It would appear that the positive character of Jar deriving from
its status as the linguistically unmarked term, is nullified or reversed by
virtue of the richer range of positive values, associated with positivity on
the interpersonal dimension, and arising out of perceptual properties,
such as size and dynamic notions relating to accessibility and approach.

ACKNOWLEDGMENTS

The final version of this paper was written while the first author was on
study leave at the Max Planck Institut fur Psicholingistik (Nijmegen, Hol-
land), supported by a C.N.R. grant. Acknowledgments are gratefully
given to Ino Flores d'Arcais for his extensive criticisms and suggestions,
and to Graham Richardson for substantial help in editing the final manu-
script. Address correspondence to: Lucia Colombo, Istituto di Psi-
cologia, Piazza Capitaniato, Padova.

REFERENCES

Bierwisch, M. Some semantic universals of German adjectivals. Foundations of Language,


1967, 3, 1-36.
92 Colombo

Chase, W. G., & Clark, H. H. Semantics in the perception of verticality. British Journal of
Psychology, 1971, 62, 311-326.
Clark, H. H. Space, time, semantics, and the child. In T. Moore (Ed.), Cognitive develop-
ment and the acquisition of language. New York: Academic Press, 1973.
Clark, H. H. Semantics and comprehension. In T. A. Sebeok (Ed.), Current trends in
linguistics, XII: Linguistics and adjacent arts and sciences. The Hague: Mouton, 1974.
Clark, H. H., Carpenter, P. A., & Just, M. A. On the meeting of semantics and perception.
In W. G. Chase (Ed.), Visual information processing. New York: Academic Press,
1973.
Gruber, J. S. Lexical structures in syntax and semantics. Amsterdam: North Holland, 1976.
Lyons, J. Semantics (Vol. 2). Cambridge: Cambridge University Press, 1977.
Miller, G. A., & Johnson-Laird, P. N. Language and perception. Cambridge: Cambridge
University Press, 1976.
Osgood, C. E., May, W. H., & Miron, M. S. Cross cultural universals ofaffective meaning.
Urbana, Illinois: University of Illinois Press, 1975.
Osgood, C. E., Sucy, G. J., & Tannenbaum, B. H. Measurement of meaning. Urbana,
Illinois: University of Illinois Press, 1957.
Seymour, P. H. K. Response latencies in judgments of spatial location. British Journal of
Psychology, 1969, 60, 31-39.
Seymour, P. H. K. Semantic equivalence of verbal and pictorial displays. In R. A. Kennedy
& A. L. Wilkes (Eds.), Studies in long term memory. London: Wiley, 1975.
Seymour, P. H. K. Human visual cognition: A study in experimental cognitive psychology.
London: Collier Macmillan, 1979.
Seymour, P. H. K. Perceptual and judgmental bias in classification of word-shape displays.
Acta Psychologica, 1971, 35, 416-477.

Das könnte Ihnen auch gefallen